Temperature-Programmed High-Performance Liquid Chromatography

Anal. Chem. 2008, 80, 7144–7152

Technical Notes Temperature-Programmed High-Performance Liquid Chromatography Coupled to Isotope Ratio Mass Spectrometry Jean-Philippe Godin,*,† Ge´rard Hopfgartner,‡ and Laurent Fay† Nestle Research Center, Nestec Ltd., Department of Bioanalytical Science, P.O. Box 44, CH-1000 Lausanne 26, Switzerland, and Life Sciences Mass Spectrometry, School of Pharmaceutical Sciences, University of Geneva, University of Lausanne, 20 bd d’Yvoy, CH-1211 Geneva 4, Switzerland The utility of liquid chromatography coupled to the isotope ratio mass spectrometry technique (LC-IRMS) has already been established through a variety of successful applications. However, the analytical constraint related to the use of aqueous mobile phases limits the LC separation mechanism. We report here a new strategy for high-precision 13C isotopic analyses based on temperature-programmed LC-IRMS using aqueous mobile phases. Under these conditions, the isotopic precision and accuracy were studied. On one hand, experiments were carried out with phenolic acids using isothermal LC conditions at high temperature (170 °C); on the other hand, several experiments were performed by ramping the temperature, as conventionally used in a gas chromatography-based method with hydrosoluble fatty acids and pulses of CO2 reference gas. In isothermal conditions at 170 °C, despite the increase of the CO2 background, p-coumaric acid and its glucuronide conjugate gave reliable isotopic ratios compared to flow injection analysis-isotopic ratio mass spectrometry (FIA-IRMS) analyses (isotopic precision and accuracy are lower than 0.3‰). On the opposite, for its sulfate conjugate, the isotopic accuracy is affected by its coelution with p-coumaric acid. Not surprisingly, this study also demonstrates that at high temperature (170 °C), a compound eluting with long residence time (i.e., ferulic acid) is degraded, affecting thus the δ13C (drift of 3‰) and the peak area (compared to FIA-IRMS analysis at room temperature). Quantitation is also reported in isothermal conditions for p-coumaric acid in the range of 10-400 ng/mL and with benzoic acid as an internal standard. For temperature gradient LCIRMS, in the area of the LC gradient (set up at 20 °C/ min), the drift of the background observed produces a nonlinearity of SD (δ13C) ∼0.01 ‰/mV. To circumvent this drift, which impacts severely the precision and * Author to whom correspondence should be addressed. Phone +41 21 785 94 43. Fax: +41 21 785 94 86. E-mail: [email protected]. † Nestec Ltd. ‡ University of Geneva, University of Lausanne.

7144

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

accuracy, an alternative approach, i.e., eluting the compound on the plateau of temperature studied was reported here. Other experiments with temperatureprogrammed LC-IRMS experiments are also reported with the presence of methanol in the injected solution to mimic residual solvent originating from the sample preparation or to slightly increase the solubility of the targeted compound for high-precision measurement. High-precision compound specific measurements of 13C/12C ratios are invaluable tools in the fields of biogeochemistry, ecology, forensics, and the biological sciences.1,2 Over the last 20 years, gas chromatography coupled to isotope ratio mass spectrometry (GC-IRMS)3,4 has been widely applied to the abovementioned fields. However, most biological molecules are nonvolatile and thus require derivatization steps prior to analysis. This physical-chemical transformation introduces a potential source of imprecision in addition to several extra carbons and longer sample preparation times.5 Although the idea of combining liquid chromatography (LC) to isotope ratio mass spectrometry (IRMS) was proposed over 10 years ago, the interface allowing the LC-IRMS coupling was only commercialized in 2004.6 The LCIRMS coupling is based on a wet chemical oxidation of organic molecules into CO2 gas. The main constraint of the LC method development for an LC-IRMS run (based on the commercialized LC Isolink interface) is the imperative absence of organic buffers in the mobile phase entering the interface.7 This limitation makes of the most widespread LC stationary mechanism, the reversed phase LC, a phase which is not suitable for the LC-IRMS technique. Several reports show that using specific and restrictive chromatography modes of elution, we can overcome all these Lichtfouse, E. Rapid Commun. Mass Spectrom. 2000, 14, 1337–44. Brand, W. A. Rapid Commun. Mass Spectrom. 1996, 31, 225–35. Meier-Augenstein, W. Anal. Chim. Acta 2002, 465, 63–79. Sessions, A. L. J. Sep. Sci 2006, 29, 1946–61. Docherty, G.; Jones, V.; Evershed, R. P. Rapid Commun. Mass Spectrom. 2001, 15, 730–8. (6) Krummen, M.; Hilkert, A. W.; Juchelka, D.; Duhr, A.; Schluter, H. J.; Pesch, R. Rapid Commun. Mass Spectrom. 2004, 18, 2260–6. (7) Godin, J. P.; Hopfgartner, G.; Fay, L.-B. Mass Spectrom. Rev. 2007, 26, 751–74. (1) (2) (3) (4) (5)

10.1021/ac8004204 CCC: $40.75  2008 American Chemical Society Published on Web 08/09/2008

constraints for various classes of compounds (e.g., as amino acids,8,9 small peptide,10 sugars,11 or volatile organic acids12). These modes of elution are based on ion exchange or mixed mode interactions with gradients of inorganic salts such as phosphate buffers, diluted sulfuric acid, or water. While the coupling of LC to IRMS offers considerable opportunities for isotopic measurements, accurate determination of 13C/12C ratio requires that the targeted compounds be baseline resolved.7 Therefore, identifying alternative LC separation mechanisms compatible with LC Isolink is an interesting concept to pursue. The concept of high temperature or temperature-programmed LC is not new,13 and this approach is seen as a powerful optimization variable for various LC separations14–18 with many advantages.19–23 The isothermal mode of elution gives a powerful tool to improve the retention factor, peak width, and peak height as well as the chromatographic selectivity in a LC separation.24 The interest in this direction is also driven by the commercialization of LC columns capable of resisting to temperatures higher than 120 °C19 and instruments allowing heating of the LC columns and the mobile phases up to 200 °C.25 Currently, HTLC is hyphenated with UV, MS, NMR and ICPMS devices.26 Last but not least, when using high temperature conditions associated with a 100% aqueous mobile phase, uncommon hyphenation is also reported with an FID detector.27,28 The HTLC-FID coupling presents interesting similarities to the LC-IRMS since both devices are highly sensitive to the organic eluent and to the LC flow rate which in LC-IRMS is limited to 700 µL/min.6 A great concern with temperature-programmed LC-IRMS is not the physical coupling itself but rather the aspect relative to the isotopic 13C/12C ratio measurement and the effect on the isotopic precision and accuracy. Typically, with LC-IRMS the isotopic precision of δ13C values is around 0.1-0.3‰,6 and (8) McCullagh, J. S.; Juchelka, D.; Hedges, R. E. Rapid Commun. Mass Spectrom. 2006, 20, 2761–8. (9) Godin, J. P.; Hau, J.; Fay, L. B.; Hopfgartner, G. Rapid Commun. Mass Spectrom. 2005, 19, 2689–98. (10) Schierbeek, H.; te Braake, F.; Godin, J. P.; Fay, L. B.; van Goudoever, J. B. Rapid Commun. Mass Spectrom. 2007, 21, 2805–12. (11) Cabanero, A. I.; Recio, J. L.; Ruperez, M. J. Agric. Food Chem. 2006, 54, 9719–27. (12) Heuer, V.; Elvert, M.; Tille, S.; Krummen, M.; Mollar, X. P.; Hmelo, L. R.; Hinrichs, K. U. Limnol. Oceanogr. 2006, 4, 346–57. (13) Miller, D. J.; Hawthorne, S. B. Anal. Chem. 1997, 69, 623–7. (14) de Boer, A. R.; Alcaide-Hidalgo, J. M.; Krabbe, J. G.; Kolkman, J.; Emde Boas, C. N.; Niessen, W. M.; Lingeman, H.; Irth, H. Anal. Chem. 2005, 77, 7894–900. (15) Plumb, R. S.; Rainville, P.; Smith, B. W.; Johnson, K. A.; Castro-Perez, J.; Wilson, I. D.; Nicholson, J. K. Anal. Chem. 2006, 78, 7278–83. (16) Riddle, L. A.; Guiochon, G. J. Chromatogr., A 2006, 1137, 173–9. (17) Vanhoenacker, G.; Sandra, P. J. Chromatogr., A 2005, 1082, 193–202. (18) Teutenberg, T.; Lerch, O.; Gotze, H. J.; Zinn, P. Anal. Chem. 2001, 73, 3896–9. (19) Vanhoenacker, G.; Sandra, P. J. Sep. Sci. 2006, 29, 1822–35. (20) Guillarme, D.; Heinisch, S.; Rocca, J. L. J. Chromatogr., A 2004, 1052, 39–51. (21) Dolan, J. W. J. Chromatogr., A 2002, 965, 195–205. (22) Smith, R. M. Anal. Bioanal. Chem. 2006, 385, 419–21. (23) Heinisch, S.; Puy, G.; Barrioulet, M. P.; Rocca, J. L. J. Chromatogr., A 2006, 1118, 234–43. (24) Dolan, J. W. J. Chromatogr., A 2002, 965, 195–205. (25) McNeff, C. V.; Yan, B.; Stoll, D. R.; Henry, R. A. J. Sep. Sci. 2007, 30, 1672–85. (26) Guillarme, D.; Heinisch, S. Sep. Purif. Rev. 2005, 34, 181–216. (27) Miller, D. J.; Hawthorme, S. B. Anal. Chem. 1997, 69, 623–27. (28) Guillarme, D.; Heinisch, S.; Gauvrit, J. Y.; Lanteri, P.; Rocca, J. L. J. Chromatogr., A 2005, 1078, 22–7.

the isotopic accuracy is less affected by errors compared to measurements carried out by GC-C-IRMS.29 A secondary concern associated with temperature-programmed LC-IRMS is the quantitative aspect. We report here the quantitation of the p-coumaric acid at 160 °C with an internal standard. Additionally, we investigate the effect of residual presence of methanol in the injected buffer solution on the isotopic ratio measurements using standard CO2 pulses. Here, we hypothesize that methanol would come from incomplete solvent removal or from the addition of an organic modifier to slightly increase the solubility. In order to improve the attractiveness of the LC separation for high-precision 13C isotopic analyses, this novel temperatureprogrammed LC-IRMS approach is illustrated with underivatized phenolic acids (p-coumaric acid and its sulfate and glucuronide conjugates and ferulic acid) and hydrosoluble fatty acids in isothermal and temperature-programmed conditions, respectively. Previous works have already highlighted the isotopic analyses of analogous compounds to give information of dietary bioavailability30 or in ecology to measure anaerobic metabolism in marine sediments.12 EXPERIMENTAL SECTION Materials and Reagents. Phosphoric acid, sodium peroxodisulfate (p.A.), potassium hydrogen phosphate (p.A.) were purchased from Merck (Darmstadt, Germany). Deionized water was prepared using a Milli-Q system (with a resistance level equal to 18.2 MΩ cm). Fatty acids and p-coumaric acid were purchased from Sigma (St. Louis, MO, USA). The sulfate and glucuronide conjugate of p-coumaric acid were chemically synthesized in house. Each solution was prepared by dissolving the compounds in the aqueous eluent and finally passed by ultrason to obtain a complete dissolution. Phosphate buffers were prepared with Milli-Q grade water. To decrease the ubiquitous CO2+ background signal (m/z 44, measured with a preamplifier feedback resistor of 300 MΩ U), all reagents and LC eluents were thoroughly degassed in an ultrasonic bath (Digitana AG, Switzerland) at 30 ± 5 °C for 15 ± 5 min (using a water vacuum). The concentration of the acidic reagent was approximately 1.0 mol L-1 phosphoric acid in water, and the concentration of sodium peroxodisulfate was approximately 0.5 mol L-1 in water. Instrumentation. Experiments were carried out on a MAT252 IRMS (Finnigan MAT, Bremen, Germany). The IRMS operated at an accelerating voltage of 8 kV. The ion source was held at a pressure of 1 × 10-6 Torr, and ions were generated by electron impact at 70 eV. Three faraday cup detectors monitored simultaneously and continuously the CO2+ signals for the three major ions at m/z 44 (12CO2), m/z 45 (13CO2 and 12C17O16O) and m/z 46 (12C18O16O). The liquid chromatographic eluents were delivered with a Rheos 2000 (Flux, Basel, Switzerland) liquid chromatography pump. The injection was performed with a CTC Analytics autosampler (Zwingen, Switzerland) equipped with a 10 µL loop. (29) Godin, J. P.; Breuille, D.; Obled, C.; Papet, I.; Schierbeek, H.; Hopfgartner, G.; Fay, L. B. J. Mass Spectrom. [Online early access]. DOI: 10.1002/ jms.1406. Published Online: March 27, 2008. (30) Setchell, K. D.; Faughnan, M. S.; Avades, T.; Zimmer-Nechemias, L.; Brown, N. M.; Wolfe, B. E.; Brashear, W. T.; Desai, P.; Oldfield, M. F.; Botting, N. P.; Cassidy, A. Am. J. Clin. Nutr. 2003, 77, 411–19.


7145

Figure 1. Schematic representation of the temperature-programmed liquid chromatography hyphenated with an isotope ratio mass spectrometer. (1) The helium and CO2 tanks used to degas the eluent buffer, and as the reference gas, respectively, (2) the LC pumps and the degasser, (3) the injection valve with a 10 µL loop, (4) the autosampler, (5) the 0.2 µm online filters, (6) the Polaratherm 9000 heater, (7) a plexiglass window, (8) the oxidant and acid reagents used in the LC-IRMS interface, (9) the LC Isolink interface, (10) the magnetic sector instrument, and (11) the data treatment.

The pump was connected to the heating module via a switching valve. One online filter (0.2 µm, Vicci, Schmidlin Labor, Switzerland) was placed between the pump and a Polaratherm 9000 series (RIC, Kortrijk, Belgium) and was connected to an LC Isolink interface (Thermo Electron, Bremen, Germany). At the entrance of the LC-IRMS interface, a second online filter (0.2 µm) was placed. Finally, the LC Isolink interface was coupled to the MAT252. Within the Polaratherm 9000, the LC mobile phase was heated at the same temperature as the column and the LC effluent (after the column) was set at 30 °C. Inside the LC-IRMS interface, the temperature of the reactor was set at 99.9 °C. The helium (99.9998%) flow rate of the separation unit was set at 1 mL/min. The LC Isolink flow rates were set at 50 µL/min for both acid and oxidant reagents (respectively, phosphoric acid and sodium peroxodisulfate). In these conditions, flow injection isotopic analysis (FIA-IRMS), i.e., injection without any LC column, and LC-IRMS analyses with LC columns were performed. The scheme of the installation is described in Figure 1. Chromatographic Conditions for Temperature-Programmed LC-IRMS. Two different LC columns were used. The first was a polymeric column packed with a sulfonated polystyrenedivinylbenzene (Coregel 87H3 from the Laubscher laboratory, Miecourt, Switzerland). The column (300 mm × 7.8 mm) is stable up to 90 °C and at 0 e pH e14. The second column used for the LC separation at higher temperatures was a porous graphitized carbon (PGC) column (Hypercarb, 100 mm × 2.1 mm, 5 µm from Thermo Fisher, Germany). The Hypercarb column is particularly resistant to high temperature (up to 200 °C) and to 1 e pH e14.31 For the LC separation of hydrosoluble fatty acids, the isotopic analyses were carried out with the Coregel 87H3 and Hypercarb columns. In isothermal mode, the Coregel worked at 40 °C, with 0.008 N sulfuric acid as the eluent and with a flow rate of 500 µL/min. With temperature-programmed mode, the Hypercarb column was used with phosphoric acid (20 mM) in isocratic mode (31) West, C.; Lesellier, E.; Tchapla, A. J. Chromatogr. A 2004, 1048, 99–109.

7146


(LC flow rate at 220 µL/min). The gradient of temperature started at 35 °C, held for 5 min, followed by a ramp of temperature of 20 °C/min until 110 °C held for 1 or 5 min. For LC separation of phenolic acids, isotopic analyses were carried out with the Hypercarb column in isothermal conditions (between 150 and 170 °C) with phosphate buffer (20 mM, pH 7.2) in isocratic mode at 500 µL/min. Injection of standard CO2 pulses (20 s of peak width at the baseline) every 30 s was also performed using the LC-IRMS interface throughout different LC runs. This allows one to mimic the elution of compounds and to study the effect of the LC conditions on isotopic precision and accuracy. Isotopic Measurements and Calculation. Data were acquired with Isodat NT software. The 13C/12C isotopic ratio is expressed as a δ13C value calibrated against the international standard (Vienna Pee Dee Belemnite, VPDB). The δ notation which measures the relative isotopic concentration is defined as δ13Csample, ‰ ) [(Rs⁄Rstd) - 1] × 1000

(1)

where Rs is the ratio of 13C in the sample and Rstd is the ratio of (13C/12C) of International Standard of Pee Dee Belemnite, Rstd ) 0.011 237 2. All the δ13C values are reported relative to the reference CO2 of a known carbon isotopic composition, introduced directly into the ion source in three pulses at the beginning of each run. The CO2 reference gas was calibrated against VPDB international standard (IAEA-CH6 through the LC Isolink interface). For the integration parameters, the slope (mV/s) used was set at 0.1 s with an individual background algorithm (with a window of 5 s). For the temperature gradient, the dynamic background was used, with a step width at 75 points.32 (32) Ricci, M. P.; Merritt, D. A.; Freeman, K. H.; Hayes, J. M. Org. Geochem. 1994, 21, 561–71.

Figure 2. Temperature-programmed LC-IRMS chromatogram of phenolic acids eluted with the Hypercarb column in isothermal conditions at 170 °C; flow rate 500 µL/min; phosphate buffer 20 mM, pH 7.2. The compounds are (1) sulfate conjugate of p-coumaric acid; (2) p-coumaric acid; (3) glucuronide conjugate of p-coumaric acid; (4) ferulic acid. The amount of carbon injected was 39, 280, 101, and 269 ng, respectively.

Table 1. 13C Isotopic Ratios of Phenolic Acids Carried out with FIA-IRMS at Room Temperature and with LC-IRMS in Isothermal Conditions at 170 °Ca

compounds

FIA-IRMS δ13C ± SD, (‰) (n ) 5)

sulfate p-coumaric acid p-coumaric acid glucuronide p-coumaric acid ferulic acid

-30.95 ± 0.71 -28.80 ± 0.14 -22.87 ± 0.20 -26.23 ± 0.13

HTLC-IRMS (170 °C) δ13C ± SD, (‰) (n ) 5)

HTLC-FIA ∆13C, (‰)

-32.04 ± 1.60 -28.88± 0.47 -22.70 ± 0.28 -22.79 ± 1.48

1.09b 0.08 0.17 3.44

a The typical LC separation is reported in Figure 2. Individual phenolic acid was also analyzed by flow injection mode-IRMS using the same eluent and flow rate at room temperature. ∆ (‰) represents the difference of δ13C values measured with the Hypercarb and the FIA-IRMS analyses at room temperature. In this condition, ferulic acid is degraded due to its long residence time at high temperature. b The isotopic values affected by the low chromatographic resolution obtained with the adjacent compound.

RESULTS AND DISCUSSION In order to use the temperature-programmed LC with the IRMS device, the choice of the size, i.e., internal diameter (i.d.) of the column, is key. Typically, with the LC-IRMS, the LC flow rate is limited to 700 µL/min due to the geometry of the commercialized interface. However with temperature-programmed liquid chromatography hyphenated to the UV or MS devices, based on the van Deemter curve, LC separation is generally associated with a high LC flow rate for narrow bore column if compared to separations carried out at room temperature.33,28 With a 2.1 mm i.d. column, the optimum flow rate is ∼145 µL/ min and reaches 600 and 1500 µL/min at 100 °C and 200 °C, respectively. In the work reported here, the chromatographic efficiency is compromised by working with nonoptimal velocity (according to the van Deemter curve) for the LC eluent to respect the constraints of the LC-IRMS interface. These circumstances might be improved by the miniaturization of the LC column using a 1 mm i.d. column. Thus, at 200 °C, the optimum LC flow rate is ∼340 µL/min, which matches the requirement of the LC-IRMS interface. However, the loading capacity is 4 times less with a 1 (33) Lestremau, F.; de Villiers, A.; Lynen, F.; Cooper, A.; Szucs, R.; Sandra, P. J. Chromatogr., A 2007, 1138, 120–31.

mm i.d column than with a 2 mm i.d column.34 This is a problem as the IRMS is sensitive to the amount of material injected (the required amount is 1 nmol),2 and when increasing the injected volume, the peak shape would also detoriates. Isothermal Mode of LC Separation in LC-IRMS. To demonstrate the utility of temperature-programmed LC-IRMS with aqueous mobile phase, phenolic acids (p-coumaric acid, its glucuronide, its sulfate metabolites, and ferulic acid) were injected at 170 °C and at room temperature by flow injection-IRMS (individually injected). With the use of the Hypercarb column in isothermal conditions with a phosphate buffer, a baseline separation was achieved for the p-coumaric acid, its glucuronide conjugate and ferulic acid. In this condition, the sulfate conjugate of p-coumaric acid was coeluted with p-coumaric acid (as illustrated in Figure 2). The comparison of δ13C values measured by flow injection-IRMS and by temperature-programmed LC-IRMS at 170 °C (see Table 1) showed excellent isotopic precision, SD(δ13C), lower than 0.3‰. As already reported in the literature, with unequal abundance of two adjacent peaks (here, 1 versus 7, for p-coumaric acid versus its sulfate), the results for the smaller peak produce substantially greater degradation of both precision (34) Abian, J. J. Mass Spectrom. 1999, 34, 157–68.


7147

Table 2. 13C Isotopic Ratios of Hydrosoluble Fatty Acids at Natural Abundance Measured by LC-IRMS in Isothermal Conditions at 40 °C and with Temperature Gradienta

citric acid tartaric acid succinic acid propionic acid butyric acid

isothermal elution (with Coregel at 40 °C) δ13C ± SD, (‰) (n ) 5)

gradient of temperature (with Hypercarb) δ13C ± SD, (‰) (n ) 3)

∆13C (isothermal - gradient) ∆13C, (‰)

-18.32 ± 0.39 -21.77 ± 0.67 -22.40 ± 0.11 -29.27 ± 0.24 -22.57 ± 0.17

-16.22 ± 1.53 -21.49 ± 0.06 -21.56 ± 0.74 -28.46 ± 0.32 -21.95 ± 0.36

2.10b 0.28 0.84b 0.81c 0.62c

a See text for the conditions. b Represents the isotopic values that are affected by the temperature gradient leading to ∆ ‰ (accuracy) higher than 0.8 ‰. c The isotopic values affected by the low chromatographic resolution obtained with adjacent compounds.

and accuracy than the larger peak.35 Thus, good isotopic accuracy (lower than 0.3‰) was obtained for p-coumaric acid and its glucuronide conjugate. However, for the sulfate metabolite of p-coumaric acid, a degraded isotopic accuracy (with a difference of ∼1.0‰) and precision, SD(δ13C) at 1.6‰, were observed. For ferulic acid, the δ13C values were not consistent between the FIAIRMS and LC-IRMS analyses showing thus a large shift in the isotopic ratio (+3‰). Moreover, the peak area analysis follows the same trend. To express this, the peak area of ferulic acid is normalized versus the amount of carbon injected (in V s ng-1 C) and also versus the sensitivity of the benzoic acid injected in the same conditions by FIA-IRMS to obtain a relative sensitivity (the sensitivity of benzoic acid here was 0.16 V s ng-1 C). Thus, the relative sensitivity of ferulic acid was 0.30 ± 0.1 (n ) 8) at 170 °C versus 1.0 ± 0.1 at room temperature. Comparatively, p-coumaric acid and its glucuronide metabolite had a relative sensitivity of 0.8 and 0.9 at 170 °C. These results showed that the oxidation efficiency was not significantly different when the temperature was increased and that at high temperatures degradation could occur according to the residence time of the compound in the LC run. This degradation seriously deteriorates the δ13C values and the peak area. A concern related to the temperature programmed in LCIRMS was the increase of the CO2 background. In isothermal conditions, generally whatever the temperature used, the CO2 background becomes stable after variable times depending on the lifetime of the column. The CO2 background varies with the temperature (using the same LC column, buffer, and flow rates). For illustration, in isothermal conditions, the variation of the CO2 background (∆ m/z 44, in mV) was around 80 and 200 mV, in the ranges 150-160 °C and 150-170 °C, respectively. For all isothermal conditions, the observed isotopic precision for the pulses reference gas, injected in the beginning of each run, was very close to SD(δ13C) at 0.045-0.06‰ (duplicate run with three pulses in each run). Results reported by Merrit and Hayes36 with a GC-C-IRMS also show that the isotopic precision was only slightly affected by the modification of the CO2 background: a ∆m/z 44 of the CO2 background around 200 mV did not affect the isotopic precision. Our results confirm the results reported above: if the CO2 drift is linear, there is no significant change in δ13C measurement. Effect of Temperature Gradient in LC-IRMS. Traditionally, the GC-C-IRMS separation is carried out with a temperature ramp (35) Goodman, K. J.; Brenna, J. T. Anal. Chem. 1994, 66, 1294–301. (36) Merritt, D. A.; Hayes, J. M. Anal. Chem. 1994, 66, 2336–47.

7148


to elute compounds with different volatility. In GC programmed temperature-IRMS, it is known that both the temperature gradient and the carrier gas flow influence the isotopic ratio.37 Here, in LC-IRMS, the temperature ramp is used to take advantage of the modification of the elution strength of the aqueous mobile phase. As pointed out by Teutenberg,38 the column bleeding (amplitude of the drift of the background) with high temperature LC conditions varies according to the stationary phases, the temperature, and the instruments used for measuring it (UV or charged aerosol detector). However, the drift of the background can be easily overcome in LC-UV by substracting the background with an appropriate LC software. As reported previously for phenolic acids, the chromatographic conditions are essential parameters for measuring isotopic ratios. To illustrate the effect of a temperature gradient (below 120 °C) on isotopic measurements, temperature-programmed LC-IRMS experiments were carried out on hydrosoluble fatty acids. To assess the isotopic accuracy, the compounds were also injected on the Coregel column in isothermal conditions at room temperature (see Table 2). With the Coregel column, all the fatty acids were baseline separated. Here, with the temperature gradient conditions tested (see Figure 3), formic and acetic acids were coeluted, while propionic, tartaric, butyric, succinic, and citric acids were separated. The accuracy of propionic acid was affected by the poor resolution of the double peak eluting before (corresponding to formic and acetic acids). Improvement of the accuracy is possible by changing the chromatography conditions or by applying other algorithms as described by Goodman.35 For the remaining compounds, the isotopic precision as well as the accuracy varied with the drift of the background (from 0.06 to 1.53‰ for the precision and from 0.28 to 2.10‰ for the accuracy). Additional experiments carried out with CO2 pulses confirmed these observations. Successive pulses of the CO2 reference gas (arbitrarily set at 0‰) were performed with a gradient of temperature followed by an isothermal period of 5 min (Figure 4A). By increasing the temperature and then the CO2 background, as reported in Figure 4B, the area over the amplitude of the background (mV) ratios dropped significantly from 0.33 to 0.05. As expected from the previous experiments with fatty acids, the isotopic precision as well the accuracy decreased severely with the increase of the background drift and with the temperature (37) Meier-Augenstein, W.; Watt, P. W.; Langhans, C. D. J. Chromatogr., A 1996, 752, 233–41. (38) Teutenberg, T.; Tuerk, J.; Holzhauser, M.; Kiffmeyer, T. K. J. Chromatogr., A 2006, 1119, 197–201.

Figure 3. Temperature-programmed LC-IRMS chromatogram of hydrosoluble fatty acids (standard mixtures) injected on the Hypercarb column with a gradient of temperature. The buffer is phosphoric acid; with a flow rate at 220 µL/min; the temperature gradient started at 35 °C, held for 5 min and then a ramp of temperature of 20 °C/min until 110 °C, held for 1 min. The compounds injected are (1) formic acid; (2) acetic acid; (3) propionic acid; (4) tartaric acid; (5) butyric acid; (6) succinic acid; and (7) citric acid. The mixture is not homogeneous in terms of amount of sample injected: It varies from 100-250 ng of carbon injected.

drop at the end of the LC run (see Figure 4C). Thus, between the run time 10 and 16 min, corresponding to the temperature gradient (at 20 °C/min) in our LC run, the SD(δ13C) calculated on the CO2 pulses was 3.65‰. When this result is normalized to the drift of the CO2 background (304 mV), the drift produced a SD(δ13C) at 0.01 ‰/mV. Additionally, in the isothermal plateau after the temperature gradient, the signal/background ratios were stable and there was a significant improvement in precision as well accuracy. The SD(δ13C) measured during the first 8 min at 35 °C was 0.11‰ (n ) 15 pulses). In the isothermal period (corresponding at 4 min of stable signal/ background ratios), the SD(δ13C) was 0.23‰ (n ) 9 pulses). Thus, to obtain a meaningful improvement of isotopic measurements with temperature gradient, an alternative solution could be to elute the compound after the temperature gradient in a temperature plateau where generally the CO2 background is stable The results of temperature-programmed LC-IRMS show here that the nonlinearity of the IRMS background impacts the isotopic precision and accuracy. These anomalies can be due to (1) the coelution of peaks which can be improved by changing the chromatographic conditions or by using a specific curve fitting algorithm35 (mainly for the restoration of the accuracy rather than precision); (2) the incorrect definition of the background either because of inaccurate choice of the integration parameters or because of inconsistent definitions for each three channels (m/z 44, 45, and 46). Despite their importance, these two points were not discussed in this manuscript. (3) The thermal degradation of the compound (as described here for ferulic acid); (4) the modification of the oxidation efficiency during the run after the modification of the temperature (as described here) or the flow rates.9

Quantitative Measurements by HTLC-IRMS of p-Coumaric Acid. With the era of LC-IRMS, the use of the IRMS device to perform quantitative measurements becomes an appropriate and complementary tool to the 13C high-precision isotopic ratio measurements. As reported by several authors, the peak area is proportional to the number of carbon injected for a specific compound. Shierbeek et al. used this approach to measure the glutathione concentration in human erythrocyte blood using an internal standard in the range of 0.2-2 mg/ mL.10 Here, we report such an approach to quantitate pcoumaric acid with the Hypercarb column at 160 °C with phosphate buffer using benzoic acid as an internal standard. In the range measured (between 95 and 2375 ng of carbon corresponding to 0.9-22.5 nmol injected or 0.01-0.4 mg/mL), a linear correlation was found with r2 ) 0.999 (by plotting in the x-axis the amount of carbon injected versus in the y-axis the ratio of peak areas between p-coumaric acid and benzoic acid). In the range of concentration studied, for p-coumaric acid and benzoic acid, δ13C was -28.50 ± 0.20‰ (n ) 15) and -29.95 ± 0.31‰ (n ) 15), respectively. Measurement of Isotopic Ratio with a Small Percentage of Residual Methanol in the Injected Solution. Problems may arise for measurement of the isotopic ratio by LC-IRMS (only with the LC Isolink interface) when the removal of organic solvent is uncompleted in the sample preparation or when organic solvent is added in the injected solution when the compound is difficult to solubilize in aqueous buffer. It can therefore be hypothesized that use of programmed temperature would solve this point. We investigated here the effect of the addition of a small percentage of methanol in the injected solution (and not in the LC buffer). We studied the δ13C values of the compound which would elute Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

7149

Figure 4. Panel A is a temperature-programmed LC-IRMS chromatogram of CO2 pulses injected every 30 s on a Hypercarb column with a temperature gradient; flow rate 220 µL/min; phosphoric acid 20 mM; temperature gradient started at 35 °C held for 5 min and then a ramp of temperature of 20 °C/min until 110 °C, held for 5 min. Panel B describes the area (V s) of the CO2 pulses divided by the CO2 background (measured on m/z 44) over the run time. Panel C represents the variation ∆13C, (‰) of δ13C through the LC run. For the second pulse, the δ13C is set at 0‰. For each determination, three successive pulses are grouped and two LC runs are used. 7150


Figure 5. Panel A is the temperature-programmed LC-IRMS chromatogram of methanol (12.5 nmol or 5% v/v) carried out with the Hypercarb column; flow rate 500 µL/min; phosphate buffer 20 mM, pH 7.2; isothermal temperature at 170 °C. Before and after the CO2 peak of methanol, CO2 pulses are injected every 30 s to mimic the elution of other compounds. Panel B is a representation of the variation of the 13C isotopic ratios of CO2 pulses before and after the elution of methanol (represented by the black bar). Arbitrarily, the reference CO2 gas was set at 0‰. Each point represents the grouping of four successive pulses within a duplicate run. The dotted line is the evolution for the injection of 1% methanol (2.5 nmol), and the bold line is the injection of 5% methanol.

after the methanol peak at 35, 85, and 170 °C (in an isothermal run). By injecting a spike of 10% v/v of methanol (∼25 nmol injected) on the Hypercarb column at 35 and 85 °C (data not shown), the resulting CO2 peak amplitude was higher than 10 V, and the peak tailing was very broad (>800 s). However, by increasing the temperature to 170 °C and by reducing the amount of methanol to 5% (12.5 nmol) and successively to 1% (v/v), the CO2 peak width decreased to ∼200 s, respectively (with an amplitude higher than 10 V for both conditions). As reported in Figure 5A, a series of standard pulses of CO2 reference gas was performed. By injecting 5% of methanol and grouping four successive CO2 pulses together for each determination (corre-

sponding to a width at baseline of 150 s) before and after the elution of methanol, the isotopic precision and accuracy of the CO2 pulses were seen to be affected 4 min after the elution of methanol (Figure 5B). The isotopic precision was shown to be 10-fold higher than for the first group of pulses in the beginning of the run and SD(δ13C) was 0.29 versus 0.04‰. After 10 min, SD(δ13C) returned to 0.04‰. Additionally, with the injection of 1% of methanol (2.5 nmol), the isotopic precision of the CO2 pulses was only slightly affected 3 min after the elution of the CO2 peak of methanol if compared to pulses before its injection, and SD(δ13C) was 0.06 versus 0.04‰. These experiments showed that the temperature-programmed LC-IRMS might also be a powerful Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

7151

tool to measure isotopic ratio of molecules when small percentages of organic solvents are present in the injected solution. CONCLUSIONS The new analytical strategy based on the coupling of the temperature-programmed LC-IRMS presents interesting insights for high-precision 13C isotopic analyses. Compared to the conventional parameters which can be optimized during a LC-IRMS method development (which are the column, the ionic strength of the buffer, and the pH), temperature can play a major role by changing the eluent strength of the aqueous mobile phase. LC separation at temperatures as high at 170 °C can be used without significantly affecting the isotopic ratio, although the CO2 background is increased. However, at this temperature, degradation may be observed for compounds with long residence times giving unreliable 13C isotopic ratios. Temperature gradients can be also used with LC-IRMS. Precaution is needed since the drift of the CO2 background affects the precision and accuracy when the targeted compound is eluted in the drift area. Moreover temper-

7152


ature-programmed LC-IRMS adds new and orthogonal stationary phase packing materials to the field of high-precision 13C isotopic analyses. While demonstrating the effectiveness of the temperature programmed for high-precision 13C isotopic measurement, this work also lays the groundwork for future isotopic applications with a new tool for deciphering complex samples. ACKNOWLEDGMENT Thanks go to Dr. L. Poquet and Prof. D. Barron for their discussion on phenolic acids and C. Parker for proof-reading of this manuscript. We would like also thank Dr. E. Varesio for the fruitful discussion on temperature gradient as well as C. Grivet for some preliminary work. We also thank the two anonymous reviewers for their valuable advice and constructive comments. Received for review February 29, 2008. Accepted June 5, 2008. AC8004204

Temperature-Programmed High-Performance Liquid Chromatography

Recommend Documents