NOTE: Publication of the Flat Panel Focus has been temporarily suspended. Please visit our Reference Library for additional information or topics of interest.

This edition of Flat Panel Focus® offers relief for your film stress—solutions for your bent or broken substrates and peeling films. Ken Nauman also answers reader questions about arc-management setup, banding, and output cable choice.
Film Stress Relief
Excessive film stress can cause problems ranging from cracking (crazing), wrinkling, curling, and peeling films to bent or shattered substrates. Even when the results are not this extreme, film stress can dramatically reduce device lifetime. This is particularly a problem for the large (3 m x 3 m) and thin (< 1 mm) substrates that are typical of FPD manufacturing. Vertical deposition techniques may exacerbate stress-related problems, and film stress is a critical issue to address for flexible substrates. Although the potential problems are many, the good news is that it’s possible to control stress and achieve the film properties that you require.
There are two primary types of stress where thin films are concerned: thermal and intrinsic.
Thermal Stress
Thermal stress is the result of the difference, during cooling, in contraction rates of the deposited film and the substrate (coefficient of thermal expansion or TCE). If the film contracts more than the substrate, the substrate will likely bend in a concave formation as its larger surface area is drawn in toward the film’s smaller surface area. On the other hand, if the substrate contracts more than the film, the substrate will likely bend in a convex formation. This is most often a problem for processes in which the device cools after a film has been deposited at elevated heat.
Unfortunately, there is nothing you can do about the inherent reaction of a material to heating and cooling. However, a few possible solutions exist. One is to switch to a different deposition method. Evaporation, for example, can be performed at lower substrate temperatures and thus produces thermal-stress-free films. However, it’s associated with possible adhesion (Figure 1) and film density problems. Another solution is to switch to substrate and film materials with similar coefficients of thermal expansion. Also consider the use of RF-superimposed DC, which is a newer technique that can be performed at lower temperatures. It produces results that are comparable to high-temperature processes.
Figure 1. Adhesion test showing proper adhesion (left) and poor adhesion (right)—The film on the right was deposited by evaporation, which often is associated with adhesion problems
Intrinsic Stress
Every material, in its native form, has an inherent molecular structure that gives it certain tendencies in terms of film stress. However, the material’s inherent properties can be influenced by controlling ion energy in order to “tune” this stress. An ion with low kinetic energy impacts the substrate with a weak force, producing low packing density. In effect, the deposited atoms are too few and far apart, causing forces of attraction across the empty spaces between them. This creates tensile stress, in which the film seems to shrink up. If the film is deposited directly on the substrate, this likely produces a concave curvature (Figure 2). If it is deposited on top another film, the likely result is poor film adhesion.
On the other end of the spectrum, ions with high kinetic energy come down very forcefully into the deposited film, producing high packing density. If ion energy is too great, the deposited atoms may be packed too tight, causing them to exert a force of repulsion against each other. This results in compressive stress, which can be strong enough to bend the substrate backwards (Figure 2). Like tensile stress, compressive stress can also result in adhesion problems for films that are sputtered on top of another deposited layer.
Figure 2. Tensile and compressive stress—Insufficient ion energy can produce tensile stress, in which the film seems to shrink. This can result in concave curvature of the substrate (left). Excessive ion energy can produce compressive stress, in which the film seems to expand, possibly resulting in convex curvature of the substrate (right)
Ion Energy Management (IEM)
There are many factors affecting ion energy and thus film stress, including pressure, source-to-substrate distance, and process power choice.
Pressure
The term mean free path refers to the average distance a particle can travel without collision. At higher pressures, more particles are present per cubic centimeter than at lower pressures; mean free path is shorter (Figure 3) and more collisions occur. Collisions reduce the kinetic energy with which atoms impact the substrate. The result is lower packing density and greater tensile film stress.
On the other hand, if pressure is too low and mean free path is excessive, the lack of collisions may result in over-energized atoms that impact the substrate with too much force. This may produce excessive packing density and compressive stress.
Figure 3. Fewer particles are present at lower pressures; mean free path and ion energy are greater. At higher pressures, more particles are present; mean free path and ion energy are lower
An equilibrium pressure must be achieved to produce the ideal film stress properties for any given process. Because film stress is so highly influenced by pressure, a good rule of thumb is to adjust pressure only to control film stress and not to manipulate sputter rate, voltage, or any other process condition.
Because film stress is so highly influenced by pressure, a good rule of thumb is to adjust pressure only to control film stress and not to manipulate sputter rate, voltage, or any other process condition.
|
Figure 4 shows the general relationship between pressure and tensile/compressive film stress. The process of determining the ideal pressure for your process involves trial and error. For example, if you’re experiencing tensile stress in your films, try turning down the pressure about 0.5 mTorr to increase ion energy. If the tensile stress remains, turn pressure down another 0.5 mTorr. In cases of compressive stress, try increasing pressure. In either case, experiment until film stress properties reach an acceptable level. Keep in mind that if the plasma extinguishes, the pressure has been turned down too low. If the plasma becomes unstable and arcing increases, the pressure has been turned up too high.
Figure 4. General relationship between pressure and film stress
Pressure Uniformity
It’s also important to have uniform pressure throughout your chamber. If different pressure zones exist, film stress will vary across the substrate. This also makes it very difficult to adjust pressure in order to achieve specific film stress properties. Figure 5 shows a chamber with extreme pressure gradients that might result in varying film stress properties across the substrate.
Figure 5. Pressure gradients resulting from system design issues can produce varying film stress properties
To improve pressure uniformity, it may be necessary to add gas manifolds or modify existing ones. Figure 6 shows a design that produces excellent pressure uniformity and thus enables better control over film stress. It allows each MFC to be adjusted independently in order to compensate for chamber design and cathode design issues. Figure 7 shows modifications that can be made to the manifold itself.
Figure 6. Schematic of a segmented gas manifold

Figure 7. The size of and space between a gas manifold’s holes can be adjusted in order to improve uniformity
Source-to-Substrate Distance
Source-to-substrate distance influences two crucial phenomena: plasma ignition and film stress. An airplane must have a certain length of runway in order to achieve the speed necessary for takeoff. Similarly, an electron must have sufficient space (mean free path) to accelerate across in order to create a collision powerful enough remove an electron from an atom. When enough ions are created, a cascade effect ignites the plasma and deposition can begin.
Similarly, atoms of target material must have a specific amount of kinetic energy so that they impact the substrate neither too forcefully (results in compressive stress) nor too feebly (results in tensile stress). Source-to-substrate distance defines the length of the “runway” for atoms of target material. A shorter runway produces lower kinetic energy, while a longer runway results in higher kinetic energy.
For every process, there is a source-to-substrate distance that produces the ideal amount of kinetic energy. In general, this usually falls somewhere between 5 to 15 cm (2 to 6″). Determining the best distance for your process is usually a process of trial and error. Start at 10 cm (4″) and increase or decrease the distance depending on the results and your desired packing density/film stress properties.
Process Power Selection
Certain types of process power produce more highly energized ions than other types. For example, pulsed DC and AC produce higher ion energy and packing density than straight DC. Figures 8 and 9 show how the film density produced by straight DC compares to that produced by pulsed-DC power. Please note that DC is an excellent option for many processes, but if you need to increase packing density, consider adding pulsing or switching power types.
Figure 8. Film quality produced by straight-DC power—Lower ion energy produces lower packing density and poor film uniformity
Source: Centre for Advanced Materials and Surface Engineering, University of Salford, U.K.
Figure 9. Film quality produced by pulsed-DC power—Higher ion energy produces greater packing density and good film uniformity
Source: Centre for Advanced Materials and Surface Engineering, University of Salford, U.K.
If you are experiencing film stress problems, first try adjusting process pressure. If that doesn’t solve the problem, try modifying source-to-substrate distance. If you continue to experience film stress issues, consider switching process power types. Please also note that if you are using pulsed DC, you can adjust pulsing frequency and duty cycle to control ion energy and thus film stress.
If you are using pulsed DC, you can adjust pulsing frequency and duty cycle to control ion energy and thus film stress.
|
Table 1 below provides general guidelines for process power selection, including ratings for packing density.
Table 1. Process power selection table
Of course, there are other factors to consider besides film stress when selecting a power supply. Please also consult the following resources for further advice on process power selection, and feel free to contact us for further assistance:
Other Film Stress Solutions
Cooling and Venting in Inert Gas
A relatively simple solution to reduce stress is to let devices cool as much as possible before venting the chamber. This prevents the newly deposited films from being traumatized by a sudden change in temperature and pressure. Such trauma seems to increase film stress. For the same reason, it also may be beneficial to vent in an inert gas such as argon. Nitrogen, a more affordable alternative, may also be used. This is very important for certain processes, such as the deposition of very thin TCOs, in order to prevent over-oxidation.
Let devices cool as much as possible before venting the chamber. This prevents the newly deposited films from being traumatized by a sudden change in temperature and pressure. Such trauma seems to increase film stress.
|
Post-Deposition Annealing
Intrinsic stress problems can sometimes be solved by annealing—heating up the film to relieve some of the material properties that cause this stress. This is done after the deposition process is complete to release some of the energy stored in the film. Of course, this is not an option for polymer substrates and color filters, which are sensitive to high temperatures. Also, thermal annealing and other high-temperature manufacturing processes can cause diffusion of underlying layers. Some manufacturers use laser annealing to overcome this problem, because this technique heats only the topmost layer.
Figure 10. Effect of temperature on film stress properties—Thermal annealing alleviates both tensile and compressive stress
Periodically Adjusting Process Conditions/Film Characteristics
If process conditions remain exactly the same throughout the deposition of an entire layer, especially when that layer is particularly thick, undesirable film properties, such as film stress, may become more pronounced. Certain periodic process adjustments can serve to “re-boot” film properties. For example, slightly changing pressure approximately every 100 Å gives a fresh start that prevents either tensile or compressive stress from dominating. Even if no extreme film stress is evident, periodic modifications of film properties may prevent problems from developing.
Interleaving is a more extreme process that is based on the same principle. Layers of film in tensile stress are alternated with layers of film in compressive stress. The net stress of the entire stack, then, is neutral. The main drawback is that this only works to a certain film thickness before adhesion problems develop.
If you have further questions about film stress or need hands-on help solving film stress in your process, please contact us.
Ask the FPD Expert!
Are you struggling to squeeze more profit out of your FPD process?
Ken Nauman, AE’s FPD applications expert, answers some of your difficult questions. Submit your question or comment to Ken at +1.970.214.6280 or
Ken.Nauman@aei.com
- How do I determine the best arc-management settings for my process?
- I’m having problems with banding across my film. What could be causing this and how can I fix it?
- What type of output cables should I use for my power supply?
- How do determine the best arc-management settings for my process?
Answer: Understanding and correctly setting up your power supply’s arc management enables you to prevent a number of process and film quality problems. This can also help you troubleshoot if arcing-related process or film problems arise. If you’re asking this question, you probably already know that the potential effects of excessive arcing can range from occasional process interruptions to devastating film damage and complete process shutdown. Therefore, it’s important to take full advantage of your power supply’s arc-management capabilities.
The table below defines key arc-management parameters. Understanding the effects of increasing or decreasing these parameters helps you make the best decision for your particular situation. High-quality power supplies have arc-management software that is easy to adjust. Once you understand the parameters and determine the best settings for your specific process, the procedure for implementing them should be easy and intuitive. Please contact us if you need further assistance determining and implementing the ideal arc-management settings for your power supply.
Table 2. Key arc-management parameters
Parameter | Definition | Effect of Increasing | Effect of Decreasing |
Detect Time
|
The length of time the power supply puts output energy into the arc
|
Cleans surface
|
Reduces particles
|
Shutdown Time
|
The length of time the power supply remains off before re-igniting the plasma |
Cools arcing area
Allows plasma to extinguish
|
Increases rate
Lowers re-ignition energy
|
Voltage Level
|
The voltage at which the power supply determines that there is an arc event
|
Increases sensitivity to arcs
|
Decreases sensitivity to arcs
|
- I’m having problems with banding across my film. What could be causing this and how can I fix it?
Answer: Banding is a type of film defect that typically extends across the substrate from top to bottom. It can be an area where deposition is too thick, too thin, or simply has the wrong film properties. There are a number of possible causes for this problem, including the following:
Drive System Malfunction
In a dynamic system, banding may result if the glass is not moving through the process smoothly. Any sudden stops, starts, or changes in speed can cause the deposition of bands of overly thick or thin film. Check the drive system and make repairs to ensure that the substrate is moving through the process at a uniform speed.
Chamber Leaks and Outgassing
The accidental addition of oxygen to the process through a chamber leak or outgassing can dramatically reduce sputter rate and result in banding. Perform a leak check on your chamber. To reduce outgassing, select vacuum-compatible materials and condition system surfaces by baking out heat-resistant materials and wiping down surfaces with anhydrous alcohol. In some cases, the “leak” may actually be an inadvertent result of process design. For example, if a chamber door opens at the end of a certain deposition phase, changes in pressure or stoichiometry may occur for a short period. Adjust your process to prevent accidental introduction of oxygen. This may involve timing the process transition precisely so that deposition is completely done before the chamber door is opened. Balancing the gas pressure and stoichiometry of adjacent chambers also may solve this problem. In addition, it’s critical to completely base out your system before starting to back fill. Pumping to high vacuum before process startup ensures that no water vapor or other substances remain that may contaminate subsequent processes.
Cathode Failure
If you are using a static deposition process, one of your cathodes may have stopped sputtering. This prevents deposition on an entire region of the substrate because each cathode is “responsible” for a given zone (See Figure 11 below).

Figure 11. Cathode failure can cause extreme banding in a static deposition process
The loss of one cathode is not as much of a problem for a dynamic process as it is for a static process. In a dynamic process, turning up the power on the remaining cathodes can compensate, with no negative effect on film quality. (Please see the Q2 2007 edition of Flat Panel Focus for a detailed discussion of this phenomenon.) However, dynamic processes are still susceptible to banding due to the particular methods of this manufacturing technique. Read below for more information.
Arcing-Related Phenomena
A variety of phenomena related to arcing can cause banding. In a dynamic process, power levels to one or more cathodes may fluctuate when excessive arcing triggers frequent power supply shutdown for arc handling. If power is shut off too often, delivered power may drop below set point. Fluctuating power levels cause corresponding fluctuations in sputter rate. As the substrate moves in front of one or more affected cathodes, bands of thinner deposition result that correspond with these dips in delivered power. Please click on the graphic below to watch an animation of this phenomenon.

Figure 12. Click the graphic above to view an animation of arc-related banding in a dynamic deposition process
A number of conditions can cause arcing at rates high enough to cause this phenomenon. Excessive arcing may simply be the result of the inherent qualities of the target material. For example, AZO is known to arc hundreds or even thousands of times per second. Furthermore, in addition to the problems described in Chamber Leaks and Outgassing above, chamber leaks, outgassing, and the failure to base out your system can introduce contaminants that cause high rates of arcing.
However, arc-related phenomena may be the cause of banding even when your arc rate isn’t high enough to cause excessive shutdown times. For example, improper arc-management setup may be at fault. If your shutdown time (see Table 2 above) is too long, power may be interrupted and deposition therefore suspended long enough to produce banding. Likewise, if shutdown time is too brief, the power supply may turn back on before the arc is extinguished, which immediately initiates another power supply shutdown. Cumulatively, power may be suspended long enough to cause a great reduction in sputter rate. Another arc-related phenomenon that may cause banding is film buildup with poor adhesion that condenses on the chamber walls, fixtures inside the chamber, or edges of the target. This buildup may flake. The flakes may short out a cathode, which the power supply will likely interpret as a constant arc and shut down power for a lengthy period. This may cause banding.
To solve any of these arc-related phenomena, the following measures are recommended:
- Choose a power supply with effective, easy-to-set arc-management capabilities. This is especially important for processes using materials that are arc-prone, such as AZO.
- Select vacuum-compatible materials to reduce outgassing. Also condition system surfaces by baking out heat-resistant materials and wiping down surfaces with anhydrous alcohol.
- Fully base out your system before backfilling to prevent contamination.
- Clean your chamber to prevent flaking from initiating inappropriate arc responses from your power supply. Implement regular cleaning to prevent excessive buildup in the first place.
- Check for and repair leaks.
- Perform a few trials to determine the optimal shutdown time setting so that a given arc is fully extinguished, but power is not suspended so long that film quality is negatively affected. Your power supply should offer effective arc-management software that enables you to easily adjust these parameters.
- Consider adding pulsing if you are using DC power and arcing is a particular problem. Pulsing periodically discharges dielectric surfaces that otherwise would likely build up a charge and eventually arc. This may be especially useful for arc-prone target materials.
- What type of output cables should I use for my power supply?
Answer: I always recommend shielded cable. Make sure it has the lowest inductance per meter possible, and ground the shield at both ends. For power usage of the magnitude used in industrial sputtering applications, grounding at one end is not sufficient and may be dangerous. The use of a cable that is well shielded and well grounded helps prevent noise, even for DC power supplies, where arcing can lead to the same noise issues commonly experienced with AC, RF, and pulsed DC. Some materials, such as AZO, may have hundreds, or even thousands of arcs per second, making the use of shielded cable absolutely vital.
DC power supplies may experience noise issues due to arcing.
|

Figure 13. Properly grounded power lines between a low-frequency power supply and cathodes in a coater; the power supply output cable is surrounded by a ground shield that is itself grounded at both ends
For more information about proper grounding practices, please see the Q1 2008 edition of AE’s Sputter Spotlight e-newsletter.