Cleanroom Technologies

Controlling particle transfer caused by cleanroom gloves

(Second in a series)

The second installment in a three-part series explores how cleanroom gloves act as a source of contamination and investigates how washing
reduces glove-induced particle transfer.

Roger W. Welker, R.W. Welker Associates

Cleanroom gloves are critical for contamination control in the semiconductor, flat-panel display, disk-drive, and other high-technology industries. They are among the most important and expensive of cleanroom consumables. Given how they are used, they are also among the most likely sources of contamination, primarily as a result of contact transfer. While many consumers rightly specify how clean the gloves they use must be when received from glove manufacturers, few of them know that the cleanliness of their gloves can be improved by washing them and that glove cleanliness can gradually degrade through use in the cleanroom. In the first part of this series (MICRO, May 1999), the use of contamination and electrostatic discharge (ESD) tests to qualify and certify cleanroom gloves was discussed.¹ This installment deals with the contamination state and performance of cleanroom gloves while they are in use. It explores the contamination behavior of natural latex cleanroom gloves and examines studies on the contamination behavior of nitrile rubber gloves. Furthermore, it discusses the as-received cleanliness level of gloves, the effects of washing gloves, the gradual recontamination of gloves during use, and the effects of in situ glove cleaning. A future article will explore the factors affecting ESD performance under realistic cleanroom use conditions.

Natural Rubber Latex Gloves

From 1985 to 1987, gloves made of natural rubber latex were the subject of scrutiny. Because high-tech products at that time exhibited relatively low sensitivity to ESD damage, natural rubber latex was the material of choice in nearly all electronics fields, including the semiconductor industry. (IC manufacturers also used static-dissipative polyvinyl chloride gloves in ESD-sensitive areas.) However, users observed that while one particular brand of natural latex gloves did not leave fingerprints on hard-disk-drive (HDD) parts, four others used in the same cleanrooms left fingerprints. Consequently, the five different brands of gloves were analyzed in a contamination control lab. In the first test, technicians touched the surface of clean microscope slides wearing several different pairs of each brand of glove. The microscope slides were then inspected at 160* magnification using dark-field light microscopy. It was found that gloves leaving <500 particles/cm² (2.0 µm diam) did not leave a visible fingerprint on HDD parts, while gloves leaving >5000 particles/cm² left a distinctly visible fingerprint.² This was the first evidence showing a correlation between contact-transferable particles and a rejectable defect on an HDD part.

Recognizing that light microscopy is a tedious method for inspecting glove cleanliness, a correlation was established between the light microscopy of glove particle extracts and the liquidborne particle counts (LPCs) of glove extracts.³ Although not reported at the time, a correlation was also established among microscope counts of particles deposited on slides, LPCs, and particles counted on filters.⁴

A foot-pedal-operated cleanroom glove wash station with soap dispensers. In the background, wall-mounted HEPA-filtered hand dryers. Photo by Lynn Liebshutz, IBM, San Jose

The LPC procedure required the development of a particle extraction method. Since preliminary tests showed that ultrasonic extraction was not an acceptable method because of the extreme erosion sensitivity of natural rubber latex, another extraction procedure was devised. First, 300 ml of DI water and 0.02% by volume detergent were placed in a glove. Then the glove was placed in a 2-L beaker containing 500 ml of DI water and 0.02% detergent and subjected to orbital shake at 120 rpm for 10 minutes. After the glove was removed from the beaker, the water was drained from the glove and poured into the water in the beaker. Then the water was ultrasonically degassed by rapidly pulsing the power on and off. As soon as this step was completed, the suspension was measured. Because the gloves in use were hand specific rather than ambidextrous, operators sometimes would invert a glove and wear it on the opposite hand when the correct glove for that hand was unavailable, thereby making the inside of the gloves nearly as important a source of transferable contamination as the outside. Thus, both the inner and outer surfaces of the gloves were measured.

In addition, preliminary tests were conducted to determine anion concentration levels on the gloves. This was accomplished by filling gloves with pure 18-M(omega)/cm DI water, sealing them with a twist tie, and placing them on a hot plate at 60° to 80°C for 1 hour. After allowing the gloves to cool to room temperature, samples were analyzed by means of ion chromatography. It was found that natural rubber latex gloves at that time had an average of >8000 particles/cm² (>=0.8 µm diam) and 2—10-µg/cm² anion contamination dominated by chlorides and sulfates.

Maintaining Cleanliness through Glove Washing

The next problem was to determine if hand washing could clean gloves. It seemed that if particles clung loosely enough to gloves to be removed by contact, simply washing the gloves by rubbing the hands together under a stream of running water might be effective. This portion of the study used a C1515 wafer-surface inspection system from Hamamatsu (Hamamatsu City, Japan). The instrument's sensitivity was adjusted so that its 50% detection efficiency was set at approximately 5 µm. An operator wearing a fresh pair of gloves touched the surface of a bare silicon wafer. The resulting fingerprint contained an average of 175 particles/cm². After the gloves were washed in a stream of running DI water and dried in a conventional (not a HEPA-filtered) hand dryer, the operator touched the wafer a second time, depositing an average of 3 particles/cm² (>=5.0 µm diam).⁵

These experiments established that glove cleanliness could be measured using contact transfer followed by microscopy or wafer inspection. They also showed that measurements of extracts from gloves using LPC proved that glove washing was effective in reducing contact transfer. The question remained: what happens to gloves during manufacturing operations? To answer this question, gloves worn during actual manufacturing operations were tested using a new form of LPC extraction by which only the outside surface of the gloves was measured, since contamination accumulated on the inside surface was not thought to pose a risk of contaminating HDD parts.

The first step of the extraction procedure involved filling a clean 500-ml beaker with ~350 ml of DI water and 0.02% detergent and ultrasonically degassing the solution. The degassed solution's LPC was measured by removing ~50 ml from the beaker, and this value was used as the LPC blank. Then the gloves were turned inside out, which happens naturally when operators remove gloves worn during the manufacturing process. The inverted gloves were filled with the remaining 300 ml of detergent-and-water solution from the beaker, sealed with a twist tie, and then undulated for 10 minutes in a 2-L beaker containing 500 ml of detergent-and-water solution. The solution inside the gloves was poured back into the 500-ml beaker and ultrasonically degassed by pulsing the power on and off rapidly. Finally, the LPC was measured without delay. The surprising results of this experiment, summarized in Table I, were confirmed by repeat testing.

Glove Condition	Particles/cm2 (=0.8 µm diam)	Percentage That Leaked
As received	8520	0
After 2 hours of use (trial 1)	8270	70%
After 2 hours of use (trial 2)	8396	57%

Table I: Higher overall particle levels were measured on natural rubber latex gloves after 2 hours of use when fewer gloves sprang leaks and thus were tested.⁵

Something was clearly wrong. How could gloves used for 2 hours in the manufacturing process be as clean as new? The answer was that gloves leaking at the fingertips after the detergent-and-water solution had been poured into them had not been LPC tested. By factoring in these gloves, the statistical dilemma was solved. It was thought that the small number of gloves that had not sprung leaks had not been used for a full 2 hours because operators had changed gloves during the operation. The new gloves had had little opportunity to accumulate or shed particles. It was disturbing to discover that a significant percentage of the operators wore gloves with unnoticed pinholes in the fingers, which was clearly undesirable from a contamination standpoint.

A commercial glove wash station with built-in dryer. Photo courtesy of Pentagon Technologies

To get a better indication of the glove recontamination rate, the contact transfer of particles to silicon wafers was measured using the wafer surface inspection system. In this procedure, gloves were first washed and measured for particulation. The operator then walked around the cleanroom for 5 minutes touching objects on the work surfaces. The contact transfer of particles to the silicon wafer was measured again. The process of touching the cleanroom work surfaces, tools, and oscilloscopes was repeated for another 15 minutes, followed by additional measurements. Finally, the gloves were washed in running DI water and measured. The results of this experiment, shown in Table II, indicate that gloves do not remain clean after washing. Within 20 minutes, their contact transfer rate has returned to approximately half that of gloves in the as-received condition. While glove washing initially reduces contact transfer, it was evident that this benefit is only temporary.

Glove Condition	Contact Transfer (particles/cm2 =5 µm diam)
As received	175
After washing	3
After 5 minutes use in the cleanroom	50
After 20 minutes use in the cleanroom	90
After rewashing	3

Table II: Changing particle levels on natural rubber latex gloves following washing, recontamination, and rewashing.

Three options for maintaining glove cleanliness during the manufacturing process were considered. The first was to instruct operators to rewash their gloves in the glove station. This was rejected as having too negative an impact on productivity. The second option was to wipe the gloves with a cleanroom wiper wetted with isopropyl alcohol. This option went untested because there was no low-cost, premoistened, disposable cleanroom wiper on the market. The alternative of using a knitted or woven cleanroom wiper was rejected as being too difficult to control and too expensive. The third option, to provide cleanroom operators with workstation sticky mats, was evaluated and eventually adopted.

The effectiveness of the workstation sticky mat in reducing contact transfer was demonstrated by cutting a standard cleanroom floor mat to approximately a 6-in. square and making it available to about 20 operators who washed their gloves and wore them for 2 hours during the manufacturing process. A set of clean microscope slides was used to collect contact transfer samples. First, the operators touched a microscope slide with a freshly washed glove and then touched a slide with a different gloved finger after touching the sticky mat. Then they touched a slide with the glove after 2 hours of use. After touching the sticky mat with a different finger, they touched another slide with this cleaned finger. The particles transferred to the microscope slides were then measured using dark-field microscopy and recorded at two different sizes: 2.0 and 10.0 µm. Table III shows the results of this experiment.

Glove Condition	Particle Size (µm)	Contact Transfer (particles/cm2)
		Before Using Sticky Mat	After Using Sticky Mat
Washed	10	0	0
	2	120	70
After 2 hours	10	5	0
	2	300	200

Table III: Results of tests demonstrating the effectiveness of using a sticky mat for the in situ cleaning of natural rubber latex gloves.

The freshly washed gloves transferred no particles larger than 10 µm to the clean glass microscope slide. Moreover, in contrast to even the cleanest unwashed natural rubber latex glove, which transfers about 500 particles/cm² (2.0 µm diam), the washed gloves transferred 120 particles/cm² (2.0 µm diam). This was reduced by about 40% after the operators touched the sticky mat. After 2 hours of use in the cleanroom, the washed gloves transferred an average of 5 particles/cm² (10.0 µm diam). After the operators touched the sticky mat, no particles >=10.0 µm were transferred to the slide, indicating that the sticky mat is 100% effective for removing the newly accumulated large particles from the gloves. After 2 hours of use in the cleanroom, the gloves transferred an average of 300 particles/cm² (2.0 µm diam).

Sticky mats thus came into use at all manual assembly workstations. Besides the effectiveness of this method for reducing the contact transfer of particles, especially those >10.0 µm, two other effects were noticed. First, sticky mats tore the fingertips off gloves that had pinholes or tiny fingertip tears. Second, the assembly operators reacted to the presence of sticky mats by remembering that their gloves must be kept clean.

Nitrile Glove Performance

Beginning in 1991, a gradual and important shift in glove selection occurred with the introduction of nitrile gloves measuring 0.003—0.004 in. thick. These inexpensive gloves, offering good dexterity, durability, and static-dissipative properties, soon became prominent in the HDD, medical, and semiconductor industries. This was especially true in the HDD industry because of the introduction of magnetoresistive (MR) heads, which are extremely ESD sensitive. The sensitivity of MR heads to electrostatic discharge made the continued use of insulative natural rubber latex unacceptable. Nitrile gloves also became increasingly important because they are more resistant to pinholes and leaks than their rubber latex predecessors.

With the introduction of nitrile, the issues of washing, recontamination, and in situ cleaning came to the fore. Tests to determine nitrile cleanliness were performed. As illustrated in Table IV, washed nitrile gloves show dramatically higher levels of particle and anion cleanliness than as-received gloves.

Parameter	Gloves as Received	Gloves after Washing
Particles/cm2 =0.5 µm	1000—4000	25—100
Anions (µg/cm2)	1—4	0.01—0.10

Table IV: Typical particle and ionic cleanliness levels for nitrile gloves as received and after washing.

Another test was performed to determine the tendency of nitrile gloves to become recontaminated. In this study, gloves worn by five operators, each of whom performed a separate manual assembly operation, were investigated. First, an entire lot of gloves was measured for particles and anions on two occasions: as the lot was received and after washing. Then two types of dryers were evaluated: a conventional heated hand dryer and a dryer with a HEPA filter—that is, a cleanroom version of a hand dryer. The outside surfaces of the gloves were measured after they had been worn 1/2, 1, 2, 4, and 8 hours. As summarized in Table V, washed gloves had more than 5 times fewer particles per square centimeter (>=0.5 µm diam) and nearly 10 times fewer anions (µg/cm²) than as-received gloves. There was little difference in cleanliness between gloves dried with a conventional heated hand dryer and those dried with a HEPA-filter hand dryer. Interestingly, the initial cleanliness level of nitrile gloves, as shown in Table V, is considerably higher than the initial cleanliness level of natural rubber latex gloves, as shown in Table I.

Glove Condition	Particles/cm2 (>=0.5 µm diam)		Anions (µg/cm2)
	Mean	Standard Deviation	Mean	Standard Deviation
As received	1120	400	0.940	0.055
Washed, conventionally dried	200	10	0.095	0.005
Washed, HEPA-filter dried	190	10	0.090	0.010

Table V: Test results showing particle and anions levels on nitrile gloves after washing and by two different drying methods.

Figure 1 demonstrates how the length of time nitrile gloves are worn in the cleanroom affects glove cleanliness. In this study, data for all workstations were averaged together, because there did not appear to be differences between the recontamination rates at different workstations. Gloves worn 1 to 2 hours had particle contamination levels equal to those of as-received gloves. In addition, particle contamination increased steadily with time. In contrast, while anion contamination levels increased over time, they did so by no more than 20% of the as-received values even after 8 hours of use (data not shown). The buildup of anion contamination was not continuous, suggesting that anion contamination levels may be dependent on the inherent variability of the anion content of the gloves rather than the effects of time. Additional study is required to determine whether nitrile gloves undergo recontamination, whether anions diffuse to the surface, or whether new surface areas become exposed, allowing the extraction of previously unavailable anions.

Figure 1: Experimental results showing the effect of wear time on the particle cleanliness of nitrile gloves.

Conclusion

Gloves do not remain clean during use. A small but significant increase in ionic contamination levels occurs during use in the manufacturing process. More important, within 1 to 2 hours gloves become recontaminated by particles to a level equal to or greater than the as-received level. The cleanliness of cleanroom gloves can be improved significantly by washing, which results in a decrease in particulate and ionic contamination of one to three orders of magnitude.

References

1. RW Welker and PG Lehman, "Using Contamination and ESD Tests to Qualify and Certify Cleanroom Gloves," MICRO 17, no. 5 (1999): 47—51.

2. RW Welker, previously unpublished laboratory data.

3. R Coplen, RW Welker, and RL Weaver, "Correlation between ASTM F312 and Liquidborne Optical Particle Counting," in Proceedings of the 34th Technical Meeting of the Institute of Environmental Sciences (Mt. Prospect, IL: Institute of Environmental Sciences and Technology, 1988), 390—394.

4. RW Welker, previously unpublished laboratory data.

5. RW Welker, "Glove Selection and Use," (papers presented at the IBM Contamination Control Course, Paris, France, April 19—21, 1994).

Roger Welker is founder and principal scientist of R.W. Welker Associates, a consulting firm specializing in contamination and electrostatic discharge control. He has 17 years of experience in high-technology development and manufacturing at IBM, Seagate, and Micropolis. He also spent 11 years in applied R&D, focusing mainly on applications of fine particles. Welker has authored or coauthored more than 60 papers and is a member of the Institute of Environmental Sciences and Technology, the American Association for Aerosol Research, the Electrostatic Overstress/Electrostatic Discharge Association, and the Data Storage Institute. He received his BS in chemistry from the University of Maryland in College Park. (Welker can be reached at 818/368-0557 or rwwlws@aol.com.)

MicroHome | Search | Current Issue | MicroArchives
Buyers Guide | Media Kit