Complex Refractive Indices of Aerosols Retrieved by Continuous

Feb 6, 2009 - down (CRD) spectroscopy provides highly sensitive measurement of aerosols' extinction coefficients from which the complex refractive ind...
0 downloads 0 Views 1MB Size
Download PDF
Anal. Chem. 2009, 81, 1762–1769

Complex Refractive Indices of Aerosols Retrieved by Continuous Wave-Cavity Ring Down Aerosol Spectrometer N. Lang-Yona,† Y. Rudich,† E. Segre,‡ E. Dinar,† and A. Abo-Riziq*,† Department of Environmental Sciences and Physics Services, Weizmann Institute, Rehovot 76100, Israel The major uncertainties associated with the direct impact of aerosols on climate call for fast and accurate characterization of their optical properties. Cavity ring down (CRD) spectroscopy provides highly sensitive measurement of aerosols’ extinction coefficients from which the complex refractive index (RI) of the aerosol may be retrieved accurately for spherical particles of known size and number density, thus it is possible to calculate the single scattering albedo and other atmospherically relevant optical parameters. We present a CRD system employing continuous wave (CW) single mode laser. The single mode laser and the high repetition rate obtained significantly improve the sensitivity and reliability of the system, compared to a pulsed laser CRD setup. The detection limit of the CW-CRD system is between 6.67 × 10-10 cm-1 for an empty cavity and 3.63 × 10-9 cm-1 for 1000 particles per cm3 inside the cavity, at a 400 Hz sampling and averaging of 2000 shots for one sample measurement taken in 5 s. For typical pulsed-CRD, the detection limit for an empty cavity is less than 3.8 × 10-9 cm-1 for 1000 shots averaged over 100 s at 10 Hz. The system was tested for stability, accuracy, and RI retrievals for scattering and absorbing laboratorygenerated aerosols. Specifically, the retrieved extinction remains very stable for long measurement times (1 h) with an order of magnitude change in aerosol number concentration. In addition, the optical cross section (σext) of a 400 nm polystyrene latex sphere (PSL) was determined within 2% error compared to the calculated value based on Mie theory. The complex RI of PSL, nigrosin, and ammonium sulfate (AS) aerosols were determined by measuring the extinction efficiency (Qext) as a function of the size parameter ((πD)/λ) and found to be in very good agreement with literature values. A mismatch in the retrieved RI of Suwannee River fulvic acid (SRFA) compared to a previous study was observed and is attributed to variation in the sample composition. The small system presented delivers high ability for fast measurements and accurate analysis, making it a good candidate for field aerosol optical properties studies. 1762

Analytical Chemistry, Vol. 81, No. 5, March 1, 2009

The impact of aerosols on the Earth’s climate and air quality is a major uncertainty in climate change models as was emphasized in the latest Intergovernmental Panel on Climate Change (IPCC) report,1 pointing to an urgent need for further studies on aerosol properties and processes. A major component of the aerosol’s climatic effect is their scattering and absorption of solar radiation, which is governed by their optical and physical properties.2-7 Common scattering aerosols in the atmosphere include inorganic salts (sea spray, sulfate aerosols).8,9 These aerosols have mainly a “cooling effect” on the climate due to a decrease in the solar radiation that reaches the Earth’s surface. Mineral dust,10-12 soot aerosols,13-18 and organic aerosols * To whom correspondence should be addressed. E-mail: Ali.Abo-riziq@ weizmann.ac.il. † Department of Environmental Sciences. ‡ Physics Services. (1) Solomon, S.; Qin, D.; Manning, M.; Alley, R. B.; Berntsen, T.; Bindoff, N. L.; Chen, Z.; Chidthaisong, A.; Gregory, J. M.; Hegerl, G. C.; Heimann, B. H.; Hoskins, B. J.; Joos, F.; Jouzel, J.; Kattsov, V.; Lohmann, U.; Matsuno, T.; Molina, M.; Nicholls, N.; Overpeck, G. R.; Ramaswamy, V.; Ren, J.; Rusticucci, M.; Somerville, R.; Stocker, T. F.; Whetton, P.; Wood, R. A.; Wratt, D. Technical Summary. In Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change; Cambridge University Press: Cambridge, U.K. and New York, USA, 2007. (2) Bellouin, N.; Boucher, O.; Haywood, J.; Reddy, M. S. Nature 2005, 438, 1138–1141. (3) Chen, W. T.; Kahn, R. A.; Nelson, D.; Yau, K.; Seinfeld, J. H. J. Geophys. Res. 2008, 113. (4) Kim, D. Y.; Ramanathan, V. J. Geophys. Res. 2008, 113. (5) Lohmann, U.; Feichter, J. Atmos. Chem. Phys. 2005, 5, 715. (6) Niranjan, K.; Sreekanth, V.; Madhavan, B. L.; Moorthy, K. K. Geophys. Res. Lett. 2007, 34. (7) Riziq, A. A.; Erlick, C.; Dinar, E.; Rudich, Y. Atmos. Chem. Phys. 2007, 7, 1523–1536. (8) Chamaillard, K.; Jennings, S. G.; Kleefeld, C.; Ceburnis, D.; Yoon, Y. J. J. Quant. Spectro. Rad. Trans. 2003, 79-80, 577–597. (9) Kiehl, J. T.; Schneider, T. L.; Rasch, P. J.; Barth, M. C. J. Geophys. Res. 2000, 105, 1441–1457. (10) Li, Z.; Goloub, P.; Blarel, L.; Damiri, B.; Podvin, T.; Jankowiak, I. Appl. Opt. 2007, 46, 1548–1553. (11) Sokolik, I. N.; Toon, O. B. Nature 1996, 381, 681–683. (12) Sokolik, I. N.; Toon, O. B.; Bergstrom, R. W. J. Geophys. Res. 1998, 103, 8813–8826. (13) Andreae, M. O. Nature 2001, 409, 671–672. (14) Bond, T. C.; Streets, D. G.; Yarber, K. F.; Nelson, S. M.; Woo, J. H.; Klimont, Z. J. Geophys. Res. 2004, 109. (15) Bond, T. C.; Bergstrom, R. W. Aerosol Sci. Technol. 2006, 40, 1–41. (16) Henson, B. F. J. Geophys. Res. 2007, 112. (17) Liua, L.; Mishchenkoa, M. I. J. Quant. Spect. Rad. Trans. 2007, 106, 262– 273. (18) Menon, S.; Hansen, J.; Nazarenko, L.; Luo, Y. F. Science 2002, 297, 2250– 2253. 10.1021/ac8017789 CCC: $40.75  2009 American Chemical Society Published on Web 02/06/2009

(“brown” carbon)19-23 are important absorbing aerosols that can lead to global and regional warming effects.18,24-28 One significant challenge in constraining the aerosols’ direct climatic effect is the inadequate knowledge of their optical properties, especially of complex aerosols that contain absorbing components.1 A clearer knowledge of aerosol characteristics is required and can be partially achieved by improving in situ measurements of extinction and other optical parameters. This requires fast and accurate measurements that can approach the uncertainty associated with fast changes in the aerosols’ chemical and spatial changes.1,29 A relatively new method used for measuring optical properties of aerosols is cavity ring down (CRD) spectroscopy. CRD spectroscopy, first developed by O’Keefe and Deacon, is a sensitive direct extinction evaluation technique.30 Briefly, in a CRD experiment a defined amount of light enters into a highly reflective optical cavity and is reflected multiple times within the cell. The time required for one periodic movement of the light is 2L/c (L is the length of the cavity and c is the speed of light). The light intensity decays exponentially due to cavity losses and small transmission of the mirrors, with a time constant τ0, defined as the empty cavity decay time. Introducing gases or particles into the cavity reduces the decay as a result of scattering and absorption.31 Because of the long optical path length achieved, CRD is most sensitive for absorption spectroscopy and has been widely used for gas phase studies.32-35 However, recently, optical properties of aerosols have been investigated using pulsed-CRD (19) Andreae, M. O.; Gelencser, A. Atmos. Chem. Phys. Discuss. 2006, 6, 3419– 3463. (20) Dinar, E.; Riziq, A. A.; Spindler, C.; Erlick, C.; Kiss, G.; Rudich, Y. Faraday Discuss. 2008, 137, 279–295. (21) Decesari, S.; Fuzzi, S., Facchini, M. C.; Mircea, M.; Emblico, L.; Cavalli, F.; Maenhaut, W.; Chi, X.; Schkolnik, G.; Falkovich, A.; Rudich, Y.; Claeys, M.; Pashynska, V.; Vas, G.; Kourtchev, I.; Vermeylen, R.; Hoffer, A.; Andreae, M. O.; Tagliavini, E.; Moretti, F.; Artaxo, P. Atmos. Chem. Phys. 2006, 6. (22) Hoffer, A.; Gelencser, A.; Guyon, P.; Kiss, G.; Schmid, O.; Frank, G. P.; Artaxo, P.; Andreae, M. O. Atmos. Chem. Phys. 2006, 6, 3563–3570. (23) Sun, H. L.; Biedermann, L.; Bond, T. C. Geophys. Res. Lett., 2007, 34. (24) Hansen, J.; Sato, M.; Ruedy, R.; Nazarenko, L.; Lacis, A.; Schmidt, G. A.; Russell, G.; Aleinov, I.; Bauer, M.; Bauer, S.; Bell, N.; Cairns, B.; Canuto, V.; Chandler, M.; Cheng, Y.; Del Genio, A.; Faluvegi, G.; Fleming, E.; Friend, A.; Hall, T.; Jackman, C.; Kelley, M.; Kiang, N.; Koch, D.; Lean, J.; Lerner, J.; Lo, K.; Menon, S.; Miller, R.; Minnis, P.; Novakov, T.; Oinas, V.; Perlwitz, J.; Perlwitz, J.; Rind, D.; Romanou, A.; Shindell, D.; Stone, P.; Sun, S.; Tausnev, N.; Thresher, D.; Wielicki, B.; Wong, T.; Yao, M.; Zhang, S. J. Geophys. Res. 2005, 110. (25) Horvath, H. Atmos. Environ. 1993, 27, 293–317. (26) Koren, I.; Kaufman, Y. J.; Rosenfeld, D.; Remer, L. A.; Rudich, Y. Geophys. Res. Lett. 2005, 32. (27) Ramanathan, V.; Chung, C.; Kim, D.; Bettge, T.; Buja, L.; Kiehl, J. T.; Washington, W. M.; Fu, Q.; Sikka, D. R.; Wild, M. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 5326–5333. (28) Ramanathan, V.; Crutzen, P. J.; Kiehl, J. T.; Rosenfeld, D. Science 2001, 294, 2119–2124. (29) Strawa, A. W.; Castaneda, R.; Owano, T.; Baer, D. S.; Paldus, B. A. J. Atmos. Ocean. Tech. 2003, 20, 454–465. (30) O’Keefe, A.; Deacon, D. A. G. Rev. Sci. Instrum. 1988, 59, 2544–2555. (31) Mazurenka, M.; Orr-Ewing, A. J.; Peverallb, R.; Ritchieb, G. A. D. Annu. Rep. Prog. Chem., Sect. C 2005, 101, 100–142. (32) Brown, S. S.; Stark, H.; Ciciora, S. J.; McLaughlin, R. J.; Ravishankara, A. R. Rev. Sci. Instrum. 2002, 73, 3291–3301. (33) Pradhan, M.; Lindley, R. E.; Grilli, R.; White, I. R.; Martin, D.; Orr-Ewing, A. J. Appl. Phys. 2008, 90, 1–9. (34) Thiebaud, J.; Fittschen, C. Appl. Phys. B: Lasers Opt. 2006, 85, 383–389. (35) Brown, S. S. Chem. Rev. 2003, 103, 5219–5238.

systems.7,20,31,36-42 Pulsed CRD systems are very sensitive but often have accuracy limitations caused by the multiple longitudinal and transverse mode excitation, leading to multiexponential decays (each mode has its own characteristic ring-down time). Most pulsed systems reported in the literature are also limited by their low repetition rate.30,36 For the best of our knowledge, only Moosmuler et. al38 used a pulsed CRD system operating at 14 kHz for cavity enhanced detection along with CRD. Continuous wave (CW) CRD has been used mainly for highresolution gaseous absorption spectroscopy (see for an example refs 33, 34, 43, and 44). Reported studies on the aerosol’s optical properties measured by CW-CRD are rather limited. A number of studies focused on the optical properties of a single aerosol particle extinction.45,46 To the best of our knowledge, only one system was constructed by Strawa et. al29,31,47 for measuring the optical properties of ambient aerosols. This small portable system examined the optical properties of aerosols at 675 and 1550 nm. It was designed with three mirrors forming a narrow triangle. The input beam in this configuration is reflected at 90° and therefore does not couple back into the laser. The ability of this setup to simultaneously measure at two different wavelengths provides estimation of the Ångstrom exponent. Furthermore, the instrument was designed to measure the scattering independently by locating a scattering detector inside the cavity at 90° relative to the beam pathway. This arrangement allowed the determination of the single scattering albedo and the absorption coefficient (Rabs). The detection limit reported for the empty cavity of this setup was 1.5 × 10-8 cm-1 at 50-100 Hz and averaging for 10 s. We demonstrate here a new and simple CW-CRD setup based on single mode CW-CRD for measuring optical properties of aerosols at 532 nm. This new system overcomes many limitations encountered with pulsed lasers CRD-aerosol spectrometers. The laser’s single cavity mode ensures high CRD sensitivity due to the true single exponential decay.7,20,39 Moreover, a fast modulation of the cavity length allows operation at high repetition rates and substantially faster averaging, enabling accurate measurements in short times. The design of this new CW-CRD is different (36) Baynard, T.; Lovejoy, E. R.; Pettersson, A.; Brown, S. S.; Lack, D.; Osthoff, H.; Massoli, P.; Ciciora, S.; Dube, W. P.; Ravishankara, A. R. Aerosol Sci. Technol. 2007, 41, 447–462. (37) Lack, D. A.; Lovejoy, E. R.; Baynard, T.; Pettersson, A.; Ravishankara, A. R. Aerosol Sci. Technol. 2006, 40, 697–708. (38) Moosmuller, H.; Varma, R.; Arnott, W. P. Aerosol Sci. Technol. 2005, 39, 30–39. (39) Pettersson, A.; Lovejoy, E. R.; Brock, C. A.; Brown, S. S.; Ravishankara, A. R. J. Aerosol Sci. 2004, 35, 995–1011. (40) Riziq, A. A.; Trainic, M.; Erlick, C.; Segre, E.; Rudich, Y. Atmos. Chem. Phys. 2008, 8, 1823–1833. (41) Rudic, S.; Miles, R. E. H.; Orr-Ewing, A. J.; Reid, J. P. Appl. Opt. 2007, 46, 6142–6150. (42) Spindler, C.; Abo Riziq, A.; Rudich, Y. Aerosol Sci. Technol. 2007, 41, 1011– 1017. (43) Macko, P.; Romanini, D.; Mikhailenko, S. N.; Naumenko, O. V.; Kassi, S.; Jenouvrier, A.; Tyuterev, V. G.; Campargue, A. J. Mol. Spectrosc. 2004, 227, 90–108. (44) Crunaire, S.; Tarmoul, J.; Fittschen, C.; Tomas, A.; Lemoine, B.; Coddeville, P. Appl. Phys. B: Lasers Opt. 2006, 85, 467–476. (45) Butler, T. J. A.; Miller, J. L.; Orr-Ewing, A. J. J. Chem. Phys. 2007, 126, 174302. (46) Miller, J. L.; Orr-Ewing, A. J. J. Chem. Phys. 2007, 126, 174303. (47) Hallar, A. G.; Strawa, A. W.; Schmid, B.; Andrews, E.; Ogren, J.; Sheridan, P.; Ferrare, R.; Covert, D.; Elleman, R.; Jonsson, H.; Bokarius, K.; Luu, A. J. Geophys. Res. 2006, 111.

Analytical Chemistry, Vol. 81, No. 5, March 1, 2009

1763

Figure 1. Schematic representation of the CW-CRD aerosols spectrometer setup for measuring aerosol optical properties.

from Strawa’s setup. It is a further improvement of our pulsed system and takes into account several considerations for improving the detection limit and accuracy of the system as described later. In this manuscript, we demonstrate a first application of the new system, emphasizing its abilities to derive the complex refractive index of aerosols and discuss some of its advantages and shortcomings compared to the pulsed CRD system used for aerosol research. EXPERIMENT The experimental setup consists of aerosol generation, sizing, and counting equipment coupled with a new CW-CRD system, as illustrated in Figure 1. CW-Cavity Ring Down (CRD). The cavity (Figure 1) is made of a 3/8 in. conductive 85 cm long tube, yielding a cavity with a small volume of 0.43 L. Two highly reflective concave mirrors (1 m radius of curvature, stated reflectivity R ) 99.995%, Los Gatos) are placed in newly designed holders that enable precise tuning of the mirror axes. The position of one mirror is fixed, and the second can move along the cavity axis by a piezo-ring actuator (Piezomechanic) while maintaining stable alignment. A small flow of dry nitrogen (0.05 and 0.1 standard liter per minute (SLM) introduced in front of the mirrors prevents mirror contamination by particle deposition. The aerosol flow (0.8 SLM) enters the cavity through two tubes at 45° from either side of the cavity to ensure a uniform aerosol distribution inside the cavity, which is important for accurate measurements, as previously shown for pulsed-CRD systems.48 The length of the cavity actually occupied by particles is 75 cm. (48) van Leeuwen, N. J.; Diettrich, J. C.; Wilson, A. C. Appl. Opt. 2003, 42, 3670–3677.

1764

Analytical Chemistry, Vol. 81, No. 5, March 1, 2009

The particles concentration at the exit of the cavity is constantly measured by a condensation particle counter (CPC, TSI 3775), and the dilution factor inside the cavity due to the purge flows is taken into account in the analysis. The output of a 50 mW 532 nm single-mode CW laser (Torus 532) is directed through an optical isolator and a telescope (lenses with 10 and 3 cm focal lengths) to an acousto-optic modulator (AOM, Brimrose). The AOM is modulated by the buildup of light intensity in the cavity, as will be described later. The first order diffracted beam is coupled to the cavity through a 75 cm focusing lens and the laser intensity emerging from the CRD cell is measured with photomultiplier (Hamamatsu H6780-02MOD). In the CW-CRD operation, the mirror mounted on the piezo travels back and forth at some frequency. When the length of the cavity matches the laser mode, the light is efficiently trapped and the rapid intensity buildup is used to directly trigger the AOM and to block the laser beam. The exponential decay of the intensity in the cavity is digitized and processed for each “pulse”. The exponential decay time for aerosol free empty cavity (only dried nitrogen) is τ0 ∼ 30 µs, implying a 99.991% reflectivity. Decay times shortened when aerosols were introduced. A 14 bit fast digitizer card (National Instruments PCI-5122) is used to acquire the signal, at rates between 5 to 20 Mega samples per second (Mspls). The acquisition is triggered when the PMT signal itself exceeds an adjustable threshold. When this happens, the card generates a digital trigger signal which is fed into the AOM driver, to block the laser beam. This setup has response times on the order of tens of nanoseconds. As the diffracted laser beam is switched off, the digitizer records the event for 50-100 µs. This setup does not require additional gating electronics. The trigger is automatically rearmed at the end of each collected event, restoring laser illumination to the cavity. The onboard memory

of the card can store thousands of consecutive decays. By transfer of all the collected decay signals at once to the computer RAM, rather than individually, it is possible to achieve good efficiency and collection rates up to several hundred events per second. The rate and quality of events are determined by the residence time of the cavity near a resonance of the laser frequency, which is spoiled by uncontrollable mechanical vibrations of the system. Ringup to a sufficient signal level takes place only if the cavity remains “almost” tuned (i.e., if the initial length of the cavity is very close to a resonance and neither the vibration nor the piezo drive change it within the ringup time). Under such conditions, the PMT signal increases in a short time after blocking the beam. Once triggered, and photons injection is shut off, the intensity decay is essentially due to the aerosol optical properties and is well exponential. The choice of the trigger level also plays a role, since a higher level causes the process to start only when the cavity is closer to a resonance, which happens less frequently than with a lower threshold but provides decay signals with a higher signal-to-noise ratio. A background mechanical noise without a piezo drive would be sufficient by itself to randomly tune the cavity and to produce a distribution of decay events. However, this would occur at random times, possibly with long lags inbetween, depending on the uncontrollable vibration characteristics, and would result in a distribution of “good” and “bad” decay events. Application of a periodic signal with an amplitude that produces a mirror excursion of roughly one laser wavelength to the piezoelectric ring is an effective way for achieving a high rate of repeatable decays and a sufficient degree of repeatability and control of the measurements.49,50 A 1 kHz, ∼10 Vpp signal was computer-generated to this extent by another National Instruments DAQ card, buffered by an audio amplifier, and applied directly to the piezoelectric mirror holder. The frequency of this driving signal cannot be too high for the cavity to remain tuned during the ringup time. Other studies (for examples see refs 51 and 48) compensated the mechanical vibrations by driving the piezoelectric ring with a feedback signal, but their setups require a much higher electronic and optical complexity than ours. Software Development. Custom-made Labview software controlled the acquisition and analysis, displaying signals graphically in real time and logging the evolution of decay (τ) and concentration (N) in the course of the experiments. The software provides a graphical, interactive method of choosing the optimal parameters for the exponential fit of the decay events, such as the trigger level for AOM shut-off, the number of decay events average, the temporal interval over which the exponential fit is sought, the statistical weighting method, and the rate of measurements. Moreover, the program can store the reference decay (τ0) and compute and display the evolution of τ as well as the extinction (Rext) and the extinction efficiency (Qext) derived from it during the measurements. Aerosol Generation and Classification. Aqueous solutions (20-500 mg L-1) of the compound of interest are nebulized using a constant output atomizer (TSI-3076, 25 psi, ∼3 SLM (49) Romanini, D.; Kachanov, A. A.; Sadeghi, N.; Stoeckel, F. Chem. Phys. Lett. 1997, 264, 316–322. (50) Vogler, D. E.; Lorencak, A.; Rey, J. M.; Sigrist, M. W. Opt. Lasers Eng. 2005, 43, 527. (51) Martinez, R. Z.; Metsala, M.; Vaittinen, O.; Lantta, T.; Halonen, L. J. Opt. Soc. Am. 2006, 23, 727–740.

flow) with dry particle-free pure nitrogen. The resulting droplets are dried in two silica gel column dryers (RH < 3%) forming a polydisperse distribution of dried aerosol. The aerosols (about 0.5 SLM) are neutralized (TSI 3012A) to obtain an equilibrium charge distribution. A monodisperse distribution is generated by an electrostatic classifier (TSI differential mobility analyzer (DMA)) operating with a 5 SLM dry (RH < 3%) clean air sheath flow. The size-selected aerosols are then diluted by dry nitrogen to obtain a 0.8 SLM flow which is directed into the CRD cell. Retrieval Algorithm Method and RI error Calculation. The retrieval algorithm for single component particles was detailed by Abo Riziq et. al7 and Lack et. al.37 Shortly, it compares the measured Qext for a given size parameter to the Qext calculated using a Mie scattering subroutine for homogeneous spheres,52 simultaneously altering the real and imaginary components of the refractive indices. A set of refractive indices is found by minimizing the “merit function” (χ2/Nps), where N

χ2 )

∑

(Qext measured - Qext calculated)i2 i2

i)1

and Nps is the number of measurements (different particle sizes) used in the fit, and ε is the standard deviation of the measured Qext (see eq 15.5.5 in ref 52). The algorithm scans through possible physical values of the indices and increases the resolution until converging to the merit function’s absolute minimum. This algorithm has been successfully applied in several recent studies.7,20 In order to estimate the error of the retrieved RI, we study the deviation of χ2 in the vicinity of χ02(which is the initial χ2 of the retrieved RI), as described by Dinar et. al.20 Assuming the errors of the measurement εi2 to be normally distributed, the values of χ2 for different measurements follow a χ2 distribution for the two degrees of freedom n and k. The quintile for the 68.3% (1σ) confidence level for this parameter set is 2.298. Any measurement that falls within the 1σ error bound of the best measurement is considered acceptable if χ2 < χ02 + 2.298. The values for n and k that fulfill this relation lie within contours around χ02. Projections of the contour lines onto the n and k plane give the standard errors ∆n and ∆k, respectively. RESULTS AND DISCUSSION The capabilities of the CW-CRD setup were validated in a series of experiments in which we tested the system’s sensitivity, limit of detection, and reliability for different type of aerosols. Detection Limit, Stability, and Accuracy. A cleaner single exponential decay is obtained due to the single mode laser and better mode matching. This dramatically improves the accuracy in determining the decay time and hence the derived extinction and system overall sensitivity. Figure 2 presents 72 single exponential decays obtained with different concentrations of 400 nm polystyrene latex (PSL) spheres. Each decay is an average of 2000 shots. The detection limit for an empty cavity is dictated by the ability to precisely measure the minimum difference between τ0 and the measured τ, given by the following equation:7 Rmin )

L √2Sτ0 cd τ 2√RT 0

Analytical Chemistry, Vol. 81, No. 5, March 1, 2009

(1) 1765

Figure 2. A total of 72 exponential decays measured with CW-CRD at different PSL concentrations (200-500 particles/cm3).

cm- 1 for τ0 ) ∼60 µs (much longer than the system studied here) for a 100 s measurement.7 Higher detection limit is attained with higher reflectivity mirrors (easily achievable), increasing τ0, decreasing Sτ0, and by carrying out the experiments at higher repetition rates. The detection limit for a cavity filled with particles, calculated following Pettersson et. al,39 ranges between 6.64 × 10-10 and 3.63 × 10-9 cm- 1 for particle number concentration (N) between 0 and 1000 particles cm-3, respectively. Stability. The stability of the retrieval quality was tested for over 1 h and with varying aerosol concentrations. Figure 3 presents N, extinction (Rext), extinction efficiency (Qext), and the decay time (τ) with respect to time for 400 nm PSL. Each point shown represents an average of 2000 shots sampled at a repetition rate of 400 Hz for 5 s. The initial PSL concentration was ∼1000 cm-3 and decreased to less than 90 cm-3. With decreasing particle concentrations, Rext decreased and τ increased while Qext remained constant. The extinction efficiency is calculated by eq 2. Qext ) 4Rext ⁄ πND2

(2)

where N is the particle number density, Rext is the extinction coefficient, and D is the particle’s diameter. While the extinction efficiency, Qext, remains constant for specific particle diameter, Rext and N are linearly correlated.7 τ and Rext are inversely proportional, τ ) L ⁄ C(1 - r + Rextd)

Figure 3. Particle number density (N) (a), extinction coefficient (Rext) (b), decay time (τ) (c), and the extinction efficiency (Qext) (d) for 400 nm polystyrene latex spheres (PSL) as a function of time. Qext remains almost constant as the concentration decreased. This measurement shows the linearity between Rext and the particle number density and the inverse proportion between Rext and τ over a long measurement time.

where L is the length of the cavity, d is the actual part of the cavity filled with the aerosol, c is the speed of light, Sτ0 is the minimum detectable change in the ring down time for one laser shot, R is the repetition rate, and T is the sampling time. In this CW-CRD system, L is 85 cm, d ) 75 cm, τ0 ) 30 µs, Sτ0 ) 0.5 µs, the repetition rate, R, is 400 Hz, and the sampling time, T, is 5 s. The detection limit of the system for these conditions is 6.64 × 10-10 cm- 1. Our previous pulsed laser CRD system has a detection limit for an empty cavity of 3.77 × 10-9 1766

Analytical Chemistry, Vol. 81, No. 5, March 1, 2009

(3)

where r is the reflectivity of the mirrors. In this experiment, Qext remained constant with an average of 2.99 ± 0.09 (standard deviation). The value of Qext (calculated by Mie theory) for 400 nm PLS is 3.05. This gives a percentage error of 2% along the entire experiment time. The reported percentage error in pulsed-CRD for Qext measurements was ∼5%. The fluctuation in Qext for low particle concentrations is higher compared to high particle concentrations probably due to larger fluctuations in the particle concentration in the laser beam volume, which will be higher for low aerosol concentration as discussed in detail by Pettersson et. al.20,39 In addition, it is possible that time lags between the concentration measured by the CPC and the corresponding τ measured by the photomultiplier during concentration changes contribute to the error. Therefore, the standard deviation (STDV) of the measured Qext increases, even though the calculated error remains low. The ratio between Rext and the particle number density is expressed by the following equation: Rext ) Nσext

(4)

where σext is the extinction cross section. It is possible to use this correlation to extract σext, as shown in Figure 4. The σext value obtained by this analysis is (3.75 ± 0.01) × 10-9 cm2 with a correlation of R2 ) 0.9974. This is compared to σext ) 3.44 × 10-9 cm2 measured using our pulsed laser system.53 The Miecalculated σext based on n ) 1.598 + i0 is σext ) (3.82 ± 0.01) × 10-9 cm2, implying an error around 2%.

Figure 4. The extinction coefficient (Rext) was measured as a function of particle number density of 400 nm PSL, at different concentrations. The extinction cross section (σext) can be extracted from the slope of the linear fit.

Figure 5. Extinction efficiency (Qext) as a function of the size parameter (χ) for PSL (a). The solid lines represent a Mie fit resulting in n ) (1.60 ( 0.02) + (i0.01 ( 0.03) which is obtained by fitting the experimental data points (9) measured with CW-CRD, while n ) (1.60 ( 0.03) + (i0.01 ( 0.05) was obtained by a fit to experimental data points which were measured with pulsed CRD,7 both at 532 nm. (b) Ammonium sulfate ((NH4)2SO4), the solid lines represent the Mie fit: n ) (1.52 ( 0.01) + (i0.00 ( 0.03) which was obtained using CW-CRD while n ) 1.52 + i0.00 is the theoretical value of the refractive index. (c) Nigrosin, the solid line represents the Mie fit: n ) (1.72 ( 0.01) + (i0.28 ( 0.01). (d) Suwannee River fulvic acid (SRFA), the solid line is the Mie fit yielding n ) (1.520 ( 0.003) + (i0.02 ( 0.01).

Refractive Indices of Various Laboratory-Generated Particles. For additional system validation, the refractive indices (RI) of four aerosol types with known characteristics were measured

(Figure 5): (a) PSL, spherical particles that only scatter at 532 nm; (b) ammonium sulfate ((NH4)2SO4), an important component of atmospheric aerosols, which is a purely scattering Analytical Chemistry, Vol. 81, No. 5, March 1, 2009

1767

Table 1. Comparison between Pulsed-CRD and CW-CRD Measurementsa CW-CRD x

Qext

Std

pulsed CRD RI

x

1.180 1.417 1.771 2.361 2.951

0.495 0.973 1.916 3.159 3.770

0.139 0.114 0.123 0.202 0.215

Polystyrene Spheres Latex (PSL) (1.60± 0.02) + (i0.01 ± 0.03) 1.180 1.417 1.771 2.361 2.951

1.771 2.066 2.361 2.656 2.951

1.401 2.096 2.542 3.206 3.414

0.031 0.128 0.093 0.135 0.160

Ammonium Sulfate (NH4)2SO4 (1.52 ± 0.01) + (i0.00 ± 0.03) 1.772 2.067 2.362 2.657 2.953

1.033 1.180 1.328 1.476 1.771 1.918 2.066 2.213 2.361 2.508 2.656 2.804 2.951

1.204 1.380 1.911 2.287 2.655 2.897 3.123 3.079 3.108 3.043 3.097 3.113 3.093

0.039 0.062 0.050 0.031 0.041 0.047 0.070 0.104 0.104 0.089 0.092 0.066 0.115

1.180 1.476 1.771 2.066 2.361 2.656 2.951

0.458 0.798 1.526 2.063 2.470 3.109 3.410

0.073 0.044 0.059 0.075 0.074 0.081 0.043

Nigrosin (1.72 ± 0.01) + (i0.28 ± 0.08)

1.181 1.329 1.476 1.624 1.772 1.919 2.067 2.214 2.362 2.510 2.657 2.805 2.953

Suwannee River Fulvic Acid (SRFA) (1.520 ± 0.003) + (i0.02 ± 0.01) 1.181 1.476 1.772 2.067 2.362 2.657 2.953

Qext

Std

RI

0.676 1.054 2.115 3.083 3.875

0.003 0.027 0.080 0.111 0.113

(1.60 ± 0.03) + (i0.01 ± 0.05)

1.491 2.026 2.486 3.182 3.467

0.073 0.059 0.034 0.025 0.057

(1.52 ± 0.01) + (i0.01 ± 0.01)

1.401 1.603 1.886 2.242 2.511 2.784 2.900 2.937 2.990 2.982 3.067 3.035 3.145

0.040 0.011 0.069 0.048 0.077 0.043 0.073 0.102 0.069 0.062 0.058 0.105 0.067

(1.65 ± 0.01) + (i0.24 ± 0.01)

0.656 1.343 2.451 2.855 3.276 4.000 4.079

0.035 0.049 0.065 0.047 0.040 0.052 0.071

(1.65 ± 0.01) + (i0.02 ± 0.01)

a For the aerosols SRFA, (NH4)2SO4, nigrosin, and PSL, the extinction efficiency (Qext), standard error, and refractive index (RI) were summarized for different size parameters (x).

material at 532 nm; (c)nigrosin, black organic material with a peak absorption at 532 nm; (d) Suwannee River fulvic acid (SRFA), a weakly absorbing organic material at 532 nm and a commonly used model for complex organic material, such as humiclike substances.7 Figure 5a shows Qext of PSL as a function of the size parameter (πD/λ, where D is the particle diameter and λ is the laser’s wavelength). The solid points represent the experimental results obtained by the CW-CRD, and the green points are from previous experiments performed using a pulsed-CRD system.7 The retrieved refractive index (RI) obtained by the described setup is n ) (1.60 ± 0.02) + (i0.01 ± 0.03) (red line) using only five different sizes for the fit. For comparison, the retrieved RI for pulsed-CRD is n ) (1.60 ± 0.03) + (i0.01 ± 0.05) (blue line) and for 10 different aerosols sizes. The measurement error for individual points in the CW-CRD system is larger than pulsed-CRD, probably due to the time lag between the concentration measured by the CPC and the corresponding τ, measured by the photomultiplier during concentration changes. This may become significant at short measurement times, as applied here. However, the total error (52) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B. R. 1992. (53) Dinar, E.; Taraniuk, I.; Graber, E. R.; Anttila, T.; Mentel, T. F.; Rudich, Y. J. Geophys. Res. 2007, 112. (54) Garvey, D. M.; Pinnick, R. G. Aerosol Sci. Technol. 1983, 2, 477–488.

1768

Analytical Chemistry, Vol. 81, No. 5, March 1, 2009

in the determined RI is smaller compared to the pulsed-CRD system. This is explained by the higher accuracy of Qext in the CW-CRD. A highly accurate RI of n ) (1.52 ± 0.01) + (i0.00 ± 0.03) (Figure 5b), is obtained for (NH4)2SO4 aerosols with only five different sizes (300, 350, 400, 450, and 500 nm). The RI we obtained is in very good agreement with previous measurements of (NH4)2SO4 at 532 nm.37 Nigrosin is used here as a model of a strongly absorbing material. Nigrosin particles were confirmed to be spherical using transmission electron microscopy20 and AFM microscopy.54 For nigrosin, Garvey and Pinnick54 reported a complex RI of n ) 1.67 + i0.26 at 532 nm. Lack et. al37 measured n ) 1.70 + i0.31 using a CRD-AS system. Dinar et. al20 derived a refractive index of n ) (1.65 ± 0.01) + (i0.24 ± 0.01). In Figure 5c, the recent results of nigrosin obtained by our CW-CRD are shown. The black points represent the measured Qext of different sizes, while the red line is the retrieved RI of nigrosin, n ) (1.72 ± 0.01) + (i0.28 ± 0.08), which is in good agreement with the measurement of Lack et al. This measurement demonstrates the ability of the system to measure strongly absorbing aerosols. In Figure 5d we show Qext of seven different sizes of SRFA aerosol obtained using a CW-CRD system and the retrieved RI of n ) (1.520 ± 0.003) + (i0.02 ± 0.01). The reported RI for SRFA is n ) (1.634 ± 0.004) + (i0.02 ± 0.005) at 532 nm.20

While the imaginary part of the complex refractive index matches the earlier observation very well, the real part differs. We attribute this difference to differences in the samples used. The original sample is not available for purchase anymore. This finding, despite the difference, is still in line with the physical and optical properties of the material at this wavelength (weakly absorbing particles at 532 nm). Table 1 summarizes the difference between Qext, standard deviations, and RI measured for SRFA, (NH4)2SO4, nigrosin, and PSL as was measured both by the pulsed laser7 and by our CW-CRD for 532 nm wavelength. In some of the measurements, the standard deviation from the CW-CRD is slightly higher. This is due to a systematic error in the CPC readings that was not corrected by the commercial software used. For the pulsed CRD, the CPC readings were done manually, and therefore, this error was excluded. CONCLUSIONS The capabilities of a newly built CW-CRD system for retrieving the optical properties of aerosols were demonstrated for scattering and absorbing aerosols. This system is substantially smaller, faster, more accurate, and consumes less power compared to an equivalent pulsed CRD system. By using a single mode CW laser, a very clear single exponential decay is obtained, hence providing substantially more accurate analysis of the decay time and all the derived optical

parameters. The new system is stable and consistent over long measurement times with varying particle concentrations. In this experiment, a good inverse relationship between Rext and τ was observed, while the Qext remained constant. However, at low aerosol concentrations (90 cm-3), larger errors exist due to concentration fluctuations in the laser beam volume. In addition, RI values of different aerosols were precisely determined with an excellent agreement compared to previous measurements conducted with pulsed-CRD and other optical systems. Care was taken in designing the system to ensure good mixing of the aerosols in the cavity. The present system’s sensitivity is almost 2 orders of magnitude higher, and its repetition rate is about 4 times faster than other reported CW-CRD systems for aerosol measurements.29 The high qualities of the system, together with its small dimensions and lower power requirements, render it suitable for construction of a portable system more suitable for field and simulation chamber studies. ACKNOWLEDGMENT This work was partially supported by the Israel Science Foundation (Grant 196/08) by the Grand Center and by the Helen and Martin Kimmel Award for Innovative Investigation. Received for review August 24, 2008. Accepted December 6, 2008. AC8017789

Analytical Chemistry, Vol. 81, No. 5, March 1, 2009

1769