Data on Dissociative Electron Attachment Accommodated in the Structure of Belgrade Collisional Database ACol

Abstract

This work presents an extension of the Belgrade ACol collisional database within the Virtual Atomic and Molecular Data Centre (VAMDC) framework to include dissociative electron attachment (DEA) processes. DEA, a low-energy electron-driven resonant mechanism leading to molecular fragmentation, is relevant in fields such as plasma science, radiation damage, nanofabrication, and EUV lithography. The ACol data model was redesigned to improve semantic clarity and flexibility by separating physical collision data from bibliographic and serialization structures. A new DataSource entity and a redefined TabulatedData–Collision relationship enable a more normalized database while preserving compatibility with the XSAMS schema through dynamic, query-time construction of hierarchical structures. The implementation is demonstrated using DEA to isoflurane, incorporating experimental data from independent studies and cataloging resulting fragment anions and energy-dependent yields. Elastic cross sections of isoflurane are included as they are immanently connected to DEA, providing an essential tool for understanding the complex DEA process. The novelty of the present implementation lies not in the first representation of DEA within VAMDC, but in its integration into a multi-process collision database based on a compact, normalized data model that separates scientific data and bibliographic provenance from the XSAMS serialization structure. The updated architecture enhances interoperability, reduces redundancy, and improves data management while maintaining compliance with VAMDC standards.

1. Introduction

In this paper, we discuss improvements made to the collisional database ACol, which is one of the Belgrade nodes within the Virtual Atomic and Molecular Data Centre (VAMDC). The VAMDC Consortium [1,2] is an association of 39 independently maintained and curated databases in the field of atomic, molecular, and optical physics whose member databases contain data on elementary collisional and radiative processes, including their cross sections, rates, and energy dependencies, as well as elementary data such as wavelengths, optical strengths, transition probabilities, etc. The existence of such databases is essential for both data providers and data users, since as they provide a reliable, centralized source of scientific data that would otherwise be dispersed across myriad scientific publications. Such databases facilitate research in many disparate areas, for example, in models and observations of the interstellar medium and planetary atmospheres [3] or in the analysis of experiments using swarm techniques [4].

The purpose of this work is to extend the coverage of collisional processes currently represented in ACol [5]. Beyond the five processes, namely, excitation, associative ionization, dissociative recombination, electron–ion–atom recombination, and Penning ionization, we now include the process called Dissociative Electron Attachment (DEA). The DEA process is relevant at low electron impact energies, where an electron interacting with a molecule forms a temporary negative ion (TNI). This is a resonant process, highly dependent on the electron impact energy, such that the TNI is also known as a resonance. Such a resonance is a short-lived species that dissociates into an anion and neutral fragment(s). Alternatively, the TNI can re-emit the electron and thus return to elastic scattering. The competing process is spontaneous electron detachment (autodetachment) [6,7]. That is why it is always important to study the elastic scattering process in addition to DEA. Elastic electron scattering experiments are used to detect the resonances that are required in DEA so that electrons may be captured at specific energy levels. Theoretical descriptions of DEA rely extensively on elastic electron–molecule scattering data. The cross section of a molecule undergoing DEA is deeply tied to the temporary trapping and scattering potentials of the target molecule.

Beyond its resonant nature, DEA is characterized by pronounced bond and site selectivity. For isolated molecules, electron attachment is strongly influenced by both the symmetry properties of the chemical bonds and the spatial arrangement of the constituent atoms. Dissociation can occur via either a direct pathway, arising from the population of repulsive σ* states, or an indirect predissociation mechanism involving transient occupation of π* orbitals. Depending on the initial electron energy and electron affinity of the molecule, TNIs may be classified as single-particle shape resonances, core-excited shape resonances, or core-excited Feshbach resonances. A comprehensive review of the various types of DEA processes, their underlying molecular mechanisms, and representative examples illustrating their selectivity is provided in [7].

Given the importance of the DEA process, dedicated DEA Club meetings are held biennially or triennially. An overview of contemporary research activities within the DEA community is provided in the report of the fourth DEA Club meeting [8]. The inclusion of DEA processes in the ACol database was motivated by discussions at the fifth DEA Club Meeting, held in 2026 in Sopot, Poland, which highlighted the need for their systematic development and integration [9].

Current studies of the DEA process include many important fields, including technological plasmas, radiation damage (RADAM), microhydrated radiosensitizers, focused electron beam-induced processing (FEBIP), and EUV lithography. A schematic illustration of different aspects of the DEA process is presented in Figure 1. The role of DEA in RADAM has been well established since 2000, when the group of Leon Sanche [10] showed that low-energy electrons may cause single- and double-strand breaks in the DNA helix via the DEA process. Recently, the complexity of DNA radiation damage has been investigated using DNA nanotechnology, illuminating the role of DEA in DNA fragmentation [11]. Radiosensitizers are molecules that increase DNA susceptibility to radiation-induced damage, which is extremely important under hypoxic conditions. The recent perspective article [12] reviews both established candidate radiosensitizers and promising future directions in radiosensitizer development. Radiosensitizing nucleosides include uracil-based derivatives or novel purine-based derivatives [13] that undergo DEA, yielding DNA strand breaks. Recent advances have highlighted the broader role of DEA processes in the design and optimization of radiosensitizers for biomedical applications. A comprehensive review by Sedmidubská and Kočišek [12] summarizes current experimental and theoretical studies of low-energy electron interactions with radiosensitizing molecules and discusses their implications for enhancing radiation-induced DNA damage. Extensive experimental and theoretical studies are also being performed on electron attachment to different molecules and clusters, both in the gas and hydrated phases [14,15]. In FEBIP, low-energy electrons play a pivotal role, with DEA inducing the dissociation of precursor molecules and forming nanostructures [6,16]. Acid generator resists for EUV lithography, operating at 13.5 nm wavelength, have been recently proposed that exploit DEA and its ability to selectively fragment molecules [17].

Although it has been known since 2021 that the VAMDC infrastructure can accommodate DEA data, when the Innsbruck Dissociative Electron Attachment Database (IDEADB) was established [18], it has remained an open question whether the Belgrade VAMDC nodes could also support this type of data. From the perspective of reaction databases, in addition to VAMDC, the LXCat project provides a repository of cross-section data for reactions needed to model electron and ion swarm parameters in non-equilibrium low-temperature plasmas [19]. The main processes covered are elastic electron scattering, electron impact excitation, and ionization. However, a complete database representation of electron and ion kinetics would require the inclusion of additional reaction types and processes, including DEA [20]. Such an extension requires the definition of a new data format, the selection of an appropriate database type, and the design of the software stack [21]. These findings underscore the importance of incorporating DEA processes into reaction databases to enable more comprehensive kinetic modeling.

DEA may produce several chemically distinct anion fragment channels from the same parent molecule, each associated with a separate energy-dependent yield and ion mass [22]. Database representation therefore requires explicit identification of individual ionic products and a serialization mechanism capable of returning multiple ion species within the appropriate XSAMS atomic structure. It appears natural that the Belgrade electron/atom (molecule) database (BEAM) [5] should accommodate the DEA process since one of the reactants is always an electron. However, another Belgrade node, ACol, Collisional Atomic Processes (Excitation–Ionization), was selected to host the DEA data because its reactant product structure is better suited to representing multiple fragmentation channels. In this paper, we present how the ACol database has incorporated DEA data. Isoflurane was selected as an initial case study due to the availability of recent measurements in which DEA data were independently obtained by two separate research groups [23,24].

2. Methods Used to Describe DEA Process in ACol Database

2.1. DEA to the Volatile Halogenated Anesthetics

Studies of volatile halogenated anesthetics are important because of their prevalence in clinical practice [25]. There is also concern about their contribution to global warming since they have a long lifetime in the atmosphere and a relatively high global warming potential (GWP) [26]. Experimental studies on the DEA process were carried out using monochromatic electron beams in a crossed-beam arrangement with mass spectrometers, either two-sector field mass spectrometers [23] or conventional quadrupole mass analyzers [24], to detect the anion fragments.

Isoflurane (CF₃–CHCl–O–CHF₂) is designated as 2-chloro-2-(difluoromethoxy)-1,1,1-trifluoroethane according to IUPAC nomenclature, but it can also be regarded as 1-chloro-2,2,2-trifluoroethyl difluoromethyl ether. All hydrofluoroethers strongly absorb in the VUV region. Absorption cross sections have been measured for isoflurane in the range from 5.0 eV to 10.8 eV (115–248 nm) [27], while infrared spectra have recently been measured in the range from 1650 to 650 cm⁻¹ [26]. Reactions with hydroxyl (OH·) radicals are generally used to elucidate the GWP of hydrofluoroethers as long-lived species [28].

2.2. Modifications of ACol Data Model and Schema Generation

The ACol [5] database and service, hosted by the Serbian Virtual Observatory1 is a VAMDC [1] node that operates within the broader ecosystem of heterogeneous, interoperable data nodes with a unified web service and centralized portal access. The implementation of a VAMDC node is typically derived from the upstream version of the NodeSoftware2, which provides a common framework for data access and XSAMS [29,30] XML serialization. Individual nodes incorporate specific data models and a range of customizations to accommodate the structure and semantics of their underlying datasets, while conforming to the serialization standard.

We revised the ACol data model and retrieval workflow to support DEA data within the existing multiprocess structure of the database while preserving compatibility with the VAMDC XSAMS standard. The updated organization separates the physical description of collision processes, numerical datasets, and bibliographic provenance more clearly, allowing multiple reaction channels and independent measurements to be represented consistently without unnecessary duplication. Figure 2 provides an overview of the principal database entities and their relationships. A detailed description of the database refactoring, relationships between individual entities, and the extended XSAMS serialization procedure is provided in Appendix A. This revision of the software exhibits a cleaner separation between domain semantics and serialization concerns, reduces redundancy, and leverages dynamic composition to bridge the gap between an optimized relational model and a rigid external XML schema.

The overall ACol workflow is summarized in Figure 3. The web interface and external VAMDC clients provide two distinct access paths to the same underlying relational database. Browser-based functions, including the newly introduced Overview dashboard and Explore Data tools, as well as the qn plots, are handled through dedicated Django views that return statistics, filtered dataset descriptions, source metadata, raw numerical values, and plot output. Standardized data retrieval is performed through the VAMDC-TAP endpoint, where VSS2 queries are translated through the node dictionaries and query functions into database operations before the results are serialized as XSAMS XML. Both paths rely on the same normalized persistence layer, while the XSAMS-specific hierarchical structures are constructed dynamically during query processing rather than stored directly in the database.

2.3. Database Access, Curation, and Data Quality

ACol is a curator-managed database rather than an open repository in which users directly upload records. New datasets are selected and entered by the ACol team, generally from peer-reviewed publications or through direct collaboration with data producers. Scientists may propose additional datasets or provide numerical data and metadata to the authors, but the final mapping of species, states, reaction channels, units, sources, and tabulated values is performed by the database curators.

Validation is carried out at two levels. First, the database representation is checked for technical consistency, including species identity, charge and mass information, reactant product assignments, units, the declared number of numerical values, relations between database objects, and successful XSAMS serialization. Second, the entered numerical arrays and bibliographic metadata are compared with the original publication, and graphical inspection is used to identify transcription errors, discontinuities, or incorrect unit assignments. This procedure verifies the faithful representation of the published data but does not constitute an independent experimental or theoretical reassessment of the source results.

Uncertainties are not generated or estimated by ACol; they must originate from the source publication. The present data model stores the reported independent and dependent numerical values, their units, descriptions, and bibliographic provenance, but does not yet provide a dedicated machine-readable array for point-wise uncertainties. Where uncertainty information is available, users should consult the linked original publication. The structured representation of uncertainties and additional experimental metadata is a possible extension of the data model.

3. Results

3.1. Data on DEA to Isoflurane Molecule

We have used the results of DEA for the isoflurane molecule obtained by two independent groups of researchers [23,24] as a case study. Within the entity Sources, all relevant data from the publication are listed: title, type of publication, all authors, year of publication, volume, page range, unique identifier, and digital object identifier. Within a class Species, both reactants and products are specified. One of the reactants is always a particle, in this case an electron, while the other is a molecule defined with its ordinary structural formula, stoichiometric formula, ion charge, chemical name, InChI, and InChIKey. Additionally, a description of the molecular state is provided; however, in the present work, only the ground state is considered. Within the Processes entity, there are two distinct paths, radiative and collisional. Collisional transitions are distinguished by a unique ID. The process itself is a class with its name, VAMDC code, and IAEA code. Data sets for DEA comprise two axes, X as electron energy and Y as anion yield. Selected fragment anions from DEA to isoflurane are listed in

Table 1. Selected fragment anions from DEA to isoflurane presented in the ACol database. Positions of peak maxima in ion yields, the ion species and values of m/z are presented.

Kopyra et al. [24] (eV)	Matias et al. [23] (eV)	Anion	m/z
6.2 & >9	3.3 & 7.1 & 9.3	F−	19
0.1	0.6 & 3.3 & 9.2 & 12.6	Cl−	35
0.7 & 2.7	3.0 & 6.9 & 9.5	HF2−	39
0.7 & 2.6	3.0 & 9.3 & 13.1	CFO−	47
1.9	2.9 & 9.2	CHF2O−	67
0.6 & 2.6	3.1	C2F3Cl−	116

together with the values of m/z and peak maxima as identified in the literature. Here, we compare results reported in two independent studies: one conducted through a collaboration between the Innsbruck, Lisbon, and Birmingham research groups [23], and the other by the Siedlce and Belgrade research groups [24]. Datasets are imported into ACol in the form in which they are originally presented in the corresponding references. Consequently, no normalization is performed across datasets from different sources; any normalization is limited to that applied within individual references. The appearance of peak positions within ion yields is shown in the first two columns. Although the two datasets differ in the positions of certain peaks, they also share several common features. The uncertainty associated with the peak energy is estimated to be ±0.2 eV, based on variations in the positions of the calibration peaks observed throughout the experiment. A comprehensive discussion of the comparison is provided in [24].

3.2. Data on Elastic Scattering by Isoflurane Molecule

Due to the role of isoflurane in clinical practice and its high global warming potential, extensive elastic electron scattering experiments and calculations have been performed [31]. Absolute differential cross sections (DCS) have been measured in a crossed-beam experiment for scattering angles from 25° to 125° and at electron impact energies from 50 eV to 300 eV. These data are deposited in the BEAM database. Calculations were performed using the Independent Atom Model and the Screening Corrected Additivity Rule with incorporated interference effects (IAM-SCAR + I). The theoretical DCS span more than five orders of magnitude, whereas the experimental data values span approximately three orders of magnitude, reflecting the fact that the data were acquired over a limited angular range. Experimental data are shown in Figure 4.

3.3. Representation in ACol Database

The revised ACol implementation improves both semantic consistency and operational flexibility. The numerical data are linked directly to the collision processes they describe, while bibliographic provenance is represented independently. The hierarchical objects required by XSAMS are not stored in the database; instead, they are created temporarily during query processing and serialization. This keeps the relational model compact while preserving full compatibility with the XSAMS output format.

Figure 5 presents the newly introduced ACol web interface used for browsing, inspecting, and retrieving atomic and molecular collision data. Whereas the previous interface was primarily oriented toward constructing queries and returning XSAMS output, the redesigned interface now supports direct, interactive exploration of the database before export. It includes an Overview dashboard with database statistics, an Explore Data section for live filtering by collision type, species, and bibliographic source, interactive inspection of individual plots and tabulated datasets, and dedicated ‘qn plots’ for selected state-resolved processes.

The Overview section provides a continuously updated summary of the scientific content currently available in ACol. It reports the total number of collisions, collision types, species, species states, and bibliographic sources, together with distributions of collision records by process type, contributing publication, and species/state occurrence. At the time of revision, ACol contained 359 collision records covering 6 collision types, 30 species, 122 species states, originating from 11 bibliographic sources.

This live overview complements the more detailed Explore Data interface and allows users to assess the scope and composition of the database before searching for individual datasets. Users can examine reaction channels, numerical ranges, source titles, and DOI links, display plots together with tabulated raw X-Y values, and then generate XSAMS output for selected processes. The interface therefore extends ACol from a standards-compliant data endpoint into a practical scientific exploration environment.

The ACol database implementation, including its database schema, software architecture, configuration files, data models, interface components, and documentation, is publicly available through the GitHub (version 2.0.0) repository. A versioned release corresponding to the present work has also been archived on Zenodo to support reproducibility and long-term accessibility [32].

4. Conclusions and Future Directions

The integration of Dissociative Electron Attachment processes into the ACol database represents a significant extension of its scientific scope within the VAMDC infrastructure. By introducing a refined data model that separates physical collision data from bibliographic and structural elements, the database achieves improved semantic clarity, reduced redundancy, and greater flexibility in data management. The adoption of a runtime construction approach for XSAMS-compliant structures preserves interoperability while allowing a more normalized and efficient internal representation. The successful implementation using isoflurane as a case study demonstrates the practical applicability of the updated framework and confirms its ability to accommodate complex, energy-dependent electron-driven processes. Overall, these developments strengthen ACol’s role as a robust and adaptable resource for collisional atomic and molecular data.

Future work on the database may focus, from a domain perspective, on further refinements of the data model to enable a more detailed treatment of molecular states, isotopologues, and excited-state dynamics. In addition, the range of supported collisional processes could be easily further extended beyond the currently implemented set.

From a technical perspective, the platform could be further enhanced as a data-intensive system through improvements in both data access and user interaction layers. This includes query-driven, interactive visualization components for efficient exploration of large collision datasets, as well as backend optimization of data retrieval and aggregation workflows. In addition, the data quality pipeline could be strengthened by combining rule-based validation at ingestion and API level with statistical and machine learning-based anomaly detection over stored data, improving consistency, reliability, and maintainability of the database at scale. Beyond quality assurance, the increasing availability of standardized DEA datasets may enable the development of predictive machine learning approaches for electron-driven molecular processes. Recent work has demonstrated the potential of machine learning methods for predicting and refining DEA properties and electron collision cross sections, including the prediction of DEA properties for halogenated molecules, highlighting the importance of interoperable, high-quality databases for supporting data-driven research in this field [33,34].