High-temperature elemental analysis and pyrolysis techniques for stable isotope analysis

Transcription

1 RAPID COMMUNICATIONS IN MASS SPECTROMETRY Rapid Commun. Mass Spectrom. 2003; 17: Published online in Wiley InterScience ( DOI: /rcm.1076 High-temperature elemental analysis and pyrolysis techniques for stable isotope analysis Matthias Gehre* and Gerhard Strauch UFZ-Centre for Environmental Research Leipzig-Halle, Laboratory of Stable Isotopes, Permoserstrasse 15, Leipzig, Germany Received 3 March 2003; Revised 30 April 2003; Accepted 30 April 2003 A universal method for pyrolysis and elemental analysis, suitable for the online determination of deuterium, carbon, nitrogen and oxygen isotopes for organic and inorganic substances, is presented. The samples are pyrolytically decomposed in a high-temperature pyrolysis (HTP) system, at a temperature exceeding 14008C, in the presence of reactive carbon. The method is suitable for the analysis of stable isotope ratios from hydrogen, carbon, nitrogen and oxygen. The instrumentation and experimental procedure are simple and cost-effective. The reproducibility of the delta values for D/H is better than 3%, and for 18 O, 13 C (organic) and 15 N (inorganic) it is 0.2%. The HTP system is suitable for solid and liquid samples and can use an autosampler for the samples. Results are presented for the isotopic composition of international reference materials and selected laboratory reference materials, which demonstrate the precision and accuracy of the method. Possible problems in the measurement of nitrates and their solutions are particularly discussed. The analyses of oxygen isotopes in selected geological samples (carbonates, silicate, biotite) are demonstrated. Copyright # 2003 John Wiley & Sons, Ltd. For 15 years online techniques for the automated determination of isotope ratios have been developed using an isotope ratio mass spectrometer (IRMS) connected to an elemental analyser system (EA). A carrier gas (He, %) transfers the gaseous products online via an open split to the mass spectrometer. At first the continuous flow (CF) elemental analyser technique was suitable for the measurement of 13 C and 15 N isotopes 1 in organic samples, and some years later for 34 S isotopes 2 in inorganic samples. These systems satisfy the requirements for a higher sample throughput and an automated sample gas preparation mode. The high-temperature pyrolysis system 3,4 complements the elemental analyser techniques with an increased temperature range up to 15008C and the possibility of 18 O and D measurements ( 13 C in organic samples with a C/O ratio 1, and 15 N in inorganic samples, are also possible). Consequently, the HTP system completes the EA technique for bulk analyses of stable isotopes on light elements. In the case of bulk analyses, an accurate sample preparation is necessary because all atoms of one element (i.e. oxygen) in the sample react to form the same sample gas (CO). It is subsequently impossible to distinguish whether, for example, the oxygen in the CO comes from BaCO 3 or from Fe 2 O 3 if the sample has been contaminated with the latter compound. The CF online methods, i.e. EA, HTP, GC systems (combustion, pyrolysis), require several different characteristics of the isotope ratio mass spectrometer (IRMS). Table 1 *Correspondence to: M. Gehre, Centre for Environmental Research Leipzig-Halle, Laboratory of Stable Isotopes, Permoserstrasse 15, Leipzig, Germany. gehre@ana.ufz.de compares some essential technical properties (i.e. system stability, ion source linearity, sensitivity) between the classical dual-inlet method and the CF methods. EXPERIMENTAL A detailed description of the general principle of the HTP technique has already been presented. 3,4 Here we show briefly the designs of the HTP system (Fig. 1) and of the HTP reactor (Fig. 2). The isotopic composition of a sample is usually expressed as: ½% Š ¼ R sample R standard 1000; R standard where R is the ratio of the heavy to light isotope such as 13 C/ 12 C, 18 O/ 16 O, etc. This paper will illustrate four special points for applications using the HTP-EA technique, using as examples determinations of isotopic composition of cellulose, water, nitrates, and some geological samples, i.e. carbonates and silicates. 5 For such measurements on inorganic compounds it is necessary to have a reactive carbon source near the sample, i.e. a Ni/C mixture (10%/90%, approx. 20 mg) in the hottest reactive zone of the reactor (see Fig. 2), or a small amount (approx. 200 mg) of Ni/C mixture placed directly in the silver capsule containing the sample. The required sample size is lower than 10 mmol for C and O and lower than 20 mmol for H and N, depending on the sensitivity of the IRMS and the open split system used. The difference in the required molar Copyright # 2003 John Wiley & Sons, Ltd.

2 1498 M. Gehre and G. Strauch Table 1. Comparison of some significant parameters between the classical dual inlet and the online technique: sample preparation, measurement, and method of calculation Offline methods (dual inlet) Online methods (continuous flow). Sample preparation external (conversion to gases). Sample conversion to gases online. Measurement of pure gases, separation of gases offline. Separation of gases using GC column. Large sample size (mg). Small sample size (mg-range). Only few problems with the homogeneity of the samples. Problems with the homogeneity of the samples are observed. Dual-inlet system pressure adjust sample gas/reference gas. Linearity and stability of the IRMS are important for the quality of measurement. Sample transfer from the bellows directly into the ion source/. Sample transfer by carrier gas (He) via open split waste. Change sample $ standard (min. 6). Only one peak per sample. Statistical mean, calculated from several values (standard sample). Calculation with one peak (integration) and the reference gas from the same sample. Calibration of the system necessary on a monthly/quarterly basis. Calibration of the system on a daily basis and during the run Figure 1. Schematic of the HTP-EA system, ConFlo II split and the IRMS. amounts is determined by the reactions of C and O to CO, and of 2N and 2H to N 2 and H 2. Organic compounds (e.g. cellulose, sugar) can generally be measured at lower temperatures for the pyrolytic conversion to CO. The problem with measurement of H-isotopes from cellulose in a temperature range below 14008C is that the conversion to H 2 gas is probably incomplete, and that methane can be formed from traces of of H and C during the pyrolytic process. Small traces of CH 4 in the ion source disturb the stability of the H þ 3 -factor during the H 2 -measurement. The detected dd values are thus misleading, and the system does not work in a stable fashion. The observed effect is shown in Figs. 3(a) and 3(b). RESULTS AND DISCUSSION Measurement of cellulose The measurement of the isotopic composition of cellulose using the HTP system is very simple. The homogenised samples are well dried, packed in silver capsules (low O-blank) and are completely pyrolysed in the reactor. We can measure up to three different isotopes during one run. The HD-isotopes are measured as molecular X 2 (X ¼ H,D) and the 13 C- and 18 O-isotopes as CO, formed from the cellulose as follows: C 12 H 22 O 11! 11H 2 þ 11 CO þ C # # D½3=2Š 13 C29=28 ½ Š 18 O30=28 ½ Š where *C represents organic carbon derived from the cellulose molecules. The computer-controlled peak jumping technique is very stable and very reproducible for the case of N 2 to CO 2. However, for the measurement of hydrogen, the IRMS does not work with sufficient stability with respect to the magnetic field jumping from HD to CO. In this case the technique has trouble with the temperature control of the magnet even at constant room temperature and consequently with the effect on the stability of the magnetic field, and/or with maintaining a constant temperature in the magnet current control board. The magnet current required for the measurement of

3 Elemental analysis and pyrolysis techniques for isotope analysis 1499 Figure 2. The HTP reactor (schematic). Carrier gas (He) ¼ 40 ml/min; T ¼ 14508C; ceramic tube and glassy carbon tube with no contact between them; reactive carbon layer of Ni/C (10:90, approx. 20 mg). hydrogen isotopes is relatively low; during hydrogen measurements, the temperature of the sense resistor at an amplifier inside the board, used to convert current to voltage for a comparison of the detected and control set-point values of the magnet current, is relatively low in comparison to the temperature during CO measurements. When the magnetic field jumps from CO to HD a false zero value, and thus a faulty jump CO, may consequently occur for the HD magnetic field. In the course of a sequence of cellulose samples the temperature in the electrical circuit may rise, and the beginning of the peak centre for the next sample starts with wrong parameters. Consequently, the jump to CO is not exact enough for an accurate measurement. Temperature control of the magnet current board could be helpful to alleviate this effect. If this effect is observed on any system, the authors recommend two separate measurement runs for HD and CO. Normally, one cannot observe this effect for peak jumping from N 2 to CO 2. The magnet temperature resulting from the magnet current is nearly the same for both masses, and the differences in the resistor temperature in the board caused by the current flow through this resistor are smaller than those caused by jumping from CO to HD. The magnetic field stability is within the range of the peak centre (plateau). Moreover, it has been discovered that different d 18 O values can arise from sample preparation techniques for oxygen isotopes. For the offline nickel and mercury methods more negative values (approx. 1%) are obtained in comparison to those obtained using the HTP technique (M. Saurer, ETH- Zurich, personal communication). Figure 3. GC for pyrolysis of acetanilide detected using a thermal conductivity detector with pyrolysis temperatures of (a) 13808C and (b) 14808C; y-scale ¼ peak intensity. Data courtesy of HEKAtech.

4 1500 M. Gehre and G. Strauch Figure 4. Water sample, 0.25 ml, syringe injected, T ¼ 14508C, He flow ¼ 40 ml/ min. (a) H 2 [HD-] chromatogram and intensity ratio and (b) CO-[ 18 O-] chromatogram and intensity ratio. The dryness of samples prior to the measurement is very important for the accuracy of the dd and d 18 O values. The hydrogen- and oxygen-isotope values from the residual water contaminant are much more depleted in D and 18 O than are those of the cellulose sample. 6 Measurement of water In principle, water can easily be measured for D- and 18 O-isotope values using the HTP system (Figs. 4(a) and 4(b)). In our opinion the standard deviation of 2% for HD-isotopes is too large for the natural water range. For D-enriched experiments the system is suitable and simple. The capacity of water samples for one reactor is nearly unlimited. Some results are shown in Table 2: H 2 O þ C! H 2 þ CO # # D½3=2Š 18 O30=28 ½ Š Here, *C represents inorganic carbon derived from the reactor (layer of Ni/C). Such isotope measurements can be performed with the HTP system even on samples of biological origin (e.g. serum, blood, urine, fruit juice). We recommend use of a normal 10-mL gas-tight syringe with a cone needle tip (HP point style). Table 2. Water analyses, syringe injection, 0.25 ml, n ¼ 5 in each case Sample Intensity 2 [V] dd vs. VSMOW [%] Std.-dev. [%] d 18 O vs. VSMOW [%] Std.-dev. [%] SMOW * * 0.07 SLAP * * 0.29 GISP IAEA-302A IAEA-302B *Regression: y ¼ ax þ b; for dd: a ¼ 1.091, b ¼ ; for d 18 O: a ¼ 1.037, b ¼

5 Elemental analysis and pyrolysis techniques for isotope analysis 1501 Figure 5. Pyrolysis chromatogram and intensity ratio for KNO 3 obtained using the HTP system and a MAT 252 IRMS. Measurement of nitrates It is also possible to measure nitrates using the HTP method. The nitrates are pyrolysed by temperatures exceeding 14008C completely to N 2 and CO: 2KNO 3 þ 8 C! N 2 þ 6 CO þ ½2KŠ # # 15 N29=28 ½ Š 18 O30=28 ½ Š Here, *C represents inorganic carbon derived from the reactor (layer of Ni/C). Some IRMS instruments have difficulty in measuring N 2 ( 15 N-isotopes) and CO ( 18 O-isotopes) in one run, although the GC peak separation of N 2 and CO is complete. The 18 O value from the CO produced is too positive when measured vs. the CO reference gas, even though the same m/z values, (30) for nitrogen and for carbon monoxide, are used. Peak jumping is only necessary in order to change the reference gases (N 2 and CO) and the calculation software of the instrument for the 15 N-isotope (vs. air) and 18 O-isotope values (vs. VSMOW; Fig. 5). Unfortunately, in some ion sources, NO (mass 14 þ 16 ¼ 30) is formed by reaction of nitrogen and oxygen (oxygen is always available in the source system, from water or small leaks) on the hot surface of the filament. This background of NO disappears very slowly in comparison to the background of nitrogen (m/z 28, 29) and influences the CO ratio (m/z 30/28) of the sample. The result is a falsely calculated 18 O value (3% too positive vs. VSMOW). One way to reduce this effect is shown in Fig. 5. It is important to insert one CO reference gas pulse before the CO peak of the sample arrives from the gas chromatograph. Using the data thus obtained together with that from the second CO reference gas peak introduced after the sample, a variable background correction is performed. This procedure can reduce the effect from the NO background to around 1% for the d 18 O values, depending on the N 2 /CO separation in the GC column. Users of the ConFlo III interface can additionally turn on the He dilution during the nitrogen peak. This effectively reduces the amount of nitrogen in the ion source and does not influence the system parameters (fractionation effects in the reference gases during dilution mode, ConFlo II). Another possibility is the separate measurement of N 2 and CO from the nitrates, with blankout of the nitrogen peak by measuring the d 18 O values using the dilution technique and CO reference during the nitrogen transfer into the ion source. Measurement of some geological samples: carbonates and silicates Some geological samples can also be measured using HTP. Applications to phosphates and sulphates have been described in detail previously. 3,4 The online analyses of carbonates and silicates are of interest for some geological applications. Both minerals are measurable using offline sample preparation techniques, i.e. the phosphoric acid method for carbonates and fluorination techniques for the silicates. The carbonates can be measured by using the normal HTP system. The observed oxygen yield is only approximately 67%. The d 18 O values are within 1% of the accepted values with a standard deviation of 0.5%. The recommended method to obtain a better sample (oxygen) conversion of the carbonates to carbon monoxide is to add a catalytic compound; 5 AgCl is suitable as an additive for carbonates. When using such additives some additional precautions are necessary, as the gaseous halogen compounds produced must be trapped before the GC column; gases containing halogen damage the column. It is possible to fix these gases with a chemical trap for acidic gases (i.e. ascarite) or with a copper furnace heated to 6008C. CaCO 3 þ 2 C þ AgCl!! 3 CO þ ½CaŠ þ ½?Cl 2 Š # 18 O½30=28Š Here, *C represents inorganic carbon derived from the reactor (layer of Ni/C), and the gases denoted as [?Cl 2 ]** must be fixed with an ascarite trap or Cu furnace. Some results of the effectiveness of the pyrolysis (oxygen yield) using AgCl as an additive (approximately 300 mg per sample, placed directly into the silver capsule with the sample) are shown in Table 3 and the isotope results are shown in Table 4. Another question concerns the measurement of silicates using HTP. Using the unmodified system we detected less

6 1502 M. Gehre and G. Strauch Table 3. Pyrolysis efficiency for oxygen yield, with and without additives, for a range of samples (T ¼ 14508C) Sample material Formula No additives O yield [%] With additives Sucrose C 12 H 22 O Cellulose 100 Organics 100 Barium sulphate BaSO Potassium nitrate KNO Silver nitrate AgNO Phosphates Barium carbonate BaCO 3 Calcium carbonate CaCO * Ferric carbonate FeCO * Ferric oxide Fe 2 O * Quartz sand SiO 2 <5 >80** Zeolite Al 2 O 3 <5 >70** Manganese oxide MnO 30 90** Additives: * 0.3 mg AgCl þ0.3 mg Ni/C; T ¼ 15008C. **0.3 mg KF or 0.3 mg PTFE þ0.3 mg Ni/C per sample, T ¼ 15008C. Note: Installation of a trap (i.e. ascarite) for fixing fluorinated gases (i.e. HF) is essential to avoid serious damage to the GC column. than 20% oxygen yield converted to CO. The solution to this problem was again to use an additive. Non-hygroscopic fluorinated compounds (e.g. KF, other salts), or carbonfluorine compounds (PTFE) in powder form, are used for silicates and permit an oxygen yield of more than 80%. The trapping of fluorinated gases in the He carrier is very important, as mere traces totally damage the GC column and can cause damage to surfaces in the ion source. The application of fluorine-bearing additives is also very problematic for all parts made of quartz, including fused-silica capillaries. Some results obtained using KF as an additive (approximately 300 mg, placed directly into the silver capsule with the sample) are shown in Table 5. SiO 2 þ 2 C þ KF!! 2 CO þ ½Si þ KŠ þ ½?F 2 Š # 18 O½30=28Š Here, *C represents inorganic carbon derived from the reactor (layer of Ni/C), and the fluorine-containing gases denoted as [?F 2 ]** must be fixed with an ascarite trap or a Cu furnace. CONCLUSIONS The high-temperature pyrolysis (HTP) technique is suitable for isotope measurements on a large range of liquid and solid, organic and inorganic, compounds. The HTP technique is usable as a multi-isotope preparation method and, in combination with modern isotope ratio mass spectrometers, is useful for measurements of deuterium, oxygen, carbon (organic, see above) and nitrogen (inorganic, see above) isotopes. Possible recombination of N and O within the ion source can influence the d 18 O values. Recommendations have been presented on how to deal with problems relevant to specific applications. Table 4. Pyrolysis results for carbonate samples (T ¼ 15008C) HTP-carbonate test (IAEA reference materials), T ¼ 15008C, d 18 O, measured vs. reference gas (calibrated against VSMOW) Weight Ampl. Area CO O-yield Mean Calib Calc þ *Accept. Difference [mg] [V] [Vs] ** [VSMOW] [VSMOW] [VPBD] [VPDB] [28] [28] (28) (%) (%) (%) (%) (%) NBS 19 (CaCO 3 ), no additive (detected O-yield 67%) Mean Std.-dev NBS 19 (CaCO 3 ), additive: 500 mg AgCl and 300 mg Ni/C Mean Std.-dev CO-9 (BaCO 3 ), additive: 500 mg AgCl and 300 mg Ni/C Mean Std.-dev LSVEC (Li 2 CO 3 ), additive: 500 mg AgCl and 300 mg Ni/C Mean Std.-dev þ d 18 O sa/vpdb ¼ *d 18 O sa /VSMOW *VPDB ¼ accepted reference value 8. **O-yield in % (in comparison to cellulose and barium sulfate, small discrepancies due to balance procedures).

7 Table 5. Pyrolysis results for silicate and biotite samples (T ¼ 15008C) HTP-test geological samples (IAEA reference materials), T ¼ 15008C, d 18 O Weight Ampl. Area CO Mean Calib IAEA-value* Difference [28] [28] [VSMOW] [VSMOW] [VSMOW] (mg) (V) (Vs) (%) (%) (%) (%) NBS 28 (SiO 2 ), additive: 300 mg KF and 300 mg Ni/C Mean Std.-dev NBS 27 (biotite)**, additive: 300 mg KF and 300 mg Ni/C Mean Std.-dev *Accepted reference value (various offline methods). 8 **Biotite milled to 30 mesh (originally 1 2 mm diameter). Elemental analysis and pyrolysis techniques for isotope analysis 1503 The results presented for carbonates and silicates show that the HTP technique is also applicable, with the use of additives as catalysts, for geological applications. An improved system is currently in preparation for particular application to geological samples. The HTP technique is also well suited to the analysis of samples for paleoclimatological, hydrological and soil sciences. Acknowledgements We would like to thank Klaus Hecker (HEKAtech GmbH, Wegberg, Germany) for the construction of the HTP system and his outstanding technical support. We thank Manfred Groening (IAEA, Vienna) for his support by organising International Reference Materials. We would like to thank Detlef Schulz-Bull (IOW Rostock-Warnemünde, Germany), Michael Böttcher (MPI Bremen, Germany), and Reiner Höfling (UFZ) for helpful comments and discussions during the experiments. Carol Kendall is thanked for her valuable support in drafting the manuscript. Last but not least, we thank the unknown reviewers for very helpful comments. REFERENCES 1. Pichlmayer F, Blochberger K. Fresenius Z. Anal. Chem. 1988; 331: Giesemann A, Jäger HJ, Normann AL, Krouse HR, Brand WA. Anal. Chem. 1994; 66: Kornexl BE, Gehre M, Hoefling R, Werner RA. Rapid. Commun. Mass Spectrom. 1999; 13: Gehre M. IAEA, TecDoc-1247, IAEA: Vienna, 2001; Gehre M. Verfahren zur massenspektrometrischen on-line Bestimmung von Sauerstoffisotopenzusammensetzungen in geologischen Proben. Deutsches Patentamt, Aktenzeichen: DE A1, Saurer M, Robertson I, Siegwolf R, Leuenberger M. Anal. Chem. 1998; 70: Coplen TB. Chem. Geol. 1988; 72: IAEA TecDoc-825. Reference and intercomparison materials for stable isotopes of light elements, IAEA: Vienna, 1995.