Characterization of thin film CO_2 ice through the infrared ν_1+ν_3 combination mode
aa r X i v : . [ a s t r o - ph . I M ] A ug MNRAS , 1–8 (2017) Preprint 16 November 2018 Compiled using MNRAS L A TEX style file v3.0
Characterization of thin film CO ice through the infrared ν + ν combination mode Jiao He, ⋆ Gianfranco Vidali, † Physics Department, Syracuse University, Syracuse, NY 13244, USA
Accepted XXX. Received YYY; in original form ZZZ
ABSTRACT
Carbon dioxide is abundant in ice mantles of dust grains; some is found in the purecrystalline form as inferred from the double peak splitting of the bending profile atabout 650 cm − . To study how CO segregates into the pure form from water-richmixtures of ice mantles and how it then crystallizes, we used Reflection AbsorptionInfraRed Spectroscopy (RAIRS) to study the structural change of pure CO ice as afunction of both ice thickness and temperature. We found that the ν + ν combina-tion mode absorption profile at 3708 cm − provides an excellent probe to quantify thedegree of crystallinity in CO ice. We also found that between 20 and 30 K, there isan ordering transition that we attribute to reorientation of CO molecules, while thediffusion of CO becomes significant at much higher temperatures. In the formation ofpure crystalline CO in ISM ices, the rate limiting process is the diffusion/segregationof CO molecules in the ice instead of the phase transition from amorphous to crys-talline after clusters/islands of CO are formed. Key words: methods: laboratory: solid state – infrared: ISM – ISM: molecules –ISM: evolution – solid state: volatile CO , along with H O and CO, is the main compo-nent of the ice mantle covering interstellar dust grainsin molecular clouds. In space, solid CO is observedin the mid-infrared by the bending mode ( ν ) at ∼
650 cm − (Gerakines et al. 1999; Pontoppidan et al. 2008;Ioppolo et al. 2013; Noble et al. 2013), the asymmetricalstretching mode ( ν ) at ∼ − (Gerakines et al. 1999;Nummelin et al. 2001; Noble et al. 2013), as well as thecombination modes ν + ν at 3708 cm − and 2 ν + ν at 3600 cm − (Gerakines et al. 1999; Keane et al. 2001). CO can also be observed via the ν mode at ∼ − (de Graauw et al. 1996; Boogert et al. 2000). The symmet-ric stretch ν is IR inactive. The abundance of solid CO with respect to H O ice ranges from 5% to 40%, witha median in the 20–30% range, in star formation regions(Boogert et al. 2015; Yamagishi et al. 2015). Because the ν mode is intense and is often saturated, the ν mode is oftenused instead to study the abundance and the physical andchemical environment of solid state CO . Pontoppidan et al.(2008) did a comprehensive survey of the ν mode of CO in various young stellar objects (YSOs) and found that the ⋆ E-mail: [email protected] † E-mail: [email protected] ν absorption profile can be separated into a few compo-nents representing different chemical and physical environ-ments for CO . Most interestingly, the double peak split-ting of the bending mode—the so-called Davydov split-ting (Davydov & Sardaryan 1962)—for pure crystalline CO is found in many lines of sight (Pontoppidan et al. 2008;Ioppolo et al. 2013; Noble et al. 2013). To use the doublepeak splitting as a thermal history probe, it is crucial tostudy in the laboratory how CO segregates into patchesof pure CO and then crystallizes. The segregation of CO from CO :H O:CH OH or CO :H O mixtures has beenstudied in several prior works (Sandford & Allamandola1990; Ehrenfreund et al. 1998, 1999; Gerakines et al. 1999;Palumbo & Baratta 2000; Hodyss et al. 2008; Ioppolo et al.2013; Isokoski et al. 2014; Cooke et al. 2016) using trans-mission FTIR spectroscopy. In these studies, and in the IRstudies of pure CO ice (see Kataeva et al. (2015) and ref-erences cited therein) the ice is much thicker than that ofactual ice mantles. It is known that crystallization is char-acterized by long range order in the solid; in thin films,it depends on the thickness, with the thinner films oftenbeing amorphous or partially amorphous with nanocrystals(Loerting et al. 2009). It is more realistic to study the segre-gation and crystallization in the thickness range comparablewith the thickness of ice mantles (less than a few tens of amonolayer (ML)) in the ISM. However, thin films transmis- c (cid:13) J. He & G. Vidali sion spectroscopy is not sensitive enough to give data withhigh enough signal to noise ratio. Reflection Absorption In-fraRed Spectroscopy (RAIRS) provides a more sensitive al-ternative to study the segregation and crystallization of CO in thin film ice mixtures, although in general RAIRS spec-tra has different absorption profile than transmission spec-tra (Baratta & Palumbo 1998). ¨Oberg et al. (2009) usedRAIRS and adopted a more realistic thickness ( <
40 ML)of CO :H O mixture to quantify the segregation of CO .They found that segregation of CO becomes significant be-tween 50 and 60 K. Most recently, He et al. (2017) mea-sured the binding energy of CO on water ice and on CO ice, and found that the former is weaker than the latter.Therefore with enough thermal energy CO is more likelyto bind to other CO molecules to form clusters instead ofbinding to water. He et al. (2017) obtained the diffusion en-ergy barrier of CO on the surface of compact amorphoussolid water. On a laboratory time scale the diffusion be-comes significant at above 60 K, which is in agreement withthe temperature found in prior studies of the segregationof CO from CO :H O mixtures (eg., Hodyss et al. 2008;¨Oberg et al. 2009).While the segregation of CO in water-based mixtureshas been studied by several groups, fewer details are avail-able of the crystallization process of pure CO ice at lowtemperature. Furthermore, there is some disagreement onthe results and their interpretation. Escribano et al. (2013)used theoretical modeling and laboratory measurements us-ing both RAIRS and transmission spectroscopy to studythe crystallization of pure CO ice. They found that evenfor CO ice deposited when the substrate is at 8 K, aslong as the ice film is not too thin, the ice is partly crys-talline. They also found that when an amorphous CO iceis heated up to 25 K, it becomes crystalline. From 25 Kto the desorption temperature, the ice structure does notchange, as inferred from the fact that there is no changein the infrared spectra in this temperature range. Thisis in contradiction with Gerakines & Hudson (2015) andIsokoski et al. (2013). Isokoski et al. (2013) measured highresolution transmission FTIR spectra of pure CO ice de-posited at 15 K as well as during heating up, and found thatthe ν mode double peak and the combination modes ν + ν and 2 ν + ν become narrower and sharper during heatingup from 15 K to 75 K. This indicates that the orderingin the ice changes during heating up. Gerakines & Hudson(2015) found that CO ice deposited at 10 K at the rate of0.1 µ m/hr, or about 200 times slower than in Isokoski et al.(2013), does not show any splitting in the bending mode.They claimed that the missing of splitting is attributedto the fact that their CO ice is amorphous. This lackof splitting in Gerakines & Hudson (2015) disagrees withboth Escribano et al. (2013) and Isokoski et al. (2013). Sincethese three groups used different thickness of CO ice, itremains a question whether the difference among them isdue to the difference in ice thickness and/or depositionrate (Escribano et al. (2013) used a slightly slower deposi-tion rate than Gerakines & Hudson (2015)). In this workwe study systematically how the CO ice crystallization de-pends on both ice thickness and temperature. We also lookfor all signatures of CO ice crystallization other than thesplitting of bending mode, hoping to find alternative probesof CO crystallinity. The experiments were carried out in a ultrahigh vacuum(UHV) setup located at Syracuse University. A base pres-sure of 2 × − torr can be obtained routinely. At the centerof the chamber there is a gold coated copper disk (the sam-ple) attached to the cold tip of a liquid helium cryostat. ALakeshore 336 temperature controller with a calibrated sili-cone diode and a resistance heater were used to measure andcontrol the temperature in the range of 8 – 500 K with an ac-curacy better than 50 mK. CO gas was deposited onto thesample disk from the background via a UHV precision leakvalve. A stepper motor controlled by a LabVIEW programwas used to drive the leak valve. The deposition dose wascalculated by the integration of pressure over time, assuming1 Langmuir (1 L, 1 × − torr · s) of exposure is equivalentto 1 monolayer (ML). Later we use ML and L interchange-ably. It is assumed that at a deposition temperature of 10 Kthe sticking of CO is unity (He et al. 2016), and the pres-sure at the ionization pressure gauge is the same as thatin front of the sample. The standard gas correction factorfor CO has been taken into account when calculating thedeposition dose. With the automated leak valve, the deposi-tion rate and dose can be controlled to relative uncertaintyless than 3%. The main uncertainty in thickness comes fromthe pressure reading of the ion gauge. A systematic error asmuch as 30% is possible in this type of gauges. For experi-ments in this work, a CO deposition rate of 4 L/minute wasadopted except for the deposition of water and CO mixture.The following thicknesses were attempted: 1, 2, 5, 10, 15, 20,30 L. After deposition at 10 K, the sample was heated up to100 K at a rate of 0.1 K/s to desorb the CO ice (Temper-ature Programmed Desorption). The infrared spectra of theCO ice was monitored by a Nicolet 6700 Fourier TransformInfraRed (FTIR) spectrometer in the Reflection AbsorptionInfraRed Spectroscopy (RAIRS) setup with an incidence an-gle of 78 degrees. The FTIR collects and averages 7 spectrafrom 600 cm − to 4000 cm − at a resolution of 1 cm − every10 seconds, both during deposition and during TemperatureProgrammed Desorption (TPD). We deposited 1, 2, 5, 10, 15, 20, and 30 L of CO onto thegold surface at 10 K, and then heated up the sample from10 K to 100 K with a ramp rate of 0.1 K/s. The asymmet-rical stretching mode ( ν ) absorption spectra for all thesethicknesses, normalized to the maximum of all the spectrafor the same thickness during TPD, are shown in Fig. 1.Fig. 2 shows the absorption spectra during TPD at selectedtemperatures for selected thicknesses (30, 15, 5, and 2 L).The following vibrational modes are shown: ν + ν modeat 3708 cm − , 2 ν + ν mode at 3600 cm − , asymmetricalstretching mode ν at ∼ − , ν mode for CO at ∼ − , bending mode ν at ∼
675 cm − . For ν and ν modes, the position of the absorption peaks are blue shiftedrespect to typical spectra measured in transmission. This isdue to the splitting of transverse optical (TO) and longitu-dinal optical (LO) modes. In RAIRS measurement, usuallythe LO mode is seen, while in transmission setup at nor-mal incidence only TO mode is excited. For polycrystalline MNRAS000
675 cm − . For ν and ν modes, the position of the absorption peaks are blue shiftedrespect to typical spectra measured in transmission. This isdue to the splitting of transverse optical (TO) and longitu-dinal optical (LO) modes. In RAIRS measurement, usuallythe LO mode is seen, while in transmission setup at nor-mal incidence only TO mode is excited. For polycrystalline MNRAS000 , 1–8 (2017) haracterization of CO ice films, in the reflection mode at normal incidence, a small LOpeak is seen as well (Kataeva et al. 2015). The LO and TOmodes are present both in amorphous and crystalline solids(Berreman 1963).We first analyze the ν mode. In Fig. 2 the second col-umn shows the ν profile for different temperatures. For allthe thicknesses, at the lowest temperature the peak is cen-tered at below 2380 cm − , typical of LO mode of disorderedCO . Depending on the thickness, at different temperaturesthe peak blue shifts to 2381 cm − , which is the typical po-sition for crystalline CO LO mode (eg., Escribano et al.2013). As seen in Fig. 1, for 30 L of CO , the first changehappens at 25 ± , it is at ∼
50 K. For the 30 L and 20 L ices, at above ∼
60 K the peakbecomes much narrower and the red wing almost disappears.This indicates the formation of long-range order (completecrystallization), as confirmed by the behavior of the ν + ν combination mode to be discussed below. The sensitivityto the macroscopic environment of the ν + ν combinationmode was noted before for nanoparticles (Bauerecker 2005).At the same time, the overall peak area of ν becomes muchsmaller. This can be explained if the orientation of CO molecules in the crystal are in a configuration that lowersthe LO absorption. If we assume in the amorphous phasethe average angle between the linear CO molecules and thesurface normal is 45 degrees, in the crystalline form thisangle is likely to be smaller than 45 degrees, and the LOmode (which excite vibration mode in the normal directionof the substrate surface) of asymmetrical stretching becomesless intense. At the same time we should expect to see anincrease in TO mode absorption. Indeed, Fig. 3 shows theincrease in TO mode absorption at 2343 cm − at ∼
60 K.In the RAIRS, TO mode absorption is usually much weakerthan LO because TO mode cancels out at the metal surfaceif the incident angle is close to 90 degrees. Another dramaticchange happens above 80 K during the desorption of CO ice. The peak red shifts from ∼ − to lower than 2380cm − , and it broadens. This may be because the symmetryis broken and the ice becomes disordered again, due to therapid movement of CO molecules during desorption.In other vibration modes, there are also significantchanges accompanying the structural changes. In the ν LOmode, amorphous CO shows a peak at ∼
676 cm − ; as theice crystallizes, it blue shifts to ∼
679 cm − . The presenceof the naturally occurring isotopic impurity CO providesanother powerful marker of the morphology of the film. The ν mode of CO shows a change from a broad band to anarrow peak at the same temperature of the onset of crys-tallization that was mentioned above about the ν mode of CO . Because of the low temperature at the transition ( ∼
25 K and higher for thinner films) there is no diffusion atthe laboratory time scale. Therefore, single CO moleculesremain in place, but their environment changes, and this isreflected by the sharpening of the ν mode. The peak sharp-ens with increasing degree of crystallinity while the positionremains the same at 2283 cm − . The combination modes ν + ν and 2 ν + ν have the same shape and their magni-tudes are proportional. In the ice deposited at low temper-ature and especially for the thinner films, the combinationmodes show a broad profile, as expected for an amorphoussolid. Here we use the word amorphous loosely, since there is no direct information on the degree of ordering of thedeposit. As an intermediate range order (Price 1996) struc-ture forms, a sharp component shows up, and becomes morepronounced as longer range order is formed. The degree ofcrystallinity is strictly positively related to the magnitudeof the sharp component. We suggest that these combinationmodes are excellent probe of CO ice crystallinity becausethey are very pronounced and easy to separate from thebroad amorphous component.Singling out one sharp component is numerically mucheasier than singling out a double peak. Another advantageof using the combination modes as a crystallinity probe isthat they show the same behavior in both transmission andRAIRS spectra (Bauerecker 2005). On the surface of dustgrains the combination modes of CO can also be directlycompared with RAIRS/transmission measurements withoutgrain shape corrections. To study the segregation of CO in ice mantle, thin ice layers are preferred and RAIRS ismore sensitive for such measurement. The ν and ν modesin RAIRS are dramatically different from the transmissionspectra. Combination modes provide a consistent profile re-gardless of the geometry of laboratory setup. The coinci-dence of sharp combination modes and bending mode split-ting in transmission spectra has been observed in prior stud-ies (Ehrenfreund et al. 1998; Hodyss et al. 2008), which ver-ifies the validity of using combination modes as indicationof crystallization, and also supports that this indication ofcrystallization applies to transmission spectra as well.Now we show how the ν + ν mode can be used asa probe to study in detail the crystallization of pure CO ice. The ν + ν absorption peak can be fit using two com-ponents, one sharp and narrow component centered at 3708cm − best fitted with a Lorentzian distribution with γ = 0 . − , one broad component fitted with a Gaussian distri-bution with σ > − . The Lorentzian component is thesignature of crystalline CO while the Gaussian componentrepresents amorphous (or disordered) CO ice. In a singlecrystalline CO ice only the Lorentzian component can beseen while in the other extreme, fully amorphous CO ice hasonly a broad Gaussian component. A typical fitting is shownin Fig. 4. Fig. 5 shows the area of both components duringthe deposition of the 30 L of CO ice when the substrate is at10 K. Below 10 L the ice is amorphous. As the thickness in-creases further the crystalline component emerges. This sug-gests that the underlying polycrystalline/amorphous goldsurface may affect the growth of crystalline CO ice. An al-ternative explanation is that partial crystallization requiresat least intermediate range order, and too thin a film at 10 Kcan not form the intermediate range order at 10 K. The re-sult in Fig. 4 can be extended to thicker ices. Isokoski et al.(2013) deposited 3000 ML of CO when the substrate wasat 15 K, and the ice is polycrystalline, which agrees qualita-tively with Fig. 4.In order to better quantify the crystallinity, we define anew parameter—degree of crystallinity ( DOC ) as:
DOC = A crystalline A crystalline + A amorphous (1)where A crystalline and A amorphous are the absorptionstrength of the Lorentzian component and Gaussian com-ponent, respectively. Fig. 6 shows the DOC as a functionof temperature during TPD for different thickness CO ices. MNRAS , 1–8 (2017)
J. He & G. Vidali t e m p e r a t u r e / K
30 L 2360238020 L 2360238015 L 23602380wavenumber / cm −
10 L 236023805 L 236023802 L 236023801 L scale 00.250.500.751.0
Figure 1.
Normalized absorption spectra of the CO asymmetrical stretching ( ν ) mode during TPDs. The thickness of the CO ice ismarked on the top of each panel. The colorbar scale, on the right side of the figure, shows relative intensity. In thinner than 5 L ices, the combination modes are toonoisy and we don’t quantify the
DOC . There is a dramaticincrease in
DOC between 20 and 30 K for all of the thick-nesses, which suggests a significant crystallinity change inthis temperature range. We attribute this change to orien-tational changes of CO molecules in this temperature range:CO molecules reorient and forms intermediate range order(nanocrystals). There are few studies on orientational order-ing in CO cubic ice (Torchet et al. 1996; Kuchta & Etters1988; Krainyukova & Kuchta 2017). Solid, crystalline CO ice has Pa¯3 symmetry in vacuum with four molecules per ele-mentary cell placed along the cubic diagonals. The moleculesperform small librations around the diagonal (with the Catom on the diagonal) (Kuchta & Etters 1988). A recentTHEED (Transmission High Energy Electron Diffraction)study on thin ( ∼
10 nm) films deposited at ∼
65 K reveals amore complicated rotational motions with the molecular tips(the oxygen atoms) hopping in 24 equivalent positions witha maximum deviation from the diagonal of about 30 degreesand decreasing from 15 to 70 K (Krainyukova & Kuchta2017).In the growth of 30 and 20 L ices, although during depo-sition the sample is at 10 K, gas-phase CO molecules pos-sess room temperature thermal energy, and therefore CO molecules can reorientate right after landing on the surfaceand form nanocrystals. For thinner ices, it is more difficultto form structures with long-range order. In the extremecase where the ice is very thin (1–2 L), molecules have todiffuse on the substrate surface and form clusters or islandsbefore an intermediate range order (nanocrystalline) struc-ture can be formed. From Fig. 1 it can be seen that for 1 Lof CO the transition temperature is at 50 K, which is closeto the temperature of diffusion of CO on a compact waterice surface (He et al. 2017). In very thin CO ices, the crys-tallization temperature is mostly limited by diffusion. Thesame conclusion should also apply for low concentration ofCO mixed in water ice. For ice thicker than 5 L, at higher temperatures, the DOC value increases gradually to unityat about 65 K, and ice becomes almost all crystalline. Theemerging ν TO mode absorption at about 60 K (Fig. 3)also supports the formation of long range order.
The infrared spectra of pure CO ice depositedat low temperature has been measured previously,e.g. (Sandford & Allamandola 1990; Edridge et al.2013; Escribano et al. 2013; Isokoski et al. 2013;Gerakines & Hudson 2015); here we compare our re-sults with some of the prior studies. In Isokoski et al.(2013), Fig. 5 shows that the 3000 ML ice grown at 15 Kalready demonstrates a certain degree of crystallinity and isin the polycrystalline form, in agreement with our results.Between 15 K and 75 K, the degree of crystallinity increases.At 75 K the ice is likely fully crystallized. In Escribano et al.(2013), Fig. 2 shows that the double peak splitting of thebending mode emerges between 20 K and 25 K; this alsoagrees with our DOC vs. temperature curve. However, Fig.2 in Escribano et al. (2013) shows almost no difference inthe bending profile above 25 K until desorption of the ice.This is different from our measurement and Isokoski et al.(2013)’s. It is probably due to the relatively low signalto noise ratio in Escribano et al. (2013); therefore smallchanges in the absorption profile can not be recognized.Fig. 1 of Escribano et al. (2013) also shows an increase indegree of crystallinity during deposition when the ice isstill thin, in agreement with Fig. 5 of our work. In Fig. 6of Escribano et al. (2013) the 200 ML of CO depositedat 14 K already shows some character of crystalline ice,and this also agrees with our results. In Edridge et al.(2013)’s experiments, CO was deposited at 28 K, whichis about the transition temperature from the amorphousphase. Their ν profiles show a sharp peak at all temper-atures during TPD, indicating (poly)crystalline structure. MNRAS000
The infrared spectra of pure CO ice depositedat low temperature has been measured previously,e.g. (Sandford & Allamandola 1990; Edridge et al.2013; Escribano et al. 2013; Isokoski et al. 2013;Gerakines & Hudson 2015); here we compare our re-sults with some of the prior studies. In Isokoski et al.(2013), Fig. 5 shows that the 3000 ML ice grown at 15 Kalready demonstrates a certain degree of crystallinity and isin the polycrystalline form, in agreement with our results.Between 15 K and 75 K, the degree of crystallinity increases.At 75 K the ice is likely fully crystallized. In Escribano et al.(2013), Fig. 2 shows that the double peak splitting of thebending mode emerges between 20 K and 25 K; this alsoagrees with our DOC vs. temperature curve. However, Fig.2 in Escribano et al. (2013) shows almost no difference inthe bending profile above 25 K until desorption of the ice.This is different from our measurement and Isokoski et al.(2013)’s. It is probably due to the relatively low signalto noise ratio in Escribano et al. (2013); therefore smallchanges in the absorption profile can not be recognized.Fig. 1 of Escribano et al. (2013) also shows an increase indegree of crystallinity during deposition when the ice isstill thin, in agreement with Fig. 5 of our work. In Fig. 6of Escribano et al. (2013) the 200 ML of CO depositedat 14 K already shows some character of crystalline ice,and this also agrees with our results. In Edridge et al.(2013)’s experiments, CO was deposited at 28 K, whichis about the transition temperature from the amorphousphase. Their ν profiles show a sharp peak at all temper-atures during TPD, indicating (poly)crystalline structure. MNRAS000 , 1–8 (2017) haracterization of CO ice . . . . . . . L . . . . . . . . . . . . . . . . . . . . . . . . . . . . L . . . . . . . . . . . . . . . . . . . . . . . . . . . . L . . . . . . . . . . . . . . . . . . . . . . L . . . . . . . . . . . . . . . . . wavenumber / cm − ν + ν ν + ν ν CO ν ν a b s o r b a n ce Figure 2.
RAIRS of 30, 15, 5, and 2 L of CO at selected temperatures. The thickness is labeled on the left side of each row, and thevibrational modes are labeled on the top of each column. Spectra are offset for clarity. The temperature for each curve is also marked.The ν mode of CO (mixed in CO with natural abundance) is also shown. Gerakines & Hudson (2015) also studied the phase of CO ice, and found that their CO ice deposited at a rate of 0.1 µ m hr − (200 times slower than in (Isokoski et al. 2013))has no sign of crystallization but doubling this depositionrate yields crystalline CO . This disagrees with our work as well as with Escribano et al. (2013)’s and Isokoski et al.(2013)’s experiments. Given the standard methods used inall these experiments, one is led to conclude that perhapsthe explanation lies in the non ultra-high vacuum condi-tions used in Gerakines & Hudson (2015)’s experiments. MNRAS , 1–8 (2017)
J. He & G. Vidali . . . . . . . − )0 . . . . . . . . . a b s o r b a n ce Figure 3.
Zoom-in of the region of CO ν TO mode for 30 LCO ice during TPD at selected temperatures. − . . . . . a b s o r b a n ce experimentalfittingcrystallineamorphous Figure 4.
A typical fitting of the ν + ν combination mode profileusing Lorentzian (cyan) and Gaussian (green) distributions. . . . . . . . a b s o r p t i o n s tr e n g h / c m − crystallineamorphous Figure 5.
The area of Lorentzian (crystalline) and Gaussian(amorphous) components during the deposition of 30 L CO ongold substrate at 10 K.
10 20 30 40 50 60 70 80temperature / K0 . . . . . . d e g r ee o f c r y s t a lli z a t i o n
30 L20 L15 L10 L5 L
Figure 6.
Degree of crystallinity (see text) in CO ices of differentthickness during TPD. This discrepancy could be due to contamination frombackground water in the vacuum chamber during CO deposition. The effect of water in CO ice has been shownto wipe out the sharp features of both ν and ν (e.g.,Cooke et al. (2016)). A clue that this might be the caseresides in Gerakines & Hudson (2015)’s experiments when,by increasing the deposition rate (and, thus, decreasingthe deposition time and degree of contamination), theamorphous-like features (broad ν profile and unresolved ν splitting) change into crystalline-like ones (sharp ν and ν split). To check the effect that water has on the CO icespectrum, we performed an experiment with a 1:10 mixtureof H O:CO at a CO deposition rate of 1 ML/minute.The result is shown in Figure 7. Figure 8 shows the same ν + ν mode for pure CO ice. Clearly the mixture withwater doesn’t show any sign of being crystalline below ∼
40 K, while the pure CO ice shows a sharp peak from10 K. When heating it, the mixture with water crystallizes.This agrees with Gerakines & Hudson (2015)’s finding thatannealing to 70 K crystallizes the ice. Therefore, we suggestthat ultrahigh vacuum conditions are important for thestudy of CO crystallization. The segregation and crystallization of CO in ice mantlesis a useful probe of the thermal history of ices in densemolecular clouds. While prior works analyzed the doublepeak splitting of bending profile in efforts to study thesegregation and crystallization CO in CO :H O mixtures(Pontoppidan et al. 2008; Ioppolo et al. 2013; Noble et al.2013) the splitting profile suffers from the difficulty in sep-arating the double peak feature from CO in other environ-ments. In addition, the shape of the bending mode profileis affected by grain shapes. We propose that the ν + ν combination mode profile is a more sensitive probe of purecrystalline CO and, because of its weakness, is not sen-sitive to dust grain shape effects as the other transitionsare (Keane et al. 2001). Another advantage of the ν + ν combination mode profile is that it is similar in both trans-mission spectra and reflection spectra, therefore can be used MNRAS000
40 K, while the pure CO ice shows a sharp peak from10 K. When heating it, the mixture with water crystallizes.This agrees with Gerakines & Hudson (2015)’s finding thatannealing to 70 K crystallizes the ice. Therefore, we suggestthat ultrahigh vacuum conditions are important for thestudy of CO crystallization. The segregation and crystallization of CO in ice mantlesis a useful probe of the thermal history of ices in densemolecular clouds. While prior works analyzed the doublepeak splitting of bending profile in efforts to study thesegregation and crystallization CO in CO :H O mixtures(Pontoppidan et al. 2008; Ioppolo et al. 2013; Noble et al.2013) the splitting profile suffers from the difficulty in sep-arating the double peak feature from CO in other environ-ments. In addition, the shape of the bending mode profileis affected by grain shapes. We propose that the ν + ν combination mode profile is a more sensitive probe of purecrystalline CO and, because of its weakness, is not sen-sitive to dust grain shape effects as the other transitionsare (Keane et al. 2001). Another advantage of the ν + ν combination mode profile is that it is similar in both trans-mission spectra and reflection spectra, therefore can be used MNRAS000 , 1–8 (2017) haracterization of CO ice − . . . . . . . a b s o r b a n ce Figure 7. ν + ν mode of 22 ML of 1:10 H O:CO mixturedeposited at 10 K and heated up at 0.1 K/s. The temperature foreach spectrum is labeled. − . . . . . . . a b s o r b a n ce Figure 8. ν + ν mode of 20 ML of pure CO ice deposited at 10K and heated up at 0.1 K/s. The temperature for each spectrumis labeled. by RAIRS measurements and provides more sensitive mea-surements of ice films with a thickness comparable with theice mantle. Although very few previous observations of icemantles provide enough high resolution data covering thecombination mode at 2.7 µ m (Keane et al. 2001), JWST willprovide high quality spectrum data covering this wavelengthregion. More reliable information of the thermal history ofice mantles can be expected from comparing observationalspectra with laboratory measurements.We also found that there is a dramatic increase in thedegree of crystallinity between 20 and 30 K, and we at-tribute it to orientational ordering of CO molecules. Weintroduce a new parameter—degree of crystallinity (DOC)—to reliably quantify crystallinity of CO ice. Above 30 K,the CO ice further crystallizes until it becomes fully crys-tallized at about 65 K. We can separate the formationof pure crystalline CO in ice mantles into two processes:1) segregation/diffusion of CO to form clusters/islands ofCO ; 2) the formation of long range order in CO clus- ters/islands. The latter process becomes efficiently at above20 K, while the former process has been found to happen at amuch higher temperature (He et al. 2017; ¨Oberg et al. 2009;Hodyss et al. 2008). We therefore concludes that the ratelimiting process is the segregation of CO from the CO :H Omixture instead of the crystallization process itself. In thecase of a low fraction of CO in the mixture, the diffusionof CO on water surface, which has been studied in detailby He et al. (2017), becomes the most important process inthe formation of pure crystalline CO . ACKNOWLEDGEMENTS
We thank SM Emtiaz, Yujia Huang, and Francis Toriello fortechnical assistance. This research was supported by NSFAstronomy & Astrophysics Research Grant
REFERENCES
Baratta G. A., Palumbo M. E., 1998, JOSAA, 15, 3076Bauerecker S., 2005, Physical Review Letters, 94, 033404Berreman D., 1963, PhRv, 130, 2193Boogert A. C. A., et al., 2000, A&A, 353, 349Boogert A. C. A., Gerakines P. A., Whittet D. C. B., 2015,ARA&A, 53, 541Cooke I. R., Fayolle E. C., ¨Oberg K. I., 2016, ApJ, 832, 5Davydov A. S., Sardaryan R. A., 1962, NucPh, 37, 106Edridge J. L., Freimann K., Burke D. J., Brown W. A., 2013,RSPTA, 371, 20110578Ehrenfreund P., Dartois E., Demyk K., D’Hendecourt L., 1998,A&A, 339, L17Ehrenfreund P., et al., 1999, A&A, 350, 240Escribano R. M., Munoz Caro G. M., Cruz-Diaz G. A., Rodriguez-Lazcano Y., Mate B., 2013, PNAS, 110, 12899Gerakines P. A., Hudson R. L., 2015, ApJ, 808, L40Gerakines P. A., et al., 1999, ApJ, 522, 357He J., Acharyya K., Vidali G., 2016, ApJ, 823, 56He J., Emtiaz S. M., Vidali G., 2017, ApJ, 837, 65Hodyss R., Johnson P. V., Orzechowska G. E., Goguen J. D.,Kanik I., 2008, Icarus, 194, 836Ioppolo S., Sangiorgio I., Baratta G. A., Palumbo M. E., 2013,A&A, 554, A34Isokoski K., Poteet C. A., Linnartz H., 2013, A&A, 555, A85Isokoski K., Bossa J.-B., Triemstra T., Linnartz H., 2014, PCCP,16, 3456Kataeva T. S., Kolomiitsova T. D., Shchepkin D. N., Asfin R. E.,2015, CPL, 641, 117Keane J. V., Boogert A. C. A., Tielens A. G. G. M., EhrenfreundP., Schutte W. A., 2001, A&A, 375, L43Krainyukova N., Kuchta B., 2017, JLTP, 187, 148Kuchta B., Etters R., 1988, PhRvB, 38, 6265Loerting T., Brazhkin V. V., Morishita T., 2009, AdChP, 143, 29Noble J. A., Fraser H. J., Aikawa Y., Pontoppidan K. M., SakonI., 2013, ApJ, 775, 85Nummelin A., Whittet D. C. B., Gibb E. L., Gerakines P. A.,Chiar J. E., 2001, ApJ, 558, 185¨Oberg K. I., Fayolle E. C., Cuppen H. M., van Dishoeck E. F.,Linnartz H., 2009, A&A, 505, 183Palumbo M. E., Baratta G. A., 2000, A&A, 361, 298Pontoppidan K. M., et al., 2008, ApJ, 678, 1005Price D., 1996, COSSM, 1, 572Sandford S. A., Allamandola L. J., 1990, ApJ, 355, 357Torchet G., de Feraudy M.-F., Boutin A., Fuchs A. H., 1996,J. Chem. Phys., 105, 3671MNRAS , 1–8 (2017)
J. He & G. Vidali
Yamagishi M., Kaneda H., Ishihara D., Oyabu S., Onaka T., Shi-monishi T., Suzuki T., 2015, ApJ, 807, 29de Graauw T., et al., 1996, A&A, 315, L345This paper has been typeset from a TEX/L A TEX file prepared bythe author. MNRAS000