On application of a sufficiently high voltage to the SBD electrode (Fig. 1a), gas breakdown was observed to occur along the edges of the grounded electrode, which spread across the dielectric surface. OES was used to identify excited states within the discharge (Fig. 1b) and showed the characteristic emission bands of the Nitrogen second positive system. No emission was observed at 777 nm and 844 nm, indicating an absence of excited atomic oxygen. Using high-resolution OES, the rotational and vibrational temperature of the discharge was assessed via comparison with simulated data at Trot = 345 K ± 35 K and Tvib = 2700 K ± 200 K, which is in line with previously reported studies employing air CAP generated using the SBD configuration [23].

Highly reactive species primarily created through electron driven dissociation (e.g., N, O, H and OH) diffuse away from the discharge layer and rapidly react to form a variety of longer-lived RONS, including O3, H2O2, NO, and other NxOy molecules [24]. FTIR gas phase analysis (Fig. 1c) revealed the composition of the gas phase effluent under low- and high-power modes of operation. Results indicated that under low-power conditions, O3 was the primary long-lived product, formed through a reaction involving atomic and molecular oxygen (Table 1—reaction R1) [24]. In contrast, under high-power conditions, the gas phase effluent was dominated by NxOy species, such as N2O and NO2.

The transition between O3-dominated to NxOy-dominated RONS composition occurs via a multi-step process and has been widely reported in the literature [17]. First, as the discharge power increases, the rate of thermal decomposition of O3 increases, resulting in less O3. Second, an increased power also results in more NO, which inhibits O3 production via direct decomposition to O2 and NO2, and through the consumption of O, a key precursor for O3 formation, described in reactions R2–R4 (Table 1).

Most of the absorption peaks shown in Fig. 1c can be directly linked to RONS widely accepted to be generated by CAP created in ambient air under different power regimes [16]. The exception is the significant absorption peak observed at ~ 1380 cm−1 (shaded grey), which does not relate to known RONS produced by CAP. Following exposure to CAP generated RONS, the absorption peak at 1380 cm−1 remains indefinitely (Fig. 1d) despite flushing the cell with filtered air for many minutes following exposure. Also shown in Fig. 1d is the infrared spectrum of KNO3 retrieved from the NIST spectral database [25], providing a clear indication that the unexpected peak is a likely result of interactions between plasma generated RONS and the KBr windows of the gas cell. Notably, further exposure of the KBr windows to RONS leads to further non-linear growth of the 1380 cm−1 absorption peak (Fig. 1e), meaning that the peak cannot easily be removed using background subtraction methods.

To understand the origin of the 1380 cm−1 absorption peak observed in Fig. 1, multiple surface analysis techniques were used. XPS was adopted to determine compositional changes to the surface of exposed KBr (Fig. 2). The control sample shows the typical spin–orbit splitting K2p (293 eV) and Br3d (69 eV) with an atomic concentration ratio close to 1:1 (Table 2). Moreover, carbon (C1s—284.5 eV) and oxygen (O1s—532 eV) peaks were detected, indicating contamination due to absorption of impurities from the ambient air [26]. No nitrogen was found on the freshly prepared KBr pellet (Fig. 2a). In contrast, the samples exposed to low-power plasma (Fig. 2b) and high-power plasma (Fig. 2c) for the same duration (480 s) show a clearly visible nitrogen N1s peak (407 eV) and a slightly increased atomic oxygen content. Interestingly, the K/Br atomic ratio of the plasma-exposed samples deviates from that of the control even when the standard error of the instrument is taken into account (Table 2). This could indicate that the anion sublattice is affected by RONS, partially substituted and the surface chemistry of the KBr ionic crystal is changed. In particular, the plasma treatment led to a binding of nitrogen and oxygen on the surface, which was further investigated by deconvolution of the XPS peaks.

XPS survey of a untreated KBr sample, b treated for 480 s with low-power CAP, and c treated for 480 s with high-power CAP

The core levels were deconvoluted using the Voigt function to decipher changes in the molecular environments of the Br3d and K2p peaks. In the case of Br (control sample, Fig. 3a), two peaks describing the 3d5/2 and 3d3/2 orbital splitting (Δ ≈ 1.0 eV and area ratio 2:3) were found at the typical positions 68.7 eV and 69.74 eV, respectively [27]. Keeping their positions fixed and taking into account the peak area ratio, the two other exposed samples were investigated (Fig. 3b and c). It can be suggested that the fitted peaks became wider and their intersection is larger when compared with the control sample. These changes could be related to the additional number of chemical bonds contributing to the Br3d peak profile.

High-resolution XPS spectra showing Voigt fits of the Br3d core levels for a untreated KBr, b treated for 480 s with low-power CAP and c treated for 480 s with high-power CAP

Similarly, the XPS profile of the K2p core level (Δ ≈ 2.8 eV and area ratio 1:2) shows both the K2p3/2 and K2p1/2 components of treated samples to be wider compared to the untreated sample (Fig. 4a), this again suggests a change in surface chemistry affecting potassium. Furthermore, this assumption could be supported by another observation. Due to the broadening of the peaks, it was not possible to adequately fit the K2p profile using two peaks when strict conditions for position and K2p3/2/K2p1/2 ratio were met. In both cases, namely, low-power plasma (Fig. 4b) and high-power plasma (Fig. 4c), the peak area ratio tends to be different from that of the control sample (K2p3/2/K2p1/2 = 2). At low power, the ratio is 2.45, and at high power it is 1.49. These changes are likely due to the interaction of plasma generated RONS with the KBr surface, leading to the formation of new chemical bonds and the modification of existing ones.

High-resolution XPS spectra showing Voigt fits of the K2p core levels for a untreated KBr, b treated for 480 s with low-power CAP and c treated for 480 s with high-power CAP

To investigate the issue further, O1s and N1s were studied for a possible positional cross-check of the NOx bonds. Prior to analysis, the nitrogen and oxygen regions must first be discussed. In the case of the N1s peak (the normalisation is adjusted to the highest peak), which is not present in the control sample even after a high-resolution scan, the XPS analysis showed a clear contribution of the asymmetric peak at about 407 eV in both cases for low and high-power plasma exposure. However, their asymmetry differs, indicating a possible contribution from more than one oxidation state of nitrogen, especially in the case of the ‘low-P’ plasma (Fig. 5a). As for oxygen (the normalisation is set to the highest peak), which was already present in the control sample because it was exposed to ambient conditions, it is more likely that a broad O1s peak occurs at about 532 eV (Fig. 5b, grey data) due to chemical bonding to carbon (which is also found in all samples and does not differ much, see inset in Fig. 6). After contact with the plasma effluent, the shape of the peak changes as well as its intensity, first broadening (red data in Fig. 5b) and then increasing. This behaviour could be correlated with the changes in the N1s peak, suggesting that the O1s peak is altered by RONS. However, to answer the question about the nature and proportion of the NOx components, it is sufficient to fit only the N1s profile (the C1s peak has no influence on the profile shape).

High-resolution XPS spectra representing a N1s and b O1s core levels for untreated KBr (grey), treated with low-power CAP (red) and treated with high-power CAP (blue)

High-resolution XPS spectra representing N1s core levels for a treated with low-power CAP and b treated with high-power CAP. Inset is a normalised carbon C1s spectra (shift-corrected to 285 eV)

After fitting, it is clear that in both cases the contribution of NO3− (407.2 eV) dominates (Fig. 6a and b), with an additional portion of NO2 (405.7 eV) at the surface [28, 29]. The NO3−/NO2 ratio for the low-power plasma sample is roughly equal to 1.3 and for the high-power plasma it is much higher, being 10.8. The apparent presence of the NO3 group and the reduced proportion of Br may indicate that the latter has been partially replaced by a nitrate group and that KNO3 is formed on the KBr surface after plasma exposure. These changes are likely to be reflected in the transparency of KBr within the mid-range IR spectrum, significantly affecting the properties and performance of the treated material. Thus, these observations are important to explain the origin of unexplained peak during CAP species analysis.

KBr pelletes subjected to a 480 s exposure of CAP showed the formation of several absorption peaks, in clear contrast to the untreated KBr spectra, with the most significant features being found in the 1300–1500 cm−1 region (Fig. 7a), which are accompanied by the appearance of additional smaller peaks spanning the 800–1800 cm−1 spectral region. As shown in Fig. 1, this region is typically used to identify key CAP generated reactive species, including O3, NO2, HNO3 and HNO2. Comparing Fig. 7a with the data available in the literature indicates the shape of the two modified curves (red and blue) are reminiscent of the NO3 anion with 14N and 15N isotopes, whose IR modes also populate a similar wavelength span [30,31,32,33,34].

a Wide-range FTIR data of the plasma exposed KBr pellets, time-resolved ATR-FTIR comparison of treated KBr sample with b low plasma power and c high plasma power

Since the grey shaded areas are the most affected, they will be the focus of further discussion. The time evolution of this portion of the spectrum (Fig. 7b-c) is quite complex, since the mutual contribution of several modes can be observed regardless of plasma power. Importantly, after prolonged treatment, the low-power (low_P) sample showed the most prominent peak at 1352 cm−1 and the largest mode in the high-power (high_P) sample was at 1380 cm−1. To study them in detail, the spectra obtained in both plasma regimes were deconvoluted with the Gaussian function.

The interval of interest (1500–1300 cm−1, NO3-nitrate anion) includes several interacting IR-active modes with different intensities, most of which are attributed to the vibrational frequencies of either the symmetric stretching of 14N16O3− (⁓ 1380 cm−1) and 15N16O3− (⁓ 1350 cm−1), or HON bending in HNO3 (Fig. 8a and b) [33,34,35,36,37]. Through detailed examination of the two peaks, it could be suggested that under high-power conditions at the beginning of the plasma treatment, the IR spectrum of the NO3− anion is overlaid by the contribution of HNO3 and 15N16O3− which is dominant up to 120 s of exposure. After that and until the end of the treatment, the contribution of the isotope 14N16O3− (⁓ 1380 cm−1) in KNO3 is the strongest. The opposite occurs under low-power conditions, where HNO3 and 15N16O3− overtake the 14N16O3− (⁓ 1380 cm−1) peak under long exposure. In both cases it can be confiremed that the KBr surface is functionalised with a cocktail of different N-containing functional groups, which also leads to the formation of  KNO3, which is confirmed via Raman analysis in the following section. The small deviation in peak position, observed by comparing the low-power and high-power spectra, could be attributed to the influence of more complex isotopes, including 15N16O318O− in the case of the lower standing peak (green shaded area in Fig. 8a and b), and 14N16O318O− for the higher standing peak (grey shaded area in Fig. 8a and b) [31]. It is important to note that the origins of the observed stable nitrogen isotopes are assumed to be from the samples being exposed to ambient air where there is an abundance of 14N plus a small amount of 15N present [38]. The isotopic nitrogen ratio was further explored with ToF–SIMS.

Deconvoluted FTIR data of treated KBr sample with a low plasma power and b high plasma power at different lengths of exposure time

Looking at the FTIR data in general, it could be postulated that a longer exposure of the KBr window plasma causes functionalisation of the metal halide due to the dominance of the NO3− anion with a small contribution of NO2 stretching vibrations (peak 1250 cm−1 in Fig. 7a and XPS data for N1s in Fig. 6a). These anions could affect the vibrational properties of the KBr lattice modes at low THz frequencies which can be studied further using Raman spectroscopy.

In addition to XPS and FTIR, Raman spectroscopy was also performed on CAP-exposed KBr pellets (Fig. 9). Complementary to the mid-IR (i.e., FTIR) analysis, Raman provides an additional insight into the far-IR range where cation and anion vibrational features of KBr are accessible. In the case of a pure KBr crystal, the Raman spectrum is composed of a mix of low-frequency longitudinal and transverse lattice modes (up to 9 THz or 300 cm−1). The remaining spectrum, namely above 300 cm−1 is free of modes as expected. In contrast, both treated samples reveal a characteristic strong and narrow peak centred at about 1050 cm−1 that can be attributed to the NO3– nitrate anion stretching vibration [39]. This strong Raman mode and two additional peaks detected at 717 cm−1 and 1334 cm−1 in the high-power plasma treated sample are consistent with the previously published Raman data for KNO3 crystals and KNO2 structures [40,41,42,43,44].

Raman spectra of a untreated KBr sample, treated with low plasma power and high plasma power (480 s), and b accompanied by Gaussian fitting of the main NO3− mode

The most intense NO3− peak was fitted (Fig. 9b) and it can be seen that the peaks obtained under low- and high-power exposure are similar but not exactly the same, with the normalised modes differing slightly in terms of the position of their maximum intensity and the contribution of the fitting components. Under high-power conditions, the maximum intensity is at 1051 cm−1, which is 3 cm−1 lower than that observed in the lower power case. The Raman profile also has a more complex shape and is generally broader, as indicated by the contributing main and shoulder fit components. In particular, the shoulders are much larger in the high-power plasma case than in the low-power plasma case. Following data from Brooker et al. and other related articles published on the subject, this mode is attributed to Ag symmetry, with a contribution from B2g symmetry and is also sensitive to the presence of defects [40, 42,43,44]. Moreover, the low-frequency interval (< 300 cm−1) is also modified. According to a calculated spectrum for KBr using the two-phonon frequency Born and Bradburn approach, this region consists of a mutual combination of 6 acoustic and optical frequency modes [45, 46]. After careful consideration of the spectra available, one might suspect that the intensity interplay between two intense Raman features at 120–125 cm−1 (‘A1g–2Eg–T2g’ symmetry mode ⁓ 3.7 THz) and 210–215 cm−1 (‘A1g–Eg’ symmetry mode ⁓ 6.4 THz) is not the same since the plasma exposes the KBr surface[46]. A peak at the higher Raman shift (210–215 cm−1) becomes smaller, while the low-frequency peak at low power first becomes sharp and then splits, giving rise to a new component at 90 cm−1 previously detected in the theoretical and experimental Raman study of KNO3 in the sub-THz region [47]. These features indicate molecular dynamics changes in the K–Br and Br–Br interactions in the cation and anion sublattices and, together with the corresponding changes in the XPS and FTIR results, provide definitive evidence for the formation of KNO3 on the surface of KBr after plasma treatment.

Three samples were compared by ToF–SIMS analyses, control (green), low-power: 480s (blue), and high-power: 480s (red) (Fig. 10). The mass spectra of negative secondary ions indicate that plasma treatment resulted in a significant increase in nitrate (III) and (V) concentrations on the surface compared to the reference sample [48]. Besides, NO2− and NO3− signals are also present in intense cluster ions of potassium and nitrates such as KNO2−, KNO3−, KNO4−, KNO5−, KNO6−, KN2O4− KN2O5− and KN2O6− (Fig. 10a). The presence of so many different cluster secondary ions indicates that nitrates of different oxidation states are present in high concentrations. Isotopic ratios of 15NO2−/14NO2− and 15NO3−/14NO3− (between 0.005 and 0.017) are slightly higher than their natural abundancies (0.0034) [49]. Even greater differences can be observed for the ratios of N16O18O−/N16O2− and N16O218O−/N16O3− (between 0.009 and 0.022) compared to the natural ratio of 18O/16O of 0.0019. However, this is not due to the actual isotopic enrichment but it is rather caused by the SIMS measurements. Signals of heavier isotopes have much lower intensity so the noise of the background contributes to their intensity in relative terms much more than in case of intense signals of lighter and most abundant isotopes of N and O. Furthermore, there is also some effect of the overlapping of signals (isobaric interferences) from HNO2−, H2NO2−, HNO3− and H2NO3− secondary ions which have almost the same mass as nitrite and nitrate ions with heavier isotopes [50]. The occurrence of H-related fractions and heavier isotopes of oxygen (18O) might alter the FTIR spectra, smearing the profile due to possible vibrational modes overlapping. The ratio of NO2− and NO3− ions compared to the KBr species is slightly higher when plasma with lower power is used, but the differences are marginal. On the other hand, the nitrate layer on the surface is visibly thicker in the case of the treatment with the lower-power plasma (Fig. 10b). Due to the lack of an appropriate standard, it was not possible to determine the exact thickness of the nitrate layer, which is in cases of both plasma treatments between 20 and 40 nm. The control sample revealed a clean surface with negligible contamination (Fig. 10a). Considering both the FTIR and SIMS analyses, the following conclusions can be drawn: as expected, there is no isotopic separation effect caused by plasma exposure, the variations in the FTIR profile can be attributed to overlapping modes arising from a mixture of multiple species, including nitric and nitrous acids, at different durations of plasma discharge time.

a Survey spectra of negative secondary ions from the samples treated with high-power plasma (red), low-power plasma (blue) and untreated reference (green). Labelled are the main secondary ions on the m/z range from 40 to 210. ToF–SIMS depth profiles of b control sample, c low-power: 480 s and d high-power: 480 s. Depth profiles were also measured while analysing negative secondary ions