 |
 |
 |
 |
|
Comments? Suggestions? Send us your feedback.
MicroMagazine.com |
CLEANROOM/FACILITIES TECHNOLOGIESEvaluating cleanroom wipers to establish performance benchmarksOsmond Atterbury, Himansu R. Bhattacharjee, Douglas W. Cooper, J. Ronald Dominique, Steven J. Paley, William R. Paley, and Howard Siegerman, Texwipe The term benchmark, meaning a surveyor's mark on a stationary object of known elevation that is used as a reference point in topographical surveys and tidal observations, has taken on the general meaning of something that serves as a standard of comparison. Thus, benchmarking in the microelectronics industry means ascertaining the best performance currently achievable by a type of product or process, against which one can compare existing capabilities and future advances. In the benchmarking study described in this article, product examples from the cleanest class of cleanroom wipers—sealed-edge laundered polyester knits—were evaluated on behalf of a major semiconductor manufacturer to determine the state of the art as well as to compare the wipers with each other. Identified in the article as products A, B, C, D, and E, the test wipers included other manufacturers' products, which were bought from distributors without any indication given that they would be part of a benchmarking program, and Texwipe products that were secured from regular manufacturing runs. The five kinds of wipers were examined for the following characteristics:
Experimental Methods and Results A total of 10 tests were performed on each product type. To examine lot-to-lot consistency, two or more bags from each of four different manufacturing lots were obtained for each product, and, to examine consistency within a given manufacturing lot, wiper samples were taken from each of two different bags. Thus, eight samples of each product were evaluated in each type of test, except for the abrasion, ion, and some NVR tests. The resulting data were analyzed by standard statistical tests for mean, standard deviation, standard error, and relative standard deviation. Tests of statistical significance also were applied.
Releasable Particles. In the test for releasable particles, the test wiper was immersed in a tray containing DI water and a small amount of surfactant, which simulated the low-surface-tension liquids often used in semiconductor manufacturing. The tray was then agitated using an orbital shaker, and particles released into the liquid were captured on a polycarbonate etched-pore (0.4-µm) filter and counted using a scanning electron microscope. Figure 1 plots the levels of releasable particles 0.5 µm or larger per square meter of fabric as the means and the means ± 1 standard error. The data on which this and most subsequent figures are based are summarized in Table I. As Figure 1 indicates, products A, B, and C exhibited fewer releasable particles than products D and E. The mean for E—129.59 particles/m2—was about a factor of five larger than the respective means for products A, B, and C, which were 25.90, 29.56, and 25.24 particles/m2, respectively.
Table I: Summary of test results. The weighted sums of normalized values at the bottom were calculated as follows: Particles + Fibers + Abrasion + (NVR—DI water + NVR—IPA)/2 + (Na+ + K+ + Cl—)/3. Releasable Fibers. To measure releasable fibers larger than 100 µm, the wipers were placed individually in a 2-L Erlenmeyer flask filled with DI water, and agitated with a biaxial shaker. The fibers released into the water were then captured on a membrane filter having 0.45-µm pores and counted using an optical microscope with image analysis software. As the results in Figure 2 show, product B was the cleanest, releasing about 600 fibers per square meter, with products A, C, and D losing about twice as many fibers and product E about four times as many.
Abrasion Resistance. The test used to measure abrasion resistance involved rubbing the test wiper against a wire mesh screen using a specified force, stroke length, and speed while air was being drawn through the mesh.1 The particles released by this process were counted by an optical particle counter using a 1-minute counting cycle. For this test only two samples from one lot of each wiper type were evaluated, for a total of 10 samples. As shown in Figure 3, which is a plot of the particle release rate, products A and B released about 1100 >=0.5 µm particles/min while product E produced about twice that amount and products C and D released nearly 4000 particles/min.
Nonvolatile Residues. To test for NVR in DI water, eight wipers of each type were boiled in water for 5 minutes. After boiling, the liquid was filtered, the filtrate was evaporated to dryness, and the residue weighed. The results were normalized by dividing by the area of the wipers (in square meters). As seen in Figure 4, all five products had mean values of less than 0.1 g/m2, and four of the five had mean values below 0.05 g/m2.
The test procedure for determining NVR in IPA was similar to that used with DI water. Four wipers of each type were boiled in electronics-grade isopropyl alcohol (IPA, 2-propanol) for 5 minutes, after which the solvent was filtered, the filtrate evaporated to dryness, and the residue weighed. The results were again normalized by dividing by the wiper area. Figure 5 shows that all five products had mean values of less than 0.15 g/m2, although these values were somewhat higher than the results for NVR in DI water.
Releasable Ions. In the test for releasable ions, 10 wipers of each type were soaked in 750 ml of hot DI water for 15 minutes. The water was then decanted and evaporated to 50 ml, after which the ion types and concentrations were determined by capillary ion electrophoresis, and these values were normalized by dividing by wiper weight. The means for sodium ions ranged from 0.211 to 0.539 ppm, the means for potassium ions ranged from 0.052 to 0.094 ppm, and the means for chloride ions ranged from 0.228 to 0.435 ppm (see Table I). Absorbent Capacity. To measure absorbent capacity, the test wiper was weighed, immersed in DI water until thoroughly wet, then removed and allowed to drip for 60 seconds before being reweighed. Figure 6 plots the absorbent capacity (wet weight minus dry weight) for the five products in milliliters per square meter. With similar mean values of about 450 ml/m2, products B, C, and E had the highest absorbencies, while product D exhibited an absorbency of about 90% of those values. Product A was least absorbent, with a mean value of about 250 ml/m2.
Absorbency Time. To test for absorbency time, an aliquot of DI water was placed on the test wiper and the time it took for the specular reflection to disappear was recorded. Products A, B, C, and E had mean absorbency times near 3 seconds; product D had a mean of 7 seconds. Statistical Analyses As previously mentioned, the data from each type of test were analyzed using various statistical techniques, primarily the Student's t-test for comparing means and the F test for comparing variances (standard deviation squared). A difference was considered statistically significant if a difference that large or larger would occur in less than 5% of the instances when sampling from identical normal distributions. The highlights brought out by these analyses included the following comparisons.
Contamination Indexes An index number is typically a ratio derived from a series of observations and used as a measure or indicator. For example, the consumer price index is the ratio of a measure of current prices to those of a base period. To look for patterns among the products in this study and to have a common measure, we formed normalized contamination levels by taking the ratio of each of the products' means from a particular test to the most desirable mean value obtained, so that the best performing product in a particular test had a normalized level of 1. Figure 7 shows the normalized levels by test type, thereby summarizing the comparative information learned from these tests. It can be easily seen, for example, that except for product E, the values for releasable particles are quite close to each other; the NVR—IPA values are very similar; and so forth. Where the spread in values for a test is large, substantial improvement is still possible for the products that performed less well.
Another kind of comparative index can be constructed by giving various weightings to the outcomes. One approach is to add the mean and the standard deviation, essentially weighting each equally, which produces what would be the 84th-percentile value if the distribution were normal. Those sums can be compared, or they can be combined across contaminant categories after normalization, and then themselves be normalized by dividing by the lowest sum in each category, so that the resulting values range up from 1.00. Even the categories can be combined by giving each a weighting. Table I lists the means, standard deviations, and normalized values for all test types except absorbent capacity and absorbency time. At the bottom of the table is one possible index, in which the particle and fiber values were added separately, the two NVR values were added after being averaged, and the three ion values were added after being averaged. On this scale, the order for the wiper types becomes A, B, C, D, E, from best performing, or cleanest, to least clean. Conclusion Ten tests that measured various types of released contaminants, absorbent capacity, and absorbency time were used to compare samples of five different sealed-edge knitted polyester cleanroom wiper products. Substantial and statistically significant differences were found in many of the comparisons made between the resulting test data. The best values obtained for each test indicate the current state of the art in cleanroom wipers and can serve as benchmarks for measuring future progress. Reference 1. Atterbury O, Bhattacharjee HR, Cooper DW, et al., "Comparing Cleanroom Wipers with a Dry Abrasion Resistance Test," MICRO, 15(9):83—100, 1997. Osmond Atterbury is an associate research chemist at Texwipe (Upper Saddle River, NJ) where he is involved in developing and applying advanced measurement techniques for testing cleanroom materials. He has a BS in chemistry and zoology from the University of the West Indies (Kingston, Jamaica) and has written numerous technical reports. Himansu R. Bhattacharjee, PhD, is laboratory director and senior scientist at Texwipe, with responsibility for technical research and new product development. He received his PhD in physical chemistry from Wayne State University and has more than a dozen patents and two dozen peer-reviewed publications. (Bhattacharjee can be reached at 201/327-9100, ext. 270.) Douglas W. Cooper, PhD, is director of contamination control at Texwipe and is involved in R&D for advanced cleaning materials. The recipient of a PhD in applied physics from Harvard University, he has written or cowritten more than a hundred peer-reviewed publications. Cooper was elected a fellow of the Institute of Environmental Sciences and Technology in 1995. (Cooper can be reached at 201/327-9100, ext. 397.) J. Ronald Dominique is an associate research chemist at Texwipe, responsible for R&D in advanced cleaning products. He received his MS in environmental engineering from the New Jersey Institute of Technology. Steven J. Paley is president of Texwipe and oversees all of the company's technical research, engineering, and new process development. He was awarded his MS in mechanical engineering product design from Stanford University. The author of numerous technical publications, Paley holds five patents with several others pending. William R. Paley is executive vice president of Texwipe, where he is responsible for North American marketing and sales. He holds an MBA from Columbia University. Howard Siegerman, PhD, is a director of marketing at Texwipe, specializing in advanced cleaning materials for microelectronics cleanrooms. He received his PhD in analytical chemistry from the University of Toronto and has published articles and patents related to contamination control.
MicroHome |
Search | Current Issue | MicroArchives Questions/comments about MICRO Magazine? E-mail us at cheynman@gmail.com. © 2007 Tom Cheyney |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Figure 1: Results of the releasable particle test for the five types of wipers.
Figure 2: Results of the releasable fiber test for the five types of wipers.
Figure 3: Results of the abrasion resistance test for the five types of wipers.
Figure 4: Results of the test for NVR in DI water for the five types of wipers.
Figure 5: Results of the test for NVR in IPA for the five types of wipers.
Figure 6: Results of the absorbent capacity test for the five types of wipers.
Figure 7: Normalized contamination levels for the five types of wipers from the 10 tests. These comparative results were obtained by taking the ratio of the mean for each product to that of the best performing product, so that the minimum mean equals one.