Vous êtes sur la page 1sur 11

EXPERIMENT 30 THIN FILM DEPOSITION

1 Equipment list
Complete deposition system consisting of

Glass vacuum chamber


Protective clear plastic cylinder
Rotary and diffusion pumps and controllers
Evaporation unit and controller
Magnetron sputtering coating unit and controller
Crystal thickness monitor

Aluminium wire for evaporation


Tungsten wire for melting the aluminium
Microscope slides on which to deposit aluminium and copper films
Multimeter for measuring film resistance

References: References are listed at the end of these notes. Before attempting the ex-
periment, you are strongly encouraged to download the papers from the Librarys e-journal
portal at http://www.library.usyd.edu.au/ejournals/

1
2 Aim
In this experiment you will use evaporative and magnetron sputtering techniques to produce
thin films of copper and aluminium onto a glass slide (substrate is the technical word).
Moreover, you will measure the resistance of the copper films during deposition in order to
calculate the percolation threshold (more on this later). As a result, you will learn about the
mechanism with which thin films are formed. From the resistance measurements you will
then determine the bulk resistivity of copper and compare it to tabulated bulk values. Use
your knowledge of the thin film deposition mechanism in order to explain any discrepancy
between the two values.

3 Introduction
Thin film deposition technology is a major area of scientific research because of its wide
range of applications such as optics, electronics (such as microprocessors), biotechnology and
the tool manufacturing industry. Examples include anti-reflection coatings, interconnects
for printed circuit boards, bio-compatible films for surgical implants and ultra-hard films
decreasing the wear rate of machinery parts. The purpose of this experiment is to become
familiar with vacuum equipment and processes, and two types of metal film deposition
techniques - evaporation and sputtering.
Almost all film deposition techniques use vacuum conditions to reduce contamination
and assist in film growth. In this experiment you will use a vacuum chamber to deposit
aluminium contacts on a glass substrate by a process known as evaporative deposition. The
contacts will then be used to monitor the conductivity of a copper film deposited using
a magnetron sputter coater. The dynamics of thin film growth and the mechanisms of
conductivity in discontinuous metal films will be investigated.

3.1 Evaporation System

Evaporation Unit

The apparatus consists of a glass cylinder vacuum chamber with a rotary vacuum pump, an
oil vapour diffusion pump and pressure gauge. An evaporation source is mounted within the
glass chamber. The source consists of twisted tungsten wires connected to a 60A current
source at one end and the other end connected to earth. A small piece of aluminium wire

2
is wound around the tungsten, which becomes very hot when the current source is turned
on. Since tungsten has a melting point of 3410 C and aluminium has a boiling point of
2467 C at atmospheric pressure the aluminium will boil before the tungsten begins to melt
(in fact both these temperatures are reduced at low pressure. In our system, Al begins to
evaporate at 900 C and violently evaporates at 1200 C). Aluminium vapour quickly fills
the space (since at our operating pressure of about 106 atmospheres, there are many fewer
gas molecules with which to collide), and condenses on any cold surface forming a metallic
film.

3.2 Magnetron Sputter Coater

Mounted in the lid of the vacuum chamber is a magnetron sputter coater. This consists
of a high voltage source for the production of an argon plasma by the initiation of a gas
discharge between two electrodes. Positively charged argon ions are accelerated toward the
negative electrode (cathode). The momentum transfer from argon ions with high kinetic
energy impacting on the cathode can be enough to overcome the cohesive energy (the
energy required to break the atoms of the solid free into isolated atoms) of the magnetron
target. Atoms are dislodged (sputtered) from the targets surface and can be recondensed
on another surface forming a thin film of the original solid. A generalised form of sputtering
yield (Sy ) may be expressed as:
M
Sy = ( ) (1)
Ec

where M is the atomic mass and Ec is the cohesive energy. The value of the constants,
and , are determined by the operational conditions, such as the bombarding species (e.g.
Ar+ ), the impact energy per ion, the current density and the impact direction. The cohesive
energies of Al and Cu are approximately 3.4 and 3.5 eV/atom. (For more on cohesive energy,
see Kittel [1]).
A toroidal (or other shaped) magnetic field can significantly improve the efficiency of
the ionization process in the discharge by confining the trajectories of secondary electrons
to helical paths around the magnetic field lines causing the electrons undergo more ionizing
collisions with neutral gases near the target than would otherwise occur. This enhances the
ionisation of the plasma near the target leading to a higher sputter rate. It also means that
the plasma can be sustained at a lower pressure. The sputtered atoms are overwhelmingly
neutrals and so are not affected by the magnetic trap.
For a given sputtering current, sputter rates are improved by an order of magnitude
by the application of a magnetic field. Magnetically assisted sputtering units are termed
magnetrons.

3.3 Crystal monitor

A vibrating-crystal film thickness monitor is mounted in the vacuum chamber. The vibration
frequency of the crystal is proportional to the mass of material deposited on its surface.
Calibrating this change in frequency to the atomic weight of the deposited material gives
an accurate measure of the deposited film thickness. This method breaks down when the
mass of the deposited material becomes too great and the calibration becomes non-linear.
In this case the crystal needs to be replaced.

3
Crystal monitor

4 Theory
4.1 Film growth
The early stages of film growth can be classified into two distinct types. Uniform coverage of
the substrate by the depositing film followed by layer-by-layer growth, is called Frank-Van
der Merwe growth mode. Film growth facilitated by nucleation and agglomeration of islands
is called Volmer-Weber growth mode. A third mode (Stranski-Krastinov mode) describes a
combination of the two. In general, metals on insulators grow via the Volmer-Weber growth
mode, depending on the surface energies of the depositing material and the substrate. For
a given depositing material (e.g. Cu), the nucleation and agglomeration processes will be
different on different substrates (e.g. glass, Si wafer, polymer).

Volmer-Weber growth

Nucleation of islands on the substrate surface occurs when a metal atom or ion imparts some
or all of its kinetic energy to the substrate causing it to adsorb to the substrate surface.
Adsorbed atoms (adatoms) are bound to the substrate by either Van der Waals or covalent

4
forces. If the adatoms have enough residual kinetic energy to overcome these binding forces
they can diffuse on the surface or desorb from the surface back into the gas phase. Diffusing
atoms will move around the surface until they lose some energy or encounter a region where
the binding forces are strong enough to pin them down. Depending on surface energy, the
diffusing atoms may be mobile on time scales of several seconds. Surface defects such as grain
boundary dislocations, step edges or surface impurities are common regions where diffusing
adatoms will become bound. Bound adatoms themselves then become points where other
diffusing adatoms can bind, thus nucleating islands.
Each nucleation site will become an island and each island will become a metal crystallite.
The density of grain boundaries is determined by the size of the metal crystallites, which is
initially determined by the distance between nucleation sites. Since bulk electrical resistance
is dependant on electron scattering from grain boundaries and surfaces as well as from
voids and defects, the nucleation and agglomeration processes directly affect the electrical
properties of the bulk material. When the film is thick enough that scattering from surfaces
is negligible the resistivity will be equal to that of the bulk material of the film [2]. Thick
enough means greater than the mean free path of the (delocalised) conduction electrons,
which, for Cu, is 39 nm.
Impurities in the sputtered material (Cu often contains significant O2 , Ag, P, Ni, etc.)
also affect electrical conductivity. In the case of Cu, will these impurities increase or decrease
conductivity?

4.2 Percolation threshold


Island growth continues until the islands coalesce with one another, forming a contiguous
film with connecting pathways. The stage at which this just occurs is often termed the
percolation threshold. Percolation is of great interest to engineers and mathematicians,
and also makes an appearance in theoretical physics, particularly in conformal field theory
and quantum gravity. (If you are interested in exploring percolation theory, there is an
excellent on-line article on the American Mathematical Society website [3]). Before the
percolation threshold, the film is not continuous (think copper islands on a glass sea),
but contains spaces that, depending on deposition conditions, may become voids as film
growth continues. Discontinuous films facilitate electronic conduction by electron tunnelling
between islands. Tunneling probability is inversely proportional to the distance between the
islands, increasing to unity at the percolation threshold. Above the percolation threshold,
the (Cu) film behaves as a conductor, i.e. conduction is facilitated by classical processes in
the conduction band of the solid.
During work at the School of Physics for his PhD, former student Tom Oates was able
to determine the percolation threshold of a titanium film with a high degree of accuracy
by using spectroscopic ellipsometry to detect the change in the films optical constants as
it changed from being a non-conductor to a conductor [4]. Ten years earlier, Maaroof and
Evans [5], had shown that is possible to determine with reasonable accuracy, (though not
quite perfectly, see Toms paper) the percolation threshold by measuring the resistance
of a metal film at regular intervals during its deposition. When the film thickness, t, is
plotted against the product of its resistance R and t2 , a minimum occurs at the percolation
threshold, i.e. at the minimum of the plot of:

t = Rt2 (2)

You will use Maaroof and Evans method to determine of the percolation threshold of your
Cu film.

5
5 Procedure
5.1 Starting the vacuum system
The diffusion pump takes about 15 minutes to warm up so it is a good idea to do this early.

Before you start, check with a demonstrator that the water is turned on. The water
provides ESSENTIAL cooling for the diffusion pump.

Switch on the main power switch on the coater unit

Switch on the rotary pump (pumps switch to rotary on)

Open the backing valve to the diffusion pump (diff pump backing valve switch to
open)

Switch on the diffusion pump (pumps switch to diff pump)

There should now be four red lights showing on the circuit diagram on the top panel of
the unit (rotary pump, pump valve, backing valve and diffusion pump) and also the power
on light at top left. If any other lights are observed ask a demonstrator to check the system.
After 15 minutes the diffusion pump should be warmed up. The vacuum system will then be
on stand-by for pumping the chamber. In the interim the glass substrates can be prepared.

5.2 Substrate preparation


Thin film growth and adhesion is fundamentally related to the substrate surface conditions.
It is therefore crucial to thoroughly clean the substrate before the deposition process. In
industry this is often done by plasma etching impurities from the surface. We will be a little
less thorough here, using alcohol and a tissue.

Clean two glass microscope slides thoroughly with alcohol and a kimwipe tissue and
allow to dry.

Now remove the stainless-steel lid of the vacuum chamber. This can be difficult and it
is important not to damage the glass cylinder. ASK A DEMONSTRATOR IF YOU
HAVE TROUBLE. This is done by holding the sides of the lid and pulling vertically
upward. When the lid is at least 5cms clear of the top of the glass cylinder, rotate the
lid around to the left and allow it to slide back down the support post.

The alligator clips and crystal monitor may still be inside the chamber from a previous
experiment. For reference, make a quick diagram of the arrangement in your logbook,
so you will know how to reinstall them in part 5.5. Reach inside the cylinder and
remove them both by loosening the blocks from the posts. The crystal monitor is
unplugged by gently pulling the plug out of its socket at the base of the chamber.
Place these parts aside until later.

Using some paper towel moistened with alcohol, clean the inside of the glass vacuum
chamber on the side closest to you and the rear side nearest the window, so you can
see what is going on inside the chamber.

Aluminium oxidises quickly and is chemically un-reactive, but it is still good practice
to avoid inhalation of metal dust.

Undo the evaporation holder earth lead.

6
Loosen the evaporation holder at the support post and slide it up and remove it from
the chamber.

Twist a 3 cm piece of aluminium wire around the tungsten wire. If the tungsten wire
is broken it will need to be replaced. The tungsten becomes thermally embrittled
and susceptible to breakage after several heating and cooling cycles. Take care not to
damage the ceramic insulator if you need to replace the tungsten wire.

Place the two clean glass slides side by side in the centre of the circular substrate
holder, avoiding touching the slide surfaces.

Place a third slide across the middle of the two clean slides. This will act as a shadow
mask for the other slides and leave a 2.5 cm square of uncoated glass with aluminium
contacts at either end.

Crossed glass slide arrangement

Return the evaporation unit to the chamber. You want to connect the evaporation
unit between the earth post and either post 2 or 3. If any of the posts is in the wrong
place then simply unscrew it and move it.

Now tighten the tungsten evaporation unit in place about 10 cm above the glass slides.

Cover the crystal monitor plugs with a small piece of aluminium foil to avoid a short
circuit occurring when aluminium deposition starts.

Replace the lid of the vacuum chamber.

5.3 Chamber evacuation


The chamber is now ready to be evacuated. The pumps should by now be warm enough to
operate.

Close the diffusion pump backing valve. This should not be left closed for more than
about ten minutes.

Check that the black air inlet valve at the bottom of the panel is screwed shut and
that the gas inlet switch beside it is off.

Open the chamber roughing valve and observe the pressure. The pressure should
begin to reduce after about 30 seconds. If not, check the lid is properly down, or ask
a demonstrator to help.

When the gauge shows a pressure of 1 101 mbar or less, close the roughing valve.

7
Open the backing valve, returning the vacuum system to stand-by.

Open the Chamber Isolation Valve. It is important that the pressure is below 1 101
mbar when the valve is opened.

When the pressure in the chamber drops to below 1 104 mbar the system is ready
to begin the evaporation process. You will need to change the detector on the pressure
gauge to see the pressure below 1 103 mbar.

5.4 Evaporative deposition of aluminium electrodes


Evaporation is achieved by resistively heating the tungsten electrode. If the current through
the tungsten is too high the tungsten will break and you will have to start again after
replacing it.

Check that the Source A Control knob is set to zero.

Switch Source A Select to Cont. on. The source A light should illuminate. If not
then you probably connected to the Source B post. In this case go to the Source
B Control knob.

Turn the control knob until the tungsten filament just begins to glow red after a few
seconds. If this process is not observed or slow at 30 A, the current can be increased
slightly however the control knob should not be turned above 50 A. If the filament has
been used many times it will become brittle and may break in which case it will need
to be replaced.

After a few seconds of the tungsten glowing you should observe the aluminium wire
begin to melt and evaporate. When the inside of the glass tube turns silver/black
so that the evaporation unit is no longer visible the current control knob should be
returned to zero.

Switch off the source (Source A select to off).

Close the chamber isolation valve, which returns the vacuum system to stand-by (four
red lights).

The chamber is now ready to be vented. It is important that the chamber is isolated from
the pumping system before air is admitted or the diffusion pump may be damaged.
Double check that the chamber isolation valve light is NOT on. The air inlet valve is
located on the bottom panel of the instrument. Slowly open the valve to admit air into
the chamber. The evaporation unit will be VERY HOT for a number of minutes after the
deposition process. Wait at least 10 minutes before you open the chamber lid, then remove
the evaporation unit, the glass tube and the glass slides.
You should now have two glass slides with aluminium contacts and an uncoated 25mm
square in the middle of each contact. The next stage will be to deposit a thin metal film
from a magnetron sputter coater whilst monitoring the conductivity of the film.

Obtain Tutors initials before proceeding:

8
5.5 Film deposition by magnetron sputtering and in-situ conductivity
measurement
Questions:

Explain why different materials exhibit different sputtering rates.

From equation (1), and assuming the same operating conditions, will Cu or Al have
the larger sputtering yield? Give reasons for your answer.

What do you expect to happen to the sputtering yield if the Cu target in the
magnetron becomes coated with Al during the evaporation process, and then exposed
to the atmosphere during the change-over from evaporation to sputtering? Explain
your reasoning.

What does the crystal monitor actually measure? Is there a constant relationship
between what the crystal monitor measures and film thickness? Give reasons for your
answer.

The calibration factor should be pre-set for copper deposition. The crystal should be the
same distance from the source as the substrate and as close as practical to it. An error in
the measurement arises at low film coverage due to a difference in the probability of atoms
sticking to different substrates (otherwise known as the sticking co-efficient).
You will now monitor the conductivity of a magnetron deposited ultra-thin copper film
in-situ. The evaporation unit should already be removed from the chamber and the vacuum
system should be on stand-by (remember to warm up the pumps if you are performing this
section on a different day to the evaporation section).

Plug the crystal monitor into the plug on the rear of the base inside the vacuum
chamber and attach the support arm to the Earth post. The crystal should be at least
10 cm from the top of the chamber and facing upright.

Switch on the crystal monitor control unit (Xtal) at the bottom of the coater unit.

Re-zero the film thickness.

Replace the alligator clips back into the chamber on the posts that connect to posts
1 and 4, which connect to the red and black wires that will be connected to the
mulitmeter.

Taking care not to touch the uncoated area of the slide, attach the two aluminised
contact pads of one of the glass slides to the two alligator clips. Make sure that the
coated side of the slide is facing upward and there is a good electrical contact with
the Al film. How would you check this?

The magnetron source is located slightly to the rear of the chamber lid. Position the
crystal monitor and the slide, side-by-side, at least 10 cm from the top of the chamber
and toward the rear of the vessel. If the crystal monitor is too close to the magnetron
source it will act as an electrode and initiate a glow discharge, which could damage
the crystal.

If the circular Cu target in the centre of the magnetron is silver coloured, rather
than the colour of copper, it has become coated with Al during evaporation. Remove
the Al coating using a ScotchBrite scrubbing pad and alcohol, and wipe clean with
a kimwipe, before attempting to use the magnetron.

9
Replace the lid of the chamber and check that the substrate and crystal monitor are
below the centre of the magnetron source and also equal distance away. The lid can
be rotated somewhat to achieve the optimal positioning.

Pump down the chamber as in section 5.3

When the pressure is below 1 104 mbar, argon can be introduced into the chamber
by switching the gas inlet switch on the bottom panel of the deposition unit. Make sure
the argon flow control tap and micrometer needle-valve on the pipe entering beneath
the left side of the chamber are are turned off before opening the gas inlet switch. Do
not turn the micrometer needle-valve off too hard or you may damage the needle in
the valve.

Adjust the argon flow until the pressure in the chamber is within in the range 1 6
102 mbar. This may take some practice as the pressure tends to overshoot.

Plug in the black and red lead to the multimeter inputs, and set the range to 20 M.

Switch on the sputtering unit power and check that the current knob is fully counter-
clockwise.

Turn the sputtering switch to continuous and slowly increase the sputter current
to 75 mA. You should observe a purple-green glow discharge initiated at the cathode
(target) surface. If the glow is completely purple without at least a green tinge, you are
not sputtering enough copper. Check that your pressure is within the correct range.
Did you check the copper target before pumping down? Note the toroidal shape of
the plasma that is concentrated in the region of high magnetic field. If a plasma is
not observed check that the auto pump-down switch is set to on. If there is still
no plasma observed ask a demonstrator to help.

Observing the film thickness on the crystal monitor, (and leaving the current set to
75 mA) stop the magnetron by turning the sputtering switch from continuous to
off when the film thickness reaches 1 nm. Any residual plasma in the chamber will
interfere with the multimeter resistance reading, so the measurements must be made
with the plasma turned off.

Obtain Tutors initials before proceeding:

Question: Explain why the plasma will interfere with the resistance measurement.

Take data of film resistance versus thickness for intervals of 0.2 nm to a resistance of
about 1 k.

Comment on any observations of changes in resistance with time as you take your
measurements.

Continue to take data in 0.5 nm intervals until the film thickness reaches 10 nm.

Then take data at 1.0 nm intervals until the resistance is 150 .

Estimate errors in your resistance values in terms of the short circuit resistance you
measured earlier.

10
Make a plot of film resistance vs thickness (R vs t).
Make a plot of Rt2 vs t (Maaroof and Evans), and estimate the percolation threshold
of Cu on glass.
By considering the film growth processes described in the Theory section and the way
in which the crystal monitor measures thickness, can you think of any uncertainties
that may be present in your estimate?
What properties of the surface and the film will affect the percolation threshold?
Multiply the resistance by the film thickness to obtain values of the film resistivity,
in m.
Plot vs t and estimate the bulk film resistivity for copper by determining where
the plot will flatten out. Compare it with the published value in the Handbook of
Chemistry and Physics. Explain any discrepancy. Is temperature relevant?
Explain why the resistivity reduces with film thickness to a minimum value.
Compare and contrast your resistivity results with those of Barnat, et al, [6].

Obtain Tutors initials before proceeding:

5.6 Shutdown
With the vacuum system in stand-by mode, switch off the diffusion pump.
Leave the backing valve open and the backing pump running for ten minutes while the
diffusion pump cools down, then switch off both of them. One of the demonstrators
will turn off the cooling water at the end of the lab.

6 References:
[1]. C. Kittel, Introduction to Solid State Physics, 6 - 9th eds., Wiley, N.Y. (1996 +).
[2]. H.-D. Liu, Y.-P. Zhao, G. Ramanath, S.P. Murarka, G.-C. Wang, Thickness dependent
electrical resistivity of ultrathin (<40 nm) Cu films, Thin Solid Films 384, 151156
(2001).
[3]. D. Austin, Percolation: Slipping through the cracks, American Mathematical Society,
http://www.ams.org/featurecolumn/archive/percolation.html (2008).
[4]. T. W. H. Oates, D. R. McKenzie, and M. M. M. Bilek, Percolation threshold in ultra-
thin titanium films determined by in situ spectroscopic ellipsometry, Physical Review
B 70, 195406 (2004).
[5]. A. I. Maaroof and B. L. Evans, Onset of electrical conduction in Pt and Ni films,
Journal of Applied Physics, 76, 2, 1047-54 (1994).
[6]. E. V. Barnat, D. Nagakura, P.-I. Wang, and T.-M. Lu, Real time resistivity measure-
ments during sputter deposition of ultrathin copper films, Journal of Applied Physics,
91, 9, 1667-72 (2002).

11