Environ. Sci. Technol. 2003, 37, 4737-4742
Real-Time PCB Monitoring Using Time-of-Flight Mass Spectrometry with Picosecond Laser Ionization YOSHIHIRO DEGUCHI,* SHINSAKU DOBASHI, NORIHIRO FUKUDA, AND KATSUHIKO SHINODA Nagasaki Research and Development Center, Mitsubishi Heavy Industries, Ltd., 5-717-1 Fukahori-machi, Nagasaki 851-0392, Japan MASATOSHI MORITA (AIA) National Institute for Environmental Studies, Onogawa 16-2, Tsukuba, Ibaraki 305-8506, Japan
The study demonstrates the applicability of laser ionization time-of-flight mass spectrometry for real-time measurement of polychlorinated biphenyls (PCBs). Picosecond 266-nm laser light ionization reduced fragmentation and provided very high PCB detection sensitivity. This high sensitivity has advantages in terms of real-time monitoring capability as compared to the conventional GC-ECD or GC-MS methods, which require at least several days for the analysis of PCBs. Detection sensitivity of under 0.01 mg/Nm3 was achieved with a 1-min measuring time; this sensitivity is superior to the exhaust gas control guideline of 0.15 mg/Nm3 by a factor of 10. A prototype PCB monitoring device has been developed and tested in a pilot PCB treatment plant. The 1-min detection time represents a substantial advance in the monitoring of exhaust gas and the workplace atmosphere in accordance with safety regulations.
Introduction PCBs (polychlorinated biphenyls) are a type of chlorinated aromatic hydrocarbons widely used in condensers and other applications prior to 1970 but are now recognized as serious environmental pollutants. The Japanese PCB Treatment Special Equipment Law, which came into effect on July 15, 2001, made PCB disposal mandatory. Many firms and research laboratories are thus engaged in research and development work on PCB disposal and processing. There are several methods to dispose of PCBs, such as hydrothermal, ultraviolet light, plasma, and Ni-Cu catalytic decompositions. Mitsubishi Heavy Industries, Ltd. (MHI) has also developed a treatment plant for its own stored PCBs and begun hydrothermal decomposition of PCBs from December 2000. PCB processing requires the treatment of PCBs (i.e., breakdown into harmless substances) as well as the cleaning of PCB containers and the processing of contaminated paper, wood, and other exposed/contaminated substances. It must be ensured in the PCB treatment process that PCBs are not released into exhaust gas or the atmosphere of the work environment. The conventional analysis method requires 2∼3 d for gas sampling and the concentration process. Faster PCB monitoring methods are needed from the standpoint of safety management. * Corresponding author telephone: +81-95-834-2322; fax: +8195-834-2165; e-mail: yoshihiro•
[email protected]. 10.1021/es034257j CCC: $25.00 Published on Web 09/10/2003
2003 American Chemical Society
The objective of the research reported here was to develop a measurement system capable of measuring PCBs within 1 min for exhaust gas and the atmosphere of the disposal work environment, thus leading to the development of the laser ionization TOFMS (time-of-flight mass spectrometry) method (1). This technique uses picosecond short-pulse laser light, which reduces the fragmentation of PCBs and raises PCB ionization efficiency. Detection sensitivity of 0.01 mg/ Nm3 was achieved, which is less than the exhaust gas control guideline of 0.15 mg/Nm3 by a factor of one-tenth. A prototype PCB monitoring apparatus was also constructed using this technology, and PCB monitoring was field tested in MHI's treatment plant for PCB hydrothermal decomposition exhaust gas.
Principle of PCB Measurement In this study, the TOFMS method with resonance-enhanced multiphoton ionization (REMPI) was used with direct introduction of exhaust gas into the vacuum chamber. Development of this method has been pursued in recent years as a means of high sensitivity detection of chlorinated hydrocarbons and dioxins (2-5). The principle of ionization underlying the REMPI TOFMS system is well-described in ref 1. Chlorinated aromatic molecules such as PCBs absorb certain wavelengths, particularly ultraviolet light of 250300 nm (2) and become excited to high energy levels. When light having high energy density (e.g., laser light) is introduced, the excited molecules then absorb this light as well and become ionized. This process is harnessed to ionize PCBs, after which TOFMS is used to separate and detect PCBs from impurities. REMPI methods for analyzing chlorinated hydrocarbons have been developed using a pulsed ultrasonic jet technique (2-5). In this technique, the measurement molecule is subjected to cryogenic cooling in the ultrasonic jet, such that the molecules are cryogenically collected at a specific energy level and limited to a resonance-enhanced ionization wavelength, thereby improving selectivity and achieving high sensitivity. However, two problems arise: (i) To tune the laser wavelength to the resonance frequency of the molecule, a wavelength-tunable laser is required. Durability generally becomes an issue when this type of laser is used in plant monitoring equipment. It should also be noted that PCBs are complex mixtures of isomers and congeners (groups of PCBs having different numbers of chlorine atoms) with different absorption wavelengths that
FIGURE 1. Structure of the ion trap ring electrode. The ion trap is composed of a hyperboloid-shaped electrode and is capable of trapping ions with a certain mass number at the center of the electrode. VOL. 37, NO. 20, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4737
FIGURE 2. Trajectory of laser-excited ions to the ion trap. The number of ions trapped and the orbit are dependent on the ion trap operating conditions. Convergence of ions at the center of the ion trap enhances the detection sensitivity. (a) Laser-excited ion trajectory through ion collection optics. (b) Ion locus inside the ion trap (optimal dynamic trapping, ion generated at point A). (c) Ion locus inside the ion trap (improper dynamic trapping, ion generated at point A). compromise the selectivity derived from the resonant wavelength. (ii) In the case of chlorinated hydrocarbons such as PCBs, the resonance ionization level easily reaches a dissociation energy level when the number of chlorine atoms on the PCB molecules increases; this means that fragmentation starts to govern rather than the ionization process and that measurement sensitivity declines. In this project, to resolve these difficulties, a measurement technique was developed using an ion-trap type TOFMS method in conjunction with a dynamic trapping technique (6-11). Picosecond level short-pulse laser light is used in order to reduce the fragmentation process in high -chlorine PCBs. By using short-pulse laser light in this manner, dissociation-induced reduction in sensitivity is countered because resonance ionization occurs more quickly than the dissociation process (13). The ion trap system also works to select target ions from other unnecessary ions. It functions as a device for the selection of mass and is capable of storing ions with certain mass numbers using the electric field inside the trap. It is also possible to distinguish ions with the same mass number using MS-MS technology (14). The use of short pulse laser ionization with an ion trap enables the enhancement of both sensitivity and selectivity. The laser excited ions are introduced into an ion trap, which operates by applying ac and dc electric fields to a hyperboloid-shaped electrode as shown in Figure 1. When dc voltage (U) and ac voltage (V cos ωt) are applied within this electrode, the electric potential (Φ) at point (r,z,t), describing cylindrical coordinates within the trap, can be expressed by (6, 14):
Φ(r,z,t) ) Φ0(t)[r 2 - 2z2]/2r02 + Φ0(t)/2
(1)
Here, for Φ0(t) ) U + V cos ωt, r0 is the minimum radius of the cell or 2z02 ) r02. In this electric field, the movement of a charged particle having mass m and electrical charge (e) becomes the same in both the r and z directions and can be expressed according to the Mathieu differential equations represented as follows (6, 15): 2
2
2
d r/dt + (e/mr0 )(U + V cos ωt)r ) 0
(2)
d2z/dt2 - (2e/mr02)(U + V cos ωt)z ) 0
(3)
The trapped ions stay in a limited domain represented by the stable solutions to formulas 2 and 3. Considering that the cell within the ion trap apparatus is of a certain size (minimum radius r0 of the hyperboloid), the trapped ion mass depends on the frequency of the ac voltage. Accordingly, species selectivity can be assigned to the ions to be trapped. 4738
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 20, 2003
FIGURE 3. Schematic diagram of the PCB monitoring apparatus. The apparatus was composed of a picosecond laser, an ion trap TOFMS device, an ion counter, and a computer. The operation of this system was automated for plant monitoring use. The trajectories of the laser-excited ions calculated using SIMION 6.0 (16) are shown in Figure 2. Panel a shows the movement of laser-excited ions, while panels b and c are ion loci inside the trap under different trapping conditions. The voltage of the ion collection optics and the dynamic trapping timing affect both the number of ions trapped and the orbit of ions inside the trap cell. Optimization of these parameters was performed using both simulated and experimental results. After the ions are held in the trap for a certain time, they are released into the TOFMS chamber. Taking the ion mass as m, the ionic valency as z, the ion distance of flight as L, the time-of-flight as t, and the acceleration electrode potential as V, the following formula can be established using the law of energy conservation:
1 L zV ) m 2 t
2
()
m)
2V t L2
(4)
Using formula 4 above, the mass and number density of the ions can be measured by means of the time-of-flight to reach the detector and the ion signal intensity, respectively.
Experimental Section The experimental apparatus in this study is shown in Figure 3. The prototype monitoring system was developed for the demonstration of industrial use. Laser light having a wavelength of 266 nm and a pulse width of 100 ps (SL-212: 10 Hz,
FIGURE 4. PCB measurement results using nanosecond and picosecond laser ionization. The laser pulse energies were set to the same level (10 mJ/pulse) in both cases. The use of a shorter pulse laser ionization source reduces the fragmentation of PCBs and enhances the detection sensitivity. 10 mJ/pulse at 266 nm; SL-2210: 250-500 Hz, 0.4 mJ/pulse at 266 nm, Ekspla) was used for PCB ionization. The sampled gas with PCBs was introduced into the ionization chamber by means of a capillary with a 0.5 mm i.d. column (see Figure 3a). Ionized molecules were introduced into the ion trap cell (C-1251, R. M. Jordan Co., Inc.) using a dynamic trapping method and were released using a high-voltage pulser (PVM4140, DEI, V4 ) +1500 V and V5 ) -1500 V in Figure 3a) after a specific time (1.5-50 µs depending on the laser frequency) into the reflectron type TOFMS chamber (C-1251, R. M. Jordan Co., Inc.; flight length: 500 mm). The ion collection optics (V1 ) 3, V2 ) 3, and V3 ) 1.5 in Figure 3a), the dynamic trapping timing, and the laser power were set to suppress the fragmentation of PCBs. The pressure of the ionization chamber was set at 10-2 Pa and that of the flight chamber was set at 10-6Pa using 0.5 and 0.25 m3/s turbo pumps. The ions were detected by the ion detector (25 mm MCP Z-gap detector, R. M. Jordan Co., Inc.), and their signals were transformed into current output after which they were amplified (SR-445, SRS) and digitized by a counter (SR-430, SRS) and stored as data in a computer. The design of the ion trap allows helium to be supplied inside the ion trap cell. The introduction of helium facilitates the cooling of the trapped ions, thus improving the trapping rate and signal time resolution. PCB vaporization and gaseous transport methods were adopted for PCB standard samples. The PCBs were placed in containers held at a constant temperature (typically 300350 K), were mixed with nitrogen at their vapor pressures, and then introduced into the apparatus. Dichlorotoluene at
FIGURE 5. PCB mass spectra using 266-nm picosecond laser ionization. Two to six chlorine PCBs were successfully detected using picosecond laser ionization. Higher sensitivity was achieved for PCBs having smaller numbers of chlorine atoms. (a) KC-300; (b) KC-500. the ppb level was also added as an internal standard for the system. By diluting the PCBs generated, it is possible to adjust the PCB concentration of standard samples in the range of 0.01-1 mg/Nm3 (PCB concentration differs according to the number of chlorine atoms on the PCB molecules).
Results and Discussion PCB Standard Gas Measurement. Figure 4 shows measurement results using the nanosecond and picosecond laser ionizations for PCBs introduced into the apparatus as standard PCB gas. Both of the laser pulse energies were set at 10 mJ/pulse. The mass number is on the horizontal axis, and the ion signal intensity is on the vertical axis. KC-300 (Kaneka, standard PCB sample containing mainly 2-4ClPCBs) was used as the PCB standard. Measurement time was 1 min, and the concentration of PCBs was evaluated by comparison with the signal from dichlorotoluene (the internal standard). Each PCB concentration was confirmed using conventional GC-MS analysis, after solvent absorption of the measured gas sample. Utilization of a laser with a pulse time width of 100 ps clearly resulted in a stronger PCB signal than for a laser with a pulse time width of 5 ns. Using the picosecond pulse width, the increases in the PCB signal were over 10 times for 2-4Cl-PCBs. These results suggest that the use of a laser light source with a shorter pulse time period reduces the fragmentation of PCBs as well. PCB fragmentation was unavoidable in nanosecond laser ionization and, regardless of the laser pulse energy, was dominant as compared with the ionization process. The mass spectra detected using VOL. 37, NO. 20, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4739
FIGURE 6. Comparison between the laser-based system and GC-MS method. A satisfactory proportional relationship was confirmed between the two methods. The conventional gas sampling + GC-MS method also encompasses measurement errors, and the standard deviations between both sets of results were within 30%. This agreement is generally satisfactory for plant monitoring applications: (a) 2Cl-PCB (dichlorinated biphenyl), (b) 3Cl-PCB (trichlorinated biphenyl), (c) 4Cl-PCB (tetrachlorinated biphenyl), (d) 5Cl-PCB (pentachlorinated biphenyl), and (e) 6Cl-PCB (hexachlorinated biphenyl). KC-300 and KC-500 (Kaneka, standard PCB sample containing mainly of 4-6Cl-PCBs) are shown in Figure 5. As indicated by the figure, satisfactory measurement of 2-6Cl-PCBs was confirmed. 4740
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 20, 2003
Figure 6 presents a comparison between the PCB standard gas measurement results using the laser-based experimental apparatus and the conventional gas sampling + GC-MS method. A satisfactory proportional relationship was con-
FIGURE 7. PCB treatment plant multi-point monitoring system diagram. The system employed a multi-sampling line system, enabling the monitoring of gases in several stages of container treatment and PCB disposal processes. firmed between the two methods. The PCB concentrations of the relevant chlorinated PCBs were in the range under 0.001 to over 1.0 mg/Nm3. For 1 min and S/N ) 3, the theoretical detection limits are 0.00011 mg/Nm3 (approximately 10 pptv) for 2Cl-PCB, 0.00048 mg/Nm3 (approximately 40 pptv) for 3Cl-PCB, and 0.00080 mg/Nm3 (approximately 60 pptv) for 4Cl-PCB, indicating that sensitivity of 0.01 mg/ Nm3 PCB concentration is achievable. Some of the data indicated a certain amount of deviation between the two methods. The deviation trend was the same in the measured PCB congeners, possibly caused by uncertainty of the gas sampling and preconcentration processes in the conventional method. The standard deviations between both results were within 30%. Considering the possible measurement errors in the conventional gas sampling + GC-MS method, this agreement is generally satisfactory for plant monitoring applications. Application to Exhaust Gas from a PCB Treatment Facility. A PCB measurement room was newly installed in a PCB treatment plant. The PCB monitoring equipment developed in the project reported here was installed in this pilot plant as shown in Figure 7, which shows the multisampling line system diagram for PCB treatment. The main components of the measured gases were N2, CO2, O2, and H2O in the hydrothermal decomposition exhaust gas and air (N2 and O2) in the workplace atmosphere. These gases contained several trace components attributed to insulating oils, cleaning chemicals, and their decomposed materials on the order of mg/Nm3. The concentrations of these trace components were dependent on the plant operating conditions and varied by several orders of magnitude. Some trace components have masses that are similar to those of PCBs, and this method can separate PCBs from most of these components using resonance ionization. On the other hand, the conventional method requires a time-consuming process for the separation of PCBs from these components. Figure 8 provides examples of hydrothermal decomposition exhaust gas measurement results obtained from the PCB monitoring system installed in the PCB treatment plant. The horizontal axis represents time, while the vertical axis shows PCB concentration. As indicated in Figure 8, the PCB concentration in exhaust gas in a low concentration (normal)
FIGURE 8. Measurement results for PCB hydrothermal decomposition exhaust gas. The concentrations of PCBs as well as trace compositions are dependent on the plant conditions. Real-time PCB monitoring under practical conditions was demonstrated using this method without serious interference from the trace compositions in the exhaust gas. (a) Low concentration gas. (b) High concentration gas. VOL. 37, NO. 20, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4741
case was under 0.01 mg/Nm3, which agreed well with the measurement results for 0.001 mg/Nm3 obtained with the conventional method. In the case of high PCB concentration (experimental condition), 3Cl-PCB concentration varied between 0.1 and 0.2 mg/Nm3 and 4Cl-PCB varied between 0.01 and 0.02 mg/Nm3, which showed good respective agreement with the results for 0.19 and 0.02 mg/Nm3 for 3Cl- and 4Cl-PCB, respectively, obtained using the conventional method. The concentrations of PCBs as well as trace compositions were dependent on the plant conditions and tended to change several orders of magnitude. Using the actual exhaust gas from a PCB treatment plant demonstration, it was demonstrated that PCB monitoring could be possible without interference from either the main gas compositions or the minor coexisting substances. Several methods are proposed for the rapid detection of PCBs such as enzyme immunoassay and low resolution GC-MS methods. It is, however, difficult for them to monitor gas-phase PCBs within 1 min because these methods require gas sampling and preconcentration of PCBs. In the future, the long-term reliability of the monitoring system will be verified, and practical applications will be pursued as an innovative tool for safety management in various types of PCB treatment plants.
Acknowledgments We express our sincere appreciation to Professor David M. Lubman of the University of Michigan for his expert advice on ion trap mass spectrometry.
Literature Cited (1) Lubman, D. M., Ed. Lasers and Mass Spectrometry; Oxford University Press: Oxfored, 1990.
4742
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 20, 2003
(2) Tembreull, R.; Sin, C. H.; Li, P.; Pang, H. M.; Lubman, D. M. Anal. Chem. 1985, 57 (7), 1186. (3) Boesl, U. J. Phys. Chem. 1991, 95, 2949. (4) Zimmermann, R.; Boesl, U.; Weickhardt, C.; Lenoir, D.; Schramm, K.-W.; Kettrup, A.; Schla, E. W. Chemosphere 1994, 29 (9), 1877. (5) Oser, H.; Thanner, R.; Grotheer, H.-H. Chemosphere 1998, 37 (9), 2361. (6) March, R. E.; Hughes, R. J. Quadrupole Storage Mass Spectrometry; John Wiley & Sons: New York, 1989. (7) Cooks, R. G.; Kaiser, R. E. Acc. Chem. Res. 1990, 23 (7), 213. (8) Mather, R. E.; Waldren, R. M.;Todd, J. F. J.; March, R. E. Int. J. Mass Spectrom. Ion Processes 1980, 33 (3), 201. (9) Eiden, G. C.; Garrett, A. W.; Cisper, M. E.; Nogar, N. S.; Hemberger, P. H. Int. J. Mass Spectrom. Ion Processes 1994, 136 (2), 119. (10) Qin, J.; Steenvoorden, R. J. J. M.; Chait, B. T. Anal. Chem. 1996, 68 (10), 1784. (11) Huang, P.; Wall, D. B.; Parus, S.; Lubman, D. M. J. Am. Soc. Mass Spectrosc. 2000, 11 (2), 127. (12) Wall, D. B.; Kachman, M. T.; Gong, S. S.; Parus, S. J.; Long, M. W.; Lubman, D. M. Rapid Commun. Mass Spectrom. 2001, 15 (18), 1649. (13) Matsumoto, J.; Lin, C.-H.; Imasaka, T. Anal. Chem. 1997, 69 (22), 4524. (14) Qian, M. G.; Lubman, D. M. Rapid Commun. Mass Spectrosc. 1996, 10 (15), 1911. (15) Dehmelt, H. G. Advances in Atomic and Molecular Physics Vol. 5; Bates, D. R., Esterman, I., Eds.; Academic Press: New York, 1969; pp 109-154. (16) He, L.; Wu, J.-T.; Parus, S.; Lubman, D. M. Rapid Commun. Mass Spectrom. 1997, 11 (13), 1467.
Received for review March 21, 2003. Revised manuscript received August 4, 2003. Accepted August 4, 2003. ES034257J
