Thermal Oxidation Reaction between NH3 and O3: Low-temperature Formation of an NH4+ -bearing Salt

NH3 has long been predicted to be an important component of outer solar system bodies, yet detection of this compound suggests a low abundance or absence on many objects where it would be expected. Here, we demonstrate that a thermally driven oxidation reaction between ammonia (NH3) and ozone (O3) in a H2O + NH3 + O3 mixture may contribute to the low abundance of NH3 on some of these objects, as this reaction efficiently occurs at temperatures as low as 70 K. We determined the overall activation energy for this reaction to be 17 ± 2 kJ mol−1, which is consistent with other chemical systems that react at cryogenic temperatures. The loss of these two compounds coincides with the formation of NH4+ and NO3− at low temperatures, both of which are observable with infrared spectroscopy. Warming our H2O + NH3 + O3 mixtures through sublimation, we find a number of higher-temperature phases, such as ammonia hemihydrate, nitric acid, and ammonium nitrate (NH4NO3). The most stable of these is NH4NO3, which remains on the substrate until temperatures near 270 K. The salt product within this sample contains near-infrared spectral features between 2.0 and 2.22 μm, which is a spectral region of interest for several outer solar system objects, including the Uranian satellites Miranda, Ariel and Umbriel, and Pluto's satellite Charon.


Introduction
Ammonia (NH 3 ) has long been predicted to be an important component of outer solar system bodies (Lewis 1972;Stevenson 1982).This possibility is especially intriguing given that laboratory experiments have shown that it can form numerous hydrated states with H 2 O (Bertie & Morrison 1980;Kargel 1992), and these hydrates can suppress the freezing point of H 2 O by up to 100 K, possibly allowing liquid H 2 O to exist in much colder regions than if NH 3 were absent (Kargel 1992).However, through its near-infrared (NIR) absorption feature near 2.2 μm, observations of NH 3 suggest a low abundance on the surface of most objects where it has been detected.NH 3 , or possibly an ammonia hydrate or an ammonium ( + NH 4 )-bearing salt which also possess NIR absorptions near 2.2 μm, has been tentatively detected on the surfaces of Charon (Brown & Calvin 2000;Cook et al. 2007Cook et al. , 2023)), Pluto (Dalle Ore et al. 2019), several Uranian satellites, including Miranda, Ariel and Umbriel (Bauer et al. 2002;Cartwright et al. 2020Cartwright et al. , 2023)), Quaoar (Jewitt & Luu 2004), Orcus (Barucci et al. 2008), possibly Enceladus (Emery et al. 2005;Verbiscer et al. 2006;Hendrix et al. 2010), and within the clouds of Jupiter (Lewis 1969).This apparent low surface abundance has led to multiple laboratory studies investigating whether other processes or interactions with other surface species could affect the NH 3 surface abundance.
Ion irradiation studies have demonstrated the efficient destruction and sputtering of pure NH 3 and mixtures of NH 3 and H 2 O (Lanzerotti et al. 1984;Loeffler et al. 2010), and the radiolytic production of nitrogen-or N-bearing compounds.Loeffler et al. (2006) demonstrated that H + irradiation and the subsequent heating of H 2 O and NH 3 mixtures resulted in the formation and eruption of N 2 and H 2 bubbles, a possible smallscale source of the H 2 O vapor and nitrogen within the plumes detected on Enceladus during NASA's Cassini flybys (Hansen et al. 2006;Waite et al. 2006).Additionally, the irradiation of NH 3 and H 2 O mixtures results in the formation of increasingly complex N-bearing molecules including hydrazine (N 2 H 4 ), hydroxylamine (NH 2 OH; Zheng & Kaiser 2007), and + NH 4 (Moore et al. 2007), and results in the amorphization of crystalline ammonia hydrates (Moore et al. 2007).Using a 193 nm argon fluoride excimer laser, Loeffler & Baragiola (2010) demonstrated that while pure NH 3 ice can be destroyed by UV photolysis, NH 3 is relatively stable when it is photolyzed in mixtures with H 2 O, suggesting that ions and electrons are likely more efficient in destroying NH 3 on extraterrestrial surfaces.
While the destruction of NH 3 via energetic particles appears to be a viable explanation for the low abundance of NH 3 found on the surfaces of many outer solar system ices, laboratory studies have also demonstrated that NH 3 participates in a number of different thermally driven reactions that could also lower its abundance, including acid-base reactions (Schutte et al. 1999;Noble et al. 2013;Loeffler & Hudson 2015) and nucleophilic additions (Bossa et al. 2009a(Bossa et al. , 2009b)).For a review of possible thermally driven reactions for a number of astrochemically relevant species, see Theulé et al. (2013).Notably, recent work by Potapov et al. (2019) demonstrates that, in general, this reactivity in the absence of ionizing radiation is likely underestimated in laboratory work due to catalytic effects of dust grain Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
surfaces, which tend to be difficult to recreate experimentally.Regardless, through these thermal reactions NH 3 reacts to produce semi-volatile ammonium-bearing ( + NH 4 -bearing) salts, including but not limited to ammonium hydrosulfide or NH 4 SH (Loeffler et al. 2016) and ammonium acetate or NH 4 HCO 2 (Kruczkiewicz et al. 2021); these salts also have an absorption feature near 2.2 μm due to + NH 4 .Many of these + NH 4 -bearing salts remain stable under a wide range of thermal conditions (Kruczkiewicz et al. 2021).The formation of + NH 4 -bearing salts may explain the relative depletion of nitrogen in some cometary ices (Poch et al. 2020;Kruczkiewicz et al. 2021).Similarly ammoniated ( + NH 4 -) phyllosilicates also agree well with spectral features identified on warmer astronomical bodies, like the dwarf planet Ceres (King et al. 1992;Singh et al. 2021) and asteroid Ryugu (Pilorget et al. 2022).
In addition to the multiple reaction pathways mentioned above, it is also possible that NH 3 could undergo oxidation reactions, similar to what has been observed for simple sulfurbearing molecules expected on several outer solar system satellites (Loeffler et al. 2010;Loeffler & Hudson 2013, 2015;Mifsud et al. 2021;Tribbett & Loeffler 2022).The strong oxidants hydrogen peroxide (H 2 O 2 ) and ozone (O 3 ) are of specific interest, because they have been detected on a number of planetary bodies (Noll et al. 1996(Noll et al. , 1997;;Carlson et al. 1999;Hand & Brown 2013;Trumbo et al. 2019Trumbo et al. , 2023)).Oxidation reactions involving NH 3 have been observed in previous gasand liquid-phase studies, suggesting that O 3 oxidizes NH 3 to nitrate (NO - 3 ; Olszyna & Heicklen 1971;Singer & Zilli 1975).However, Loeffler & Hudson (2015) found that NH 3 and H 2 O 2 ice mixtures undergo a simple acid-base proton transfer.This difference suggests that NH 3 and O 3 should be revisited to further investigate the chemical reactivity of NH 3 .Here, we present evidence for a thermally driven oxidative reaction between NH 3 and O 3 at temperatures as low as 70 K in both H 2 O-dominated and NH 3 -dominated H 2 O + NH 3 + O 3 ice mixtures.We identify the reaction products using a combination of infrared (IR) spectroscopy and mass gravimetry, provide estimates for the overall activation energy of this reaction, and place our results in the context of extraterrestrial environments.

Experimental Methods
We performed all experiments at Northern Arizona University in an ultra-high vacuum (UHV) chamber with a base pressure of 10 −9 torr (∼10 −10 within the thermal radiation shield) that we have described in a recent publication (Tribbett & Loeffler 2022).To prepare ice samples, we codeposited H 2 O (high-performance liquid chromatography grade, Sigma Aldrich), NH 3 (99.99%purity, Matheson Gas), and O 3 from separate mixing lines onto an optically flat gold mirror electrode of a quartz crystal microbalance (QCM) at 50 K, where we estimate the absolute accuracy of the temperature to be within 0.5 K.We produced O 3 by sparking a glass manifold containing ∼750 torr of O 2 (99.999%,Matheson Gas) with a Tesla coil for ∼20 minutes.O 3 condenses into a portion of the manifold submerged in liquid nitrogen, and the residual O 2 was pumped out leaving the residual O 3 to be leaked into the UHV chamber.
We determined the relative abundances of molecules in our ice samples following the QCM technique described in Tribbett & Loeffler (2022).We estimate the thickness of our ice samples to be ∼1.1 μm, assuming a density of 0.82 g cm −3 for H 2 O (Westley et al. 1998), 1.65 g cm −3 for O 3 (Teolis et al. 2007a;Raut et al. 2011), and 0.68 g cm −3 for NH 3 (Hudson et al. 2022).We note that discrepancies exist in the literature regarding the density of amorphous NH 3 at low temperatures (Wood & Roux 1982;Satorre et al. 2013;Hudson et al. 2022), and we opt to use the most recent literature values available.
We also performed reference experiments using nitric acid (HNO 3 ).In those experiments, we deposited 15 M HNO 3 (70% v/v, Sigma Aldrich) from the glass manifold after degassing it with two liquid nitrogen freeze thaw cycles.Despite the concentration of the HNO 3 solution, we expect the vapor composition, and consequently the deposited ice composition, to be ∼1:1 based on previous laboratory work characterizing HNO 3 and H 2 O ice mixtures of varying molar ratios (Hanson & Mauersberger 1988;Ritzhaupt & Devlin 1991).For clarity, we summarize the compositions and other important parameters calculated for the ice mixtures used in this study in Table 1.
After deposition, we warmed our ice mixtures and monitored their evolution using a combination of IR spectroscopy and microbalance gravimetry.We note that the instrument details are described in depth in Tribbett & Loeffler (2022).We warmed our ice samples following one of two different heating treatments: linear heating and isothermal annealing.In the case of the linear heating experiments, initially we warmed our ice samples from 50 to 290 K at 1 K minute −1 to monitor for evidence of thermally driven reactions in the IR spectra.Once we established that we could clearly observe a reaction, we repeated the experiments using a significantly slower warming rate (0.1 K minute −1 ) to improve the temporal resolution in our QCM data.For the isothermal annealing experiments, we warmed our ice samples from 50 K to an annealing temperature between 75 and 90 K at 5 K minute −1 .After reaching the prescribed annealing temperature, we monitored the sample with IR spectroscopy for 24 hr, which allowed us to derive the reaction kinetic parameters

Note.
a Assumed to be 1:1 based on the work presented in Hanson & Mauersberger (1988).Corresponding HNO 3 and H 2 O column densities were determined from the QCM-derived total areal mass of the ice sample.
following a technique common in the astrochemical literature (Bossa et al. 2008;Loeffler & Hudson 2013;Tribbett & Loeffler 2022; see Section 4 for details).Several of our previous studies involving sulfur use the O 3 combination band at 2104 cm −1 as a proxy for the extent of the reaction (Loeffler & Hudson 2013, 2016;Tribbett & Loeffler 2022) rather than the stronger ozone fundamental band at 1034 cm −1 (Teolis et al. 2007a) since the combination band is well separated from the reaction products formed in those chemical systems.For the ammonia ice mixtures, we find that reaction products strongly overlap with the O 3 combination band (see Table 2).Thus, here we opt to use the O 3 fundamental band as a proxy for the extent of the reaction, taking into account that the strength of this band decreases by ∼5% between 50 and 90 K.

H 2 O-rich Mixtures
Initially, we performed experiments with H 2 O-rich mixtures.We chose the composition of these mixtures to reflect the general expectation that H 2 O is the most cosmically abundant ice and that NH 3 would be more abundant than O 3 , where O 3 can be produced from O 2 , a radiolytic byproduct of water (Bennett & Kaiser 2005;Teolis et al. 2006).Given that the abundance of NH 3 ranges widely depending on the astrochemical application (Tielens 2005), we chose an abundance of ∼25% by number.
Figure 1 shows the mid-infrared (MIR) spectra of a waterdominated H 2 O + NH 3 + O 3 ice mixture after deposition at 50 K and during warming (123 and 207 K) at 1 K minute −1 .The 50 K spectrum is dominated by several large overlapping spectral features attributed to H 2 O (O-H stretch, 3295 cm −1 ), NH 3 (N-H stretch, 3376 cm −1 ), and intermolecular interactions between the H 2 O and NH 3 (O-H-N, 2912 cm −1 ; Moore et al. 2007).At longer wavelengths, the broad H 2 O lattice mode overtone (2250 cm −1 ), bending mode (∼1654 cm −1 ), and lattice mode (835 cm −1 ) are accompanied by several more narrow NH 3 features (1629 cm −1 , 1112 cm −1 ) and O 3 features (2108, 1034 cm −1 ).We note that the 2250 cm −1 band has also been attributed to a possible combination bend and lattice Notes.Note that the -NO 3 band at 800 cm −1 has also been denoted as ν 8 .a Peak positions listed in parentheses are from the H 2 O-dominated ice mixtures.b Assignments indicated with the primes are attributed to the cation ( + NH 4 ) to remain consistent with the cited reference.
vibration mode (Giguère & Harvey 1956).In the NIR, we find two additional NH 3 combination features at 5014 and 4526 cm −1 .The broad baseline structure underlying these combination features is consistent with NH 3 well mixed in H 2 O-dominated mixtures (Moore et al. 2007).The spectral assignments and references for each reactant are shown in Table 2.We note that CO 2 contamination is evident from the absorption feature (C=O stretch) at 2343 cm −1 .Although we cannot easily determine the column number density of CO 2 based on the band strength of CO 2 (Gerakines et al. 1995;Gerakines & Hudson 2015) due to the interference effects typical of thin-film reflectance spectroscopy (Teolis et al. 2007b), we can approximate the CO 2 concentration from previous calibration experiments described in Tribbett & Loeffler (2022).We determine a CO 2 contamination upper limit of ∼1.10 × 10 15 molecules cm −2 , or ∼0.04% of our sample by number.Regardless, previous studies have demonstrated that CO 2 is not oxidized by O 3 (Loeffler & Hudson 2016;Tribbett & Loeffler 2022) and is unlikely to react with H 2 O or NH 3 in the absence of ionizing radiation (Zheng & Kaiser 2007;Chen et al. 2011).At 123 K, our sample looks qualitatively similar to our sample at 50 K, showing no obvious product features.However, we note a decrease in intensity of each NH 3 and O 3 feature and a slight broadening of the H 2 O bending feature centered at 1654 cm −1 .At higher temperatures (e.g., 207 K), most of the sample has sublimated.However, several unresolved features appear in both the 3600-2600 cm −1 (Figure 1, inset) and 1600-1300 cm −1 regions.QCM data (not shown here) confirm the presence of 0.5 μg cm −2 of material present on the substrate (∼0.6% of the initial sample mass), which does not begin to sublimate until 270 K.While the small absorption features in the 3200-2800 cm −1 region (see Figure 1, inset) make unique identification of this residue difficult, its presence along with the observed spectral changes prior to sublimation suggest that a reaction occurs within this ice mixture.

NH 3 -rich Mixtures
Given that H 2 O and O 3 do not appear to react during warming in reference experiments, the stable high-temperature product shown in the inset of Figure 1 is likely due to a reaction between NH 3 and O 3 .Thus, to better investigate these potential reactions, we transitioned to studying NH 3 -dominated ice mixtures.
Figure 2 shows the IR spectra of a H 2 O + NH 3 + O 3 ice mixture after deposition at 50 K and during warming at 0.1 K minute −1 .The initial spectrum at 50 K is qualitatively similar to the H 2 O-dominated ice mixture, although the NH 3 peaks are sharper and the positions of several H 2 O and NH 3 bands are shifted.These shifts are generally consistent with the spectral assignments of features within H 2 O and NH 3 ice mixtures of varying relative concentrations (Moore et al. 2007).At temperatures between 70 and 90 K, we see the formation of three distinguishable, sharp peaks at 1364, 1348, and 800 cm −1 , in addition to a shoulder at 1506 cm −1 and two broader features centered around 1725 and 2058 cm −1 .The sharp doublet feature (1364 and 1348 cm −1 ) is consistent with the -NO 3 ion, previously identified in -NO 3 -bearing salts (Keller & Halford 1949;Theoret & Sandorfy 1964), in solid-phase pure HNO 3 (Smith et al. 1991)  mixtures with NH 3 (Ritzhaupt & Devlin 1977).The broad feature centered near 1725 cm −1 is consistent with a feature that has been observed in hydrated HNO 3 (Ritzhaupt & Devlin 1991;Smith et al. 1991).We attribute the 1506 cm −1 shoulder feature to the + NH 4 ion (Loeffler & Hudson 2015), and the broad 2058 cm −1 feature to + NH 4 in the presence of excess NH 3 (Corset et al. 1968;Huston et al. 1983a).The 800 cm −1 is consistent with a feature previously attributed to the -NO 3 ion and the charge-transfer intermediate - O 3 , or ozonide.We discuss this feature in more detail in Section 4.
Above 110 K, two new absorption features at 1549 and 753 cm −1 appear, which are consistent with the formation of ammonia hemihydrate (2NH 3 :H 2 O; Moore et al. 2007).To better illustrate these similarities, we compare the IR spectrum of a H 2 O + NH 3 + O 3 (20:75:5) after warming to 120 K to the IR spectrum of a H 2 O + NH 3 ice mixture after warming to 123 K (Figure 3).Our binary ice mixture is consistent with the 2NH 3 :H 2 O spectra published by Moore et al. (2007), which suggests that at higher temperatures (110-140 K) the hemihydrate is contributing to the spectra of the mixture containing O 3 .However, more importantly, these hemihydrate features are visually distinguishable from the features that we have attributed to the -NO 3 and + NH 4 ions, indicating that, while these two processes are likely occurring simultaneously, the formation of hydrates is not responsible for the perceived reaction products.We note that as NH 3 and H 2 O begin to sublimate above 140 K (data not shown here), our IR spectra do not present evidence for ammonia monohydrate (NH 3 :H 2 O) or ammonia dihydrate (NH 3 :2H 2 O).This is consistent with the work by Moore et al. (2007), which showed ammonia monohydrate is more readily formed at deposition temperatures higher than those used in our study and also found no evidence of the dihydrate.
In addition to the emergence of several new absorption features, Figure 4 shows the fundamental O 3 absorption band during warming.We expect O 3 to be thermally stable over this temperature range (50-120 K), as blank experiments with only H 2 O and O 3 show negligible O 3 desorption until the ice is warmed to ∼145 K (Figure 5, dashed line) and there is only a slight decrease in the band area due to its temperature dependence.Thus, the decrease in band area suggests that O 3 is reacting to oxidize NH 3 to NO - 3 .Figure 5 shows the QCM-derived mass-loss rate during warming of a H 2 O + NH 3 + O 3 ice mixture between 50 and 290 K at a rate of 0.1 K minute −1 .As noted previously, there is negligible mass loss below 115 K. Above 115 K, we observed three regions of sublimation, beginning at approximately 120, 180, and 230 K.In the first region, we typically see five to six peaks.Interestingly, we saw that the relative intensity of these peaks varied significantly from sample to sample, as well as a slight variation in the peak position, even though the initial IR spectra and estimated compositions were nearly identical.We attribute this to the number of competing phases present and crystallization during sublimation.The first mass-loss peak in this first region occurs at 120 K and is likely due to the release of O 2 as the residual NH 3 becomes increasingly mobile and crystallizes.This desorption temperature is consistent with previous studies that demonstrate the release of O 2 from proton-irradiated H 2 O ices (Bahr et al. 2001).The next desorption peak at 130 K is largely due to the desorption of unreacted NH 3 .We observe four additional peaks between 135 and 165 K.These desorption peaks and the underlying structure within the peaks are likely due to the residual H 2 O and the presence of ammonia hydrates, each with slightly different binding energy to H 2 O.However, due to the relatively thin samples, we cannot confirm the identity of these hydrates using their corresponding IR spectra.In the next two sublimation regions, we observe a single peak, with the first one occurring near 190 K and the second one occurring near 275 K.We estimate the areal mass of the first peak to be ∼0.3 μg cm −2 and the second peak to be ∼5.1 μg cm −2 .Interestingly, the areal mass of the second peak is approximately 10× more material than we found at high temperatures in the H 2 O-dominated ice.
Figure 6 shows an IR spectrum of this residue at 200 K.The features between 1500 and 1300 cm −1 are characteristic of a semi-volatile -NO 3 compound, while the stronger features between 3200 and 2800 cm −1 resemble spectra of the residue produced in the H 2 O-dominated ice (Figure 1 inset).Given the presence of excess NH 3 in our NH 3 -dominated mixtures, the  likely counterion is + NH 4 , suggesting the formation of the salt ammonium nitrate (NH 4 NO 3 ).Moreover, the high-temperature salt spectrum (Figure 6) is consistent with the IR spectrum of low-temperature phase NH 4 NO 3 crystals produced by Theoret & Sandorfy (1964).Interestingly, we found that the desorption temperature of this thermally stable product can shift by ∼10 K between samples, which likely reflects slight variations in the initial ice composition, changing the amount of salt present, and the highly porous and nonuniform nature of the salt residue, which could alter its desorption characteristics.This porosity and nonuniformity is similar to the formation of pure, highly porous O 3 ice samples from irradiated and sublimed O 2 ices (Teolis et al. 2007a).
Given that the higher-temperature residue appears to be a salt, we speculated that the smaller peak near 190 K was due to nitric acid or hydrated nitric acid.To verify this, we grew a HNO 3 + H 2 O mixture (∼1:1) and warmed it at 1 K minute −1 , while monitoring the IR spectra and QCM.We find that the entire sample had sublimated prior to 200 K (Figure 5), which is reasonably consistent with previous studies examining the IR spectra of nitric acid and water clusters (Ritzhaupt & Devlin 1991;Smith et al. 1991), and supports that the desorption peak near 190 K is due to nitric acid or hydrated nitric acid.Our IR measurements also support this conclusion, as we find that although the IR spectrum (Figure 3) is consistent with the amorphous thin film of nitric acid and water produced by Ritzhaupt & Devlin (1991), it does not match any major product features we observed in our warmed H 2 O + NH 3 + O 3 mixture, suggesting that any form of HNO 3 produced would be in relatively small amounts.

Reaction Chemistry and Spectral Assignments
Due to the importance of NH 3 -bearing atmospheric particulates and the presence of NH 3 in waste water, the oxidation of NH 3 by O 3 has been studied previously in both the gas (Olszyna & Heicklen 1971;De Pena et al. 1973;Olszyna et al. 1974) and liquid phases (Singer & Zilli 1975;Kuo et al. 1997;Khuntia et al. 2013).Olszyna & Heicklen (1971) demonstrated that gaseous mixtures of O 3 and excess NH 3 could produce O 2 , H 2 O, N 2 O, N 2 , and solid NH 4 NO 3 at 30 • C. The mechanism described involved single-oxygen-atom transfers and required a free-radical chain-reaction mechanism, assuming an OH radical chain carrier (Equations ( 1)-( 6)): While the authors provide constraints on the reaction mechanism, the radical chain initiation step remains unclear, and the majority of these intermediate N-bearing species were not detected (Olszyna & Heicklen 1971).Subsequent gas-phase studies have largely focused on the product, NH 4 NO 3 , and kinetics of particulate growth (De Pena et al. 1973;Olszyna et al. 1974) assuming an overall reaction of More recent efforts have coupled this overall reaction to photochemical models to produce steady-state distributions of NH 3 and -NO 3 aerosols globally and in regions of high localized pollution (Feng & Penner 2007;Wen et al. 2015).The aqueous literature, with an emphasis on waste-water treatment, demonstrates a similar oxidation reaction between O 3 and the free NH 3 within + NH 4 -salt solutions, NH 4 Cl (Singer & Zilli 1975) or NH 4 SO 4 (Khuntia et al. 2013), producing the -NO 3 ion in solution: Interestingly, H 2 O not participating in the reaction scheme proposed in the aqueous literature (Equation ( 8)) supports our finding that H 2 O-dominated mixtures show a lower abundance of products than the NH 3 -dominated mixtures, suggesting that in our case H 2 O acts as a dilutant, inhibiting the efficiency of the overall reaction.
Unlike the studies mentioned above, which examined this chemical system at or near room temperature, Huston et al. (1983b) found that a reaction between NH 3 and O 3 also occurs readily at cryogenic temperatures in the solid phase.This reaction produced NH 4 NO 3 in the presence of excess NH 3 through the intermediate ammonia ozonide ( + NH 3 O - 3 ), which was identified based on an IR absorption band at 800 cm −1 and a yellow color change: Similarly, Herman & Giguère (1965)  Given that the two proposed mechanisms to form the salt require a slightly different stoichiometry and involve different intermediate products, we use our QCM and IR data to determine whether either mechanism is favored.For singleatom addition (1-6), 4-5 O 3 molecules react to form NH 4 NO 3 , while only 2-3 react in reactions involving the ozonide intermediate.We can estimate the molar ratio of these two compounds in our experiments by calculating the initial O 3 column density from the IR and the mass of the hightemperature NH 4 NO 3 residue from the QCM.Across several experiments, we calculate a molar ratio of 4.3 ± 1.2 O 3 /NO - 3 , suggesting that single-atom addition may be important.However, at low temperatures we do not see the IR-active NO - 2 ion, nitrite, that forms in Equation (4), which should have an absorption shifted from the -NO 3 features (Weston & Brodasky 1957).Additionally, at higher temperatures (e.g., 200 K), our ammonium salt spectrum is also consistent with the lowtemperature (∼213 K) NH 4 NO 3 spectrum (Theoret & Sandorfy 1964) between both 3200-2800 cm −1 and 1600-1300 cm −1 , while showing no additional structure indicating the presence of NH 4 NO 2 , which is also IR active and shifted to lower wavenumbers (Waddington 1958).The absence of NO - 2 is also consistent with aqueous work (Equation ( 8)), where Singer & Zilli (1975) note the absence of detectable nitrite in solution despite a O 3 /NO - 3 ratio of 4. Interestingly, we also observe a feature near 800 cm −1 above 90 K, which has been previously attributed to ozonide in mixtures of NH 3 and O 3 (Herman & Giguère 1965;Huston et al. 1983b).While it is possible that this is ozonide, it may also be a weaker band of nitrate, as its appearance and increase correlate well with the stronger nitrate doublet absorptions at higher wavenumbers.In fact, Keller & Halford (1949) identified a -NO 3 in-plane bending feature in NH 4 NO 3 samples near 830 cm −1 , and Smith et al. (1991) demonstrated that this feature may be redshifted in mixtures of HNO 3 and H 2 O.While this 800 cm −1 feature further increases in intensity and shifts to slightly higher wavenumbers at higher temperatures (∼120-130 K), this is due to the formation of ammonia hemihydrate, which also has a spectral feature near 800 cm −1 (Moore et al. 2007).These spectral changes correspond to the appearance of a new feature at 1549 cm −1 , another 2NH 3 :H 2 O feature.Thus, while we cannot rule out the importance of the ozonide pathway from the IR data, we can conclude that if the single-atom-addition pathway is favored, as is suggested by the QCM results, then the intermediate reactions must occur very quickly, rendering the nitrite undetectable.Regardless of the mechanism and the unstable intermediate, which matter significantly less to observations of planetary bodies where thermal reactions occur over millions of years, we demonstrate that NH 3 reacts with O 3 at low temperatures to produce the refractory + NH 4 -bearing salt NH 4 NO 3 .

Reaction Kinetics
In addition to investigating the mechanism and products of this reaction, we are interested in determining the kinetic parameters, which provide constraints on the reaction time at lower, astronomical temperatures outside of the range of temperatures studied here.We follow the isothermal annealing technique, which has been used previously to determine the overall activation energy of other astrochemically relevant systems (Bossa et al. 2008;Loeffler & Hudson 2013;Tribbett & Loeffler 2022).Figure 7 shows the IR spectrum, including the 1033 cm −1 O 3 absorption band, during warming to and isothermal annealing at 90.6 K (t = 0 at 50 K).As a proxy for the extent of the reaction, we integrated the band area of the 1033 cm −1 O 3 absorption band after removing the continuum with a linear baseline.We note that the total mass loss of the sample over the 24 hr annealing period was zero within the uncertainty of the QCM, as expected from Figure 5. Figure 8 shows the normalized integrated O 3 band area as a function of time (t = 0 at 50 K) for three of our isothermal annealing temperatures (75.6, 80.6, and 90.6 K).After we warmed the ice to the annealing temperature, which occurs within 460 s at our highest annealing temperature (90.6 K), the normalized O 3 band area decays approximately exponentially for each temperature studied.Consequently, we assume first-order kinetics for this reaction, and we write the rate of reaction (k) as a function of O 3 column number density: where N 0 is the initial O 3 column number density, and t is the time elapsed.We calculate the rate of reaction (k) at the time at which half of the O 3 is consumed: From the rate of reaction, we can calculate the overall activation energy: shows the Arrhenius plot (i.e., the rate of reaction for various temperatures) for several different isothermal annealing temperatures between 75 and 90 K.The linear fit of this data is proportional to the overall reaction activation energy, which we determine to be 17 ± 2 kJ mol −1 with a preexponential factor of (1.1 ± 0.19) × 10 7 s −1 .The activation energy is consistent with values determined for other chemical systems using this experimental technique (Bossa et al. 2008;Loeffler & Hudson 2013;Tribbett & Loeffler 2022).We note that this reaction has a lower activation energy than the analogous sulfur oxidation reactions (Loeffler & Hudson 2013;Tribbett & Loeffler 2022) but requires more overall energy than many single-step acid-base reactions involving NH 3 .For example, the reaction between HCN and NH 3 forming the salt NH 4 CN requires an activation energy of

Astrophysical Implications
The main focus of this study dealt with identifications of reaction products and constraints on the reaction kinetics in NH 3 -rich ice mixtures, which are unlikely to be found on typical icy extraterrestrial surfaces.However, although our experiments suggest that H 2 O in astrophysical environments may dilute this reaction, we expect that any ice containing both NH 3 and O 3 will produce the same reaction products that we observed in the NH 3 -rich ices, including NH 4 NO 3 , given the appropriate temperature and time.Keeping this in mind, below we place our results from our NH 3 -rich ices in the context of a number of astrophysical environments.
We find that NH 3 ice is readily oxidized by the radiolytically produced oxidant O 3 to produce the N-bearing anion NO - 3 , which upon warming forms a + NH 4 -bearing salt (NH 4 NO 3 ).Moreover, this reaction occurs in the absence of radiation, at temperatures as low as 70 K on a laboratory timescale.For example, at 75 K, half of the O 3 is consumed within approximately 24 hr, while half is consumed within 30 minutes at 85 K. Critically, several Uranian satellites contain observable NIR bands that have been attributed to ammonia hydrates and ammonium-bearing salts (Cartwright et al. 2020(Cartwright et al. , 2023)), and have been correlated to geologically active terrain.These satellites have average dayside surface temperatures between 70 and 85 K (Hanel et al. 1986;Grundy et al. 2006), although the temperature extremes extend above (90 K) and well below (20-40 K) that range (Sori et al. 2017;DeColibus et al. 2022).
Our results indicate that this thermally driven reaction can quickly produce the NH 4 NO 3 salt at these average temperatures, and likely at lower temperatures over geologic times.We note that the NH 4 NO 3 features shown in the inset of Figure 6 are at 2.03 and 2.14 μm, which is shifted from features tentatively attributed to ammonia hydrates and ammoniumbearing salts on the Uranian satellites.Specifically, the reported reflectance spectra of both Ariel and Umbriel possessed absorption features at 2.20 and 2.24 μm (Cartwright et al. 2020(Cartwright et al. , 2023)).Umbriel also showed an absorption feature at 2.14 μm; however, this feature was attributed to an organic molecule with an amine or nitrile functional group (Cartwright et al. 2023).Our experiments provide an additional + NH 4 -bearing salt candidate, NH 4 NO 3 , for the 2.14 μm absorption feature.While direct comparisons of the peak positions of our residue to remote-sensing observations are limited in applicability since we expect that both changes in temperature and physical structure (e.g., different crystalline or amorphous phases) may shift the absorption features of our NH 4 NO 3 samples, which we acquired at ∼200 K, we note that this reaction provides an additional candidate for the + NH 4 -bearing salts responsible for the NIR spectral features in the 2.0-2.24μm region.NH 4 NO 3 may provide a considerably more refractory nitrogen-atom sink compared to ammonia or hydrated ammonia, similar to other NH 4 -bearing salts which have been postulated to contain a reservoir of nitrogen atoms on comet nuclei (Filacchione et al. 2019;Poch et al. 2020;Kruczkiewicz et al. 2021).
NH 3 , NH 3 hydrates, and + NH 4 -bearing salts have also been postulated to be on the surface of Pluto's satellite Charon (Cook et al. 2023).While Charon is expected to be significantly colder than the Uranian satellites (15-60 K; Grundy et al. 2016b), this reaction could proceed over a geologic time frame  75.6, 77.2, 78.7, 80.6, 82.7, 84.1, 85.6, 87.7, and 90.6 K. or during a heating or resurfacing event (Stern et al. 1993;Grundy et al. 2016a).We note that the absorption features identified in the reflectance spectra of Charon (Cook et al. 2023) are also at longer wavelengths (2.21 and 2.24 μm) than our NH 4 NO 3 sample spectrum (2.03 and 2.14 μm).
In addition to the aforementioned satellites, recent measurements using the Jovian Infrared Auroral Mapper (JIRAM) on NASAʼs Juno spacecraft show evidence for a spectral feature at 2.22 μm on Ganymede, which has been postulated to be due to one of several different salts, including + NH 4 -bearing salts (Tosi et al. 2024).Ganymede provides an exciting test for the thermally driven chemistry demonstrated here not only because both NH 3 (Molyneux et al. 2022) and O 3 (Noll et al. 1996;Hendrix et al. 1999) have been tentatively detected on the surface, in addition to the O 3 precursor O 2 (Spencer et al. 1995;Trumbo et al. 2021), but also because the surface temperatures are considerably warmer on Ganymede (85-145 K; Hanel et al. (1979)) than on the Uranian and Plutonian satellites.While Tosi et al. (2024) ruled out -NO 3 based on the absence of a strong feature at wavelengths longer than 2.5 μm, the NO - 3specific NH 4 NO 3 features appear outside of the wavelength range probed by JIRAM (6.5-7.5 μm).Recently, Bockelée-Morvan et al. (2024) performed JWST NIRSpec and MIRI observations of Ganymede and acquired spectra that contained residual structure between 5.5 and 7 μm after model subtraction.However, the quality of these spectra is somewhat low due to flux discontinuities near the MIRI channel edges and the limited number of binned spectra.Future analyses of MIR spectra of Ganymede may be able to provide additional, more diagnostic evidence for nitrate salts, and for NH 4 NO 3 specifically.
In addition to the outer solar system, + NH 4 features have also been detected in the interstellar medium (ISM) near young stellar objects by their absorptions at 3.26, 3.48, and 6.85 μm (Schutte & Khanna 2003, their   NH 4 (6.85 μm) in the ISM toward dense cores; however, they also found evidence of cyanate (OCN − ), suggesting the possibility of NH 4 OCN.While this does not completely rule out interstellar NH 4 NO 3 , NH 4 OCN may be more likely here, as NH 3 and HNCO can react to form NH 4 OCN at temperatures as low as 8 K (Demyk et al. 1998;Van Broekhuizen et al. 2004), which corresponds to an activation energy of only 0.4 kJ mol −1 (compared to 17 kJ mol −1 for NH 3 and O 3 ).Given that this reaction occurs readily at astrochemical temperatures and the resultant salt contains spectral features consistent with features detected on several outer solar system objects, we intend to derive optical constants for this salt in the NIR and MIR spectral regions in the near future.

Conclusion
Here, we provide evidence for an efficient thermally driven oxidation reaction between NH 3 and O 3 at temperatures as low as 70 K.We determined the overall activation energy of this reaction to be 17 ± 2 kJ mol −1 , which is consistent with other chemical systems that react at cryogenic temperatures.This reaction produces the -NO 3 anion at low temperatures, which interacts with excess NH 3 to produce the cation + NH 4 .Warming our H 2 O + NH 3 + O 3 mixtures through sublimation, we find a number of higher-temperature phases, such as 2NH 3 :H 2 O, HNO 3 , and NH 4 NO 3 .The most stable of these is the NH 4 NO 3 salt, which remains on the substrate until temperatures near 270 K. Notably, the salt product within this sample contains NIR spectral features between 2.0 and 2.22 μm, which is a spectral region of interest for several outer solar system objects, including the Uranian satellites Miranda, Ariel and Umbriel, and Plutoʼs satellite Charon.Finally, Ganymede provides an exciting test case for this thermally driven reaction since both NH 3 and O 3 have been tentatively detected on its surface and key -NO 3 absorption features may be accessible using state-ofthe-art MIR instruments.

Figure 1 .
Figure 1.IR spectra of a H 2 O + NH 3 + O 3 sample (68:24:8) during warming at a rate of 1.0 K minute −1 .Spectra correspond to 50 K (black), 123 K (red), and 207 K (blue).The 50 and 123 K spectra are vertically offset for clarity.Inset: highlighting the fundamental O-H stretching region of the 207 K spectrum (note the significantly smaller optical depth scale).
oxidized N-bearing compounds (including NO - 3 ) in UVphotolyzed mixtures of H 2 O, CO 2 , NH 3 , and O 2 .Moreover, they discuss the possibility of the -NO 3 and + NH 4 through ion and electron irradiation, and through thermally driven surface chemistry.Here, we demonstrate the efficacy of the postulated thermally driven surface chemistry.More recently, McClure et al. (2023) detected +

Table 1
Summary of All Experiments Including the Initial Reactant Column Number Densities (Molecules cm −2 ), Relative Abundances, and Prescribed Heating Treatments