Chinese Journal of Chemical Physics  2020, Vol. 33 Issue (1): 101-106

#### The article information

Yong Zhang

Are Fulleranes Responsible for the 21 Micron Feature?

Chinese Journal of Chemical Physics, 2020, 33(1): 101-106

http://dx.doi.org/10.1063/1674-0068/cjcp1911206

### Article history

Accepted on: December 5, 2019
Are Fulleranes Responsible for the 21 Micron Feature?
Yong Zhang
Dated: Received on November 14, 2019; Accepted on December 5, 2019
School of Physics and Astronomy, Sun Yat-sen University, Zhuhai 519082, China Laboratory for Space Research, Faculty of Science, The University of Hong Kong, Hong Kong, China
Abstract: Recent detections of C$_{60}$, C$_{70}$, and C$_{60} $$^+ in space induced extensive studies of fullerene derivatives in circumstellar environments. As the promising fullerene sources, protoplanetary nebulae (PPNe) shows a number of unidentified bands in their infrared spectra, among which a small sample exhibits an enigmatic feature at \sim 21 \mathtt{μ} m. Hydrogenation converts fullerenes into fulleranes, which breaks the symmetry of fullerene molecules and produces new infrared bands. In this work, we investigate the possibility of fulleranes (C _{60} H _m ) as the carrier of the 21 \mathtt{μ} m feature in terms of theoretical vibrational spectra of fulleranes. The evidences favoring and disfavoring the fullerane hypothesis are presented. We made an initial guess for the hydrogen coverage of C _{60} H _m that may contribute to the 21 \mathtt{μ} m feature. Key words: Fulleranes Infrared Asymptotic giant branch and post-asymptotic giant branch Circumstellar matter Stars Ⅰ. INTRODUCTION The "21 μm" feature refers to an infrared emission band peaking at 20.1 \mathtt{μ} m, which was first discovered by Kwok et al. [1] in four circumstellar envelopes of evolved stars based on a measurement of the Spitzer/IRS spectra. Sloan et al. [2] derived a longer central wavelength of (20.47 \pm 0.10) \mathtt{μ} m, . This feature is rare compared to other circumstellar dust features. Thus far, it has been detected toward only 27 evolved stars, including 18 in the Galaxy and 9 in the Large and Small Magallenic Clouds [3]. The 21 \mathtt{μ} m sources have some properties in common. Their optical-infrared spectra reveal two blackbody components, corresponding to radiation from the stellar photosphere and the dust shell. This is an evidence that stellar outflows have terminated and the envelope has been detached from the central star. Therefore, they represent a short evolutionary stage between the asymptotic giant branch (AGB) and the planetary nebula (PN), and usually are denoted as protoplanetary nebula (PPN). All the 21 \mathtt{μ} m sources are carbon rich, exhibiting absorption features from C _2 , C _3 , and CN [4] as well as the aromatic C - H bands at 3.3 and 11.3 \mathtt{μ} m [5] . Moreover, a broad emission band around 30 \mathtt{μ} m always appears along with the 21 \mathtt{μ} m feature [5] . The identification of the 30 \mathtt{μ} m feature is another unsolved problem [6] . It is worth noting that the spectra of two supernova remnants show a strong dust feature exactly peaking at 21 \mathtt{μ} m [7] . Although its peak wavelength slightly differs from the 21 \mathtt{μ} m feature, it would be interesting to investigate whether their carriers belong to a similar molecular family. Identification of the 21 \mathtt{μ} m feature is vital for understanding circumstellar chemistry and matter cycle in galaxies. Although a number of candidate materials have been proposed as its carrier [6], no consensus emerges. Some of them can be rejected. Zhang et al. [8] found that S-, Si-, and Ti-containing compounds are unlikely responsible for the 21 \mathtt{μ} m feature as a cause of low abundances of these elements. The strength of the 21 \mathtt{μ} m feature suggests that its carrier is composed of rich elements. Fe oxides can be ruled out because they would emit too broad 21 \mathtt{μ} m feature or some subfeatures that were never detected [8, 9]. A carrier candidate that has not been adequately examined by these authors is hydrogenated fullerene (fullerane). Since the discovery of C _{60} [10], fullerenes or their derivatives have been long conjectured to be ubiquitous throughout interstellar and circumstellar space. This was subsequently confirmed by the detection of C _{60} and C _{70} in PNe [11, 12], and the convincing assignment of a few diffuse interstellar bands as C _{60}$$ ^+$ [13]. C$_{60}$ can be rapidly formed in PPN stage [14]. García-Hernández et al. [12] suggested that the environments of forming fullerenes could be hydrogen-rich. Fullerenes have high proton affinities. When mixed with atomic hydrogen, C$_{60}$ can be efficiently hydrogenated into fulleranes in laboratory [15, 16]. Given the high stability of fullerene structure as well as the high abundance of cosmic carbon and hydrogen, it is attractive to investigate the vibrational spectra of fulleranes and search for their existence in circumstellar envelopes.

Based on a simple force-field model, Webster [17] found that fulleranes can radiate a broad band in the wavelength range of 19–23 $\mathtt{μ}$m, and raised the possibility of fulleranes as the carrier of the 21 $\mathtt{μ}$m feature. In laboratory environments, only specific fulleranes can be synthesized, making it hard to verify this hypothesis. Recently, more accurate calculations of vibrational spectra can be performed with the development of computational chemistry methods and computing facilities, allowing us to re-examine whether fulleranes can account for the 21 $\mathtt{μ}$m feature.

Ⅱ. COMPUTATIONS

In order to test the proposal of C$_{60}$H$_{m}$ as the carrier of the 21 $\mathtt{μ}$m feature, we calculated the vibrational spectra of selected C$_{60}$H$_{m}$ with even numbers of hydrogen atoms [18]. It is computationally impossible to derive the spectra of all C$_{60}$H$_{m}$ because of the enormous isomer numbers. In laboratory conditions, C$_{60}$ can be readily hydrogenated into C$_{60}$H$_{36}$ and then form C$_{60}$H$_{18}$ through thermal annealing [16]. With increasing hydrogen coverage, the carbon hybridization goes from sp$^2$ to sp$^3$. This reduces the stability of the carbon cage. Moreover, fulleranes are exposed to ultraviolet (UV) photons which are abundantly present in PPNe, and thus undergo dehydrogenation. As a result, C$_{60}$H$_{m}$ with large $m$ number is unlikely to exist in PPNe. A total of 55 isomers belonging to 11 kinds of C$_{60}$H$_m$ species ($m$ = 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, and 36) were selected for the computations. These isomers should represent those with lowest energy and thus be the most stable ones for a given $m$ number. For computational convenience, we did not consider fulleranes with odd number of hydrogen atoms and their ionized counterparts although they are possibly present in astronomical objects. A theoretical investigation of those species will be the subject of a follow-up study.

We performed density functional theory (DFT) calculations using the B3LYP and BH & HLYP hybrid functionals in combination with polarization consistent basis set PC1. The vibrational frequencies were obtained using a double-scaling-facter scheme [19]. This method can well reproduce the experimental spectrum of C$_{60}$ [18]. Assuming that C$_{60}$H$_{m}$ infrared spectra thermally were excited in circumstellar environments, we derived the fluxes through scaling the computed intrinsic strengths by a Boltzmann factor at $300$ K. A Drude profile with a fixed width of 0.3 $\mathtt{μ}$m was convolved to broaden the features, for the purpose of comparing the computed spectra with the observed ones. Finally, the spectrum of each C$_{60}$H$_{m}$ was obtained by co-adding its isomer spectra.

Ⅲ. RESULTS AND DISCUSSION

The theoretical spectra of C$_{60}$H$_{m}$ in the 5–30 $\mathtt{μ}$m range are presented in FIG. 1. Although the vibrational modes are spread out over a wide wavelength range, strong emission features mainly concentrate between 5–10, 13–17, and 18–23 $\mathtt{μ}$m. The 3–4 $\mathtt{μ}$m wavelength range, where the C$-$H stretching modes can produce strong features, has been extensively discussed in a previous paper [18], and thus is not included in this figure. For comparison, this figure also shows the observational spectrum of a PPN IRAS 04296+3429, which was taken from the Spitzer archive [20]. As shown in FIG. 1, a strong 21 $\mathtt{μ}$m emission band exists in the spectrum of IRAS 04296+3429, which is well confined in the wavelength range of 18–23 $\mathtt{μ}$m. It is clear that a weighted combination of different C$_{60}$H$_{m}$ spectra provides a potential to reproduce the 21 $\mathtt{μ}$m feature. The features grouping in the wavelength of 5–10 $\mathtt{μ}$m are also detected in the observational spectrum, providing a futher support for C$_{60}$H$_{m}$ as the carrier of the 21 $\mathtt{μ}$m feature. However, no strong feature in the 13–17 $\mathtt{μ}$m range is detected in the observational spectrum, which instead exhibits features in 10–13 $\mathtt{μ}$m range. The 10–13 $\mathtt{μ}$m features are not seen in the theoretical spectra of C$_{60}$H$_{m}$, and usually are attributed to silicon carbides in the literature [11].

 FIG. 1 The Spitzer/IRS spectrum of the PPN IRAS 04296+3429 (upper panel) and the theoretical spectra of C$_{60}$H$_m$ (lower panel). Note that the continuum has not been subtracted for the observational spectrum. The vertical lines mark the wavelength range of the observed 21 $\mathtt{μ}$m band

Any single C$_{60}$H$_{m}$ spectrum is unable to reproduce the observed 21 $\mathtt{μ}$m feature. Fulleranes, if being responsible for the 21 $\mathtt{μ}$m feature in PPNe, should be a mixture of various isomers with different hydrogenation degrees. It is natural to expect that their spectral pattern may vary from source to source. However, astronomical observations have shown that the 21 $\mathtt{μ}$m features have a remarkably consistent profile [21]. The same problem exists for the mixtures of polycyclic aromatic hydrocarbon molecules as the carrier of unidentified infrared emission bands, and has been extensively discussed [22]. In order to explain the observations, we conjecture that only a certain fullerane family can be formed and survive in the exclusive PPN environment. If the UV radiation is absent, hydrogen is mainly in the molecular state, and thus hydrogenation of C$_{60}$ is unlikely to occur; if it is too strong, it may substantially dehydrogenate fulleranes, or even break the carbon cage. The rigorous environments required for the existence of fulleranes are compatible with the rareness of the 21 $\mathtt{μ}$m feature in astronomical objects.

FIG. 1 indicates that C$_{60}$H$_{36}$ can be ruled out as the carrier from the 21 $\mathtt{μ}$m feature due to the lack of features lying in the 18–23 $\mathtt{μ}$m wavelength range. C$_{60}$H$_m$ with $m$ = 2–8 is unlikely responsible for the 21 $\mathtt{μ}$m feature because their strong features have peak wavelengths shorter than 20 $\mathtt{μ}$m. Moreover, C$_{60}$H$_{2}$ can produce a strong feature around 27 $\mathtt{μ}$m, which has never been detected. FIG. 2 presents the contributions from each C$_{60}$H$_m$ isomer to the total intrinsic strength of the features lying in the 18–23 $\mathtt{μ}$m range. Except C$_{60}$H$_{2}$ and C$_{60}$H$_{36}$, the strengths increase with increasing $m$ values, suggesting that moderately hydrogenated C$_{60}$ among fulleranes produces the strongest vibrational bands in the wavelength range encompassing the 21 $\mathtt{μ}$m feature.

 FIG. 2 Intrinsic strengths of the C$_{60}$H$_{m}$ features lying in the wavelength range of 18–23 $\mathtt{μ}$m versus the $m$ values. The filled circles, squares, and triangles represent the slightly ($m$ = 2–10), moderately ($m$ = 12–20), and heavily ($m$ = 36) hydrogenated C$_{60}$, respectively

In FIG. 3, we examine the intensity-weighted wavelengths ($\lambda_{21}$) of C$_{60}$H$_m$ as a function of $m$, which is defined as

 FIG. 3 The intensity-weighted wavelengths of the features lying in the wavelength range of 18–23 $\mathtt{μ}$m versus the $m$ values of C$_{60}$H$_{m}$. Symbols are the same as those in FIG. 2. The horizontal line denotes the wavelength position of the 21 $\mathtt{μ}$m feature
 $\begin{eqnarray} \lambda_{21} = \frac{ \sum\limits_i \lambda_iF_i}{ \sum\limits_i F_i} \end{eqnarray}$ (1)

for 18 $\mathtt{μ}$m$\le $$\lambda_i$$ \le$23 $\mathtt{μ}$m, where $F_i$ is the intrinsic strength of the $i$th mode at the wavelength of $\lambda_i$. An inspection of FIG. 3 reveals that there is an approximately linear trend of longer intensity-weighted wavelengths with increasing hydrogenation degrees. This is in agreement with the predictions by the force-field model [17]. Comparing with the observed peak wavelength (20.1 $\mathtt{μ}$m), we infer that moderately hydrogenated fulleranes with 10$< $$m$$ <$20 are the promising carriers for the 21 $\mathtt{μ}$m feature.

While the existence of fulleranes in astronomical environments seems plausible, there is as yet no unambiguous detection. Díaz-Luis et al. [23] failed to detect the C$-$H stretching bands at 3.4–3.6 $\mathtt{μ}$m in two PNe exhibiting strong C$_{60}$ emissions, suggesting that fulleranes might have been destroyed by strong UV radiation or mostly ionized. Based on a comparison between the laboratory spectrum of gaseous C$_{60}$H$^+$ and the observations, Palotás et al. [24] speculated that C$_{60}$H$^+$ might contribute to the spectra of two C$_{60}$-containing PNe. Presumably, exposed to very strong UV radiation, moderately and heavily hydrogenated fulleranes cannot survive in PNe. This is compatible with the non-detection of the 21 $\mathtt{μ}$m feature in PNe. Zhang and Kwok [25] reported a tentative detection of fulleranes in the C$_{60}$ source IRAS 01005+7910, which is a PPN about to enter the PN stage. IRAS 01005+7910 is not assigned as a 21 $\mathtt{μ}$m source in the previous literature. However, a closer view of its infrared spectrum clearly reveals a faint feature exactly peaking at 20.1 $\mathtt{μ}$m (FIG. 4). This feature is much weaker than the C$_{60}$ bands, and has not been noted previously. If it is attributed to fulleranes, we can hypothesize that fulleranes in this PPN are undergoing dehydrogenation, resulting in a transition from moderately hydrogenated C$_{60}$ to slightly and none hydrogenated C$_{60}$. The enrichment of slightly hydrogenated C$_{60}$ may partly explain the 15–20 $\mathtt{μ}$m plateau emission in the spectrum of IRAS 01005+7910.

 FIG. 4 The Spitzer/IRS spectrum of the PPN IRAS 01005+7910. The vertical solid and dashed lines mark the positions of the 21 $\mathtt{μ}$m feature and the four C$_{60}$ bands, respectively. The inset panel shows the continuum-subtracted spectrum

A criticism of the fullerane hypothesis was advanced by Posch et al. [26], who pointed out that the actual wavelength of the C$_{60}$ feature is not coincident with that of the emission band in the meteoritic nanodiamonds shown by Hill et al. [27]. The argument stems from the fact that nanodiamonds and fulleranes have a similar hybridization structure. However, it is not always appropriate to expect that nanodiamonds and fulleranes emit bands at the same wavelength positions. The fullerane hypothesis refers to a combination of numerous C$_{60}$H$_m$ isomers with different $m$ values, making it sufficiently flexible to match the observational spectrum.

A significant criterium for band identification is to examine whether associated subfeatures from the carrier candidate are visible in observed spectra. The C$-$H stretching mode around 3.4 $\mathtt{μ}$m has been detected in almost all the 21 $\mathtt{μ}$m sources. However, the theoretical spectra of fulleranes imply that there is no correlation between the intensities of the 3.4 $\mathtt{μ}$m and the 21 $\mathtt{μ}$m features, as shown in FIG. 5. The 3.4 $\mathtt{μ}$m/21 $\mathtt{μ}$m intensity ratios largely vary among different isomers, and strong 21 $\mathtt{μ}$m sources might exhibit weak features at 3–4 $\mathtt{μ}$m. This is conceivable since the 21 $\mathtt{μ}$m feature dominantly arises from C$-$C deformation vibrations. Therefore, the feature around 3.4 $\mathtt{μ}$m may not serve as a good tracer to examine the fullerene hypothesis.

 FIG. 5 Intrinsic strengths of the C$_{60}$H$_{m}$ features lying in the wavelength range of 3–4 $\mathtt{μ}$m versus those of 18–23 $\mathtt{μ}$m. Symbols are the same as those in FIG. 2

The theoretical spectra of fulleranes reveal that the 5–10 $\mathtt{μ}$m feature has an intensity positively correlating with that of the 18–23 $\mathtt{μ}$m feature (FIG. 6). This is in good agreement with the observational results that the spectra of 21 $\mathtt{μ}$m sources usually show a broad peak at 7–9 $\mathtt{μ}$m as well [2]. This makes the fullerane hypothesis quit plausible. However, the 5–10 $\mathtt{μ}$m feature has been commonly detected in various circumstellar envelopes, and cannot been entirely attributed to fulleranes.

 FIG. 6 Intrinsic strengths of the C$_{60}$H$_{m}$ features lying in the wavelength range of 5–10 $\mathtt{μ}$m versus those of 18–23 $\mathtt{μ}$m. Symbols are the same as those in FIG. 2

According to our calculations, an issue of the fullerane hypothesis is that the 13–17 $\mathtt{μ}$m feature arising from fulleranes is hardly visible in the observational spectra. As shown in FIG. 7, the 13–17 $\mathtt{μ}$m feature should have a comparable intensity with the 21 $\mathtt{μ}$m one, which is in contrast to the observations. FIG. 7 shows that the 13–17 $\mathtt{μ}$m feature is likely to peak at 15 $\mathtt{μ}$m. Zhang et al. [18] examined the spectra of all C$_{60}$-containing objects, but did not make any unambiguous detection of the 15 $\mathtt{μ}$m feature (this feature was marginally seen in only two C$_{60}$-containing objects, which cannot be considered as a solid detection). It doesn't seem to support the expect that fulleranes coexist with fullerenes. Nevertheless, it is notable that DFT calculations reproduce the spectra of free-flying molecules, which can interpret the band wavelengths of experimental spectra much better than the band strengths. If the material is in condensed phase, some prominent bands in the theoretical spectra may be strongly suppressed. From a close inspection of the laboratory spectrum of C$_{60}$H$_{18}$ (see FIG. 3 in Ref.[16]), we can see some bands lying in the 13–17 $\mathtt{μ}$m range with wavelengths coincident with the theoretical calculations. But these bands appear much weaker than those lying in the 5–10 $\mathtt{μ}$m and 18–13 $\mathtt{μ}$m ranges, contrary to the theoretical results. Furthermore, an appropriate excitation model needs to be employed to accurately predict the observed intensities. As a result, the non-detection of the 13–17 $\mathtt{μ}$m feature is insufficient to invalidate the fullerane hypothesis.

 FIG. 7 Intrinsic strengths of the C$_{60}$H$_{m}$ features lying in the wavelength range of 13–17 $\mathtt{μ}$m versus those of 18–23 $\mathtt{μ}$m. Symbols are the same as those in FIG. 2. The insert shows the correlation between the strengths (in arbitrary units) of the 15.8 $\mathtt{μ}$m and the 21 $\mathtt{μ}$m features, as found by Zhang et al. [20]. Dashed lines represent the linear fittings

Although no strong band has been detected in the 13–17 $\mathtt{μ}$m range in astronomical objects, a weak feature at 15.8 $\mathtt{μ}$m appears in the spectra of all the 21 $\mathtt{μ}$m sources [5] . Zhang and Kwok [14] found that there is a loose correlation between the intensity of the 15.8 $\mathtt{μ}$m and the 21 $\mathtt{μ}$m features (FIG. 7), suggesting that the two features may arise from associated materials. Sloan et al. [2] presented that the associated 15.8 $\mathtt{μ}$m feature may arise from aliphatic chains with alkyne bonds, and thus the carrier of the 21 $\mathtt{μ}$m feature should be related to aliphatic hydrocarbons. This favors the fullerane hypothesis.

The theoretical spectra of fulleranes do not exhibit strong bands around 30 $\mathtt{μ}$m. Therefore, the 30 $\mathtt{μ}$m feature detected in AGB stars, PPNe, and PNe is unlikely due to fulleranes. While the 21 $\mathtt{μ}$m feature, without exception, is accompanied by the 30 $\mathtt{μ}$m feature, the number of the 30 $\mathtt{μ}$m sources is much larger, and the strengths of the two features are not correlated with each other. A common property of the two features is that both are detected in C-rich environments. Strikingly, the 30 $\mathtt{μ}$m feature was also viewed in the spectra of the PPNe and PNe detected in C$_{60}$ [25]. It is tempting to conjecture that the carrier of the 30 $\mathtt{μ}$m feature can be synthesized during the AGB stage, and then when exposed to UV radiation at PN and PPN stages it is fragmented and partly converted into fullerenes and fulleranes through a top-down scenario.

Ⅳ. CONCLUSION

We investigated the theoretical spectra of C$_{60}$H$_{m}$ with the aim of examining whether fulleranes are responsible for the 21 $\mathtt{μ}$m feature in PPNe. Based on a comparison of wavelengths and intensities, we infer that moderately hydrogenated C$_{60}$ is a promising carrier material producing this feature. A mixture of specific C$_{60}$H$_{m}$ isomers is required to fit the observed profile of the 21 $\mathtt{μ}$m feature. The survival of fullerane is very sensitive to the UV radiative field, and thus can give a natural explanation for the emergence of the 21 $\mathtt{μ}$m feature on very short timescales. The main issue is that the theoretical spectra of fullerane pose strong features in the 13–17 $\mathtt{μ}$m range which are absent in 21 $\mathtt{μ}$m sources. However, given the limits of computation and model complexity, this is not complete enough to rule out the fullerane hypothesis. The analysis and results proposed here can provide a guideline for future experimental efforts on the identification of the 21 $\mathtt{μ}$m feature.

Ⅴ. ACKNOWLEDGMENTS

I am grateful to Dr. SeyedAbdolreza Sadjadi for his help on the computations of C$_{60}$H$_{m}$ spectra. This work was supported by the National Natural Science Foundation of China (No.11973099). I also acknowledge the Science and Technology Development Fund of Macau Special Adminastrative Region for support through grant 0007/2019/A.

Reference
 [1] S. Kwok, K. M. Volk, and B. J. Hrivnak, Astrophys. J. Lett. 345, L51 (1989). [2] G. C. Sloan, E. Lagadec, A. A. Zijlstra, K. E. Kraemer, A. P. Weis, M. Matsuura, K. Volk, E. Peeters, W. W. Duley, and J. Cami, Astrophys. J. 791, 28(2014). DOI:10.1088/0004-637X/791/1/28 [3] A. Mishra, A. G. Li, and B. W. Jiang, Astrophys. J. 825, 68(2016). DOI:10.3847/0004-637X/825/1/68 [4] E. J. Bakker, H. J. G. L. M. Lamers, L. B. F. M. Waters, and T. Schoenmaker, Astrophys. Space Sci. 224, 335(1995). DOI:10.1007/BF00667868 [5] B. J. Hrivnak, K. Volk, T. R. Geballe, and S. Kwok, Proc. Int. Astronom. Union 4, 213(2008). DOI:10.1017/S1743921308021583 [6] B. W. Jiang, K. Zhang, and G. A. Li, Earth, Planets Space 62, 105(2010). DOI:10.5047/eps.2008.12.004 [7] J. Rho, H. L. Gomez, A. Boogert, M. W. L. Smith, P. O. Lagage, D. Dowell, C. J. R. Clark, E. Peeters, and J. Cami, Mon. Not. Roy. Astronom. Soc. 479, 5101(2018). DOI:10.1093/mnras/sty1713 [8] K. Zhang, B. W. Jiang, and A. G. Li, Mon. Not. Roy. Astronom. Soc. 396, 1247(2009). DOI:10.1111/j.1365-2966.2009.14808.x [9] A. G. Li, J. M. Liu, and B. W. Jiang, Astrophys. J. 777, 111(2013). DOI:10.1088/0004-637X/777/2/111 [10] H. W. Kroto, J. R. Heath, S. C. O'Brien, R. F. Curl, and R. E. Smalley, Nature 318, 162(1985). DOI:10.1038/318162a0 [11] J. Cami, J. Bernard-Salas, E. Peeters, and S. E. Malek, Science 329, 1180(2010). DOI:10.1126/science.1192035 [12] D. A. García-Hernández, A. Manchado, P. GarcíaLario, L. Stanghellini, E. Villaver, R. A. Shaw, R. Szczerba, and J. V. Perea-Calderón, Astrophys. J. 724, L39 (2010). [13] E. K. Campbell, M. Holz, D. Gerlich, and J. P. Maier, Nature 523, 322(2015). DOI:10.1038/nature14566 [14] Y. Zhang, and S. Kwok, Astrophys. J. 730, 126(2011). DOI:10.1088/0004-637X/730/2/126 [15] F. Cataldo and S. Iglesias-Groth, Mon. Not. Roy. Astronom. Soc. 400, 291 (2009). [16] S. Iglesias-Groth, D. A. García-Hernández, F. Cataldo, and A. Manchado, Mon. Not. Roy. Astronom. Soc. 423, 2868(2012). DOI:10.1111/j.1365-2966.2012.21097.x [17] A. Webster, Mon. Not. Roy. Astronom. Soc. 227, 1555 (1995). [18] Y. Zhang, S. A. Sadjadi, C. H. Hsia, and S. Kwok, Astrophys. J. 845, 76(2017). DOI:10.3847/1538-4357/aa71ac [19] M. L. Laury, M. J. Carlson, and A. K. Wilson, J. Comput. Chem. 33, 2380(2012). DOI:10.1002/jcc.23073 [20] Y. Zhang, S. Kwok, and B. J. Hrivnak, Astrophys. J. 725, 990(2010). DOI:10.1088/0004-637X/725/1/990 [21] K. Volk, S. Kwok, and B. J. Hrivnak, Astrophys. J. 516, L99 (1999). [22] S. Kwok, and Y. Zhang, Astrophys. J. 771, 5(2013). DOI:10.1088/0004-637X/771/1/5 [23] J. J. Díaz-Luis, D. A. García-Hernández, A. Manchado, and F. Cataldo, Astron. Astrophys. 589, A5 (2016). [24] J. Palotás, J. Martens, G. Berden, and J. Oomens, Nat. Astron. 494 (2019). [25] Y. Zhang, and S. Kwok, Earth, Planets Space 65, 1069(2013). DOI:10.5047/eps.2013.06.003 [26] T. Posch, H. Mutschke, and A. Andersen, Astrophys. J. 616, 1167(2004). DOI:10.1086/425113 [27] G. H. M. Hill, A. P. Jones, and B. L. D'Hendecourt, Astron. Astrophys. 336, L41 (1998).