Experimental Setup. The developed spectroscopic system consists of the MIR SC source, whose beam transmits through a multipass cell (MPC) containing the products of the plasma reaction at a low pressure. The transmitted beam is sent towards a custom-built Fourier Transform Spectrometer (FTS) with a spectral resolution of 3 GHz, allowing detection of narrow molecular absorption lines of gas species at a low pressure. The experimental setup is presented in Fig. 1. The light source is a newly developed, fiber-coupled MIR SC source (DTU Fotonik, average power ~ 86 mW, pulse duration ~ 0.5 ns, repetition frequency 3 MHz37). The MPC has 31.2 m effective interaction length (Thorlabs, HC30L/M-M02) and is connected to the discharge cell, such that the reaction products of the plasma are sent to the MPC for interaction with the SC beam. The custom-built FTS has been demonstrated and discussed in detail in our previous work33. A brief description is presented in the Methods section, as well.
Source characterization. The spectral coverage of the MIR SC source was characterized with the FTS system by sending the SC light through the MPC under vacuum. The resulting power spectral density is shown in Fig. 2a and covers the spectral region between 1300–2700 cm− 1 (3.7–7.7 µm). While most of the spectral power is between 1700–2100 cm− 1, spectroscopy in the other wavelength ranges is still feasible with high SNR, due to the high average power of the MIR SC source and sensitive photodetectors. Highly absorbing water (H2O) and CO2 lines can be observed in the spectrum, despite the measurement being performed under vacuum conditions. Outside the MPC, the SC beam path was shielded and purged continuously with N2 gas. In Fig. 2a, the etalon fringes are also visible in the spectrum due to the partial overlap of the SC spots of the consecutive reflections on the MPC mirrors. However, the measured absorbance spectrum is not affected as the etalon fringes are stable and are canceled out by normalizing the sample spectra by the background spectra33. The simulated spectral line intensity of the relevant species from the HITRAN2020 database38 between 1300–2500 cm− 1 are shown in Fig. 2b. Due to the overlap of the absorption lines, a low pressure (16.5 mbar) and a high spectral resolution (3 GHz) are required to distinguish between the absorption lines.
System validation. To initially evaluate the performance of the MIR SC FTS-based system, we measured the spectrum of a gas mixture of 495 ppm CO2 in N2 at 16.5 mbar pressure. Here, we dilute a calibrated mixture of 5% CO2 in N2 (Linde Gas) with pure N2, down to 495 ppm, using two flow controllers. The measured spectrum is shown in Fig. 3 (in black, 500 averages in ~ 16 minutes) alongside a fitted, modelled CO2 spectrum (in red, inverted for clarity). The model spectrum is calculated using the HITRAN database parameters and a Voigt profile, convolved with a sinc function. The retrieved concentration from the fit is 485 ± 12 ppm. The uncertainty is calculated from the standard deviation of the noise in the residual of the fit. In Fig. 3a, the full rotational-vibrational band of CO2 is shown. To demonstrate the agreement between the measurement and the fitting routine, an enlargement of the spectral features between 2357 cm− 1 and 2363 cm− 1 is displayed in Fig. 3b. The residuals are shown in the bottom panels (Fig. 3c and 3d). The rather featureless residuals demonstrate the high precision of the frequency calibration, as well as the good quality of the fitting routine.
The linear response of the system to different applied CO2 concentrations was evaluated in a dynamic range between 0.05–2.5% of CO2 in N2, by diluting the 5% CO2 in N2 mixture further using pure N2 gas. Each measurement consists of 500 averaged spectra, measured in ~ 16 minutes. The reference spectra for the linear fitting routine were simulated from the HITRAN database as described before. The retrieved concentrations from the fit versus the applied concentrations are shown in Fig. 4, together with the corresponding errors, exhibited in the lower panel. The linear fit shows a Pearson correlation coefficient square value of 0.9995, demonstrating a very good agreement between the measured concentration values. The corresponding relative errors are within a ± 4% margin for the entire range, with an average error of 2%.
Product analysis of CO 2 /N 2 discharge plasmas. To further assess the performance of the system and to demonstrate its ability to quantify complex mixtures of reaction products, the outflow of a discharge was used. The discharge was generated in a flowing mixture of 50% CO2 in N2 (flow of 2 ln/h) in a discharge cell at a pressure of 25 mbar. The applied voltage was 17.5 kV with a current of 10 mA that results in a specific energy input of 7.1 MJ/mol. The outflow of the discharge cell was guided to the MPC, which had a controlled pressure of 16.5 mbar. In Fig. 5, the measured absorbance spectra of nitrogen dioxide (NO2), nitric oxide (NO), nitrous oxide (N2O), CO, CO2 and H2O are shown (in black) together with corresponding simulated spectra (inverted, in color, using HITRAN parameters). The concentrations of the species retrieved from fitting the simulations to the absorption spectra are shown in Table 1. The practically featureless residual of the fit, shown in the bottom panel of Fig. 5, indicates a very good fit for all detected species.
A wide variety of different species are detected over a range of concentration levels from hundreds of ppm to percentage-level. The broad spectral coverage does not only allow for detection of absorption lines of various compounds, but also to select absorption lines of a certain species with an appropriate line strength for the given concentration, preventing limitations arising from absorption lines which absorb almost 100% of the light.
Table 1
Retrieved concentrations of electrical discharge products
| Retrieved concentration (% or ppm) |
| 50% CO2 / 50% N2 | 30% CO2 / 70% CH4 |
Carbon dioxide (CO2) | 13.0 ± 0.6% | 13.9 ± 0.7% |
Carbon monoxide (CO) | 14.6 ± 0.7% | 32.1 ± 1.7% |
Nitrous oxide (N2O) | 570 ± 30 ppm | -- |
Nitrogen dioxide (NO2) | 154 ± 13 ppm | -- |
Nitric oxide (NO) | 0.33 ± 0.01% | -- |
Ethylene (C2H4) | -- | 0.73 ± 0.03% |
Formaldehyde (H2CO) | -- | 0.13 ± 0.01% |
Acetone (C3H6O) | -- | 0.3 ± 0.1% |
Acetaldehyde (C2H4O) | -- | 0.3 ± 0.1% |
Product analysis of CO 2 /CH 4 discharge plasma. A bigger challenge in the spectroscopic analysis of the complex mixture of reaction products is the dry reforming of methane, as more complex molecules are formed, which cannot be found in HITRAN. Therefore, to extend the evaluation of the system, a CO2/CH4 discharge is generated. A discharge voltage of 18 kV with a current of 15 mA provided a specific energy input of 11 MJ/mol for a mixture of 70% CH4 and 30% CO2 (flow 2 ln/h, 19 mbar pressure). The absorbance features of the detected products are presented in Fig. 6. Here, a broad absorbance feature is visible in the 1700–1800 cm− 1 wavenumber region. In general, this indicates the presence of molecular species with large number of closely spaced rotational transitions in the vibrational band, which cannot be resolved spectroscopically at this pressure and temperature. Using the PNNL database, this specific absorbance profile was found to be likely belonging to acetone (C3H6O) and acetaldehyde (C2H4O), both with a concentration of 0.3 ± 0.1%. The concentration could however not be determined exactly, as PNNL is constructed with experimental data at 1 atmosphere pressure, which is significantly different from our experiment. We confirmed the presence of these two molecular species in our mixture using proton-transfer-reaction mass spectrometry (PTR-MS)39 and gas chromatography–mass spectrometry (GC-MS). Moreover, the residual shows the ability of the system to detect overlapping absorbance features of multiple species, as both the acetone, acetaldehyde, formaldehyde (H2CO) and most H2O spectral features are fitted well with their reference spectra. Furthermore, the absorbance lines of ethylene (C2H4) around 1880 cm− 1 are not included in the HITRAN database. Therefore, the GEISA database40 was used to create a simulated reference spectrum for C2H4, indicating a concentration of 0.73 ± 0.03 %
In summary, this experiment demonstrates the system’s ability to accurately detect numerous molecular species created in electrical discharges of CO2/CH4, even for the ones with overlapping spectral features.
Plasma with varying ratio of CO 2 and CH4. To demonstrate the possibilities of the system for plasma analysis and study, a quantitative analysis of the products formed in the electrical discharge is made, using a series of measurements with a varying CO2/CH4 ratio (gas flow 2 ln/h, 18kV, 15 mA, specific energy input of 11 MJ/mol). In Fig. 7a, the measured CO concentration is displayed for varying mixtures of CO2/CH4, along with the assumed limit in CO production, calculated from the number of carbon (in red) and oxygen atoms (in blue) available in the system to react to CO. These limits are calculated from the retrieved CO2 concentrations, which are shown in Supplementary Fig. 1. From the difference between the CO2 concentrations with the discharge on or off, the maximum number of carbon and oxygen atoms available for conversion is determined. The highest CO production is found for a 50/50-mixture of CO2/CH4. The CO values are within the expected maximum available carbon and oxygen atoms generated in the plasma, indicating a correct reaction balance. As the concentration of CO (~ 53%) is higher than the concentration of converted CO2, this indicates that CO is not only formed by removal of an oxygen atom from CO2, but also by recombination of the removed oxygen atom from CO2 with a dehydrogenated carbon atom from CH4. Furthermore, there is additional production of C2H4 and H2CO (Fig. 6), of which the retrieved concentrations are presented in Fig. 7b.