Detection and Quantification of Aromatic Contaminants in Water and

types of soil investigated (sand, humus, clay) showed differences in the desorption efficiency, none of them produced mass spectrometric interferences...
0 downloads 0 Views 150KB Size
Download PDF
Anal. Chem. 2002, 74, 4861-4867

Articles

Detection and Quantification of Aromatic Contaminants in Water and Soil Samples by Means of Laser Desorption Laser Mass Spectrometry Christian Weickhardt,* Karen To 1 nnies, and Daniel Globig

Department of Physical and Analytical Chemistry, Brandenburg University of Technology, Erich-Weinert-Strasse 1, 03044 Cottbus, Germany

The combination of laser desorption of untreated soil samples and subsequent selective laser ionization followed by time-of-flight mass analysis results in an ultrafast technique for the quantitative detection of aromatic contaminants in soil samples. The method allows for high sample throughput, because the complete measurement is finished within about 1 min. Although the different types of soil investigated (sand, humus, clay) showed differences in the desorption efficiency, none of them produced mass spectrometric interferences when an ionization laser wavelength of 266 nm was used. Quantification was carried out by relative measurement with respect to an internal standard and gave satisfactory results over 4 orders of magnitude of analyte concentration. Although the detection of polycyclic aromatic hydrocarbons could successfully be carried out using nanosecond laser pulses, the quality of the mass spectra obtained for labile substances, for example, nitrotoluenes, could be greatly improved by the use of ultrashort pulses in the subpicosecond range. With the preliminary setup, detection limits in the low micrograms-per-gram range were achieved. The identical setup can be used for the analysis of liquids, in particular, water, when the soil sample is replaced by a solid, porous adsorber medium onto which the sample is applied. Activated carbon proved to be a useful adsorber for IR laser desorption, whereas for the UV, granular clay or lime/sand mixtures are preferable. The characterization of a military or industrial hazardous waste site is an important precondition for its successful remediation. Because of the inhomogeneous distribution of the contaminants typical for hazardous waste sites, usually a large number of soil * Corresponding author. Fax: +49-355-693985. E-mail: [email protected]. 10.1021/ac020189s CCC: $22.00 Published on Web 08/22/2002

© 2002 American Chemical Society

samples have to be analyzed in order to obtain a clear picture of the situation. As far as organic analysis is concerned, such a vast number of individual samples can hardly be dealt with by conventional soil analytical techniques, because most of them require time-consuming, costly sample preparation steps.1 As a result of this situation, there exists a need for new techniques for soil analysis, in particular, screening techniques that allow high sample throughput and can avoid lengthy preparation procedures. A further goal is the reduction of the number of sample manipulations and the corresponding sources of quantitative error. A direct analysis of the contaminants within a short time frame would be desirable. The demands on a suitable technique for the detection of organic contaminants in soil, for example, selective trace detection in a complex matrix, multicomponent capability, high sample throughput, and simple sample preparation, point toward laser mass spectrometry2,3 as a possible alternative. This rather new analytical technique combines the selectivity and high efficiency of resonant multiphoton ionization4,5 (REMPI) with the speed and high sensitivity of time-of-flight mass spectrometry.6,7 It is capable of performing multicomponent analysis with high selectivity in even complex mixtures8 thereby maintaining high temporal (1) Soil Analysis, 2nd ed.; Smith, K. A., Ed.; Dekker: New York, 1991. (2) Lubman, D. M. Lasers and Mass Spectrometry; Oxford University Press: New York, 1994. (3) Weickhardt, C.; Moritz, F.; Grotemeyer, J. Eur. Mass Spectrom. 1996, 2, 151-160. (4) Letokhov, V. S. Laser Photoionization Spectroscopy; Academic Press: Orlando, FL, 1987. (5) Powis, I.; Baer, T.; Ng, C.-Y. High-Resolution Laser Photoionization and Photoelectron Studies; John Wiley & Sons: Chichester, 1995. (6) 1994 Time-of-Flight Mass Spectrometry; Cotter, R. J., Ed.; American Chemical Society: Washington, DC, 1994. (7) Weickhardt, C.; Moritz, F.; Grotemeyer, J. Mass Spectrom. Rev. 1996, 15, 139-162. (8) Boesl, U. J. Mass Spectrom. 2000, 35, 289-304.

Analytical Chemistry, Vol. 74, No. 19, October 1, 2002 4861

resolution9 or sample throughput, depending on the analytical problem to be solved. In certain applications, REMPI’s selectivity is sufficiently high to make a chromatographic preseparation unnecessary. In these cases, sample treatment and preparation steps can be reduced to a minimum or even avoided completely. A precondition for the application of REMPI in mass spectrometry is that the analyte molecules are in the gas phase. Therefore, if solid-state material is to be analyzed, a desorption step has to precede the analysis by laser mass spectrometry. Pulsed methods, such as laser desorption,10,11 providing a high density of neutrals during the ionization event and leaving even labile molecules intact to a large extent are perfectly adapted to laser mass spectrometry and consequently minimize the sample handling required. Because of its high potential for trace analysis in solids and surfaces, the two-step technique resulting from the combination of laser desorption and resonant laser mass spectrometry has already been applied to a variety of analytical questions, such as the analysis of aromatic surface coverings,12 the analysis of large biochemical compounds,13 and the detection of pesticides in agricultural samples.14,15 Other groups have used two-step laser mass spectrometry for the chemical mapping of surfaces or as a microprobe technique.16-21 In addition, its applicability to the detection of soil contaminants has already been investigated on principle.22,23 In addition to soil, often water seeping through the ground has to be analyzed in order to characterize a hazardous waste site.24 Although, for example, TNT is rather persistent in the ground and not washed out, it can be degraded by microorganisms to aminotoluenes, which are soluble and may contaminate groundwater. Therefore, it would be highly desirable to use the same instrument for soil as well as water analysis. However, because of vacuum requirements, the vapor pressure of the water sample has to be reduced. This can be achieved both by cooling and freezing the sample and by adsorbing it on solid material. Because in the latter case a plain surface is able to fixate only a small amount of sample, it is advantageous to use porous media in this context. Although the detection of aromatic compounds in (9) Weickhardt, C.; Boesl, U.; Schlag, E. W. Anal. Chem. 1994, 66, 10621069. (10) Becker, C. H.; Gillen, K. T. Anal. Chem. 1984, 56, 1671. (11) Meijer, G.; deVries, M. S.; Hunziker, H. E.; Wendt, H. R. Appl. Phys. B 1990, 51, 395-403. (12) Zenobi, R. Chimia 1994, 48, 64-71. (13) Grotemeyer, J.; Schlag, E. W. Acc. Chem. Res. 1989, 22, 399. (14) Orea, J. M.; Besco´s, B.; Montero, C.; Gonzalez, U. A. Anal. Chem. 1998, 70, 491-497. (15) Orea, J. M.; Montero, C.; Jime´nez, J. B.; Uren ˜a, A. G. Anal. Chem. 2001, 73, 5921-5929. (16) Zenobi, R.; Philippoz, J.-M.; Buseck, P. R.; Zare, R. N. Science 1989, 246, 1026-1029. (17) Haefliger, O. P.; Zenobi, R. Rev. Sci. Instrum. 1998, 69, 1828-1832. (18) Mahajan, T. B.; Ghosh, U.; Zare, R. N.; Luthy, R. G. Int. J. Mass Spectrom. 2001, 212, 41-48. (19) Gillette, J. S.; Ghosh, U.; Mahajan, T. B.; Zare, R. N.; Luthy, R. G. Isr. J. Chem. 2001, 41, 105-110. (20) Zenobi, R. Chimia 2001, 55, 773-777. (21) Kalberer, M.; Morrical, B. D.; Zenobi, R. Anal. Chem. 2002 in press. (22) Dale, M. J.; Jones, A. C.; Pollard, S. J. T.; Langridge-Smith, P. R. R. Environ. Sci. Technol. 1993, 27, 1693-1695. (23) Fye, J. L.; Nelson, H. H.; Mowery, R. L.; Baronavski, A. P.; Callahan, J. H. Anal. Chem. 2002, 74, 3019-3029. (24) Maughan, J. T. Ecological Assessment of Hazardous Waste Sites; VanNostrand: New York, 1993.

4862

Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

frozen water samples by means of LD-LAMS has already been studied,25,26 the use of a solid adsorber medium was used in our investigations because of the reduced instrumental efforts. The whole sample preparation and the measurement can be carried out at room temperature, and a cooling system is not required. EXPERIMENTAL SECTION Sample Preparation. Soil samples were collected from various locations around Cottbus, with their types ranging from sandy via humus and loamy soil to clay. For quantitative measurements, weighed out amounts of chemicals (anthracene, pyrene, fluorene, 9-chloroanthracene, phenazine) which were obtained in technical quality and used without further purification were added to the soil samples. Afterward, the mixtures were homogenized in a ball mill for 10 min. The soil samples were then pressed to small rods with a diameter of 3 mm, introduced into the mass spectrometer through a vacuum lock and placed in position for laser desorption by means of a manipulator. Water samples were prepared by spiking tap water with weighed out amounts of contaminants and homogenizing the solutions in an ultrasonic bath for 1 min. The solutions were dropped onto adsorber material pressed into a cup at room temperature. The adsorber materials investigated were activated carbon, granular clay, sodium sulfate, and a mixture of calcium carbonate and sand. Mass Spectrometer. The instrument used allowed the laser desorption to be carried out directly in the ion source of the mass spectrometer or, alternatively, in front of a pulsed nozzle, with subsequent acceleration of the desorbed material toward the ionization region by means of the supersonic jet (see Figure 1). Although the first possibility results in a high density of neutrals at the point of ionization, the latter is advisable when cooling of the molecular degrees of freedom is required in order to increase the selectivity of the ionization. All results presented and discussed in this work were obtained using the first setup, that is, by direct laser ionization in the desorption plume without supersonic beam cooling. After photoionization, the ions were transferred into a homebuilt gridless reflectron time-of-flight mass spectrometer with a total drift length of ∼1.5 m and equipped with a Wiley-McLarentype ion source. The signal of the detector, which was based on a microsphere plate, was directly transmitted to a digital oscilloscope coupled to a PC for data storage and processing. Laser Systems. Laser desorption was carried out using the fundamental (1064 nm) or the frequency-quadrupled (266 nm) output of a Nd:YAG laser with pulse energies up to 50 mJ in the IR and 3 mJ in the UV. The laser beam was unfocused and irradiated the whole cross section of the sample (diameter, 3 mm). For multiphoton ionization, either the frequency-doubled output of a nanosecond dye laser system, the fourth harmonic of a Nd:YAG laser (wavelength 266 nm), or the fourth harmonic of a Ti:sapphire laser system generating pulses with a temporal width of ∼170 fs could be used. Typical pulse energies used were ∼0.5 mJ for the nanosecond and 20 µJ for the femtosecond laser. The (25) Bucheli, T. D.; Haefliger, O. P.; Dietiker, R.; Zenobi, R. Anal. Chem. 2000, 72, 3671-3677. (26) Alimpiev, S. S.; Mlynski, V. V.; Belov, M. E.; Nikiforov, S. M. Anal. Chem. 1995, 67, 181-186.

Figure 1. Different laser desorption setups for REMPI: (a) laser desorption coupled to the supersonic expansion in front of a pulsed nozzle and (b) laser desorption direct within the ion source.

Figure 2. Laser mass spectra of pyrene (concentration: 10 µg/kg) detected in different types of soil by laser desorption laser mass spectrometry. Note the absence of possibly disturbing signals.

beams of both lasers were focused into the ionization region by means of a quartz lens with a focal width of 200 mm. The experiment was run at a repetition rate of 10 Hz. RESULTS AND DISCUSSION Soil Analysis. In the first step, a possible influence of the soil type on the measurements was investigated. Samples from three different soilsssandy, humus, and clayswere prepared. They were studied in respect to desorption efficiency, possible mass spectrometric interferences, and their influence on the quantification. To make sure that the laser desorption and all other parts of the instrument were working properly, pyrene was added to each sample in a concentration of 10 µg/kg, and each mass spectrum was checked for the presence of the pyrene signal. In Figure 2, the time-of-flight spectra obtained using an ionization wavelength of 266 nm and nanosecond laser pulses are shown for the different soil types. In all three mass spectra, the pyrene signal is clearly present, although with different intensities, and it is the only signal recorded. This indicates that a certain influence of the soil type on the desorption efficiency is present, but that no interfering signals originating from the soil matrix complicate the detection of aromatic compounds by the technique described here, at least down to the concentration values investigated here.

In connection with the analysis and remediation of military waste sites, the substances of greatest interest are explosives and chemical warfare agents. From the perspective of resonant multiphoton ionization, such chemicals are somehow problematic, because they tend to quickly dissociate after photoexcitation. For nitroaromatic or metal organic compounds, these dissociation processes may take place at rate constants significantly larger than 109 sec-1, thus giving rise to the detection of only small and uncharacteristic fragment ions in the mass spectrum when conventional nanosecond pulses are used for ionization. Several groups have demonstrated within the last years that by the use of ultrashort light pulses with durations in the subpicosecond region, it is possible to overcome the problem related to fast intramolecular relaxation processes, which appears to be a rather general one for a broad variety of analytically important substances.27 These short and intense pulses are aimed at finishing the ionization process before a significant percentage of the excited molecules can undergo relaxation processes. Figure 3 compares the laser mass spectra of 3-nitrotoluene and 2,4,6-TNT obtained by multiphoton ionization using conventional nanosecond laser pulses (left side) to those measured after ionization by light pulses with a duration of ∼150 fs (right side). In both cases, the wavelength was set to 206 nm, the longest UV wavelength available from our femtosecond-laser system. At this wavelength, all nitrotoluenes can be ionized by a two-photon process. Although for the monosubstituted toluene, the molecular ion signal is still the base peak of the mass spectrum, the dissociation of TNT is so fast that even under the conditions of ultrafast multiphoton ionization, no molecular ions are observed. However, TNT can readily be identified by its OH-loss signal. In the nanosecond case, the high degree of fragmentation makes a clear identification of any of the compounds impossible. Quantification of the measurements was obtained by performing relative measurements with respect to an internal standard added to the soil sample as early as possible, that is, prior to homogenization in the ball mill. It should be kept in mind that the method applied here is basically a surface-sensitive technique, because the material liberated by laser desorption comes from a (27) For a review, see: Ledingham, K. W. D.; Singhal, R. P. Int. J. Mass Spectrom. 1997, 163, 149-168.

Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

4863

Figure 3. Laser mass spectra of 3-nitrotoluene (upper spectra) and 2,4,6-trinitrotoluene (lower spectra) using nanosecond laser pulses (left column) and 150-fs laser pulses (right column). Wavelength: 206 nm.

Figure 4. Temporal course of the decacyclene ion signal in a laser desorption laser-MS measurement on clay. When the desorption laser is blocked for single shots, the signal vanishes completely and returns immediately at the next shot hitting the sample.

thin surface layer of the sample. Furthermore, the interaction of the desorption laser with the sample surface may change its chemical composition and stoichiometry. Against this background, it is obvious that reliable quantitative data can be obtained from only the first few laser shots onto the sample. In Figure 4, the temporal course of the signal of laser-desorbed and laser-ionized decacyclene in clay is plotted for a series of 6000 laser shots applied within 10 min (upper trace). As can be seen from the signal of the photodiode that monitors the laser activity on the sample (lower trace), the ion signal is at a maximum immediately after starting the desorption sequence. Then it first decreases with a time constant of ∼150 s () 1500 laser shots) and afterward remains almost constant at ∼20% of the maximum level. To check whether part of the signal is due to a continuous evaporation of analyte as a result of heating of the sample and its holder, the desorption laser was blocked from time to time for one or two 4864 Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

Figure 5. Comparison of the phenanzine concentration in clay samples determined by LD-LAMS with the values prepared by weighing. The solid line represents a least-squares fit to the data points.

shots, and the resulting ion signal was compared to the average signal level at that point in time. These events show up in the lower trace in Figure 4 as sudden signal setbacks. Because the blocking of the desorption laser reduces the signal to baseline at all times after the start of the measurement, continuous evaporation can be ruled out. We, therefore, assume that the intense signal at the beginning of the measurement can be attributed to analyte molecules located within a thin surface layer from which they can easily escape into the vacuum. This surface layer is depleted within several hundreds of laser shots. Afterward, the analyte molecules can be transferred into the gas phase only at the rate with which the laser digs a hole into the soil material and sets free lower layers. This process leads to a reduced signal but can continue for a long time. A comparison of the temporal behavior of the ion signal of two different substances shows that their surface concentrations are decreased at different rates. Consequently, the quantification has to be carried out within a series of laser shots short enough

Figure 6. Comparison of the laser mass spectra obtained by laser desorption of a water sample containing 2,6-diaminotoluene at a concentration of 100 µg/mL from various adsorber media: (a) mixture of lime and sand, (b) granular clay, (c) activated carbon, and (d) sodium sulfate. Two different desorption laser wavelengths were used: left side, 266 nm (pulse energy, 3 mJ); and right side, 1062 nm (pulse energy, 50 mJ).

to ensure that the ratio of the two signals is constant to a good approximation. At the desorption laser parameters used in these experiment, this was usually the case for up to 30-50 pulses. Therefore, the way to obtain good quantitative results is a tradeoff between getting rid of shot-to-shot fluctuations by averaging over many individual measurements and finishing the measurement within a time short enough to ensure that the relative signal intensity of analyte and internal standard does not change

significantly. This means of quantifying the results is based on the assumption that neither chemical reactions between calibration and pollutant compounds nor replacement effects between them occur. Although such processes are probably negligible for trace concentrations of these compounds, they cannot be ruled out at higher concentrations and should be studied in detail. We determined the concentration of phenazine in clay using 9-chloroanthracene at a concentration of 10 mg/kg as internal Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

4865

standard. Phenazine concentrations ranged from 1 mg/kg to 10 g/kg. Figure 5 compares the values determined by laser desorption laser mass spectrometry with the concentrations as they were prepared by weighing. The value obtained for the highest phenazine concentration was used for calibration. All data points are located within a (20% range around the solid line, indicating the one-to-one correspondence. This is, of course, far from being a precision measurement, but it is acceptable for a rapid scanning technique that requires only ∼1 min of time for the processing of each sample. In this way, it is possible to quickly locate the “hot spots” of a hazardous waste site, which then can be investigated in detail by standard methods. Water Analysis. For the analysis of water samples, the sample carriers used for soil analysis were filled with various porous adsorber materials. For this application, it was important that all materials were mechanically stable when dry as well as when wet. A drop of water was applied and allowed to soak for some seconds. Afterward, the sample was introduced into the mass spectrometer and analyzed in the same way as the soil samples. The suitability of the four different adsorber materials was investigated using a desorption laser wavelength of either 1064 or 266 nm. The samples were spiked with either diaminotoluenes or aminophenols in various concentrations. Both can be efficiently ionized using the fourth harmonic of a Nd:YAG laser (wavelength, 266 nm; pulse duration, 5 ns). Figure 6 compares the results obtained for 2,6-diaminotoluene, which can be regarded as typical. Although for the calcium carbonate/sand mixture and the granular clay, the best results were observed when the UV wavelength was used, activated carbon showed slightly better results with the IR and a comparatively high degree of fragmentation when the UV was applied. These observations may be explained by the different absorptivity of the pale calcium carbonate/sand mixture and the clay compared to the black carbon at the two wavelengths. Because the pulse energy of the desorption laser in the IR was 12 times that in the UV, the latter turns out to be much more efficient for the desorption of aromatic substances; however, all three of these materials proved to be useful for the laser desorption of aromatic water contaminants. Sodium sulfate did not give satisfactory results at any wavelength. In contrast to the temporal development of the ion signal in the case of soil samples, the porous adsorbers showed an increase in the rate of desorbed material within the first few hundred laser shots, followed by a steady decrease comparable to that observed with soil samples. Furthermore, the signal does not vanish when single-desorption laser pulses are blocked. We interpret this observation in such a way that because of heating of the adsorber material, analyte molecules are also thermally desorbed, which results in a nonvanishing signal for a short period. A large part of the sample is confined in the pores of the adsorber and not exposed to the laser light. Therefore, the laser heating of the sample is supposed to have an important effect on the diffusion of the analyte in the sample, making it available for laser desorption. Only after the adsorber is warmed can it be efficiently set free and analyzed. To quantify the results obtained with water samples, an internal standard was added to the sample together with the contaminant. 4866 Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

Figure 7. Comparison of the concentration of 2,4-diaminotoluene in water determined by LD-LAMS using 2,6-diamino-3-chlorotoluene as internal standard compared to the concentration prepared by weighing. The solid line is a linear least-squares fit to the data points.

Diaminochlorobenzenes turned out to fulfill all requirements for an internal standard and were used in a concentration of 100 µg/ mL. The concentration of a contaminant was determined by averaging the mass spectra of 100 laser shots and a relative measurement in respect to the signal of the internal standard. Figure 7 compares the concentrations of 2,4-diaminotoluene determined by laser mass spectrometry with those as prepared by weighing. It can be seen that the results can be fitted by a linear function within the concentration range investigated (50900 µg/mL). Each data point represents the average value and the standard deviation of a series of 10 individual measurements on the same solution. The precision as well as the reproducibility for all concentrations is within (10%. The detection limit defined as the analyte concentration resulting in a signal-to-noise ratio of 3 was found in the low micrograms-per-milliliter range for the diaminotoluenes and diaminophenols. Thus, the sensitivity determined for liquid samples is comparable to that for solids. It should be mentioned that neither the mass spectrometer nor its ion source used in these experiments was optimized with respect to sensitivity. A brief estimation shows that an optimized ion source geometry in combination with a shorter reflectron and optimized ion optics should result in a decrease in the detection limit by about 2 orders of magnitude; however, the technique described here does not reach the sensitivity achieved by chromatographic methods, which is necessary for ultratrace analysis, particularly in water. CONCLUSION In this work, we demonstrated the capabilities of laser desorption laser mass spectrometry for the detection and quantification of aromatic trace substances in liquid and solid environmental matrixes. Its major advantage over established techniques is the fact that while offering multicomponent capability and delivering a large quantity of information, it requires no time- and manpower-consuming sample pretreatment. Measurements on a soil or water sample can be completed typically within a minute. Thus, high sample throughput can be achieved, which is the basis

for the rapid screening and characterization of a suspected waste site. The detection limits achieved up to now are far from the trace sensitivity required for ground and surface water monitoring, but they are acceptable for a soil analytical screening technique. These characteristic features make the method developed an ideal completion of the chromatographic techniques used in environmental analysis, which reach extremely high sensitivities but are time-consuming and expensive. The combination of both allows the fast location of “hot spots” by laser mass spectrometry and their detailed investigation by routine procedures and instrumentation. Of course, it has to be kept in mind that soil is not at all a well-defined matrix, but may vary drastically in composition, structure, and homogeneity. Therefore, until now, the analytical methods developed in this context could be investigated and tested using only a few typical soil types. Although these examples were very different, it cannot be ruled out that the results presented here may not hold for certain very special soils. It was demonstrated that the application of ultrashort laser pulses for ionization offers advantages when quickly dissociating

molecules are to be detected. However, a large number of aromatic compounds can readily be ionized using conventional and economic solid state nanosecond lasers. Together with a compact time-of-flight or quadrupole ion trap mass spectrometer, they can easily fit into a field-portable device for on-site measurements, thus minimizing the efforts for sample stabilization and transport. ACKNOWLEDGMENT The authors thank Professors J. Grotemeyer and J. Reif for their loan of instrumentation. This work was supported by a grant from the German Federal Foundation for the Environment (Deutsche Bundesstiftung Umwelt), which is gratefully acknowledged.

Received for review March 25, 2002. Accepted July 18, 2002. AC020189S

Analytical Chemistry, Vol. 74, No. 19, October 1, 2002

4867