SPUTTERINPGR OCESSES

Samer Elatrache V-17810600 estudiante de CRF

For convenience we divide sputtering processes into four cateogories: (1) dc,
(2) RF, (3) magnetron, (4) reactive. We recognize, however, that there are
important variants within each category (e.g., dc bias) and even hybrids
between categories (e.g., reactive RF). Targets of virtually all important
materials are commercially available for use in these sputtering processes. A
selected number of target compositions representing the important classes
of solids, together with typical sputtering applications for each are listed in
Table 3-6.
In general, the metal and alloy targets are fabricated by melting either in
vacuum or under protective atmospheres, followed by thermomechanical processing.
Refractory alloy targets (e.g., Ti-W) are hot-pressed via the powder
metallurgy route. Similarly, nonmetallic targets are generally prepared by
hot-pressing of powders. The elemental and metal targets tend to have purities
of 99.99% or better, whereas those of the nonmetals are generally less pure,
with a typical upper purity limit of 99.9%. In addition, less than theoretical
densities are achieved during powder processing. These metallurgical realities
are sometimes reflected in emission of particulates, release of trapped gases,
nonuniform target erosion, and deposited films of inferior quality. Targets are
available in a variety of shapes (e.g., disks, toroids, plates, etc.) and sizes.
Prior to use, they must be bonded to a cooled backing plate to avoid thermal
cracking. Metal-filled epoxy cements of high thermal conductivity are employed
for this purpose.

DC Sputtering
Virtually everything mentioned in the chapter so far has dealt with dc sputtering,
also known as diode or cathodic sputtering. There is no need to further
discuss the system configuration (Fig. 3-13), the discharge environment (Section
3.5), the ion-surface interactions (Section 3.6. l), or intrinsic sputter
yields (Section 3.6.2). It is worthwhile, however, to note how the relative film
deposition rate depends on the sputtering pressure and current variables. At
low pressures, the cathode sheath is wide and ions are produced far from the
target; their chances of being lost to the walls are great. The mean-free
electron path between collisions is large, and electrons collected by the anode
are not replenished by ion-impact-induced cathode secondary emission. Therefore,
ionization efficiencies are low, and self-sustained discharges cannot be
maintained below about 10 mtorr. As the pressure is increased at a fixed
voltage, the electron mean-free path is decreased, more ions are generated, and
larger currents flow. But if the pressure is too high, the sputtered atoms
undergo increased collisional scattering and are not efficiently deposited. The
trade-offs in these opposing trends are shown in Fig. 3-18, and optimum
operating conditions are shaded in. In general, the deposition rate is proportional
to the power consumed, or to the square of the current density, and
inversely dependent on the electrode spacing.

RF Sputtering
RF sputtering was invented as a means of depositing insulating thin films.
Suppose we wish to produce thin SiO, films and attempt to use a quartz disk
0.1 cm thick as the target in a conventional dc sputtering system. For quartz
p = 10l6 Q-cm. To draw a current density J of 1 mA/cm2, the cathode needs
a voltage V = 0.1 p J. Substitution gives an impossibly high value of 10l2 V,
which indicates why dc sputtering will not work. If we set a convenient level
of I/ = 100 V, it means that a target with a resistivity exceeding lo6 Q-cm
could not be dc-sputtered.
Now consider what happens when an ac signal is applied to the electrodes.
Below about 50 kHz, ions are sufficiently mobile to establish a complete
discharge at each electrode on each half-cycle. Direct current sputtering
conditions essentially prevail at both electrodes, which alternately behave as
cathodes and anodes. Above 50 kHz two important effects occur. Electrons
oscillating in the glow region acquire enough energy to cause ionizing collisions,
reducing the need for secondary electrons to sustain the discharge.
Secondly, RF voltages can be coupled through any kind of impedance so that
the electrodes need not be conductors. This makes it possible to sputter any
material irrespective of its resistivity. Typical RF frequencies employed range
from 5 to 30 MHz. However, 13.56 MHz has been reserved for plasma
processing by the Federal Communications Commission and is widely used.
RF sputtering essentially works because the target self-biases to a negative
potential. Once this happens, it behaves like a dc target where positive ion
bombardment sputters away atoms for subsequent deposition. Negative target
bias is a consequence of the fact that electrons are considerably more mobile
than ions and have little difficulty in following the periodic change in the
electric field. In Fig. 3-13b we depict an RF sputtering system schematically,where the target is capacitively coupled to the RF generator. The disparity in
electron and ion mobilities means that isolated positively charged electrodes
draw more electron current than comparably isolated negatively charged
electrodes draw positive ion current. For this reason the discharge currentvoltage
characteristics are asymmetric and resemble those of a leaky rectifier
or diode. This is indicated in Fig. 3-19, and even though it applies to a dc
discharge, it helps to explain the concept of self-bias at RF electrodes.
As the pulsating RF signal is applied to the target, a large initial electron
current is drawn during the positive half of the cycle. However, only a small
ion current flows during the second half of the cycle. This would enable a net
current averaged over a complete cycle to be different from zero; but this
cannot happen because no charge can be transferred through the capacitor.
Therefore, the operating point on the characteristic shifts to a negative voltage
-the target bias-and no net current flows.
The astute reader will realize that since ac electricity is involved, both
electrodes should sputter. This presents a potential problem because the
resultant film may be contaminated as a consequence. For sputtering from only
one electrode, the sputter target must be an insulator and be capacitively
coupled to the RF generator. The equivalent circuit of the sputtering system
can be thought of as two series capacitors-one at the target sheath region, the
other at the substrate-with the applied voltage divided between them. Since
capacitive reactance is inversely proportional to the capacitance or area, more
voltage will be dropped across the capacitor of a smaller surface area.
Therefore, for efficient sputtering the area of the target electrode should be
small compared with the total area of the other, or directly coupled, electrode.
In practice, this electrode consists of the substrate stage and system ground,
but it also includes baseplates, chamber walls, etc. It has been shown that the
ratio of the voltage across the sheath at the (small) capacitively coupled
electrode (V,) to that across the (large) directly coupled electrode V.