Nitrogen Compound Characterization in Fuels by Multidimensional Gas Chromatography
Özet
Here, we present a method utilizing two-dimensional gas chromatography and nitrogen chemiluminescence detection (GCxGC-NCD) to extensively characterize the different classes of nitrogen-containing compounds in diesel and jet fuels.
Abstract
Certain nitrogen-containing compounds can contribute to fuel instability during storage. Hence, detection and characterization of these compounds is crucial. There are significant challenges to overcome when measuring trace compounds in a complex matrix such as fuels. Background interferences and matrix effects can create limitations to routine analytical instrumentation, such as GC-MS. In order to facilitate specific and quantitative measurements of trace nitrogen compounds in fuels, a nitrogen-specific detector is ideal. In this method, a nitrogen chemiluminescence detector (NCD) is used to detect nitrogen compounds in fuels. NCD utilizes a nitrogen-specific reaction that does not involve the hydrocarbon background. Two-dimensional (GCxGC) gas chromatography is a powerful characterization technique as it provides superior separation capabilities to one-dimensional gas chromatography methods. When GCxGC is paired with a NCD, the problematic nitrogen compounds found in fuels can be extensively characterized without background interference. The method presented in this manuscript details the process for measuring different nitrogen-containing compound classes in fuels with little sample preparation. Overall, this GCxGC-NCD method has been shown to be a valuable tool to enhance the understanding of the chemical composition of nitrogen-containing compounds in fuels and their impact on fuel stability. The % RSD for this method is <5% for intraday and <10% for interday analyses; the LOD is 1.7 ppm and the LOQ is 5.5 ppm.
Introduction
Before use, fuels undergo extensive quality assurance and specification testing by refineries to verify that the fuel they are producing will not fail or cause equipment problems once disseminated. These specification tests include flash point verification, freeze point, storage stability, and many more. The storage stability tests are important as they determine if the fuels have a tendency to undergo degradation during storage, resulting in the formation of gums or particulates. There have been incidences in the past when F-76 diesel fuels have failed during storage even though they passed all specification tests1. These failures resulted in high concentrations of particulate matter in the fuels that could be detrimental to equipment such as fuel pumps. The extensive research investigation that followed this discovery suggested that there is a causal relationship between certain types of nitrogen compounds and the particulate formation2,3,4,5. However, many of the techniques used to measure nitrogen content are strictly qualitative, require extensive sample preparation, and provide little information on the identity of the suspect nitrogen compounds. The method described herein is a two-dimensional GC (GCxGC) method paired with a nitrogen chemiluminescence detector (NCD) that was developed for the purpose of characterizing and quanitifying trace nitrogen compounds in diesel and jet fuels.
Gas chromatography is used extensively in petroleum analyses and there are over sixty published ASTM petroleum methods associated with the technique. A wide range of detectors are combined with gas chromatography such as mass spectrometry (MS, ASTM D27896, D57697), Fourier-transform infrared spectroscopy (FTIR, D59868), vacuum ultraviolet spectroscopy (VUV, D80719), flame ionization detector (FID, D742310), and chemiluminesence detectors (D550411, D780712, D4629-1713). All these methods can provide significant compositional information about a fuel product. Since fuels are complex sample matrices, gas chromatography enhances compositional analysis by separating out sample compounds based on boiling point, polarity, and other interactions with the column.
To further this separation ability, two-dimensional gas chromatography (GCxGC) methods can be utilized to provide compositional maps by using sequential columns with orthogonal column chemistries. Separation of compounds occur both by polarity and boiling point, which is a comprehensive means to isolate fuel constituents. Although it is possible to analyze nitrogen-containing compounds with GCxGC-MS, the trace concentration of the nitrogen compounds within the complex sample inhibits identification14. Liquid-liquid phase extractions have been attempted in order to use GC-MS techniques; however, it was found that the extractions are incomplete and exclude important nitrogen compounds15. Additionally, others have used solid phase extraction to enhance the nitrogen signal while reducing the potential for the fuel sample matrix interference16. However, this technique has been found to irreversible retail certain nitrogen species, especially low molecular weight nitrogen-bearing species.
The nitrogen chemiluminescence detector (NCD) is a nitrogen-specific detector and has been successfully used for fuel analyses17,18,19. It utilizes a combustion reaction of nitrogen-containing compounds, the formation of nitric oxide (NO), and a reaction with ozone (see Equations 1 & 2)20. This is accomplished in a quartz reaction tube that contains a platinum catalyst and is heated to 900 °C in the presence of oxygen gas.
The photons emitted from this reaction are measured with a photomultiplier tube. This detector has a linear and equimolar response to all nitrogen-containing compounds because all nitrogen-containing compounds are converted to NO. It is also not prone to matrix effects because other compounds in the sample are converted to non-chemiluminescence species (CO2 and H2O) during the conversion step of the reaction (Equation 1). Thus, it is an ideal method for measuring nitrogen compounds in a complex matrix such as fuels.
The equimolar response of this detector is important for nitrogen compound quantitation in fuels because the complex nature of fuels does not allow for calibration of each nitrogen analyte. The selectivity of this detector facilitates the detection of trace nitrogen compounds even with a complex hydrocarbon background.
Protocol
CAUTION: Please consult relevant safety data sheets (SDS) of all compounds before use. Appropriate safety practices are recommended. All work should be performed while wearing personal protective equipment such as gloves, safety glasses, lab coat, long pants, and closed-toed shoes. All standard and sample preparations should be done in a ventilated hood.
1. Preparation of standards
- Prepare a 5,000 mg/kg (ppm) solution of carbazole (calibration standard, minimum of 98% purity) by placing 0.050 g in a vial and bring the total mass of each solution to 10.000 g with isopropyl alcohol. Cap the vial immediately to prevent loss of isopropyl alcohol. This is the calibration stock solution.
- Prepare a carbazole solution with 100 ppm nitrogen content by diluting 1.194 mL of stock solution to 5 mL with isopropyl alcohol. This is designated as "100 ppm nitrogen carbazole" and is used to create the calibration standards.
NOTE: The concentrations of the calibration standards indicate the concentration of nitrogen in the standard, not the carbazole concentration - Prepare the following calibration standards by serial dilution:
20 ppm nitrogen carbazole
10 ppm nitrogen carbazole
5 ppm nitrogen carbazole
1 ppm nitrogen carbazole
0.5 ppm nitrogen carbazole
0.025 ppm nitrogen carbazole - Put 1 mL of the calibration standards into separate GC vials (6 vials total).
- Prepare individual 10 ppm solutions of each standard compounds listed in Table 1 in isopropyl alcohol. Place 1 mL of each standard solution in separate GC vials (10 vials total).
NOTE: The standard compounds listed in Table 1 will be used to classify the unknown nitrogen compounds as 'light nitrogen compounds', 'basic nitrogen compounds', or 'non-basic nitrogen compounds'.
| Standard Compound | Elution Time Classification Group |
| Pyridine | Group 1 – light nitrogen compounds |
| Trimethylamine | Group 1 – light nitrogen compounds |
| Methylaniline | Group 1 – light nitrogen compounds |
| Quinoline | Group 2 – basic nitrogen compounds |
| Diethylaniline | Group 2 – basic nitrogen compounds |
| Methylquinoline | Group 2 – basic nitrogen compounds |
| Indole | Group 2 – basic nitrogen compounds |
| Dimethylindole | Group 2 – basic nitrogen compounds |
| Ethylcarbazole | Group 3 – Non-basic nitrogen compounds |
| Carbazole | Group 3 – Non-basic nitrogen compounds |
Table 1: Nitrogen standards and their elution classification groups.
2. Sample preparation
- For diesel fuels: In a GC vial, add 250 µL of fuel sample and 750 µL of isopropyl alcohol.
- For jet fuels: In a GC vial, add 750 µL of fuel sample and 250 µL of isopropyl alcohol.
NOTE: If the total nitrogen concentration of either diesel or jet fuel falls below the calibration curve (0.025 ppm nitrogen) when diluted as instructed above, do not dilute. If the nitrogen concentration in a specific nitrogen group in either diesel or jet fuel falls above the calibration curve (20 ppm nitrogen), dilute sample further.
3. Instrument setup
- Instrument configuration
- Auto-sampler: Ensure that the autosampler tray and tower are installed with a splitless inlet and wash vials in place.
- Nitrogen Chemiluminescence Detector: Ensure that the nitrogen chemiluminescence detector is installed with the appropriate gas lines (i.e., helium and hydrogen). A hydrogen generator can be utilized instead of a tank, if available.
- Duel Loop Thermal Modulator: Ensure that the duel loop thermal modulator is installed and aligned properly so that the column loop will be centered between the cold and hot jet flows during modulation.
- Column installation
- Ensure that the instrument is in maintenance mode (i.e., all burners and gas flows are turned off).
- Insert the 30 m primary column into the GC oven and connect to the splitless inlet.
- Measure and cut 2.75 m of the secondary column. Put a mark on the secondary column at 0.375 m and 1.375 m using a white-out pen.
- Place the secondary column into the Zoex modulator column holder and use the marks as guides for creating a 1 m loop within the holder for modulation.
- Connect the shorter end of the secondary column to the primary column using a micro-union. Check for a successful connection by turning on the gas flow and inserting the open end of the column into a vial of methanol. A successful connection is confirmed by the presence of bubbles.
- Place the column holder into the modulator and adjust the loops as necessary so that the loops line up properly with the cold and hot jets, as pictured in Figure 1.
- Insert the other end of the column into the NCD burner. Then turn on all burners and gas flows to ensure there are no leaks.
- Turn the oven on at the maximum temperature limit for a minimum of 2 h in order to bake-out the columns. Once completed, verify that there are no new leaks. Then, cool the oven.
Figure 1: Schematic representation of the GCxGC-NCD instrumentation. This figure has been reprinted from Deese et al. Please click here to view a larger version of this figure.
- Method parameters
- Using the computer software, set the instrument to the parameters listed in Table 2.
- Set the initial oven temperature to 60 °C with a ramp rate of 5 °C/min to 160 °C, and then change the ramp rate to 4 °C until 300 °C. The total run time is 55 minutes per sample.
- Set the hot jet temperature to be 100 °C higher than the oven temperature at any point in time. Thus: set the initial hot jet temperature to 160 °C with a ramp rate of 5 °C/min to 260 °C, and then change the ramp rate to 4 °C until 400 °C.
- Set the ancillary liquid nitrogen Dewar connected to the GC to stay between 20% and 30% full during the run.
| Instrument Parameters | ||
| NCD | Nitrogen Base Temperature | 280 °C |
| Nitrogen Burner Temperature | 900 °C | |
| Hydrogen flow rate | 4 mL/min | |
| Oxidizer flow rate (O2) | 8 mL/min | |
| Data collection rate | 100 Hz | |
| Inlet | Inlet Temperature | 300 °C |
| Inlet liner | Splitless | |
| Purge flow to split vent | 15 mL/min | |
| Septum Purge Flow | 3 mL/min | |
| Carrier gas | He | |
| Carrier gas flow rate | 1.6 mL/min | |
| Syringe size | 10 µL | |
| Injection volume | 1 µL | |
| Modulator | Modulation time | 6000 ms |
| Hot pulse duration | 375 ms | |
| Columns | Flow | 1.6 mL/min |
| Flow type | Constant flow | |
Table 2: Instrument parameters.
4. Instrument calibration
- Place the GC sample vials containing the prepared carbazole standards and load the previously configured method into the GC software.
- Create a sequence that aliquots the blank (isopropyl alcohol) at the beginning followed by the prepared carbazole standards by increasing concentration.
- Verify that the liquid nitrogen Dewar is between 20-30% full and all instrument parameters are in "ready" mode. Start the sequence.
- Once the calibration standard set analysis is completed, use the GCImage software to load each chromatogram, background correct, and detect each carbazole peak or blob.
NOTE: In GCImage, the detected peaks within the chromatogram are termed as "blobs" by the software. - In a spreadsheet program, plot the response (blob volume) against the nitrogen concentration (ppm) of each calibration standard to create a calibration curve (see Figure 2). The trend line of the curve should have R2 ≥ 0.99.
Figure 2: Example GCxGC-NCD carbazole calibration curve. Please click here to view a larger version of this figure.
5. Sample analysis
- Place the GC sample vials in the autosampler tray and load the previously configured method.
- Create a sequence that has a blank (isopropyl alcohol) at the beginning and then every 5 subsequent samples to limit any build-up of fuel within the columns.
- Verify that enough liquid nitrogen is available in the modulator's Dewar and that all instrument parameters are in "ready" mode. Then, start the sequence.
6. Data analysis
- Open the chromatogram in the GCImage software for data analysis and perform a background correction
- Detect blobs using the following filter parameters:
Minimum area = 25
Minimum volume = 0
Minimum peak = 25
NOTE: These parameters are subject to change based on instrument response or sample matrix. - Use the GCImage template function to create or load a template to group nitrogen compound classes based on the elution times of the known standards (see Table 1).
NOTE: Further explanation of template usage can be found in the representative results and Figure 8. - Once the compounds have been grouped, export the "blob set table" into a spreadsheet program. Use the sum the volume of all blobs/peaks within each compound class group, and the calibration equation determined in Section 4.4 to calculate the concentration in ppm for the nitrogen compounds in each group.
- If desired, use the following density calculations to correct for differences in the injection volume of sample vs. standards for quantitation:
NOTE: *percent difference between the ng N injected in the sample matrix vs. the standard matrix - Sum all nitrogen content in each compound class to obtain the total nitrogen content of the sample, if desired. If the total nitrogen content is determined to be above 150 ppm nitrogen or a compound class bin is outside of the calibration range, dilute the sample further for analysis. Compare these results with total nitrogen content as determined by ASTM D462913 for quantification verification.
Representative Results
The nitrogen-containing compound, carbazole, was used in this method as the calibration standard. Carbazole elutes at approximately 33 min from the primary column and at 2 s from the secondary column. These elution times will vary slightly depending on the exact column length and instrumentation. In order to obtain a proper calibration curve and, subsequently, good quantitation of nitrogen compounds within a sample, the calibration peaks should not be overloaded nor have any nitrogen contaminants. The primary and secondary column chromatograms of the carbazole calibration standard containing 0.025 ppm N are shown in Figure 3. There is not any tailing and the standard response is outside of the noise.
Figure 3: Representative chromatograms of 0.025 ppm N carbazole calibration standard on the primary (left) and secondary (right) columns. Please click here to view a larger version of this figure.
Figure 4 is an example of a GCxGC-NCD chromatogram with a carbazole standard and the resulting blob table. As can be seen, there are two detected blobs that are not within the carbazole elution time, and they have been excluded from the blob table. Extraneous peaks or blobs should not be included in the calibration curve.
Figure 4: Representative GCxGC-NCD chromatogram of a carbazole standard diluted with isopropyl alcohol. The extraneous peaks are circled in yellow. Please click here to view a larger version of this figure.
Figure 5 illustrates a typical chromatogram obtained by using this method on a diesel fuel sample and Figure 6 is a typical chromatogram of a jet fuel sample. Typically, jet fuel has fewer nitrogen compounds at lower concentrations than a diesel fuel, which can be clearly seen when comparing the two chromatograms. The peaks or "blobs" in these chromatograms are oval-shaped (little-to-no 'streaking' or too much retention on either column) and are easily distinguished from each other. It is clear that different classes of nitrogen compounds are present in diesel fuel when compared to jet fuel.
Figure 5: Representative GCxGC-NCD chromatogram containing the nitrogen compounds found in a diesel fuel. Please click here to view a larger version of this figure.
Figure 6: Representative GCxGC-NCD chromatogram containing the nitrogen compounds found in a jet fuel. Please click here to view a larger version of this figure.
In contrast to the previous examples, Figure 7 illustrates two failed sample measurements. The image on the left occurs when the modulation time is incorrect for the oven temperature, consequently resulting to wrap-around in the column. The solution to this type of failure is to either increase the modulation time or increase the temperature of the oven. The chromatogram on the right illustrates a "streaking" effect of the blobs. This occurs when the compounds are retained on the sample for too long and it destroys any compound separation. From experience, this tends to be caused by a build-up of compounds within the column. This problem can be repaired by running multiple blanks and "burning out" the column by increasing the temperature of the oven to 300 °C and allowing it to sit for several hours at that temperature.
Figure 7: Representation of failed chromatograms. Wrap around caused by incorrect modulation time (left) and peak degradation caused by sample retention on the column (right). Please click here to view a larger version of this figure.
Standards (as listed in Table 1) can be utilized to determine the groups associated with each nitrogen compound class. An example of these standard groups can be seen in Figure 8. The retention times of the standards may differ slightly on different instrumentation or different column sets. Hence, it is imperative to run the standards each time an instrument parameter is changed.
Figure 8: An example of the retention times of the standards listed in Table 1. Please click here to view a larger version of this figure.
A template can be created in GCImage to separate the nitrogen compounds found in fuel by different nitrogen classes. The template should be built from the elution times determined by the standards and then overlaid on each fuel chromatogram. Figure 9 is a representation of a template with the three groups as determined by the standard elution times. Once the template is overlaid, the blob set table will indicate the number of blobs and the total volume within each classified group.
Figure 9: Representative GCxGC-NCD chromatogram with overlaid template and blob set table. Please click here to view a larger version of this figure.
The response factor from the calibration curve should then be used to calculate the concentration of nitrogen compounds within each nitrogen class. Figure 10 depicts the concentration in ppm for nitrogen compounds detected in each of the three classes for a batch of diesel fuel samples.
Figure 10: Representative results of the nitrogen concentration (ppm) in diesel fuels by group. Please click here to view a larger version of this figure.
Supplementary File 1: The intraday and interday repeatability of total nitrogen concentration by GCxGC-NCD for four fuels. Please click here to view this file (Right click to download).
Discussion
The purpose of this method is to provide detailed information on the nitrogen content of diesel and jet fuels without extensive sample preparation such as liquid extractions. This is achieved by pairing a two dimensional GC system (GCxGC) with a nitrogen-specific detector (nitrogen chemiluminescence detector, NCD). The GCxGC provides significant separation of the compounds relative to traditional one-dimensional GC. The NCD provides trace nitrogen compound detection without any background interferences. Other nitrogen-specific detectors that have been used in the past, such as a nitrogen phosphorus detector (NPD), are interfered by the fuel's hydrocarbon matrix. In contrast, this method has little-to-no matrix interferences.
This GCxGC method uses a reverse-phase (polar-to-nonpolar) column setup so the compounds in the first dimension are separated by polarity while in the second dimension they are separated by boiling point. The second dimension separation is controlled by a thermal modulator that re-concentrates compounds via cryo-focusing and then separates the compounds further. The secondary column within the modulator must be accurately placed in order to achieve optimal separation. If the column loop is not centered between the hot and cold jet, the peaks will not have the proper shape or elute correctly. Furthermore, helium is used as the carrier gas for this system. Although hydrogen gas could be used as a carrier gas, there is a possibility that it can create active sites, which will interact with the nitrogen compounds. In order to completely eliminate that possibility, helium is highly recommended.
Because of the trace nature of the nitrogen compounds found in these fuels, mass spectrometry characterization is difficult to obtain. The most effective way to identify the nitrogen compound classes with this system is by injecting a variety of known nitrogen-containing compounds and creating a nitrogen class template based on these standards (see Table 1). The elution times of these compounds may vary slightly depending on the instrument used. Thus, it is imperative that the standard set is measured on each instrument and a unique template is created. This template can then be utilized for the fuel samples in order to characterize the classes of nitrogen compounds in a fuel and to provide quantitative information.
The ideal method for quantification of these compounds is to sum the total blob volume within each class, use the calibration equation to calculate the concentration of nitrogen per class, and then sum the class content to obtain the total nitrogen concentration. The repeatability of these measurements for analyses on both the same day and over different days has been found to be < 20% RSD (see Supplementary File 1). The highest limit of detection (LOD) and limit of quantitation (LOQ) has been found to be 1.7 ppm and 5.5 ppm, respectively (see Supplementary File 1).
To the best of our knowledge, the purpose of the method detailed is to provide significant characterization of the classes of nitrogen compounds in diesel and jet fuels. Other nitrogen characterization methods require the use of liquid extractions (which have been found to exclude imperative nitrogen compounds) and detection schemes that have significant matrix interferences. Both jet and diesel samples can be measured by using the same method and instrument configuration, the only difference is the extent of dilution of the samples before measurement. There are current efforts underway to use this GCxGC-NCD method as a way of further characterizing fuels (in addition to the published ASTM methods) in order to determine and predict fuel quality. This characterization project includes increasing the number of nitrogen standards utilized to create a reliable template to enhance the chemical compositional analysis of fuels containing nitrogen compounds, which will further refine the understanding of compounds that are detrimental to fuels in long term storage.
Açıklamalar
The authors have nothing to disclose.
Acknowledgements
Funding support for this work was provided by the Defense Logistics Agency Energy (DLA Energy) and the Naval Air Systems Command (NAVAIR).
This research was performed while an author held an NRC Research Associateship award at the U.S. Naval Research Laboratory.
Materials
| 10 µL syringe | Agilent | gold series | |
| 180 µm x 0.18 µm Secondary Column | Restek | Rxi-1MS | nonpolar phase column, crossbond dimethyl polysiloxane |
| 250 µm x 0.25 µm Primary Column | Restek | Rxi-17SilMS | midpolarity phase column |
| Autosampler tray and tower | Agilent | 7963A | |
| Carbazole | Sigma | C5132 | 98% |
| Diethylaniline | Aldrich | 185898 | ≥ 99% |
| Dimethylindole | Aldrich | D166006 | 97% |
| Duel Loop Thermal Modulator | Zoex Corporation | ZX-1 | |
| Ethylcarbazole | Aldrich | E16600 | 97% |
| Gas chromatograph | Agilent | 7890B | |
| GC vials | Restek | 21142 | |
| GCImage Software, Version 2.6 | Zoex Corporation | ||
| Indole | Aldrich | 13408 | ≥ 99% |
| Isopropyl Alcohol | Fisher Scientific | A461-500 | Purity 99.9% |
| Methylaniline | Aldrich | 236233 | ≥ 99% |
| Methylquinoline | Aldrich | 382493 | 99% |
| Nitrogen Chemiluminescence Detector | Agilent | 8255 | |
| Pyridine | Sigma-Aldrich | 270970 | anhydrous, 99.8% |
| Quinoline | Aldrich | 241571 | 98% |
| Trimethylamine | Sigma-Aldrich | 243205 | anhydrous, ≥ 99% |
Referanslar
- Garner, M. W., Morris, R. E. Laboratory Studies of Good Hope and Other Diesel Fuel Samples. ARTECH Corp. Report No. J8050.93-FR. , (1982).
- Morris, R. E. Fleet Fuel Stability Analyses and Evaluations. ARTECH Corp. Report No. DTNSRDC-SME-CR-01083. , (1983).
- . Analysis of F-76 Fuels from the Western Pacific Region Sampled in 2014. Naval Research Laboratory Letter Report 6180/0012A. , (2015).
- Westbrook, S. R. Analysis of F-76 Fuel, Sludge, and Particulate Contamination. Southwest Research Institute Letter Report. Project No. 08.15954.14.001. , (2015).
- Morris, R. E., Loegel, T. N., Cramer, J. A., Leska Myers, K. M., A, I. Examination of Diesel Fuels and Insoluble Gums in Retain Samples from the West Coast-Hawaii Region. Naval Research Laboratory Memorandum Report. No. NRL/MR/6180-15-9647. , (2015).
- Maciel, G. P., et al. Quantification of Nitrogen Compounds in Diesel Fuel Samples by Comprehensive Two-Dimensional Gas Chromatography Coupled with Quadrupole Mass Spectrometry. Journal of Separation Science. 38 (23), 4071-4077 (2015).
- Deese, R. D., et al. Characterization of Organic Nitrogen Compounds and Their Impact on the Stability of Marginally Stable Diesel Fuels. Energy & Fuels. 33 (7), 6659-6669 (2019).
- Lissitsyna, K., Huertas, S., Quintero, L. C., Polo, L. M. Novel Simple Method for Quanitation of Nitrogen Compounds in Middle Distillates using Solid Phase Extraction and Comprehensive Two-Dimensional Gas Chromatography. Fuel. 104, 752-757 (2013).
- Machado, M. E. Comprehensive two-dimensional gas chromatography for the analysis of nitrogen-containing compounds in fossil fuels: A review. Talanta. 198, 263-276 (2019).
- Adam, F., et al. New Benchmark for Basic and Neutral Nitrogen Compounds Speciation in Middle Distillates using Comprehensive Two-Dimensional Gas Chromatography. Journal of Chromatography A. 1148, 55-65 (2007).
- Wang, F. C. Y., Robbins, W. K., Greaney, M. A. Speciation of Nitrogen-Containing Compounds in Diesel Fuel by Comprehensive Two-Dimensional Gas Chromatography. Journal of Separation Science. 27, 468-472 (2004).
- Yan, X. Sulfur and Nitrogen Chemiluminescence Detection in Gas Chromatographic Analaysis. Journal of Chromatography A. 976 (1), 3-10 (2002).
