Authors: Dr. Joseph Brindley GENCOA and Dr. Anas Ghailane (AVALUXE)
Reactive Magnetron Sputtering
In reactive magnetron sputtering, sputtering of a target is conducted in the presence of a reactive gas (e.g., oxygen, nitrogen) that reacts with a sputtered material and forms a compound film on the substrate.
The reactive gas is typically used in smaller quantities compared to the inert gas, but the ratio can be varied to control the film properties.
Common applications are:
Flat-panel displays for televisions and cell phones
Photovoltaic coatings on solar cells
Optical coatings
Decorative coatings on hardware and automotive components
Solar insulating coatings on architectural glass
The most common reactively sputtered films are oxides and nitrides.
What is the Challenge?
The reaction between reactive gas and the sputtered material is known to cause process instabilities which means that the only stable operating area is in the presence of high levels of reactive gas (excess gas). This leads to reduced deposition rates as the very high partial pressure of the reactive gas results in compound formation on the whole of the metal target surface (target poisoning).
Most coatings used for wear resistance need to be close to stoichiometry to obtain a max. hardness. In order to obtain stoichiometry, the reactive process should be operated in the transition region of the hysteresis curve. However, the transition region is unstable as shown above in the graphic, whereupon decrease of the reactive gas, the output value (here partial pressure of the gas), does not return to its original value obtained during the increase of the flow, at the same time, but with a delay. Therefore two outputs are possible to one flow rate which translates to the instability of the process. To circumvent this instability, a fast feedback control device is needed.
Sputter yields from poisoned targets are typically 3x to 5x less than from a metallic target.
Solution: Reactive Gas Controller
The solution is to use a feedback control system that can very quickly adjust the reactive gas flow in response to the plasma conditions, in order to hold the process in high rate metallic or transition mode. A reactive gas controller as GENCOA’s Speedflo provides such an automatic feedback control and high-speed gas control to help prevent a runaway situation leading to target poisoning. With such a device it is possible to control a reactive process in what would usually be an unstable region.
The graph above shows an example of a process control at a given set point of a relevant sensor such as the target voltage. The actuator maintains the process in control through the mass flow controller.
Gas control during reactive sputtering strongly influences the deposition rate and film properties of the compound being deposited. Reactive gases can trap the target in poisoned mode unless the partial pressures of the reactive gas(es) are individually monitored and controlled at high speed. The dynamics of a reactive sputtering system typically requires a closed-loop feedback speed of control in the 10’s of msec range. Active feedback control reacts and adjusts the reactive gas flow control valves within 1 msec. Depending upon gas line lengths and system size, the gas will then take typically between 10-100 msecs to enter the area in-front of the sputter target. With a closed loop feedback control time of <100msec most reactive processes can be maintained at high rate and with good control in the ‘transition’ region.
We see how the Al metal signal from the PEM follows its set point (40%) that has bee specified by the user, keeping the process in control at 40% set point.
Operation of the process in the ‘transition’ region between the elemental and poisoned states of the target ensures that the metal target material is sputtered at high rate, and there is ‘just’ enough reactive gas present to ensure that the correct stoichiometry of compound layer is formed on the substrate.
Operating in this region is the ideal situation to achieve higher deposition rates and stable film properties.
Sensoring
An important part of the control system is the choice of ‘sensor’ to provide feedback from the process of the effect of the reactive gas changes. The sensor signal is the ‘input’ to the controller and a fast and stable input signal makes achieving good control more straightforward.
Target Voltage
For some metal and gas combinations, the sputter target operating voltage is a suitable input signal (works with Al and Si target materials with O2 and N2 gas). Using the target voltage simplifies the hardware and hence reduces the cost of the overall control solution.
Characteristics of target voltage sensoring:
• Often easily available
• Only works for some materials (e.g. Cu, Si, Al)
• Not possible to use for uniformity control
The target voltage sensor follows the set point 20%. The other process outputs such as metal signal (PEM), oxygen pressure (lambda) get some values as a result of the target voltage set point being at 20%.
PEM + filter
Other material combinations require a gas or metal ‘plasma emission’ signal from the process area or from a remote plasma sensing head (gas-only signal). The plasma emission is universal, in that any material or gas can be monitored and can be used to provide local sensing and gas control for very large deposition chambers (multi-channel sensing to tune uniformity over large areas). The plasma emission signal is ‘carried’ to the controller via a fiber-optic link, hence the signal is very fast – speed of light.
However, the light intensity needs to be converted into a digital voltage signal for input into the control architecture. This fast conversion is typically via a narrow bandpass optical filter and a photomultiplier tube (PMT). The PMT detector method maintains the sub 1 msec signal processing which is critical for good feedback control.
Characteristics:
Fast response time
Large area uniformity control possible
Easily disturbed by moving substrates or plasma
Spectrometer
An alternative method to convert the light signal is via a CCD-type spectrometer. The spectrometer can provide multiple gas input signals, but the integration time is typically >50 msec, so the closed-loop control speeds are much longer. This slower response speed and CCD array drift mean spectrometers are typically only used for process development rather than industrial process control.
Characteristics:
Great for full spectrum data – maximum flexibility
Limited by speed – signal intensity is a function of integration time
Lambda-probe
Many sputtering processes just use oxygen as the reactive gas, and hence another class of sensor comes into play – Lambda oxygen ‘sniffing’. Lambda sensors are used for automotive engine management to control the efficiency of the combustion. The lambda device converts the amount of group 16 gas present (in this case O2) into a proportional voltage output. With good control of the sensor temperature and adaption for use in a vacuum environment (see VacGasG16 type), this can be a convenient method to create a gas signal to use for a successful reactive gas feedback control.
Summary: Benefits of Reactive Gas Control
Much improved deposition rates are achieved (factor 2.5 – 4, dependant on process*), for lower cost production
High thin film stoichiometry and properties control – eliminates process drift
High uniformity of coating – important in serial coating
More precisely controlled film uniformity, 1.5% over large areas (ideal uniformity for double low E glazing)
More energy efficient processes – the energy consumption can be reduced by up to 70%
* Material properties of the coated surface determine the maximum rate achievable, e.g., for architectural glass, too fast a deposition rate leads to a more metallic coating which can lead to future corrosion problems.
Comments