The Astrochemistry Low-energy Electron Cross-Section (ALeCS) database I. Semi-empirical electron-impact ionization cross-section calculations and ionization rates

Context. Electron-molecule interaction is a fundamental process in radiation-driven chemistry in space, from the interstellar medium to comets. Therefore, knowledge of interaction cross-sections is key. There have been a plethora of both theoretical and experimental studies of total ionization cross-sections spanning from diatomics to complex organics. However, data is often spread over many sources, or not public or readily available. Aims. We introduce the Astrochemistry Low-energy Electron Cross-Section (ALeCS) database, a public database for electron interaction cross-sections and ionization rates for molecules of astrochemical interest. In particular, in this work, we present the first data release comprising total ionization cross-sections and ionization rates for over 200 neutral molecules. Methods. We include optimized geometries and molecular orbital energies at various levels of quantum chemistry theory. Further, for a subset of the molecules we have calculated ionization potentials. We compute total ionization cross-sections using the binary-encounter Bethe model and screening-corrected additivity rule, and ionization rates and reaction network coe ffi cients for molecular cloud environments. Results. We present the cross-sections and reaction rates for > 200 neutral molecules ranging from diatomics to complex organics, with the largest being C 14 H 10 . We find that the screening-corrected additivity rule cross-sections generally significantly overestimate experimental total ionization cross-sections. We demonstrate that our binary-encounter Bethe cross-sections agree well with experimental data. We show that the ionization rates scale roughly linearly with the number of constituent atoms in the molecule. Conclusions. We introduce and describe the public ALeCS database. For the initial release, we include total ionization cross-sections for > 200 neutral molecules and several cations and anions calculated with di ff erent levels of quantum chemistry theory, the chemical reaction rates for the ionization, and network files in the formats of the two most popular astrochemical networks, the Kinetic Database for Astrochemistry and UMIST. The database will be continuously updated for more molecules and interactions.


Introduction 1
Observational studies of molecular gas within the Milky Way 2 have revealed a diverse zoo of about 300 molecules, from simple 3 diatomics such as H 2 and CO to ever more complex molecules 4 such as NC 4 NH + (Agúndez et al. 2023), NH 2 C(O)CH 2 OH (syn-5 glycolamide, Rivilla et al. 2023), 2-C 9 H 7 CN (2-Cyanoindene, 6 Sita et al. 2022), and H 2 CCCHC 3 N (cyanoacetyleneallene, Shin-datasets and databases represent a significant advancement in the availability of cross-section data for astrochemical use.However, they generally do not include molecules containing heavier atoms, and do not report on computed and recommended reaction rate coefficients either.Furthermore, databases comprising a large number of ionization cross sections often either do not use a standard evaluation process or do not provide the data in a readily accessible way. A number of different electron-impact total ionization cross sections have been introduced in the literature (see e.g.Kim et al. 1997;Deutsch et al. 2000;Blanco et al. 2010).For large molecules, in which the computing electronic structure can become untenable, various different additivity rules have been introduced, which introduce a sort of algebra for adding atomic ionization cross sections and atomic or molecular orbits.One of the most important models is the binary-encounter dipole and its simplification, the binary-encounter Bethe (BEB) cross section (Kim & Rudd 1994;Kim et al. 1997), which we detail below in Section 2.2.3.It has performed well compared to experimental data (Kim et al. 1997;Zhou et al. 2019;Zhong et al. 2021, e.g.) and has thus become a standard.All of the above databases also present the BEB cross sections as a major result.
In this paper, we present a new public database containing electron ionization cross sections and cosmic-ray ionization rate coefficients for over 200 molecules of astrochemical interest.We report the cross sections for low-energy electrons (10 eV ≤ E e ≤ 5000 eV) for each of these molecules using both a screeningcorrected additivity rule (SCAR) and the BEB model cross section, the molecular orbital binding and kinetic energies we used, and the recommended total reaction rate coefficients for molecular cloud environments.Crucially, the cross sections and coefficients are evaluated for nearly all molecules using the same procedure, providing homogeneous datasets to enable better comparisons and consistency.

Methods
We present here the methods for the different approaches of calculating the total electron-impact ionization cross sections.Then, quantum chemistry computations are performed, and the resulting ionization rates are calculated.In brief, we calculated three different models of the electron-impact total ionization rate: the SCAR, the BEB, and the damped-BEB (dBEB) model.In particular, we explored the accuracy of the SCAR method because it may become more applicable for larger molecules due to the computational expense required for accurately computing BEB-related cross sections.The latter models require knowledge of the electronic structure, while the former only requires knowledge of the geometry and a basis set of electron-atom ionization cross sections.For our molecule selection, we chose the primary molecules used in astrochemical networks with reaction rate data on the kinetics database KIDA.
We emphasize that the total ionization cross sections below are for single-ionization events.At high energies, multiple ionization events can occur, in particular, through Auger ionization.Nishimura et al. (1999) estimated the possible impact of multiple ionizations by doubling the contribution of the cross section from inner shell electrons and found that their inclusion shifted the peak by 5% and toward higher impact energies.The contributions of multiple ionization events are beyond the scope of this initial release, and they will be considered for future releases.Furthermore, molecules are assumed to be ionized from their ground state.
Article number, page 2 of ?? The semi-empirical method we use requires a basis of electron-135 atom ionization cross sections.We obtain these cross sections 136 by fitting a polynomial to experimental data from the NIFS 137 AMIDIS-ION 5 database, where the electron-atom ionization 138 cross section takes the form where a 0 is the Bohr radius, E e is the electron energy, IP i is the where the summation is over the different atoms, i.The cross sections are computed using the SCAR following 155 Blanco & García (2003); Blanco et al. (2010).This method en- The cross section for a specific molecule, m, is given by where the sum is carried out over all constituent atoms, and σ i (E) is the electron-atom ionization cross section for an atom i.

163
Under the simplest additivity rules, as shown above, s i = 1.For 164 the SCAR method, the additivity coefficients, s i , are where N k is the number of perturbation terms included.The coefficients, ε (k) i , follow the recursive relation where α ji = max(4πr 2 i j , σ i , σ j ), and r i j is the distance between 168 atoms i and j.The terms in Equation 5 amount to higher-order 169 correction factors accounting for the overlap of the cross sections 170 of all the individual items.Due to the recursive nature of the coefficients (Eq.6), including higher-order correction terms leads to an exponential increase in computational cost.However, this is alleviated by building a dictionary cache during the recursion to avoid recomputing the same screening terms.This leads to a sublinear increase in computing time against maximum k-atom screening correction included.

BEB cross sections
The semi-empirical BEB cross section was developed as a simplification of the binary-encounter dipole cross section (Kim & Rudd 1994;Hwang et al. 1996).This cross section has been found to provide a reasonable match with experimental data, and it only requires knowledge of the molecular orbital energies.
The BEB cross sections are the base of numerous cross section databases, in particular, the NIST database.The BEB cross section is defined as where the sum is over orbitals, indexed , Here, B and U are the orbital binding and orbital kinetic energies of the ejected electron, respectively, n is the orbital occupation number, and R ∞ is the Rydberg constant.The parameters B and U are generally computed using electronic structure methods.
We also propose a modified version of the BEB cross section (dBEB) to dampen the impact of orbitals with binding energies greater than the ionization potential.In dBEB cross section formulation, the B is scaled by an exponential function such that the modified binding energy of orbital is given by (10) As we demonstrate below, this prevents the BEB cross section from overestimating with respect to experimental values, which are generally upper limits.In general, BEB cross sections underpredict experimental results at higher energies since they only account for single ionizations, while experimental data, in which the measured signal is generally the ion current, include contributions from multiple ionization events that contribute more to the measured signal per event than single ionizations.

Quantum chemistry computations
We selected a set of 156 neutral species including atoms and molecules to calculate the BEB and dBEB cross sections.The initial structures were taken from NIST Computational Chemistry Comparison and Benchmark DataBase7 (CCCBDB) database (Johnson 2022) or ChemSpider8 .The equilibrium geometries were obtained at MP2/aug-cc-pVTZ method and basis set.To validate the accuracy of this level of theory, we also optimized a group of 50 structures using the highly accurate DF-CCSD(T)-F12/cc-pVDZ-F12 (U-CCSD(T)-F12/cc-pVDZ-F12 for radical species).The set of electron binding energy (eBE), B , taken as the negative of the energy of orbital , was computed by means of electron propagator theory (EPT) in The depth-dependent electron ionization rate of a molecule, 265 m, is given by adding the contributions of primary protons, 266 ζ p,m (N H 2 ), using the approximation primary electrons, and secondary electrons from both primary protons and elec-269 trons, where se represents the secondary electrons.The factors of 2π 271 and 4π account for fluxes, which are assumed to be plane par-272 allel and isotropic, respectively.The total ionization rate for a 273 molecule is then The factor of 4π 274 comes from treating the secondary electrons as an isotropic local 275 source.

276
We used the "High" primary proton and proton-induced sec-277 ondary electron spectra as a function of column density from 278 Padovani et al. (2022), who computed the electron spectrum 279 down to 1 eV following the recent more rigorous theory of 280 Ivlev et al. (2021).The High proton spectrum was calibrated to 281 match diffuse gas observations of the H 2 ionization rate, which 282 are not reproduced with a Voyager-like spectrum (Ivlev et al. 283 2015;Padovani et al. 2018).We also used an interstellar pri-284 mary electron spectrum from Padovani et al. (2018) and their 285 induced secondary electrons.The secondary electron computa-286 tion assumes that the gas is fully molecular and includes en-287 ergy losses from Coloumb interactions, as well as for H 2 ion-288 izations and electronic and rovibrational excitations.We defined 289 our column-dependent total ionization rate coefficient for species 290 m, c m,T (N H 2 ) from where ζ H 2 ,T (N H 2 ) is the total H 2 ionization rate including all pri-292 mary, secondary, and tertiary processes.We report the column-293 density average coefficient, cm,T , for the cloud column density 294 Article number, page 4 of ?? but also suggest that the error related to the electronic structure is very small.However, for the deep-lying orbitals, no reliable EPT-eBEs can be computed.We therefore chose to use the HF values for these orbitals.
Our HF-BE data and the NIST electron-impact cross-section database agree well.The database uses data from Hwang et al. (1996) computed at the HF/6-311-G level for the eBE.The agreement is expected because the only difference between the NIST and our HF orbital binding energies is a larger atomic orbital basis set in the latter case.However, the incorporation of EPT-BE into our database enhances the overall data quality compared to the existing NIST data.We thus conclude that our calculations of orbital binding energies are derived from a robust theoretical framework, offering the required level of accuracy given the underlying assumptions in our cross-section calculations.
Finally, we also included the optimized geometries at the MP2/aug-cc-pVTZ and DF-CCSD(T)-F12/cc-pVDZ-F12 levels presented here and the sample from Heathcote & Vallance (2018) in the database.The geometries are not appreciably different from each other.All geometries that represent minimum energy structures were verified by ensuring that no imaginary frequencies in the diagonalized Hessian matrix are present.

Low-energy electron cross sections
As described earlier, we have computed the cross sections using three different approaches: the SCAR, the BEB model, and the new dBEB.
The SCAR method is inherently recursive, so that care must be taken to avoid exponential increases in computational time with the number of atoms.We used recursive caching to speed the calculations up, and we only computed to a maximum of the 10-atom screening correction.Figure 4 shows the individual screening corrections, the magnitude of the correction, and the total time as a function of the maximum screening terms kept.The screening terms past the 8-atom screening correction are negligible contributions to the total cross section, which is dominated by the first few correction factors.
We show results for a subset of the molecules, ranging from simple to complex, to compare the difference between the SCAR and BEB cross sections.Figure 5 shows them for a subsample of ten molecules, ranging in size from CO to c-C 6 H 6 .We find that the SCAR approach generally overestimates the cross section, although the impact is most pronounced in c-C 6 H 6 because the molecular orbitals overlap.However, given the speed with which these cross sections were computed, they may be useful for first investigations when the optimized molecule geometry is known a priori.While we encourage the use of BEB cross sections, we include a Python script in the database to compute the SCAR cross section from a provided optimized geometry.
We investigated the model predictions for a subset of these molecules in more detail for which experimental data are available.Figure 7 shows a comparison with experimental data for a subset of molecules, including the simple molecule CO 2 , symmetric ring c-C 6 H 6 , prebiotic species of interest NH 2 CHO, and carbon chains.Experimental and theoretical cross section data are known to deviate quite significantly (see the discussion in Zhou et al. 2019).In particular, experimental cross sections include double ionizations and Auger ionizations, and they therefore overestimate the single-ionization cross section.The cross sections we present here are for single-ionization events.Multiple ionizations are left for future work.With the data we computed, the BEB slightly overesitmates the cross section in lowenergy regions and underestimates it in higher-energy regions.
Nonetheless, the values from the BEB calculation agree far better with experimental results than the SCAR or the AR model.
Because the experimental curve should serve as an upper limit for the theoretical BEB model, and noting that even with highly accurate eBEs it still overestimates the cross section, we suggest a modified BEB model in which the the contribution from deeper-lying electron ionizations are scaled, to better balance its weight in the total cross section.As expected, we find that the dBEB corss-section underpredicts the experimental value.Future calculations using the more complicated binary-encounter dipole model (Kim & Rudd 1994) will be investigated for future releases.In our further analysis, we use the BEB cross sections to calculate the reaction rates because toward higher energies, they tend to agree better with the data.

Molecular cloud rate coefficients
We discuss here the molecular cloud reaction rate coefficients, scaled to the total ionization rate (see Section 2.4).Figures B.9 -B.12 show the coefficients c m,T for our database sample.There is a general trend of an increase in c m,T with the size of the molecule.However, for a given number of atoms, there is still substantial scatter due to the geometry and composition.Our reaction rate is marginally lower than that reported by UMIST for CO, originally from Black (1975), who reported c CO,T = 3, whereas we find c CO,T = 1.943.We report the mean coefficients in Table A.7, where the mean is taken of the coefficients for total hydrogen nucleus column densities ranging from 10 20 -10 23 cm −2 .We note that these reaction rates are tailored for molecular cloud-like environments, such as those whose prescribed external cosmic-ray spectrum matches the "High" model from Padovani et al. (2022) (see also Ivlev et al. 2015) and whose total hydrogen nucleus column densities lie between 10 20 ≤ N H ≤ There is a substantial difference in the reaction rates found with the SCAR and BEB methods.While the additivity rules account for the geometric overlap of the atomic cross sections, it does not account for the structure of the molecule orbitals nor for differences in eBEs due to molecular bonds.Therefore, we 444 do not recommend the use of additivity rules to compute the 445 cross sections for use in astrochemical modeling, unless it is not 446 practical to compute molecular orbitals and use the BEB cross 447 section.

448
Figure 8 shows the mean c m,T coefficient as a function of the 449 number of atoms.Similar to Figure 6, we find a general increase 450 in the reaction rate with the number of constituent atoms.The 451 figure also shows a best-fit relation, where N atom is the number of constituent atoms.The relation 453 reproduces the general trend for N atom ≤ 20.We also fit the co-454 efficients as a function of the number of valence electrons, which 455 produces a slightly better fit, where N v,elec is the number of valence electrons in the molecule.457 This fit was only constrained for molecules with 9 ≤ N v,elec ≤ 458 70.In both cases, there is significant scatter of approximately 459 a factor of 2 around the fit trends, so that caution should be 460 used when these are used for molecules that are not listed in 461 the database.where the first row gives the number of electron energy bins.The 484 later rows give the data in (two) three columns.For BEB cross 485 sections, the columns provide electron impact energy (eV), BEB 486 cross section in units of σ 2 0 and, if available, the damped-BEB 487 cross section in units of σ 2 0 .For SCAR cross sections, the three 488 columns provide energy (eV) and SCAR cross section in units 489 of σ 2 0 .

490
The orbital information is stored in two directories contain-491 ing the orbitals in the NIST format, NIST_orbitals/ and in-492 cluding all orbitals, full_orbitals/.where the B column gives the eBE (eV), U gives the average electron kinetic energy (eV), N gives the orbital occupation number and Q is a scaling factor to include higher-order ionization effects.We also show an example of the full orbital file, co.forbWe include the computed ionization potentials in the ips/ directory for the compiled NIST and computed CAM-B3LPY and CCSD(T) molecules.These are two-column files with the molecule and ionization potential in eV.We also include two network files containing all the new molecular and atomic ionization rate coefficients in the networks/ directory.These network files are in the KIDA and UMIST formats, alecs.kida.inand alecs.umist.d,respectively.We recommend users to only include ionization rates for molecules in which there are associated recombination rates for the ion.In our rate files, we assume that all ionizations occur in the manner AB + e -− −− → AB + + e -+ e -, and we emphasize that potential users should check their chemical networks for other branches such as AB + e -− −− → A + B + + e -+ e -.We leave these differences for future database releases.
Finally, in the geoms/ folder, we include all molecule geometries computed at the HF, MP2, and CCSD(T) levels.The MP2 and CCSD(T) geometries are formatted as Pdb files and the HF geometries as Xyz.

Astrochemical modeling
We include the new calculations in a model that uses the KIDA reactions framework, providing a zero-dimensional model as in Sect.3.1 of Wakelam et al. (2015).It is important to note that in our context, this test has not been designed to quantify the impact of the new rates in an astrophysical environment, but to show that the new rate equations are compatible with previously established formats.Therefore, the cross sections may play a more important role in ice chemistry, disks, planetary atmospheres, exomoons, or cometary environments, for instance.However, this analysis is beyond the aims of the present paper, and it will be discussed in a forthcoming work.
We modeled a gas with total hydrogen density n H,tot = 2 × 10 4 cm −3 , gas temperature T = 10 K, initial conditions as in Tab. 8, and no dust (e.g., see Hincelin et al. 2011;Loison et al. 2014).We computed the ionization rate ζ(N H 2 ) including protons, primary and secondary electrons, and the secondary from primary electrons.The ionization rate is a function of the column 563 density N (see Eq. 14), and we assumed that the visual extinction 564 (necessary for the photochemistry) is A V = 1.0638 × 10 −21 N, 565 where N is in units of cm −2 .To model the time-dependent evo-566 lution of the chemical abundances for 10 Myr, we made use 567 of kmarx (Grassi in prep., commit 58f6ac9), a Python-based 568 database that allowed us to solve the chemical ordinary dif-569 ferential equations with a standard BDF solver14 (Hindmarsh 570 1983).We include the HTML output produced by the code in 571 the database as a zip file in the chem_models/ folder, where 572 the reactions present in this work are listed under the class 573 CRReactionAdv.

574
When using our chemical network, we replaced the ioniza-575 tion reactions from KIDA where possible (i.e., H, He, N, O, H 2 , 576 and CO) and added the missing ionization rates that affect the 577 molecules listed in Tab.8.We did not add the species that cre-578 ate sinks or sources in the chemical network, that is, species that 579 only appear in the products or in the reactants.We also note that 580 for the sake of comparison, the cosmic-ray reaction rates from 581 KIDA were scaled so that the H 2 ionization matches the ion-582 ization rate of molecular hydrogen in our database.The aim of 583 this scaling is to avoid discrepancies determined by different as-584 sumptions in the cosmic-ray spectra employed.585 Fig. 13 shows the results obtained with the KIDA database 586 reaction rates (solid lines) and calculated with the rates present 587 in this work (dashed lines).For the sake of clarity, we only plot 588 the species that show a difference larger than one-tenth of an or-589 der of magnitude and reach at least n(t) = 10 −8 n H,tot during their 590 evolution.Only a few species present a negligible discrepancy 591 between the two databases.The extent of these variations is very 592 small because the critical cosmic-ray-driven reactions are very 593 similar in ALeCS and in KIDA, but the results might be less in-594 terchangeable in different environments.Moreover, we note that 595 we assumed that the constituents of the spectra match, that is, 596 that both include protons, electrons, and secondary processes, 597 hence the scaling mentioned above.

599
We presented the initial data release for the Astrochemistry Low-600 energy electron cross-section (ALeCS) database.In this release, 601 we include the total ionization cross sections and ionization rate 602 coefficients for over 200 neutral molecules of astrochemical in-603 terest calculated using three different semi-empirical methods: 604 the SCAR (Blanco & García 2003;Blanco et al. 2010), the BEB 605 model (Kim & Rudd 1994;Hwang et al. 1996), and a new dBEB 606 presented here.The last model dampens orbitals deep within the 607 potential well and was demonstrated to help prevent the BEB 608 model from overestimating the ionization cross section when 609 compared with experimental results.We also presented the ion-610 ization rate coefficients and molecular ionization rates scaled to 611 a reference total H 2 ionization rate.

612
The database is fully public and will include the ionization 613 data described above, along with the molecule orbitals and op-614 timized geometries.In this current release, we only include the 615 semi-empirical ionization cross sections and reference experi-616 mental data.Future releases will include more sophisticated ion-617 ization cross-section calculations along with excitation and mo-618 mentum transfer.Finally, the database will include open-source 619 software tools necessary to couple these processes to astrochem-620 ical codes.

Energy (eV)
Ar + e Ar + Fig. 1: Electron ionization cross sections.The solid lines give the fits from Equation 2, and the black points show the data from the NIFS database.
Article number, page 9 of ??        Fig. 13: Comparison of the chemical evolution using KIDA reaction rates (solid lines) and our rates (dashed lines).We report the species that present a difference of at least half an order of magnitude and reach at least n(t) = 10 −8 n H,tot .The ionization rate of H 2 is 1.64×10 −16 s −1 , and the rate for model "High" and with the assumptions described in the text corresponds to A V =10.6 and N=10 22 cm −2 .
Article number, page 14 of ??          Article number, page 27 of ??

140
ionization potential of the atomic species i, x = E e /IP i , and c k 141 are fitting coefficients.Figure1shows the experimental data and 142 the fit cross sections. 598

Fig. 2 :
Fig. 2: Distribution of the number of atoms in each molecule for this data release.

Fig. 3 :
Fig. 3: Comparison of the ionization potentials presented here with the NIST database.Left: NIST database vs. CCSD(T) and CAM-B3LYP DFT calculations (solid and empty squares, respectively).Right: CCSD(T) vs. CAM-B3LYP DFT ionization potentials.The dashed line represents the one-to-one ratio.

Fig. 4 :Fig. 5 :Fig. 6 :Fig. 7 :Fig. 8 :
Fig.4: Benchmark of the number of k-atom screening corrections included using C 5 H 10 .Left: k-atom screening corrections as a function of electron energy for C 5 H 10 .Right: Time to compute σ SCAR (E e ) for a given maximum k-atom screening correction (black) and the magnitude of the maximum correction (blue) as a function of the maximum k.
Table1shows the results of the fits of Equa- 146The simplest semi-empirical cross section is the simple additiv-147 ity rule (AR).Here, the total molecule ionization cross section 148 of species, m, is 149

Table 1 :
Electron-atom ionization cross-section fit coefficients for Equation2.Coefficients are given in the form a(b) = a × 10 b . 264

Table 2 :
Summary of the quantum chemistry methods and references.N H 2 ≈ 10 20 − 10 23 cm −2 , although we note that the co-of the orbitals, and thus facilitates detachment for an electron.According to Koopmans' theorem, the lowest eBE is equal to the ionization potential of the molecule.The values of the EPT-eBE and the IP computed at our best level of theory (CCSD(T)/CBS) agree excellently.The discrepancies are about 0.2 eV and are much improved with respect to HF-eBEs.Given that the lowest eBE carries the most weight in the BEB cross section, it is paramount to ensure that it is accurate.Our results not only exhibit a marked improvement over canonical HF values, 295efficients only marginally scale with column density.Hereafter, 296 we denote cm,T as c m,T , due to the marginal scaling with column 297 density.In general, since the BEB cross sections perform better 298 than the SCAR data in comparisons against experimental data, 299 we report here only the coefficients using the BEB cross section, 300 although all cross sections are available in the database.3013.Results302In total, we have computed the structure, orbitals, and cross sec-303 tions for 202 unique (by composition, not counting isomers) 304 neutral molecules ranging in size from 2 to 24 atoms, includ-305 ing data augmented by the results from Heathcote & Vallance 306 (2018).Figure 2 shows the distribution of the number of atoms 307 for the molecules in our database.While most of the molecules 308 have fewer than 6 atoms, we include some with up to 24 atoms 309 (C 14 H 10 ).When KIDA does not specify an isomer but multiple 310 isomers exist, we took the most stable isomer.We detail a sum-311 mary of the results below, with all of the data available online in 312 the public database 13 .313 3.1.Ionization potentials, electron binding energies, and 314 electron kinetic energies 315 We present here the ionization potentials computed at both the 316 DFT (CAM-B3LYP) and CCSD(T)/CBS levels, and we com-317 pare them to the experimental value recommended in NIST.Ta-318 ble A.4 shows the resulting ionization potentials in eV for the 319 molecules.Figure 3 shows a comparison between the differ-320 ent calculations and the NIST database.In general, our calcu-321 lations for molecules with fewer than seven atoms are consistent 322 with the values presented in the NIST database.For molecules 323 with higher ionization potentials, our calculated IPs are generally 324 higher.The agreement between the CAM-B3LYP and CCSD(T) 325 calculations is relatively good, but there is a general trend that 333 the latter method are smaller than those obtained with HF since 334 they are corrected to include the instantaneous electron-electron 335 repulsion (electron correlation) in its energy, which increases the energy An example of the NIST 493 formatted file, co.norb

Table 3 :
Unique molecules included in this database (202), sorted by their number of constituent atoms.

Table 5 :
Molecular orbitals of CO computed at the MP2/aug-cc-pVTZ level.

Table 6 :
Same as Table 5, but for H 2 O.
Article number, page 15 of ??

Table A
Article number, page 16 of ??

Table A
Article number, page 17 of ??

Table A
Article number, page 18 of ??

Table A
Article number, page 19 of ??

Table A
Article number, page 20 of ??

Table A
Article number, page 21 of ??

Table A
Article number, page 22 of ??

Table A
Heathcote & Vallance (2018)fficients following Equation14for all neutral molecules in the database.The boxes show the minimum and maximum coefficients in the column density range 10 20 cm −2 ≤ N H 2 ≤ 10 23 cm −2 , and plus signs show the mean.The cyan boxes denote molecules using the MP2 calculations, and the red boxes denote calculations fromHeathcote & Vallance (2018).