UV photoprocessing of NH₃ ice: photon-induced desorption mechanisms

Abstract

Ice mantles detected on the surface of dust grains towards the coldest regions of the interstellar medium can be photoprocessed by the secondary ultraviolet (UV) field present in dense cloud interiors. In this work, we present UV-irradiation experiments under astrophysically relevant conditions of pure NH₃ ice samples in an ultra-high vacuum chamber where solid samples were deposited on to a substrate at 8 K. The ice analogues were subsequently photoprocessed with a microwave-discharged hydrogen-flow lamp. The induced radiation and photochemistry led to the production of H₂, N₂ and N₂H₄. In addition, photodesorption to the gas phase of the original ice component, NH₃, and two of the three detected photoproducts, H₂ and N₂, was observed thanks to a quadrupole mass spectrometer (QMS). Calibration of the QMS allowed quantification of the photodesorption yields, leading to Y_pd (NH₃) = 2.1|$^{+2.1}_{-1.0}$| × 10⁻³|$\frac{{\rm molecules}}{{\rm incident\ photon}}$|⁠, which remained constant during the whole experiments, while photodesorption of H₂ and N₂ increased with fluence, pointing towards an indirect photodesorption mechanism involving energy transfer for these species. Photodesorption yield of N₂ molecules after a fluence equivalent to that experienced by ice mantles in space was similar to that of the NH₃ molecules (Y_pd (N₂) = 1.7|$^{+1.7}_{-0.9}$| × 10⁻³|$\frac{{\rm molecules}}{{\rm incident\ photon}}$|⁠).

methods: laboratory: molecular, ISM: clouds, ISM: molecules

1 INTRODUCTION

Infrared observations, mainly performed by space telescopes like the Infrared Space Observatory (ISO), Spitzer and AKARI, have revealed the presence of ice mantles on the surface of interstellar dust grains in dense interstellar clouds (see e.g. Gibb et al. 2004; Öberg et al. 2011, and references therein). Water is the major component of the ice mantles, but five more species are confirmed to be present with varying abundances relative to H₂O: CO, CO₂, CH₃OH, NH₃ and CH₄; and several more have been tentatively detected (Boogert et al. 2015 and references therein).

Ice molecules are processed by cosmic rays and the secondary ultraviolet (UV) field produced in the interior of dense clouds when the cosmic rays interact with the gas phase H₂ molecules (Cecchi-Pestellini & Aiello 1992; Shen et al. 2004). In particular, UV photons induce photochemical reactions by photolysis of absorbing molecules and recombination of the produced radicals (see Öberg et al. 2016 and references therein), and also photodesorption processes, which represent one of the proposed non-thermal desorption mechanisms necessary to explain the presence of gas phase molecules in the coldest regions of the interstellar medium (ISM), where they should be completely depleted forming these ice mantles (e.g. Wilacy & Langer 2000; Bergin et al. 2001).

Depending on whether the photodesorbing molecule is the same ice molecule which absorbs the UV photon (or at least a fragment of the molecule), or a different molecule, one can distinguish between direct photodesorption mechanisms, and photodesorption mechanisms involving energy transfer, respectively.

In the case that the photon is absorbed by a surface ice molecule which does not subsequently photodissociate, the absorbing molecule can directly photodesorb to the gas phase. This mechanism has been proposed by van Hemert, Takahashi & van Dishoeck (2015) for the photodesorption of CO molecules from a pure CO ice based on molecular dynamics simulations. However, in previous works, this process had been proven to be inefficient during experimental simulations with CO and N₂ ices when compared to indirect photodesorption mechanisms (see, e.g. Muñoz Caro et al. 2010; Bertin et al. 2012, 2013), although it may have a contribution in other cases (Dupuy et al. 2017). Alternatively, if the absorbing surface molecule photodissociates, the resulting photofragment can desorb provided that it is formed with enough kinetic energy, or it may recombine with another fragment, leading to the formation of a photoproduct which also could photodesorb, thanks to the excess energy of the parent photofragments and/or the exothermicity released during the recombination reaction (see e.g. Andersson & van Dishoeck 2008; Fayolle et al. 2013; Fillion et al. 2014; Bertin et al. 2016). Since the direct photodesorption of photoproducts through this mechanism (which is referred to as photochemical desorption or photochemidesorption in previous works; Cruz-Díaz et al. 2016; Martín-Doménech et al. 2016) takes place immediately after their formation, photoproducts cannot be accumulated on the ice surface prior to their photodesorption, and the measured photodesorption yield during experimental simulations remains constant with fluence.

On the other hand, if the photon is absorbed by a molecule below the ice surface leading to its electronic excitation, the energy can be redistributed or transferred to a surface molecule¹ which could break the intermolecular bonds and photodesorb. This process is usually known as desorption induced by electronic transitions (DIET) followed by subsequent energetic transfer,² this mechanism is sometimes referred to as indirect DIET or photon-induced co-desorption, although no distinctions are made, in general. (Fayolle et al. 2011, 2013; Bertin et al. 2012, 2013; Fillion et al. 2014), and it is referred to as kick-out photodesorption in van Hemert et al. (2015). In the case that the absorbing molecule dissociates, the excess energy can be transferred by the resulting photofragment to a surface molecule, leading to its photodesorption. When the photofragment is an H atom, it can diffuse through the ice and transfer its momentum to a surface molecule. This is known as kick-out photodesorption³ (Andersson & van Dishoeck 2008). Alternatively, the excess energy from the recombination of the photofragments can also be transferred to a surface molecule leading to its photodesorption (Andersson & van Dishoeck 2008; Fillion et al. 2014). When a previously formed photoproduct photodesorbs through any of these indirect mechanisms, accumulation in the ice prior to their photodesorption leads to an increasing photodesorption yield with fluence, as measured during experimental simulations (Martín-Doménech et al. 2015, 2016; Cruz-Díaz et al. 2016).

Experimental simulations aiming to study the photodesorption of molecules taking place during photoprocessing of pure ices made by the six species confirmed to be present in interstellar ice mantles are a necessary first step prior to the study of the photodesorption in more realistic multicomponent ice analogues. These studies have been reported for pure H₂O (Cruz-Díaz et al. 2017, and references therein), CO (Öberg et al. 2007, 2009a; Muñoz Caro et al. 2010, 2016; Fayolle et al. 2011; Bertin et al. 2012, 2013; Chen et al. 2014), CO₂ (Öberg et al. 2009a; Bahr & Baragiola 2012; Yuan & Yates 2013; Fillion et al. 2014; Martín-Doménech et al. 2015), CH₃OH (Öberg et al. 2009b; Bertin et al. 2016; Cruz-Díaz et al. 2016), CH₄ (Dupuy et al. 2017) and NH₃ (Nishi et al. 1984; Loeffler & Baragiola 2010) ices. In this work, we present a series of experiments simulating the UV photoprocessing of pure NH₃ ices focused on the subsequent photodesorption of ice molecules. Irradiation was carried out with a multiwavelength UV lamp, instead of a laser at a given wavelength used in Nishi et al. (1984) and Loeffler & Baragiola (2010). In addition, we include the quantification of the different photodesorbing molecules measured directly from the gas phase, which was not reported in the previous works mentioned above, thanks to the calibration of the quadrupole mass spectrometer (QMS) used in our experimental set-up (presented in Section 2) to detect the photodesorbing molecules. These results are presented in Section 3, and their astrophysical implications are discussed in Section 4. Finally, the conclusions are summarized in Section 5.

2 EXPERIMENTAL SET-UP

The photodesorption of molecules from a pure NH₃ ice has been studied through a series of experimental simulations carried out with the InterStellar Astrochemistry Chamber (ISAC), an ultra-high-vacuum (UHV) set-up located at the Centro de Astrobiología, with a working pressure of ∼4 × 10⁻¹¹ mbar, similar to that found in the interiors of dense interstellar clouds. A brief description of the experimental set-up and the protocol followed during the experimental simulations is provided below (see Muñoz Caro et al. 2010; Martín-Doménech et al. 2015, 2016; Cruz-Díaz et al. 2016 for more details). Pure amorphous ammonia ice samples were deposited from the gas phase on to a KBr window at 8 K used as the substrate, upon introduction of NH₃ (gas, Praxair 99.999 per cent) into the chamber. The deposited ices were subsequently irradiated using an F-type microwave-discharged hydrogen flow lamp (MDHL) with a vacuum-ultraviolet (VUV) flux of 2 × 10¹⁴ photons cm⁻² s⁻¹ at the sample position (Muñoz Caro et al. 2010). The total irradiation time varied between 60 and 1100 min, leading to the incident fluences indicated in Table 1. For a given fluence, the ice thickness determined the percentage of absorbed photons according to the VUV-absorption cross-section reported in Cruz-Díaz et al. (2014a),⁴ as well as the incident and absorbed dose of photons per molecule. The results in Section 3 were not found to strongly depend on these parameters. The emission spectrum of the MDHL has been characterized in situ using a VUV spectrophotometer, and is reported in Chen et al. (2010, 2014) and Cruz-Díaz et al. (2014a). It is similar to the secondary UV field of dense cloud interiors calculated by Gredel et al. (1989). Finally, the ice samples in experiments 1–3 and 6 were warmed up following irradiation to room temperature at a rate of 2 K min⁻¹.

Table 1.

Experimental parameters of the UV photoprocessing of pure NH₃ ices.

Experiment	Ninitial(NH3)	Fluenceincident	Fluence|$_{{\rm absorbed}}^{a}$|	Dose|$_{{\rm incident}}^{b}$|	Dose|$_{{\rm absorbed}}^{a}$|	Heating rate
	(×1015 molecules cm−2)	(×1018 photons cm−2)		(photons molecule−1)		(K min−1)
1	237	2.16	1.32	9.11	5.58	2
2	835	2.88	2.78	3.45	3.33	2
3	912	2.88	2.80	3.16	3.07	2
4	913	0.72	0.70	0.79	0.77	–
5	1882	1.24	1.24	0.66	0.66	–
6	2264	13.27	13.27	5.86	5.86	2

Notes.^aThe number of absorbed photons has been calculated taking into account the initial composition of the ice, i.e. the average UV absorption cross-section for a pure NH₃ ice, provided by Cruz-Díaz et al. (2014a).

^bRelative to the initial number of NH₃ molecules.

During the experimental simulations, in situ Fourier-transform infrared (FTIR) transmittance spectroscopy was used to monitor the solid sample. IR spectra of the ices were collected after deposition, after every irradiation period, and every five minutes during warm-up, with a spectral resolution of 2 cm⁻¹. The initial column density N in molecules cm⁻² of the NH₃ ices was calculated from the optical depth (τ_ν) of the N-H stretching absorption IR band at ∼3300 cm⁻¹, using the formula:

\begin{equation} N=\frac{1}{A}\int _{{\rm band}}{\tau _{\nu } \ {\rm d}\nu }, \end{equation}

(1)

where A is the band strength in cm molecule⁻¹ (2.2 × 10⁻¹⁷ cm molecules⁻¹; Schutte et al. 1996). The ice thickness of the samples is usually expressed in monolayers (ML). One ML is assumed to be 10¹⁵ molecules cm⁻².

At the same time, all the desorbing species, including ammonia and the resulting products of the induced photochemical reactions, were detected in the gas phase by a Pfeiffer Prisma QMS. The molecules reaching the QMS were ionized by ∼70 eV electron bombardment, which led to fragmentation following a given pattern. The main mass fragments coincided in this case with the molecular ions, and were used to monitor the presence of NH₃ (m/z = 17), and the photoproducts N₂ (m/z = 28), H₂ (m/z = 2) and N₂H₄ (m/z = 32) in the gas phase. The conversion from the integrated ion currents measured by the QMS for every irradiation period (A(m/z)) into photodesorbing column densities (N(mol)) was carried out using the equation:

\begin{eqnarray} N ({\rm mol}) &=& \frac{A (m/z)}{k_{{\rm CO}}} \times \frac{\sigma ^{\!+\!} ({\rm CO})}{\sigma ^{\!+\!} ({\rm mol})} \times \frac{I_{{\rm F}} ({\rm CO}^{\!+\!})}{I_{{\rm F}} (z)} \times \frac{F_{{\rm F}} (28)}{F_{{\rm F}} (m)} \times \frac{S (28)}{S (m/z)} . \nonumber\\ \end{eqnarray}

(2)

The meaning of the different parameters in equation (2) is described in Martín-Doménech et al. (2015, 2016) and Cruz-Díaz et al. (2016, 2017). The values used in this work are summarized in Table 2, and in the text. The constant k_CO and the sensitivity of the QMS are regularly calibrated (see Martín-Doménech et al. 2015 for more information on the calibration process).

Table 2.

Values of the parameters in equation (2) used to convert integrated QMS signals into photodesorbing column densities. Photodesorbing H₂ was not quantified due to contamination issues (see the text).

Factor	NH3	N2	CO
σ+(mol) (Å2)a	3.036	2.508	2.516
fragment	NH|$_3^{+}$|	N|$_2^{+}$|	CO+
m/z	17	28	28
IF(z)	1b	1b	1b
	0.460c
FF(m)	0.514d	0.933a	0.949a
	0.498e
|$k^{*}_{{\rm QMS}} \cdot S (m/z)$|5 (A mbar−1 Å−2)	1.78 × 1015	1.03 × 1015	1.03 × 1015
Srel(mol)f	1.1	1.0	1.0

Notes.^aExtracted from the online data base of the National Institute of Standard and Technologies (NIST).

^bWe assumed that no double ionization of the molecules took place in the QMS.

^cCalculated during the deposition in experiment 1.

^dCalculated during the deposition in experiments 2,3,4 and 6.

^eCalculated during the deposition in experiment 5.

^fCalculated with equation (3).

Since the pumping speeds in the ISAC set-up are not the same for all molecules, N(mol) from equation (2) must be corrected to take into account the different pumping speeds of the species of interest with respect to the CO molecules used as reference to extract k_CO:

\begin{equation} N^{{\rm real}} ({\rm mol}) = N ({\rm mol}) \times S_{{\rm rel}} ({\rm mol}), \end{equation}

(3)

where the relative pumping speed (S_rel(mol)) can be calculated as

\begin{equation} S_{{\rm rel}} ({\rm mol}) = 1.258 - 9.2 \times 10^{-3} \times M ({\rm mol}) , \end{equation}

(4)

(see Kaiser et al. 1995; Martín-Doménech et al. 2016 for more information). Photodesorption yields are finally calculated dividing N^real(mol) by the fluence.

3 EXPERIMENTAL RESULTS AND DISCUSSION

3.1 Photon-induced chemistry of pure NH₃ ice

3.1.1 IR spectra of NH₃ ice during irradiation

The IR spectrum of the pure and amorphous NH₃ ice presents three different features: a wide band centred at ∼3300 cm⁻¹ (see black curve in left-hand panel of Fig. 1) corresponding to the N-H stretching modes (the symmetric mode peaks at ∼3375 cm⁻¹, and the anti-symmetric mode at ∼3200 cm⁻¹); a weaker band at ∼1625 cm⁻¹ (see black curve in right-hand panel of Fig. 1) corresponding to the N-H bending mode; and a third band at 1070 cm⁻¹ (not shown) corresponding to the so-called umbrella mode. During photoprocessing of pure NH₃ ices, photodissociation and photodesorption of the NH₃ molecules decrease the absorbance of these bands, as shown in Fig. 1 for the former two features.

Figure 1.

IR spectra of the ice sample after every irradiation interval during photoprocessing in experiment 3 (see Table 1). Similar results were found for the rest of the experiments. Left: IR band corresponding to the N-H stretching modes of the NH₃ molecules. The decrease in the IR absorbance is due to the photodissociation and photodesorption of NH₃ molecules during irradiation. The weak shoulder above 3000 cm⁻¹ corresponds to the N-H stretching mode of the NH^{· ·} radical. Right: IR band corresponding to the N-H bending mode of the NH₃ molecules (smoothed). The peak observed at ∼1500 cm⁻¹ corresponds to the formation of the NH|$_2^{\cdot }$| radical.

Photodissociation in the gas phase of NH₃ molecules readily leads to the formation of NH|$_2^{\cdot }$| and NH^{· ·} radicals due to the loss of one or two hydrogen atoms, respectively, although the formation of ground-state NH radicals is spin-forbidden (Okabe 1978). Formation of NH|$_2^{\cdot }$| during irradiation of pure NH₃ ice was reported in Gerakines, Schutte & Ehrenfreund (1996) and Loeffler & Baragiola (2010), thanks to an IR peak detected above 1500 cm⁻¹ on the red wing of the NH₃ bending band. This peak is observed in the right-hand panel of Fig. 1. In addition, a decrease in the intensity ratio between the symmetric and antisymmetric N-H stretching modes in the left-hand panel of Fig. 1 is probably due to the contribution of the absorption corresponding to the antisymmetric N-H stretching mode of NH|$_2^{\cdot }$| (Milligan & Jacox 1965). At the same time, a wide shoulder at ∼3100 cm⁻¹ on the red wing of the N-H stretching band shown in the left-hand panel of Fig. 1 could be attributed to the imidogen radical (NH^{· ·}; Rosengren & Pimentel 1965). During the warm-up phase of the experiment, after photoprocessing of the ice sample is complete, the mobility of these radicals is increased. Diffusion and subsequent reaction of the radicals led to a decrease of their IR features. The IR peak at ∼1500 cm⁻¹ corresponding to the bending mode of the NH|$_2^{\cdot }$| radical, disappears between 26 and 73 K, as shown in the right-hand panel of Fig. 2. At 67 K, the intensity of this feature is roughly half of that at 8 K, and has completely disappeared at 73 K. In addition, a slight decrease in the shoulder at ∼3100 cm⁻¹ corresponding to the imidogen radical is observed at temperatures above 67 K (left-hand panel of Fig. 2).

Figure 2.

IR spectra of the ice sample during the TPD after irradiation in experiment 3 (see Table 1). Similar results were found for the rest of experiments. Left: Above 67 K, a change in the NH₃ ice matrix structure leads to a change in the N-H stretching band profile (see also Satorre et al. 2013; Giuliano et al. 2014). At the same time, the shoulder above 3000 cm⁻¹ slightly decreases, probably due to the diffusion and subsequent reactions of the NH^· radical. Right: Above 26 K, the intensity of the peak corresponding to the NH|$_2^{\cdot }$| radical also decreases due to its diffusion and subsequent reactions, and has completely disappeared at a temperature of 73 K, indicating that no free radicals are left in the ice sample at that temperature. Spectra are shifted for clarity.

During irradiation, NH^{· ·} and NH|$_2^{\cdot }$| radicals can be dissociated by subsequent photons leading to the production of H and N atoms, which recombine to form H₂ and N₂ molecules (see Loeffler & Baragiola 2010, where additional pathways to the formation of these photoproducts are reported), which are IR inactive due to the lack of an electric dipole moment. Formation of these molecules is confirmed, thanks to the temperature-programmed desorption (TPD) of the irradiated ices during the warm-up phase of the experiments (see Section 3.1.2). In addition, formation of hydrazine (N₂H₄) by recombination of two NH|$_2^{\cdot }$| radicals, or reaction of an NH^{· ·} radical with an NH₃ molecule (Loeffler & Baragiola 2010), is also reported at a temperature of 10 K in Gerakines et al. (1996). Although no IR absorption corresponding to this molecule was detected during our experimental simulations, formation of this molecule was also confirmed during the TPD of the irradiated ices (Section 3.1.2).

3.1.2 TPD of the irradiated NH₃ ice

After photoprocessing of the ice samples, a constant heating rate was applied to perform the TPD of the irradiated ices. Thermal desorption of ammonia and the photoproducts H₂, N₂ and N₂H₄ was detected by the QMS at a given temperature according to their volatility, as shown in Fig. 3. In particular, desorption of H₂ molecules was observed at a wide range of temperatures, since the very beginning of the warm-up phase of the experiment, due to the high volatility of this species. A desorption peak corresponding to most of the photoproduced N₂ molecules was observed at 29 K (slightly higher than the desorption temperature reported in Fayolle et al. 2016 for pure N₂ ices, probably due to the effect of the NH₃ matrix). On the other hand, the hydrazine molecules finally desorbed at 172 K, a desorption temperature 10–20 K lower than that reported in Roux & Wood (1983) for experiments performed in a high-vacuum chamber with a base pressure five orders of magnitude higher compared to the one in the ISAC set-up.

Figure 3.

Thermal desorption peaks detected by the QMS for different m/z fragments during the TPD phase of experiment 6 (see Table 1, similar results were found for the rest of experiments) corresponding to NH₃ (m/z = 17, black) and the three irradiation products: H₂ (m/z = 2, purple), N₂ (m/z = 28, blue) and N₂H₄ (m/z = 32, red). TPD curves are shifted for clarity. The desorption peak observed for m/z = 32 at T = 117 K is probably due to an ion-molecule reaction of ammonia inside the mass spectrometer, since it is also observed in experiments with no ammonia-ice irradiation.

3.2 Photodesorption from pure NH₃ ice

During irradiation of the ice samples, the photodesorbing molecules were detected in the gas phase by the QMS. In particular, photodesorption led to an increase in the signal of the m/z fragments of the photodesorbing molecules during every irradiation interval. The measured ion current of the main mass fragments corresponding to NH₃ (m/z = 17), and the three photoproducts: H₂ (m/z = 2), N₂ (m/z = 28) and N₂H₄ (m/z = 32), are shown in Fig. 4. Photodesorption of the former three species was detected, as in Nishi et al. (1984) and Loeffler & Baragiola (2010), while N₂H₄ was not found to photodesorb. The integrated QMS ion currents were converted into photodesorbing column densities using Equations (2) and (3), the parameter values shown in Table 2, and a k_CO = 5.55 × 10⁻¹² A min ML⁻¹. The ratio of the photodesorbing column densities in every irradiation period and the incident fluence during the same period led to the average photodesorption yield for every irradiation interval shown in Table 3. Quantification of the photodesorbing H₂ molecules was not included since the measured values suffered from background H₂ contamination, which is the most usual contaminant in UHV chambers.

Figure 4.

Photodesorption of NH₃ (black), and the photoproducts H₂ (purple) and N₂ (blue) detected by the QMS during irradiation of the ice sample in experiment 1 (see Table 1, similar results were found for the rest of experiments). Irradiation intervals are indicated with vertical dashed lines. The increase in the photodesorption of H₂ and N₂ suggests that these molecules photodesorb mainly through an indirect mechanism (see the text), while the photoproduct N₂H₄ (red) was not found to photodesorb. Signals are shifted for clarity. Fluctuations in the signals are due to instabilities on the VUV flux during long irradiation periods.

Table 3.

Photodesorption yields measured for the different irradiation intervals during photoprocessing of a pure NH₃ ice in experiment 1, see Table 1. Similar results were found for experiments 2–6.

Irradiation period	Fluencea	Ypd (NH3)b	Ypd (N2)b
(min)	×1017 (photons cm−2)	×10−3 (⁠|$\frac{{\rm molecules}}{{\rm incident\ photon}}$|⁠)
0−10	1.2	2.05	1.51
10−20	2.4	2.12	1.36
20−30	3.6	1.89	2.18
30−40	4.8	2.28	2.41
40−50	6.0	1.92	2.64
50−60	7.2	2.01	3.05
60−180	21.6	1.74	5.86

Notes.^aTotal fluence at the end of every irradiation period.

^bAveraged for every irradiation period. A factor of 2 is assumed as the error in the photodesorption yield values due to the uncertainties in all the parameters of equation (2) (see also Martín-Doménech et al. 2016).

Photodesorption of NH₃ molecules took place with an average photodesorption yield of 2.1 × 10⁻³|$\frac{{\rm molecules}}{{\rm incident\ photon}}$|⁠, which remained constant during the whole experiment (see Table 3). This is similar, within errors, to the photodesorption yield indirectly measured in the solid phase by Loeffler & Baragiola (2010) with a quartz-crystal microbalance (QCM)⁶ during photoprocessing of a pure NH₃ ice grown at 40 K using a 193 nm ArF laser.

During irradiation of pure ice samples, it is not possible to elucidate whether photodesorption of the original ice component takes place through a direct mechanism or through a mechanism involving energy transfer attending only to the evolution of the photodesorption yield with fluence. This can only be done for the photodesorption of photoproducts, as explained in Section 1. Nishi et al. (1984) measured the energy distribution of the photodesorbing NH₃ molecules from a pure NH₃ ice grown at 90 K and irradiated with a 193 nm ArF laser, using a time-of-flight (TOF) mass spectrometer. Two energy components with mean translational energies of 0.17 and 0.65 eV, respectively, were found. The low energy component was associated with NH₃ molecules photodesorbing through a mechanism involving energy transfer from the absorbing molecule in the bulk of the ice to a surface molecule. This is similar to the DIET mechanism involving energy transfer described in Section 1. In this case, absorption of a photon led to a transition to a Wannier exciton state which subsequently propagated through the crystal approaching a weakly bound edge molecule which acquired linear momentum due to the electronic repulsive force. On the other hand, the high-energy component, which accounted for half of the photodesorbing molecules, was associated with excited surface molecules, probably desorbing after recombination of a previously photodissociated NH₃ molecule.

Photodesorption of H₂ and N₂ molecules took place with an increasing photodesorption yield with fluence, as reported in Loeffler & Baragiola (2010). The photodesorption yield of N₂ was found to be 1.4 × 10⁻³|$\frac{{\rm molecules}}{{\rm incident\ photon}}$| (i.e. on the order of that of NH₃) for a fluence of up to 2.4 × 10¹⁷ photons cm⁻², and then gradually raised, being ∼4 times higher when the total fluence was 2.2 × 10¹⁸ (see Table 3). In this case, the increasing photodesorption yield is related to the accumulation of photoproduced H₂ and N₂ molecules in the ice sample prior to their desorption into the gas phase. Therefore, there is a significant contribution of indirect photodesorption mechanisms involving energy transfer. These mechanisms require the absorption of a new photon after the formation of the photoproducts, as explained in Section 1, in contrast to the direct photodesorption mechanisms where desorption occurs immediately after the formation of the molecules, without the need of absorbing a new photon. In that case, the photoproduct molecules could not accumulate on the surface of the ice before their desorption, leading to a constant photodesorption yield with fluence (see Martín-Doménech et al. 2015, 2016; Cruz-Díaz et al. 2016, 2017).

4 ASTROPHYSICAL IMPLICATIONS

Ammonia ice is thought to form early at the onset of the diffuse cloud collapse which leads to the formation of dense clouds where ice mantles are usually detected, by hydrogenation reactions of N atoms on the surface of dust grains. The abundance of NH₃ with respect to H₂O in the ice mantles is around 5 per cent in dense clouds and also in the cold circumstellar envelopes around protostars (see Boogert et al. 2015, and references therein). Its importance, though, is beyond all doubts since it is the only confirmed source of N atoms in interstellar ices. The detection of ammonia in the gas phase of the cold ISM was already reported in Cheung et al. (1968).

NH₃ ice molecules are photoprocessed along with all the ice components in the interior of dense clouds thanks to the secondary UV field mentioned in Section 1, leading to photochemical reactions that produce new ice molecules and photodesorption processes that allow the presence of molecules in the gas phase of cold interstellar regions where thermal desorption is negligible. These processes can be simulated in the laboratory under astrophysically relevant conditions. For the VUV flux of the lamp used (see Section 2), the ice samples experience a fluence of ∼3 × 10¹⁷ photons cm⁻² after ∼30 min of irradiation in our experiments, which is similar to the fluence experienced by the ice mantles during the expected lifetime of a molecular cloud, assuming a secondary UV flux of ∼10⁴ photons cm⁻² s⁻¹(Shen et al. 2004). The six components confirmed to be present in the interstellar ices are distributed over two distinct ice layers as the result of the formation process (see Boogert et al. 2015, and references therein). A polar ice layer is formed first on the surface of dust grains, dominated by H₂O and including NH₃, CH₄, and a fraction of the CO₂ ice molecules. On top of this polar ice layer, an apolar layer is subsequently formed, mainly composed by CO along with the rest of the CO₂ molecules and probably CH₃OH. In any case, irradiation experiments of pure ices are used as benchmarks in the study of the photoprocessing of more realistic multicomponent ices.

In our experimental simulations using an MDHL, whose emission spectrum is similar to that expected in the interior of dense clouds, photodesorption of NH₃ molecules is observed to proceed with a constant yield with fluence during irradiation of a pure NH₃ ice (Y_pd (NH₃) = 2.1|$^{+2.1}_{-1.0}$| × 10⁻³|$\frac{{\rm molecules}}{{\rm incident\ photon}}$|⁠). This value is of the order of the observed H₂O photodesorption during photoprocessing of pure H₂O ices using the same UV lamp (Cruz-Díaz et al. 2017), while it is one order of magnitude lower than the photodesorption yield of CO molecules during photoprocessing of a pure CO ice (Muñoz Caro et al. 2010), and one order of magnitude higher than that measured for CO₂ (see Table 4) under similar conditions (Martín-Doménech et al. 2015). Photodesorption of NH₃ molecules from an ice mixture dominated by water will be addressed in a forthcoming paper, and is expected to be lower than the value presented in this paper for segregated (pure) NH₃ ice (Loeffler & Baragiola 2010).

Table 4.

Photodesorption yields measured for pure ices during irradiation with a MDHL.

Ice component	Ypd	Reference
	×10−3 (⁠|$\frac{{\rm molecules}}{{\rm incident\ photon}}$|⁠)
H2O	1.3 ± 0.2	Cruz-Díaz et al. (2017)
CO	54.0 ± 5.0	Muñoz Caro et al. (2010)
CO2	0.1 ± 0.04	Martín-Doménech et al. (2015)
CH3OH	≤0.03	Cruz-Díaz et al. (2016)
NH3	2.0|$^{+2.1}_{-1.0}$|	This work

On the other hand, we also observed photodesorption of the produced N₂ molecules, with a yield which increased for fluences higher than 2.4 × 10¹⁷ photons cm⁻². For a fluence similar to that experienced by the ice mantles in the expected lifetime of a dense cloud (∼3 × 10¹⁷ photons cm⁻²), the average photodesorption yield of N₂ molecules was similar to that observed for the NH₃ molecules (Y_pd (N₂) = 1.7|$^{+1.7}_{-0.9}$| × 10⁻³|$\frac{{\rm molecules}}{{\rm inciden\ photon}}$|⁠). Photodesorption of N₂ molecules from a pure N₂ ice is expected to be negligible, due to the low VUV-absorption cross-section of this ice (the average absorption cross-section is 7.0 × 10⁻²¹ cm², while for NH₃ ices it is 6.1 × 10⁻¹⁸ cm²; Cruz-Díaz et al. 2014a,b). Detection of gas phase N₂ was reported in Knauth et al. (2004) from far-UV observations towards the star HD 124314, with an abundance that could not be explained with gas phase chemical models for either diffuse or dense clouds.

5 CONCLUSIONS

We have performed experimental simulations of the UV photoprocessing of NH₃ ice using a MDHL with an emission spectrum similar to that expected to be present in the interior of dense clouds. IR spectroscopy in transmittance was used to monitor the solid sample during irradiation. Formation of NH|$_2^{\cdot }$| and NH^{· ·} radicals due to the photodissociation of NH₃ molecules was observed thanks to the appearance of IR features at ∼1500 and ∼3100 cm⁻¹, respectively. The photoproducts H₂ and N₂ are not IR active, and the N₂H₄ IR features did not reach the sensitivity limit in the spectra of the irradiated ices, but they were detected in the gas phase after thermal desorption during the TPD of the photoprocessed ice thanks to a QMS.

In addition, photodesorption of NH₃ and the photoproducts H₂ and N₂ were detected by the QMS during photoprocessing of the ice samples. Photodesorption of NH₃ took place with a constant yield with fluence, as it is the case for other species during irradiation of their pure ices. On the other hand, the photodesorption yield of H₂ and N₂ were observed to increase with fluence, pointing towards a significant contribution of indirect photodesorption mechanism involving energy transfer from the absorbing molecule to a previously formed H₂ or N₂ molecule, which subsequently desorbed. This kind of mechanism allows accumulation of the photoproducts prior to their desorption, leading to an increase of the number of molecules available for the photodesorption and, therefore, an increase of the photodesorption yield with fluence.

The calibration of the QMS allowed us to quantify the photodesorption yields for the desorbing molecules. NH₃ molecules were found to photodesorb with an average yield of 2.1|$^{+2.1}_{-1.0}$| × 10⁻³|$\frac{{\rm molecules}}{{\rm incident\ photon}}$|⁠, which is of the same order than the photodesorption measured for H₂O molecules during irradiation of pure water ice (Cruz-Díaz et al. 2017). N₂ molecules are not expected to efficiently photodesorb from pure N₂ ices (Y_pd (N₂) ≤ 2 × 10⁻⁴|$\frac{{\rm molecules}}{{\rm incident\ photon}}$|⁠; Öberg et al. 2009a), probably due to their low VUV-absorption cross-section (Cruz-Díaz et al. 2014b). During irradiation of a pure NH₃ ice, a photodesorption yield similar to that of NH₃ molecules was measured for N₂ molecules, for a fluence equivalent to that experienced by ice mantles during the expected lifetime of dense clouds (Y_pd (N₂) = 1.7|$^{+1.7}_{-0.9}$| × 10⁻³|$\frac{{\rm molecules}}{{\rm incident\ photon}}$|⁠), although photodesorption yields up to four times higher were measured for higher fluences.

Acknowledgements

This research was financed by the Spanish Ministry of Economy and Competitivity (MINECO) under projects AYA2011-29375 and AYA2014-60585.

1

The transfer of energy could also take place between molecules on the surface of the ice.

2

The term DIET could, in principle, encompass all photodesorption processes, and does not necessarily imply energetic transfer. However, these two concepts tend to be related in the literature (see Fayolle et al. 2011; Bertin et al. 2012). In addition, whren the absorbing and the photodesorbing molecules belong to different species during energy-transfer mediated processes, this mechanism is sometimes referred to as indirect DIET or photon-induced co-desorption, although no distinctions are made in general.

3

Not to be confused with the kick-out photodesorption in van Hemert et al. (2015).

4

The average absorption cross-section of pure NH₃ ice is 6.1 × 10⁻¹⁸ cm² (Cruz-Díaz et al. 2014a), which led to an absorption close to 100 per cent of incident photons in most of the experiments (see Table 1).

5

The ratio S(m/z)/S(28) is the same as the ratio |$k^{*}_{QMS} \times S (m/z)$|/|$k^{*}_{QMS} \times S (28)$|⁠.

6

The QCM measured the total ice mass-loss, which was assigned mostly to ejected NH₃ molecules, although it included any other photodesorbing product.