Chemical variations across the TMC-1 boundary: molecular tracers from translucent phase to dense phase

By comparing the observational abundances of the molecules with the calculations from the models, we could determine the best fit model and the chemical evolutionary timescale of the molecular cloud. We use the following Equation 1, which is known as the distance of disagreement (Vidal, et al., 2017), to calculate the mean logarithmic difference for all detected molecules at every time step.

where $X_{mod,i}(t)$ and $X_{obs,i}$ is the modeling abundance for species $i$ at time $t$ and observed abundance, respectively, and $N_{obs}$ is the total number of the observed species.

Figure 1 shows the calculated results for the initial carbon abundance of 2.5 $\times$ 10^-4 and a C/O ratio of 0.5. It can be seen that the best fit time is located at $\sim$5.3 $\times$ 10⁵ year. This best fit time is in well agreement within the range of 4.2$\pm$2.4 $\times$ 10⁵ year estimated from the CO depletion timescale toward 13 cores in Taurus (Pineda et al., 2010). Moreover, in order to show the diversity of the abundance between observational value and modeling value of each species, we plot in Figure 2 the logarithmic abundance ratio difference for each observed species. In the figure, the left endpoint and right endpoint for each molecule is the minimum and maximum value in the time range of 10⁴ to 10⁷ year respectively. The red cross mark corresponds to the time at best agreement, i.e. 5.3 $\times$ 10⁵ year. The vertical grey area between -1 and +1 represent one order of magnitude lower and higher respectively when compared with the observed values.

In Figure 2, there are 66 out of 93 species, which is 71 percent whose logarithmic abundance is within the range of -1 to +1 of observation at the best fit time. This percentage is similar to the results of earlier studies with a total observed species of $\sim$60 (Agúndez & Wakelam, 2013; Wakelam, et al., 2015). Specifically, for the 23 O group molecules, C₃O and CH₃O are about two orders of magnitude overproduced, while the large COMs (CH₃CHO, HCOOCH₃ and CH₃OCH₃) are underproduced. For the 22 C group molecules, there are 8 species out of the range from -1 to +1. Most of these out-of-range species are C_nH^- and C_nH₂ with n = 4, 6, 8. Note that all of the small hydrocarbons with fewer than 3 carbon elements are much closer to the observations than the large hydrocarbons. This trend is also seen in other models, which may indicate that the inefficient destruction pathways for the larger hydrocarbons when they grow complex. Nitrogen-bearing molecules in the N group are the most detected species in the TMC-1 CP. Among them, there are 5 out of 28 that are outside the range from -1 to +1. They are CN, C₅N, H₂CN, NH₂CHO and HCN, which are arranged from large to small disagreement with respect to their corresponding observed values. Comparing with the sequential arrangement for the overproduced small hydrocarbons and large hydrocarbons, there is no clear such an arrangement for these nitrogen-bearing molecules.

Finally, about a half of S group species (9 out of 20) are outside the range of one order of magnitude when compared with their observational values at the best fit time. Among them, CS⁺ is the most underproduced species, the predicted peak fractional abundance with respect to H is 2.2 $\times$ 10^-11 at $\sim$250 year and severely decreased after that. While the observed abundance of CS⁺ is 1.5 $\times$ 10^-10. As a comparison with CS⁺, CS is overproduced by two orders of magnitude at its best fit time. The predicted peak abundance of CS is 1.9 $\times$ 10^-6 at $\sim$3.9 $\times$ 10⁴ year and also decreases after that. While the observed abundance for CS is 1.5 $\times$ 10^-9. In the chemical reaction network, CS is mainly produced by HCS⁺, and destructed by H⁺ and H₃⁺. On the other hand, CS⁺ is mainly produced by reaction of CH with S⁺, and destructed by electron and H₂. Most calculated abundances of the sulfur-bearing species are larger than their observed values. The discussion of these discrepancies will be presented later.

We noticed that the predicted abundances of S group species are closer to their observed values when the C/O ratio increases. While the trend for other group species is on the opposite. The variation of C/O on the chemistry as expressed by the numbers of the disagreement species for the four groups can be seen in Figure 3, which corresponds to the model with an initial carbon abundance of 2.5 $\times$ 10^-4. The trends for the four groups are also presented for the low and intermediate initial carbon abundance models. The disagreed species are counted when the logarithmic abundance ratio differences between models and observations are out of the range from -1 to +1 (see Figure 2). Note that the counts are calculated for each model at their own best fit time.

From Figure 3 it can be seen that the C and O group species favor a C/O ratio of 0.5, and with the increase of the C/O ratio the disagreement rises highly. As opposed to the C and O groups, S group species favor a C/O ratio larger than 1.2, while the overall N group species favor a low C/O ratio similar to the C and O groups. However, the favored low C/O ratio for the overall N group species seems contradiction with the result of Maffucci, et al. (2018), which shows that a C/O ratio of 1.2 can better reproduce cyanopolyynes. We plot the mean logarithmic abundance difference for cyanopolyynes in Figure 4 using Equation 1 with $N_{obs}$ only include HC_nN, where n = 3, 5, 7, 9 at the time determined from the total observed species, i.e. the same evolutionary time used in the Figure 3. It can be seen that the minimum value is located at C/O = 1.0, which is close to 1.2, while our best fit model of C/O = 0.5 is in another minimum location. Generally, as concluded by Maffucci, et al. (2018), the increase of C/O enhances the production of cyanopolyynes.

Chemically speaking, at the carbon-poor, oxygen-rich condition (C/O $<$ 1), most of carbon are locked with oxygen either in the gas phase or on the grain surface mainly in the form of CO, CO₂, and H₂CO. While at the carbon-rich, oxygen-poor condition (C/O $>$ 1), part of carbon-bearing molecules is hydrogenated, thus both in the gas phase and on the grain surface, C_nH_m and HC_nN are overproduced, where n ranges from 3 to 9, and m = 2, 3, 4. The initial nitrogen is in the neutral atomic form and less abundant than carbon. It first reacts with hydrocarbons to form nitriles. Most of species in the N group are nitrile-containing molecules, thus the chemistry of N group species is also affected by the variation of C/O ratio and shows similar trend as with the C and O group species. However, for the subset of HC_nN with n = 3, 5, 7, 9 in the N group, their abundances could be increased by the overproduced large hydrocarbons when C/O $>$ 1, thus show a better fit with the observations than that at the conditions of C/O $<$ 1. The effect of the variation of C/O on the chemistry of sulfur-containing species in the S group is more complex as the elemental compositions of S group species are largely different from each other, which indicates that the reactions related to S group species are well connected with other group species. From Figure 3 it can be seen that the difference between the maximum and minimum counts for S group species is only 2, which seems not significant when compared with other groups. There are three species, H₂CS, H₂S and SO, whose abundances approach to their observed values as C/O increased. Instead, the abundance of C₃S move away from its observed value. For H₂CS and H₂S, they share similar production pathways when C/O ratio changes. However, their destruction involves C⁺ or C so that they overproduced when C/O $<$ 1. As for SO, it is produced by the reaction between S and OH, which could be affected by the C/O ratio. Generally, at C/O $<$ 1, most S group species are slightly overproduced; on the other hand, with the increase of C/O ratio, the abundances of these species tend to approach their observational values at the best fit time.

As the dominated material, hydrogen is the most abundant element in the ISM both in its atomic form and in its molecular form. In the initial condition of the models, we assume all hydrogen exists in its molecular form. Under this condition, all other species are evolved with time. We determine the best fit time (i.e 5.3 $\times$ 10⁵ year) for TMC-1 CP by fitting the abundances of all detected species with their predicted values. At this time, it can be seen from Figure 5 that the corresponding atomic hydrogen abundance at A_V $<$ 10 mag is $\sim$10^-3. The mean abundance of atomic hydrogen is 10^-2.8 in TMC and other molecular cloud regions determined by HI Narrow Self Absorption (HINSA, (Li & Goldsmith, 2003; Krčo & Goldsmith, 2010)). This consistency between predicted and observed HI abundance suggests that the above assumption is reasonable. If all hydrogen exists in its atomic form at the initial stage, the conversion timescale for hydrogen to its molecular form could be longer.

The ionization fraction, or the electron abundance ($x(e^{-})$), is a basic physical characteristic of the ISM that has a wide influence on the star formation from the driving of the ion-neutral chemistry to the dynamics between the gas and magnetic field. Goicoechea et al. (2009) estimated the gradients of ionization fraction in the Horsehead PDR regions from the outer edge to the shielded cores where the value changes from 10^-4 to 10^-9 respectively. Unlike the strong UV field in Horsehead PDR regions (35 in habing unit, (Goicoechea et al., 2009)), the UV field in TMC-1 regions is relatively weak (3.8 in habing uint). There has also a similar ionization fraction gradient (see the right panel in Figure 5). At the edge of TMC-1, the value of $x(e^{-})$ is $\sim$10^-5, which is contributed from C, S and cosmic-ray induced ionization of H₂. In the denser part of TMC-1, the ionization fraction is mainly controlled by the cosmic-ray induced ionization of H₂ because UV photons cannot penetrate into the clouds.

Except for the conversion of atomic hydrogen to molecular hydrogen in the transitional regions, the species of C⁺, C and CO are also widely recognized as the evolutionary tracers of these regions. Figure 6 shows the calculated abundances of these three species as a function of A_V. It can be seen that the abundance of C⁺ gradually decreases from the outer low extinction regions to the inner high extinction regions. Meanwhile, C⁺ are partially converted to C and CO, and the abundance of C peaks around $\sim$4 mag. The conversion of C⁺ to CO is efficient because of less strength of UV field in these regions and the shielding from H₂ and CO itself. With the continuing increase of A_V, eventually most of carbon are locked in the form of CO. Meanwhile, at high A_V and density, gas-grain chemistry plays an important role, and gas phase CO begins to adsorb onto the grain surface. Thus, we see a depletion of CO after A_V $>$ 10 mag. The rate of depletion depends on the gas density, with the higher the gas density, the greater the CO depletion rate. In the horizontal extensions of TMC-1 CP, it can be seen that at the location of offset 3 from the emission peak, CO is severely depleted, and this position represents the highest density. The effect of gas density also has an influence on the other gas phase species as can be seen in the following. Molecules that can be easily synthesized in the gas phase adsorb more quickly onto the grains when the gas density is high; while for the gas species with low synthesis rate, such as some large molecules, their abundances trend to be increased under high density conditions. In comparison, the densities for the horizontal extensions of TMC-1 C and TMC-1 NH₃ are generally decreased from the inner to the outer region. It should be noted that our results are presented under the same evolutionary times. Some studies have noticed the possible different evolutionary stages for the TMC-1 C and TMC-1 CP emission peak (Choi et al., 2017; Navarro-Almaida et al., 2021). The state of the chemistry could be affected by the different physical conditions and different evolutionary stages of the clouds. However, it is difficult to decouple the effects of these two factors on the chemistry as the different evolutionary stages may have implied the different physical conditions, and it is the focus of our next paper to explore this problem.

The main carriers of oxygen are the atomic oxygen (OI) and CO in the gas phase at A_V $<$ 10 mag. It can be seen in Figure 6 that the abundance of atomic oxygen is slightly higher than that of CO before its depletion. Recently, Goldsmith et al. (2021) explored the OI emission and absorption in the W3 high-mass star formation regions. They found that OI is dominated by the foreground absorption of low-excitation atomic oxygen, and its abundance could be comparable to, or greater than that of CO. The highest fractional abundance of OI with respect to H is 6 $\times$ 10^-5 under an initial oxygen abundance of 2.3 $\times$ 10^-4 in Goldsmith et al. (2021). In our results, the best fit model uses an initial oxygen abundance of 5.0 $\times$ 10^-4, and the abundance of OI could be maintained at a level of $\sim$2.0 $\times$ 10^-4 when A_V $<$ 10 mag. Thus, the atomic oxygen, along with C and C⁺, could be used as the chemical tracers to evaluate the physical properties of the translucent ISM.

Xu, et al. (2016a); Xu & Li (2016b) also showed that the primary hydrides CH and OH could be used as the gas tracers across the TMC boundary. They derived a fractional abundance (with respect to H₂) of $\sim$2 $\times$ 10^-8 and $\sim$2 $\times$ 10^-7 for CH and OH at A_V = 2 mag, respectively. Compared with our modeling results, our predictions for the abundances of CH and OH are underproduced by a factor of 5 and 30, respectively. Xu, et al. (2016a) interpreted the reason for the large discrepancy between the modeled and observational OH abundance as the shock, which is originated by the compression of the gas in the cloud’s edges. However, the above estimation is under the assumption that all hydrogen is in the form of H₂, i.e. the abundance of H₂ with respect to the total nuclei is 0.5. If we assume that a fraction of hydrogen is in the atomic form, which is reasonable under the cold neutral medium condition, then the above discrepancy between our predictions and the observations for the abundances of CH and OH could be replaced by a factor of 25 and 5, respectively, when a value of 0.3 for the abundance of H₂ with respect to the total nuclei is used. It can be seen that the fraction of atomic hydrogen could be important for the primary gas contents and their chemistry. Therefore, to better understand the chemistry of CH and OH, direct measurement of primary gas contents, such as using the HINSA to determine the property of clouds (Zuo et al., 2018), could be helpful.

The cosmic value of the sulfur elemental abundance is used so that we could study the sulfur chemistry across a wide range of A_V. We demonstrate in Figure 7 the calculated most abundant gas phase and grain surface S-bearing molecules. Similar to the trends of C⁺ and C as a function of A_V, S⁺ (shown in empty markers) remain its ionic atomic form when A_V $<$ 3 mag, and are converted to neutral atomic form (shown in filled markers) at higher A_V. When A_V $>$ 10 mag, sulfur is mostly adsorbed onto the grains and hydrogenated to JHS and JH₂S subsequently (hereafter, “J” stands for grain surface species) as shown in the upper panels of Figure 7. For example, at the TMC-1 CP, which corresponding to an A_V of 18.2 mag, 59 percent of the total sulfur abundance is in the form of JHS and JH₂S, and 20 percent is occupied by other grain surface species such as JHSCN, JHSO, JOCS and JNS, whereas 15 percent by the gas phase neutral atomic sulfur, and the remaining small amount is occupied by other S-bearing species. For the gas phase S-bearing species, they are mainly produced via the gas phase reactions with the exception of H₂S, which could be produced by the non-thermal desorption of grain surface JH₂S. Among them, CS, SO and SO₂ reach their peak abundances when A_V are around or less than 10 mag, while H₂S, CCS and OCS reach their peak abundances in the inner part of the cloud, which has large A_V and density. Moreover, it can also be seen in Figure 7 that the abundances of SO and SO₂ are more sensitive to the physical conditions at certain A_V, while other species are much less affected.

Compared with other abundant gas phase tracers, such as CO, there is a lack of abundant S-bearing molecular tracers in the gas phase for the sulfur chemistry. The abundance of CS reaches 10^-6 in the hot corino condition (Codella et al., 2021), and depletes to 10^-8 in the hot core and dark cloud (Luo et al., 2019; Bulut et al., 2021). In Figure 7, it can be seen that the CS abundance could reach the highest value of 10^-6 at A_V = 4 mag. However, the dominant sulfur-carriers above this level are atomic sulfur in the gas phase and JHS and JH₂S on the grain surface. The reason is twofold: the initial elemental S abundance is less than that for C, O and N, and S is heavier than C, O and N, making it easier to adsorb on the grains.

Figure 8 shows the gas phase N (in empty markers), N₂ (in filled markers) and other nitrogen-bearing species as a function of A_V. The initial nitrogen is in the neutral atomic form due to its slightly higher ionization potential (14.5 eV) than that of H atom. It can be seen in Figure 8 that the majority of nitrogen remains in its neutral atomic form when A_V is less than 10 mag, and mostly convert to N₂ at higher A_V. N₂ is not observable at infrared or millimeter-wavelength transitions, but can be estimated by the main production channel of N₂H⁺: N₂ + H${}_{3}^{+}$ $\rightarrow$ H₂ + N₂H⁺. Womack, Ziurys, & Wyckoff (1992) estimated a gas phase N₂ abundance of 4 $\times$ 10^-6 through N₂H⁺ under the dark cloud conditions, and suggested that gas phase N₂ as the dominant nitrogen-bearing species. In our models, for A_V $>$ 10 mag, the N₂ abundances varies between 5 $\times$ 10^-6 and 1 $\times$ 10^-5, which is less than 3 times of the observational value. N₂ is mainly produced by the reaction of N + CN $\rightarrow$ C + N₂ at earlier time ($\sim$5 $\times$ 10⁴ year), and by the cosmic-ray desorption of grain surface JN₂ at later time. Two of the destruction pathways of N₂ are the reaction of N₂ + H${}_{3}^{+}$ $\rightarrow$ H₂ + N₂H⁺ and the adsorption onto the grain surface. The production of NH₃ is also starting from N₂. Firstly, N⁺ is produced by the reaction of N₂ + He⁺ $\rightarrow$ N⁺ + N + He. Then, following the sequence of N⁺, NH⁺, NH${}_{2}^{+}$ and NH${}_{3}^{+}$ with the reaction of H₂, NH${}_{4}^{+}$ is produced. Finally, NH₃ is produced by the reaction of NH${}_{4}^{+}$ + e^- $\rightarrow$ H + NH₃. The family of cyanopolyynes, HC_nN (n = 3, 5, 7, 9) are produced by the reaction of CN with hydrocarbons of C_nH₂ (n = 2, 4, 6, 8). Overall, except for N₂, it can be seen from Figure 8 that the abundances of most nitrogen-bearing species gradually increase with larger A_V and peak at the highest density. Generally, the nitrogen-bearing species trace the regions with higher A_V and larger n_H.

By comparing the gas phase N₂ in Figure 8 with the grain surface JN₂ in the following Figure 10, it can be seen that the trends for N₂ and JN₂ are very similar, both having a turning point around A_V of 4–5 mag. The grain surface JN₂ is mainly produced by the grain surface reaction of JN + JNO $\rightarrow$ JO + JN₂, and also by the gas phase N₂ adsorption. The major consuming channel for JN₂ is its cosmic-ray desorption to the gas phase, which is independent of the A_V. Thus, the gas phase N₂ and grain surface JN₂ are well coupled through the adsorption and desorption mechanism. By comparing the estimated N₂ abundances in the dark clouds and in the warm clouds, which shows no difference, Womack, Ziurys, & Wyckoff (1992) suggested that dust grains may not play an important role in the N₂ chemistry. However, according to our above analysis, we suggest that dust grains indeed could play an important role in the N₂ chemistry by keeping N₂ on the grain surface and releasing back to the gas phase at a later time.

Complex organic molecules (COMs), which are defined as carbon-containing molecules with at least six atoms (Herbst & van Dishoeck, 2009), represent the advanced complexity during the development of the chemistry. Some COMs, such as CH₃OH and CH₃CHO, have been detected in the diffuse clouds toward Galactic center (Thiel et al., 2017) and in the translucent clouds (Turner, 1998; Turner, Terzieva, & Herbst, 1999). Thiel et al. (2017) estimated the abundances with respect to H₂ for CH₃OH, CH₃CHO and NH₂CHO in the diffuse clouds toward the line-of-sight of Sagittarius B2 of 8.1 $\times$ 10^-8, $<$3.2 $\times$ 10^-9 and $<$8.5 $\times$ 10^-10, respectively at a $\upsilon_{LSR}$ of 9.4 km/s. While the fractional abundances with respect to H₂ for CH₃OH and CH₃CHO in translucent clouds are in the range of 10^-9 to 10^-8 (Turner, 1998; Turner, Terzieva, & Herbst, 1999). However, the origin of these COMs in the diffuse and translucent clouds is still in debate. Liszt, et al. (2018) argued that the nature of the host gas environment could complicate this situation. Indeed, as suggested in Thiel et al. (2017), the diffuse clouds chemistry could be enriched by the mixing between the dense part and diffuse part of the gas.

Figure 9 shows the modelled abundances of the three COMs as a function of A_V. At intermediate A_V (around 5 mag), the modeling CH₃OH abundance is consistent with the observations, while CH₃CHO and NH₂CHO are underproduced. CH₃OH is first synthesized on the grain surface and then desorbed into the gas phase by the chemical process. The main chemical process for the production of gas phase CH₃OH is the chemical desorption of the exothermic surface reactions. Under a grain temperature of $\sim$10 K, the thermal desorption is not efficient, and other desorption mechanisms, such as the cosmic-ray desorption, also contribute less due to the high desorption energy of grain surface methanol. The origin of gas phase NH₂CHO is also in a similar way. However, CH₃CHO is mainly produced by the gas phase reactions of O + C₃H₇ $\rightarrow$ CH₃ + CH₃CHO and O + C₂H₅ $\rightarrow$ H + CH₃CHO. The destruction pathways for these COMs are mostly by the abundant atomic carbon and by the UV photodissociation. The high radiation field in the diffuse and translucent environment is one of the reasons for the low abundances of the COMs. On the one hand, COMs could be dissociated by the high energy photons. On the other hand, the grain surface chemistry, which is one of the channels that produce the formation of COMs, could also be suppressed by the high radiation. Nevertheless, our modeling results show that COMs indeed could be produced at an abundance of higher than 10^-11 in the translucent clouds with A_V down to 3 mag.

Water ice is widely present in the dense clouds based on the observations toward the line-of-sight dense part of clouds toward the background stars (Gibb et al., 2000, 2004). However, there was no strong evidence for the existence of solid-state water in the diffuse-dense transition region (Poteet, Whittet, & Draine, 2015). Recently, Potapov et al. (2020) shows that water ice could be expected in the diffuse ISM by the laboratory experiments. Although the mixture of grain ice mantle compositions is hard to determine, it is interesting to investigate the variation of grain ice compositions under different conditions for the exploration of grain surface chemistry and the role of gas-grain interactions. In Figure 10 we show the predicted abundances of grain surface species as a function of A_V. Water ice (JH₂O) is the most abundant species when A_V is higher than 4 mag. However, most JH₂O are dissociated by UV photons when A_V is less than 4 mag, thereby making its decrease in the grain ice mantles. Under this condition, the three most abundant ice compositions are JCO₂, JCO and JN₂. Among them, JCO₂ are the dominant ice composition at low A_V because of the reaction between JCO and JOH and between JO and JHOCO, where JOH is provided by the photodissociation of JH₂O. The abundance of JCO has a turning point around 4–5 mag resulting from the competition between its destruction and formation pathways. The destruction pathways of JCO are the reactions with JH and JOH, while the dominant formation pathways of JCO are the UV photon dissociation and cosmic-ray induced UV photon dissociation of grain surface JCO₂. Note that when A_V is less than 5 mag, the contribution of the gas phase CO adsorption is unimportant. When A_V is smaller than 5 mag, the abundance of JCO gradually increased because of the increased strength of UV photon and cosmic-ray induced UV photon dissociation of JCO₂. On the contrary, starting from the turning point, with the increase of A_V the effect of photodissociation of JCO₂ becomes weaker and gas phase CO adsorption becomes more important, thus leading to the increase of JCO abundance. After A_V $>$ 10 mag, part of JCO is hydrogenated to produce JHCO, JH₂CO and JCH₃OH gradually. It can be seen in Figure 10 that the abundances of JHCO, JH₂CO and JCH₃OH are more dependent on A_V and reach their highest values at the largest A_V considered in the models. On the other hand, JH₂O, JCO₂, JHCOOH, JHCN and JN₂ are generally kept constant when A_V is higher than 4 mag. JHCN and JN₂ are produced by the grain surface reaction of JH₂ + JCN $\rightarrow$ JH + JHCN and JN + JNO $\rightarrow$ JO + JN₂, respectively. They are both the dominant nitrogen-bearing molecules on the grain surface.

The grain ice mantle composition depends on many factors, such as the radiation field and n_H density. We show in Figure 11 the development of the grain ice mantle layers as a function of time in four positions at the horizontal line of TMC-1 C and its offset 3, 4 and 5 (see Table 1). It can be seen that the ice layers gradually decrease with the decrease of A_V. The ice layers for A_V of 2.2 mag are thicker than that for A_V of 4.8 mag at the time before 10⁵ year. This is due to two times higher of n_H density toward the position at A_V of 2.2 mag than that toward the position at A_V of 4.8 mag. We list in Table 3 the most abundant fractional ice composition at these positions. In comparison, we also list the observational values of the fractional ice comparison toward Elias 16 where the location is in TMC and the dust-embedded young stellar object W33A (Gibb et al., 2000). Water ice is the dominated ice composition at the dense part of the clouds. CO and CO₂ ice show comparable fractions in the ice mantles. Meanwhile, the hydrogenated H₂CO, CH₃OH, NH₃ and CH₄ ingredients increased the ice composition complex. However, in the diffuse-translucent part of the clouds, with the effect of the changes of physical conditions, such as radiation field and n_H density on the grain ice mantles, the grain ice compositions are greatly altered. Under these conditions, the CO₂ ice becomes the dominant ice composition, with H₂O, N₂ and CO ice as the main ingredients.