Systematic investigation of CO2 : NH3 ice mixtures using mid-IR and VUV spectroscopy – part 2: electron irradiation and thermal processing

Many experimental parameters determine the chemical and physical properties of interstellar ice analogues, each of which may influence the molecular synthesis that occurs in such ices. In part 1, James et al., RSC Adv., 2020, 10, 37517, we demonstrated the effects that the stoichiometric mixing ratio had on the chemical and physical properties of CO2 : NH3 mixtures and the impact on molecular synthesis induced by thermal processing. Here, in part 2, we extend this to include 1 keV electron irradiation at 20 K of several stoichiometric mixing ratios of CO2 : NH3 ices followed by thermal processing. We demonstrate that not all stoichiometric mixing ratios of CO2 : NH3 ice form the same products. Not only did the 4 : 1 ratio form a different residue after thermal processing, but O3 was observed after electron irradiation at 20 K, which was not observed in the other ratios. For the other ratios, the residue formed from a thermal reaction similar to the work shown in Part 1. However, conversion of ammonium carbamate to carbamic acid was hindered due to electron irradiation at 20 K. Our results demonstrate the need to systematically investigate stoichiometric mixing ratios to better characterise the chemical and physical properties of interstellar ice analogues to further our understanding of the routes of molecular synthesis under different astrochemical conditions.


Introduction
Interstellar ice analogue experiments play an important role in understanding molecular synthesis in the interstellar medium (ISM), and systematic investigations into the experimental parameters can provide a wealth of information. In James et al., 1 henceforth referred to as RJ20, we demonstrated the impact that one discrete experimental parameter, the stoichiometric mixing ratio, has on both the chemical and physical properties of CO 2 : NH 3 ice mixtures from deposition at 20 K and throughout thermal processing. For example, the NH 3 -rich CO 2 : NH 3 mixtures (e.g. 1 : 3, 1 : 10) formed higher amounts of residue material where more NH 3 crystallite grain boundaries existed, suggesting that structural diffusion of reactants may be linked to enhanced reactivity of this system.
In this follow-up paper, we extend our study of the discrete experimental parameter of stoichiometric mixing ratios of CO 2 : NH 3 ices to include 1 keV electron irradiation at 20 K followed by thermal processing. Non-thermal processing due to electrons is thought to occur in the ISM due to the interaction of cosmic rays with solids releasing secondary electrons. 2 Electron irradiation of CO 2 : NH 3 ice analogues has been previously studied 3,4 as well as other methods of non-thermal processing such as UV photons 5 and ions. 6 These previous studies are summarised in Table 1. Ammonium carbamate ([NH 4 ][H 2 NCO 2 ]) and carbamic acid (H 2 NCOOH) were reported as UV processing products at 10 K, ammonium carbamate was reported as a product for 9-20 eV electron irradiation at 10 K and carbamic acid as a product for 144 keV S 9+ ion processing at 16 K. We note that ammonium carbamate, or both ammonium carbamate and carbamic acid, were identied as the thermal products in previous non-irradiated studies. 1,3,[5][6][7][8][9][10][11][12] In addition to ammonium carbamate and/or carbamic acid CO, 4,5 OCN À , 4,5 N 2 O (ref. 6) and ammonium formate 5,6 were reported. In the previous studies of non-thermal processing of CO 2 : NH 3 ices summarised in Table 1 only two different stoichiometric mixing ratios were investigated, a 1 : 1 ratio and a 0.75 : 1 ratio. As such, this paper presents the rst study dedicated to investigating the non-thermal processing of different stoichiometric mixing ratios of CO 2 : NH 3 ices. Similar to RJ20, in this paper, we present combined mid-IR and vacuum-ultraviolet (VUV) studies of CO 2 : NH 3 ice mixtures. For the studies summarised in Table 1, in situ mid-IR spectroscopy was used to study the evolution of the CO 2 : NH 3 ices. For some studies, mass spectrometry 5 and high-resolution low energy electron loss spectroscopy 3 were also used. VUV spectroscopy not only provides an additional technique to study the formation of products, but variations in Rayleigh scattering tails in CO 2 : NH 3 ices observed in VUV spectra in RJ20 can provide insight into the physical changes occurring with the ice analogues due to initial stoichiometric mixing ratios.
In this paper, we present the results of CO 2 : NH 3 mixtures deposited at 20 K and then irradiated with 1 keV electrons at 20 K, and subsequent thermal processing. Henceforth ices processed in this way are referred to as e-irradiated. Mid-IR spectra of CO 2 : NH 3 mixtures with differing ratios of 4 : 1, 2 : 1, 1 : 1, 1 : 2, 1 : 5 & 1 : 10 are complemented by a VUV spectroscopic study of mixtures with ratios of 4 : 1, 2 : 1 & 1 : 3. The results presented in RJ20, which were also thermally processed, but not irradiated with electrons, are used as a set of control experiments to make direct comparisons and, therefore, investigate the effect of electron irradiation on the chemical and physical properties of the ice mixtures. The astrophysical implications of this study combined with that of RJ20 will be presented in a further publication.

Experimental
Both the mid-IR and VUV experiments were performed using The Open University Portable Astrochemistry Chamber (PAC). This was the same experimental system used in RJ20, using a Nicolet Nexus 670 FTIR spectrometer with an external MCT detector for mid-IR measurements and the AU-UV beamline on ASTRID2 (Aarhus University, Denmark) to record the VUV spectra. A more detailed description can be found in Section 2 of RJ20 and Section S1.1 of the ESI for RJ20. Physical vapour deposition of the ices was conducted at a base pressure of low 10 À9 mbar and a base temperature of 20 K. CO 2 (99.8%, BOC) and NH 3 (99.96%, ARGO International Ltd) were premixed in the gas line prior to deposition and deposited onto a cooled substrate (mid-IR: ZnSe, Crystran; VUV: MgF 2 , Crystran). Aer deposition, the ice samples were irradiated with 1 keV electrons using a Kimball Physics FRA-2X1-5549 electron gun with a current of 10 mA. Electron irradiation was conducted at set intervals, cumulating in a total electron irradiation time of 30 min or a total uence of 3.37 Â 10 16 e À cm À2 , and spectroscopic measurements were taken aer each interval. The acquisition time for each mid-IR spectrum was approximately 2 min. Aer electron irradiation, the ices in the mid-IR study were allowed to rest for $1 h. For the VUV spectrum, the acquisition time is dependent on the step size used: for spectra below 110 K, this corresponded to $1 h, and for temperatures at or above 120 K, this corresponded to $10 min. The ice samples were then thermally processed to set temperatures and held at this temperature while spectroscopic measurements were taken.
The lm thickness and ratios of the CO 2 : NH 3 mixtures were calculated in the same way as described in the ESI of RJ20. Deposition rates were between 0.8-1.9 nm s À1 for both mid-IR and VUV spectroscopic studies. The average lm thickness for the mid-IR study was 437 nm. Thinner lms for the VUV measurements were required to prevent saturation of the absorption peaks, and the average lm thickness was 204 nm for the VUV study. The spectra of the mixtures were normalised to a specic thickness when compared, 400 nm for the mid-IR samples and 200 nm for the VUV samples, and indicated in the gure captions. The individual sample thickness and normalisation factors are given in Table S1 in the ESI. † All mid-IR and VUV spectra are freely available on the Open Research Data Online (ORDO) Repository. 13

Penetration depth of 1 keV electrons
To negate any substrate effects which may occur during electron irradiation the ice samples were grown to a thickness larger than the electron penetration depth. The penetration depths of the 1 keV electrons within the ice samples were estimated using the CASINO (monte CArlo SImulation of electroN trajectory in sOlids) program. 14 The CASINO program requires inputs of the electron energy, the angle of the electron beam with respect to the sample, the composition, and density of the sample. A weighted density was used for the CO 2 : NH 3 mixtures which was calculated using the density of pure CO 2 (1.11 g cm À3 ) and Table 1 Summary of the results of previous studies conducted on the non-thermal processing of CO 2 : NH 3 ice mixtures at low temperatures (10-30 K pure NH 3 (0.74 g cm À3 ) 15 and specic values are given in Table  S2 of the ESI. † The estimated average penetration depth of the electrons was 65 nm, well below the sample thicknesses.

Deposition at 20 K
The following CO 2 : NH 3 mixtures were deposited at 20 K: 4 : 1, 2 : 1, 1 : 1, 1 : 2, 1 : 5 & 1 : 10. These ratios are similar to the ratios used in the mid-IR study of RJ20 (3 : 1, 2 : 1, 1 : 1, 1 : 3 & 1 : 10), with slight differences arising due to practical difficulties during gas mixing prior to deposition. In this study, the 4 : 1 ratio is comparable to the RJ20 3 : 1 ratio, and the 1 : 2 & 1 : 5 ratios are comparable to the RJ20 1 : 3 ratio. Deposition spectra are shown in Section 2.1, Fig. S2.1 of the ESI, † and band assignments and positions are given in Table S3 of the ESI. † The same general trends in the mid-IR spectra were observed for the ratios used in this study and for the ratios used RJ20. In brief, the stoichiometric mixing ratio affected the CO 2 bonding environment. For the NH 3 -rich 1 : 10 mixture, CO 2 was essentially a defect within the NH 3 ice and existed in the form of isolated CO 2 molecules. For the NH 3 -rich 1 : 5 mixture CO 2 complexed to NH 3 to form a CO 2 : NH 3 molecular complex as well as existing as isolated CO 2 molecules. Within the other ratios (4 : 1, 2 : 1, 1 : 1 & 1 : 2) CO 2 was in the form of CO 2 dimers, CO 2 : NH 3 molecular complexes and isolated CO 2 . For a detailed characterisation of the CO 2 : NH 3 mixtures deposited at 20 K, see Section 3.1 of RJ20.

Electron irradiation at 20 K
Aer deposition at 20 K, the CO 2 : NH 3 mixtures were irradiated with 1 keV electrons at discrete intervals cumulating in a total uence of 3.37 Â 10 16 e À cm À2 . Fig. 1 shows the mid-IR spectra of a CO 2 : NH 3 mixture in a 1 : 1 ratio aer set intervals of 1 keV electron irradiation. Mid-IR spectra of the irradiated ices in other ratios can be found in Section S2.2, Fig. S4 to S7 of the ESI. † For reference, the 1 keV electron irradiation mid-IR spectra of pure CO 2 and pure NH 3 are shown in Fig. S2 and S3 of the ESI, † respectively.
Irradiation with 1 keV electrons induced several changes within the 1 : 1 CO 2 : NH 3 mixture as shown in Fig. 1. The intensity of all the CO 2 absorption bands decreased throughout electron irradiation. Within the O-H/N-H stretching region, between 3550-2900 cm À1 (Fig. 1a), an overall increase in intensity was observed with the appearance of shoulders on the higher wavenumber side of the NH 3 n 3 absorption band and on the lower wavenumber side of the NH 3 n 1 absorption band. A broad feature around 2800 cm À1 also appeared. The appearance of several distinct features was also observed in Fig. 1c: OCN À (2170 cm À1 ), CO (2140 cm À1 ) and HCN (2092 cm À1 ). Several broad features also appeared between 1750-1250 cm À1 : broad shoulders on both sides of the NH 3 n 2 absorption band, and broad features centred around 1485 cm À1 and 1343 cm À1 .
Differences between the stoichiometric ratios were observed, and the positions at which new features formed are listed in Table 2. Notably, the new absorption features due to electron irradiation in the region between 1800-1250 cm À1 were not present in all ratios and are discussed further in Section 3.2.1. The new absorption features which were present also varied in position. The ratio-dependent formation of CO and OCN À are discussed in Section 3.2.2.
3.2.1 New absorption peaks between 1800-1250 cm À1 . Fig. 2 shows the region between 1800-1250 cm À1 for all ratios aer irradiation with 1 keV electrons aer a total uence of 3.37 Â 10 16 e À cm À2 . The new absorption features formed are dependent on the stoichiometric ratios of the initial CO 2 : NH 3 mixtures. For example, on the blue wing of the NH 3 n 4 Fig. 1 Example mid-IR spectra of a CO 2 : NH 3 mixture in a 1 : 1 ratio irradiated with 1 keV electrons at 20 K to a total fluence of 3.37 Â 10 16 e À cm À2 . See Section S2.2 of the ESI † for the mid-IR spectra of the 4 : 1, 2 : 1, absorption band one distinct absorption feature was identied in the 4 : 1 ratio at 1715 cm À1 . For the 2 : 1, 1 : 1 & 1 : 2 ratios, the absorption feature on the blue wing of the NH 3 n 4 absorption band was broad with two shoulder features. On the red wing of the NH 3 n 4 absorption band, two distinct absorption features were observed near 1583 and 1550 cm À1 but only in the NH 3 -rich mixtures. Broad, asymmetric features were observed in the NH 3 -rich mixtures at $1505 cm À1 shiing to $1480 cm À1 in the other mixtures. Absorption features at $1380 cm À1 were observed for the NH 3 -rich mixtures. Broad absorption bands were observed for all ratios, except for the 4 : 1 ratio, near 1340 cm À1 . An absorption feature was observed near $1300 cm À1 for all ratios except the 1 : 10 ratio.
Absolute band assignments of these irradiation features are very difficult without theoretical calculations and are beyond the focus of this paper. Broadly, the overall prole of the mixtures shown in Fig. 2 is similar for all ratios, except for the 4 : 1 ratio, indicative of products with similar functional groups. However, we do note that NH 3 -rich mixtures had additional absorption features at $1583 cm À1 , $1550 cm À1 , and Table 2 Band assignments and positions of new features after 1 keV electron processing to a total fluence of 3.37 Â 10 16 e À cm À2 of pure CO 2 ice (1 : 0), pure NH 3 ice (0 : 1) and CO 2 :NH 3 mixtures (4 : 1, 2 : 1, 1 : 1, 1 : 2, 1 : 5 & 1 : 10) deposited at 20 K
Position (cm À1 ) 1 : 0 4 :  $1380 cm À1 which could either indicate a more complex product, additional products or both. 3.2.2 Formation of CO and OCN À . The formation CO and OCN À was dependent on the initial mixing ratio as shown in Fig. 3, as was the relative amounts of CO and OCN À . CO forms from the direct dissociation of CO 2 due to electron irradiation, and the amount of CO in the CO 2 : NH 3 ratios is in proportion to the concentration of CO 2 within the mixtures. The 4 : 1 ratio had the highest amount of CO, and the 1 : 10 ratio had the lowest amount of CO. The formation of OCN À occurs via several steps, rst requiring CO formation. 4,5 The highest amount of OCN À formed was in the 1 : 5 ratio followed by, in descending order, the 1 : 2, 1 : 10, 1 : 1 & 2 : 1 ratios. While the amount of OCN À seems to depend on the concentration of NH 3 , the 1 : 10 does not t this trend. This suggests that the formation of OCN À depends on an N-bearing intermediate but requires a minimum amount of CO, of which there is not enough in the 1 : 10 mixture. No detectable amounts of OCN À were observed in the 4 : 1 ratio. The position of the OCN À peak is also ratiodependent, blueshiing as the concentration of NH 3 decreases in the mixing ratios.

Thermal processing
Aer 1 keV electron irradiation at 20 K, the CO 2 : NH 3 mixtures were thermally processed and analysed at discrete temperatures until desorption. Fig. 4 shows the mid-IR spectra of the thermal processing results of a CO 2 : NH 3 mixture in a 1 : 1 ratio. Mid-IR spectra of the other ratios can be found in Section S2.3 of the ESI. † Thermal processing aer electron irradiation for pure CO 2 and pure NH 3 ices are given for reference in Fig. S8 and S9 of the ESI, † respectively. Thermal processing, aer electron irradiation, induced several changes within the 1 : 1 CO 2 : NH 3 mixtures as shown in Fig. 4. Similar to the non-irradiated 1 : 1 CO 2 : NH 3 mixture in RJ20, segregation of the homogeneous mixtures was identied through blue shis in the CO 2 vibrational modes towards pure CO 2 wavenumbers between 60-70 K ( Fig. 4a and b). Again similar to the non-thermally processed mixtures in RJ20, splitting of the n 2 fundamental mode of NH 3 between 60-70 K signied a phase change in NH 3 . The CO peak area decreased throughout thermal processing until the CO desorbed between 130-150 K, whereas the OCN À peak area increased until it reached a maximum near 130 K before decreasing until the residue material desorbed near 296 K. Between 70-80 K, a thermally induced reaction was initiated similar to the nonirradiated results in RJ20 with the absorption bands of the CO 2 : NH 3 molecular complex also disappearing at this temperature. While CO 2 desorbed at a similar temperature in both the e-irradiated and non-irradiated 1 : 1 mixture of RJ20, the NH 3 in the e-irradiated mixture desorbed at a higher temperature compared to NH 3 in the non-irradiated mixture. A residue material was present aer both CO 2 and NH 3 had desorbed with slight changes observed in the residue mixture between 150 and 200 K. The residue material from the eirradiated mixtures desorbed at higher temperatures compared to the non-irradiated mixtures in RJ20.
The temperatures at which these changes occurred were dependent on the stoichiometric mixing ratio and are listed in Table 3.
An overall comparison of the ratios from the e-irradiated CO 2 : NH 3 mixtures with the non-irradiated mixtures of RJ20 shows that the temperatures at which segregation of the CO 2 : NH 3 mixtures commenced and the onset phase change temperature of NH 3 were very similar. The estimated electron penetration depth was at most 17%. Therefore, the observed shi in the CO 2 absorption bands associated with segregation and the splitting pattern of the n 3 absorption band of NH 3 associated with the onset of phase change of NH 3 was likely representative of the non-irradiated part of the CO 2 : NH 3 ice samples.
The desorption temperatures of CO 2 and NH 3 were generally higher when subjected to electron irradiation compared to the desorption temperatures of CO 2 and NH 3 from the nonirradiated study of RJ20. Refractory material overlaying a more volatile material can elevate desorption temperatures due to trapping of molecules beneath the refractory layer 21,22 and products formed from e-irradiation likely formed a more refractory layer in our ice mixtures.

Residue
In addition to e-irradiation induced reaction at 20 K from 1 keV electrons, a thermally induced reaction was observed for all CO 2 : NH 3 mixtures near $90 K. Electron irradiation introduces the formation of new molecules (e.g. CO and OCN À ), but the electrons are estimated to only penetrate to a depth of 17% of the thickness of the ice. Although sputtering will reduce the ice  thickness, the majority of the mixtures were still mainly composed of CO 2 and NH 3 , as evidenced through the strong absorption bands of CO 2 and NH 3 observed aer e-irradiation shown in Fig. 1 and S10-S14 of the ESI. † The thermal reaction present in the e-irradiated CO 2 : NH 3 mixtures was, therefore, likely to be the same as the non-irradiated ices in RJ20. Similar to RJ20, ammonium carbamate was identied via the COO À asymmetric and symmetric stretches, and carbamic acid via the C-O and C]O stretches. Fig. 5 shows the residue spectra at 150 K and 200 K for all CO 2 : NH 3 ratios. Ammonium carbamate and carbamic acid were identied in the residue spectra at 150 K and 200 K for all mixtures except the 4 : 1 ratio. Among these ratios, the 1 : 5 ratio had the highest amount of residue material, followed by the 1 : 2 and 1 : 1 ratios, with the 1 : 10 and 2 : 1 ratios forming the least amount of residue. Ammonium carbamate to carbamic acid conversion was observed between 150-200 K in all mixtures except the 4 : 1 ratio.
From Fig. 5 it is clear that the 4 : 1 residue at 150 K lacks the strong COO À asymmetric and symmetric stretches suggesting that little to no ammonium carbamate formed, the overall residue spectral prole is different from the other ratios. Focussing on the region between 1900-800 cm À1 , at 150 K, the 4 : 1 residue spectra has broad absorption peaks centred around 1470 cm À1 and 1368 cm À1 and a much weaker absorption peak at 1262 cm À1 . Thermal processing to 200 K redshis the broad peaks observed at 150 K to 1448 cm À1 and 1332 cm À1 and the weaker absorption peak to 1257 cm À1 . The appearance of the absorption peak at 1612 cm À1 suggests thermal conversion occurred between 150 K and 200 K. While it is clear that the 4 : 1 residue was different from the other ratios, identication of this residue is not possible without further experimental and computational investigations. Fig. 6 shows the comparison between the residue spectra of the mid-IR 2 : 1, 1 : 1 & 1 : 10 ratios at 150 K and 200 K for the non-irradiated mixtures of RJ20 and the e-irradiated mixtures of this paper. The 1 : 10 residues at 150 K were remarkably similar in prole, suggesting that the thermally induced reaction, initiated at $90 K, was not signicantly impacted by electron irradiation at 20 K. At 200 K, we observed a thermal conversion of ammonium carbamate to carbamic acid in the RJ20 1 : 10 residue through the increase in the C]O and C-O stretches at $1707 cm -1 and $1327 cm À1 , respectively. However, very little thermal conversion of ammonium carbamate to carbamic acid was observed in the e-irradiated 1 : 10 ratio. Additionally, OCN À was observed at 150 K and 200 K for the e-irradiated 1 : 10 residue. The e-irradiated residues at 150 K and 200 K were much more intense compared to the non-irradiated residues of RJ20, indicating the formation of higher amounts of residue material.
Electron irradiation at 20 K clearly inuences the thermal reactivity of the 2 : 1 & 1 : 1 ratios. In the non-irradiated study of RJ20 the 1 : 1 and CO 2 -rich ratios formed lower amounts of residue material compared to the NH 3 -rich ratios. This was attributed to the different bonding environment of CO 2 within the different CO 2 : NH 3 mixtures. CO 2 dimers, which formed mainly in the CO 2 -rich and 1 : 1 mixtures, desorbed at lower temperatures compared to isolated CO 2 within an NH 3 matrix and CO 2 : NH 3 complexes found largely in NH 3 -rich mixtures. As discussed in Section 3.3, e-irradiation of the CO 2 : NH 3 mixtures appeared to elevate the desorption temperatures of CO 2 and NH 3 increasing the probability for a thermal reaction to occur before desorption of CO 2 and NH 3 . Also, the residence of CO may play a part, with CO desorption delayed to higher temperatures in the 1 : 1 & 1 : 2 and 2 : 1 & 1 : 5 mixtures.

Electron irradiation and thermal processing of pure CO 2 and NH 3
The VUV spectra of pure CO 2 ice and pure NH 3 ice deposited at 20 K and irradiated with 1 keV electrons at discrete intervals to a total uence of 3.37 Â 10 16 e À cm À2 are shown in Fig. 7a and 8a, respectively. We also present the subsequent thermal processing of the e-irradiated CO 2 and NH 3 ices in Fig. 7b-d and 8b, respectively. Fig. 7a shows the result of e-irradiation of pure CO 2 at 20 K with 1 keV electrons at discrete intervals. The formation of CO was observed by the appearance of an absorption band centred around 147 nm, which had intense vibrational structure, and was due to the A 1 P ) X 1 S + transition. [23][24][25] We tentatively assign the observation of O 3 to the broad absorption peak centred around 258 nm, which is known as the Hartley band. 26, 27 We note that this broad absorption peak may also be ascribed to the Cameron band of CO 28 or a combination of both the Hartley band of O 3 and the Cameron band of CO. However, interpretation of our e-irradiated CO 2 : NH 3 mixtures in Section 4.3, suggests that this band was more likely due to the presence of O 3 . Aer electron irradiation at 20 K pure CO 2 was thermally processed, and VUV spectra were acquired at discrete temperatures until desorption. VUV spectra of the thermal processing of e-irradiated CO 2 ice is shown in Fig. 7b with close-ups of the CO absorption band in Fig. 7c and O 3 in Fig. 7d. The absorption band of CO due to the A 1 P ) X 1 S + transition disappeared between 80-90 K, as did the absorption band of the Hartley band of O 3 with CO 2 desorbing around 100 K. Fig. 8a shows the irradiation of pure NH 3 ice at 20 K with 1 keV electrons at discrete intervals. A new absorption feature centred around 150 nm was observed aer an irradiation uence of 1.12 Â 10 16 e À cm À2 . The intensity of the NH 3 absorption peaks decreased signicantly throughout electron irradiation. Due to the relatively strong VUV absorption crosssection of NH 3 compared to CO 2 , 24 a thinner ice sample was required to prevent saturation of the absorption peaks. The estimated electron penetration depth of the NH 3 ice was 54% compared to the CO 2 ice which was at 23%. Aer electron  irradiation at 20 K, pure NH 3 was thermally processed, and spectra were acquired at discrete temperatures until desorption. VUV spectra of the thermal processing of e-irradiated NH 3 ice is shown in Fig. 8b. The e-irradiated absorption features shown in Fig. 8b persisted as a residue until 200 K. We believe a good candidate for this residue prole is hydrazine (N 2 H 4 ). Previous studies identify hydrazine as a product of non-thermal processing (e.g. electrons, UV photons) of pure NH 3 (ref. [29][30][31][32] with a desorption temperature between 155-172 K. 30,32,33 The gas-phase VUV spectra of N 2 H 4 and N 2 D 4 are characterised by three absorption peaks with maxima near 144 nm, 168 nm & 194 nm. 34,35 The three absorption peak maxima in Fig. 8b are at 129 nm, 152 nm & 176 nm, which may represent these the N 2 H 4 bands since a shi in the absorption peaks from the gas phase to the solid phase is common. 24

Deposition at 20 K
The following CO 2 : NH 3 mixtures were deposited at 20 K: 4 : 1, 2 : 1 & 1 : 3 and were the same as the ratios used in the VUV study of RJ20. For detailed characterisation of the CO 2 : NH 3 mixtures deposited at 20 K see Section 3.1 of RJ20. Deposition spectra are shown in Fig. S3.2 of the ESI. † Briey, the absorption band due to theÃ 1 A 00 2 )X 1 A 0 1 electronic transition of NH 3 largely overlaps and obscures the CO 2 absorption bands due to the 1 P g ) 1 S g + and 1 D u ) 1 S g + electronic transitions in the 1 : 3 ratio. However, in the 4 : 1 & 2 : 1 ratios the absorption band due to the 1 P g ) 1 S g + electronic transition of CO 2 is visible.

Electron irradiation at 20 K
Aer deposition at 20 K the CO 2 : NH 3 mixtures were irradiated with 1 keV electrons at discrete intervals. Fig. 9-11 show the VUV spectra of the electron irradiation of CO 2 : NH 3 mixtures in a 4 : 1, 2 : 1 & 1 : 3 ratio at 20 K, respectively.  Similar to the mid-IR spectra, the VUV spectra showed changes aer electron irradiation. CO formed in all ratios, and was observed via the appearance of an absorption band centred around 147 nm, which was due to the A 1 P ) X 1 S + electronic transition. This transition had vibrational structure, which was most intense for the 4 : 1 ratio, followed by the 2 : 1 ratio and the 1 : 3 ratio. All of the ratios showed a decrease in the intensity of the absorption band due to theÃ 1 A 00 2 )X 1 A 0 1 electronic transition of NH 3 , with the excess NH 3 ratio (1 : 3), having the highest relative decrease.
In addition, the 4 : 1 ratio had a broad absorption peak centred around 258 nm. We believe this absorption peak is due to the Hartley band of O 3 , rather than the Cameron band of CO. If the peak was due to the Cameron band of CO, we would expect to see it in the 2 : 1 ratio where relatively large amounts of CO were still observed. The formation of O 3 was not observed in the mid-IR spectra due to the intense absorption peak of the n 2 absorption band of NH 3 obscuring the O 3 n 3 absorption peak at 1038 cm À1 .

Thermal processing
Aer electron irradiation at 20 K the CO 2 : NH 3 mixtures were thermally processed and VUV spectra were measured at discrete temperatures until desorption. Fig. 12-14 show the VUV thermal processing spectra of the e-irradiated 4 : 1, 2 : 1 & 1 : 3 ratios, respectively.
For the 4 : 1 ratio, the VUV spectra changed quite signicantly between 80-90 K. Desorption of CO was marked by the disappearance of the intense vibrational structure on the absorption band due to the A 1 P ) X 1 S + electronic transition between 80-90 K corroborating the mid-IR results shown in Section 3.3 and consistent with the CO 2 desorption temperature in pure irradiated CO 2 ice. The Hartley band of O 3 also disappeared at the same temperature as CO, indicating O 3 desorption. The absorption band due to theÃ 1 A 00 2 )X 1 A 0 1 Fig. 11 VUV spectra of a CO 2 : NH 3 mixture in a 1 : 4 ratio deposited at 20 K and then processed with electrons at discrete intervals to a total fluence of 3.37 Â 10 16 e À cm À2 . Bottom panel shows a close-up of the region between 120-175 nm including the absorption band due to CO A 1 P ) X 1 S + electronic transition. electronic transition of NH 3 also decreased signicantly between 80-90 K and an absorption band centred around 205 nm also formed at this temperature, indicating a thermal reaction, which continued to increase in intensity until 150 K. The thermal processing of the VUV spectra for the 2 : 1 and 1 : 3 ratios did not change as signicantly as the 4 : 1 ratio. The vibrational structure on the absorption band of CO due to the A 1 P ) X 1 S + electronic transition disappeared between 110-120 K for the 2 : 1 ratio and 120-150 K for the 1 : 3 ratio, corroborating the mid-IR results presented in Section 3.3. The absorption band due to theÃ 1 A 00 2 )X 1 A 0 1 electronic transition of NH 3 gradually decreased throughout thermal processing until 150 K where it disappeared for the 2 : 1 ratio. For the 1 : 3 ratio, the absorption band due to theÃ 1 A 00 2 )X 1 A 0 1 electronic transition of NH 3 had mostly disappeared aer electron irradiation at 20 K before thermal processing. At 150 K, an absorption peak was observed centred around 145 nm in the 2 : 1 & 1 : 3 ratios had vibrational structure. At 200 K, the vibrational structure disappeared in the 2 : 1 ratio but remained in the 1 : 3 ratio. This vibrational structure was not associated with CO, which had desorbed between 110-150 K.
In RJ20, a phase change of NH 3 was observed via the formation of a Wannier-Mott exciton peak at 194 nm in the 1 : 3 ratio when thermally processed to 90 K. However, no Wannier-Mott exciton was observed in the e-irradiated 1 : 3 ratio. The Wannier-Mott exciton peak intensity is associated with the presence of many NH 3 crystallite grain boundaries (i.e., smaller crystallites). Due to electron irradiation, about 42% of the 1 : 3 ratio ice sample was processed, which reduced the number of pure NH 3 crystallite grain boundaries in the non-irradiated part of the sample, and hence there was no observation of the NH 3 Wannier-Mott exciton peak.

Residue
Ammonium carbamate and carbamic acid were identied at 150 K and 200 K for all mid-IR ratios except the 4 : 1 ratio, which exhibited a different residue spectrum (Fig. 5). Fig. 15 shows the VUV residue spectra at 150 K and 200 K for the 4 : 1, 2 : 1 & 1 : 3 ratios. At 150 K and 200 K, the 1 : 3 residue proles were similar. Although, the intensity of the absorption peaks centred around 125 nm and 150 nm decreased at 200 K. A slight redshiing of the 150 nm peak to 152 nm also occurred. The 2 : 1 residue has an absorption peak centred around 145 nm at 150 K, which Fig. 13 VUV spectra of the thermal processing of a 2 : 1 CO 2 : NH 3 mixture after 1 keV electron irradiation to a total fluence of 3.37 Â 10 16 e À cm À2 . Bottom panel shows a close-up of the region between 120-175 nm including the absorption band due to the CO A 1 P ) X 1 S + electronic transition.
Fig. 14 VUV spectra of the thermal processing of a 1 : 3 CO 2 : NH 3 mixture after 1 keV electron irradiation to a total fluence of 3.37 Â 10 16 e À cm À2 . Bottom panel shows a close-up of the region between 120-175 nm including the absorption band due to CO A 1 P ) X 1 S + electronic transition. disappeared at 200 K, revealing a peak centred around 154 nm. A peak around 125 nm resolved upon thermal processing. The 4 : 1 residue has peaks centred around 145 nm and 205 nm at 150 K. The peak at 145 nm decreased in intensity upon thermal processing to 200 K, but the peak centred around 205 nm remained at a similar intensity. Fig. 16 shows a direct comparison between the VUV 4 : 1, 2 : 1 & 1 : 3 residue spectra of the CO 2 : NH 3 residues at 150 K and 200 K of the non-irradiated study of RJ20 and the eirradiated mixtures in this study. In agreement with the mid-IR residue spectra comparison (Fig. 6), the residue spectra for the e-irradiated residues were more intense than the nonirradiated VUV residue spectra of RJ20. At both 150 K and 200 K, the e-irradiated residues lacked the absorption peak at $175 nm observed in the RJ20 residues. Similar to the 4 : 1 mid-IR residue spectra, the 4 : 1 ratio of the e-irradiated residue has a different spectral prole to the equivalent RJ20 ratio.
A tentative assignment of the absorption peaks was given in RJ20 and summarised here. Tentatively, an absorption peak at $150 nm was assigned as arising due to an electronic transition of ammonium carbamate, and an absorption peak at $170 nm was assigned as arising due to an electronic transition of carbamic acid. No absorption peak was observed at $170 nm for the e-irradiated 2 : 1 and 1 : 3 residues, further supporting the assignment of carbamic acid being responsible for the transition at $175 nm. Small amounts of carbamic acid were probably present within the 2 : 1 and 1 : 3 residues but obscured by the ammonium carbamate transition at $150 nm. The absence of an absorption peak at $170 nm in the e-irradiated 4 : 1 ratio also conrmed the mid-IR results of Section 3.4 which showed no evidence of the formation of carbamic acid. The e-irradiated residue spectra at 200 K for all mixtures contained a transition at $150 nm. This transition was tentatively assigned as being due to ammonium carbamate in RJ20. No strong COO À stretches were present in the mid-IR 4 : 1 residues (Fig. 5), suggesting that ammonium carbamate was not present in this mixture.

Rayleigh scattering tails
Rayleigh scattering tails are observed in VUV spectra when particle sizes are less than l\10, and the intensity of scattered light is proportional to l À4 . Similar to RJ20, Rayleigh scattering tails were observed because of the rough, clumpy surface of the ice lm as opposed to scattering tails observed in astrophysical ices, which did not fully wet the surface. 36,37 The Rayleigh scattering tails were tted using the following model introduced in RJ20: where c is a constant of proportionality and is dependent on the particle size, the refractive index and also the number density of scatterers present within the sample.
Similar to RJ20 we present the Rayleigh scattering tails as a fractional change in the constant of proportionality of the processed ice relative to the constant of proportionality at deposition (Dc):

Dc
where c p is the constant of proportionality of the processed (electron or thermal) ice sample and c d is the constant of proportionality of the ice sample at 20 K. Fig. 17 shows the thermal evolution of Dc aer 1 keV electron irradiation for pure NH 3 (0 : 1) and the CO 2 : NH 3 mixtures (4 : 1, 2 : 1 & 1 : 3). The scattering tails for pure CO 2 were outside the Rayleigh scattering regime. In Fig. 17 there are two scatter points marked at 20 K for each ratio. The white crosshair scatter points indicate the Dc values aer 1 keV electron irradiation to a total uence of 3.37 Â 10 16 e À cm À2 . For all CO 2 : NH 3 ratios and pure NH 3 the Dc value was signicantly higher aer electron irradiation compared to before deposition. Molecular dissociation and the formation of new products due to electron irradiation of the samples appeared to disrupt the ice structure. These VUV scattering results provide direct evidence that as well as inducing chemical changes, electron irradiation also caused macroscopic changes in ice structure and morphology. A signicant decrease in the Dc value at 90 K, for the 4 : 1 ratio, and to a lesser extent the 2 : 1, corresponded to the combined desorption of CO and CO 2 , consequently leading to rearrangement within the ice. The 1 : 3 had the least amount of CO 2 and CO and their desorption appeared to have little effect on the Dc value. The Dc value of pure NH 3 also remained relatively constant throughout thermal processing aer electron irradiation at 20 K.

Discussion
We set out with the aim of demonstrating the impact that the different chemical and physical properties of a range of stoichiometric CO 2 : NH 3 ice mixtures can have on the molecular synthesis induced by electron irradiation and subsequent thermal processing.
In agreement with three out of four previous studies on the non-thermal processing of CO 2 : NH 3 ices below 16 K, we observed the formation of CO and OCN À aer 1 keV electron irradiation at 20 K. [4][5][6] Bertin et al. did not observe CO or OCN À aer 9-20 eV electron irradiation at 30 K and noted that if CO had formed, it would have desorbed at their working temperature of 30 K. Given that OCN À requires the formation of CO, this would explain the lack of OCN À . 4,5 For the other products reported in previous studies, i.e. ammonium carbamate, ammonium formate, carbamic acid, NH + 4 and N 2 O, most of their identifying functional groups are within the 3500-2800 cm À1 or 1800-1250 cm À1 regions where signicant overlap occurs with the strong NH 3 absorption bands. We nd it difficult to make absolute assignments of the IR absorption peaks within these two regions. Overlap from the intense NH 3 absorption bands, similar functional groups present within molecules (e.g. ammonium carbamate and ammonium formate), and matrix-isolation effects due to the different stoichiometric ratios of CO 2 : NH 3 mixtures are all contributing factors. As such, we do not identify molecules within this region at e-irradiation temperatures.
Our e-irradiation of stoichiometric ratios revealed that the 4 : 1 ratio formed O 3 , not observed in the other ratios. O 3 formation in pure CO 2 is a multi-step process which is summarised below: Our VUV spectroscopic study was particularly useful for observing the formation of O 3 , which was otherwise obscured by the intense n 2 absorption band of NH 3 in the mid-IR study.
Other than the identication of CO, OCN À and O 3 , no identication of products due to e-irradiation was made. Mid-IR spectra obtained for the stoichiometric mixing ratios of CO 2 : NH 3 indicated a ratio-dependence on products formed from e-irradiation with NH 3 -rich mixtures showing more complex spectra within the region between 1800-1200 cm À1 . Comparison of our 1 : 1 CO 2 : NH 3 residue results with the residue of Jheeta et al., who also used 1 keV electrons to irradiate their 1 : 1 CO 2 : NH 3 mixture albeit at a slightly higher temperature of 30 K, showed good agreement. 4 While Jheeta et al. only identied ammonium carbamate, an absorption peak at $1700 cm À1 is indicative of a C]O stretch suggesting that carbamic acid was also present. Our 1 : 1 results were also in agreement with the 1 : 1 residues of Bossa et al. 5 and Bertin et al. 3 who identied ammonium carbamate and carbamic acid within their residue material. No comparison was made with Lv et al. 6 as they did not thermally process their irradiated CO 2 : NH 3 mixtures.
By combining our e-irradiated results with our nonirradiated thermal study of CO 2 : NH 3 mixtures, presented in RJ20, we were able to further elucidate the impact of electron irradiation. Excluding the 4 : 1 ratio, all of the other e-irradiated CO 2 : NH 3 ratios formed a similar residue to the non-irradiated residues in RJ20. This suggests that the e-irradiated residues, apart from the 4 : 1 residue, are comprised of products mainly induced by the thermal reaction occurring at approximately 80-90 K. However, higher amounts of residue material formed in the e-irradiated CO 2 : NH 3 mixtures compared to the nonirradiated mixtures of RJ20. We believe this was due to the refractory layer formed from electron irradiation at 20 K, which consequently elevated the desorption temperature of CO 2 , thus allowing more time for a thermal reaction to occur. This corresponds to the practices within the literature which use more refractory molecules as 'plugs' to prevent the desorption of more volatile molecules. 21,22 We also observed that less conversion of ammonium carbamate to carbamic acid occurred in the e-irradiated CO 2 : NH 3 mixtures upon thermal processing between 150-200 K. For example, the non-irradiated CO 2 : NH 3 2 : 1, 1 : 1, 1 : 3 & 1 : 10 mixtures showed an increase in the C]O and C-O stretch in Fig. 5 of RJ20, whereas, very little change was observed in Fig. 5 for the thermally processed e-irradiated CO 2 : NH 3 mixtures. Ammonium carbamate appears to be stabilised by the products in the e-irradiated mixtures. At 150 K and 200 K, the eirradiated product of OCN À is present, which we suggest stabilises the ammonium carbamate, thus preventing decomposition to carbamic acid at higher temperatures.
The Rayleigh scattering tails obtained from VUV spectra gave direct evidence of physical changes within the CO 2 : NH 3 ice mixtures due toelectron irradiation.

Conclusions
We systematically investigated the effect of the stoichiometric mixing ratio on 1 keV e-irradiated CO 2 : NH 3 ices and subsequent thermal processing using mid-IR and VUV spectroscopy. Our work is the rst time that a study has focussed on the nonthermal processing of stoichiometric mixing ratios of CO 2 : NH 3 ices. We show that the small e-irradiation products of CO and OCN À are strongly dependent on the initial mixing ratio of CO 2 and NH 3 , with no observable OCN À formed in the 4 : 1 ratio. However, the 4 : 1 ratio did form O 3 aer e-irradiation at 20 K which was not observed in the other CO 2 : NH 3 mixtures. We also show that the CO 2 -rich, e-irradiated 4 : 1 CO 2 : NH 3 ratio formed a markedly different residue upon thermal processing compared to the other ratios. The other ratios formed similar residues between themselves and similar residues to their thermal processing counterpart residues from RJ20. However, ammonium carbamate to carbamic acid conversion was arrested in the e-irradiated residues. The astrophysical implications of these results, along with the results of RJ20, will be discussed in a forthcoming paper.

Conflicts of interest
There are no conicts to declare.