Investigation of airborne molecular base and ammonium off-gassed from cleanroom gloves
06/01/2001
AMC
By Andrew J. Magenheim, Ph.D., and Mike Hansen
DO GLOVES REPRESENT A RISK TO BASE-SENSITIVE PROCESSES IN PHOTOLITHOGRAPHY?
Airborne molecular contamination (AMC) is evolving as a major concern for semiconductor manufacturers. In particular, chemically amplified DUV photoresist is sensitive to extremely low levels of basic airborne molecular contamination.1 Airborne molecular bases include ammonia, amines or N-methyl pyrrolidone (NMP) among others.
These species originate from many items found in cleanroom environments, including construction materials, filters, caulks, cleaning solutions and personnel.2 In many facilities, the wafer must travel from an exposure tool, to a develop tool, and finally to an oven for the baking process. Critical dimension shifts (as much as 10-20 nm) and "T-topping" (widening in the top of the image profile) have been observed during a 10-minute interval (between steps) where the total molecular base (TMB) contamination levels were greater than 18 ppb.2 For reference, some observations of TMB in cleanrooms are summarized in Figure 1.
Kishkovich et al. (1999) showed that latex gloves could be a source of airborne base in cleanroom environments using a real-time TMB monitoring experiment.3 Significant variability was demonstrated between products suggesting that gloves and other cleanroom consumables may be sources of basic AMC. The following investigation was conducted to expand the information presented in the context of gloves marketed for cleanroom use.
 |
Figure 1: Ranges of TMB from various environments/objects. Data is from Kishkovich et al 1999; Kishkovich and Larson 2000, and this paper. Vertical line represents 20 ppb threshold observed for the onset of line defects.
Two types of thin-walled latex gloves are commonly used in cleanroom environments: natural rubber latex (NRL) and Nitrile (Acrylonitile Butadiene). Both latex materials may use minor amounts NH4OH or amines as stabilizing agents for the raw latex emulsions. Thus, it is conceivable that both NRL and nitrile gloves could contribute to the TMB in the controlled environments.
Polyvinyl Chloride (PVC or vinyl) gloves are sometimes used in cleanrooms. However, mobile components (such as DOP) in PVC gloves limit the use to applications insensitive to contamination by plasticizer residue. Unlike latex emulsions, the PVC processes do not involve base-stabilization and thus would be expected to have lower risk to base-sensitive processes.
Experimental
Ten glove products marketed for sale in cleanroom environments were evaluated for TMB using the real-time monitoring methods described by Kishkovich et al (1999).3 One piece of each product was measured (with duplicates run for quality control). Due to the limited sample size, the results and conclusions presented should be taken as a survey of general trends and not the absolute performance of a specific product.
 |
Figure 2: Example results from Total Molecular Base (TMB) real-time monitoring experiment for three Nitrile cleanroom gloves. Note all profiles show an initial peak tailing to a steady state value of TMB. The results show significant differences in both the magnitude of the initial peak and the steady state evolution of TMB between different brands of Nitrile gloves. Note the relative high background TMB concentration for Sample N-1.
The experiments are conducted by passing dry air at room temperature through a test chamber holding the specimen (a glove in this case) to monitor the passively off-gassed contaminants in the air stream. The effluent is analyzed in real-time using a chemiluminescence detector configured to detect total airborne molecular bases.4
 |
Results for total molecular base evolution and extractable ammonium content for cleanroom gloves. Samples include NRL-Natural Rubber Latex, Nitrile, and Polyvinyl Chloride glove products. TMB results include both the maximum airborne concentration in part per billion (ppb, µL TMB/L air) and the cumulative TMB evolved in micromoles (µmoles) over the two-hour duration of the experiment. Water extractable ammonium in µmoles is based on extraction in Deionized water at 40°C for 4 hours).
The procedures are identical to those previously used with the addition that the gloves were conditioned by opening the bag in a low TMB cleanroom for one hour. This conditioning step was added to eliminate potential artifacts for products that are packed in vacuum-sealed bags in comparison to those that are not. Functionally this simulates standard procedures that involve placing gloves in dispensers in the gowning room prior to use. The experiments produce output in terms of TMB concentration in ppb (µL TMB per L air) at a given time and are recorded for a two-hour period. Results were integrated over the two-hour period to obtain the cumulative TMB off-gassed in micromoles (µmole).
The real-time monitoring experiments are very informative, but require unique equipment and specialized configuration. For this reason we analyzed the same products using standard Ion Chromatography to determine if the TMB evolved was coincident with extractable ammonium. Ammonium was evaluated in deionized water-based extracts at 40 degrees Celsius for 4 hours. This is meant to reflect worst-case conditions in comparison to those encountered during use (assuming body temperature of 36 degrees Celsius and a minimum of 2 glove changes per 8-hour shift).
Results/discussion
The TMB real-time monitoring analyses produce a profile that includes an initial spike followed by a drop to steady state TMB levels for all products (see Figure 2). The real-time experiments indicate significant differences among products available for cleanroom use.
The maximum values for individual products ranged from 13 ppbv to 129 ppbv. While these maximum values are not in excess of that observed for humans (Kiskovich et al, 1999), gloves may represent a greater risk if operators are allowed to handle wafers at critical stages of the DUV photolithography process.
For many of the products tested, the initial spike returns to values at or below the baseline within 20 to 30 minutes (e.g. Figure 1 sample N-5). For others the steady state achieved indicates a continuous release of TMB to the atmosphere of 15 to 20 ppb (N-1 and N-4). This continuous release of
TMB represents an uncontrolled source of contamination for these specific products.
As NRL and nitrile gloves are manufactured with ammonia (or other amines) as a stabilizing agent, we investigated the water-extractable ammonia to determine whether there is a correlation to cumulative TMB. Figure 3 shows the cumulative TMB evolved plotted against the ammonium extracted from gloves using heated deionized water as the solvent. Most results fall near the theoretical 1:1 line indicating that NH3 may account for the bulk of the TMB evolved. For samples with highest TMB (N-1 and N-4) the NH4 falls below this line, suggesting that these products out-gassed basic molecules other than ammonia. Some manufacturers add amines or other topical "anti-stats" to enhance the electrostatic properties of their products for ESD-sensitive applications. Identification of the basic species awaits further work.
Summary
This study of evolution of airborne molecular base from gloves was conducted to evaluate whether gloves represent a risk to base-sensitive processes in photolithography and other areas of semiconductor manufacturing. The TMB varied among different products.
Some products showed continuous release of basic species for periods in excess of two hours. As ammonium hydroxide is often used in the production of latex gloves. The TMB results were compared to ammonium contents determined via water-extraction. For many of the products, ammonia can account for the bulk of the TMB released. Some products show deficits of NH4 relative to TMB suggesting another basic additive to these products.
 |
Figure 3: Comparison of the Cumulative TMB from the experiments (conducted by Extraction Systems Inc.) to the NH4+ extracted in DI water at 40°C for 4 hours (conducted by Kimberly Clark/Safeskin R&D). Note the deficit in NH4 relative to TMB for samples N-1 and N-4. This indicates that these products evolve unidentified basic species in addition to ammonia.
The maximum TMB concentration for every product analyzed is near or above the 20 ppb threshold observed for the onset of line defects (see table, p. 25). This maximum is a transient peak and may reflect an initial off-gassing event during experimental equilibration. The steady state value for TMB also varies widely. Several products exhibited off- gassing even after two hours—suggesting that these products are a continuous source of contamination during use (see Figure 2).
The products evaluated in this study appear to emit TMB in the range observed for operators and other items present in cleanroom environments. Thus, common measures used to minimize risk (airflow exchange, handling precautions and process isolation) should minimize impact of glove use in the cleanroom.
Overall, the results suggest that evaluation of TMB and/or NH4 should be conducted when evaluating gloves for use in base-sensitive applications. There are clear differences among the products tested here—indicating that raw material and processing could impact the final performance of the finished glove. Analysis of ammonium in water extracts could be used as a surrogate for TMB measurements if the correspondence between TMB and NH4 is demonstrated. Alternative glove materials that do not use ammonia or other volatile basic additives could present a solution if other concerns such as particles and non-volatile residues (NVR) are acceptable to the user.
Andrew J. Magenheim, Ph.D., conducted this research while a senior scientist at Kimberly-Clark Scientific and Industrial Business, where his responsibilities included providing critical evaluation of cleanroom glove contamination performance.
Mike Hansen works for Abbie Gregg Inc. (Tempe, AZ), an engineering and consulting concern that provides services for the semiconductor, flat panel display and other microelectronics industries.
References
- McDonald et al. (1993). Airborne chemical contamination of Chemically Amplified Resist. 1. Identification of the problem. Chemical Materials 5, 348-356.
- Dean and Carpio (1994). Contamination of positive Deep-UV photoresists. OCG Microlithography Seminar, Interface '94. 199-212.
- Kischkovich et al. (1999) Real time methodologies for monitoring airborne molecular contamination in modern cleanroom environments. SPIE Microlithography '99.
- Kishkovick and Larson (2000) Amine control for DUV Lithography: Identifying Hidden Sources. SPIE Microlithography 2000.
- Kiskovich, 1999.