Article pubs.acs.org/ac
Real-Time Shape-Based Particle Separation and Detailed in Situ Particle Shape Characterization Josef Beranek,† Dan Imre,‡ and Alla Zelenyuk*,† †
Pacific Northwest National Laboratory, Richland, Washington 99354, United States Imre Consulting, Richland, Washington 99352, United States
‡
S Supporting Information *
ABSTRACT: Particle shape is an important attribute in determining particle properties and behavior, but it is difficult to control and characterize. We present a new portable system that offers, for the first time, the ability to separate particles with different shapes and characterize their chemical and physical properties, including their dynamic shape factors (DSFs) in the transition and free-molecular regimes, with high precision, in situ, and in realtime. The system uses an aerosol particle mass analyzer (APM) to classify particles of one mass-to-charge ratio, transporting them to a differential mobility analyzer (DMA) that is tuned to select particles of one charge, mobility diameter, and for particles with one density, one shape. These uniform particles are then ready for use and/or characterization by any application or analytical tool. We combine the APM and DMA with our single-particle mass spectrometer, SPLAT II, to form the ADS and demonstrate its utility to measure individual particle compositions, vacuum aerodynamic diameters, and particle DSFs in two flow regimes for each selected shape. We applied the ADS to the characterization of aspherical ammonium sulfate and NaCl particles, demonstrating that both have a wide distribution of particle shapes with DSFs from approximately 1 to 1.5.
P
also not an intrinsic particle property, but a measure of particle behavior that depends on pressure, flow regime, particle size, and particle orientation.4−7 At present, microscopy remains the most common particleshape characterization method. It is, however, an offline, laborintensive, and rather destructive analysis method that yields two-dimensional information on a small number of particles. Angular-resolved light scattering is another technique that has been used with limited success to identify particle asphericity during in situ sampling.8,9 Recent developments in aerosol mass spectrometry and in particular the development of our single particle mass spectrometer, SPLAT II (hereafter SPLAT),10,11 with its ultrahigh sensitivity and sizing precision, made it possible to combine it with a differential mobility analyzer (DMA) and conduct high precision measurements on mobility-selected particles. We, and others, have demonstrated that this approach yields real-time, in situ information on particle asphericity,12 fractal dimension,10,13 and average DSFs.4,5 Additionally, this experimental system measures particle size, composition, density, effective density (ρeff), morphology, etc.10 Using the combined DMA/SPLAT system, we measured average DSFs for ammonium sulfate and NaCl particles and showed that their shapes are size-dependent. These studies suggested that aerosol samples include a wide range of shapes, but due to the inherent limitations of the DMA/SPLAT system,
article shape plays a central role in determining particle properties and behavior, providing information about the particle phase and optical properties, in addition to affecting the particle transmission in inlets. As an example, the plasmon resonance properties of metal nanoparticles depend not only on whether individual particles are cubic or spherical but also on the sharpness of the cube’s edges.1 The inability to isolate particles with identical size and shape can render nanoparticle preparation methods impractical. A large portion of small particles in various environments are aspherical. Nevertheless, because of the difficulty in determining particle shape experimentally, data interpretation is commonly reliant on the assumption that the particles are spherical. This assumption introduces errors that are difficult to evaluate. For example, when converting measured particle diameters of aspherical particles such as rod-shaped gold nanoparticles2 or fractal soot particles3 into volumes, the assumption of sphericity leads to an overestimation of their volumes by a factor of more than 10. Particle shape also has a direct impact on the interpretation of commonly measured particle attributes. Unlike mass (mp), density (ρp), and volume equivalent diameters (dve), particle mobility (dm), aerodynamic (da), or vacuum aerodynamic (dva) diameters are not intrinsic particle properties but are instead equivalent diameters that describe particle behavior in different flow regimes.4,5 To account for the effect of the particle shape on behavior, a correction factor, termed dynamic shape factor (DSF or χ), is used. The DSF is the ratio of the resistance force on an aspherical particle to the resistance force on a sphere of equivalent volume moving with the same velocity.6 The DSF is © 2012 American Chemical Society
Received: October 2, 2011 Accepted: January 4, 2012 Published: January 4, 2012 1459
dx.doi.org/10.1021/ac202235z | Anal. Chem. 2012, 84, 1459−1465
Analytical Chemistry
Article
only a single, approximate DSF (χ)̅ could be obtained.5 Moreover, χ̅ is based on measurements conducted by the DMA at atmospheric pressure, in the transition regime, and by SPLAT, in the vacuum or free-molecular regime. Because the DSFs in the two regimes are unequal and generally could not be separately determined by the SPLAT/DMA system, we were forced to assume that the DSF is independent of flow-regime and calculated a single approximate DSF.4,5 We further demonstrated that the DMA/SPLAT system can be used for agglomerates of uniform spheres, whose dve is known by construction, to measure the DSFs of particles in the transition and free-molecular regimes.5 Here, we show that adding an aerosol particle mass analyzer (APM) to the DMA/SPLAT system made it possible to measure the DSFs of all particles in the transition and freemolecular regimes. An APM classifies particles based on their mass-to-charge ratio by balancing centrifugal and electrostatic forces14 to yield particle mass that together with density can be used to calculate particle dve, which makes it possible to characterize the DSFs in the transition and free-molecular regimes for all particles. Moreover, we demonstrate that the combined APM/DMA/ SPLAT system (ADS) makes it possible to separate, from a mixture, particles with the same mass, size, charge, and shape for use by any instrument. Typically, the APM is used to measure particle mass concentrations.15,16 When the APM is combined with the DMA, it measures particle density or effective density17−20 based on which particle fractal dimension21 and porosity can be calculated.20 In all but one of the studies that utilized the DMA/APM system, particles were first classified by the DMA, and then their mass distributions were measured by the APM, operating in a stepping mode that takes ∼35 min/scan. Malloy et al.18 demonstrated it is possible to significantly improve temporal resolution by reversing the order and measuring dm distributions of APM-classified particles, which takes only ∼1− 2 min/scan. We will show that using the APM first, under our operating conditions, also improves the size resolution of the system. Here, we present the ADS, a new portable system that takes advantage of the recently developed second-generation compact APM, and the high sizing precision and sensitivity of the DMA/SPLAT system, with its capability for online multidimensional single particle characterization.10 In the ADS, APM mass-selected particles are classified by the DMA and characterized by SPLAT to yield, in a single measurement, particle composition, mp, dve, dva, dm, ρp, ρeff, and for the first time, particle DSFs in the transition and free-molecular regimes.
d m = d ve χt
Cc(d m) Cc(d ve)
(2)
where χt is the DSF in the transition regime and Cc (dm) and Cc (dve) are the Cunningham slip correction factors. For spherical particles, dm = dve (χ = 1), and for aspherical particles, dm depends on dve and χt, with larger dm corresponding to larger χt. The DMA is then tuned to select from the APM-classified particles, particles with a narrow dm distribution, transporting a narrow distribution of shapes to SPLAT, where their mass spectra and dva are measured: ρp d ve d va = ρ0 χv (3) where χv is the particle DSF in the free-molecular regime and ρ0 is unit density.4 Note that when APM-classified spherical particles are characterized, tuning the DMA to larger dm yields larger dva that correspond to particles with larger dve within the narrow APM-selected mp distribution. In contrast, tuning the DMA for aspherical particles off peak center to larger dm selects particles with larger DSF and smaller dva. In the absence of APM, dve is unknown, and the dva/dm ratio can only be used to calculate a single DSF, under the approximation that χv = χt = χ,̅ according to
ρp 1 Cc(d va χvρ0 /ρp) d va = dm Cc(d m) ρ0 χvχt ρp 1 Cc(d va χ̅ρ0 /ρp) = ρ0 χ̅ 2 Cc(d m) (4) It is, however, not clear how good this approximation is and how it changes with particle shape and size.
■
EXPERIMENTAL SECTION A detailed description of the processes of particle generation and characterization by SPLAT, DMA, and APM is provided in the Supporting Information. In the ADS, the APM is used to classify particles with a narrow distribution of mass-to-charge ratios. These particles are transported to the DMA. The DMA classifies, from the narrow distribution of particle masses, particles with a narrow distribution of mobility diameters. Note that for particles with uniform density, all the DMA-classified particles have the same mass and shape. The dva distribution and individual particle mass spectra of APM and DMA-classified particles are characterized by SPLAT. When the aerosol contains a mix of particles with different compositions and densities, the APM- and DMA-classified particles all have the same mass and dm but different shapes. In these cases, the presence of particles with a distribution of densities manifests itself in the individual particle mass spectra and broader line-shapes of the dm and dva distributions. The individual particle mass spectra are used to classify particles based on their composition23 prior to data analysis. When the material density is unknown, it can be estimated based on the measured particle compositions or composition-resolved vacuum aerodynamic size distributions.24 Particle sphericity/asphericity is identified in real-time using the combined DMA/SPLAT12 or APM/SPLAT systems or by SPLAT alone.25 Below, we present application of the ADS to
■
THEORETICAL BACKGROUND A detailed discussion of the relationships between particle dm, dva, dve, ρp, ρeff, mp, and DSFs is provided elsewhere.4,5,22 Here, we provide only the relevant equations and limit the discussion to particles with uniform density. In the ADS system, we start by selecting, with the APM, particles by mass: π m p = (d ve)3 ρp (1) 6 The dm distribution of APM-classified particles is measured by scanning DMA. The particle dm is 1460
dx.doi.org/10.1021/ac202235z | Anal. Chem. 2012, 84, 1459−1465
Analytical Chemistry
Article
1 shows that, in addition to the singly charged particles, the DMA selected doubly charged particles with a dm = 304.5 nm, whose dva = 367 nm, yielding a density of 1.210 g cm−3. The inset in Figure 1 shows an expanded scale of the dva distribution of singly charged particles, which has a full width at halfmaximum (FWHM) of 7%. Here, FWHM is defined as 100 × Δd/d0, where Δd is the full width of the size distribution at halfmaximum and d0 is the peak position. When the DMA is operated with a sheath-to-sample flow-rate ratio of 20:1, the width of the dva distribution narrows to 5% (blue trace). The narrow width of the dva distribution and its relation to the DMA operating conditions provide direct evidence that the density distribution of SOA particles is insignificant. On the basis of the measured particle density, the APM was tuned to select SOA particles with a dve = 190 nm. SPLATmeasured dva distributions of these particles are shown in Figure 1 in the green and black traces. The perfect agreement between the dva of DMA- and APM-classified particles demonstrates that the correct SOA particle density was used to tune the APM. Also shown are the APM-selected multiply charged particles, whose masses are integer multiples of the mass of particles with a dve = 190 nm and their dva distributions peak at 289, 330, and 364 nm for the doubly, triply, and quadruply charged particles, respectively. The expanded scale in the inset shows that when the APM performance parameter is set to λ = 1 and λ = 3 (see the Supporting Information for additional information on the APM performance parameter, λ), the dva distributions of the APM-classified particles have FWHM values of 4% and 3%, respectively. Operating the APM with λ = 3 brings about a significant decrease in particle number concentrations and limits the range of APM mass selection. λ = 0.32 (not shown) yields particles with a size distribution broader than the DMA. On the basis of these results, we chose to operate the APM with a λ = 1. Note that consecutive classification with the APM and DMA eliminates multiply charged particles from the final aerosol population. Here again, the line-shape of the dva distribution provides direct evidence that the density distribution is insignificant. Figure 2 presents the results of the characterization of spherical silica particles using the ADS. Figure 2a shows the overall dm distribution of the aerosolized spherical silica particles (blue trace), with a FWHM of 17%. The presence of a small fraction of silica dimers and trimers is indicated by
characterize spherical quasi-solid secondary organic aerosol (SOA) particles and solid silica particles and two types of solid aspherical particles with two crystalline shapes (NaCl and ammonium sulfate).
■
RESULTS AND DISCUSSION Spherical Particles: Quasi-Solid α-Pinene SOA and Solid Silica Particles. We begin by taking advantage of SPLAT’s high sizing resolution11 to characterize the DMA and APM performance. Figure 1 shows the comparison between the
Figure 1. Comparison between dva distributions of SOA particles classified by DMA and APM, each operated in two different conditions. Inset shows measurements for singly charged particles on the expanded scale.
SPLAT-measured dva distributions of DMA and APM-classified spherical α-pinene SOA particles, which we have previously shown to be spherical quasi-solids.26 The red trace in Figure 1 is the dva distribution of DMA-classified SOA particles (dm = 190 nm) with the DMA operating at standard conditions (sheath to sample flow rates ratio of 10:1). These particles have a dva = 228.5 nm and a density of 1.205 g cm−3, which is in excellent agreement with our previous measurements.12 Figure
Figure 2. Spherical silica particles: (a) dm distribution of aerosolized particles and of APM-classified particles with a deconvolution that illustrates the presence of dimers and trimers; (b) dva distributions of APM-classified particles with a dve = 185 nm and by the DMA tuned to the dms indicated in the legend; (c) dm distribution of the dimers and dva/ρ distributions of the dimers generated three different ways. 1461
dx.doi.org/10.1021/ac202235z | Anal. Chem. 2012, 84, 1459−1465
Analytical Chemistry
Article
Figure 3. Characterization of ammonium sulfate particles with the ADS: (a) dm distribution of 100 nm APM-classified particles. The dashed blue line illustrates the asymmetry of the dm distribution. (b) The dva distribution of 100 nm APM-classified ammonium sulfate particles that were subsequently classified by the DMA at indicated dm values; (c) relationships between calculated DSFs in the transition and free-molecular regimes.
the green trace. Using silica density of 1.75 g cm−3, as determined by DMA/SPLAT, the APM was tuned to select particles with a dve = 185 nm, whose dm distribution is shown in the red trace in Figure 2a. The narrow dm size distribution provides direct evidence that the density distribution of these particles is insignificant. Figure 2b shows that when the DMA is tuned to select from the narrow distribution of APM-classified particles that have slightly larger or smaller diameters than those at the center with a dm =185 nm (red trace), the corresponding changes in the dva follow changes in the dm. In other words, spherical particles with a larger dm have a larger dva. This again provides direct evidence that the density distribution of silica particles is insignificant. In cases where the width of the dm distribution of the APM-classified particles is determined by the presence of particles with a distribution of densities, tuning the DMA to select particles with a larger dm classifies particles with lower density, whose dva, as a result, would be smaller. The black trace in Figure 2c illustrates the dm distribution of APM-classified silica dimers, with the APM tuned to select particles with double mass and double volume dve = 21/3 × 185 = 233 nm. The dimers’ dm distribution peaks at 241 nm, yielding according to eq 2 χt of 1.05. The dva distribution of these particles was characterized by SPLAT three different ways: (1) APM (dve = 233 nm) to SPLAT; (2) APM (dve = 233 nm) to DMA (dm = 241 nm) to SPLAT; and (3) DMA (dm = 241 nm) to SPLAT. As Figure 2c demonstrates, all three of the dva distributions marked in red, green, and blue traces, respectively, peak at the same position (dva = 362 nm) to yield a χv of 1.13. These measured DSFs for silica dimers in the transition and free-molecular regimes are in good agreement with our previous measurements of dimers of uniform latex spheres and with the predicted DSFs.5 Note that the FWHM of the dva distributions of the dimers is 11%, a factor of almost 3 broader than that for spherical silica monomers, reflecting the fact that in SPLAT aspherical particles are randomly oriented and that the DSF depends on particle orientation. Broadening of the dva distribution due to particle asphericity provides a simple and direct method to distinguish online between spherical and aspherical particles using SPLAT.12 Aspherical Particles: Crystalline Ammonium Sulfate and Sodium Chloride. Ammonium sulfate and NaCl particles provide examples of common, aspherical particles, with small
DSFs.5 Because our previous studies of these particles were conducted using the DMA/SPLAT system and did not include the APM, we were only able to calculate the average DSF for each ammonium sulfate and NaCl particle size. A single DSF value, representing the shapes of all of the particles in the sample, gives the erroneous impression that all the particles have the same DSF and the same shapes, providing no information on the DSF dependence on the flow regime. Nevertheless, in those studies we found that the average DSF of ammonium sulfate is size-dependent and increases from 1.03 to 1.07 as particle size increases from 150 to 500 nm. Similarly, the average DSF of NaCl particles was found to increase from 1.06 to 1.16 as particle size increased from 160 to 1000 nm. We begin with ammonium sulfate. The red trace in Figure 3a shows the dm distribution of APM-classified ammonium sulfate particles with a dve = 100 nm and a density of 1.77 g cm−3, peak at dm = 102.5 nm, translating to χt = 1.035. In addition, there is a small peak at dm = 86.5 nm (marked by the green trace) that corresponds to doubly charged particles with double mass and double volume (dve = 21/3 × 100 nm = 126 nm) that were selected together with the singly charged dve = 100 nm particles (solid blue trace) by the APM. Because these dve = 126 nm, particles retain their double charge during transport from the APM to the DMA, they appear in the dm distribution as particles with a smaller dm. A similar peak displaying the presence of doubly charged APM-classified particles was observed in the study by Malloy et al.18 but was not considered. The dashed blue line in Figure 3a, whose half-width is 5%, exemplifies the fact that the dm distribution is asymmetric. While the 5% half-width of the left side of the dm distribution is determined by the resolution of the scanning DMA, the 12% half-width on the right side reflects the presence of ammonium sulfate particles with a larger χt. The measured dva distributions are shown in Figure 3b. The thick red line shows the dva distribution of APM-classified particles that were directly transmitted to SPLAT. It exhibits two distinct peaks that correspond to singly and doubly charged particles. Both peaks are asymmetric with a sharp edge on the right side, whose half-width is 3%, and with long tails on the left. The position of the sharp right edges are determined by the dva of nearly spherical particles with a dve = 100 and 126 nm and a density of 1.77 g cm−3. The narrow width, on the right side, indicates that all particles have the same density. The long tails 1462
dx.doi.org/10.1021/ac202235z | Anal. Chem. 2012, 84, 1459−1465
Analytical Chemistry
Article
Figure 4. Characterization of NaCl particles with the ADS: (a) dm distribution of 100 nm APM-classified particles. The dashed blue line illustrates the asymmetry of the dm distribution. (b) The dva distribution of 100 nm APM-classified NaCl particles that were subsequently classified by the DMA at indicated dm values; (c) relationships between calculated DSFs in the transition and free-molecular regimes.
on the left side of the dva distributions reflect the presence of particles with larger χv. The other traces in Figure 3b correspond to the dva distributions of particles with different shapes. As before, these particles were classified by the APM with a dve = 100 nm, sent to the DMA, where they were classified again to select particles with the dms listed in the figure’s legend and then sent to SPLAT. The figure shows that the DMA can exclusively select either singly or doubly charged particles. When the DMA is tuned to select singly charged ammonium sulfate particles with dm = 99 nm (nearly identical to their dve) the dva distribution is narrower and peaks at the largest dva (orange trace). When the DMA is tuned to select particles with larger dm, it selects particles with a larger DSF and a smaller dva. Each DMA position selects particles with a specific χt, whose value is calculated using eq 2. The measured position of the dva distribution of the same particles yields, according to eq 3, the corresponding χv. Figure 3b also shows that doubly charged particles exhibit similar behavior. As in the case of the singly charged particles, tuning the DMA to select particles with larger dm selects particles with a larger DSF and a smaller dva. Figure 3c shows the relationships between χt and χv calculated based on the ADS measurements of ammonium sulfate particles, whose sizes are indicated in the figure legend. To illustrate the measurements’ reproducibility, the figure includes results of two repeated experiments for particles with a dve = 200 nm. Figure 3c shows that the APM-classified ammonium sulfate aerosol contains particles with a remarkably wide range of DSFs. It also shows that DMA can select particles based on their shape. While particles with DSF of ∼1.03 represent the most abundant ammonium sulfate particles in the sample, particles with DSF as large as 1.5 are also present. Additionally, the figure shows that for smaller particles of 100 and 126 nm, χt ≅ χv. For 200 nm particles with small DSFs, χt ≅ χv, but for particles with larger DSFs, χt < χv. It is important to point out that the relatively large vertical bars for χv, which look like error bars in Figure 3c, do not represent measurement uncertainties or the presence of a wide distribution of particle shapes. The latter is determined by the DMA and is marked in the figure as a horizontal bar. Instead, the large spreads in χv reflect the fact that DSFs are orientationdependent and that in SPLAT particles are randomly oriented,
as we discussed in the case of the silica dimers shown in Figure 2 and in previous publications.5,12 NaCl particles are typically assumed to have a cubic shape, with a DSF of 1.08 in the continuum regime and a calculated DSF of 1.25 in the free-molecular regime.27,28 As illustrated in Figure 4, here we use the ADS to separate NaCl particles of different shapes and to measure their DSFs in the transition and free-molecular regimes. Figure 4a shows, in the red trace, the measured dm distribution of APM-classified NaCl particles with a dve = 100 nm and a density of 2.165 g cm−3, which also includes doubly charged particles (dve = 126 nm, green trace). The dm distribution of singly charged particles peaks at 105.5 nm, yielding χt = 1.08. The dm distribution of doubly charged particles peaks at 88 nm, which translates to 131.5 nm for singly charged particles, and yields χt = 1.07. The dashed blue line in Figure 4a completes the symmetric half of the left side of the dm distribution for singly charged NaCl particles to produce a lineshape with 10% FWHM, consistent with the DMA scanning resolution. As in the ammonium sulfate particles case, particles with larger dm (on the right side of the distribution) have larger DSFs. The position, asymmetry, and breadth of the line-shapes of the dm distributions shown in Figures 3a and 4a provide a simple illustration of the fact that APM-classified aspherical particles, while having a narrow dve distribution, contain particles with different shapes. It offers an explanation for the results of a recent study by Tajima et al.,29 in which DMAclassified spherical, polystyrene latex particles, and aspherical NaCl particles were used to characterize the performance of the first-generation APM. Their study reports that in contrast to the spherical particles, for which calculated APM line-shapes are nearly the same as those we observed, the measured line widths for aspherical NaCl particles were ∼1.3 to 1.5 larger than calculated. According to our data, the DMA-selected NaCl particles with narrow dm distribution include particles with a range of shapes and a wider distribution of the dve and mp than expected. In the same study, it was found that the APM-measured masses of DMA-classified spherical particles were in excellent agreement with those calculated. In contrast, the authors point to significant discrepancies between the measured and calculated masses of aspherical NaCl particles. When they assumed the NaCl particles were spherical, their calculated 1463
dx.doi.org/10.1021/ac202235z | Anal. Chem. 2012, 84, 1459−1465
Analytical Chemistry
Article
and NaCl particles have broad distributions of shapes and successfully separated particles with the same charge, size, mass, and shape and then characterized their properties in detail. In addition to the data presented here, we have successfully applied the ADS to characterize fractal soot and dust particles. These results will be presented in separate publications. We characterized the performance of the newly manufactured, second-generation APM, a compact field-deployable instrument that can be used to measure particle mass distributions or to select particles with a narrow distribution of masses. We demonstrated significantly improved temporal and sizing resolution by reversing the traditional manufacturerrecommended DMA-APM approach by using the APM to classify particles with a narrow distribution of masses followed by the DMA in a scanning mode to measure their mobility diameters. We showed that under our operating conditions, the APM selects particles with a size distribution of almost a factor of 2 narrower than that selected by the DMA. We also demonstrated that it is possible to classify particles in sequence by the APM and then by the DMA to select particles of one charge, thereby eliminating the problems associated with the ever-present multiply charged particles. We established that the very same approach offers the unique capability to separate particles of different shapes in situ and in real-time to be used by different applications or to be characterized by various particle analysis techniques like microscopy, mass spectrometry, and spectroscopy.
masses were significantly larger than those measured with the APM. In an attempt to take particle shape into account, the authors calculated the NaCl DSFs using the only literature available approach27,28 and used these values to calculate particle masses that were ∼20% lower than those measured with the APM. If we instead use our measured NaCl DSF (χt = 1.08) to calculate the masses of NaCl particles characterized by Tajima et al., we find them to be in perfect agreement with their APM measurements. Thus far, we focused only on the most probable NaCl DSF in the transition regime. Figure 4b shows the measured dva distributions of APM- and DMA-classified NaCl particles. The thick red line corresponds to the dva distribution of APMclassified particles that were transmitted directly to SPLAT. Two distinct peaks corresponding to singly and doubly charged particles are clearly visible. The line-shapes of both particle sizes have sharp edges on the larger dva side with a 3% half width, indicating insignificant material density distribution. Peak positions are consistent with the χv = 1.08 and tail-off on the left side of the dva distributions, indicating the presence of particles with larger DSFs. The other traces shown in Figure 4b correspond to the dva distributions of particles with different shapes and the dm values listed in the figure legend. Figure 4c is a plot of the χv vs χt for three NaCl particle sizes. It provides a clear graphic display of the range of DSFs that these particles have. We find the most probable χt for these particle sizes is 1.08 ± 0.01. In the free-molecular regime, the DSFs are 1.08 ± 0.02 and 1.11 ± 0.02 for 100 nm and for the two larger NaCl particles sizes, respectively. Figure 4c shows that particles, with DSFs as large as 1.5, are also present in this sample. It is important to keep in mind that the values reported here refer to the χv in the random orientation and that the broad line-shapes of the dva distributions shown in Figure 4b are the result of the orientation-dependent DSF and provide direct information on the relationship between orientation and DSF. As in the case of ammonium sulfate particles, the horizontal bars in the χt dimension are due to the width of the DMA, and the vertical bars in the χv dimension reflect the fact that the DSF depends on the particle orientation. Figure 4c suggests that the DSFs in the two flow regimes are nearly the same for the 100 and 126 nm NaCl particles. For larger particles, the χv is slightly larger than χt.
■
ASSOCIATED CONTENT
* Supporting Information S
Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected].
■
ACKNOWLEDGMENTS This work was supported by the U.S. Department of Energy (DOE) Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences, and Biosciences. The research was performed using the Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by the U.S. DOE’s Office of Biological and Environmental Research and located at Pacific Northwest National Laboratory (PNNL). PNNL is operated by the U.S. DOE by Battelle Memorial Institute under Contract No. DE-AC06-76RL0 1830. Special thanks to Darrell Herling, George Muntean, and additional support by the U.S. DOE’s Office of Energy Efficiency and Renewable Energy.
■
CONCLUSIONS We presented the ADS, a new portable system that makes it possible, for the first time, to separate a mixture of aspherical particles, particles with the same sizes, masses, and shapes, and to measure their DSFs in the transition and free-molecular flow regimes with very high precision. The ADS first uses the APM to select particles with a narrow dve distribution, followed by the DMA to select particles with a narrow dm distribution from that population, thus with narrow χt and shapes. These particles are then characterized by SPLAT to yield the individual particle compositions and dva, from which the χv is calculated. We illustrated the utility of the newly developed ADS to characterize aspherical ammonium sulfate and NaCl particles with different sizes and shapes. Most importantly, we measured, for the first time, the DSFs in two different flow regimes for ammonium sulfate and NaCl particles, two common atmospherically relevant salt particles. We showed that the most probable NaCl DSFs in the transition and free-molecular regimes are different than their predicted values. We demonstrated that even APM-classified ammonium sulfate
■
REFERENCES
(1) Orendorff, C. J.; Sau, T. K.; Murphy, C. J. Small 2006, 2 (5), 636−639. (2) Song, D. K.; Lenggoro, I. W.; Hayashi, Y.; Okuyama, M.; Kim, S. S. Langmuir 2005, 21 (23), 10375−10382. (3) Lall, A. A.; Friedlander, S. K. J. Aerosol Sci. 2006, 37 (3), 260− 271. (4) DeCarlo, P. F.; Slowik, J. G.; Worsnop, D. R.; Davidovits, P.; Jimenez, J. L. Aerosol Sci. Technol. 2004, 38 (12), 1185−1205. (5) Zelenyuk, A.; Cai, Y.; Imre, D. Aerosol Sci. Technol. 2006, 40 (3), 197−217. 1464
dx.doi.org/10.1021/ac202235z | Anal. Chem. 2012, 84, 1459−1465
Analytical Chemistry
Article
(6) Hinds, W. C., Aerosol Technology: Properties, Behavior, and Measurement of Airborne Particles. 2nd ed.; Wiley-Interscience: New York, 1999. (7) Zelenyuk, A.; Imre, D. Aerosol Sci. Technol. 2007, 41 (2), 112− 124. (8) Dick, W. D.; Ziemann, P. J.; Huang, P. F.; McMurry, P. H. Meas. Sci. Technol. 1998, 9 (2), 183−196. (9) Perry, R. J.; Hunt, A. J.; Huffman, D. R. Appl. Opt. 1978, 17 (17), 2700−2710. (10) Zelenyuk, A.; Imre, D. Int. Rev. Phys. Chem. 2009, 28 (2), 309− 358. (11) Zelenyuk, A.; Yang, J.; Choi, E.; Imre, D. Aerosol Sci. Technol. 2009, 43 (5), 411−424. (12) Zelenyuk, A.; Yang, J.; Song, C.; Zaveri, R. A.; Imre, D. Environ. Sci. Technol. 2008, 42 (21), 8033−8038. (13) Slowik, J. G.; Stainken, K.; Davidovits, P.; Williams, L. R.; Jayne, J. T.; Kolb, C. E.; Worsnop, D. R.; Rudich, Y.; DeCarlo, P. F.; Jimenez, J. L. Aerosol Sci. Technol. 2004, 38 (12), 1206−1222. (14) Ehara, K.; Hagwood, C.; Coakley, K. J. J. Aerosol Sci. 1996, 27 (2), 217−234. (15) Lall, A. A.; Ma, X. F.; Guha, S.; Mulholland, G. W.; Zachariah, M. R. Aerosol Sci. Technol. 2009, 42 (11), 1075−1083. (16) Park, K.; Kittelson, D. B.; McMurry, P. H. Atmos. Environ. 2003, 37 (9−10), 1223−1230. (17) Kuwata, M.; Kondo, Y. Atmos. Chem. Phys. 2009, 9 (16), 5921− 5932. (18) Malloy, Q. G. J.; Nakao, S.; Qi, L.; Austin, R.; Stothers, C.; Hagino, H.; Cocker, D. R. Aerosol Sci. Technol. 2009, 43 (7), 673−678. (19) McMurry, P. H.; Wang, X.; Park, K.; Ehara, K. Aerosol Sci. Technol. 2002, 36 (2), 227−238. (20) Lee, S. Y.; Widiyastuti, W.; Tajima, N.; Iskandar, F.; Okuyama, K. Aerosol Sci. Technol. 2009, 43 (2), 136−144. (21) Zhang, R. Y.; Khalizov, A. F.; Pagels, J.; Zhang, D.; Xue, H. X.; McMurry, P. H. Proc. Natl. Acad. Sci. U.S.A. 2008, 105 (30), 10291− 10296. (22) Zelenyuk, A.; Cai, Y.; Chieffo, L.; Imre, D. Aerosol Sci. Technol. 2005, 39 (10), 972−986. (23) Zelenyuk, A.; Imre, D.; Cai, Y.; Mueller, K.; Han, Y. P.; Imrich, P. Int. J. Mass Spectrom. 2006, 258 (1−3), 58−73. (24) Vaden, T. D.; Imre, D.; Beranek, J.; Zelenyuk, A. Aerosol Sci. Technol. 2011, 45 (1), 125−135. (25) Vaden, T. D.; Imre, D.; Beranek, J.; Zelenyuk, A. Aerosol Sci. Technol. 2011, 45 (1), 113−124. (26) Shrivastava, M.; Zelenyuk, A.; Imre, D.; Beranek, J.; Easter, R.; Zaveri, R.; Fast, J. Atmos. Chem. Phys. Discuss. 2011, 11 (7), 20107− 20139. (27) Dahneke, B. E. J. Aerosol Sci. 1973, 4 (2), 147−161. (28) Dahneke, B. E. J. Aerosol Sci. 1973, 4 (2), 163−170. (29) Tajima, N.; Fukushima, N.; Ehara, K.; Sakurai, H. Aerosol Sci. Technol. 2011, 45 (2), 196−214.
1465
dx.doi.org/10.1021/ac202235z | Anal. Chem. 2012, 84, 1459−1465
