Speciation of Methylcyclopentadienyl Manganese Tricarbonyl by High

Methylcyclopentadienyl manganese tricarbonyl (MMT) is a fuel additive that has been marketed for use in unleaded gasoline since December 1995...
0 downloads 0 Views 70KB Size
Download PDF
Anal. Chem. 1999, 71, 5379-5385

Speciation of Methylcyclopentadienyl Manganese Tricarbonyl by High-Performance Liquid Chromatography-Diode Laser Atomic Absorption Spectrometry David J. Butcher,† Aleksandr Zybin, Michail A. Bolshov, and Kay Niemax*

Institute of Spectrochemistry and Applied Spectroscopy, Bunsen-Kirchhoff-Strasse 11, D-44139, Dortmund, Germany

Methylcyclopentadienyl manganese tricarbonyl (MMT) is a fuel additive that has been marketed for use in unleaded gasoline since December 1995. The widespread use of this additive has been suggested to cause health risks, but limitations in data regarding its degradation products and their toxicity prevent an accurate evaluation. To monitor the organomanganese compounds, it is clearly advantageous to employ low-cost, high-sensitivity, manganese-specific instrumentation to perform speciation. In this work, instrumentation fitting these criteria was obtained by the combination of high-performance liquid chromatography (HPLC) with diode laser atomic absorption spectrometry (DLAAS) and was used to determine MMT, its nonmethylated derivative, cyclopentadienyl manganese tricarbonyl (CMT), and inorganic manganese. DLAAS was shown to be a versatile analytical technique for total Mn determination, with a detection limit of 1 ng/ mL and a linear dynamic range (LDR) of almost 5 orders of magnitude. Analytical figures of merit for HPLC-DLAAS included a detection limit of 2 ng(as Mn)/mL, a LDR of 3 orders of magnitude, and an analysis time of three minutes. The organometallic compounds are characterized by rapid photolysis in sunlight, and hence, experiments were performed to evaluate whether normal laboratory lighting is suitable for their determination. Our results showed that normal laboratory protocols may be employed except that the organomanganese compounds should be stored away from light except during sample introduction procedures. The ability of the instrumentation to selectively preconcentrate organomanganese compounds while removing inorganic manganese was demonstrated. Sufficient resolution was obtained to determine a 20-fold excess of CMT compared with MMT. The ability of the system to do practical analysis was demonstrated by the accurate determination of MMT in spiked samples of gasoline, human urine, and tap water. These results demonstrate the suitability of HPLC-DLAAS for the speciation of MMT and its derivatives in industrial, toxicological, and environmental samples. Recently, considerable controversy has developed over the potential toxicity of an organomanganese fuel additive, methyl* Corresponding author: (phone) +49-231-1392-101; (fax) +49-231-1392-310; (e-mail) [email protected]. † Current address: Department of Chemistry and Physics, Western Carolina University, Cullowhee, North Carolina 28723. 10.1021/ac990570l CCC: $18.00 Published on Web 11/02/1999

© 1999 American Chemical Society

Figure 1. Chemical structures of relevant organomanganese compounds: cyclopentadienylmanganese tricarbonyl (CMT), methylcyclopentadienylmanganese tricarbonyl (MMT), and carboxycyclopentadienyl manganese tricarbonyl (CCMT).

cyclopentadienyl manganese tricarbonyl (MMT) (Figure 1).1-6 From 1977 to 1995, MMT was approved only for limited use in leaded gasoline. In 1995, the manufacturer was able to successfully challenge the denial in federal court. Consequently, MMT has been marketed in the United States since December 1995. Although manganese is considered a nutritionally essential trace element, it is known that very high levels of inhaled manganese induce neurobehavioral and respiratory effects.1,2 However, the health risks associated with the widespread introduction of MMT into the U.S. gasoline supply, which would probably cause relatively small increases in inhaled manganese, remain unknown because of limitations in data. These include limited knowledge of the combustion products of MMT, limited knowledge of the toxicity of MMT and its combustion products, and considerable uncertainty in exposure assessments of MMT. In addition to the combustion of manganese, human exposure to MMT and its degradation products may also occur through surface water and groundwater ecosystems.3 One of the deficiencies of most previous studies of MMT is that the analysis typically consisted of the measurement of total levels of manganese, rather than of the particular compound present. MMT is considerably more toxic than most inorganic (1) Davis, J. M. Environ. Health Perspect. Suppl. 1 1998, 106, 191-201. (2) Frumkin, H.; Solomon, G. Am. J. Ind. Med. 1997, 31, 107-115. (3) Garrison, A. W.; Cippolone, M. G.; Wolfe, N. L.; Swank, R. R. Environ. Toxicol. Chem. 1995, 14, 1859-1864. (4) Lynam, D. R.; Pfeifer, G. D.; Fort, B. F.; Ter Haar, G. L.; Hollrah, D. P. Sci. Total Environ. 1994, 146/147, 103-109. (5) Lenane, D. L.; Fort, B. F.; Ter Haar, G. L.; Lynam, D. R.; Pfeifer, G. D. Sci. Total Environ. 1994, 146/147, 245-251. (6) Egyed, M.; Wood, G. C. Sci. Total Environ. 1996, 189/190, 11-20.

Analytical Chemistry, Vol. 71, No. 23, December 1, 1999 5379

forms of manganese. Differences in toxicity exist among inorganic manganese compounds, with manganese(III) being more toxic than manganese(II). The quantitative determination of the species or form of a metal is referred to as speciation.7 At the present time, most techniques for speciation involve relatively complicated, expensive instrumentation. For example, Walton et al.8 determined MMT and related compounds by highperformance liquid chromatography (HPLC) with laser-excited atomic fluorescence spectrometry (LEAFS) detection using a flame as the atom cell and obtained a detection limit of 0.5 ng/ mL (10 pg). This system used an expensive and unreliable excimer laser pumped dye laser system, which limits the practicality of the technique. So far, the most widely used hyphenated technique for speciation analysis is on-line coupling of different separation techniques (gas or liquid chromatography, capillary electrophoresis, etc.) with element-selective detection of the separated species by inductively coupled plasma mass spectrometry, ICPMS, and for some elements, ICP optical emission spectrometry (ICPOES).9 ICPMS provides sufficient sensitivity for a large variety of samples and species to be analyzed, but the purchase price and cost of operation are high. The widespread introduction of MMT into the United States gasoline supply has the potential to be a serious threat to public health. It may be necessary to routinely monitor gasoline, groundwater, and human fluids (e.g., urine) to evaluate potential human exposure. Toxicological experiments of MMT and its metabolites on animals are required to assess the effects of these compounds on human health, with particular emphasis on “at risk” populations which include service-station attendants, children, and pregnant women. The development of high-sensitivity, low-cost, reliable instrumentation will allow research and monitoring laboratories to make these important measurements in many places. Semiconductor diode lasers (DL) offer high reliability, small size, and easy operation10 and, consequently, allow low-cost, highsensitivity elemental analysis by atomic absorption spectrometry (AAS). In addition, they emit light continuously, providing a high duty cycle, and can be easily tuned by changing the temperature (slow tuning) and current (fast tuning). Diode lasers provide several advantages compared with conventional hollow cathode lamps (HCLs) for excitation. In contrast to hollow cathode lamps, a DL emits a prominent, single, narrow line under normal operating conditions, which dramatically simplifies the spectral isolation of the absorption signal. One does not need the monochromator which is necessary with HCLs for isolation of the analytical line from the spectral lines of the HCL buffer gas and unwanted lines of the cathode material. The wavelength of DLs can be easily modulated at frequencies as high as in the megahertz by modulation of the diode current.10-12 Wavelength modulation (WM) of the DL with (7) Ebdon, L. C.; Hill, S.; Ward, R. W. Analyst (Cambridge, U.K.) 1987, 112, 1-16. (8) Walton, A. P.; Wei, G.-T.; Liang, Z.; Michel, R. G.; Morris, J. B. Anal. Chem. 1991, 63, 232-240. (9) Montaser, A. Inductively Coupled Plasma Mass Spectrometry ; Wiley-VCH: New York, 1998. (10) Niemax, K.; Zybin, A.; Schnu ¨ rer-Patschan, C.; Groll, H. Anal. Chem. 1996, 68, 351A-356A. (11) Franzke, J.; Schnell, A.; Niemax, K. Spectrochim. Acta Rev. 1993, 15, 379395.

5380

Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

detection of absorption at the harmonic of the modulation frequency, 2f, 4f, ..., greatly reduces low-frequency noise (flicker noise) in the baseline and provides improved limits of detection.12 In addition, WM of the radiation source provides an almost ideal correction for nonspecific absorption and significantly improves the selectivity of the AAS technique.10 The combination of wavelength modulation with a logarithmic amplifier produces a linear dynamic range of several orders of magnitude,13 which is a significant improvement over that of 3 orders of maganitude obtained with HCLAAS. If the optimal experimental conditions are realized and the fundamental shot-noise limit is achieved, the minimum absorption which can be measured is extremely lows about 10-6. The coupling of DLAAS with high performance liquid chromatography (HPLC) for speciation of chromium(III) and chromium(VI) was recently reported.14-16 Using a simple setup involving a single-beam configuration and linear processing of the detector output, a detection limit of 1 ng/mL was obtained.14 An improvement in detection limit to, 30 pg/mL, was obtained by use of a higher power of frequency-doubled laser light, a doublebeam configuration, and logarthmic processing of the output of the detector.15 Separation of the two species was obtained with analysis times of two minutes. The purpose of this study was to evaluate the suitability of HPLC-DLAAS for the determination of MMT, its combustion products, its metabolic products, and its environmental degradation products. This investigation focuses upon two commercially available organomanganese compounds: MMT and its nonmethylated analogue, cyclopentadienyl manganese tricarbonyl (CMT, Figure 1). These compounds are characterized by degradation in sunlight,3 and hence, experiments were performed to evaluate their stability in laboratory light to assess whether special conditions (e.g., instrumental operation in a dark room) are required for the analyses. Analytical figures of merit, including detection limits, linear dynamic ranges, and analysis times, were obtained by DLAAS with and without HPLC. The chromatographic separation of the compounds was optimized, and preconcentration of these compounds on the column was investigated. EXPERIMENTAL SECTION Instrumentation. Much of the HPLC-DLAAS instrumentation employed in this study has been described elsewhere.15 The 806.2nm output of a laser diode (model SDL-5311-G1 SDL, Inc.) (90 mW), driven by a power supply (Profile model LDC200), was frequency doubled with a 20-mm-long LiIO3 crystal to produce 170 nW of blue light at 403.1 nm (Figure 2). The wavelength of the laser diode was modulated with the optimal amplitude (3.9 times the full width at half-maximum of the absorption line12) at 3.5 kHz (f) produced by the internal generator of a lock-in amplifier (SR-830 Stanford Instruments). The blue light passed through an air-acetylene flame, which served as the atom cell, was separated from the fundamental by two prisms (not shown in Figure 2), (12) Silver, J. A. Appl. Opt. 1993, 31, 707-717. (13) Zybin, A.; Liger, V. V.; Kuritsyn, Yu. A. Spectrochim. Acta, Part B 1999, 54, 613-619. (14) Groll, H.; Schaldach, G.; Berndt, H.; Niemax, K. Spectrochim. Acta, Part B 1995, 50B, 1293-1298. (15) Zybin, A.; Schaldach, G.; Berndt, H.; Niemax, K. Anal. Chem. 1998, 70, 5093-5096. (16) Koch, J.; Zybin, A.; Niemax, K. Appl. Phys. B 1998, 67, 475-479.

Figure 2. Experimental arrangement for HPLC-DLAAS instrumentation. SHG: second harmonic generation.

and was passed on to a photomultiplier tube (PMT, 1P-28, Hamamatsu), where the electrical output was sent to a logarithmic amplifier. The output signal of the logarithmic amplifier was further amplified by the lock-in amplifier (model 830, Stanford Instruments) at the fourth harmonic of wavelength modulation frequency (4f). The effective time constant was about 2 s. Data processing was performed with a personal computer. We want to emphasize here the specific technique of the signal processing by WM-spectroscopy. The value of absorbance can be directly calculated from the size of the electrical signal after logarithmic conversion. The diode laser can be considered as a monochromatic source and the light intensity at the photodetector, according to the Lambert-Beer law, is given by

I ) I0 exp{-[κ(λ)l + κbl]}

(1)

where κ(λ) is the wavelength-dependent atomic absorption coefficient, κb is the wavelength-independent background absorption coefficient, and l is the absorption path length. The output voltage of the logarithmic converter is given by

V ) A ln I0 - A[κ(λ) + κbl]

(2)

where A is the a dimensional coefficient depending on the characteristic of the logarithmic converter.13 In wavelength modulation, the only modulated part in eq 2 is κ(λ)l, which selectively measures the atomic absorption. The alternative component of the voltage at the modulation frequency directly gives the value of absorbance. For 4f detection at optimum modulation amplitude, A ) 0.18 (kT/e), where k is the Boltzmann constant, T the temperature of the logarithmic transducer, and e the electron charge. This single-beam/4f detection approach differs from our previous speciation work which employed double-beam/2f detection.15 The new approach is simpler experimentally, because it is not necessary to use two photodetectors and logarithmic amplifiers. However, under high laser power conditions, the doublebeam scheme can provide a better signal-to-noise ratio by compensating for the excess laser noise, the offset, and its fluctuation. At the relatively low laser powers employed here, we were limited by shot noise, which would not be significantly

improved by double-beam detection. The use of 4f detection requires a modulation amplitude twice as large as 2f detection, which cannot be used at all laser diode wavelengths, since the spectral output of diode lasers is characterized by regions of continuous tuning and by mode hops as the wavelength is varied.11 Consequently, the 4f detection scheme cannot be used at wavelengths near to a mode hop. The details of the single-beam/ 4f detection scheme will be published elsewhere. Sample introduction into the flame was performed with hydraulic high-pressure nebulization (HHPN) using an HPLC pump (Knauer). For DLAAS, samples were measured by the continuous introduction of sample into the flame. For HPLCDLAAS, a reversed-phase C18 column (Eurosphere) was placed between the flame and the HPLC pump. The samples were injected into the system using a 0.2-mL injection loop. The eluent for chromatography was 65:35 methanol/aqueous pH 4 buffer unless specified otherwise. The HPLC peak height was used as the analytical signal. The time constant of the lock-in amplifier was 2 s. No additional smoothing was applied to the signals. Chemicals and Standard Preparation. Methylcyclopentadienyl manganese tricarbonyl (MMT) and cyclopentadienyl manganese tricarbonyl (CMT) were purchased from Aldrich Chemical and used as received. The MMT was contaminated by approximately 5% CMT, as has been previously reported in the literature.3 Stock solutions of the organomanganese compounds were prepared in HPLC grade methanol and stored in the dark. Working solutions were prepared daily in a dark room in 65:35 methanol/aqueous pH 4 buffer (0.05 M ammonium acetate, made by mixing Merck analytical grades of ammonia and acetic acid). The solutions were stored in the dark except during sampleintroduction procedures. Preparation of Samples Spiked with MMT. Gasoline and urine were spiked with MMT to give concentrations of 8.3 and 4.0 µg(as Mn)/mL, respectively. Determinations were performed by dilution of the samples to give concentrations between 40 and 100 ng/mL. It was necessary to dilute the gasoline in methanol (1 + 9) before dissolution in 65:35 methanol/aqueous pH 4 buffer to completely dissolve gasoline in the buffer. A saturated solution of MMT in water was prepared, allowed to equilibrate for several hours, diluted, and analyzed by HPLC-DLAAS to estimate the solubility of MMT in water. A summary outline of the most important experimental details is given in Table 1. RESULTS AND DISCUSSION Figure 3 shows a typical chromatogram of 50 ng/mL of inorganic manganese, 50 ng/mL of CMT, and 50 ng/mL of MMT with 65:35 methanol/pH 4 buffer. The inorganic manganese was not retained on the column and passed directly through with a retention time of 0.5 min. CMT was eluted at 1.8 min, followed by MMT at 2.5 min. Removal of the HPLC column allowed determination of total manganese in less than one minute. Stability of Organomanganese Compounds. Garrison et al.3 reported that MMT in deionized water photolyzed rapidly in sunlight with a half-life of 0.93 min. Hence, our first consideration was to evaluate the stability of the organomanganese compounds in normal laboratory light. Because we were working with concentration levels in the parts-per-billion range, these studies were performed in polyethylene bottles. Polyethylene is generally Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

5381

Table 1. Experimental Conditions for Determination of Inorganic Manganese, CMT, and MMT by HPLC-DLAAS laser diode: laser power: second harmonic generation crystal: atom cell: detection system: HPLC column: mobile phase:

mobile phase flow rate: injection volume:

SDL-5311-G1 (SDL, Inc.) 90 mW at 806.2 nm (fundamental); 170 nW at 403.1 nm (SHG) LiIO3 lean air/acetylene flame with hydraulic high-pressure nebulization (HHPN) photomultiplier tube (PMT) for optimal linear dynamic range, the PMT output was sent to a logarithmic amplifier 5-µm reversed-phase C18 column (Eurosphere) 65:35 methanol/aqueous pH 4 buffer (0.05 M ammonium acetate) for best resolution, 65:35 methanol/aqueous pH 4 buffer (0.05 M ammonium acetate) for preconcentration of CMT and MMT on the column, aqueous pH 4 buffer (0.05 M ammonium acetate) 1 mL/min 0.2 mL

Figure 3. Chromatogram of 50 ng/mL of inorganic manganese, 50 ng/mL of CMT, and 50 ng/mL of MMT by HPLC-DLAAS using 65:35 methanol/aqueous pH 4 buffer as eluent.

regarded as a better choice than glass for trace analysis. Bottles containing one of 50 ng(as Mn)/mL CMT or MMT were placed on a laboratory bench in what we considered normal lighting conditions (fluorescent tube lamps), and aliquots were removed for analysis. The absorption signals were normalized by comparison with those of 50 ng(as Mn)/mL of CMT or MMT stored in darkness. No measurements of the spectral characteristics or the intensity of light were performed. The results of the exposure of 50 ng/mL of CMT and 50 ng/ mL of MMT are shown in Figure 4. Both compounds show minimal degradation over the first 3 h of the experiment. This indicates that solutions prepared in plastic bottles can be exposed to room light during sample-introduction/injection procedures without degradation over the course of the expected stability of trace solutions (several hours). However, to minimize the exposure to light, the solutions were stored in light-tight containers except during sample-introduction procedures. After 3 h, observable degradation of CMT and MMT to inorganic manganese occurred in approximately a linear relationship. Interestingly, even after more than 2 days, complete decomposition of the organomanganese compounds was not observed. Garrison and co-workers3 reported that the manganesecontaining degradation product of photolysis was an unidentified manganese carbonyl compound. They also reported that the manganese carbonyl further degraded upon exposure to oxygen to give trimanganese tetroxide. With the chromatography system employed here, it was not possible to identify the forms of 5382 Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

Figure 4. Effect of exposure of organomanaganese compounds in polyethylene bottles to light versus time: (a) MMT and (b) CMT. Inorganic manganese (9) was produced by photoinduced degradation of CMT and MMT (b).

inorganic manganese present. Although this identification was outside the scope of the current project, future work will include the development of instrumentation to identify the inorganic forms of manganese produced during photolysis. Analytical Figures of Merit. Analytical figures of merit for DLAAS are compared with those for other common techniques in Table 2. The detection limits were defined as 3 times the standard deviation of the blank. Sixteen injections of the blank were used to evaluate the standard deviation. Our detection limit of 1 ng/mL, obtained by continuous introduction of sample into the flame, was a factor of 3 better than the best-reported value17 by hollow cathode lamp atomic absorption spectrometry (HCLAAS) which, however, is difficult to obtain routinely. Fur(17) Welz, B. Atomic Absorption Spectrometry, 3rd ed.; Wiley and Sons: New York, 1998.

Table 2. Comparison of Analytical Figures of Merit of DLAAS to Those of Other Techniques for Determination of Inorganic Manganese, CMT, and MMT DLAASa 1 1 4.7

figure of merit Mn detection limit, ng/mL CMT/MMT detection limit, ng(as Mn)/mL linear dynamic range

HCLAASb,17 3 NRe 3

ICPOESc,9 0.3 NRe 5

ICPMSd,9 0.0005 NRe 8

a This work; DLAAS ) diode laser atomic absorption spectrometry. b HCLAAS ) hollow cathode lamp atomic absorption spectrometry. c ICPOES ) inductively coupled plasma optical emission spectrometry. d ICPMS ) inductively coupled plasma mass spectrometry. e NR ) not reported.

Table 3. Comparison of Analytical Figures of Merit of HPLC-DLAAS to Those of HPLC-LEAFS for Determination of Inorganic Manganese, CMT, and MMT figure of merit Mn detection limit, ng/mL (pg) CMT/MMT detection limit, ng(as Mn)/mL (pg as Mn) linear dynamic range analysis time, min

HPLC-DLAASa 2 (400) 2 (400)

HPLC-LEAFSb,8 NRc 0.5 (10)

3 3 ) routine 6 ) higher resolution

3 20

a This work, HPLC-DLAAS ) high-performance liquid chromatography - diode laser atomic absorption spectrometry. b HPLC-LEAFS ) high-performance liquid chromatography - laser excited atomic fluorescence spectrometry. c NR ) not reported.

thermore, our detection limit was a factor of 3 worse than that of inductively coupled plasma optical emission spectrometry (ICPOES) and a factor of 2000 worse than that of ICP mass spectrometry (ICPMS). In addition to the obviously optimized experimental conditions for Mn measurement by HCLAAS,17 the relatively small difference in sensitivity between HCLAAS and DLAAS is caused by two factors. First, HCLAAS employs the 279.5-nm wavelength, which is approximately 10 times more sensitive than the 403.1-nm line used in this work.17 Second, our laser power was relatively low (170 nW). Using a detection bandwidth of 1 Hz, the absorption corresponding to the shot noise limitation for this laser intensity was approximately 2 × 10-4. A modest increase in laser power is expected to cause a significant improvement in the signal-to-noise ratio and detection limit by reducing the shot noise.15 However, even with the less sensitive line and the limited laser power employed here, our detection power is sufficiently high to allow measurement of inorganic and organic manganese compounds at concentration levels present in environmental and toxicological studies.8 DLAAS has intermediate performance on the basis of linear dynamic range (LDR). We obtained a linear range of almost 5 orders of magnitude, from 1 ng/mL to 50 µg/mL. This value is 2 orders of magnitude better than HCLAAS, the same as ICPOES, and three orders worse than ICPMS. Of course, the LDR will be improved for DLAAS if higher laser powers become available, allowing lower detection limits to be obtained. A comparison of our analytical data to previous speciation studies involving these organomanganese compounds is given in Table 3. The HPLC-DLAAS detection limit (2 ng/mL) was a factor of 2 worse than the detection limit without chromatography (1 ng/mL), because a transient signal was observed in the former experiment. The HPLC-DLAAS concentration detection limits are within a factor of 4 of the values obtained by Walton et al.8 (0.5 ng/mL for CMT and MMT) by HPLC-LEAFS. However, our

absolute detection limit (400 pg) was a factor 40 times higher than their values (10 pg). The disparity between concentration and absolute detection limits was induced by our use of a larger injection loop (0.2 mL compared with 0.02 mL). However, for many toxicological and environmental studies, we do not feel that the necessity of employing 0.2 mL per injection is a major disadvantage. The linear dynamic range (LDR) of the HPLC-DLAAS system was 3 orders of magnitude, which was the same as the value Walton et al.8 obtained by HPLC-LEAFS. The LDR was limited by the inability of the chromatographic system to resolve the very broad peaks obtained at high concentration levels. The analysis time for our system was very rapid using a mobile phase of 65:35 methanol/aqueous pH 4 buffer. The separation of inorganic manganese, CMT, and MMT were performed in three minutes, as illustrated in Figure 3. Better resolution was obtained using 60:40 methanol/aqueous pH 4 buffer as the mobile phase, as discussed below. Walton and co-workers’8 chromatographic system required a much longer analysis time of 20 min, due to the use of a different chromatographic column. Preconcentration of Organomanganese Compounds. We performed a simple series of experiments in order to evaluate the ability of the instrumentation to preconcentrate the organomanganese compounds, as performed by flow-injection techniques.18 We initiated the experiment by injecting 50 ng/mL of inorganic manganese, 50 ng/mL of CMT, and 50 ng/mL of MMT onto the HPLC system using 100% aqueous pH 4 buffer as the mobile phase. Under these conditions, the organomanganese compounds were completely retained on the column, and only the inorganic manganese was eluted and detected. We performed a second injection using 100% pH 4 buffer as the mobile phase and, again, observed elution of inorganic manganese. We then changed the mobile phase back to 65:35 methanol/aqueous pH 4 buffer. The organomanganese compounds were eluted with twice the integrated area as in Figure 3, as expected, demonstrating the effectiveness of the preconcentration procedure. These experiments demonstrate the potential of flow injection preconcentration procedures to improve the sensitivity of the HPLC-DLAAS. Improvement of Resolution for CMT and MMT. Although the chromatographic conditions shown in Figure 3 are adequate for most concentration levels of organomanganese compounds, an exception would be when one component is present in much higher concentration than another. We investigated the effect of eluent composition in order to separate a mixture of 200 ng/mL of CMT and 10 ng/mL of MMT. We were unable to study 10 ng/mL of CMT with 200 ng/mL of MMT, because, as discussed (18) Fang, Z.-L. Flow Injection Atomic Absorption Spectrometry, Wiley and Sons: Chichester, 1995.

Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

5383

Figure 5. Separation of 200 ng/mL of CMT and 10 ng/mL of MMT using 60:40 methanol/aqueous pH 4 buffer as eluent. Table 4. Analysis of Samples Spiked with MMT

sample gasoline human urine tap water

added concentration, µg(as Mn)/mL

reported solubility in water,3 µg(as Mn)/mL

determined value by HPLC-DLAAS, µg(as Mn)/mL

10, 29, 70

8.0 ( 0.8 4.2 ( 0.3 36 ( 2

8.3 4.0

a The uncertainties given are standard deviations from five replicate injections.

in the Experimental Section, the MMT contained 5% CMT as an impurity. We first employed 50:50 methanol/aqueous pH 4 buffer, which gave long analysis times (retention time for CMT was 8 min) and broad peaks. The use of 60:40 methanol/aqueous pH 4 buffer gave acceptable resolution for this mixture, as shown in Figure 5, with a total analysis time of 6 min. These results demonstrate the ability of the system to accurately determine larger differences in organomanganese compounds. Determination of MMT in Spiked Samples. To investigate the ability of the system to perform real sample analysis, MMT was added to samples of gasoline, human urine, and tap water and determined by HPLC-DLAAS (Table 4). The spiked samples were diluted to give MMT concentration levels between 40 and 100 ng/mL. The preparation of the spiked samples is decribed above (see Experimental Section). Calibration was performed by the construction of a calibration graph using standards prepared in 65:35 methanol/pH 4 buffer. The standards were injected on the HPLC in the same way as the samples were. The maximum allowable level of MMT in gasoline in the United States corresponds to a concentration of 8.3 µg (as Mn)/ mL, and hence, the gasoline was spiked to contain this MMT concentration. Our determined value of 8.0 ( 0.8 was in good agreement with the added concentration of MMT. Human urine was selected as an example of a biological sample that has been employed in toxicological studies. Walton et al.8 analyzed rat urine from animals that had been administered MMT subcutaneously. The only manganese compound observed in the rat urine was carboxycyclopentadienylmanganese tricarbonyl (CCMT, Figure 1). Their results suggested that CCMT may be responsible for organ damage resulting from MMT ingestion, although other undetected compounds could also be responsible 5384

Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

for the toxicity. This compound is not commercially available and was not available to us for these experiments. Therefore, we spiked human urine with MMT, which we expect to be similar enough to CCMT to evaluate the sensitivity of the HPLC-DLAAS system for toxicological studies. Walton and co-workers8 reported a detection limit for CCMT (1 ng/mL) within a factor of 2 of the value for MMT (0.5 ng/mL), justifying this assumption. MMT was added to the urine to give a final concentration of 4 µg(as Mn)/mL, the approximate concentration of CCMT reported by Walton et al.8 Our measured value of 4.2 ( 0.3 µg/ mL is in good agreement with the amount added to the sample. These results indicate that this system is suitable for the determination of MMT and its metabolites in biological samples. The ability to determine MMT in water is of significance because of its relative stability in groundwater. Garrison et al.3 reported that under anaerobic conditions in contact with aquifer material, MMT may have as long a half-life as 568 days. These results imply that leakage from underground gasoline tanks may induce the accumulation of MMT in groundwater, suggesting the importance of methods to determine MMT in water. We obtained a value of 36 ( 2 µg(as Mn)/mL for the solubility of MMT in water. This value is within the range of three values (10 µg/mL, 29 µg/mL, and 70 µg/mL) reported in the literature. For all spiked samples, the matrix did not affect the retention time and peak shapes, probably because our detection limits were rather low, allowing large dilution factors to be used. CONCLUSIONS HPLC-DLAAS was shown to be a suitable technique for the speciation of MMT, CMT, and inorganic manganese in gasoline, urine, and water samples. Its characteristics include low detection limits (2 ng/mL), a long linear dynamic range (3 orders of magnitudes), rapid analysis time (3 min), and low cost of instrumentation and operation. With appropriate modifications to the instrumentation described here, flow injection techniques may be employed to further improve the sensitivity. Minor modifications to the eluent composition have been reported that can be used to improve the separation when one organomanganese compound is present at much higher concentration than the other. Finally, it was demonstrated that, although the organomanganese compounds should be stored out of direct laboratory light, it is not necessary to make modifications to the lighting of laboratories for the determination of MMT. Despite the already impressive analytical performance of HPLC-DLAAS, further improvements in sensitivity are expected soon. Recently a new generation of laser diodes, which emit 5 mW in the blue spectral range (390-410 nm) are commercially available from Nichia Chemical Industries (Tokushima, Japan).19 These blue GaN laser diodes (type: NLHV 500) are currently being tested in our laboratory for manganese determination. Prelimary measurements provided, already, a significant improvement in the sensitivity of DLAAS. Instead of 2 × 10-4 absorption, as found with 170 nW of frequency doubled laser radiation, we were able to detect a manganese specific absorption of 5 × 10-5 with the new laser diode. However, we used the Mn 403.3 nm absorption line which is about 30% weaker than the 403.1-nm line (19) Nakahara, S.; Senoh, M.; Nagahama, S.; Iwasa, N.; Matsushita, T.; Kiyoku, H.; Sugimoto, Y.; Kozaki, T.; Umemoto, H.; Sano, M.; Chocho, K. Appl. Phys. Lett. 1998, 72, 211-213.

used throughout the present investigation. Assuming that shot noise is limiting, the increase in laser power should improve the detection limit by more than 2 orders of magnitude, on the basis of our experience with red laser diodes and DLAAS measurements in modulated plasmas.10 However, taking into account our experience with DLAAS in analytical flames, other noise sources become limiting at absorption values below 1 × 10-5.20 The reason no lower absorption was measured yet with the higher laser power will be discussed in detail elsewhere. In conclusion, a considerable amount of information is required in order to evaluate the environmental and toxicological impact of MMT. For example, the degradation products of MMT and CMT must be identified in the presence and absence of light. The toxicity values of these compounds and their metabolites need to be examined through the use of animal models. The combustion products of MMT must be identified, and the toxicity of these (20) Groll, H.; Schnu ¨ rer-Patschan, C.; Kuritsyn, Yu.; Niemax, K. Spectrochim. Acta, Part B 1994, 49, 1463-1472.

compounds must be evaluated.1 Clearly, an enormous amount of interdisciplinary research and analysis are required to evaluate the impact of MMT in gasoline. The results presented here demonstrate that HPLC-DLAAS can play a valuable role in these investigations. ACKNOWLEDGMENT This paper is dedicated to Professor Dr. Dieter Klockow on the occasion of his 65th birthday. D.J.B. would like to acknowledge support from NSF Award 9902679. Furthermore, funding by the Ministerium fu¨r Schule und Weiterbildung, Wissenschaft und Forschung of the state Northrhine-Westphalia and the Ministerium fu¨r Bildung, Wissenschaft, Forschung und Technologie of the Federal Republic of Germany is gratefully acknowledged. Received for review May 27, 1999. Accepted September 21, 1999. AC990570L

Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

5385