|   |   |   |

To receive the next issue by e-mail, visit My Account and confirm you have opted-in to receive e-mail from AE.

The Q3 2008 edition of Sputter Spotlight^® aims to demystify crazing—to put cracks in its veneer, so to speak—by describing key phenomena that underlie this type of damage. This two-part article discusses general causes and remedies for crazing, as well as a number of key cases that may provide useful insights and general problem-solving strategies. Please refer back to this newsletter in Q1 2009 for the second part of Sputter Spotlight’s “Cracking the Crazing Code.”

Cracking the Crazing Code: Part One

Despite its whimsical name, crazing is a very real and frustrating problem that many industries deal with, including architectural and residential glass, FPD, and solar. It’s a phenomenon that seems to have an aura of mystery, perhaps because there is such a wide variety of possible triggering factors and therefore no standard solution.

However, for the most part, there is no mystery to the visual identification of crazing. Although this type of damage can vary somewhat in visual appearance, generally, you know it when you see it: those characteristic cracks that branch out across a surface in lightning-like patterns (Figure 1). The next time you find yourself standing on an old, worn concrete slab, notice the similarities in the pattern of cracks under your feet (Figure 2) to that in an FPD SiO₂ barrier layer that has crazing damage. In fact, the same term, “crazing,” is used in the U.S. to refer to similar patterns of cracking in ceramic materials such as concrete, dishware, and even teeth. It’s been said that the term actually originated in the pottery field. Regardless of the term’s origin, crazing is perhaps unique in its ability to cause consternation in such a great variety of contexts and industries.

Figure 1. Crazing in a thin film

Figure 2. Crazing in a concrete slab

In a thin film, the patterns of significant thinning and fissures that characterize crazing can cause catastrophic device failures if not detected early. Either luckily or unluckily (depending on your perspective), when crazing occurs, it’s usually dramatic enough to be obvious before the affected device continues too far along in the manufacturing process. However, dealing with the problem after the fact is obviously an insufficient strategy. Like every other type of film damage, crazing must be traced back to its root cause in order to devise a way to eradicate it.

Process Diagnosis and Repair

Most of the time, the concept of an unwanted electrical discharge refers to a phenomenon that occurs on the cathode surface, what we commonly refer to as arcing. This usually is solved with a given set of known methods: effective power supply arc-management technology and other familiar process adjustments. However, in the case of crazing, the damaging electrical discharge occurs not at the cathode, but on other remote surfaces, such as shielding, carriers, or, in the worst cases, the substrate. Electrical discharges on the substrate surface can’t be detected, let alone handled, by any power supply, no matter how advanced its arc-management technology. This is due to the fact that there may not be a dramatic change in voltage or current at the cathode. Thus, the power supply has no means of detection.

Beyond the fact that arc-management technology does not apply here, when crazing occurs, process diagnosis and repair is particularly difficult and time consuming compared to other types of film damage. This is because a variety of phenomena can cause crazing. Furthermore, once a diagnosis is made, devising a solution also takes a great deal of time and testing. Unlike most cases of arcing at the cathode, the electrical discharges at the substrate surface that cause crazing aren’t solved with a specific set of known methods. Today’s requirements for high throughput and small system footprint necessitate specific system designs and conditions that may unintentionally lead to crazing. Therefore, a hard-won, custom solution is usually necessary because of the fact that system conditions vary from process to process.

Although each case of crazing is unique, the following sampling of crazing origins and solutions may provide some general insights and strategies that you can apply to your own situation. At the very least, these cases hopefully will make the prospect of solving crazing in any process a little less daunting.

Common Sources of Crazing

Clues that can help indicate the conditions that are triggering crazing in a given process include the following:

• The location of the damage on the film surface
• The particular visual look of the damage
• The manner of onset

The section below describes a major potential source of crazing. Also included are the symptoms most often associated with the given instigating factor, as well as a detailed explanation of the problem and suggestions for repair. Additional sources of crazing will be discussed in the Q1 2009 edition.

Lack of Chamber Cleanliness

Symptoms:

• Damage distributed across the substrate
• Characteristic “crazed” visual appearance
• Relatively sudden onset of damage after an extended period of satisfactory film quality

The buildup of film material on chamber surfaces is perhaps the most common cause of crazing. In the case of a metallic film, this buildup creates unwanted electrical paths that lead to electrical discharges where you don’t want them. Dielectric film deposition can also cause unwanted discharges because of its ability to retain and build an electrical charge. Crazing usually is the result when these discharges occur on the surface of the substrate.

Lack of cleanliness is most likely the cause of crazing when a new system has been operating satisfactorily for a certain period of time and then suddenly starts producing crazed films. A new system is, of course, completely clean and free of any kind of material buildup. However, as soon as it is put into operation, material begins to condense on all of the surfaces inside the chamber. Over time, this deposition builds until it is substantial enough to bridge gaps between components. If the buildup is metallic, this creates unwanted electrical paths. These paths can connect the substrate to ground, which promotes electrical discharges at the surface of the substrate and leads to crazing. For crazing caused by dielectric buildup, an insulative film such as SiO₂ builds up over time on chamber components such as a shield or carrier. These components become electrically charged and, because of their proximity, discharge at the substrate. Substrate crazing can also result from the reverse process, in which an electrically charged substrate discharges at a grounded carrier, roller, or other adjacent chamber component.

Of course, the phenomena described above aren’t limited to pristine, new systems that have never been operated. The same thing can happen to an older system if a significant period has passed since the last cleaning.

To solve crazing triggered by material buildup, it’s necessary to physically remove the unwanted coating from all internal chamber surfaces. Although most factories clean their chambers whenever the targets are replaced, in order to avoid crazing, it may be necessary to clean more often.

Please note that film buildup can exacerbate any of the instigating factors that will be described in our Q1 2009 newsletter, so it’s a good idea to clean your chamber at the first sign of crazing, even if additional causes may exist.

Conclusion

Please watch in Q1 2009 for the extensively illustrated conclusion to this article, which includes more crazing cases and solutions. In the mean time, please contact Doug Pelleymounter at Sputtering@aei.com or Ken Nauman at Ken.Nauman@aei.com 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 sputtering applications. He is a major contributor to AE's PV Sun Times^® and Sputter Spotlight^® e-newsletters. In this column, Doug helps you answer some of your difficult application questions. E-mail your sputtering applications questions to Doug at Sputtering@aei.com. 

1. I have a question concerning the Pinnacle^® Plus+ 5 kW (325 to 650 VDC). We are using it with a small 3" molybdenum target (in DC mode) with a magnetron for sputtering moly films. However, our current process runs only at 300 W (at about 400 VDC; 0.75 A), which is well below the specified repeatability ("0.1% from 10% to 100% of rated power"). I'm a bit concerned about the overall process stability. Do you have any idea on the output accuracy and repeatability of the 5 kW power supply, operating at only 300 W? Would it be better to use a 500 W DC model?
2. How far apart do the magnetrons need to be when you are using an AC supply and a dual-magnetron system (DMS)?
3. What rate can I expect from a dual-magnetron system and AC supply compared to a DC supply and one magnetron?

1. I have a question concerning the Pinnacle^® Plus+ 5 kW (325 to 650 VDC). We are using it with a small 3" molybdenum target (in DC mode) with a magnetron for sputtering moly films. However, our current process runs only at 300 W (at about 400 VDC; 0.75 A), which is well below the specified repeatability ("0.1% from 10% to 100% of rated power"). I'm a bit concerned about the overall process stability. Do you have any idea on the output accuracy and repeatability of the 5 kW power supply, operating at only 300 W? Would it be better to use a 500 W DC model? –Jeorg Winkler
   Answer: Yes, you do have a problem there. While Pinnacle power supplies and their Pinnacle Plus+ cousins can indeed run “down in the weeds,” I would advise against it. AE supplies are current sources and need some sort of amperage to lock their control circuits onto. I have had a couple of other folks try to turn the Pinnacle Plus+ on at this low power who are having some strange and wonderful things happen to their processes. The current is too low for the control circuits, and accuracy will go out into the 5% range at this low power. Repeatability might not be sufficient either, as the supply might be thinking it is in a constant arc.
   
   I agree that you should run with a Pinnacle 3 kW or an MDX 1.5 kW with a Sparc-le^® V. We know that the Sparc-le V needs about 110 to 150 W just to “heat its innards.” Knowing this, you can make the appropriate adjustments and get much better repeatability.
   
   
   



2. How far apart do the magnetrons need to be when you are using an AC supply and a dual-magnetron system (DMS)?
   Answer: This is a really good question. When you are using planar magnetrons, they can be as close as 2.5 cm (1″) and as far as 1 m (3'). The key is that you need a good path for electron flow between the magnetrons. Nothing other than the darkspace shields should be in the way of the electron flow. If you are using planar magnetrons side by side (facing the same direction or even slightly tilted toward each other), you will need to “rotate the tires”—swap target faces or even rotate them on the same magnetron, because the racetrack that is closest to the other magnetron will wear very fast, which reduces the target utilization.
   
   
   



3. What rate can I expect from a dual-magnetron system and AC supply compared to a DC supply and one magnetron?
   Answer: If we say that DC is always on and therefore we will get 100% rate of deposition, then an AC supply with a DMS will have about 80 to 85% of this because it crosses zero when it changes sputtering magnetrons. Due to the constantly changing AC signal, it cannot give a consistent signal to the magnetrons for sputtering. When it nears zero, the sputtering decreases, and when it actually hits zero, sputtering ceases. The other magnetron must re-light the fire, so there is a lag in current here.