ConFlo III An Interface for High Precision d 13 C and d 15 N Analysis with an Extended Dynamic Range

Transcription

1 RAPID COMMUNICATIONS IN MASS SPECTROMETRY ConFlo III An Interface for High Precision d 13 C and d 15 N Analysis with an Extended Dynamic Range Roland A. Werner*, Beate A. Bruch and Willi A. Brand Max-Planck-Institut für Biogeochemie, P.O. Box , Jena, Germany A newly developed interface coupling a CHN combustion device (elemental analyser EA ) to an isotope ratio mass spectrometer is described and evaluated. The purpose of the device is to extend the dynamic range of d 13 C and d 15 N analysis from less than 2 orders of magnitude to more than 3 orders of magnitude. Carbon isotope ratio measurements of atropine as a model compound have been performed analysing between 1 mg to 5 mg C with acceptable to excellent precision (0.6 to 0.06%, -notation). The correction due to the blank signal is critical for sample amounts smaller than 4 mg C. The maximum sample weight is determined by the combustion capacity of the EA. Larger sample amounts are measured using dilution of a small part of the EA effluent with helium. The dilution mechanism works virtually free of isotope fractionation. Bulk combustion at temperatures beyond 1000 C (Dumas combustion) can be applied for the on-line measurement of 15 N/ 14 N, 113 C/ 12 C 2 and 34 S/ 32 S 3,4 ratios of most organic and a number of inorganic sample materials. Combined measurement of 13 C and 15 N values from the same sample combustion have been reported using either cryogenic separation followed by dual-inlet measurement 5 or on-line with He as carrier gas, 6,7 in particular for small sample amounts. 8 Recently isotope ratio monitoring (îrm) of 18 O/ 16 O 9 12 and 2 H/ 1 H 13 ratios by pyrolysis, using elemental analysers at elevated temperatures, has been described. In special cases intra molecular 13 C values of substances are also accessible with this techniques. 10,14 For such purposes, automatic combustion/pyrolysis units (elemental analysers, EA) have been coupled on-line with isotope ratio mass spectrometers (irm-eams, EA-IRMS or bulk sample isotope analysis 15 (BSIA), refer also to Fig. 1). Provided the combustion/reduction or the pyrolysis reaction is quantitative the product gases (CO 2,N 2,SO 2 and H 2 O for combustion, CO and H 2 for pyrolysis) represent the bulk isotopic composition of the sample. For isotope ratio determination the product gases must be purified or separated from each other before being introduced into the mass spectrometer. Almost all organic materials contain only small amounts of nitrogen and a large amount of carbon. For combined 13 C and 15 N measurements of the same sample this poses a major challenge. In order to measure nitrogen with the required precision for natural abundance studies enough nitrogen must enter the ion source. Thus the carbon peak will be far too large and must be attenuated. The attenuation has to be designed in a way that isotope fractionation is completely avoided. *Correspondence to: R. A. Werner, Max-Planck-Institute für Biogeochemie, P.O. Box , Jena, Germany. CCC /99/ $17.50 EXPERIMENTAL Instrumentation An NA 1110 CN elemental analyser (CE Instruments, Rodano, Italy) was interfaced to a MAT 252 isotope ratio mass spectrometer (Finnigan MAT, Bremen, Germany) via a modified Finnigan ConFlo II 2 open split interface 16 (Fig. 1). Our ConFlo III system allows simultaneous high precision measurement of the -values of GC-separated N 2 and CO 2 peaks generated from a single sample (Fig. 2) with a greatly exhanced dynamic range compared to the ConFlo II. 17 The commercially available NA 1110 analyser is equipped with a pressure regulator followed by a flow controller for the He carrier gas. In order to minimise the air blank and to reduce the effect of the regulating action upon the baseline (traces of m/z 28 and 29) the flow controller of the EA 1110 was switched off. The He carrier gas pressure as well as the O 2 pressure were set to 100 kpa allowing a He flow of approximately 80 ml/min; the O 2 flow for the autosampler inlet purge line was adjusted to a flow between ml/min. Principle of dilution Due to their high C/N ratios, combined 15 N and 13 C measurements of plant and carbon rich soil samples require a strong attenuation of the CO 2 peak. In the ConFlo III interface a small aliquot of CO 2, from the GC system inside the NA 1110 elemental analyser, is diluted with an appropriate amount of pure He gas carefully avoiding isotope fractionation (Fig. 3). Reference gas pulses of varying duration can be introduced under computer control into the system at any time during the chromatogram from a separate open split similar to the design in the irm-gcms interface described earlier. 18

2 1238 CONFLO III AN INTERFACE FOR PRECISION 13 C AND 15 N ANALYSIS Figure 1. Schematic of the experimental setup: Samples are weighed in tin or silver capsules and placed into the autosampler of the CHN combustion unit (EA). The sample carbon is quantitatively combusted to CO 2 at 1020 C in the presence of pure oxygen; NO x are reduced to N 2 on Cu at 650 C. After GC separation (40 C, Porapak PQS) the sample gas peaks enter the isotope ratio mass spectrometer (Finnigan MAT model 252) via the ConFlo III interface. Isotope ratio measurement is made on-line. The ConFlo III The ConFlo II effluent stream divider and the combined reference and sample gas inlet port with the tube-in-tube assembly are substituted by a system with two separate open split inlets. The reference gas inlet is exactly the same as described elsewhere 18,19 for irm-gcms purposes. The schematic of the sample gas inlet together with the dilution arrangement is shown In Fig. 3. The sample inlet contains an active (piston operated) dilution capillary and a fixed sniffer capillary, as well as the fixed sample capillary which is placed upstream from the sniffing capillary. The effluent of the EA is passed through a pneumatic valve (MOVPT valve, SGE, Weiterstadt, Germany). In case of dilution, the pneumatic valve (vent 1) is opened and about 50 ml/min of the EA flow are diverted. Simultaneously the piston (SMC Pneumatic, Leipzig, Germany) with the dilution capillary moves upstream beyond the sniffing point. The laminar gas flow from the EA (typically around 80 ml/min) is split in a fixed relation depending on the valve status. The resulting gas flows in both cases are denoted in Fig. 3. In dilution mode ( dilution on ) only 0.5 ml of the EA flow, in straight mode about 8 ml of the EA flow, reaches the sample inlet port (flow 1). The He dilution flow can be varied manually from 5 to 50 ml/min resulting in a dilution effect between 10 and 100 fold. The 100 mm i.d. capillary to the mass spectrometer allows for a flow between 300 and 400 bar.ml/min into the ion source. During dilution off the dilution capillary is parked outside the glass tube of the open split assembly. Due to the Figure 2. Flow scheme and functionality of the ConFlo III interface. In order to measure 15 N values of samples with the required precision for natural abundance studies a minimum amount of N 2 is needed in the ion source. For the combined measurement of the 15 N and 13 C values of the same sample, the carbon peak of samples with large carbon surplus needs to be attenuated without altering its 13 C/ 12 C ratio. The introduction of reference CO 2 and N 2 is supported. The sample and reference gas peaks are shown as they elute from the GC and ConFlo III interface, respectively. 17

3 CONFLO III AN INTERFACE FOR PRECISION 13 C AND 15 N ANALYSIS 1239 Figure 3. Detailed layout of the gas flows inside the ConFlo III interface. Vent 1 and dilution action are switched simultaneously, resulting in the dilution of a further reduced portion of the EA effluent (Flow 1). The open split assembly has an internal diameter of less than 1 mm. presence of spikes on the m/z traces during valve operation the dilution should not be switched while sample or reference gas peaks are passing into the ion source. Peak shapes Representative ion currents corresponding to N 2 and CO 2 produced by combustion of atropine using the ConFlo III coupling are shown in Figs 4(a) and 4(b). For nitrogen, m/z 28 and 29 traces are shown; for CO 2, the respective traces representing the different isotopes are m/z 44, 45 and 46. The CO 2 peak shown in Fig. 4(b) is diluted 1:16. The slight tailing of the CO 2, more pronounced with increasing dilution when the sample mass is higher, is caused by a progressive overload of the packed GC column (3 m 1/4@ Porapak PQS, CE Instruments) of the elemental analyser. Materials, methods and blank correction The Sn capsules (4 6 mm, part. no ) and the Ag capsules (4 6 mm, part. no ) were purchased from Lüdi AG (Flawil, Switzerland). Samples of atropine (C 17 H 23 NO 3, purchased from CE Instruments) were weighed into the Sn or Ag capsules, loaded into an AS 128 autosampler (CE Instruments) and combusted in the EA using a 10 ml oxygen loop. The pressure in the oxygen feed (O 2 purity %) was kept closely identical to the He pressure (He purity %) in order to minimise surges Figure 4. Chromatogram of a single combustion consisting of two separate parts, (a) for N 2 and (b) the other for CO 2. All isotopic traces are shown. The square peaks at the start and at the end of the chromatograms are reference gas pulses generated from the ConFlo III interface.

4 1240 CONFLO III AN INTERFACE FOR PRECISION 13 C AND 15 N ANALYSIS Figure 5. Schematic of the total peak area (CO 2 peak, trace of m/z 44) resulting from the combustion of atropine corresponding to 1 mgcina mg Sn capsule. The total peak area is 12 masec, the blank area deriving mainly from the C impurities in the capsule is 5 masec. Thus the pure sample area is 7 nasec (1 masec = 10 9 coulomb). that would deteriorate the N 2 baseline. The small sized atropine samples were solved in analytical grade MeOH (Merck, Darmstadt, Germany) and injected manually with a 10 ml syringe (Hamilton, Reno, USA) into the capsules. The solvent was evaporated at 50 C in a drying oven. Routine operation irm-eams uses a sample range >50 mg C. For smaller samples (50 1 mg) the carbon blank contributions (mainly impurities in the capsule material) become increasingly important (Fig. 5). Direct measurements of the size of the blank yielded nmol C in Sn capsules and nmol C in Ag capsules. The measured 13 C values can be described by a two-component mixing model (sample carbon and blank carbon), the blankcorrected values are obtained from Eqn. 1: 13 C Sample ˆ 13 C Total A Total 13 C Blank A Blank =A Sample 1 A is the area of the m/z 44 peak, in Coulombs, representing the measured sample weight (Fig. 5). RESULTS AND DISCUSSION A number of sample s of atropine were weighed into tin or silver capsules with sample weights ranging from about 1.5 mg to 7 mg substance or 1 to 5000 mg carbon. Each sample peak was calibrated individually with a reference Table 1. Blank corrected d 13 C values obtained from repeated analysis of atropine and blank samples. Standard deviation (S.D.) and number of analyses (n) are given Sample amount [mg C] Dilution Mean 13 C [% V PDB ] S.D. n Blank : : : : Figure C[% V PDB ] values of atropine as a function of sample size. Only the lower end of the sample size range is shown. For samples smaller than 20 mc, blank correction is mandatory. Great care must be taken to ensure that the blanks are uniform in size and isotopic composition. gas pulse preceding the sample peak in the case of nitrogen or following the sample peak for CO 2 (see also Figs 4(a) and (b)). Two or more samples within each suite of 32 analyses were acetanilide and caffeine, both laboratory reference materials which were carefully calibrated using (NBS) 22 ( 13 C= 29.74% V PDB ), (USGS) 24 ( 13 C= 15.99% V PDB ) and (IAEA)-N1 ( 15 N= 0.43% Air N2 ). In addition each measurement series also had 3 or 4 blank samples placed randomly into the sample carousel of the autosampler (AS 128, CE Instruments). All analyses were made as combined 15 N and 13 C measurements from each single sample. At the 10 to 5 mgn level 15 N of atropine was: 26.36% Air N2 with a precision of 0.12%. For the measurement of small amounts of N 2, correction due to the blank signal turned out to be crucial. The blank is caused mainly by small leakages and pressureinduced fluctuations of the m/z 28 and 29 baseline. In the case of a high CO 2 peak of the preceeding sample arising in automated sequence measurements the reference N 2 peak is superimposed on a slowly decreasing background of CO ions. These are produced via unimolecular decay of CO 2. in the ion source. Due to the large difference in the m/z 29/28 ratio between N 2 and CO the quantitative correction for the changing CO background contribution to the 29/28 reference peak ratio is rather difficult to ascertain. For the measurement of 15 N values at natural abundance the simultaneously recorded trace of m/z 30 ( 15 N 2,NO ) bears information about the status of the reduction furnace refer to Fig. 1. An incomplete reduction of NO x formed during the combustion gives rise to increased 30/28 ratios. 15 N/ 14 N results are affected due to the missing nitrogen and, more importantly, 13 C values may be wrong due to the presence of N 2 O during elution of the CO 2. As a diagnostic tool we have observed that incomplete reduction of NO x preceeds the breaking through of surplus oxygen when the copper of the reduction furnace is close to exhaustion. Table 1 shows the mean blank corrected 13 C values [% V PDB ] of all analyses made. Carbon blank correction is particularly important for sample amounts <20 mg C. For measurements above 20 mg C, the scatter of the data is close to 0.1%, sometimes even smaller. Below 20 mg C, the variability of the blank progressively dominates the

5 CONFLO III AN INTERFACE FOR PRECISION 13 C AND 15 N ANALYSIS 1241 precision of the data. In our experiments, the blank was isotopically similar to the analyte so that the data still look very reasonable. Figure 6 clearly shows the increase in scattering of the 13 C values below 25 mg C. Below 4 mg C the accuracy of the measured -values critically depends upon the assessment of the blank signal and and its correction. The data at 1 mg C (see also Fig. 5) correspond to about 0.4 nmol C in the ion source with a signal height of about 0.5 V or 1.5 na (m/z 44). REFERENCES 1. T. Preston and N. J. P. Owens, Analyst 108, 971 (1983). 2. T. Preston and N. J. P. Owens, Biomed. Mass Spectrom. 12, 510 (1985). 3. F. Pichlmayr and K. Blochberger, Fresenius Z. Anal. Chem. 331, 196 (1988). 4. A. Giesemann, H.-J. Jäger, A. L. Norman, H. R. Krouse and W. A. Brand, Anal. Chem. 66, 2816 (1994). 5. B. Fry, W. Brand, F. J. Mersch, K. Tholke and R. Garritt, Anal. Chem. 64, 288 (1992). 6. H. Avak, A. Hilkert and R. Pesch, Isotopes Environ. Health Stud. 32, 285 (1996). 7. Micromass Ltd., Technical Note No. TN 311/LA Version 2 (1996). 8. B. Fry, R. Garritt, K. Tholke, C. Neill, R. H. Michener, F. J. Mersch and W. Brand, Rapid Comm. Mass Spectrom. 10, 953 (1996). 9. J. Santrock and J. M. Hayes, Anal. Chem. 59, 119 (1987). 10. R. A. Werner, B. E. Kornexl, A. Roßmann and H.-L. Schmidt, Anal. Chim. Acta 319, 159 (1996). 11. J. Koziet, J. Mass Spectrom. 32, 103 (1997). 12. G. D. Farquhar, B. K. Henry and J. M. Styles, Rapid Comm. Mass Spectrom. 11, 1554 (1997). 13. S. D. Kelly, I. G. Parker, M. Sharman and M. J. Dennis, J. Mass Spectrom. 33, 735 (1998). 14. M. J. Dennis, P. Wilson, S. Kelly and I. Parker, J. Anal. Appl. Pyrolysis 47, 95 (1998). 15. K. Habfast, Advanced isotope ratio mass spectrometry I: Magnetic isotope ratio mass spectrometers In: I. T. Platzner} (Ed.): Modern Isotope Ratio Mass Spectrometry, ISBN Volume 145 in Chemical Analysis, J. D. Winefordner (Series Eds). Wiley, Chichester, pp (1997). 16. W. Brand and K. Habfast German Patent DE C 2 (1993). 17. W. A. Brand, J. Oesselmann and K. Habfast, 15 N/ 13 C and N/C determination on a single sample by coupling an elemental analyser with the new ConFlo interface to a desktop IRMS. In: IAEA (ed.), Nuclear Techniques in Soil-Plant Studies for Sustainable Agriculture and Environmental Preservation, IAEA- SM-334/2, pp (1995). 18. W. A. Brand, J. Mass Spectrom. 31 (1996) D. A. Merrit, W. A. Brand and J. M. Hayes, Org. Geochem. 21, 573 (1994).