 |
 |
 |
 |
|
Comments? Suggestions? Send us your feedback.
MicroMagazine.com |
Ultrapure Water AnalysisVerifying the accuracy of low-level TOC measurementsKaren L. Franklin, Slava Petropavlovskikh, and Stephen Poirier, Ionics Sievers An on-line analyzer can provide verifiably accurate measurements of TOC in ultrapure water in conditions of low dissolved oxygen or excess hydrogen.Managing the level of contaminants in ultrapure water has enabled semiconductor manufacturers to improve product quality and maximize yields. Advances in water purification technology and innovations in instrumentation for monitoring contaminants have played a synergistic role in improving and measuring the quality of ultrapure water. These advances, in turn, have spurred lower specifications for contaminant levels in ultrapure water. Total organic carbon (TOC) analysis has proven to be a powerful technique for monitoring organic contaminants in ultrapure water. Over the past few decades, specifications for TOC have decreased by several orders of magnitude, resulting in specifications in the low-parts-per-billion range. With future-generation chip technology projected to reach narrower linewidths, semiconductor manufacturers continue to demand that their ultrapure water contain ever-lower levels of organic contamination. Projected specifications for TOC are expected to decrease by another order of magnitude over the next 15-year horizon. This trend toward lower TOC specifications will drive innovations from instrumentation suppliers and suppliers of water-purification equipment. Developed in the mid-1980s, on-line TOC instrumentation allowed the semiconductor industry to monitor ultrapure water to determine contaminant level trends. While these on-line systems can detect contaminant limits well into the parts-per-trillion range, instrument accuracy cannot be verified at low-parts-per-billion levels. Without the ability to verify low-level instrument accuracy, the industry lacks confidence in ultrapure TOC measurement. Moreover, some semiconductor facilities have experienced unnecessary, costly shutdowns because of artificially high TOC readings attributed to interferences from older systems. Clearly, the older on-line TOC instruments do not address the industry's increasing need for accurate low-level TOC analysis. The issue of low-level measurement accuracy was addressed by Sievers Instruments (Boulder, CO), a subsidiary of Ionics. Featuring an automated, on-line standard addition mechanism, its Ultrapure PPT instrument enables users to verify the accuracy of TOC measurements on-line at the low levels present in ultrapure water. This verification technique is particularly useful for measuring TOC in conditions of low dissolved oxygen, which can compromise the accuracy of the measurements. In addition, the TOC instrument technology introduces a new parameter to monitor ultrapure water: TOx indication. This parameter monitors organic compounds containing halogens, nitrogen, sulfur, or phosphorus. Data from beta sites at semiconductor facilities have illustrated how TOx indication can reveal changes in the chemistry of ultrapure water systems. This article discusses the beta-site investigation of TOC measurement accuracy in ultrapure water systems that contain various levels of dissolved oxygen. The data from this investigation illustrate the utility of verifying TOC measurement accuracy in specific ultrapure water matrices. Instrument Accuracy The accuracy of analytical instrumentation is typically assessed by verifying instrument calibration. Calibration verification usually involves analyzing a known gravimetric standard and comparing the observed instrument response to the theoretical value of the standard. To maximize confidence in analytical accuracy, calibration verification is performed in the range of measurement. However, to verify instrument calibration at low levels of analyte, alternative techniques must be used. Because low-level standards are particularly susceptible to contamination, they cannot be prepared accurately. An analytical technique commonly used to verify calibration of instrumentation at low levels of analyte is the method of standard additions, which is performed by adding a small aliquot of a known, characterized standard to a sample matrix. Complete recovery of the standard in the sample matrix confirms the accuracy of measurement in the sample matrix. The method of standard additions has been applied previously to verify the accuracy of low-level TOC measurement.1
The TOC analyzer being investigated features an integrated, automated standard addition system that enables the on-site calibration of TOC and conductivity and the verification of TOC measurement accuracy. The standard addition system, depicted in Figure 1, consists of an integrated syringe assembly, which delivers small aliquots of a known, concentrated standard into the on-line water stream. An external port accommodates a 30-ml vial containing a concentrated TOC or conductivity standard. Through a series of automated, menu-driven prompts, users perform on-line TOC calibrations, conductivity calibrations, or TOC verification. TOC calibrations consist of multiple additions of a concentrated TOC standard, and conductivity calibrations consist of additions of a conductivity standard. A calibration curve is automatically calculated and the user is prompted to accept new calibration constants. Similarly, single-level calibration verifications are performed at user-selectable additions ranging from 1 to 50-ppb TOC. Prior to each standard addition, internal flow rates are automatically measured and used to calculate the expected TOC step change. The measured and expected step change values are displayed. Through this automated verification technique, users are able to verify calibrations on-site and on-line. In addition to instrument calibration, the incomplete recovery of organics, levels of low dissolved oxygen in the sample matrix, and false positives from particular contaminants in the sample matrix influence the accuracy of TOC measurement. Confidence in the accuracy of ultrapure water TOC measurement is achieved by verifying the accuracy of TOC measurement in any given sample matrix. The complete recovery of all organics is critical to accurate TOC measurement. TOC instrumentation must be able to fully oxidize all organic materials to carbon dioxide. Conditions of low dissolved oxygen or excess hydrogen can prohibit complete oxidation. The mechanism of UV-promoted oxidation relies on hydroxyl radicals (HO*), which oxidize organics to carbon dioxide. These hydroxyl radicals are generated from the irradiation of water with UV energy, as illustrated in the following reactions:
The irradiation of water molecules also produces hydrogen radicals (H*), which react with dissolved oxygen to form more hydroxyl radicals. In sample matrices with low dissolved oxygen, hydrogen radicals are not consumed, and as a reducing species these hydrogen radicals may prohibit the complete oxidation of organics. Thus, under low-dissolved-oxygen conditions, the accuracy of TOC measurement may be compromised. Previous studies, the results of which are illustrated in Figure 2, have demonstrated how low dissolved oxygen can cause low TOC responses.2
Because instrument technologies have different oxidation efficiencies, the extent to which low-dissolved-oxygen levels affect measurement accuracy varies according to the type of TOC instrument technology in use. Furthermore, recoveries depend not only on the level of dissolved oxygen but also on the concentrations and types of organic contaminants analyzed. Taking these factors into account, it is difficult to anticipate a priori the accuracy of TOC measurement based on the level of dissolved oxygen. Thus, it is imperative that users empirically determine TOC measurement accuracy in their specific sample matrices. The selectivity of the instrument detector can also affect TOC measurement accuracy. TOC detectors must be selective for carbon dioxide. To relate conductivity and TOC accurately, conductometric technology, developed nearly 20 years ago, assumes that all conductivity results from carbonic acid or hydronium and bicarbonate ions. Nonselective conductometric detection does not discriminate between the conductivity contributed by carbonic acid and other ions present in solution. Consequently, this detection method is susceptible to chemical interferences from TOx compounds, organics containing halogens, phosphorus, nitrogen, or sulfur. Upon oxidation, TOx compounds produce not only carbon dioxide but also acidic by-products, as shown in Figure 3. With nonselective detection, acidic by-products inflate conductivity and therefore produce artificially high TOC readings, or false positives.3 As illustrated in Table I, nonselective detection technology yields artificially high TOC readings in the presence of TOx compounds, greatly compromising the accuracy of TOC measurement.4
Table I: Nonselective detection technology yields artificially high TOC readings in the presence of TOx compounds, compromising TOC measurement. The TOC analyzer contains selective membrane conductometric detection technology that is not prone to interferences from TOx compounds.3,5 The oxidized sample passes through a membrane module prior to the conductivity measurement. The selectively permeable membrane module allows carbon dioxide to cross the membrane, and the conductivity of the resulting solution is measured, as illustrated in Figure 1. Ionic by-products from the oxidation of TOx compounds do not interfere with the conductivity measurement. As a result, the analyzer accurately measures the TOC levels of sample matrices containing TOx compounds. The TOC analyzer features TOx indication, which monitors the presence of TOx compounds. TOx indication is a nonspecific measurement that refers to the amount of organics containing halogens, nitrogen, phosphorus, or sulfur. As depicted in Figure 1, the analyzer derives TOx from the difference between the selective and nonselective conductivity measurements obtained from conductivity cells Nos. 3 and 2, respectively. This difference is attributable to ionic by-products from oxidation. To calculate the concentration of TOx from conductivity, the TOC analyzer assumes that the conductivity difference is a result of hydrogen chloride and expresses TOx in parts-per-billion carbon as chloroform. Monitoring TOx in addition to TOC helps reveal not only the changes in the amounts of organic contaminants but also the changes in the types of organic contaminants. Results With its integrated standard additions apparatus, the analyzer allows users to verify the accuracy of their TOC measurement. The utility of this accuracy verification was demonstrated at beta sites at two semiconductor facilities whose ultrapure water contained low levels of dissolved oxygen. At both facilities, TOC analyzers were installed in parallel with another analyzer (designated A) in the polish loop. Dissolved oxygen levels at semiconductor facility I were in the 0.1- to 0.3-ppb range while semiconductor facility II maintained dissolved oxygen levels of <10 ppb.
Table II: On-line data from beta sites at IC facilities with low dissolved oxygen. As demonstrated in Table II, on-line data from semiconductor facility I showed that the TOC analyzer's response was approximately 5.9 ppb, approximately 2.7 ppb higher than analyzer A's response, which was 3.2 ppb. Similarly, at semiconductor facility II the response of the TOC analyzer was approximately 2.6 ppb, approximately 1.4 ppb higher than analyzer A, which found 1.2 ppb. In both facilities, analyzer A showed a significantly lower response to TOC contaminant levels than the TOC analyzer. The lower response from analyzer A suggests that it may not completely recover organics in the low-dissolved-oxygen sample matrix.
Table III: Verification data from semiconductor facilities I and II. To assess whether the TOC analyzer fully recovers organics in the low-dissolved-oxygen sample matrix, verifications were performed, the results of which are presented in Table III. These verifications were performed by introducing a standard of 4 ppm carbon as sucrose and allowing the TOC analyzer to perform automated standard additions at approximately the 1- and 10-ppb TOC levels. Verifications were performed at 10 ppb (verification I) and 1 ppb (verification II). Two sets of verifications were performed at semiconductor facility I and one at semiconductor facility II. These verification results demonstrated that the TOC analyzer was indeed able to accurately recover organics in the low-dissolved-oxygen sample matrix. Previous studies showed analyzer A's low TOC response in conditions of low dissolved oxygen.2 These verifications also corroborate data from an external standard additions apparatus through which standard additions of sucrose, isopropanol, and urea were introduced to challenge the oxidative efficiencies of the TOC analyzer and analyzer A.2 Data generated from the beta sites also revealed that TOx indication is useful for managing ultrapure water systems. Figure 4 demonstrates that both the primary and the polish loops in one of the facilities showed increased TOC and TOx levels, which were a result of changing the feedwater. The similar profiles of these loops indicate that the increased concentration of TOx compounds from the feedwater was reflected in the final polish water despite the water system's various purification processes.
Notably, the response of analyzer A in the primary loop was higher than that of the TOC analyzer, as shown in Figure 3. The higher response may be attributable to significant levels of TOx in the water system. Upon oxidation with analyzer A, TOx compounds cause inflated TOC levels. Interestingly, analyzer A did not respond to the increase in TOx organics coming through the polish loop. This can be explained by certain TOx compounds present in the feedwater and the primary and polish loops that were not oxidized by analyzer A. This hypothesis is supported by studies demonstrating that analyzer A was unable to recover some TOx compounds such as urea.2 The intensity of the polish ultraviolet reactor at the other beta site appeared to decrease at a steeper slope than usual and then subsequently increased for a brief duration, as depicted in Figure 5. As UV intensity increased, TOC and TOx increases were observed downstream in the polish analytical loop. Furthermore, these increases in TOC and TOx occurred after a reverse osmosis membrane in the primary loop underwent a routine cleaning.
This ultraviolet intensity profile suggests that there was a buildup of UV-absorbing TOx compounds in the UV reactor. Because the sensor for ultraviolet intensity is located on the opposite side of the reactor from the UV source, a buildup of UV-absorbing compounds would apparently decrease the ultraviolet intensity. When TOx compounds were released downstream into the polish water, an event possibly catalyzed by residual chemicals used in the reverse osmosis cleaning process, ultraviolet intensity increased and caused increases in TOC and TOx downstream. In contrast, analyzer A, located parallel to the TOC analyzer, was unable to detect a discernible change in TOC. This lack of response suggests that analyzer A was unable to recover the TOx organics present in the sample matrix. The TOC analyzer was able to detect a real change in polish water TOC when routine maintenance on the water system was performed. Conclusion Featuring an automated, integrated standard addition system, the TOC analyzer provides the semiconductor industry with verifiably accurate TOC measurement. This accuracy is verified in the specific sample matrix, which enables users to assess the accuracy of their TOC measurement when low-dissolved-oxygen or excess hydrogen conditions are present. Performance data from beta sites at semiconductor facilities illustrate the utility of accuracy verification and demonstrate the analyzer's high sensitivity to real changes in TOC. Furthermore, the analyzer's TOx indication parameter monitors the concentration of TOx compounds, organics containing halogens, nitrogen, phosphorus, or sulfur. Monitoring changes in TOC and TOx provides users with more information about the chemistry of their water systems. Information on changes in the chemistry of water systems can be especially useful in managing the effects of regular maintenance procedures on ultrapure water quality. Ultimately, this can help improve ultrapure water quality itself. References 1. MJ Bollinger et al., "A Novel Approach to Verifying TOC Instrument Accuracy," in Proceedings of the Semiconductor Pure Water and Chemicals Conference (Sunnyvale, CA: Semiconductor Pure Water and Chemicals Conference, 1997), 97—110. 2. R Godec and K Franklin, "The Verification of Analytical Ultrapure Water Instrumentation Performance Using an Automated Standard Addition Apparatus," in Proceedings of the Semiconductor Pure Water and Chemicals Conference (Sunnyvale, CA: Semiconductor Pure Water and Chemicals Conference, 1999), 91—110. 3. RD Barley, RS Hutte, and K O'Neill, "Application of TOC Monitoring in Semiconductor Manufacturing," Ultrapure Water 11, no. 6 (1994): 20—24. 4. T Chu, "Trihalomethanes Can Cause RO/DI System Problems," in Proceedings of the Semiconductor Pure Water and Chemicals Conference (Sunnyvale, CA: Semiconductor Pure Water and Chemicals Conference, 1989), 229—255. 5. R Godec, K O'Neill, and R Hutte, "New Technology for TOC Analysis in Water," Ultrapure Water 9, no. 9 (1992): 17—22. Karen L. Franklin, PhD, is a product specialist for Sievers Instruments, a division of Ionics, in Boulder, CO. Her responsibilities include TOC product marketing and applications. Franklin is the author of many technical papers and presentations on TOC instrumentation. She is also the author and coauthor of technical papers on chemoenzymatic chemistry and on the design and synthesis of enzyme inhibitors. She received her PhD in organic chemistry from the Scripps Research Institute in La Jolla, CA. (Franklin can be reached at 303/444-4491, ext. 241, or kfranklin@sieversinst.com.) Slava Petropavlovskikh is TOC product line manager at Sievers Instruments. With more than 10 years experience in the analytical instrumentation field, he was a principal developer of the PPT TOC analyzer. He received an MS in applied physics from Moscow Physical Technical Institute and an MS in engineering mathematics from Moscow Military Academy. (Petropavlovskikh can be reached at 303/444-2009, ext. 167, or slava@sieversinst.com.)
Steve Poirier is vice president of strategic planning and business development at Sievers Instruments. He has served in a variety of roles in emerging companies involved with high-technology products and services marketed to the semiconductor and pharmaceutical industries. In 1979 he joined Balazs Analytical Laboratory serving as laboratory director and vice president. He was a founder of Anatel, a process instrumentation company, and DataTrax Systems, a data management software company, where he served as the company's president. He has a BS in chemistry and has performed graduate work in chemistry at Northeastern University in Boston. He also has an MBA from the University of Phoenix in Santa Clara, CA. (Poirier can be reached at 303/444-2009, ext. 211, or spoirier@sieversinst.com.)
MicroHome |
Search | Current Issue | MicroArchives Questions/comments about MICRO Magazine? E-mail us at cheynman@gmail.com. © 2007 Tom Cheyney |
