On-Line Technique To Determine the Isotopic Composition of Total

Oct 19, 2007 - Dries Huygens,*,† Pascal Boeckx,† Jan Vermeulen,† Xavier De Paepe,‡ Andrew Park,§. Sam Barker,§ and Oswald Van Cleemput†...
0 downloads 0 Views 324KB Size
Download PDF
Anal. Chem. 2007, 79, 8644-8649

On-Line Technique To Determine the Isotopic Composition of Total Dissolved Nitrogen Dries Huygens,*,† Pascal Boeckx,† Jan Vermeulen,† Xavier De Paepe,‡ Andrew Park,§ Sam Barker,§ and Oswald Van Cleemput†

Laboratory of Applied Physical Chemistry ISOFYS, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium, Shimadzu Europe, Duisburg, Germany, and SerCon Limited, Cheshire, United Kingdom

A new on-line analytical setup for 15N measurements of total dissolved nitrogen (TDN) has been developed through the coupling of a high-temperature catalytic (Ce(IV)O2) oxidation furnace, a Cu reduction furnace, and an isotope ratio mass spectrometer. The detection limit for accurate δ15N measurements is 20 mg of N L-1. For N-containing compounds dissolved in water, a standard deviation on N concentration measurements of 0.2 mg of N L-1, independent of N concentration, has been found. Reproducibility on δ15N increased with increasing N concentration, with standard deviations varying from 0.8 to 0.1‰ in the range of 20-100 mg of N L-1. Salt matrixes, in which the N species might be dissolved, could influence the analysis capacity and continuity, mainly at concentrations above 0.1 M. To our knowledge, this system is the first successful on-line setup capable of performing routine δ15N and N concentration measurements of the TDN pool. Potential applications are large and are believed to result in an increased insight in N cycling and dissolved organic nitrogen behavior in terrestrial and aquatic ecosystems. Improvements in stable isotope analysis techniques have often been proposed as a potential means for increased understanding of biogeochemical nutrient cycling. Specifically, McDowell1 stated that analytical research on the isotope signal of dissolved organic matter (DOM) should be undertaken in order to resolve its functional role in terrestrial ecosystems. Many studies suggest a key role for DOM in terrestrial and aquatic ecosystems over a broad range of latitudes of the world.2-5 The potential susceptibility of DOM to climate change6 and human impact7 emphasizes the need to include DOM in ecosystem studies. On balance, DOM * To whom orrespondence should be addressed. Tel: +32 9 264 99 07. Fax: +32 9 264 62 30. E-mail: [email protected]. † Ghent University. ‡ Shimadzu Europe. § SerCon Limited. (1) McDowell, W. H. Geoderma 2003, 113, 179-186. (2) Jones, D. L.; Shannon, D.; Murphy, D. V.; Farrar, J. Soil Biol. Biochem. 2004, 36, 749-756. (3) Murphy, D. V.; Macdonald, A. J.; Stockdale, E. A.; Goulding, K. W. T.; Fortune, S.; Gaunt, J. L.; Poulton, P. R.; Wakefield, J. A.; Webster, C. P.; Wilmer, W. S. Biol. Fertil. Soils 2000, 30, 374-387. (4) Neff, J. C.; Chapin, F. S.; Vitousek, P. M. Front. Ecol. Environ. 2003, 1, 205-211. (5) Zhong, Z.; Makeschin, F. Soil Biol. Biochem. 2003, 35, 333-338. (6) van Breemen, N. Nature 2002, 415, 381-382. (7) Chantigny, M. H. Geoderma 2003, 113, 357-380.

8644 Analytical Chemistry, Vol. 79, No. 22, November 15, 2007

affects many processes in soil and water, including soil and water pollution and global warming.8 Though advances have been made in isotope determination of the dissolved organic carbon pool,9,10 no precise and routine method is currently available for dissolved organic nitrogen (DON).11 Isotope DON analysis can be achieved by measuring the isotopic signal of total dissolved nitrogen (TDN) and applying isotope correction for dissolved inorganic nitrogen (DIN) pools.12,13 Alternatively, DIN components can be removed14 from the sample before analysis. Existing attempts to determine TD15N involve offline sample preparation, including persulfate oxidation15-17 or Kjeldahl digestion,18 followed by diffusion steps.19 The off-line preparation techniques are time-consuming while accuracy and reproducibility often do not suffice. A first on-line method was presented by Russow et al.20 but showed significant shortcomings, compelling applying correction factors. Moreover, the isotopic detection limits of this method equaled 0.5 atom %, precluding natural-abundance isotope studies. In order to meet scientific needs, a new method has been developed to determine the 15N content of the TDN pool using a high-temperature catalytic oxidation (HTCO) furnace, a heated dehumidification system, a reduction furnace, and an isotope ratio mass spectrometer (IRMS) in series. Preliminary results11 indicated the potential of this system for accurate and precise 15N measurements up to natural-abundance levels. In this paper, the final system configuration will be described, and a complete set (8) Kalbitz, K.; Kaiser, K. Geoderma 2003, 113, 177-178. (9) Bouillon, S.; Korntheuer, M.; Baeyens, W.; Dehairs, F. Limnol. Oceanogr. Methods 2006, 4, 216-226. (10) St-Jean, G. Rapid Commun. Mass Spectrom. 2003, 17, 419-428. (11) Huygens, D.; Boeckx, P.; Vermeulen, J.; De Paepe, X.; Park, A.; Barker, S.; Pullan, C.; Van Cleemput, O. Rapid Commun. Mass Spectrom. 2005, 19, 3232-3238. (12) Lehmann, M. F.; Bernasconi, M.; McKenzie, J. A. Anal. Chem. 2001, 73, 4717-4721. (13) Silva, S. R.; Kendall, C.; Wilkison, D. H.; Ziegler, A. C.; Chang, C. C. Y.; Avanzino, R. J. J. Hydrol. 2000, 228, 22-36. (14) Feuerstein, T. P.; Ostrom, P. H.; Ostrom, E. Org. Geochem. 1997, 27, 363370. (15) Dail, D. B.; Davidson, E. A.; Chorover, J. Biogeochemistry 2001, 54, 131146. (16) Seely, B.; Lajtha, K. Oecologia 1997, 112, 393-402. (17) Whalen, J. K.; Parmelee, R. W.; Subler, S. Biol. Fertil. Soils 2000, 32, 347352. (18) Bremmer, J. M.; Mulvaney, C. S. In Methods in soil analysis; Page, A. L., Miller, R. A., Keeney, D. R., Eds.; ASA and SSSA: Madison, WI, 1982, pp 595-624. (19) Stark, J. M.; Hart, S. C. Soil Sci. Soc. Am. J. 1996, 60, 1846-1855. (20) Russow, R.; Hupka, H.-J.; Go ¨tz, A.; Apelt, B. Isot. Environ. Health Stud. 2002, 38, 215-225. 10.1021/ac070607z CCC: $37.00

© 2007 American Chemical Society Published on Web 10/19/2007

Figure 1. On-line setup of the described TDN-IRMS system.

of results will be presented in order to address the application potential of the analytical setup in N research. Analytical Setup. A general scheme of the presented analytical method is shown in Figure 1. The coupled system consists of a TOC analyzer (Shimadzu TOC-VCSH, Shimadzu), an elemental analyzer (ANCA GSL, PDZ Europe), and an isotope ratio mass spectrometer (20-20, SerCon). This analytical setup will be named TDN-IRMS throughout the text. We refer to Huygens et al.11 for basic system description. Helium, at a flow rate of 120 mL min-1, is used as carrier gas, and an oxygen pulse, just before sample combustion, is admitted for 15 s to provide an oxygen-rich combustion environment. Sample combustion occurs in the HTCO furnace of the TOC analyzer at 900 °C, where transformation of all N components to nitric oxide (NO) and nitrogen dioxide (NO2) takes place, while carbon is transformed into CO2. Cerium(IV) oxide (granules, 1-2 mm in diameter; Merck) is used as a catalyst in the HTCO furnace, topped with 2 cm of quartz wool. As injector block movement before sample injection introduces atmospheric air into the combustion column,11 a He atmosphere is created around the sliding injector block. This is accomplished by surrounding the block with a “He chamber” purged with a constant He flow of 300 mL min-1 at 0.5 bar of overpressure.11 This purge flow is the countercurrent He gas of the Nafion dehumidification system (see further) from which the water is removed using a combination of a water trap and an anhydrous magnesium perchlorate scrubber. After sample combustion, the temperature of the gas stream is controlled and maintained at 110°C by heating tape (Fisher Scientific Bioblock), controlled via a thermocouple, wound around the stainless steel tube. The temperature controlled gas stream is passed through a heated Nafion (MDH gas dryer, Perma pure) dehumidification system. The combination of the heated stainless steel tubing and Nafion

system replaces the standard Peltier dehumidification element in the commercial TOC analyzer. The Nafion tube is able to dehumidify a gas stream without loss of NOx components21 until equilibrium between current and countercurrent He gas (300 mL min-1) is reached. The heating tape and Nafion housing are necessary to prevent condensation of the gas stream before dehumidification, an effect that would result in failure of the Nafion system. Afterward, the NOx gas leaving the dehumidification system is transferred to the reduction furnace of an elemental analyzer via stainless steel tubing. The combustion products pass through a furnace containing Cu at 600 °C, where nitrogen oxides are reduced to elemental nitrogen, excess oxygen is absorbed, and halogens are scrubbed. Excess water is removed by a trap containing anhydrous magnesium perchlorate, and CO is transformed into CO2 using Schultz reagent, after which CO2 is trapped by Carbosorb. The gas stream then passes through a gas chromatograph (GC) column for separation of the N2. Finally, the gas stream is admitted to the IRMS, where the N2 isotopic species are ionized (28N2+, 29N2+, 30N2+) and separated to their mass-tocharge ratios in a magnetic field. The configuration of the IRMS inlet crimp was modified in order to direct maximum ion quantities toward the universal triple collector of the IRMS. The ion current intensities corresponding to the three isotopic species are detected in the universal triple collector of the IRMS and used for the calculation of the total N mass and 15N enrichment. A specific software package (Callisto version 4.0.22, SerCon) was installed to automate the coupled system. The software controls autosampler (AIM 3300, A.I. Scientific) position, TOC analyzer start and stop functions, and oxygen pulse delivery. (21) Bateman, J.; Klemn, R.; Brookhaven National Laboratory, New York, 1977.

Analytical Chemistry, Vol. 79, No. 22, November 15, 2007

8645

In order to remove atmospheric N2 dissolved in the sample matrix, the sample is purged with He for 1 min just before injection using the “nonpurgeable organic carbon” mode of the TOC analyzer. A batch of 60 samples can be analyzed in one IRMS run, while total analysis time for one sample equals 405 s. Experimental Tests. To check proper functioning of the online technique, the following issues were tested: (1) combustion efficiency and accuracy of N concentration measurements; (2) accuracy of δ15N measurements; (3) accuracy and reproducibility of δ15N values in function of N concentration; (4) cross-contamination; (5) the effect of salts in the sample matrix. In the first set of experiments, pure and mixtures of standard N-containing compounds (SNCs) were dissolved in MilliQ water. The SNCs (NaNO3, NH4Cl, (NH4)2SO4, urea, 4-nitrophenol, N-acetyl-D-glucosamine, L-glutamine, DL-R-alanine, 2,6-pyridinedicarboxylic acid) included reduced and oxidized molecules varying in molecular weight, complexity, and structure, reflecting typically occurring N-containing compounds in environmental ecosystem studies. Using two international 15N standards (I15NS) (IAEA-NO3 and USGS-32) with a δ15N value versus air of 4.7 ( 0.2 22 and 180.0 ( 1.0‰,22 respectively, the analysis accuracy was checked at an N concentration of 20 mg of N L-1. Sample combustion efficiency was tested for the pure and mixtures of SNCs at 20 mg of N L-1 after deducing a calibration curve relating N2 peak area to N concentration of the IAEA-NO3 I15NS. Accuracy of the 15N analysis was tested at an N concentration of 20 mg of N L-1 by comparing the δ15N value of the water-dissolved SNCs analyzed by TDNIRMS to the δ15N value of the SNCs measured in its solid state using an EA-IRMS configuration (ANCA-SL, PDZ Europe; 2020, SerCon). δ15N accuracy and reproducibility was tested at different N concentrations (2.5, 5, 10, 20, 40, and 100 mg of N L-1) for N-acetyl-D-glucosamine and (NH4)2SO4. The effect of crosscontamination between samples was checked in specific “memory effect” tests, where a 20 mg of N L-1 -8.7‰ (δ15N) standard (2,6pyridinedicarboxylic acid) and a 20 mg of N L-1 180‰ (δ15N) standard (USGS-32) were injected, each followed by 6 injections of a 20 mg of N L-1 4.7‰ (δ15N) standard (IAEA-NO3). As salts are inherently associated with ecosystem samples in both aquatic and terrestrial ecosystem studies, the influence of salt concentrations was tested by measuring the evolution of N concentration and δ15N of a 40 mg of N L-1 N-acetyl-D-glucosamine standard in a 0.1 and 1 M KCl matrix. As such, the effect of salts on analytical reproducibility and accuracy could be checked. For all measurements, a standard injection volume of 1000 µL was used. All different parameters were studied in at least five repetitions. Data Analysis. Data on isotope values are reported in permil (‰) as δ15N versus air:

δ15N )

Rsample - Rreference × 1000 (‰) Rreference

where Rsample is the ratio of 15N/14N in the sample and Rreference is the ratio of the international reference, being atmospheric nitrogen (22) Bo¨hlke, J. K.; Coplen, T. B. Reference and intercomparison materials for stable isotopes of light elements; IAEA: Wien, 1995; pp 51-66.

8646

Analytical Chemistry, Vol. 79, No. 22, November 15, 2007

(0.003 676 5). The data reported for the TDN-IRMS system are measured using IAEA-NO3 as a working reference, dissolved at the same N concentration and in the same sample matrix as the measured SNCs or I15NS. It should be noted that a dissolved I15NS was used as reference (isotopic composition assumed to be equal to its solid state δ15N value) as no δ15N certified reference for TDN is available. Data reported on the EA-IRMS configuration (solid SNCs) are measured using flour (δ15N ) 2.69 ( 0.2‰, certified by Iso-Analytical) as working standard. Blank correction for TDN-IRMS was done by subtracting the I28 and I29 beam areas (averaged over 5 repetitions) from the I28 and I29 beam areas of the batch references and batch samples. RESULTS AND DISCUSSION Blank Measurements and Peak Shape. Figure 2 shows standard TDN-IRMS chromatograms for the described on-line setup of a 1000-µL injection of a MilliQ H2O sample (Figure 2a), and a 1000-µL injection of a 20 mg of N L-1 solution (Figure 2b). The blank N2 peak area (sum of N2 isotopic species 28N2+, 29N2+, and 30N2+) had an average value (n ) 5) of 5.18 × 10-9 A‚s, with a standard deviation of about 2.1 × 10-10 A‚s. The blank 29N2+/ 28N + ratio showed a considerable variation (raw data varied from 2 0.006 977 to 0.007 122, corresponding to absolute 15N/14N values in the natural-abundance range close to 0.003 676 5). The 20 ppm N sample had an average N2 peak area of 5.62 × 10-8 A‚s (n ) 5). We expect that both movements of the sliding injector block (Figure 1) and changes in flow rate after sample injection could explain the baseline instability and associated blank N2 peak areas (Figure 2a). In preliminary tests, the He purge pressure and flow rate into the He chamber were varied in order to minimize the blank N2 peak area. Despite optimization, the blank 29N2+/28N2+ ratio close to natural-abundance N2 values still suggest minimal N2 inflow during sliding injector block movements. Moreover, we noticed changes in flow rate and internal system pressures due to water expansion and desorption after sample injection onto the heated catalyst structure, an effect potentially causing baseline instability. The use of a relatively high inlet pressure and limiting the flow rate via a restriction before the GC column could potentially further minimize blank N2 peak areas. Combustion Efficiency. Figure 3 shows the measured N concentrations for different mixtures of pure and mixtures of SNCs at 20 mg of N L-1 (n ) 5). It is shown that all components are measured accurately with a fair reproducibility, indicating 100% combustion efficiency for the different components. An average standard deviation of 0.2 mg of N L-1 was found, corresponding to a standard error of 1%. Sample concentration linearity was checked for KNO3 (IAEANO3), resulting in a linear (N2 peak area (A‚s) ) 2.61 × 10-9 × N concentration (mg of N L-1) + 4.69 × 10-9, r2 ) 0.9995) range for the tested concentrations from 0 to 100 mg of N L-1 (n ) 5). The standard deviation of the N2 peak area varied between 4 × 10-10 and 9 × 10-10 A‚s. A small offset, comparable to the blank N2 peak area, was noticed. The standard deviation was independent of the N concentration of the measurement, and equals on average 0.2 mg of N L-1. System Accuracy. Measuring an I15NS (USGS-32, δ15N ) 180 ( 1.0‰) at 20 mg of N L-1, using a different I15NS (IAEA-NO-3, δ15N ) 4.7 ( 0.2‰) as reference confirmed the accuracy of isotope

Figure 2. (a) Standard TDN-IRMS chromatogram of a MilliQ H2O blank (1000-µL injection). The upper line indicates the chromatogram of 28N + (left axis), the lower line the chromatogram of 29N + (right axis). (b) A standard TDN-IRMS chromatogram of a 20 mg of N L-1 standard 2 2 (1000-µL injection). The upper line indicates the chromatogram of 28N2+ (left axis), the lower line the chromatogram of 29N2+ (right axis).

analysis by TDN-IRMS (δ15N measured ) 180.6‰, standard deviation 0.6‰, n ) 7). Accuracy and Reproducibility of δ15N in Function of N Concentration. The δ15N values of the pure and mixtures of SNCs (20 mg of N L-1) dissolved in MilliQ water and measured as solid samples are shown in Figure 4 and Table 1 (n ) 5). Over a broad range of δ15N values (-8.7 to 1.3‰), the pure and mixtures of SNCs showed a good correspondence in δ15N values (indicated by the position of the SNCs on the y ) x line). The latter indicates that accuracy of δ15N measurements is maintained during TDNIRMS analysis. Possibly, the slightly higher standard deviations for 4-nitrophenol and 2,6-pyridinedicarboxylic acid are a result of the limited solubility of the latter components, decreasing sample homogeneity. The standard deviation on δ15N value was slightly higher for the water-dissolved SNCs (on average 0.9‰) compared to the solid standards (on average 0.3‰). The analytical accuracy and reproducibility in function of N concentration is shown in Figure 5 for N-acetyl-D-glucosamine and (NH4)2SO4 (n ) 5). In a sensitivity analysis, blank N2 peak area

and blank isotope ratio were altered within the obtained measurement reproducibility. From Figure 5, it can be deduced that the detection limit of the TDN-IRMS system is situated at ∼20 mg of N L-1 for isotope N measurements at natural abundance. Similarly, a clear trend toward higher reproducibility in function of N concentration can be noticed for both components (δ15N standard deviation on average 5.4, 3.3, 1.8, 0.8, 0.4, and 0.1‰ for, respectively, 2.5, 5, 10, 20, 40, and 100 mg of N L-1) (Figure 5). The high variability associated with the blank 29N2+/28N2+ ratio (see above) is expected to be responsible for the lack of accuracy and reproducibility of δ15N measurements at low N concentrations. The contribution of the blank to the total N2 peak area decreases exponentially with increasing N concentration (varying from 48.5% at 2.5 mg of N L-1 to 1.9% at 100 mg of N L-1), explaining the increased accuracy and reproducibility with increased N concentration. The detection limit of 20 mg of N L-1 is relatively high for environmental N studies. As the blank N2 peak area is independent of the sample volume, system capabilities can eventually be Analytical Chemistry, Vol. 79, No. 22, November 15, 2007

8647

Table 1. δ15N Values for the Different Pure and Mixtures of Standard N-Containing Compounds (SNCs) Measured in MilliQ Water (20 mg of N L-1, TDN-IRMS) and in Solid Form (EA-IRMS) (Standard Deviations in Parentheses, n ) 5) δ15N H2O dissolved (‰) urea

DL-R-alanine

2,6-pyridinedicarboxylic acid N-acetyl-D-glucosamine NaNO3 4-nitrophenol NH4Cl (NH4)2SO4 L-glutamine mix Aa mix Bb mix Cc mix Dd Figure 3. Measured N concentrations for a 20 mg of N L-1 solution of the different pure and mixtures of the standard N-containing compounds dissolved in MilliQ H2O using the TDN-IRMS setup (error bars indicate plus minus one standard deviation, n ) 5) (mix A ) 50% NH4Cl, 50% NaNO3; mix B ) 30% 4-nitrophenol, 30% urea, 40% DL-R-alanine; mix C ) 30% 2,6-pyridinedicarboxylic acid, 30% N-acetyl-D-glucosamine, 40% (NH4)2SO4; mix D ) 20% NaNO3, 20% 4-nitrophenol, 20% urea, 20% 2,6-pyridinedicarboxylic acid, 20% NH4Cl).

-1.52 -0.15 -7.36 -2.15 -0.32 -5.35 0.34 1.16 -0.08 -0.51 -1.43 -3.23 -3.05

(1.03) (1.50) (1.79) (0.81) (0.89) (1.26) (0.57) (0.21) (0.54) (0.85) (0.67) (1.17) (0.89)

δ15N solid (‰) -1.87 0.49 -8.66 -2.18 -0.06 -4.94 0.32 1.28 0.47 0.13 -1.85 -2.74 -2.85

(0.79) (0.45) (0.19) (0.12) (0.51) (0.20) (0.17) (0.14) (0.38) (0.27) (0.30) (0.09) (0.20)

a Mix A: 50% NH Cl, 50% NaNO . b Mix B: 30% 4-nitrophenol, 30% 4 3 urea, 40% DL-a-alanine. c Mix C: 30% 2,6-pyridinedicarboxylic acid, 30% N-acetyl-D-glucosamine, 40% (NH4)2SO4. d Mix D: 20% NaNO3, 20% 4-nitrophenol, 20% urea, 20% 2,6-pyridinedicarboxylic acid, 20% NH4Cl.

Figure 5. Measured δ15N values in function of the N concentration for N-acetyl-D-glucosamine (b) and (NH4)2SO4 (O) using the TDNIRMS system (error bars indicate plus minus one standard deviation) (gray zones are indicative for the δ15N value of the respective standard N-containing compound measured in solid form using EA-IRMS). Figure 4. Relationship between δ15N values measured using TDNIRMS in MilliQ H2O dissolved form (20 mg of N L-1) and δ15N values measured in solid form (EA-IRMS) for the different pure and mixtures of standard N-containing compounds (error bars indicate plus minus one standard deviation, n ) 5) (pyridine ) 2,6-pyridinedicarboxylic acid; 4-nitrophenol ) 4-nitrophenol; mix C ) 30% 2,6-pyridinedicarboxylic acid, 30% N-acetyl-D-glucosamine, 40% (NH4)2SO4; mix D ) 20% NaNO3, 20% 4-nitrophenol, 20% urea, 20% 2,6-pyridinedicarboxylic acid, 20% NH4Cl; mix B ) 30% 4-nitrophenol, 30% urea, 40% DL-R-alanine; glucosamine ) N-acetyl-D-glucosamine; urea ) urea; NaNO3 ) sodium nitrate; mix A ) 50% NH4Cl, 50% NaNO3; NH4Cl ) ammonium chloride; alanine ) DL-R-alanine; glutamine ) Lglutamine; (NH4)2SO4 ) ammonium sulfate).

increased by raising the sample volume to a maximum of 2000 µL. However, in order to maintain catalyst functionality over a longer period and to control internal pressures inside the TOC analyzer, it is advised to refrain from this option. Many TDN concentration techniques have been described for environmental N studies such as free evaporation (nitrogen blowdown),23 rotary 8648 Analytical Chemistry, Vol. 79, No. 22, November 15, 2007

evaporation,24 or freeze-drying.25 However, all these techniques are documented to show small losses or gains of N compounds,26,27 possibly introducing fractionation effects, and limiting its use for isotopic TDN measurements up to natural abundance level. Memory Effects. Injection of an -8.7‰ (δ15N) standard had no influence on the δ15N value (4.7‰) of the upcoming component, indicating the absence of cross-contamination when measuring 15N at natural abundance (Figure 6). However, after the injection of the 180‰ (δ15N) standard, the system needed three successive injections to recover to accurate δ15N values for the 4.7‰ (δ15N) standard (Figure 6). This indicates the need to insert blank samples when analyzing 15N at enriched level. (23) Becker, G.; Colmsjo, A.; Ostman, C. Environ. Sci. Technol. 1999, 33, 13211327. (24) Macrae, J. D.; Hall, K. J. Environ. Sci. Technol. 1998, 32, 3809-3815. (25) Lehmann, J.; Lilienfein, J.; Rebel, K.; do Carmo, Lima, S.; Wilcke, W. Soil Use Manage. 2004, 20, 163-172. (26) Alavoine, G.; Nicolardot, B. Anal. Chim. Acta 2001, 445, 107-115. (27) Cheng, C. C. Polycyclic Aromat. Compd. 2003, 23, 315-325.

Figure 6. Evolution in δ15N signal of a 4.7‰ (δ15N) standard (IAEANO3) after injecting respectively a 180‰ (δ15N) standard (USGS32, b) and a -8.7‰ (δ15N) standard (2,6-pyridinedicarboxylic acid, O) (test performed using 20 mg of N L-1 standards).

Figure 7. Box plots for the TDN-IRMS measured δ15N values (box plot on the left side) and N concentrations (box plot on the right side) for 27 injections of N-acetyl-D-glucosamine (40 mg of N L-1) dissolved in 0.1 M KCl (δ15N N-acetyl-D-glucosamine ) 2.2 ( 0.1‰ determined via EA-IRMS). The lower box plot boundary indicates the 25th percentile; the line within the box plot indicates the median; the upper box plot boundary indicates the 75th percentile. The whiskers (error bars) above and below indicate the 90th and 10th percentile, respectively. The dots indicate outliers.

System Sustainability and Maintenance for Water Samples with Low Salt Solutions. The lifetime of the oxidation and reduction tubes limits the continuous use of the TDN-IRMS system. The Cu reduction column should be replaced after ∼60 samples. Every 180 samples, it is advised to rinse the heated dehumidification system and stainless steel tubing via the repeated, alternated injection of a 1 M HCl and H2O solution and to replace the Ce(IV)O2 catalyst. Afterward, the newly installed catalyst should be conditioned via 20 1000 µL injections of a H2O sample, and this without connecting the TOC analyzer to the EAIRMS (Figure 1). As such, the internal pressures can be controlled in order to stick the quartz wool layer to the Ce(IV)O2 catalyst. Sample Matrix Effects. The influence of sample salt matrix was checked by dissolving a 40 mg of N L-1 SNC (N-acetyl-Dglucosamine, δ15N ) -2.2 ( 0.1‰) in a 0.1 M KCl matrix (Figure 7). Accuracy of δ15N could be maintained for a long period of ∼27 samples with a similar isotope reproducibility (average δ15N equals with -2.3‰ with a standard deviation of 0.5‰) to H2O-dissolved

standards. However, the standard deviation of the N concentration appeared to be slightly higher compared the H2O-dissolved standards (on average 0.8 mg of N L-1). Likewise, it was noticed that the IRMS background level and flow rate varied slightly during the 0.1 M KCl batch. From the 28th sample on, a clear decrease in flow rate and N2 peak area could be noticed, resulting in a decreased analytical performance, especially on the N concentration. The latter suggests catalyst or tube clogging, indicating the need to stop the batch, to replace or regenerate the catalyst structure, and to perform system maintenance. Measuring standards in a 1 M KCl matrix had a drastic effect on peak shape, flow rate continuity, N2 peak area, and isotope N ratio (data not shown), indicating the failure of the TDN-IRMS setup at high salt levels. Potential of the Analytical Setup for Isotope Ecosystem N Studies. Results indicate that the proposed on-line TDN-IRMS setup is capable of oxidizing and measuring the N concentration of a wide range of components while accuracy of the δ15N measurements is maintained. Standard deviation of N concentration measurements for water-dissolved SNCs equals 0.2 mg of N L-1, independent of the sample concentration. The detection limit of the system for accurate δ15N determinations is situated at ∼20 mg of N L-1 whereas the δ15N reproducibility increases in function of the N concentration (standard deviation equals 0.8‰ at 20 mg of N L-1, 0.1‰ at 100 mg of N L-1). It was shown that the salt matrix, in which the N species might be dissolved, could influence the analysis capacity and continuity, especially at high salt levels. We advise working with low salt concentrations. Dissolving SNCs (40 mg of N L-1) in 0.1 M KCl resulted in a slight effect on analytical reproducibility (standard deviation of 0.8 mg of N L-1 and 0.5‰ for N concentration and δ15N, respectively). To our knowledge, this system is the first successful on-line setup capable of performing δ15N and N concentration measurements of the TDN pool. N concentration and δ15N measurements of TDN can be performed simultaneously and on-line, requiring a fair analysis time of less than 7 min per sample. The continuous use of the system is limited to ∼25 samples when analyzing 0.1 M KCl extracts. The analytical performance of the TDN-IRMS system allows 15N tracer studies, as well as 15N natural-abundance studies as a potential application. To conclude, we claim that the described on-line TDN-IRMS setup has a high potential and applicability in ecosystem N research. Especially, studies focusing on the functional role of DON should benefit from this advance in analytical N isotope research. ACKNOWLEDGMENT This research was supported by the Fund for Scientific ResearchsFlanders (Belgium) (FWO) (Grant G.0426.04). We thank Shimadzu Europe for the supply of spare parts for the TOCVCSH analyser. Received for review March 27, 2007. Accepted August 7, 2007. AC070607Z

Analytical Chemistry, Vol. 79, No. 22, November 15, 2007

8649