Lower Thermal Budgets Affect Contamination

Tool builders and fab operators are constantly seeking ways to lower thermal budgets to speed throughput and use less resources, utilities and energy. For example, in some epitaxial deposition processes, engineers have been able to operate the processes at lower temperatures and still meet rigorous thin-film performance specifications. That means the process takes place on a more rapid turnaround basis because the chamber doesn't need to be heated to over 1000°C, as was the case a few years ago. However, what engineers have discovered is that with the lower-temperature process, contaminants that were not there at higher temperatures are increasingly present and remain in the process tools, pressure vessels, carriers and, in some cases, on the surface of the wafers. Hence, users have had to find ways to monitor the presence of those contaminants more closely.

The key contaminants of concern are moisture and oxygen — what some refer to as the "twin terrors of semiconductor manufacturing." Interestingly, a couple of years ago, there was not a great deal of agreement over what levels of contaminants were considered too high and at what points in the process one wanted to measure those levels of contaminants. But the industry has moved to increasing concurrence on those questions. One of the agreements has been that oxygen, although a serious contaminant, is driven out of the process vessels and tools faster than moisture is. The focus has increasingly moved toward moisture detection, not to the exclusion of oxygen detection, but preferentially more so than oxygen detection.

In terms of determining the best places to monitor, the process engineers want to measure moisture in the transfer chamber before they introduce wafers into the process tool. Most process engineers also want to measure at the exhaust of the process chamber after etch to make sure the gases have been acceptably purged out. Then there are disagreements about trying to measure in-process or not. In-process measurements involve changing gas mixtures, temperature levels and pressure levels — all of which represent challenges to any type of analyzer technology.

With moisture, a notoriously sticky molecule, analyzer users were faced with the challenge of waiting for the gas lines to equilibrate and thereby produce a gas flow into the analyzer that represented a true level of moisture in the gas, as opposed to outgassing or adsorption in the gas lines because of environmental effects. Today, the analyzer speed-of-response is so fast (one measurement per second) that the analyzer is no longer the limiting factor. But changing moisture levels will not present themselves to the analyzer at that pace. So we still have customers that find changing moisture levels caused by diurnal effects (the difference in temperatures between daytime and nighttime), pressure effects, flow effects and seasonal effects (the difference between the warmth of the earth and the effect on gas lines and gas tanks in the spring and summer is quite different than the fall and winter in most countries). Therefore, the need for the analyzer to immediately react has been met. Now it's a question of the analyzer waiting for the sample gas conditions to change for the analyzer to respond.

Contaminants arise from a variety of sources. The gas is typically brought into the fab in giant cylinders, which are specified to ultralow levels of contaminants. Even so, oftentimes it is desirable to periodically verify that the cylinder has the stated level of purity because, as a cylinder of gas is depleted, moisture clings to the cylinder walls and will release or outgas, depending on temperature, draw rate out of the cylinder and pressure in the cylinder. That means that a cylinder that meets its spec might, in fact, contribute more moisture to the lines, simply because of the low level of gas remaining in the cylinder, a fast draw-down rate or, bizarrely enough, a hot day. Also, moisture is incredibly sticky, so it clings to surfaces, even electropolished stainless steel.

Some of the gases used in semiconductor processing, hydrogen chloride (HCl) for instance, are quite difficult to purify. It is important to verify that HCl purifiers have not broken through and are still doing their job. There is also the possibility of minute leaks in the gas supply system, fittings and the tools themselves, which have to be reconditioned after a certain number of cycles for leak tightness. There are many opportunities for contaminants to get into the chamber and land on the wafer. Even opening the doors to FOUPs and bringing wafers into the chamber brings in cleanroom air, which, of course, contains a great deal of moisture.