Non-Thermal Ammonia Decomposition for Hydrogen Production over Carbon Films under Low-Temperature Plasma—In-Situ FTIR Studies

Due to easy storage and transportation, liquid hydrogen carriers will play a significant role in diversifying the energy supply pathways by transporting hydrogen on a large scale. Thus, in this study, amorphous carbonaceous materials have been employed for hydrogen production via ammonia decomposition under non-thermal plasma (NTP) conditions. The adsorption and splitting of ammonia over carbons differing in the chemical structure of surface functional groups have been investigated by in situ spectral studies directly under NTP conditions. As a result of NH3 physical and chemical sorption, surface species in the form of ammonium salts, amide and imide structures decompose immediately after switching on the plasma environment, and new functionalities are formed. Carbon catalysts are very active for NH3 splitting. The determined selectivity to H2 is close to 100% on N-doped carbon material. The data obtained indicate that the tested materials possess excellent catalytic ability for economical, COx-free hydrogen production from NH3 at a low temperature.


Introduction
The growth of technological development has caused a huge increase in energy demands. People have used fossil fuels such as coal, natural gas, or petroleum for ages. Now we know that burning them contributes to environmental degradation through the emission of greenhouse gases, especially CO 2 [1]. The increase in ecological awareness has caused a constantly growing need for alternative and renewable energy resources [2]. Solar panels and windmills are extremely favorable from the environmental point of view, but they need specified conditions to achieve good efficiency. Because of that, the question arises of how to obtain green energy to satisfy our needs. The answer is hydrogen, especially green hydrogen [3].
Nowadays, hydrogen production is no longer a problem; it is possible with water electrolysis. This was explored by Liu et al. [4] who investigated kinetics and oxidation pathways using Fe 3+ as a catalyst. Literature data show that water electrolysis is the most preferred method of producing green hydrogen [5,6]. However, the main problem of hydrogen use is its distribution, which requires a lot of energy. There are a few different ways to store and transport hydrogen. One way is by compressing hydrogen gas, but it is energy inefficient. The second way is a carriage of liquefied H 2 , but it requires an extremely low temperature −250 • C. Recently, significant attention has been paid to obtaining hydrogen from ammonia decomposition, the third method to store H 2 . It is the most advantageous way, as mentioned above.
Pinzón et al. [7] obtained CO x -free hydrogen by catalytic ammonia decomposition using Ru supported on reduced graphene oxide (rGO). Ruthenium is often used as a catalyst; nevertheless, its cost makes the process unprofitable. Bell et al. [8] developed a Int. J. Mol. Sci. 2022, 23, 9638 2 of 10 cheaper catalyst which was cobalt deposited on γ-Al 2 O 3 . They investigated the dependence between cobalt particle size and catalytic activity.
One of the most commonly used catalysts utilized in the ammonia decomposition are iron and cobalt (e.g., [9]). Unfortunately these metals tend to sinter thus need a promoter to maintain high activity. In this case, the authors [9] used lanthanum oxides.
To our knowledge, only a few studies refer to ammonia decomposition by non-thermal plasma (NTP). Lu et al. [10] investigated the influence of dielectric barrier discharge plasma on the hidden active phase. They compared thermal catalytic ammonia decomposition and splitting in plasma. El-Shafie et al. [11] obtained hydrogen by using plasma in a plasma membrane reactor. The authors applied a palladium-copper membrane to separate hydrogen. This work focused on the research of energy and exergy of this process. El-Shafie et al. [12] explored alumina particles as a catalyst in the ammonia decomposition process by a dielectric barrier discharge reactor. Akiyama et al. [13] obtained pure hydrogen without by-products by using non-thermal, atmospheric-pressure plasma. They applied different types of electrodes.
Bearing in mind (i) high stability under the H 2 atmosphere, (ii) high surface area, (iii) welldefined pore structure, (iv) moderate N-affinity, we decided to look closer at the possibility of using the carbonaceous materials for NH 3 splitting under the NTP environment.
The present study aimed to prove the catalytic ability of carbon materials. In this work, for the first time, the splitting of ammonia over carbons differing in the chemical structure of surface functional groups has been investigated by in situ FTIR spectroscopy. As a result of carbon surface exposition to the NH 3 atmosphere, the physically and chemically adsorbed ammonia molecules are detected on the surface of carbon films. The evolution of these functionalities under NTP conditions is observed.

Carbon Films Preparation
The infrared spectrum of C des -carbon sample outgassed at 800 • C (Figure 1a) shows the presence of two main groups of signals sharp at ca. 1600 1/cm and mutually overlapping in the range 1100-1500 1/cm. These signals are due to C=C and C-C stretching modes that are weakly active in the IR because of the breakdown of selection rules [14]. Moreover, some C-H stretching vibrations can be detected at 3059 1/cm-a characteristic region of unsaturated and aromatic units. Note that there are no aliphatic functionalities in the carbon structure.
Doping nitrogen (5.31 m/m% based on elemental analysis) to the material (CdesN) does not drastically change the carbon FTIR spectrum (Figure 1b). Only the decrease in the background level and a shift of continuous absorption limit in the direction of shorter waves is observed due to the electrodonating character of newly formed surface species.
Further blue-shift of continuous absorption limit is observed for the carbon sample after oxygen doping. Additionally, the presence of absorption bands at 1830, 1760, and 1200, 1/cm for this sample (Figure 1c) indicates that some acidic surface groups are cyclic anhydrides. These surface structures were discussed previously, e.g., [15][16][17][18].
The spectral changes caused by ammonia adsorption (Figure 1a'-c') on these materials are negligible; thus they are much more visible as the differential spectra and will be discussed in the following sections.

In-Situ Plasma-Assisted Catalytic NH3 Conversion Investigations
The FTIR spectra of the gaseous products formed as a result of the NH3 decomposition over tested carbon films (30 min in static conditions) at 1 kW/m are shown in Figure  2. The NH3 conversion, as was expected, is NTP power-dependent. The higher power, the higher conversion close to 100% was reached for both high temperature-treated carbons (Cdes and CdesN). The differences between the carbons were observed for these samples under lower NTP power. The highest catalytic activity was determined for the nitrogendoped sample (CdesN). It is important to note that this material is characterized by the highest H2 selectivity (over 98%). Only ca. 40 ppm of CH4 and no C2H2 were detected after 30 min reaction under the most extreme conditions. The presence of surface oxygen functionalities causes the lower activity and selectivity of oxidized material (Cox). It will be discussed in the following paragraphs.
Bearing in mind the high resistance of carbons to the reducing atmosphere (including H2) it wasn't a surprise that both treated at 600 °C catalysts (Cdes and CdesN) show no structural changes caused via NH3 splitting reaction. Raman spectra, as well as high-resolution SEM pictures, remain unchanged, proving no effect of the process on sample morphology ( Figure 3).
In contrast, for the oxidized material (Cox), the intensity of the D-band slightly rises as the G-band decreases in intensity. Also, the SEM analysis reveals that the reaction somewhat destroyed the surface. Thus it seems reasonable to look closer at the overall mechanism.

In-Situ Plasma-Assisted Catalytic NH 3 Conversion Investigations
The FTIR spectra of the gaseous products formed as a result of the NH 3 decomposition over tested carbon films (30 min in static conditions) at 1 kW/m are shown in Figure 2. The NH 3 conversion, as was expected, is NTP power-dependent. The higher power, the higher conversion close to 100% was reached for both high temperature-treated carbons (C des and C des N). The differences between the carbons were observed for these samples under lower NTP power. The highest catalytic activity was determined for the nitrogen-doped sample (C des N). It is important to note that this material is characterized by the highest H 2 selectivity (over 98%). Only ca. 40 ppm of CH 4 and no C 2 H 2 were detected after 30 min reaction under the most extreme conditions.  The presence of surface oxygen functionalities causes the lower activity and selectivity of oxidized material (C ox ). It will be discussed in the following paragraphs.
Bearing in mind the high resistance of carbons to the reducing atmosphere (including H 2 ) it wasn't a surprise that both treated at 600 • C catalysts (C des and C des N) show no structural changes caused via NH 3 splitting reaction. Raman spectra, as well as highresolution SEM pictures, remain unchanged, proving no effect of the process on sample morphology ( Figure 3).  While ammonia is adsorbed on carbonaceous material, some characteristic surface species are formed. As a result of NH3 adsorption ( Figure 4, red spectrum), broad, mutually overlapped IR bands in the region of OH and NH stretching vibrations are formed at 3456, 3381, 3207, 3031 1/cm. The shape (wideness) of these bands and the presence of a signal at 2800 1/cm indicate the formation of hydrogen bonds. Their origin may be both: (i) hydrogen bonds between adsorbed NH3 molecules and oxygen-functional groups of Cox, or (ii) adsorbate-adsorbate-type interactions between physically adsorbed NH3 and previously chemisorbed molecules.
The spectral changes in the region of C=O stretching vibrations are caused by ammonia chemisorption on the oxidized carbon surface. The disappearance of band doublet at 1841 and 1753 1/cm is due to the reaction between NH3 and surface anhydrides. One of In contrast, for the oxidized material (C ox ), the intensity of the D-band slightly rises as the G-band decreases in intensity. Also, the SEM analysis reveals that the reaction somewhat destroyed the surface. Thus it seems reasonable to look closer at the overall mechanism.
While ammonia is adsorbed on carbonaceous material, some characteristic surface species are formed. As a result of NH 3 adsorption ( Figure 4, red spectrum), broad, mutually overlapped IR bands in the region of OH and NH stretching vibrations are formed at 3456, 3381, 3207, 3031 1/cm. The shape (wideness) of these bands and the presence of a signal at 2800 1/cm indicate the formation of hydrogen bonds. Their origin may be both: (i) hydrogen bonds between adsorbed NH 3 molecules and oxygen-functional groups of C ox , or (ii) adsorbate-adsorbate-type interactions between physically adsorbed NH 3 and previously chemisorbed molecules.
The spectral changes in the region of C=O stretching vibrations are caused by ammonia chemisorption on the oxidized carbon surface. The disappearance of band doublet at 1841 and 1753 1/cm is due to the reaction between NH 3 and surface anhydrides. One of the products thus formed is surface ammonium salts of carboxylic acids. In their IR spectrum, an intense band at 1471 1/cm (δ s of NH 4 + ion) and bands at 1551 1/cm (ν as of COO − ), and 1397 1/cm (ν s of COO − ) are formed. Such observations suggest that adsorption on Brönsted acid sites becomes dominant on this surface. However, the presence of the signal at 1655 and ca. 1300 1/cm (visible as a right shoulder of 1397 1/cm band) suggests that some Lewis sites are also present on the C ox surface. The IR band at 1107 1/cm can be attributed to physically adsorbed NH 3 . the products thus formed is surface ammonium salts of carboxylic acids. In their IR spectrum, an intense band at 1471 1/cm (δs of NH4 + ion) and bands at 1551 1/cm (νas of COO − ), and 1397 1/cm (νs of COO − ) are formed. Such observations suggest that adsorption on Brönsted acid sites becomes dominant on this surface. However, the presence of the signal at 1655 and ca. 1300 1/cm (visible as a right shoulder of 1397 1/cm band) suggests that some Lewis sites are also present on the Cox surface. The IR band at 1107 1/cm can be attributed to physically adsorbed NH3. When NTP is turned on, the IR bands of structures mentioned above disappear immediately, indicating that the NH3 decomposition occurs effectively.
In the NH stretching region, the IR peaks at 3456 and 3381 1/cm of -NH2 structures remain stable. A new signal at 3528 1/cm appears. In the fingerprint spectral region, the intensity of the negative doublet bands decreases, meaning that surface anhydrides are at least partially rebuilt. Interestingly, new IR signals appear at 1695 and 1220 1/cm. Such observations indicate that surface amines (C-NH2) structures are formed.
A somewhat different IR spectrum was registered when NH3 was adsorbed on outgassed at 600 °C carbon film ( Figure 5-red spectrum). In this case, besides the physical adsorption (1096 1/cm), the interactions with Lewis acid centers dominate (1588, 1260 1/cm) combined with strong hydrogen bonds-a broad band in the range of 3300-2400 1/cm.
Very interesting seems to be the presence of the IR bands below 900 1/cm. While the signal at 867 1/cm can be attributed to NH3 δ-the others at 661 and below 370 1/cm prove the formation of the secondary and tertiary amines, due to dissociative adsorption of NH3 on a positively charged surface.
After switching on the NTP, the physically adsorbed NH3 molecules (1096 1/cm) decompose while the bands attributed to adsorption on Lewis sites (1553 and 1201 1/cm) increase in intensity. Similarly, the 661 and 425 1/cm bands rise in intensity, too. Note that the signal attributed to C=C stretching mode at 1600 1/cm decreased in intensity.
Similar spectral changes confirming dissociative NH3 adsorption and splitting under NTP conditions were observed during the reaction over CdesN surface ( Figure 6). Some When NTP is turned on, the IR bands of structures mentioned above disappear immediately, indicating that the NH 3 decomposition occurs effectively.
In the NH stretching region, the IR peaks at 3456 and 3381 1/cm of -NH 2 structures remain stable. A new signal at 3528 1/cm appears. In the fingerprint spectral region, the intensity of the negative doublet bands decreases, meaning that surface anhydrides are at least partially rebuilt. Interestingly, new IR signals appear at 1695 and 1220 1/cm. Such observations indicate that surface amines (C-NH 2 ) structures are formed.
A somewhat different IR spectrum was registered when NH 3 was adsorbed on outgassed at 600 • C carbon film ( Figure 5-red spectrum). In this case, besides the physical adsorption (1096 1/cm), the interactions with Lewis acid centers dominate (1588, 1260 1/cm) combined with strong hydrogen bonds-a broad band in the range of 3300-2400 1/cm. Very interesting seems to be the presence of the IR bands below 900 1/cm. While the signal at 867 1/cm can be attributed to NH 3 δthe others at 661 and below 370 1/cm prove the formation of the secondary and tertiary amines, due to dissociative adsorption of NH 3 on a positively charged surface.
After switching on the NTP, the physically adsorbed NH 3 molecules (1096 1/cm) decompose while the bands attributed to adsorption on Lewis sites (1553 and 1201 1/cm) increase in intensity. Similarly, the 661 and 425 1/cm bands rise in intensity, too. Note that the signal attributed to C=C stretching mode at 1600 1/cm decreased in intensity.
Similar spectral changes confirming dissociative NH 3 adsorption and splitting under NTP conditions were observed during the reaction over C des N surface ( Figure 6). Some differences arising from the other surface composition caused by N-doping were observed in the fingerprint region-below 900 1/cm. differences arising from the other surface composition caused by N-doping were observed in the fingerprint region-below 900 1/cm.

Discussion
Due to easy storage and transportation, liquid hydrogen carriers will play a significant role in diversifying the energy supply corridor by transporting hydrogen on a large

Discussion
Due to easy storage and transportation, liquid hydrogen carriers will play a significant role in diversifying the energy supply corridor by transporting hydrogen on a large Figure 6. FTIR differential spectrum of C des N carbon film after NH 3 adsorption at 25 • C (red spectrum-0 min) and time-resolved spectra when NTP is switched on from 0 to 30 min.

Discussion
Due to easy storage and transportation, liquid hydrogen carriers will play a significant role in diversifying the energy supply corridor by transporting hydrogen on a large scale. Low or even no-carbon footprint is the crucial factor for their application. Of all the liquid hydrogen carriers, ammonia has proven to be a carbon-free and high-energy density sustainable candidate. The NH 3 splitting has the advantages of one-way transport.
Produced in this way, N 2 does not need to be directly recovered and recycled after the dehydrogenation step.
In this work, we proved that ammonia cracking might generate environment-friendly nitrogen and hydrogen not only on the metal-based catalysts, as was stated previously by Wei et al. [19] and the group of Ohtsuka [20]. It has been reported that the Fe-catalyzed NH 3 decomposition may proceed through a mechanism involving the N-containing intermediate species. We showed here that a similar mechanism is also possible on carbon catalysts but contrary to metallic systems reaction leads here on the surface maintaining the carbon structure (Figures 3-6).
We proved that effective NH 3 splitting on the carbon surface was observed on both high-temperature treated materials. This reaction was examined in the literature over Fe and Co containing active sites [10,21,22]. However no work was done for pure carbons, nor for NTP conditions.
Outgassed carbon surfaces (C des and C des N) played the role of nitrogen scavengers. As a consequence, functionalities similar to carbon nitrides were formed. Interestingly, doping oxygen hampered the overall process, most probably via competition of O and N connected to the active center. Thus, a bare or N-containing carbon surface seemed ideal for the dissociative adsorption of NH 3 under NPT conditions. Once the NH 3 decomposed, the isolated N or H adsorption became important. The properties of the carbon surface made the interaction of N much stronger than that of H with C. Nitrogen atoms tended to occupy the active sites by preference, which could lead to catalytic deactivation. Nevertheless, such an effect was not observed in this study; the selectivity on the C des N catalyst was as high as 98%. However, further examination of the circulation of N "on" and "in" the C-surface needs to be carried out.

Carbon Films Synthesis
Carbon films utilized in this study were prepared from cellulose. The tested materials shows the 2-dimensional nature of, i.e., unlimited length and width (in this study, 35 × 15 mm) and a thickness of about 20 µm (Figure 1). The charring experiments were set up as follows: cellulose films were carbonized at 600 • C for 1 h under a dynamic vacuum. These samples were labeled-C des . To prepare an oxidized carbon film (C ox ), the C des samples were exposed to 100 kPa of O 2 at 300 • C for 1 h and then evacuated at 200 • C. The nitrogen-doped carbons (C des N) were prepared from C ox materials by annealing in an NH 3 atmosphere (100 kPa) at 600 • C for 1 h. For catalytic tests, the powdered carbon samples were prepared similarly.

Materials Characterization
The full characteristics of carbon films, including elemental and thermogravimetric analysis, low-temperature N 2 adsorption, 13 C NMR, and structure modeling, were presented recently [23][24][25]. The results indicated that carbon films are microporous solids with a bimodal pore size distribution. The pores are grouped around two main diameters. The first diameter approaches 0.5-0.6 nm, and the second diameter occurs in the range 1.3-1.5 nm. Thus, the studied carbon films possess interesting features, i.e., nearly homogeneous microporosity, that gives the carbon films a molecular sieve character.
The surface area and structure of carbon films contribute to the catalytic activity only if they exert some control on the accessibility of reactants towards the active centers. There is no enormous difference between the pore size distribution of doped and non-oxidized carbon samples [26]. Thus we could treat the carbon films as a microporous material with similar porosity and focus more on the surface chemistry instead of micropore structure.

In-Situ Plasma-Assisted Catalytic NH 3 Conversion Investigations
The IR spectroscopic studies were carried out in a vacuum cell described previously [20,[26][27][28], and plugged into the vacuum line (Figure 7). For the in situ IR study, ammonia (constant pressure 1.38 Torr) was exposed to a carbon film at 1 kW/m normalized power NTP atmosphere. The spectra were collected during a 30 min reaction. Typically, 32 scans were collected with an instrument resolution 4 1/cm. The IR cell was made of quartz with KBr windows. The cell construction enabled the thermal treatment of the carbon film up to 1000 • C in any controlled atmosphere or in a vacuum.

In-Situ Plasma-Assisted Catalytic NH3 Conversion Investigations
The IR spectroscopic studies were carried out in a vacuum cell described previously [20,[26][27][28], and plugged into the vacuum line (Figure 7). For the in situ IR study, ammonia (constant pressure 1.38 Torr) was exposed to a carbon film at 1 kW/m normalized power NTP atmosphere. The spectra were collected during a 30 min reaction. Typically, 32 scans were collected with an instrument resolution 4 1/cm. The IR cell was made of quartz with KBr windows. The cell construction enabled the thermal treatment of the carbon film up to 1000 °C in any controlled atmosphere or in a vacuum. The composition of the gas phase in contact with the catalysts was also monitored by Mattson Genesis II FTIR spectrophotometer. A carbon film sample in the IR cell could be moved in and out of the infrared beam. Single beam spectra were recorded with the carbon film sample either in the path of the infrared beam or withdrawn from the beam, as well as with and without the gases in the cell. Subtracting the spectrum of the gas phase (recorded within 15 s) from that of the combined gas and surface species gives the spectrum due to surface species adsorbed on carbon samples.
The other important element of the system is the glow discharge (GD) reactor operating under 10 kV (30 mA) and 30 kHz conditions. The distance between electrodes was 30 cm. The effluent gases (NH3 and CxHy) were analyzed quantitatively using a multiplepass gas cell (with the pathlength of 2 m) attached online to the flow system (Figure 7). The selectivity was calculated from relation H2/(H2 + ΣCxHy). Note that no carbon oxides were detected within all the experiments. The composition of the gas phase in contact with the catalysts was also monitored by Mattson Genesis II FTIR spectrophotometer. A carbon film sample in the IR cell could be moved in and out of the infrared beam. Single beam spectra were recorded with the carbon film sample either in the path of the infrared beam or withdrawn from the beam, as well as with and without the gases in the cell. Subtracting the spectrum of the gas phase (recorded within 15 s) from that of the combined gas and surface species gives the spectrum due to surface species adsorbed on carbon samples.
The other important element of the system is the glow discharge (GD) reactor operating under 10 kV (30 mA) and 30 kHz conditions. The distance between electrodes was 30 cm. The effluent gases (NH 3 and C x H y ) were analyzed quantitatively using a multiple-pass gas cell (with the pathlength of 2 m) attached online to the flow system (Figure 7). The selectivity was calculated from relation H 2 /(H 2 + ΣC x H y ). Note that no carbon oxides were detected within all the experiments.

Conclusions
In this work, for the first time, we examined a promising method of obtaining pure hydrogen by ammonia decomposition in a non-thermal plasma environment over carbon catalysts. We showed the direct experimental evidence that carbon works as an N scavenger, but the bonding was not so efficient due to the low electronegativity difference.
The process was performed only at the surface; we did not interfere with the Cstructure. Based on the presented results, one could expect that the used catalysts were extremely stable under NTP conditions, although further studies are needed.
Carbons can be harnessed effectively as catalysts for hydrogen production. The data obtained indicate that the tested materials possess excellent catalytic ability for economical, CO x -free hydrogen production from NH 3 decomposition at a low temperature.