Study of Residual Particle Concentrations Generated by the Ultrasonic

Jul 30, 2003 - storage container walls may lead to increased dissolved impurities ... and polymer materials.11 In general, container-dependent solvent...
0 downloads 0 Views 115KB Size
Anal. Chem. 2003, 75, 4486-4492

Study of Residual Particle Concentrations Generated by the Ultrasonic Nebulization of Deionized Water Stored in Different Container Types Mark Knight and Giuseppe A. Petrucci*

Department of Chemistry, University of Vermont, Burlington, Vermont 05405-0125

A scanning mobility particle sizer has been used to quantify residual particle number and mass concentrations generated by ultrasonic nebulization of deionized (DI) water stored in a variety of bottles. High variability of residual particles was found not only between different bottle types but also between different bottles of the same type. Degradation of the water quality, quantified as increased residual mass and number concentrations as a function of time, occurred to varying degrees for water stored in different bottle types. Overall, glass bottles showed the highest residual particle concentrations and exhibited the poorest stability over time. After a storage period of 3 weeks, DI water stored in Pyrex bottles showed average increases in particle mass and number densities in the aerosol of over 250% and 60%, respectively. Total dissolved impurity levels in the water increased from 110 to 290 ng mL-1 over the 3-week period. It is hypothesized that leaching from the bottle walls increases impurity levels in the water over time. Leaching was observed for both glass and polymer bottles. Contrary to this trend, residual particle concentrations from deionized water stored in Teflon bottles showed a net decrease during the measurement period. With respect to absolute residual particle concentrations and storage stability, a Teflon bottle yielded the best performance. Total residual particle mass and number densities for Teflon were less than a factor of 15% and 1%, respectively, as compared to residual particle levels observed for the Pyrex bottle. Absolute dissolved impurity levels in the water for the Teflon bottle decreased from 7.8 to 3.7 ng mL-1 over the 4-week period. The choice of sampling and storage containers can be a primary decision in many applications. Containers present one of the earliest possibilities of contamination and can have major consequences on the application results. Solution nebulization is a widespread method of converting liquid samples to aerosols for subsequent analysis or processing. It has previously been shown that nebulization of pure solvents leads to the generation of high number densities of ultrafine particles, with aerodynamic diam* To whom correspondence should be addressed: (e-mail) Giuseppe. [email protected].

4486 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

eters less than 100 nm.1-4 Residual particle formation has been observed from the nebulization of water5,6 and several organic solvents, including acetone,7 isopropyl alcohol,8 and methanol.9 The origin of residual particles has been attributed, in most cases, to impurities native to the aqueous or organic solvent; for example, see Blackford et al.5 When a solution is nebulized, droplets are formed and dispersed in a gas. As described by eq 1, upon evaporation of the

d3 ) (Cv/F)D3

(1)

solvent, each droplet leaves behind a residual solid particle, regardless of solvent purity. The residual particle diameter, d, depends on the total dissolved impurity concentration, Cv, in the solvent and the solvated droplet diameter, D. The density (g cm-3) of the residual particle core is given by F. The manner of nebulization and quality of the solvent will have an impact on the number and size of residual particles present after evaporation. As the impurity level increases, the mode of the residual particle size distribution will shift to higher diameters as governed by eq 1. Ho et al.4 and Wen et al.8 have suggested that leaching from storage container walls may lead to increased dissolved impurities in the solvent, which in turn lead to higher residual particle mass and number concentrations upon nebulization. Reports have also appeared in the literature on the effect of solutions of varying composition and pH on leaching of specific elements from glass10 and polymer materials.11 In general, container-dependent solvent impurities are not well understood but remain a relevant source of variable background for many applications. (1) LaFranchi, B. W.; Knight, M.; Petrucci, G. A. J. Aerosol Sci. In press. (2) Krames, J. H.; Buttner, F. E. J. Aerosol Sci. 1991, 22, s15. (3) Kaye, B. H. Kona 1997, 15, 68. (4) Ho, J.; Kournikakis, B.; Gunning, A.; Fildes, J. J. Aerosol Sci. 1988, 19, 1425-1428. (5) Blackford, D. B.; Kerrick, T. A.; Schuermann, G. Ultrapure Water 1994, 11, 57-63. (6) Blackford, D. B. J. Process Anal. Chem. 1999, 4, 92-98. (7) Kinney, P. D.; Pui, D. Y. H.; Liu, B. Y. H.; Kerrick, T. A.; Blackford, D. B. J. Inst. Environ. Sci. 1995, 38, 27-35. (8) Wen, H. Y.; Kasper, G.; Chesters, S. Microcontamination 1986, 43, 3368. (9) Allen, L. B.; Koropchak, J. A.; Szostek, B. Anal. Chem. 1995, 67, 659-666. (10) Scholze, H. J. Non-Cryst. Solids 1988, 102, 1-10. (11) Moody, J. R.; Lindstrom, R. M. Anal. Chem. 1977, 49, 2264-2267. 10.1021/ac034355n CCC: $25.00

© 2003 American Chemical Society Published on Web 07/30/2003

Residual particles may present a limitation to the use of solution nebulization for aerosol generation in many areas, for example, the semiconductor and coating industries, and the health professions. Our particular interest concerns the use of ultrasonic nebulization for the generation of reference particles for calibration of aerosol mass spectrometers. Aerosol mass spectrometry,12 presently under development in many laboratories, is an emerging method for single-particle analysis that is sensitive not only to particle mass but also to the number of particles present in the sample aerosol. Aerosol mass spectrometers are often calibrated with monodisperse aerosols generated by nebulization of dilute solutions of polystyrene spheres with a narrow, well-characterized diameter range. In general, solution nebulization is a reliable method for generating calibration particles with diameters greater than 100 nm. Below this size, we have found residual particles make a significant contribution to the reference particle numbers. The presence of residual particles is of greatest concern in cases where the aerosol sample to be analyzed is dilute. As an example, Phares et al.13 described difficulties associated with residual particles in applying the ART-2a particle classification algorithm to aerosol mass spectrometry of laboratory-generated particle standards. Approaches described in the literature, such as diffusion filtering,9,14,15 used to selectively reduce residual particle numbers cannot be used for reference particles below 100 nm because diffusion filtering cannot differentiate between residual and sample particles of the same diameter. Because residual particles cannot be removed, they can become a significant source of background. Therefore, it is important to understand the origin and behavior of residual ultrafine particles. A scanning mobility particle sizer (SMPS) classifies particles in terms of aerodynamic diameter and counts them with a condensation particle counter. The output of the SMPS is a particle size distribution of particle number density as a function of particle aerodynamic diameter. Different weighted distributions, for example, surface area and mass, can be calculated directly from the number distribution. A detailed description of SMPS theory and operation is beyond the scope of this paper. The interested reader is referred to Kinney and Pui16 and Mertes et al.17 for an in-depth discussion of the instrumental components of the SMPS. As a first approximation, a SMPS classifies all particles within a set diameter range, without regard for chemical composition. In this regard, SMPS systems are useful to measure the total number and mass concentrations of residual particles generated by solution nebulization. A SMPS can be used effectively to evaluate different solvents and nebulization systems with respect to purity and suitability for specific applications. Residual particle monitors based on the SMPS have already been proposed and adopted as an ASTM Standard Test Method for On-Line Measurement of Residue After Evaporation of High-Purity Water.18 (12) Suess, D. T.; Prather, K. A. Chem. Rev. 1999, 99, 3007-3035. (13) Phares, D. J.; Rhoads, K. P.; Wexler, A. S.; Kane, D. B.; Johnston, M. V. Anal. Chem. 2001, 73, 2338-2344. (14) Cheng, Y. S.; Yeh, H. C. J. Aerosol Sci. 1980, 11, 313-320. (15) Sadain, S. K.; Koropchak, J. A. J. Liq. Chromatogr. Relat. Technol. 1999, 22, 799-811. (16) Kinney, P. D.; Pui, D. Y. H. TSI Tech. Pap. 1991, A74, 147-176. (17) Mertes, S.; Schroeder, F.; Wiedensohler, A. Aerosol Sci. Technol. 1995, 23, 257-261. (18) ASTM. ASTM Designation: D 5544-94 (Reapproved 1999) 1999, 728733.

Because residual background particle concentrations depend on the choice of sampling and storage containers, one must carefully consider the choice of a container that is best suited for the application at hand. Considerations such as the nature and length of sample storage, analytical method, and accuracy, container cost, and availability can play a significant role in this choice. The work presented here is the first of its kind in that it is an in-depth study of the magnitudes and variability of residual particle concentrations generated by nebulization of deionized water stored in bottles made of different materials. A preliminary study is also presented that supports the hypothesis of leaching as the principal mechanism contributing to increases over time in total dissolved impurities in the water. Overall particle mass and number densities of residual particles are measured and compared for a variety of bottle materials. EXPERIMENTAL SECTION All residual particle distributions were measured with a scanning mobility particle sizer (model SMPS 3936, TSI, Inc., St. Paul, MN) equipped with proprietary software provided from the manufacturer. The SMPS consists of a differential mobility analyzer (DMA model 3080) and condensation particle counter (CPC model 3010). The SMPS was operated using a sample flow rate of 0.400 L min-1 and a sheath flow rate of 4.0 L min-1, resulting in a measurement range of particle aerodynamic diameters of 12.4-562 nm. Carrier gas flows were filtered on-line to remove all particles producing zero air and verified using the SMPS. Aerosol generation was accomplished with either an ultrasonic nebulizer (model UT5000+, CETAC Technologies, Omaha, NE) or a concentric pneumatic nebulizer (Meinhard Glass Products, Golden, CO). The pneumatic nebulizer was equipped with an external diffusion dryer. The ultrasonic nebulizer was equipped with a desolvation stage that was used at all times. In all cases, a peristaltic pump (model Minipuls 3, Gilson, Inc., Middleton, WI) was used to force feed solution samples to the nebulizer at a rate of 0.833 mL min-1. The nebulizer-SMPS system required ∼20 min to stabilize after changing samples. All data presented are the average (( one standard deviation) of eight consecutive size distributions recorded after the stabilization period. Two identical sets of different bottle types were run on consecutive days for each measurement time point over a total of 4 weeks. Particle size distribution measurements were conducted immediately after filling the bottles and again after storage periods of 7, 14, and 31 days. Three independent 4-week experiments were conducted over the span of 18 months. All 12 bottle types (see Table 1) were either purchased from Fisher Scientific or provided directly from the manufacturer. Bottle volumes ranged from 100 to 250 mL. Immediately prior to filling, each new bottle was rinsed seven times with the same stock deionized water used to fill the bottles. Deionized water (DI) was generated with a Barnstead Water Nanopure System (model D1794, Barnstead International, Dubuque, IA). Different DI water stocks were used for each of the three 4-week experiments. All polymer bottles were Nalgene Labware (Nalg Nunc International, Rochester, NY). Pyrex is a registered trademark of Corning Inc. (Corning Inc., Corning, NY). Kimax is a registered trademark of Kimble Glass (Kimble Kontes, Vineland, NJ). These bottles represent a wide cross section of different materials that might Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

4487

Table 1. Summary of Bottle Types Used, Absolute Residual Particle Densities, Changes in Said Densities over a 4-Week Measurement Period, and Overall Bottle Performance

type

material

initial number densitya (×104 cm-3)

Pyrex Kimax HDPE LDPE PMP F-HDPE Op-Tef FEP PFA PP PC LP-HDPE

glass glass high-density polyethylene low-density polyethylene poly(methylpentene) Fluorinated HDPE opaque Teflon Teflon Teflon polypropylene polycarbonate low-particle HDPE

18 ((0.6) 16 ((1) 8.8 ((0.4) 9.7 ((0.2) 14.9 ((0.2) 7.66 ((0.06) 6 ((1) 5.0 ((0.2) 8.0 ((0.5) 11.0 ((0.5) 12 ((1) 9.18 ((0.09)

initial mass densitya (pg cm-3)

total dissolved impuritiesb (ng cm-3)

number stabilityc (%)

mass stabilityc (%)

overall performanced

45 ((5) 30 ((6) 7.0 ((0.7) 8.8 ((0.3) 45 ((3) 7.3 ((0.2) 3.1 ((0.9) 2.1 ((0.1) 4.8 ((0.4) 8 ((1) 12 ((3) 7.8 ((0.1)

112 ((12) 76 ((14) 17 ((2) 22.2 ((0.8) 113 ((7) 18.4 ((0.4) 7 ((2) 5.2 ((0.3) 12 ((1) 21 ((3) 31 ((7) 19.7 ((0.3)

+66 ((2) +35 ((3) -16 ((1) -23 ((1) -12 ((1) +56 ((1) -46 ((9) -7 ((1) -12 ((1) +16 ((1) +6 ((1) -23 ((1)

+259 ((27) +193 ((36) -13 ((1) -30 ((1) -42 ((4) +149 ((4) -52 ((15) +15 ((1) -16 ((2) +58 ((11) +16 ((4) -18 ((1)

--+++ + + ++ +++ +++ + +

a Total residual particle number and mass densities in the aerosol recorded by ultrasonic nebulization of deionized water immediately after filling the bottles. b Total impurity concentrations referenced to the solvent. Calculated according to eq 2. c Percent change in total number and mass densities in the aerosol over the 4-week measurement period. d Overall evaluation of bottle performance, including absolute magnitudes of residual particle densities, stability of residual particle densities, bottle cost, and availability; +++ excellent, + good, - poor, and - - very poor.

be employed for a variety of applications. Bottles were not dried prior to analysis and were stored at room temperature. No periodic agitation or stirring of solution was conducted during the analysis. All bottles were sealed during storage, sampling, and measurement with the exception of a small hole in the bottle cap to introduce the sampling tube. Since the Pyrex and Kimax bottles used contained only 100 mL of solution, a fourth measurement day (day 31) was not possible. RESULTS AND DISCUSSION When considering the importance of residual particles, it is instructive to differentiate between two analytical paradigms: analysis of particles with respect to mass and also with respect to number. If one follows the traditional mass paradigm, as is encountered in ultratrace methods of analysis, the measure of importance is the total analyte mass in the residual particles. For example, in the case of atomic spectroscopic methods, one is concerned with measuring a specific or limited set of elements. Residual particles that do not contain these elements do not contribute to the background. In this respect, the scientific community has compiled a largely empirical database on effective cleaning procedures to stabilize solutions of trace elements and minimize analyte-specific background.19,20 Referring to Figure 1, the percentage mass contribution of total residual particles from DI water to the NaCl analyte mass is negligible (∼1%, even if the entire residual particle mass were NaCl) as compared to the number contribution (∼9%). If, on the other hand, an analytical method is subject to a number paradigm, residual particles can make a significant background contribution. On average, due to the small diameters of residual particles, it is found that they make a much larger contribution to the total particle number density than to the total mass density. The importance of residual particles as a source of background will (19) O’Haver, T. C. In Trace Analysis: Spectroscopic Methods for Elements; Winefordner, J. D., Ed.; Wiley: New York, 1976; Vol. 46, pp 63-75. (20) Thiers, R. E. Contamination in Trace Element Analysis and Its Control; Interscience: New York, 1957.

4488 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

Figure 1. (a) Number and (b) mass particle size distributions recorded for the ultrasonic nebulization of (s) deionized water, (- ) distilled water, (‚‚‚) tap water, and (- ‚ -) 5 µg mL-1 NaCl. Mass values for the water samples were calculated based on an average particle density of 1.2 g cm-3. Mass values for the NaCl aerosol were calculated based on a particle density of 2.17 g cm-3.

depend on whether one is operating under the number or mass paradigm. In general, residual particles will have the greatest bearing on methods responsive to the number of particles in the aerosol regardless of chemical identity and mass. In light of this distinction, total mass and number densities of residual particles are treated separately in the ensuing discussion because conclusions differ whether one is considering the number or mass paradigm.

The discussion that follows is divided into three main sections: total impurity in the solvent; residual particle number density; and residual particle mass density. Bottle performance is evaluated in terms of absolute residual particle numbers and masses, variability of the residual particle background for different bottle types, variability of residual particle concentrations for analogous bottle types, and variability of residual particle concentrations over time. The data presented are for one of the 4-week runs. Averaging of data sets from the three independent 4-week experiments is not possible due to the high degree of variability in absolute magnitudes of residual particle concentrations between the different DI water stocks used. Also the three full data sets were obtained over an 18-month period, with several experimental interruptions due to system up-grade and maintenance. Therefore, averaging the three data sets would not provide a meaningful measure of error between the three data sets. Nevertheless, the general trends and conclusions described for the data set reported in this paper are valid for all three sets of data. Errors in initial number and mass densities reported in columns 3 and 4 of Table 1 reflect one standard deviation of eight consecutive measurements from the same bottle (after system stabilization) not the standard deviation between different bottles of the same type. This is the instrumental error associated with the nebulizer-SMPS system. The error in total dissolved impurity levels (column 5) is propagated from errors in initial mass density (column 4). Errors reported in the number and mass density stability (columns 6 and 7) are the result of propagation of error from the standard deviations reported in initial densities. Owing to the small data set for each type of container material, a statistical analysis was not possible for comparison of bottles of the same type. Before detailed data analysis, it is important to put the absolute magnitudes of residual particle concentrations in perspective. Panels a and b in Figure 1 compare particle number and mass distributions, respectively, obtained from ultrasonic nebulization of deionized water, distilled water, tap water, and a 5 µg mL-1 NaCl solution. The size distribution plots show the total number or mass of particles in the aerosol as a function of particle diameter. All particle size distributions are automatically fitted to a log-normal distribution by the SMPS software. The number densities and geometric mean diameter of residual particles from tap water are much higher relative to deionized water, distilled water, and 5 µg mL-1 NaCl solution owing to higher concentrations of dissolved salts in the tap water. The increase in the geometric mean of the distributions is more striking in the mass distribution shown in Figure 1b. The deionized water, due to the lowest concentration of impurities, exhibits the smallest massweighted geometric mean, 78 nm. The geometric mean increases to 135 nm for the distilled water sample and 228 nm for the tap water. The high number densities of residual particles constitute a large and variable background. Even deionized water produced more than 105 particles cm-3. This residual particle background, especially the variability, may limit solution nebulization as a feasible means of generating reference and sample aerosols of ultrafine particles. From the data shown in Figure 1, it is calculated that the contribution of residual particle mass from DI water to

the mass of NaCl particles is less than 1%; however, the contribution to particle number is greater than 9%. Within any 4-week experiment, daily changes in total residual particle number and mass densities between samples were as small as 1% and 3%, respectively. To verify the stability of the measurement system over the 4-week period, number and mass density measurements were compared with and without normalization to the particle size distribution obtained by nebulization of a 5 µg mL-1 NaCl solution. This concentration of NaCl was chosen as the normalization reference for two reasons. First, to be effective for normalization, solvent residual particle concentrations must not make a significant contribution to the NaCl particle distribution. As seen in Figure 1 and discussed above, residual particle mass and number constitute approximately 1% and 9% of the NaCl particle number and mass, respectively. Therefore, anticipated variations in residual particles are not expected to significantly alter the NaCl particle distribution over time. Second, the NaCl concentration had to be sufficiently dilute so that the entire particle distribution was within the operating diameter range of our SMPS (12.4-562 nm). Also, to avoid excessively long cleanout times between analysis of the normalization solution and deionized water samples, the lowest concentration NaCl solution was used that satisfied the two criteria discussed above. Within the data set studied over a 4-week period, no difference was observed between the normalized and raw data. Consequently changes in measured residual particle number and mass densities will be the result of bottle material and not measurement system variability. Because the system is stable over the 4-week measurement period, absolute residual particle number and mass densities are reported to provide a semiquantitative measure of residual particle concentrations. The discussion that follows should not be taken as a quantitative description of residual particle effects for the different bottle types. The discrepancies in absolute magnitudes of residual particle concentrations between the different data sets taken over the span of 18 months allows one to draw only semiquantitative conclusions regarding these effects. The data presented is from one of the 4-week experiments. Nevertheless, the general conclusions drawn are representative of the effect of bottle type on solution storage stability and residual background. I. Total Solvent Impurities. Measured particle number and mass densities in the aerosol are dependent on several experimental parameters, including solution uptake rate, carrier gas flow rate, and nebulization efficiency. The discussion that follows is specific to the use of ultrasonic nebulization. Analogous trends were observed in our experiments using a Meinhard concentric pneumatic nebulizer followed by a diffusion dryer for desolvation. Absolute residual particle concentrations from the pneumatic nebulizer were a factor of ∼10 lower, in accordance with the lower nebulization efficiency of the pneumatic nebulizer.21 To remove these dependencies and provide a more intuitive comparison of bottle performance, eq 2 may be used to convert

ci ) (Qg/Qsn) dM

(2)

all measured residual mass concentrations in the aerosol to a total dissolved impurity concentration, ci, in the deionized water. (21) Boumans, P. W. J. M. In Chemical Analysis; Winefordner, P. J. E., Ed.; John Wiley & Sons: New York, 1987; Vol. 90.

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

4489

Figure 2. (a) Total residual particle mass density (M) in the aerosol; (b) total dissolved impurity concentrations (ci) in deionized water from different bottle types calculated according to eq 2; (c) total residual particle number density (N) in the aerosol.

In this expression, Qg and Qs are the carrier gas and liquid sample volumetric flow rates (cm3 min-1), respectively. n is the nebulizer efficiency and dM the total integrated mass concentration (µg cm-3) of residual particles in the aerosol. n can be calculated by nebulizing a solution of known concentration. For our experiments, Qg was 400 cm3 min-1 and Qs was 0.833 cm3 min-1. The average aerosol mass concentration obtained from 11 replicate measurements of a 5 µg mL-1 solution of NaCl was 1.54 ((0.03) × 10-3 µg cm-3. Substituting this value into eq 2 yields an n of 0.145 ((0.003), in agreement with reported values of nebulization efficiencies for ultrasonic nebulization.22 Figure 2a shows values of the integrated mass (pg cm-3 carrier gas) of residual particles contained in the aerosol generated by nebulization of deionized water stored in each bottle type. Figure 2b shows corresponding total dissolved impurity concentrations (µg cm-3 water) in the DI water for the bottle types tested. These values were measured by nebulizing the deionized water immediately after filling the rinsed bottles from a common stock solution. There is a great variability in total dissolved impurities between the different bottle types. While one can expect the deionized water to possess a background level of dissolved impurities, the large variation between the bottle types indicates that a large component of the impurities has another source. Total solvent impurities include both suspended particles, such as silica nanoparticles that are a product of the deionization process,5 and dissolved materials. Because residual particle size distributions encompass the size range of the silica nanoparticles, the silica particles are hypothesized to contribute to the residual (22) Tarr, M. A.; Guancxuan, Z.; Browner, R. F. Appl. Spectrosc. 1991, 45, 14241432.

4490 Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

particle background of deionized water. However, the silica nanoparticle contribution is expected to remain constant because the same stock deionized water solution was used for all experiments. Therefore, variability in residual particle concentrations is hypothesized to be due entirely to solvent-surface interactions, such as leaching and adsorption, and directly reflects the stability of the bottles for sampling and storage of aqueous solutions. The results of the present study are summarized in Table 1 and discussed below. Owing to the rather unconventional nature of methods sensitive to the number paradigm, each section of the ensuing discussion separates number and mass effects for clarity. As will be shown, the choice of sampling and storage containers depends on the operative paradigm. II. Residual Particle Number versus Mass Densities. a. Residual Particle Number Densities. Figure 2c shows the total residual particle number density (particles cm-3 of carrier gas) produced by nebulization of deionized water stored in different bottle types. Residual particle number densities resulting from dissolved impurities native to the deionized water showed a high variability between bottles made of different materials. A high degree of variability in total residual particle number concentrations was also measured from nebulization of DI water stored in analogous bottle types. Glass bottles produced the overall highest residual particle number densities ((13.4-19.6) × 104 cm-3). HDPE and LDPE bottles exhibited much lower residual particle number densities ((9.4-9.6) × 10-4 cm-3). Variability in number density from analogous bottle types was also lowest for the HDPE and LDPE bottles (∼25%). Teflon bottles resulted in the lowest absolute residual particle number densities ((5-8) × 104 cm-3). On the other hand, PMP bottles exhibited relatively high residual particle concentrations (15.6 × 104 cm-3) and variability (50%). Of the bottle types tested, F-HDPE resulted in the lowest absolute residual particle number density (4.0 × 104 cm-3); however, the variability was high (39%). The LP-HDPE bottle showed the best compromise performance between absolute number density and variability (9.4 ((11%) × 104 cm-3). b. Residual Particle Mass Densities. Similar trends were observed for residual particle mass densities recorded immediately after filling the bottles. From the results summarized in Table 1, we observe that glass bottles exhibited the highest residual particle mass densities and polymer-based bottles the lowest. The observed variability (between analogous bottle types) in residual particle mass densities was analogous to or smaller than that observed for number densities. This trend is in contrast to that observed when the variability in residual particle mass densities between different bottle types is compared. In this case, the variability in mass densities is considerably greater than that observed for residual particle number densities for the different bottle types. Recall that the particle mass density in the aerosol is not element or compound specific, but rather a total particle mass density assuming an average particle density of 1.2 g cm-3. If one is operating under a mass paradigm, this high background and its associated variability may or may not be significant depending on the chemical composition of the residual particles. It is clear from the data in Table 1 that, with the exception of PMP, glass bottles on average exhibit much higher residual particle mass densities (30-45 pg cm-3) when compared to polymeric bottles (7.0-12 pg cm-3). Teflon bottles exhibit the

Figure 3. (a) Total number and (b) mass densities measured for different bottle types over a 4-week period: (right to left hatch) day 1, left to right hatch) day 14, (cross hatch) 21, and (horizontal hatch) day 31.

lowest residual particle mass densities of all the bottles analyzed (2.1-4.8 pg cm-3). III. Temporal Degradation of Stored Solutions. Even though a bottle might exhibit relatively high initial residual particle concentrations, the stability of the bottle over time is potentially of greater concern. Reports have appeared in the literature implying that solvent leaching from bottle walls contributes to total solvent impurities.4,8 A significant temporal degradation of water quality, as indicated by increased number and mass concentrations of residual particles, was observed for some bottles over a 4-week period (Figure 3). The result obtained for the FEP bottle is anomalous in that a modest net decrease (7%) was observed in particle number density with a corresponding increase (15%) in residual particle mass. Upon closer inspection of the mass-weighted particle size distribution for this bottle on the last day of measurement (day 31), it was observed that the geometric mean of the distribution was shifted to much larger diameters than on previous days. None of the particle size distributions measured for other bottles immediately prior to and after this bottle exhibited this same shift. No hypothesis can be advanced at this time to explain this anomalous result. It is hypothesized that increased number and mass concentrations of residual particles generated by nebulization of pure water are a result of leaching from bottle walls. It should be reiterated that all bottles were kept sealed during periods of storage. The bottles were also sealed during sampling and measurement with the exception of a small hole in the cap to introduce the sampling tube. Work is presently underway, using electron microscopy and aerosol mass spectrometry, to determine the chemical composition

of the residual particles to gain a better understanding of leaching mechanisms operative for the different bottle types. Decreases in number and mass concentrations are hypothesized to be due to adsorption of dissolved impurities onto the bottle surface and, possibly, absorption into the bulk of the bottle wall. a. Particle Number Densities over Time. Over the 4-week measurement period, glass bottles showed the greatest absolute change (+66%) in residual particle number densities as a function of time. Interestingly, the F-HDPE bottle also showed a dramatic increase in total residual particle number (56%) density and mass (149%) density over time. This increase is in contradiction to the behavior of the Teflon bottles, which showed a net decrease in both particle number (12%-46%) and mass (16%-52%) over time. All other polymer-based bottles also showed a net decrease in residual particle densities over time. Pyrex and Kimax glasses showed poor stability over the entire 4-week measurement period. Relative to other bottles tested, glass also exhibited the highest variability in initial particle density levels. The high degree of variability implies that the stability of glass in water depends on the composition of the glass. The formation of a hydrated layer at the glass-water interface has already been reported in the literature.23 This layer can extend as much as 25 µm into the glass surface, providing conditions for an active physical and chemical exchange between the water and the glass. Compared to glass bottles, polymer bottles were more stable over time. Residual particle number densities for F-HDPE, LP-HDPE, LDPE, PC, and PP remained relatively constant over the 4-week measurement period. Opaque Teflon, FEP, PFA, and HDPE showed a net decrease over time. One apparent difference between the mass and number data in Figure 3 is the magnitude of change between consecutive mass and number analyses. There is a much greater change in the magnitude of residual particle mass data than for residual particle numbers. Intuitively, it is straightforward to understand how an increase in impurity level in the solvent can result in larger particles and thus an increase in mass. It is perhaps not as straightforward to explain the observed increase in particle numbers as a function of total dissolved impurities. On the basis of eq 1, a change in total dissolved impurities should not result in a change in particle number densities. Because each water droplet generated by the ultrasonic nebulizer results in a solid residual particle composed of a nonvolatile core, only a change in nebulization parameters could cause a change in residual particle number densities. One explanation for the observed increase in total particle number densities as a function of Cv may be due to a measurement artifact introduced by the condensation particle counter. The CPC has a low-diameter threshold, at ∼10 nm, below which particles are not counted. Small solvent droplets with diameter D1 may lead to residual particles with diameter d1 that is below the CPC counting threshold. According to eq 1, as Cv, increases the same solvent droplets, D1, will lead to residual particles with diameters that are above the CPC threshold and are therefore counted. In this manner, there is an apparent increase in total residual particle number density as a function of dissolved impurity concentration. Clearly, this is one possibility. The dynamics of nebulization and transport are more complicated than presented here. A more detailed study of the effect of solution (23) Sheng, J.; Luo, S.; Tang, B. Waste Manage. 1999, 19, 401-407.

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

4491

concentration on particle size distributions is needed to further elucidate this point. b. Particle Mass Densities over Time. Total residual particle mass densities are a better measure for comparison of solvent quality and stability than are number densities. This is due to a mode shift of the mass particle size distribution to higher diameters relative to the number size distribution. Hence, changes in total particle mass density are not significantly affected by changes in number densities at the lower diameters, which may in effect be an experimental artifact caused by the CPC. Glass bottles exhibited the largest increase (193%-259%) in residual particle mass of all the bottles tested. On average, residual particle mass densities of the Teflon-based bottles were twice as low in comparison to HDPE. This is not unexpected when one considers the excellent chemical resistivity and greatly reduced wettability of Teflon in comparison with other materials. LDPE, HDPE, LP-HDPE, PMP, Op-Tef, and PFA all showed a net decrease (13%-52%) in residual particle mass density. On the other hand, the F-HDPE (+149%), PP (+58%), and PC (+16%) bottles all showed an increase in residual particle mass. The PFA showed the highest stability over the whole measurement period with a slight reduction in both particle mass (16%) and number (12%) densities. Overall, Teflon bottles had the best performance in absolute magnitude of residual particle mass with PFA having the best stability of all bottles analyzed. If these bottles are to be evaluated on the basis of mass, then the higher cost of these specialized bottles may be justified. If, on the other hand, the type of analysis requires reduction or stabilization of particle number, then the enhancement seen for the specialized bottles is not significant and the added cost of the Teflon bottles may not be warranted. The HDPE bottles performed equally well when particle numbers are of concern and in some instances outperformed the Teflon bottles. It should also be kept in mind that the performance of a bottle depends strongly on its history of use. Increases in residual particle mass and number concentrations of over 100-fold were observed for used glass bottles even after they had undergone an extensive washing and pretreatment process with solutions of 1%-10% nitric acid. c. Specialized Storage Bottles. Nalgene offers several specialized bottle types for the storage of high-purity chemicals. Two of these bottles were investigated for residual particle formation and temporal degradation of stored deionized water. The first bottle was a low-particulate HDPE bottle designed for storing high-purity chemicals. This bottle has a purported average colloidal particle level of less than 30 particles cm-3 at diameters of 0.3 µm and greater. While the primary source of residual particles in the deionized water is dissolved impurities, suspension of colloidal particles present on the surface of these bottles into the water is a possible secondary source. The low-particulate HDPE bottle is designed to reduce colloidal particles. Residual particle number (9.18 × 104 cm-3) and mass (7.8 pg cm-3) were of the same order of magnitude as observed for the HDPE and LDPE bottles. Analogous to the HDPE and LDPE, a decrease in total residual particle number (23%) and mass (21%) was measured over a 4-week period. The second bottle examined was a Nalgene HDPE bottle with all surfaces fluorinated to improve material impermeability to aqueous solutions. Residual particle number (7.66 × 104 4492

Analytical Chemistry, Vol. 75, No. 17, September 1, 2003

cm-3) and mass (7.3 pg cm-3) densities were not significantly lower than residual particle densities measured for untreated HDPE bottles. In addition, contrary to all other polymeric materials, the fluorinated HDPE bottle showed a substantial increase in overall residual particle number (56%) and mass (149%) concentration over the 4-week period. CONCLUSIONS A scanning mobility particle sizer has been used to measure residual particle densities resulting from the nebulization of deionized water stored in bottles composed of different glass and polymer materials. The results, summarized in Table 1, show that glass bottles are the poorest performers in terms of both absolute magnitudes of residual particle densities and stability of the deionized water stored over a 4-week period. The sharp increases observed in residual particle densities are hypothesized to be due to leaching from the bottle walls. Teflon bottles exhibited the lowest and most stable residual particle densities, although, in comparison to the polyethylene bottles, the performance enhancement was modest. Most of the polymer bottles exhibited a net decrease in residual particle densities, most likely as a result of adsorption or absorption of dissolved impurities onto the bottle surface. Further work is underway to elucidate solvent-bottle exchange mechanisms operative for the different materials and to assess means of minimizing the exchange, thereby stabilizing and minimizing absolute residual particle densities. From the data in Table 1, the distinction is clear between the number and mass paradigms when the best sampling and storage container is selected. In all cases, however, Pyrex, Kimax, PMP, and PP bottles appear inadequate for applications where residual particles can impact the application at hand. For applications subject to the mass paradigm, FEP and PFA bottles are favorable because they provide a significant reduction in absolute residual particle mass densities. These bottles also showed the greatest stability in residual particle mass densities over time. Polyethylene bottles showed a factor of 2 increase in absolute residual particle mass densities; however, the residual particle mass decreased substantially as a function of time in the 4-week measurement period. It is hypothesized that an extended passivation period may reduce both the temporal variation and absolute magnitude of residual particle mass. Under the number paradigm, the distinction between the specialized Teflon bottles and the commonplace polyethylene bottles becomes less obvious. All exhibited low absolute residual particle number densities (less than 1 × 105 cm-3) and improved temporal stability, as compared to other bottle types. Again, a negative trend was measured in residual particle number densities, indicating that a passivation period may serve to further improve the performance of these bottles. ACKNOWLEDGMENT This work was funded in part by grants from the Vermont NSFEPSCoR Program.

Received for review April 7, 2003. Accepted May 27, 2003. AC034355N