By Robert P. Donovan
Relative humidity is one of the environmental conditions usually specified for cleanroom operations. Relative humidity in a semiconductor cleanroom is typically controlled to a target value somewhere within the 30 to 50 percent range, and with tolerances as narrow as ± 1 percent for some areas, such as photolithography—or even less for deep ultraviolet processing (DUV)—and relaxing to ± 5 percent in other areas.
Keeping makeup air within these specifications throughout the year entails capital and operating expenses. But why does control of relative humidity in a cleanroom warrant such costs?
Quite simply, because relative humidity influences a number of factors that could degrade overall cleanroom performance, including:
- Bacteria growth;
- Personnel comfort zone;
- Static charge build-up;
- Metal corrosion;
- Moisture condensation;
- Photolithographic degradation;
- Water absorption.
Bacteria and other biological contaminants (mold, viruses, fungi, mites) thrive in relative humidity above 60 percent. Some bacterial populations also increase when relative humidity is greater than 30 percent. A 40 to 60 percent range in relative humidity minimizes the impact of bacteria and respiratory infections.1
The 40 to 60 percent range of relative humidity also falls well within the personal comfort zone. Higher humidity becomes oppressive; humidity lower than 30 percent leads to dryness, cracked skin, respiratory discomfort and unhappy personnel.
High humidity actually minimizes the buildup of static electrical charge on a cleanroom surface—a highly desirable effect. It's low humidity that favors charge build-up and the potentially damaging electrostatic discharges. Static charge begins to dissipate rapidly as the relative humidity exceeds 50 percent, but persists for long times on nonconductive or ungrounded surfaces when below 30 percent.
While a relative humidity of 35 to 40 percent may be an acceptable compromise, semiconductor cleanrooms typically incorporate additional control features for limiting static electrical charge build-up.
The rate of many chemical reactions, including corrosion processes, increases with increasing relative humidity. All surfaces exposed to ambient cleanroom air are rapidly covered with at least a monolayer of water. When these surfaces are thin-metal films of a composition that reacts with water, high humidity feeds that reaction. Fortunately, some metals, such as aluminum, form a protective oxide with water that blocks further oxidizing reaction; copper oxides, on the other hand, are not protective, so copper surfaces are vulnerable to high humidity.
At high relative humidity, capillary forces created by condensed water form a bonding bridge between the particle and the surface, and can increase particle adhesion to a silicon surface. This effect—Kelvin condensation—is unimportant at relative humidity less than 50 percent but can constitute the dominant force of particle adhesion in relative humidity as low as 70 percent.
By far, the most pressing need for humidity control in a semiconductor cleanroom comes from photoresist sensitivities. It's photoresist that demands the tightest (most expensive) control limits because its properties are so sensitive to relative humidity.
Actually, both relative humidity and temperature are critical for resist stability and precise dimensional control. Even at constant temperature, photoresist viscosity decreases rapidly with increasing relative humidity. Changing viscosity, of course, changes the thickness of a resist film spun-on by a fixed coating recipe. Reference 2 cites an experimental demonstration in which a three percent variation in relative humidity produced a thickness variation of 59.2 A (sic) in resist thickness.
In addition, resist swelling following a bake cycle can be aggravated by water absorption at high relative humidity. Resist adhesion can also be adversely affected by high relative humidity; low humidity (~30 percent) facilitates resist adhesion even without polymeric modifiers, such as hexamethyldisilazane (HMDS).
Relative humidity control in a semiconductor cleanroom is not optional. But, from time to time, it's good to review the reasons and bases for common, universally accepted practices.
Robert P. Donovan is a process engineer assigned to the Sandia National Laboratories and is a monthly columnist for CleanRooms magazine. He can be reached at: [email protected]
- Sterling, E. M., A. Arundel and T. D. Sterling, “Criteria for Human Exposure to Humidity in Occupied Buildings,” ASHRAE Transactions 91, Part 1, 1985 (CH-85-13, No. 1), pp. 611-622
- Gurer, E., T. Zhong, J. Lewellen, M. Krishna and E. Lee, “Photoresist Processing Tool-Based Advanced Technologies for DUV Lithography and Low-k Spin on Dielectrics of 200/300 Wafers,” Semiconductor FABTECH, 12th edition, July 2000, pp. 161-165