Probing Microporous ASW with Near-infrared Spectroscopy: Implications for JWST's NIRSpec

At cold temperatures (<100 K), water vapor condenses to form amorphous solid water (ASW; Blackman & Lisgarten 1958). ASW will contain pores, and the characteristics of these pores will depend strongly on the growth conditions (Sceats & Rice 1982; Brown et al. 1996; Stevenson et al. 1999; Horimoto et al. 2002). Porous ASW is believed to be of particular importance for astronomical environments, as the pores within the ice can trap volatile gas species and may promote gas-solid phase chemistry altering the chemical history of the ice (Bar-Nun et al. 1985; Mayer & Pletzer 1986). One consequence of ASW's porosity and large internal surface area is that the intermolecular bonding at the internal surfaces can differ compared to what occurs in the bulk of the ice (Mayer & Pletzer 1986). More specifically, the water molecules at the pore surface partially dangle off into the pore and cannot participate in lattice bonding. These dangling bonds (DBs) are characteristic of porous ASW and are typically identified in laboratory studies using the pair of infrared (IR) bands centered at 3696 and 3720 cm⁻¹ (or 2.705 and 2.688 μm; Buch & Devlin 1991; Rowland et al. 1991). These bands have been assigned to vibrations of three-coordinated (one free hydrogen bond) and two-coordinated (two free hydrogen bonds) water molecules.

Although the temperatures within the interstellar medium (ISM) and the outer solar system are sufficiently low to form porous ASW from ambient condensation of H₂O, the DB IR spectral features have not been detected anywhere. Previous laboratory studies have demonstrated that energetic particle bombardment can efficiently destroy the DB absorption bands (Palumbo 2006; Raut et al. 2007b; Dartois et al. 2013; Mejía et al. 2015; Behr et al. 2020), as well as compact the pore volume (Palumbo 2006; Raut et al. 2007b, 2008; Behr et al. 2020). In fact, most recent laboratory estimates (Behr et al. 2020) suggest that the DB features will be destroyed 10–100 times faster than the average lifetime of a molecular cloud (Blitz & Shu 1980), which is consistent with these features' absence in remote sensing spectra.

The absence of the DB features in remote sensing spectra taken to date could also be a consequence of other more basic factors. For instance, the proximity between the fundamental DB bands and the fundamental OH stretch of H₂O-ice could inhibit clear detection of these features, especially in spectra that do not have the high signal-to-noise ratios (SNRs; ∼10–100) that are typically observed in laboratory spectra. Moreover, IR-active species, such as CO₂, which have absorption features in this spectral region, may also inhibit clear detections of DB features (Dartois et al. 2013). Thus, in this study, we were interested in determining whether we could confidently identify any other absorption bands present in the spectrum of ASW that could be attributed to the DBs and hence identify porous ASW. We note that near-infrared spectral features for the DBs, where overtones and combination modes typically occur, are expected. Yet, to our knowledge, there are only a few instances in literature where these potential features are either seen or tentatively identified (Schmitt et al. 1998; Mastrapa et al. 2008; Zheng et al. 2009). For instance, two absorption features are observed near 5320 and 7230 cm⁻¹ by Schmitt et al. (1998) but not identified, while in more recent works, one or two features are seen and tentatively assigned to the DB absorptions (Mastrapa et al. 2008; Zheng et al. 2009).

In an effort to confidently identify any new potential DB absorption bands, we deposited ASW films between ∼20 and 150 K and studied their near-infrared spectra in detail after growth, during warming, and during adsorption of methane. Finally, we point out that this work was also motivated by the upcoming launch of the James Webb Space Telescope (JWST). The near-infrared spectrometer (NIRSpec) instrument on JWST is expected to have a significantly better sensitivity and resolution than its predecessors (Bagnasco et al. 2007; Milam et al. 2016), opening up new opportunities for detection of previously unidentified species, as well as more robust characterizations of the physical properties of the icy materials in extraterrestrial environments.

Experiments were performed within a stainless steel ultra-high vacuum chamber with a base pressure of 1.5 × 10⁻⁹ Torr. The specifications of this chamber are detailed in Tribbett & Loeffler (2021); modifications to include reflectance spectroscopy are similar to those described in Loeffler et al. (2020). To prepare our ASW samples, we background deposited H₂O (high-performance liquid chromatography (HPLC) grade, Sigma Aldrich) at 21 K onto an optically flat gold-mirror electrode of an Inficon IC6 quartz-crystal microbalance (QCM) as described in our previous studies (Loeffler et al. 2020). Note that in all cases, the thermal-radiation shield was removed to allow molecules from all angles to deposit on the sample, which should more closely mimic an extraterrestrial environment, while maximizing the sample's porosity (Loeffler et al. 2016). Samples were deposited at a rate of ∼4 × 10¹⁴ H₂O cm⁻² s⁻¹ to a column density of 4.13 × 10¹⁸ H₂O cm⁻² or an ice thickness of ∼1.76 μm assuming a density of 0.70 g cm⁻³ (Behr et al. 2020).

For analysis, the specular reflectance between 7500 and 2000 cm⁻¹ was acquired using a halogen light source from a Thermo-Nicolet iS50 Fourier Transform Infrared Spectrometer (FTIR) and an MCT-A detector at a spectral resolution of 2 cm⁻¹. The incident angle of the light source was 37.5° with respect to the surface normal of our gold-mirror QCM. IR spectra reported here are given in units of optical depth, −ln(R), where R is the ratio of the intensity of reflected light from the sample divided by the intensity reflected from the bare substrate. All reflectance spectra were collected without rotating the substrate (i.e., collected at the same viewing geometry).

To determine temperature dependence of the spectral features, H₂O-ice samples were warmed at a rate of 2 K minute⁻¹ and allowed to equilibrate at the acquisition temperature before taking an IR spectrum. To quantify the band area of any DB absorption bands, we used a fifth-order polynomial to reproduce and remove the baseline continuum from the absorption region prior to integration, as we have done in previous studies (Loeffler et al. 2006a; Behr et al. 2020). In some experiments, we also deposited methane (ultra-high purity, 99.97%, Matheson Gas) onto the ASW samples via background deposition at 21 K at a rate of ∼2 × 10¹⁴ CH₄ cm⁻² s⁻¹, which was monitored using the QCM. The low deposition temperatures allowed us to avoid competition between adsorption and desorption of the CH₄ (Raut et al. 2007a).

3.1. Thermal Stability of DB Absorption Bands

Figure 1 shows the near-infrared spectrum of an ASW sample after deposition at 21 K and during warming to 150 K. As expected, the spectra are dominated by two broad absorption bands, which are due to combination (ν₂ + ν₃; 5041 cm⁻¹) and overtone (2ν₃; 6748 cm⁻¹) modes of H₂O-ice (Gerakines et al. 2005; Mastrapa et al. 2008). The sharp rise beginning near ∼3800 cm⁻¹ is due to the fundamental stretching mode of H₂O-ice. These near-infrared features change slightly during warming with the most drastic changes being due to the crystallization of ASW that occurs at higher temperatures (Jenniskens & Blake 1994).

Zoom In Zoom Out Reset image size

In addition to these large absorption features, there are three spectral regions that contain smaller absorption bands that appear at higher wavenumbers than each large feature, which we show in Figure 2. It is well-known that the bands located at 3720 cm⁻¹ (DB1) and 3696 cm⁻¹ (DB2) are indicative of dangling bonds present in two- and three-coordinated H₂O molecules (Buch & Devlin 1991; Raut et al. 2007a; Dartois et al. 2013; Mejía et al. 2015; Behr et al. 2020). In addition, we also assign these other smaller features present in Figures 2(a) and (b) to dangling bond absorptions (see Table 1 for more details). The absorption band centered at 5326 cm⁻¹ (DB3) is consistent with what was previously tentatively identified as a DB combination feature at 5313 cm⁻¹ (Zheng et al. 2009) and at 5320 cm⁻¹ (Mastrapa et al. 2008), as well as what was observed in one of our recent works (see Figure 3 in Loeffler et al. 2020). Furthermore, we suspect that the DB3 combination band is due to both two- and three-coordinated H₂O molecules, as previous studies have shown that the peak position of the O–H stretching and H₂O bending modes shift in opposite directions as the coordinate state changes (Falk 1984), which likely prevents the two transitions from being resolved in our spectra. Finally, there appear to be two absorptions located at 7285 cm⁻¹ (DB4) and at 7235 cm⁻¹ (DB5). Similar to DB3, this is consistent with the tentative identification of a feature at 7230 cm⁻¹ (Mastrapa et al. 2008), although the smaller absorption feature (DB4) was not reported in that work. By comparing the band areas of each DB absorption region, we estimate that DB3 and DB4+DB5 are ∼20 times smaller than DB1+DB2, which is typical of an overtone or combination mode absorption (Hudson et al. 2017). Admittedly, it would be better to refine this estimate in future laboratory measurements, as the relative strength of the absorption features can change with film thickness in our setup, which is a consequence of thin film interference effects present in reflectance spectroscopy (Teolis et al. 2007).

Zoom In Zoom Out Reset image size

Table 1. Compilation of Previous Laboratory Work Characterizing Peak Positions (cm⁻¹) of the NIR DB Features: ^aMastrapa et al. (2008), 25 K, ^bZheng et al. (2009), 10 K, and ^c this Work, 21 K

Identification	References			SNRd
	a	b	c
DB2 (DB1)+ ${\nu }_{2}^{e}$	5320	5313	5326	31.06
2DB2	7230	⋯	7235	10.83
2DB1	⋯	⋯	7285	3.08

Note. See Palumbo (2006) and Mejía et al. (2015) for a review of the fundamental DB features. ^dSNRs are calculated by dividing the absorption feature's peak height by the peak-to-peak signal adjacent to the absorption feature (noise). These SNR values are for this work only. Note that these features can also be seen in Figure 3 in Schmitt et al. (1998); however, band centers are not reported. ^eAssignment for ν₂ is from Hagen et al. (1981).

Download table as:  ASCIITypeset image

As can be most clearly seen in Figure 2, all DB features weaken during warming, eventually dissipating. Both shoulder bands (DB1 and DB4) are unidentifiable at temperatures warmer than 70–90 K, supporting that they are both indicative of two-coordinated water molecules. DB2, DB3, and DB5 dissipate by 150 K; generally agreeing with the temperature dependence of the larger DB feature (DB2) shown in earlier studies (Horimoto et al. 2002). We point out all of these DB features are absent, as expected, within the crystalline H₂O-ice reference sample (Figure 2, bottom spectrum), which was background deposited at 150 K and acquired at 21 K.

We also studied the decrease in the DB absorption bands during warming more quantitatively by measuring the normalized band areas for each DB region as a function of temperature (Figure 3(a)). Band areas were normalized to unity based on the band areas at 21 K. Note that both DB pairs (DB1, DB2, and DB4, DB5) overlapped enough that we calculated the total band area for those spectral regions. As can be seen in Figure 3(a), the normalized band area for each region decreases as a function of temperature at a nearly identical rate, further supporting that all these features are indeed due to dangling bonds in ASW.

Zoom In Zoom Out Reset image size

3.2. Gas Adsorption

To further confirm that these new absorption bands are due to dangling bonds in ASW, we adsorbed CH₄ onto the surface of our ASW samples after deposition at 21 K. This effort was motivated by previous work that has shown the position of DB1 and DB2 shifts to lower wavenumbers as gas attaches to the DBs (Rowland et al. 1991; Chaabouni et al. 2000; Palumbo & Strazzulla 2003; Raut et al. 2007a). Figure 4(a) shows the spectra of DB1 and DB2 during deposition of CH₄. Previous studies have shown that CH₄ adsorbs onto the pore surfaces of ASW prior to condensing on the surface of the sample as a thin film (Raut et al. 2007a). The appearance of the 2904 cm⁻¹ CH₄ absorption feature (Figure 4(b)) during adsorption supports this expectation. This feature is only found in amorphous CH₄ or amorphous H₂O + CH₄ mixtures (Hudson et al. 2015; Mejía et al. 2020) and our relatively high deposition temperature (21 K) should produce crystalline CH₄ if it deposited as a multilayered thin film. Thus, we estimate the CH₄ uptake into the pores with our QCM data. As expected, DB1 and DB2 shift by approximately 30 cm⁻¹, agreeing with previous studies (Palumbo 2006; Raut et al. 2007a). Interestingly, the percentage of CH₄ required to induce a spectral shift in our sample is greater in our experiments than reported previously (Raut et al. 2007a). While this difference is not entirely clear, it seems likely due to slightly different experimental parameters (thickness, deposition temperature, etc.) employed in each study. Analogously, DB3 also shifts approximately 30 cm⁻¹ as CH₄ adsorbs to the ice surface (Figure 5(b)). In addition, we also observe an absorption band at 5384 cm⁻¹, which is a combination band of CH₄ (Grundy et al. 2002). Figure 5(a) shows the DB4 and DB5 region with increasing CH₄ uptake. Admittedly, this spectral region is not as easily interpreted as the other spectral regions, due to the appearance of multiple combination bands of CH₄ that occur near and overlap the DB bands (Calvani et al. 1992; Quirico & Schmitt 1997; Grundy et al. 2002). However, after the smallest amounts of CH₄ uptake, we can clearly resolve the shifted DB5 feature at 7185 cm⁻¹ (see arrow in Figure 5(a)). There may also be two additional CH₄ absorptions that overlap with the two DB features in the spectrum with the highest CH₄ uptake (Figure 5(a), top spectrum), yet their presence is not obvious here, likely because they are two to three times weaker than the CH₄ absorption at 7130 cm⁻¹ (Calvani et al. 1992; Grundy et al. 2002). Future studies using a gas other than CH₄ would likely better elucidate these shifts in DB4 and DB5.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.3. Estimating the Fraction of Water with a Dangling Bond

As noted previously, there have been no detections of the fundamental DB absorption bands (DB1 and DB2) in the spectra of interstellar ices. While the particular susceptibility of these features to energetic particle irradiation may be largely responsible for the absence of these features (see above), it also seems reasonable that there may be some regions in interstellar space where porous ASW exists, such as in very young molecular clouds. Dartois et al. (2013) showed spectra in the fundamental DB region (∼3700 cm⁻¹) of several younger stellar objects, finding that CO₂ absorptions and spectral noise prevented any detection of porous ASW. While the increased sensitivity predicted for JWSTs NIRSpec, compared to prior generation telescopes (Milam et al. 2016), may enable future detections of porous ASW through the identification of the fundamental DB absorptions, these features may still be obstructed by other IR-active species. Our new results, which clearly show two additional spectral regions that could be used to identify porous ASW through detection of DB absorption bands, may serve as an alternative way to identify porous ASW with remote sensing spectroscopy. The 5326 cm⁻¹ DB band (DB3) may be particularly useful given that unlike the fundamental and overtone DB bands, this combination band is well separated from CO, CO₂, and CH₄ absorption features in this spectral region (Gerakines et al. 2005; Dartois et al. 2013).

Admittedly, these near-infrared DB absorption features are ∼20 times weaker than the fundamental DB absorption features. As the column density of our samples is similar to what is typical of H₂O-ice detected in the ISM ∼1 × 10¹⁹ H₂O cm⁻² (Whittet et al. 1996; Gibb et al. 2000), we suspect that the DB3 and DB5 (Figures 2(b) and (c)) features will have optical depths on the order of ∼10⁻³. This small of an optical depth would make these features difficult if not impossible to detect with even large prior generation telescopes, where previous estimates (Sandford & Allamandola 1993) suggested that integration times on the order of hours were required to observe an absorption feature with an optical depth of ∼0.03 at a SNR of 3 (3σ confidence). However, given that the minimum detectable flux estimated for JWST's NIRSpec is about a factor of 10–100 better than any prior generation telescopes (Milam et al. 2016; Rivkin et al. 2016), it is of interest to estimate the integration time that would be required to observe these new dangling bond features with JWST. To do this, we used the JWST Exposure Time Calculator for JWST's NIRSpec operating in its fixed slit observation mode with the G235M/F170LP grating/filter pair (Pontoppidan et al. 2016), which has a resolving power of ∼1000 for wavelength coverage that includes DB3, which is the most likely of the new DB features to be well separated from other absorption features potentially present (see above). Similar time estimates are expected for the DB5 absorption band. Given a typical continuum flux of 40 mJy at 2 microns for the bright embedded source W33A (Capps et al. 1978), which is a common target for weak ice absorption features (Whittet et al. 1996; Gibb et al. 2000), and binning every four wavelength resolution elements (∼0.002 μm), we estimate that this feature should be observable at a SNR of 3 after an integration time of ∼4.5 hr. Thus, we suspect that these features should be large enough to be identified by JWST's NIRSpec in future observing campaigns.

Finally, we also speculate that much of the ASW in the ISM will have molecules trapped in its pores, which will alter the position of each DB feature, as shown here for CH₄ or for other gases previously (Rowland et al. 1991; Chaabouni et al. 2000; Palumbo & Strazzulla 2003; Loeffler et al. 2006b). This complication may, in some cases, inhibit the success of using remote sensing spectroscopy to identify porous ASW and any adsorbed species present, while in other cases, it may enhance it. In the former case, adsorbates may appear at similar peak positions as these shifted DB features, as we show for DB4 and DB5 in Figure 5(a), making unique identifications difficult. In the latter case, where adsorbates do not have a feature appearing near or overlapping with the DB features, the peak position of the shifted DB band may not only identify ASW but also possibly the adsorbed species, as the magnitude of the shift will vary depending on the adsorbed molecule. This characteristic of the DB absorption bands may be particularly valuable for cases where the adsorbates are weakly absorbing, such as with homonuclear molecules (e.g., H₂, N_2, and O₂), which are likely important components of many interstellar ice mantles (Ehrenfreund et al. 1993; Sandford & Allamandola 1993).

This research was supported by NASA Astrophysics Research and Analysis Award #80NSSC20K0359. All data shown in the paper, as well as the raw spectral data (i.e., before continuum removal or vertically shifting the spectra) can be found in Northern Arizona University's long-term repository (https://openknowledge.nau.edu/5542/).