Anal. Chem. 1997, 69, 2098-2102
A Renewable Liquid Droplet Method for On-Line Pollution Analysis by Multi-Photon Ionization Vladimir V. Gridin, Iris Litani-Barzilai, Michal Kadosh, and Israel Schechter*
Department of Chemistry, Technion-Israel Institute of Technology, Haifa 32000, Israel
A multi-photon ionization based fast conductance (MPIFC) technique was applied to detect combustion byproduct aerosols. These PAH-polluted aerosols were on-line sampled by means of renewable water microdroplets. The environmental cases considered here have involved such common air contaminants as motor car exhaust gas and cigarette smoke. The possibility of obtaining useful calibration curves has been addressed. Two droplet contamination regimes were clearly observed. These have been argued to be associated with either a volume uniform (i.e., a bulk type) or a surface-favored contamination. The latter regime is possible whenever the increasing droplet contamination extends beyond the solubility saturation of the PAH compounds. Detection limits as low as 1 pg were obtained for pyrene-contaminated renewable microdroplets. Laser-induced multi-photon ionization (MPI) has been frequently promoted as a very sensitive analytical tool with fair selectivity in its resonant mode.1-5 In particular, considerable investigative efforts have been devoted to studying various aspects of MPI-based techniques and their applications to detection of polycyclic aromatic hydrocarbon (PAH) molecules in polar and nonpolar liquids.6-21 In several experimental schemes the, socalled, fast conductance (FC) signal-detecting apparatus were
used. This resulted in a MPI-FC technique that has been successfully applied for highly sensitive detection of PAHs in various systems of environmental interest such as metals,22,23 polar and nonpolar solvents,16,24-27 water,28,29 and soil samples.30,31 In these applications, the data acquisition facilities are of a reasonably low cost as well as of a user-friendly operational simplicity. Recent reports32-35 suggest an efficient application of water microdroplets for renewable in situ sampling, e.g., of such gaseous chemicals as nitrogen dioxide32 and chlorine.33 Various aspects associated with the reproducibility and dynamics of droplet creation have been extensively studied (see, for review, McMillan et al.36-38), following the thorough investigation of those pioneered by Tate39 in 1864. In a recent study of Ogawa and co-workers,29 a high sensitivity MPI-FC method was performed on minute liquid samples. There, in the case of pyrene contamination, a detection limit of 0.3 pg was reached. It is clear, therefore, that the liquid droplet technique offers an easily engineered renewable sampling of gaseous pollutants and aerosols. As was pointed out by Liu and Dasgupta,34 due to the diffusophoresis caused by the Stefan flow, a sampling by means of renewable water droplets would be expected to favor the gaseous pollutants over the air-borne microparticle pollutants. The diffusive tendency of the latter to settle onto a droplet would be impeded by the vapor counterflow (a liquid evaporation from
(1) Letokhov, V. S. Laser Analytical Spectrochemistry; Adam Hilger: Bristol, PA, 1986. (2) Lambropoulos, P., Smith, S. J., Eds. Multiphoton Ionization; Proceedings of the 3rd International Conference, Iraklion, Crete, Greece, September 1984; Springer-Verlag: Berlin, 1984. (3) Laser Applications to Chemical Analysis; Technical Digest Series; Optical Society of America: Washington, DC, 1990; Vol. 2. (4) Brophy, J. H.; Rettner, C. T. Opt. Lett. 1979, 4, 337-339. (5) Rettner, C. T.; Brophy, J. H. Chem. Phys. 1981, 56, 53-61. (6) Yamada, S.; Ogawa, T.; Zhang, P. Anal. Chim. Acta 1986, 183, 251-256. (7) Yamada, S.; Yoshida, S.; Kawazumi, H.; Nagamura, T.; Ogawa, T. Chem. Phys. Lett. 1985, 122, 391-394. (8) Ogawa, T.; Sato, M.; Tachibana, M.; Ideta, K.; Inoue, T.; Nakashima, K. Anal. Chim. Acta 1995, 299, 355-360. (9) Vauthey, E.; Haselbach, E.; Suppan, P. Helv. Chim. Acta 1987, 70, 347353. (10) Holroyd, R. A.; Preses, J. M.; Bottcher, E. H.; Schmidt, W. F. J. Phys. Chem. 1984, 88, 744-749. (11) Yamada, S. J. Photochem. Photobiol. A: Chem. 1991, 62, 45-52. (12) Hall, G. E.; Kenney-Wallage, G. A. Chem. Phys. 1978, 28, 205-214. (13) Sander, M. U.; Luter, K.; Troe, J., Ber. Bunsenges. Phys. Chem. 1993, 97, 953-961. (14) Siomos, K.; Christophorou, L. G. Chem. Phys. Lett. 1980, 72, 43-48. (15) Siomos, K.; Kourouklis, G.; Christophorou, L. G. Chem. Phys. Lett. 1981, 80, 504-511. (16) Ogawa, T.; Kise, K.; Yasuda, T.; Kawazumi, H.; Yamada, S. Anal. Chem. 1992, 64, 1217-1220. (17) Yamada, S.; Sato, S.; Kawazumi, H.; Ogawa, T. Anal. Chem. 1987, 59, 27192721. (18) Johnson, M. E.; Voigtman, E. Anal. Chem. 1992, 64, 551-557. (19) Yamada, S. Anal. Chem. 1989, 61, 612-615.
(20) Voigtman, E.; Jurgensen, A.; Winefordner, J. D. Anal. Chem. 1981, 53, 1921-1923. (21) Voigtman, E.; Winefordner, J. D. Anal. Chem. 1982, 54, 1834-1839. (22) Ogawa, T.; Yasuda, T.; Kawazumi, H. Anal. Chem. 1992, 64, 2615-2617. (23) Kawazumi, H.; Yasuda, T.; Ogawa, T. Anal. Chim. Acta 1993, 283, 111114. (24) Li, Y. Q.; Inoue, T.; Ogawa, T. Anal. Sci. 1996, 12, 691-693. (25) Inoue, T.; Masuda, K.; Nakashima, K.; Ogawa, T. Anal. Chem. 1994, 66, 1012-1014. (26) Yamada, S. Anal. Chem. 1991, 63, 1894-1897. (27) Ogawa, T.; Sumi, S.; Inoe, T. Instrum. Sci. Technol. 1995, 23, 311-315. (28) Chen, H.; Inoue, T.; Ogawa, T. Anal. Chem. 1994, 66, 4150-4153. (29) Ogawa, T.; Sumi, S.; Inoe, T. Anal. Sci. 1996, 12, 455-458. (30) Gridin, V. V.; Korol, A.; Bulatov, V.; Schechter, I. Anal. Chem. 1996, 68, 3359-3363. (31) Gridin, V. V.; Bulatov, V.; Korol, A.; Schechter, I. Anal. Chem., in press. (32) Liu, H.; Dasgupta, P. K. Anal. Chem. 1995, 67, 4221-4228. (33) Cardoso, A. A.; Dasgupta, P. K. Anal. Chem. 1995, 67, 2562-2566. (34) Liu, H.; Dasgupta, P. K. Anal. Chem. 1996, 68, 1817-1821. (35) Barnes, M. D.; Whitten, W. B.; Ramsey, J. M. Anal. Chem. 1995, 67, 418A423A. (36) McMillan, N. D.; Finlayson, O.; Fortune, F. M.; Fingleton, M.; Townsend, D.; McMillan, D. D. G.; Dalton, M. J. Meas. Sci. Technol. 1992, 3, 746764. (37) McMillan, N. D.; Fortune, F. M.; Finlayson, O.; McMillan, D. D. G.; Townsend, D. E.; Daly, D. M.; Fingleton, M.; Dalton, M. G.; Cryan, C. V. Rev. Sci. Instrum. 1992, 63, 3432-3454. (38) McMillan, N. D.; O’Mongain, E.; Walsh, J.; Breen, L.; McMillan, D. D. G.; Power, M. J.; O’Dea, J. P.; Kinsella, S. M.; Kelly, M. P.; Hammil, C.; Orr, D. Opt. Eng. 1994, 33, 3871-3890. (39) Tate, T. Philos. Mag. 1864, 27, 176-180.
2098 Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
S0003-2700(96)01254-1 CCC: $14.00
© 1997 American Chemical Society
Figure 1. Schematics of MPI-FC experimental setup used for renewable water droplet sampling of combustion byproduct aerosols. The principal components of the aerosol contamination circle are also indicated.
the droplet’s surface). Usually, such a mechanism is expected to be by far more significant the smaller the aerosol particles involved. In this regard, however, some poorly dissolving PAH compounds could eventually produce a microparticle-contaminated water droplet. Such a regime occurs whenever the process of droplet contamination has been extended beyond the aqueous solubility limit of a particular PAH compound captured within. In this paper, we address the potential of the renewable water droplet sampling of gaseous/aerosol pollutants for the purpose of MPI-FC analysis. The method is demonstrated using an online droplet contamination process by real environmental targets. These include motor vehicle exhaust gases and cigarette smoke, known to contain PAH pollutants. Pyrene-contaminated droplets were used to address the nonlinearity of the established calibration curves. A need for a proper compensation algorithm for such nonlinear effects is given consideration. The light scattering/ absorbing capacity of microparticle aerosols is shown to be important in this regard. EXPERIMENTAL SECTION Experimental Setup. The major components of the MPIbased fast conductance facility are shown in Figure 1. The MPI processes were induced by a pulsed nitrogen laser emitting 1.5 mJ pulses at 337.1 nm in 0.6 ns (PL2300, Photon Technology International). The beam was focused (to about 1 100 µm2 spot) onto the middle of the sampling droplet by means of an optical setup of several lenses. The measured laser energy at the droplet was 120 µJ/pulse. A positive high-voltage bias was supplied to the electrodes (PS325, Stanford Research Systems) at a 2500 V/0 V, on/off, operational mode. A current preamplifier (Keithley 428) was used in a typical gain range of 107-108 V/A and in the filter-off mode of the device throughout the study. Data were collected by a digital oscilloscope (LeCroy 9300) triggered with a photodiode (rise time of 20 ns). Routine integrations of the resulting photoionization currents were performed, and each readout has been obtained from 50 laser pulses. The zero bias background signals were subtracted from the photoionization current curves, I(t), as well as from the respective charges, Q, (represented as the area under these curves) that are dealt with in the Results and Discussion section.
A free-hanging water (triply deionized) sampling droplet was produced in a renewable manner by means of a standard 50 µL (Hamilton Co., Reno, NV) syringe. The droplet size [under the usual laboratory conditions, i.e., room temperature (20 ( 2 °C), ambient air pressure (750 ( 5 Torr) and composition, and ambient laboratory moisture level (65 ( 10%)] was defined by a small (0.7 mm diameter) copper wire loop (30 AWG). This was positioned just beneath the injection hole of a stainless steel needle (supplied with the syringe). Under stationary conditions, a pendant droplet with a mean sampling volume, Vs = 14 µL, was obtained. The copper loop and the needle were in electrical contact with the negative electrode that, in turn, was attached to the current preamplifier, as shown in Figure 1. Sample Preparation. Organic traces present in (a) the exhaust gases of an idle operating gasoline-powered motor vehicle engine and those found in (b) a free-burning cigarette smoke were routinely collected. They entered, on-line, an aerosol reservoir. There, a dynamic equilibrium of combustion byproduct aerosol contamination was reached. Then a step pump was used to introduce the contaminated air into the measurement chamber in a continuous manner. Volume doses of 20 mL each (consisting of the above aerosols) were used to contaminate the overall (ionization chamber + mixing vessel + air step pump + connecting pipes) experimental volume, VE ) 4286 mL. The combined weight of combustion byproducts (in the form of aerosol particles) found in each such dose, contained, respectively, about 0.7 µg/L in the former and 4.6 µg/L in the latter case of the aforementioned gaseous pollutants. Clearly, in the above procedure, the water droplet was used to continuously collect PAH as aerosols. Pyrene-contaminated hexane solutions were prepared from an analytical grade n-hexane (Frutarom Ltd.) and pyrene (99%, Aldrich) without further purification. Pyrene concentration varied in the range from 0.1 ng/mL to 1 mg/mL. The depositions, made in 2 µL shuts of pyrene/hexane solutions, were brought into physical contact with the water droplet. In the case of 0.1 ng/ mL, this produces, e.g., a 0.2 pg pyrene contamination. By virtue of surface tension forces, such a microshut gets pulled in by the droplet. Hexane quickly evaporates and a pyrene-contaminated water droplet results. Since the reported saturation water solubility of pyrene under normal conditions is of the order of 0.1-0.2 ppm,40 in this way we also were able to produce pyrene microparticle-contaminatedc water droplets. RESULTS AND DISCUSSION In Figure 2 we present typical MPI-FC data that correspond to an on-line contamination of such renewable water droplets. Pollution of the ambient air was achieved by either (a) motor vehicle exhaust gas or (b) cigarette smoke. The observed fast initial rise, which is followed by a gradual falloff of a photoionization current profile, is similar to welldocumented features22-31 of the usual readouts in various FC experimental setups. Now, in order to establish a quantitative calibration curve for such a sampling system, we proceed similar to the recently reported procedure.30,31 The method has been introduced for pyrene-contaminated soil probes. There, a number of sporadic effects, associated with a hardly reproducible sample preparation (40) Futoma, D. J.; Smith, S. R.; Smith, T. E.; Tanaka, J. Polycyclic Aromatic Hydrocarbons in Water Systems; CRC Press, Inc.; Boca Raton, FL, 1981; p 16, Table 2, and references therein.
Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
2099
Figure 2. Photoionization current, I ) I(t), obtained by MPI-FC from renewable water droplet for gasoline-powered motor vehicle exhaust gas and cigarette smoke. Note that the area under the respective I(t) curves corresponds to the detected photocharge, Q, which, in turn, is proportional to the amount of photobsorbing organic traces present in aerosol probes excited by 337.1 nm of the pulsed nitrogen laser.
routine, were dealt with and reasonably well accounted for. A sample matrix-independent calibration curve was shown to exist there too. The renewable water droplet-sampled aerosols are probed here by a monochromatic laser energy flux, J. Let the total detected photoionization charge, Q, result from the background contribution, QB, and from the PAH aerosol concentration-dependent term, QP. The former is due to the aerosol sampling liquid (water in our case) and those ambient air contaminants whose concentration level in the droplet is not monitored during the trce detection run. Let nR be a variable number density of a type R aerosol; hereon R stands for the exhaust gas, cigarette smoke, or pyrene. With ∑fβ ) 1 we write QP ) nR∑fβJkΓβk. Here we have assumed that, within a particular nR, an off-resonance MPI process occurs for each trace β there. Clearly, the ionization process order, k, is expected to be β dependent also. The same is also true for Γβk. This latter corresponds to the ionization cross section for the kth-order process of a particular pollution constituent, β. In the above, fβ’s represent the fractal number densities of various UV-active organic traces in nR. Quite obviously, for pyrene, i.e., a one-component aerosol pollutant, fβ ) 1. In the above, the summations are over all the organic traces (molecules), β, that constitute the investigated aerosol pollutant. The latter could be of a very complex composition as are the environmentally relevant pollutants used here. Then one obtains
∑f J Γ
Q ) QB + QP ) QB + nR
k
β
k
β
(1)
Worth noting is that an off-resonance MPI-FC method used here does not provide the selectivity needed to separately trace down and specify each one of the β’s present in the aircontaminating aerosols. Despite this, a renewable droplet technique allows for the very presence of such pollutants to be conveniently (on-line) detected. In this regard, we define30 a function Ψ ≡ Q - QB, whereas Ψ* is the value of Ψ obtained at the maximal dose, nR*, of a 2100 Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
Figure 3. Plot of normalized photocharges, Ψ/Ψ*, as a function of normalized laser beam energy, J/J*. The slope of 2 line indicates an effective two-photon process.
particular contaminant R; in terms of aerosol weight per liter of air units these were nR* ) 21 µg/L and nR* ) 64 µg/L for the exhaust gas and cigarette smoke contamination, respectively. The corresponding Ψ* values were 79 fC in the former and 248 fC in the latter case, respectively. For pyrene in water, these figures were nR* ) 144 mg/L and Ψ* ) 3.38 nC. In Figure 3, we present in logarithm coordinates a plot of Ψ/Ψ* as a function of the normalized laser intensity. The slope of 2 compares well with the data, indicating an effective two-photon ionization process, as would be for the most of PAH contaminants. Next, with the data presented as Ψ/Ψ* versus nR/nR*, the sample-independent calibration curves were obtained. The results are shown in Figure 4a, for both the car exhaust- and cigarette smoke-contaminated air intakes. At fixed J, according to eq 1, such plots should produce calibration curves of slope 1. Despite one’s anticipation, the slope of the best linear fit through the data is smaller than 1. The most likely explanation of such an observation seems to be related to the very nature of the on-line sampling of combustion byproduct aerosols. Indeed, renewable water microdroplets could dissolve only a limited amount of PAH aerosols. Some of the sampled materials would remain in the form of PAH microparticles. These microparticles could be either uniformly distributed (suspended) inside the sampling droplet or, to a certain degree, favor floating near its outer shell. Additionally to such PAH microparticles, one cannot avoid sampling nonorganic as well as MPI-passive aerosols. Noteworthy is that, with the ambient condition performing MPI-FC facility, an ever increasing amount of ambient air aerosols (nonorganic particulates) should get accumulated within the droplets, anyway. All these would act upon the incoming photon flux as a light-scattering (absorbing) matrix. Therefore, the ambient condition mode of operation with such sampling droplets ought to bring about a Beer-Lambert type decrease of the energy flux available for the MPI processes. Such behavior requires further investigation. In this regard, we also note that, even for the same trace level, a variation in the irradiation angle of the incoming laser beam might result in an alteration of the calibration curve. This has been recently reported by Ogawa et al.29 for pyrene in liquids. Research efforts in this direction are currently undertaken in our laboratory. The goal would be to introduce appropriate compen-
a
b
Figure 4. (a) PLot of Ψ/Ψ* versus nR/nR* for the combustion byproduct aerosols captured by the renewable water droplets; nR* ) 21 µg/L (exhaust gas), nR* ) 64 µg/L (cigarette smoke). Both are given in terms of aerosol weight per liter of air units. The corresponding Ψ* values are 79 and 248 fC in the former and the latter case, respectively. (b) Same as Figure 4a. Here, for pyrene in water, nR* ) 144 mg/L and Ψ* ) 3.38 nC. The broken line (slope equal to that of Figure 4a) that is drawn through the low-concentration part of the pyrene data points out the similarity of the MPI-FC responses for the studied aerosols and water droplet depositions of pyrene. The arrow indicates the saturation solubility entry of pyrene in terms of nR/nR* units. The slope variation (solid lines, best linear fits drawn for the respective data) around the solubility saturation value is believed to be associated with the variation in the droplet contamination regime, i.e., from the bulk (low-concentration part) toward the surface (high-concentration part) one.
sation algorithms for the aforementioned sample-independent calibration curves. Clearly, at first, such investigation efforts should attempt studying the simplest system possible. In order to partially comply with these requirements, pyrene was chosen here to play a model role of a single-PAH component aerosol. Hence, a final entry of this report is to illustrate the above stipulated point regarding the influence of the undissolved aerosols on the sample-independent Ψ/Ψ* calibration curves. To be precise, the observed slope reduction tendency of those was found to be supported by microparticle formation in pyrene-contaminated water droplets. Indeed, upon reaching an equilibrium solubility of pyrene in water, one would anticipate that a further deposition of pyrene onto droplet should result in appearance of polycrystalline pyrene microparticles. Given a smaller then that of water material density of pyrene, their space distribution is expected to favor the droplet’s outer shell.
In any case, besides contributing to MPI processes, such microparticles should act as light scatterers/absorbers there. In accordance with the above, the slope of the calibration curve is expected to be altered at the onset of aqueous solubility of pyrene. The slope’s decrease is anticipated there too. The slope should go flat, we believe, as a “monolayer” coverage of the droplet’s outer shell would be established. At such instant, the laser irradiation would probe the outer shell-deposited pyrene only and the situation should eventually evolve into a saturated one. In Figure 4b, we present experimental results on pyrene that was used to externally contaminate water droplets. In a sense, our data are complementary to that of Ogawa et al.29 Instead of varying laser alignment onto the droplet, however, we have kept this fixed. In turn, the concentration of pyrene that was forced to settle there was increased; see Sample Preparation subsection above. Surface evaporation of hexane eventually produced pyrenecontaminated water droplet, and due to the low water solubility of pyrene, the microparticle contamination regime was effectively forced onto the sampling system. A sudden change in the calibration slopes is evident in this figure (an arrow specifies the saturation solubility of pyrene in water, which is taken here as the mean value of 0.144 ppm among the various reported entries40). In Figure 4b we also drew the calibration slope of the aerosols (compare with Figure 4a) through the low concentration part of pyrene-related data. Since this region corresponds to a low-microparticle-concentration regime for pyrene (only the ambient air aerosols should be increasingly accumulated there) and by virtue of the similarity of these slopes, we conclude that the same regime should exist for the material quantities of the combustion byproduct aerosols studied. CONCLUSIONS In this paper, a potential of renewable sampling of gaseous/ aerosol PAH compounds by water droplets is addressed. An MPIFC facility was used. The principal action of this setup is based upon detecting short-lived photoionization currents. These are created when a pulsed laser flux is used to probe organic traces in a given host matrix. The detection system is both simple and low-cost, suggesting possible development of an in situ instrumentation. The environmentally important aerosols used here were those found in the typical motor vehicle exhaust gas and cigarette smoke. Both of these complex pollutants contain many PAH traces and their overall effective response to a fixed line laser excitation is more likely to exhibit an off-resonance MPI features. In the case of “pyrene aerosol”, the calibration curve obtained for water droplet extends to about 1 pg levels. A tendency of microparticle formation to alter the slopes of calibration curves was partially addressed by studying such pyrene-contaminated water droplets. It has been shown that a “universal” calibration curve can be obtained for PAHs in the low-contamination regime (where “bulk” droplet contamination is achieved). At higher concentrations, where surface deposition is relevant, a lower calibration curve slope is obtained. The feasibility of on-line analysis by microdroplet sampling followed by MPI detection has been proven, although further research and engineering development is still necessary. Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
2101
ACKNOWLEDGMENT This research was supported, in part, by the Israeli Ministry of the Environment, by the James-Franck Program for Laser Matter Interaction, and by the Technion V.P.R. fund. V.V.G. acknowledges partial financial support provied by the Israeli Ministry of Absorbtion to the scientists regarded as return citizens.
2102
Analytical Chemistry, Vol. 69, No. 11, June 1, 1997
Received for review December 10, 1996. Accepted March 6, 1997.X AC961254Z
X
Abstract published in Advance ACS Abstracts, April 15, 1997.
