WorksPhysics and Astronomy
Rotational Excitation of HCNH by He: Collision Dynamics and Astrophysical Applications
Abstract
The protonated cyanopolyyne HCNH is an interstellar molecular ion of astrochemical interest, and understanding its rotational excitation is essential for interpreting its emission in molecular clouds. In this work, we construct a two-dimensional potential energy surface (PES) for the HCNH–He system using the CCSD(T)-F12B/cc-pVTZ-F12 method. Rotationally inelastic cross sections are calculated using the quantum close-coupling (CC) method, and the corresponding collisional rate coefficients are obtained by thermal averaging over a Maxwell–Boltzmann distribution. These rate coefficients are incorporated into a non-local thermodynamic equilibrium (non-LTE) radiative transfer model to investigate the excitation of HCNH in the cold dark cloud TMC-1. A global minimization analysis of seven observed transitions yields best-fitting values of cm and cm, with a minimum reduced of 0.87. The modelled line intensities are in good agreement with the observations. This study provides the first-principles collisional data needed to model the rotational excitation of HCNH and offers new constraints on its physical and spectroscopic properties in TMC-1.
Introduction
Protonation gives molecules large permanent dipole moments and small rotational constants, making the resulting ions detectable radio tracers in cold dark clouds and providing indirect evidence for their neutral parents (Agúndez & Wakelam 2013). Among the carbon-chain species found in these environments, cyanopolyynes and their protonated counterparts, HCN and HCNH (), form an important homologous series. While neutral cyanopolyynes have been extensively studied (McGuire 2018), their protonated counterparts have been far less extensively detected. To date, the members reported in the protonated series include HCNH and the longer-chain species with –3 (Ziurys & Turner 1986; Kawaguchi et al. 1994; Marcelino et al. 2020; Cabezas et al. 2022), with their column densities generally decreasing as the carbon-chain length increases.
HCNH is the focus of this work. Its proposed interstellar detection is based on seven lines with nearly harmonic frequency spacing found in the Yebes 40 m survey of TMC-1 (Marcelino et al. 2020), with HCNH as their most likely carrier. This assignment is supported by theoretical rotational constants and by the similarity between the abundance ratios of HCN/HCNH and HCN/HCNH. However, in the absence of laboratory rotational spectroscopy, its detection remains tentative. Laboratory infrared spectroscopy has provided direct frequency references for the identification of HCNH (Lee & Amano 1987; Kawaguchi et al. 1990), whereas the assignments of longer-chain protonated species have relied largely on rotational constants and dipole moments predicted by high-level coupled-cluster calculations (Botschwina et al. 1997; Heyl et al. 1997). We therefore adopt this assignment as a working hypothesis for the collisional and radiative transfer analyses.
The observationally inferred column densities of these ions depend on the rotational level populations, and collisional excitation in cold dark clouds is generally insufficient to maintain local thermodynamic equilibrium (LTE). Studies of HCNH collisions with He and H, together with hyperfine-resolved data for its isotopologues, have demonstrated how dedicated potential energy surfaces (PESs) and state-to-state rate coefficients affect spectral predictions (Bop & Lique 2023; Bop et al. 2024, 2026). However, because accurate rate coefficients for long-chain molecules are extremely demanding to compute, quantum scattering results have so far been reported only for HCNH–He [3], and none are available for HCNH or longer chains. Moreover, the effects of increasing chain length on the potential anisotropy and transition propensities cannot be directly extrapolated from shorter-chain systems.
In this work, we construct a two-dimensional ab initio PES for HCNH–He and perform CC calculations to obtain rotationally inelastic cross sections and low-temperature rate coefficients. The resulting collisional data are then used in non-LTE radiative transfer calculations and fitted to the observed line intensities towards TMC-1, in order to determine its excitation properties and constrain its column density. The remainder of this paper is organised as follows. Section 2 describes the construction and Legendre expansion of the HCNH–He PES. Section 3 presents the calculated cross sections and rate coefficients. Section 4 discusses the astrophysical applications and the non-LTE radiative transfer results, and § 5 summarises our conclusions.
Theoretical methods and computational details
Potential energy surface
Within the rigid-rotor approximation, the molecular geometry of HC₅NH⁺ is fixed at the equilibrium structure of its ground state, and only its overall rotation is considered. The lowest vibrational frequency of the ion is approximately 99 cm⁻¹ (Marcelino et al. 2020). At a typical interstellar temperature of 20 K, the Boltzmann factor for the first vibrationally excited state, , is only about 0.1%. Vibrational excitation can therefore be neglected over the temperature range considered in this work (Lique et al. 2006; Faure et al. 2020). This approximation has been widely adopted for similar weakly interacting systems, including SiC₄–He (Chefai et al. 2024), HNSi/HSiN–He (Tian et al. 2026a), and CaO/CaS–He (Tian et al. 2026b).
HC₅NH⁺ is a linear conjugated polyyne ion. In the centre-of-mass frame, the position of He relative to the linear HC₅NH⁺ molecule in its electronic ground state is described by the Jacobi coordinates , where is the distance between He and the centre of mass of the ion, and corresponds to the arrangement HC₅NH⁺—He, as illustrated in Figure 1. The bond lengths are calculated at the CCSD(T)-F12B/cc-pVTZ-F12 level of theory and are as follows: Å, Å, Å, Å, Å, Å, and Å.

Figure 1. Geometry of the HC₅NH⁺–He collision system and the Jacobi coordinates.
The interaction energy is calculated using the explicitly correlated coupled-cluster method CCSD(T)-F12B and the cc-pVTZ-F12 basis set (Knizia et al. 2009; Werner et al. 2012). The radial grid covers –30 Å, spanning the short-, intermediate-, and long-range regions. The angle is sampled uniformly from to in steps of , yielding a total of 2553 geometries. The basis-set superposition error (BSSE) is corrected using the Boys–Bernardi counterpoise scheme:
where , , and are the energies of the complex, the HC₅NH⁺ monomer, and the He atom, respectively, all calculated within the same set.
The resulting PES is shown in Figure 2. It exhibits pronounced axial asymmetry and a lateral double-minimum structure. Its global minimum lies on the axis at the protonated end (), reaching a depth of 194.55 cm⁻¹ at Å. By contrast, the well at the N end () is much shallower. Laterally, the interaction exhibits two asymmetric minima, separated by a shallow saddle point near with a depth of about 58 cm⁻¹; consequently, the 60 cm⁻¹ contour in the lateral region splits into two closed loops. The PES exhibits four distinct low-energy regions associated with the H₁ (+0.45680), C₂ (+1.30156), C₄ (+1.63499), and H₂ (+0.29715) sites of HC₅NH⁺, corresponding to the four main low-energy regions in the contour map. Their spatial distribution reflects the pronounced anisotropy and end-group asymmetry of the interaction potential, with the deepest well located near the protonated H₂ end.

Figure 2. Contour plot of the HC₅NH⁺–He interaction potential (in cm⁻¹). Blue contours indicate positive values, while red contours indicate negative values.
Legendre expansion of the PES
In the short-range region, the raw interaction energy exhibits an unphysically steep increase due to orbital overlap and the numerical behaviour of the Boys–Bernardi correction. To address this issue, before performing the Legendre expansion, we apply the smoothing and scaling method of González-Sánchez et al. (2024) to the raw ab initio energies. Denoting the raw and smoothed energies by and , respectively, the procedure is defined as
where the threshold parameters are cm⁻¹ and cm⁻¹, and the switching function is . This scheme ensures that and its first derivative are continuous at the joining points, while limiting the high-energy repulsive wall to a finite value, thereby avoiding numerical instabilities in the scattering calculations.
For use in the scattering calculations, the two-dimensional PES is expanded in Legendre polynomials:
where is the Legendre polynomial of order and are the radial coefficients. To reduce the computational cost while maintaining accuracy, the expansion is truncated at , for which the mean deviation of the reconstructed PES from the ab initio values is less than 0.5%.
Figure 3(a) shows the dependence of the first five radial coefficients. Among the anisotropic terms with , has a markedly larger amplitude than the others, followed by , indicating the dominance of the even-order terms. The odd terms and have appreciable amplitudes only at short range and decay rapidly with increasing . The dominance of the even-order terms reflects the strong long-range induction and dispersion attraction experienced by He as it approaches the conjugated carbon chain from the side, which also explains the propensity for transitions during collisions.

Figure 3. (a) Radial dependence of the first five Legendre radial coefficients (–4, in cm), as functions of for the HCNH system. (b) Radial dependence of the long-range interaction potential.
The long-range fit provides a further test of the asymptotic behaviour of the PES. Figure 3(b) shows that, in the long-range region Å, the ab initio points agree well with both and the fit, consistent with an asymptote dominated by the charge-induced-dipole interaction (Buckingham 1978). For Å, the ab initio points lie systematically above the asymptote, indicating that contributions from repulsion and higher-order terms such as are no longer negligible. This also explains why the analytic potential adopts the asymptotic form only in the long-range region, while the inner region is described by a Legendre expansion. The two branches are joined such that their first derivatives are continuous, avoiding unphysical kinks or barriers in the low-energy scattering calculations. The deviation between the ab initio points and the fit decreases rapidly with increasing and tends to zero, supporting the use of this analytic PES in subsequent low-energy scattering calculations.
Scattering cross sections and rate coefficients
Using the PES described above, we employ the MOLSCAT code (Hutson & Le Sueur 2019) to calculate the inelastic scattering of an atom with HCNH within the CC framework (Arthurs & Dalgarno 1960), obtaining state-to-state cross sections for collision energies up to 300 cm. For HCNH, we adopt the rotational constant cm and the centrifugal distortion constant cm (Marcelino et al. 2020).
The partial-wave sum for the integral cross section from to is given by:
where is the collision energy, is the total angular momentum, and are the orbital angular momenta of the initial and final channels, respectively, and is the Kronecker delta. The channel wavenumber is defined as , where amu (unified atomic mass units) is the reduced mass of the HCNH–He system. The rotational energy levels of the molecule are given by:
To ensure the convergence of transition cross sections involving rotational levels up to , the rotational basis truncation parameter was systematically optimized for different energy ranges. Accordingly, an energy-dependent rotational basis was adopted: for cm and for cm. For the transitions considered, the numerical uncertainty of the computed cross sections using these values is less than 5%, validating that the adopted basis provides sufficiently converged cross sections for the target levels. The corresponding integration step size control ensures 50, 20, and 10 steps per half-wavelength for –20, 20–100, and 100–300 cm, respectively. To adequately sample the threshold and resonance regions, the collision-energy grid spacings are set to 0.05, 0.1, 1, and 5 cm for the energy intervals of , , , and cm, respectively. At each energy point, the total angular momentum , controlled by JTOTU, is iteratively increased until convergence is reached (with an upper limit set to 99999). The summation process terminates automatically when the maximum changes in both diagonal and off-diagonal cross sections fall below their respective thresholds (DTOL = 0.1 Å, OTOL = 0.005 Å). To fully resolve the resonance features, the radial coupled equations are divided into two propagation stages: a log-derivative propagator is used in the short-range region (Manolopoulos 1986), and a variable-step Airy propagator is used in the long-range region (Alexander & Manolopoulos 1987). All calculations are solved using the MOLSCAT program (Hutson & Le Sueur 2019).
Figure 4 shows the de-excitation dynamics for high- initial states, and Figure 4(a) presents the de-excitation cross sections for –19. Below 10 cm, shape resonances associated with tunnelling through the centrifugal barrier and Feshbach resonances arising from quasi-bound states in the potential wells produce dense oscillations [3]; at very low energies, the cross section decreases rapidly with increasing energy, approximately following the long-range capture scaling (Gioumousis & Stevenson 1958). Above 30 cm, the cross sections depart from this scaling and decline slowly, indicating that long-range capture is no longer the limiting factor and that the dynamics become governed by the short-range anisotropic potential and exchange repulsion. At the low-energy end, the cross section is the largest, whereas the cross section is the smallest over most of the energy range; as the energy increases, the decrease in the cross section becomes markedly less pronounced and its cross section becomes the largest, with the clearest advantage at high energy. This energy-dependent selectivity is consistent with the anisotropy of the PES being dominated by the term.

Figure 4. (a) De-excitation integral cross sections for initial states –19 as functions of collision energy (on a log–log scale). (b) The corresponding de-excitation rate coefficients as functions of temperature.
Maxwell–Boltzmann averaging of these state-to-state cross sections over a 1 K grid from 3 to 50 K yields the state-to-state (de-)excitation rate coefficients for .
where is the de-excitation rate coefficient from the upper level to the lower level , and is the Boltzmann constant. The corresponding excitation rate coefficients, , are obtained from the principle of detailed balance (En Abdelhadi et al. 2012):
Figure 4(b) shows the temperature dependence of the de-excitation rate coefficients: the channel increases monotonically and substantially with temperature, becoming dominant at higher temperatures. This behaviour is directly related to its cross section, which decreases most slowly at high collision energies and forms a plateau above 30 cm (Figure 4a): as the temperature rises, high-energy collisions gain weight and the rate continues to grow. The cross section decreases steadily with energy, and its thermally averaged rate coefficient becomes nearly constant above approximately 20 K, consistent with ion–induced-dipole Langevin capture (Gioumousis & Stevenson 1958); the channel has the smallest rate and the weakest temperature dependence. Differences among initial states within the same are small; only for does the rate coefficient increase slightly with increasing initial , following the same trend as the high-energy cross sections in Figure 4(a). Thermal averaging smooths out the dense resonances at low collision energies, resulting in smooth temperature dependences of the rate coefficients. The weak dependence on the initial rotational level also indicates that this system is less sensitive to the rotational quantum number than neutral molecular systems.
Astrophysical applications
In low-density clouds such as TMC-1, collisional excitation is generally insufficient to fully thermalise the level populations (Bop & Lique 2025), so the column density derived by Marcelino et al. (2020) under the LTE assumption may be subject to systematic bias. Using the rate coefficients derived above, we can quantitatively assess this effect with non-LTE modelling.
Non-LTE radiative transfer simulation is computed with RADEX (van der Tak et al. 2007). The code adopts the escape-probability approximation and iteratively solves the statistical equilibrium equations for the level populations, assuming a uniform-sphere geometry. Because no direct HCNH–H collisional data are currently available, the H rate coefficients are obtained by scaling those for HCNH–He. Since He is not a direct surrogate for H, the scaling must account for differences in both reduced mass and polarisability:
where and are the polarisabilities of He and H, respectively; the calculated scaling factor is (González-Sánchez et al. 2024). Because calculations involving He are computationally less demanding, He is often used as a proxy when H collision data are unavailable (Bop & Lique 2025). The linewidth is set to the mean of the seven transitions listed in Table 1 of Marcelino et al. (2020), approximately 0.65 km s.
Figure 5(a) and Figure 5(b) present the non-LTE radiative transfer results for the representative transitions (top) and (bottom), respectively. Each transition is shown in four panels corresponding to , 10, 15, and 20 K. Solid lines show the brightness temperature , long-dashed black lines the excitation temperature , and short-dashed purple lines the optical depth multiplied by 100; the three colours correspond to the three column densities. Because in all models, the lines are entirely optically thin and radiative trapping is absent, and is approximately proportional to the column density at low densities.

Figure 5. and for the (top) and (bottom) transitions as functions of the H volume density at three HCNH column densities.
At the low-density (), both transitions are subthermally excited (). As increases, collisions become increasingly dominant, and approaches . levels off for . The variation of with density merely reflects differences in the populations of the upper and lower levels. Under these conditions, the plateau level of is determined mainly by , which has approached at these densities.
A comparison of the two transitions reveals a clear rotational-level effect: because involves a higher upper-level energy and a larger energy gap, a higher density is required for it to thermalise. At the same , as long as the transitions are not fully thermalised, of always exceeds that of ; the difference is most pronounced at intermediate densities and then narrows as both lines approach . For the K model, at , of is already about 16.6 K, whereas that of is only about 8.4 K, roughly half as high, showing that the latter thermalises significantly later. As the density rises further (e.g. to ), the transition still lags behind and approaches the kinetic temperature only at about , indicating that the system is progressing towards a near-LTE state.
The density at which these plateaus occur can be understood in terms of critical density (). For a transition , the critical density is the collider density at which the rates of radiative and collisional de-excitation are equal (van der Tak et al. 2007):
where and denote the initial and final levels, is the Einstein A coefficient for spontaneous emission from to . For the transitions considered here, we calculated the Einstein A coefficients for all channels within ; the data are available as described in the Data availability section. Here, is the rate coefficient for collisional de-excitation from the initial level to each lower level , the summation in the denominator runs over all possible de-excitation channels.
| 5 K | 10 K | 15 K | 20 K | |
| 13 → 12 | ||||
| 14 → 13 | ||||
| 15 → 14 | ||||
| 16 → 15 | ||||
| 17 → 16 | ||||
| 18 → 17 | ||||
| 19 → 18 |
Table 1. Critical densities of the seven observed transitions at 5–20 K.
Table 1 lists the values for the seven observed transitions at 5–20 K. As the rotational quantum number increases, increases rapidly with the cube of the transition frequency, and grows from for to for ; for a given transition, the varies by less than 4% over 5–20 K, indicating a much weaker temperature dependence than that found in neutral molecular systems (e.g. HCN–He; Wernli et al. 2007). The typical density of TMC-1 (about ) lies within this range: the low- transitions are nearly thermalised but not fully, with values slightly below , and their line intensities mainly constrain the column density; the high- transitions remain subthermally excited and are sensitive to density, providing the density information. This analysis indicates the density sensitivity of each line without detailed fitting, and guides the density range of the radiative transfer grid and the joint fit. Furthermore, it demonstrates that non-LTE treatment is essential under the physical conditions of TMC-1 and that the density sensitivity is confined to the subthermal regime, where the line ratios respond to collisions.
To constrain the observational parameters, we perform a joint fit to the seven lines observed by Marcelino et al. (2020) using the statistic of Lamption et al. (1976):
where and are the calculated and observed integrated line intensities, respectively, and is the observational uncertainty.
Figure 6 shows the two-dimensional minimisation grid, where the colour scale represents and the silver star marks the best-fitting solution; the contours, from the innermost to the outermost, correspond to , 6.18, and 11.83, i.e. the , , and confidence regions for a two-parameter fit (Lamption et al. 1976). The best fit is located at and . The contours form a horizontal band: the region constrains the column density to approximately – and extends towards higher densities. Because the seven lines have different and respond differently to density in the subthermal regime, the joint fit sets a lower limit on . At the best-fit density of , which exceeds all , the of the lines approaches , and the contours flatten along the density direction; the resulting density degeneracy is therefore a natural consequence of the loss of diagnostic information after thermalisation, rather than a deficiency of the fitting procedure itself.

Figure 6. Distribution of in the – plane at K. The best-fitting solution is marked by the silver star. The white solid, black dashed, and red short-dashed contours (from inside out) denote the , , and confidence level (, 6.18, and 11.83, respectively).
To further evaluate the goodness-of-fit and compare it with previous studies, Figure 7 displays the one-dimensional curve as a function of column density at the best-fit density of . The curve exhibits a single minimum at , yielding a reduced . A reduced value is not uncommon in astrochemical analyses, typically indicating either overestimated observational uncertainties or an excellent fit to the data, as seen in other non-LTE studies (e.g. Agúndez et al. (2023)). Compared to the value of cm derived from the LTE rotational diagram by Marcelino et al. (2020), our result is lower by a factor of , yet the two values are consistent within . The direction and magnitude of this offset align exactly with the expected systematic difference from the LTE assumption: LTE tends to overestimate the populations of high- levels, and consequently the column density (Agúndez et al. (2023)). The agreement in magnitude indicates that the current rate coefficients introduce no significant systematic bias in the radiative transfer modelling, thereby further validating the non-LTE treatment and the spectral line assignment.

Figure 7. Variation of with the column density at the best-fitting value of . The solid blue curve shows the fit, the dashed red line marks the best-fit column density, and the grey band shows the column density range derived from the LTE rotational diagram of Marcelino et al. (2020).
Conclusions
We have investigated the rotationally inelastic collisions of HCNH with He and examined their implications for the excitation of HCNH in interstellar clouds. The calculated collision dynamics exhibit a clear propensity for transitions, which is closely related to the dominant contribution of the component to the PES anisotropy. This highlights the important role of the interaction potential in determining the rotational transition propensities of the system.
The non-LTE analysis towards TMC-1 indicates that the observed HCNH lines are subthermally excited. While the derived column density agrees with the LTE estimate within , the volume density is constrained only from below, suggesting that HCNH is therefore a robust column-density tracer, while its density diagnostic power is limited by thermalisation. Future studies could extend this work to other homologous ions (e.g. ) and isotopologues (e.g. ) to investigate whether the “even- dominance” is a general collisional characteristic of long-chain protonated cyanopolyynes. Furthermore, the highly accurate state-to-state collisional rate coefficients presented here will provide a robust dataset to support radio observations and non-LTE analyses of and related species in molecular clouds.
Data Availability
The collisional rate coefficients and the associated molecular data required for the RADEX calculations will be deposited in the Zenodo repository and submitted to the LAMDA database upon acceptance for publication.
Acknowledgements
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
This paper has been typeset from a \TeX/\LaTeX file prepared by the author.
References
- [3]Al Mogren M. M., Ben Abdallah D., Al Harbi S. D. A., Al Salhi M. S., Hochlaf M., 2021, Mon. Not. R. Astron. Soc., 503, 2902