Anal. Chem. 1999, 71, 1083-1086
Optimization of 15N Detection with an Atomic Emission Detector Nanette A. Stevens*,† and Michael F. Borgerding†
Department of Chemistry, Virginia Tech, Blacksburg, Virginia 24061
Gas chromatography with atomic emission detection is a useful tool for the detection of stable isotope labels in complex samples. While papers involving the analysis of D and 13C are numerous, little work has been done in the area of 15N detection. For 15N isotope detection, three reagent gases are used: H2, O2, and CH4. In this work, the reagent gas flows were varied to optimize the sensitivity of 15N detection without sacrificing isotope selectivity. The optimal gas flows determined in this work produce the following ratios of the spectral peak areas: O 725 area/He 728 area ) 0.039 with only O2 flowing; H 486 area/He 492 area ) 12 with only H2 flowing; C 496 area/ He 502 area ) 0.41 with O2, H2, and CH4 flowing for C and no gases flowing for He. When using these gas settings, the 15N sensitivity is increased by nearly 2 orders of magnitude relative to the manufacturer-recommended settings. It was also demonstrated that the presence of a compound in both the labeled and unlabeled forms in the same sample does not affect the response. The ratios of 15N to 14N in standards, calculated from calibration plots (which are linear for both isotopes), agree well with the actual values. A tobacco smoke sample containing various 15N-labeled compounds was used to show the utility of the GC-AED for indicating which compounds in a complex sample contain the label. This sample also demonstrates the necessity for optimal sensitivity when dealing with samples containing small amounts of compounds with low incorporation levels. It is often important to determine the resulting products from a compound undergoing complex reactions, e.g., metabolism or pyrolysis. This is particularly difficult if the reactions occur within a complex matrix. When the sample to be analyzed contains hundreds, or even thousands, of compounds, it is nearly impossible to determine which components originated from the compound of interest. The problem is simplified by isotopically labeling the precursor compound. One common approach is radiolabeling.1 The main concerns with this technique are safety due to the radioactivity and radioactive contamination of analytical research instruments. More recently, to avoid the safety concerns of radiolabeling, precursor-fate work has been done using stable isotope labels.2,3 The main difficulty when using stable isotopes †Current address: R. J. Reynolds Tobacco Co., Bowman Gray Technical Center, P.O. Box 1487, Winston-Salem, NC 27102. (1) Jenkins, R. W., Jr. Beitr. Tabakforsch. 1990, 14, 353-378. (2) Chace, D. H.; Abramson, F. P. Anal. Chem. 1989, 61, 2724-2730.
10.1021/ac980997c CCC: $18.00 Published on Web 01/12/1999
© 1999 American Chemical Society
is selective detection. While isotope ratio mass spectrometry can be used to detect isotope enrichment in each compound,2,3 it is not convenient for screening a sample consisting of several hundred compounds to find those which contain the label. Gas chromatography with atomic emission detection (GCAED) has been used as a tool to screen samples containing many compounds to determine precursor-fate relationships of stably labeled components.4,5 For example, Deruaz et al. demonstrated that the urinary metabolites of 13C-labeled caffeine could easily be determined.4 Oshita et al. used GC-AED to examine the fate of 13C-labeled amino acids during yeast brewing of whisky.5 GCAED can be used to detect labeled compounds due to a difference in the emission spectrum of each isotope.6 Although the hydrogen isotopes are the only ones for which the atomic emission spectra can be resolved, 12C, 13C, 14C, and 14N, 15N can be distinguished by their molecular emission spectra following reactions in the AED plasma to produce CO and CN, respectively.6 To use GC-AED for isotope detection in the determination of precursor-fate relationships, several concerns must be addressed. First, the instrument must be able to selectively detect the isotope of interest when it is present at levels greater than the natural abundance (0.37% for 15N). Second, it must be sensitive enough to detect minor components in complex matrixes. Much work has been done to establish the selectivity and sensitivity of the GCAED for analyses of the hydrogen and carbon isotopes.6-9 However, there has been little done in the area of nitrogen isotope detection.10 In this work, sensitivity for 15N was optimized, nitrogen isotope selectivity was examined, and the effect of coelution of labeled and unlabeled compounds on the 15N response was determined. EXPERIMENTAL SECTION A Hewlett-Packard (HP) 5921A atomic emission detector (AED) coupled to a HP 5890 series II gas chromatograph was (3) Sohns, E.; Gerling, P.; Faber, E. Anal. Chem. 1994, 66, 2614-2620. (4) Boukraa, M. S.; Deruaz, D.; Bannier, A.; Desage, M.; Brazier, J. L. J. Pharm. Biomed. Anal. 1994, 12, 185-194. (5) Oshita, K.; Kubota, M.; Uchida, M.; Ono, M. Proc. Congr.-Eur. Brew. Conv., 25th 1995, 387-394. (6) Quimby, B. D.; Dryden, P. C.; Sullivan, J. J. Synth. Appl. Isot. Labeled Compd. 1991, Proc. Int. Symp., 4th 1992, 128-32. (7) Quimby, B. D.; Dryden, P. C.; Sullivan, J. J. Anal. Chem. 1990, 62, 25092512. (8) Deruaz, D.; Bannier, A.; Desage, M.; Brazier, J. L. Anal. Lett. 1991, 24, 1531-1543. (9) Leclerc, F.; Deruaz, D.; Bannier, A.; Brazier, J. L. Anal. Lett. 1994, 27, 1325-1338. (10) Deruaz, D.; Bannier, A.; Pionchon, C.; Boukraa, M. S.; Elbast, W.; Brazier, J. L. Anal. Lett. 1995, 28, 2095-2113.
Analytical Chemistry, Vol. 71, No. 5, March 1, 1999 1083
Table 1. Recommended Reagent Gas Settings (psi) for Detection of 15N ref
O2 setting
H2 setting
CH4 setting
11 12a 10c present work
70 80 toward 7.25 10
25 40 toward 29 41.5
40b toward 58 18
CH4/N2 setting 70
a This recommendation is for the use of 100% methane for the detection of N 388. No separate recommendation is given for the detection of 15N. b CH4 is set by comparing the spectral peaks for C 496 and He 492. The value given is specific to the instrument used in the present work. c Optimization by experimental design was used in this work. Thus, the recommendation tends toward the maxima and minima used in the study, rather than specific values.
used for these studies. The gas chromatograph was equipped with a HP 7673 autosampler, a split/splitless injector, and an on-column injector with electronically programmable pressure. The instrument was operated using the HP G2630AA ChemStation software. Separations were performed with a 30-m, 0.25-mm-i.d. DB-WAX capillary column of 0.25-µm film thickness (J&W Scientific) coupled via fused silica press fit connectors (J&W Scientific) to a 1-m deactivated fused silica precolumn (0.53 mm i.d.) and a 2-m deactivated fused silica transfer line (0.32 mm i.d.). All injections were 1 µL in volume. The on-column injector was operated using oven-track. The splitless injector was maintained at 250 °C with the purge valve off for 30 s. The liner used was a 2-mm-i.d. deactivated splitless liner (Hewlett-Packard 5181-8818). For the optimization study, on-column injection was used with a column flow rate of 1.5 mL/min. For the tobacco smoke samples and coelution study, splitless injection was used with a column head pressure of 23 psi. The GC oven temperature was held at 37 °C for 2 min and then was taken to 230 °C at 2.5 °C/min, where the temperature was held for 21 min. Gas flows for the AED were set such that the nitrogen spectrometer purge was 2 mL/min, the cavity pressure was 1.5 psi, and the cavity vent was 30.0 mL/min with the spectrometer window purge, all reagent gases, and the column flow turned off. 2-Methylpyrazine and nitrobenzene were used as received from Aldrich Chemical Co. Nitrobenzene-15N (99%+) was used as received from C/D/N Isotopes. Smoke samples were collected by smoking 40 cigarettes made from Burley tobacco grown in the presence of K15NO3 fertilizer on a 20-port Borgwaldt smoking machine using 55-mL puffs of 30 s duration. The total particulate matter was dissolved in 20 mL of methanol following collection by electrostatic precipitation. Safety Note. The detection of 15N with the AED requires the use of pure methane. Methane is a highly flammable gas. Installation of the gas cylinder must not be performed in the presence of open flames or sparks. Every connection in the methane gas line should be checked for leaks prior to operation of the instrument. RESULTS AND DISCUSSION Optimization of Sensitivity. The sensitivity of the GC-AED can be changed by varying the flow rates of the reagent gases into the plasma. For 15N detection, three reagent gases are used, 1084 Analytical Chemistry, Vol. 71, No. 5, March 1, 1999
viz., H2, O2, and CH4. The flow rates for these gases entering the HP 5921A AED are controlled by setting the tank pressures. Several recommendations have been made concerning the pressures of the reagent gases for the detection of 15N when using the HP 5921A AED, as shown in Table 1. The first appears in the manufacturer’s literature.11 The recommended pressure settings cannot be used, however, for the literature does not indicate a pressure for CH4, but rather for a mixture of CH4 in nitrogen. If this were used, the nitrogen background from the reagent gas would overwhelm any 14N signal from the sample. Due to the background subtraction necessary for 15N, it would also make this isotope undetectable. The confusion concerning the gases used likely results from the software package that runs the instrument. The original HP Pascal Chemstation software for the HP 5921A AED did not provide recipes for the detection of isotopes, except D and H, or for nitrogen at λ ) 388 nm. Therefore, no detection wavelengths which required the use of pure CH4 were available. Thus, the recommendation is probably a misprint. The upgraded software package (HP G2630AA), however, does allow the detection of isotopes and nitrogen at λ ) 388 nm with the HP 5921A AED. The second recommendation for reagent gas flows was provided by the instrument manufacturer when the updated software package was installed.12 This indicates that, when 100% CH4 is used, the H2 should be at 40 psi, and the O2 should be at 80 psi. The CH4 pressure is then set by comparing the spectral peaks of C 496 and He 492. For the instrument used in this work, this corresponds to a CH4 pressure of 40 psi. The third recommendation in Table 1 gives reagent gas settings for this instrument following optimization using an experimental design. Deruaz et al.10 indicated that the CH4 should be high (toward 58 psi), O2 should be low (toward 7.25 psi), and H2 should be high (toward 29 psi). Since the recommendations differ greatly, it was deemed necessary to experimentally optimize the reagent gas flows to produce maximum sensitivity. To perform the optimization, a mixture of nitrobenzene-15N (3.648 × 10-4 M) and 2-methylpyrazine (4.118 × 10-4 M) in chloroform was prepared. The 2-methylpyrazine was added in order to provide a compound with only natural abundance of 15N for background subtraction purposes. GC-AED was performed on this mixture using 67 different reagent gas pressure combinations. The areas of nitrobenzene-15N vs the reagent gas pressures are shown as contour plots in Figure 1. The data in panel A were obtained with a constant hydrogen pressure (that which is used for analysis of other common atoms with this instrument). There is a clear maximum visible in the plot at O2 ) 10 psi and CH4 ) 18-19 psi (as shown by the black area). Also, in general, the response with O2 ) 10 psi is greater than that at higher oxygen pressures for a constant methane pressure. This is in agreement with the work of Deruaz et al.,10 which indicated that the oxygen pressure should be low. In the present work, pressures lower than 10 psi were not used due to inaccuracy of the pressure regulators. Panel B shows a similar contour plot with O2 held constant at 10 psi, and CH4 and H2 varied. Again, there is a clear maximum at H2 ) 40-41 psi and CH4 ) 18-19 psi (as shown by the black (11) Quimby, B. D.; Larson, P. A.; Dryden, P. C. Hewlett-Packard Application Note 228-363, Hewlett-Packard: Palo Alto, CA, 1996. (12) Contract work by Diablo Analytical, Inc. for Hewlett-Packard. Module 16: GC-AED Maintenance; Hewlett-Packard: Palo Alto, CA, 1997; pp 127-132.
area). It is interesting to note that, for a constant methane pressure, even large changes in the hydrogen pressure have little effect on the 15N response. This also agrees well with the work of Deruaz et al.10 The conditions recommended by the manufacturer (O2 ) 80 psi, H2 ) 40 psi, CH4 ) 40 psi) produced a peak area of 7. This is very small compared to the maximum determined in this work (O2 ) 10 psi, H2 ) 41.5 psi, CH4 ) 18 psi), which gave a peak area of 668 for the same sample. Data could not be collected under the conditions suggested by Deruaz et al. due to the high amount of carbon repeatedly extinguishing the plasma. This is likely due to the plumbing of the gas lines for the two different instruments causing different flow rates of the gases into the plasma for the same tank pressure. For universality, the optimal reagent gas flows can be set by considering spectral peak areas at wavelengths corresponding to elements in the reagent gas. Thus, the conditions recommended by this work can be expressed as the following:
O 725 area ) 0.039; with only O2 flowing He 728 area H 486 area ) 12; with only H2 flowing He 492 area C 496 area ) 0.41; with O2, H2, and CH4 flowing for He 502 area C and no gases flowing for He By setting the gases in this way, the actual flows of reagent gases into the plasma are taken into account, rather than simply considering pressures at the tank. Selectivity. Once the gas flows were set to optimize the sensitivity, the isotope selectivity of the instrument was evaluated. Figure 2 shows chromatograms for detection of 14N and 15N in the same sample used in the optimization above. Chromatogram A is for 14N detection. This demonstrates good selectivity, as only the peak corresponding to 2-methylpyrazine is observed. Chromatogram B is for 15N detection. Again, good selectivity is present as only the nitrobenzene-15N peak is seen. The background subtraction mechanism employed by the instrument is essential to accomplishing selective detection of the minor isotope. At the detection wavelength for 15N, there is overlap of the emission band for 14N. Without background subtraction, peaks would be present for every compound containing nitrogen. Since there is no peak observed for 2-methylpyrazine, the background subtraction mechanism is effectively compensating for overlap of the emission spectra. Thus, only compounds with 15N greater than the natural abundance (which is below the limit of detection for the conditions used) produce peaks. Effect of Coelution of Labeled and Unlabeled Compounds. Although the GC-AED shows good selectivity for compounds containing a 15N label, as shown above, in precursor-fate studies involving a complex matrix, a new concern arises. Specifically, the labeled compounds must be able to be determined accurately in the presence of the unlabeled version of the same compound. The unlabeled compounds may arise from the same precursor naturally occurring in the sample undergoing the same reactions as the added labeled precursor, or from completely different reaction mechanisms. Thus, it was necessary to determine if the presence of the unlabeled compound affects the 15N response.
Figure 1. Contour plots of nitrobenzene-15N peak area vs (A) O2 tank pressure and CH4 tank pressure with H2 pressure constant at 41.5 psi and (B) H2 tank pressure and CH4 tank pressure with O2 pressure constant at 10 psi. Legend: (1) peak area ) 0-55, (2) peak area ) 55-110, (3) peak area ) 110-220, (4) peak area ) 220330, (5) peak area ) 330-440, (6) peak area ) 440-495, (7) peak area ) 495-550, (8) peak area ) 550-605, (9) peak area ) 605660, (black) peak area > 660.
Figure 2. Chromatograms from a 1-µL injection of 4.1 × 10-4 M 2-methylpyrazine (retention time ) 7.50 min) and 3.6 × 10-4 M nitrobenzene-15N (retention time ) 10.03 min) in chloroform detecting (A) 14N and (B) 15N.
A study was done using nine solutions containing various ratios of nitrobenzene and nitrobenzene-15N. The total nitrogen content was held constant at 5.0 × 10-4 M. Linear calibration plots were observed for both 14N (R2 ) 0.996) and 15N (R2 ) 0.999). This Analytical Chemistry, Vol. 71, No. 5, March 1, 1999
1085
Table 2. Determination of Nitrobenzene-15N in the Presence of Nitrobenzene actual % 15N 100 80 70 60 50
calcd % 15N 100 80 69 61 52
actual % 15N 40 30 20 0
calcd % 15N 41 31 22 0
Figure 3. GC-AED chromatograms of a smoke sample generated by machine smoking cigarettes made from tobacco grown in the presence of K15NO3 fertilizer. (A) 15N response using manufacturerrecommended reagent gas flows12 (B) 15N response using optimized reagent gas flows determined in the present work.
seems to indicate that the presence of the unlabeled compound does not affect the 15N response. Calculations were performed to determine the percentage of the compound present in the labeled form from the peak areas. Table 2 shows the comparison between the actual and calculated values. Clearly the calculated values are in good agreement with the actual values. (13) Personal communication with Dr. William M. Coleman, III, R. J. Reynolds Tobacco Co.
1086 Analytical Chemistry, Vol. 71, No. 5, March 1, 1999
15N
in Smoke Sample. The importance of selectivity, maximized sensitivity, and detection of a labeled compound in the presence of the unlabeled version can be seen by considering a “real” sample, i.e., a sample with a complex matrix. A smoke sample was prepared by machine smoking 40 cigarettes made from Burley tobacco grown in the presence of K15NO3 fertilizer. Thus, 15N label was incorporated into the tobacco leaves. GC-AED analysis was then performed on the smoke samples using both the optimized reagent gas flows determined in this work and the manufacturer-recommended gas settings. The results are shown in Figure 3. Panel A displays the 15N response for the manufacturerrecommended gas setting. There is only one small peak at 44 min corresponding to nicotine, the major component of the smoke. A very different chromatogram is obtained using the optimized reagent gas flows determined in this work, as shown in panel B. The nicotine peak of panel B is very large compared to the peak in panel A (note the difference in scale between panels A and B). A variety of minor peaks are also seen, many of which are larger than the nicotine peak using the manufacturer-recommended gas flows. It was of interest to determine the percent incorporation of 15N in the nicotine present in the sample. A value of 7% was determined, which agreed well with previous work involving extraction of nicotine from the tobacco and GC-MSD analysis.13 CONCLUSIONS Optimization of reagent gas flows into the AED plasma produced enormous gains in sensitivity for the detection of 15N without sacrificing isotope selectivity. The gas settings determined in this work produced an increase, relative to the manufacturerrecommended settings, in 15N response of nearly 2 orders of magnitude for both standards and a sample with a complex matrix. The presence of a compound in both a labeled and an unlabeled form in the same sample does not affect 15N response when using the optimized reagent gas flows. As seen for the smoke sample, optimum sensitivity is extremely important when dealing with samples containing small amounts of compounds with low incorporation levels. It has been demonstrated that GC-AED is useful in the detection of stable isotope labels in complex matrixes.
Received for review November 30, 1998. AC980997C
September
4,
1998.
Accepted
