NOTE: Publication of the Sputter Spotlight has been temporarily suspended. Please visit our Reference Library for additional information or topics of interest.
The Q1 2009 edition of Sputter Spotlight® presents part two of our discussion of crazing’s causes and cures. Please refer to the Q3 2008 edition for the first part of this article, which describes the main phenomena that trigger crazing, and discusses the importance of chamber cleanliness in remedying and preventing this type of film damage. The following list completes the sampling of “crazing cases” that we began in our Q3 2008 newsletter.
Cracking the Crazing Code: Part Two
Chamber Design and Cathode Spacing (Multi-Cathode Processes)
Symptoms:
- Damage distributed across the substrate, with higher concentration near affected cathodes
- Characteristic “crazed” visual appearance
- Immediate onset of crazing when system is put into operation
In the majority of glass and web coating processes, cathodes are spaced well apart from one another and placed in separate compartments. These two factors isolate them from one another enough to deter electrical crosstalk. However, other applications, including FPD and solar, do not typically utilize cathode compartments and often place cathodes very close together. These conditions make cathode crosstalk likely. FPD often employs large substrates, which may require a dozen or more cathodes per chamber. In order for the cathodes to fit into a single chamber, the space between them must shrink. When cathodes are at different potentials and are spaced closely together with no shielding between them, they are apt to electrically couple and interfere with one another. This can lead to the unwanted electrical discharges at the substrate that cause crazing.
RF presents an additional challenge with regard to crosstalk because it is simply more prone to this phenomenon than lower-frequency power. Figure 1 shows an inline RF coating system for FPD. The two cathodes and their associated plasmas are at different potentials, causing unwanted plasma interactions. These interactions lead to substrate electrical discharges that result in crazing.
Figure 1. Inline RF coating system in which unwanted plasma interactions cause substrate electrical discharges that result in crazing. The solution is to synchronize power supply output with your power supply’s CEX feature in order to increase plasma localization.
The Cesar® RF power supply’s CEX (common exciter or phase synch) feature synchronizes the output of power supplies that are connected to the same chamber, matching the output waveforms so that the cathodes and their associated plasmas are all phased properly. This increases plasma containment by discouraging electrical interactions between adjacent cathodes and plasmas. In certain complex situations, processes benefit from the use of a phase shifter in addition to CEX. Please see the Q2 2008 PV Sun Times® e-newsletter for more information about this technique.
CEX controls the power supplies’ output waveforms so that their cathodes and associated plasmas are all phased properly, which increases plasma containment by discouraging electrical interactions between adjacent cathodes and plasmas.
|
A crosstalk phenomenon similar to that described above can occur in processes using multiple AC power supplies. Please see our Enhanced Plasma Containment for Inline Sputtering Systems application note for instructions on using the PEII low-frequency power supply’s unique CEX capability to create an alternating cathode arrangement that significantly deters crazing and improves process control, overall film quality, and uptime.
Please note that the crazing-deterrent benefits of CEX aren’t limited to processes using RF or fixed-frequency AC power supplies. The Pinnacle® Plus+ pulsed-DC power supply, as well as the Pulsar® and Sparc-le® V DC pulsing accessories, feature CEX to synchronize the output of multiple pulsed-DC units.
For additional information about CEX, please see the following resources and contact us for information on how it can benefit your particular process.
Inline System Design (Low-Frequency Oxide Process After DC Metal Process)
Symptoms:
- Damage at the leading and/or trailing edges of the substrate
- Characteristic “crazed” visual appearance
- Immediate onset of crazing when system is put into operation
A specific phenomenon can cause crazing at the leading and/or trailing edges of a substrate in inline systems depositing alternating metal and oxide layers, such as for architectural and residential glass coating. As the substrate moves through the system, the deposited metal layer can conduct electricity, causing the DC cathode that has deposited the metal layer to couple with the AC cathode down the line that is depositing the oxide film. The result is “plasma plumes” and unwanted electrical discharges at the substrate. This ultimately causes crazing at the leading and trailing edges of the glass where the plasma plumes and unwanted electrical discharges occur (Figure 2).
Figure 2. Coupling of the DC metal and AC oxide cathodes across a metalized glass in an inline coating system causes plasma plumes, substrate electrical discharges, and crazing at the leading and trailing edges of the substrate. (Click graphic to view animation.)
Crazing specifically occurs at the substrate’s leading and trailing edges for a particular reason. Coupling is strongest after the substrate has been coated with the metalized film when its leading edge is under an AC cathode while at the same time its trailing edge is under a DC cathode. This position allows the metal film to serve as a bridge to electrically connect these two cathodes and promote crosstalk, as the charges are greatest at the edges of the substrate.
Figure 3. Cathode coupling is strongest when the substrate is in the position shown here. It has been coated with the metal film and its leading edge is under an AC cathode while at the same time its trailing edge is under a DC cathode. As the substrate approaches this particular position, the likelihood of crosstalk increases, and as it moves out of this position, the possibility of crosstalk ceases. (Click graphic to enlarge.)
Unlike the phenomenon shown in Figure 3, CEX is not a possible solution here because it obviously can’t synchronize the output of a DC power supply with that of an AC power supply. However, AE has innovated a solution that effectively prevents the AC and DC cathodes from coupling and thus deters crazing. Please contact us for more information about implementing this unique solution in your process.
Chamber Fixture Design
Symptoms:
- Damage at edges of the substrate adjacent to the fixture
- Very dark, burned areas on film, sometimes accompanied by etching of actual substrate
- Immediate onset of crazing when system is put into operation
Electrical discharges between the substrate and chamber components, such as a plasma shield or metal carrier, are another factor that can instigate crazing. In a dynamic process, the physical shape and material makeup of the carrier that moves the substrate through the system can be a critical issue. Metal carriers with sharply pointing ends can be the source of unwanted discharge that causes crazing at the edges of the substrate near the carrier (Figure 4). These sharp points create correspondingly strong electrical fields that can lead to extremely powerful discharges. The result may be very dark, burned areas of crazing on the film, as if it had been struck by lightning. Etching of the glass or the carrier can even occur because of the intensities involved.
Figure 4. Side view of a metal carrier holding a substrate; Pointed ends on metal carriers create powerful electric fields that can result in electrical discharges between the carrier and substrate. This leads to crazing and even etching of the glass substrate itself. As the film material thickens on top of the substrate, it closes the gap between the carrier and the substrate, further increasing the likelihood of electrical discharge.
One possible solution is to replace the metallic point with ceramic, Teflon®, or another material that is less conductive than metal. This decreases the potential strength of the electric field and reduces the chance of unwanted electrical discharge between the carrier and substrate. To the same end, it’s also important to change the physical shape of the carrier by machining down the point so that it is smooth and rounded. In addition, regular cleaning of the chamber and fixture surfaces prevents buildup of metallic material on the carriers, which would return the newly rounded, electrically insulative ends of the carrier to their former highly conductive state.
Figure 5. Side view of a carrier holding a substrate; The points on the carrier have been rounded to deter the electrical discharges that cause crazing.
Another option for averting carrier-to-substrate discharge is to construct the carrier of multiple parts instead of one continuous piece. These parts must be electrically isolated from one another. By reducing total continuous surface area, this construction reduces the carrier’s total possible electrical charge. Therefore, the possibility and strength of any electrical discharge from the carrier is greatly reduced.
In glass coating processes, the substrate itself may have sharp cut edges that produce strong electrical fields that are similar to those created by the pointed metal ends on a carrier. Like pointed metal carriers, these sharp glass edges result in electrical discharges at the substrate and crazing. In this case, the solution is to chamfer or otherwise modify the ends of the glass so that they are more rounded and thus produce a weaker electrical field, which is less prone to discharge.
In cases of substrate-to-chamber-component crazing that involve objects besides the carrier or glass substrate, it may be useful to determine if any sharp points on the relevant components can be rounded or otherwise altered to prevent the phenomena described above. Please contact us if you are experiencing a similar problem and need help devising solutions.
For more information about key issues for dynamic manufacturing processes as compared to static methods, please see our Q2 2007 edition of Flat Panel Focus.
Conclusion
This article hopefully has cut the topic of crazing down to size and provided a few general guidelines that will help you approach the problem if you ever experience it in your manufacturing operation. Please also contact us if you need further information or assistance with finding the cause and cure for crazing in your process.
Ask Doug!
Doug Pelleymounter is a senior application engineer at Advanced Energy and has more than 33 years of hands-on experience working with all kinds of challenging thin-film applications. He is a major contributor to AE's PV Sun Times® and Sputter Spotlight® e-newsletters. E-mail your sputtering applications questions to Doug at Sputtering@aei.com.
- I have a temperature-sensitive substrate. Do you have any suggestions for reducing heat in my process?
- I have a batch coater, and my runs last about 30 minutes. My target tends to arc a lot and the voltage fluctuates throughout the process. Toward the end of a run, these problems begin to subside, but by that time, I’m done with my process and have to vent again. What can I do to stabilize my process sooner?
- I recently replaced my old power supply. Why am I getting more arcing? Does this mean I’m getting more film damage?
- You mean my power supply’s arc-management software can’t be set to reduce the process arc rate?
- You’re saying that arcs that are handled by the power supply don’t cause damage. Is this true only for micro arcs, or hard arcs as well?
- I have a temperature-sensitive substrate. Do you have any suggestions for reducing heat in my process?
Answer: In almost any process, substrate heating improves film properties. For example, it can reduce film stress and improve adhesion. It also can encourage the release of gases and other contaminants from the substrate material that otherwise would alter process chemistry during deposition.
However, some substrates are damaged by elevated temperatures. Polymer substrates may melt, as might certain materials within color filters. Excessive heat also may cause diffusion of underlying doped layers. In temperature-sensitive applications such as these, heat must be carefully controlled. In the ideal scenario, the substrate is heated sufficiently to improve film properties, but not enough to damage underlying deposited layers or the substrate itself.
The main process factors affecting substrate heating are process power type, magnetron type, and process pressure.
Process Power
In the context of vacuum coating, the term stochastic heating refers to heat generated by the oscillation of the voltage waveform. DC has no stochastic heating because the sheath is constant. Any power method with a moving sheath, such as AC or RF, puts energy into the plasma via this movement. This can increase substrate heating.
Therefore, DC may be the best choice for processes with specific thermal budgets. However, many factors must be considered when choosing a process power method. Please consult The Art of Choosing a Power Supply from our Q1 2007 edition Sputter Spotlight for more information.
Magnetron Type
In general, the use of balanced magnetrons results in lower substrate temperatures than the use of unbalanced magnetrons. Use a gauss meter to check the balance of your magnetrons. The combined strength of the magnets on the edges of an unbalanced magnetron is greater than that of the magnet in the middle. Figure 6 shows an unbalanced magnetron in which the strength of the center magnet is 600 Gauss and the combined strength of the edge magnets is 800 Gauss. For a balanced magnetron (Figure 7), the combined strength of the edge magnets is equal to that of the center magnets. Unbalanced magnetrons increase ion energy and widen plasma throw distance. They result in an unconfined plasma because electrons are able to follow magnetic field lines to the substrate. These effects increase substrate heat. On the other hand, balanced magnetrons reduce ion energy, contain the plasma, and focus throw distance. These effects reduce heat.
Figure 6. Unbalanced magnetrons
Figure 7. Balanced magnetrons
Process Pressure
It’s true that generally, as pressure increases, so does temperature. However, pressure affects a number of key process issues, including molecular collision rate, ion energy, plasma throw distance, adhesion, and film stress. Therefore, your goal should always be to attain an optimal pressure that will create the best conditions possible for all of these issues. In other words, it’s not a good idea to turn down pressure just to lower temperature.
Please watch for an upcoming article that details the complexities of process pressure, and also see our discussion of the relationship between pressure and film stress/adhesion in the Q4 2008 edition of Flat Panel Focus.
- I have a batch coater, and my runs last about 30 minutes. My target tends to arc a lot and the voltage fluctuates throughout the process. Toward the end of a run, these problems begin to subside, but by that time, I’m done with my process and have to vent again. What can I do to stabilize my process sooner?
Answer: It sounds like your problem is related to target conditioning. Every time you vent the chamber or put in new cathodes, you have to sputter the target surface clean before you begin processing again. The same is true if your target has been idle for a substantial period or if you’re going to change the process chemistry. Target conditioning removes oxides and other contamination that condense on the surface of the target. It also forces the release of gases permeating the target material. If not eliminated, these impurities alter process chemistry and quality. The arcing and voltage fluctuations that you’re experiencing are probably due to the fact that the target is releasing contaminants during this time. The film quality produced under these conditions is likely poor, as well. These problems subside once these impurities have been purged from the target surface. The only problem is that target conditioning consumes almost your entire processing period and leaves little or no time for sputtering after the target is clean.
I strongly suggest the use of pulsed DC. You can either add a DC pulsing accessory to an existing DC power supply or use a DC power supply with on-board pulsing capabilities. Pulsing dramatically reduces target burn-in (conditioning) time. For example, it has been known to reduce the typical time to condition an Al target from 30 to 3 minutes—by a factor of 10. This means that only 3 minutes of a 30 minute run would be consumed with target conditioning, allowing 27 minutes of good sputtering with a clean target. As you can see, this substantially increases throughput.
Pulsing dramatically reduces target burn-in time. For example, it has been known to reduce the typical time to condition an Al target from 30 to 3 minutes—by a factor of 10.
|
The need to reduce target conditioning time is especially pronounced for batch systems, where venting must be performed frequently, and targets must be re-conditioned just as often. Obviously, the shorter your runs are, the greater your need is to abbreviate burn-in time. If you only sputter for 20 minutes and target conditioning takes 30, your whole process is taken up with no optimal sputtering accomplished—plus, your target doesn’t even get fully conditioned. It sounds like, in your process, you actually are fully conditioning your target and performing a few minutes of processing with a clean target. However, process stability and film quality are probably so poor that throughput is suffering greatly.
Please also note that while you’re burning in your target, the entire sputtering zone also is being conditioned. Contaminants are driven off of the substrate, chamber fixtures, and other surfaces, as well as the target. This is an important factor in achieving a stable process and good film properties. The goal is to achieve an equilibrium temperature and pressure so that all impurities have been driven off before you start depositing film onto the substrate.
Besides throughput improvements related to faster target conditioning, pulsing offers additional benefits, including better target utilization, reduced arcing, and enhanced film quality. Please see the following resources for additional details about these benefits:
- I recently replaced my old power supply. Why am I getting more arcing? Does this mean I’m getting more film damage?
Answer: I only have good news for you on this one. If the only thing that has changed about your process is the power supply, you have no more arcs in your process now than you did with your old unit.
Power supplies detect and report arc events. They do not create arcs. Depending on the quality of their arc-management software, power supplies vary in their “talent” for detecting arcs. Your old power supply was simply incapable of detecting a certain number of arcs. It was less sensitive to the presence of arcs than your new unit and therefore reported a lower arc rate. Your new power supply is now identifying and therefore extinguishing most or all of these formerly undetected arcs.
A comparison of the arc rates reported by the two power supplies will tell you the number of arcs that your old power supply was allowing to sneak through unhandled. This will also give you an idea of the amount of potential film damage that your new power supply is preventing. It is only the ignored arcs that potentially cause damage. Therefore, your new power supply is preventing damage to a greater degree than your old unit.
Simply put, power supplies don’t create arcs. A power supply with effective arc-management capabilities can detect and extinguish arcs after they already exist, but to reduce the actual arc rate, you need to change certain process conditions, such as target material, pressure, voltage, and degree of cleanliness.
|
Simply put, power supplies don’t create arcs, and no power supply’s arc-management software can prevent arcs from occurring in the first place. A power supply with effective arc-management capabilities can detect and extinguish arcs after they already exist, but to reduce the actual arc rate, you need to change certain process conditions, such as target material, pressure, voltage, and degree of cleanliness.
At first, the thought that your power supply’s arc management capabilities don’t prevent arcing may seem alarming. However, once you understand that arcs that are handled properly don’t pose any damage to the process or film, your alarm will hopefully subside. AZO is a good example that illustrates not only the independence of a process’s arc rate from the power supply’s functionality, but also the ability of a quality power supply to negate the effects of arcing—even when that arcing occurs at a remarkably elevated rate.
AZO has an inherent tendency to arc during sputter deposition. Any process depositing AZO, no matter what type of power supply is involved, and no matter what the arc-management settings on that power supply, will exhibit hundreds or even thousands of arcs per second, simply because of this material’s intrinsic properties. However, this doesn’t mean that all processes depositing AZO produce highly damaged films. In fact, high-quality power supplies have arc-management software that is capable of detecting and handling arcs even at this high rate. Every single arc detected by your power supply is extinguished, and properly handled arcs cause no pinholes, particles, or any other kind of damage. In fact, because of arc handling, arcs are not a major cause of pinholes. This enables the deposition of highly arc-prone materials, such as AZO, with low incidence of damage. So, to answer your second question, there has been no increase in your film damage even though you’ve changed power supplies and your new power supply is reporting a higher arc rate.
Every single arc detected by your power supply is extinguished, and properly handled arcs cause no pinholes, particles, or any other kind of damage. In fact, because of proper arc handling, arcs are not a major cause of pinholes. This enables the deposition of highly arc-prone materials, such as AZO, with low incidence of damage.
|
Please note that although arcs are not precipitated by the power supply, there is one particular type of process power that potentially reduces arc rate and is therefore highly recommended for the deposition of AZO. Pulsed DC discharges dielectric surfaces on the reverse pulse. Therefore, any charges built up on these surfaces are periodically neutralized, greatly reducing arcing.
- You mean my power supply’s arc-management software can’t be set to reduce the process arc rate?
Answer: It’s a common misconception that certain settings can be implemented via the power supply’s arc-management software to reduce process arc rate. As discussed above, arc rate can only be reduced by altering process conditions such as target material, voltage, cleanliness, and pressure. The power supply provides data about your process’s arc rate. It’s important to understand that this data is only a report, not a parameter that can be set through the power supply in order to reduce the number of arcs that occur in your process.
Indeed, you can reduce reported arc rate, but you are only affecting the power supply’s report, not the number of actual arcs in the process itself. (If you decrease the voltage trip level on your arc-management software, your power supply will detect fewer arcs and the reported arc rate will go down.) However, this is obviously a bad idea, since doing so allows more arcs to go unhandled by hindering your power supply’s ability to sense them.
It’s best to see a high arc rate reported by your power supply not with a sense of alarm, but with an understanding that this is indicating your power supply’s powerful capacity to respond even to the demands of extreme arcing. The more pressing issue is the possibility that your power supply may be reporting an artificially low arc rate, which means that it is failing to handle a certain number of arcs. If you are experiencing unacceptable levels of arc-related film damage, such as crazing and pinholes, or if your process is highly unstable, increase your voltage trip level so that your power supply detects more arcs. If this doesn’t solve the problem, you may need to switch to a power supply with more highly developed arc-management capabilities.
With regard to setting your power supply’s arc-management settings, in general, you want your power supply to respond as quickly and as often as possible to arcs without sacrificing rate. For more information about setting your arc-management parameters, please see How do I determine the best arc-management settings for my process? from the Q4 2008 edition of Flat Panel Focus.
- You’re saying that arcs that are handled by the power supply don’t cause damage. Is this true only for micro arcs, or hard arcs as well?
Answer: This is another great question, because there is a common misconception that hard arcs are more powerful and damaging than micro arcs. This is simply not true. Hard arc and micro arc are just terms that describe two different power supply responses to unwanted electrical discharges in the process. There is no inherent difference between an arc that elicits a micro-arc response and one that elicits a hard-arc response from the power supply. Therefore, a hard arc cannot cause more damage than a micro arc, since these terms refer only to power supply behavior and not actual process events. It helps to add the term “response” after the terms “hard arc” and “micro arc” to avoid the misconception that there are different types of arcs. Exactly like arcs that elicit a micro arc-response, arcs that elicit a hard-arc response have little energy, and do not cause pinholes or particle generation if properly handled by a power supply.
Hard arc and micro arc are just terms that describe two different power supply responses to unwanted electrical discharges in the process. There is no inherent difference between an arc that elicits a micro-arc response and one that elicits a hard-arc response from the power supply.
|
So, what is the difference between a micro-arc response and a hard-arc response? What’s the reason for this distinction? A micro-arc response is what you probably think of as a typical arc response—the power supply detects an arc, momentarily shuts down to extinguish it, and then turns back on to resume normal power delivery. A power supply performs a hard-arc response if, after performing this typical arc response, the original arc has not been extinguished, or a new arc has already developed in its place.
A hard-arc response differs from a micro-arc response only in shutdown time and recovery algorithm. The hard-arc shutdown time is slightly longer to allow the surface to cool and any lingering arcs to fully extinguish. The recovery algorithm is also different in a hard-arc response. That is, the method the power supply uses to determine that it is “safe” to return to full delivered power is more sensitive with a hard-arc response than with a micro-arc response. The following flow chart depicts the typical event sequences for a micro-arc and then a hard-arc response.
Figure 8. Typical event sequence for a micro-arc and hard-arc response (Click graphic to enlarge.)