Introduction

One of the main goals of chemistry is to understand processes in nature at the deepest atomic and molecular level. Moving beyond the traditional kinetics studies, which follow concentration of chemical species in time, modern reaction dynamics investigations can reveal the motion of atoms in chemical processes using either experimental or theoretical methods. Crossed-beam experiments1,2,3 are capable to detect the outcome of single collision events, while classical and/or quantum simulations4,5,6,7,8,9,10,11,12,13,14,15 can follow the atomic-level reaction pathways from reactants to products. The accurate simulations began with the H + H2 reaction4 in the 1970s and arrived to the reactions of small organic molecules such as CH45,6,7,8, CH3Cl9, CH3I3,10, C2H611,12, CH3OH8,13, CH3CH2Cl14,15, etc. with simple radicals like H6,8, F7,8,13, Cl5,8,11, OH8,12 and/or ions like F−9,10,14, Cl−3, OH−15 in the past 10–20 years. In the case of polyatomic reactants mode-specific dynamics investigations are also possible, where the influence of the initial excitation of different vibrational modes on the dynamics is studied5,7,8,9,11,13. Moreover, even state-to-state dynamics can be described, thereby tracking the mode-specific energy flow from reactants to products2,5,16,17.

In the present study we increase the complexity of the reactant by studying the dynamics of the reaction of glycine (H2NCH2COOH) with the OH radical. The glycine molecule has eight conformers with different structures and energies18,19, which introduce a novel aspect to the dynamics as the various conformers may have different reactivities. Conformer specificity is not an unknown concept20,21,22,23, though it has not been as thoroughly explored as vibrational mode-specificity. Therefore, we aim to set a new level of insight about conformer-specific reactivity by investigating the reaction of different glycine conformers with OH showing how the initial selection affects the conformer distribution of the products. Furthermore, the title reaction has multiple competitive reaction pathways describing hydrogen abstraction from the various sites (functional groups) of glycine, which adds to the complexity of the system. So far only the conformers of the reactants18,19,24,25,26, products27,28,29,30,31,32,33 and transition states and post-reaction complexes28,34,35 of the OH + glycine reaction were characterized in the literature, dynamics simulations have not been reported. Of course, such simulations are extremely challenging due to the above-mentioned complexity of the title reaction. One has to sample a large configuration space in 30 dimensions, and deal with complicated electronic structure computations of an open-shell system to obtain a potential energy surface (PES) which governs the motion of the nuclei in the chemical reaction. In the present study we carry out such simulations using the gold-standard of electronic structure methods36 and an analytical representation of the PES allowing efficient dynamics simulations revealing site and conformer specificity in the atomic-level mechanisms of the title reaction. Furthermore, we show how these simulations can uncover new, unexpected chemistry in a reaction.

Results and discussion

During the mapping of the PES for the OH + glycine H-abstraction reaction, we considered three distinct product channels35. In each case, the hydroxyl radical removes a H atom from one of the three functional groups of glycine—the amino (NH2), methylene (CH2), or carboxyl (COOH) group—resulting in the formation of H2O and a dehydrogenated glycine molecule, with the abstraction site dictating the specific product formed. Interestingly, the present quasi-classical trajectory (QCT) simulations on a newly-developed high-level ab initio full-dimensional PES reveal that the H-abstraction reaction at the carboxyl group can follow a more complex mechanism than previously anticipated. Instead of stopping at the initial step, the dehydrogenated glycine molecule formed via COOH abstraction can undergo further decomposition, breaking apart into CO2 and CH2NH2. Figure 1 presents the structures and relative classical energies of the stationary points along all identified OH + glycine H-abstraction reaction channels, comparing the benchmark potential energy values from our previous study35 (shown in black) with the analytical PES-derived energies (shown in green). The energies of the previously unconsidered products (H2O + CO2 + CH2NH2) are also determined at the same theoretical level as the benchmark mapping. The multi-reference characters of the stationary points (Supplementary Fig. 1) and the root-mean-squares fitting errors are shown in Supplementary Tables 1 and 2, respectively. The small energy differences observed between most of the benchmark and PES energies further validate the accuracy of the developed PES. The somewhat larger deviation in the case of the three-fragment products is due to the fact that this channel was automatically developed without manually adding any geometry to this region. The PES also well describes the various conformers of the stationary points as shown in Supplementary Table 3 in the case of the glycine molecule. Among the three H-abstraction pathways, only the reaction at the carboxyl group exhibits a very low barrier. However, when zero-point energy (ZPE) is taken into account, the carboxyl-mediated pathway also becomes kinetically barrierless. It can be concluded that all three reactions are exothermic. When considering only the dehydrogenation step, H abstraction from the CH2 group releases the most energy, followed by the NH2 and COOH pathways. However, taking into account the subsequent decomposition of the glycine molecule dehydrogenated at the carboxyl group, this process becomes thermodynamically more favorable than H abstraction at the NH2 group.

Fig. 1: Schematic potential energy surface of the OH + glycine H-abstraction reaction.
figure 1

Reaction pathways showing the most accurate geometries of the lowest-energy conformers of the stationary points and displaying the previously-determined35 benchmark classical relative energies in black and the values calculated using the present analytical PES in green.

Full size image

Figure 2 provides insight into how the conformer-specific integral cross section (ICS) varies with increasing collision energy (Ecoll) for each reaction channel, distinguishing eight cases based on trajectories initiated from distinct glycine conformers. The statistical uncertainty of the ICSs can be assessed based on Supplementary Table 4. When the hydroxyl radical targets the carboxyl group of glycine, the reaction rarely stops at the initial dehydrogenation step. More often, it continues with the decomposition of the dehydrogenated glycine molecule, ultimately yielding H2O, CO2, and CH2NH2 as products. Furthermore, at low collision energies, particularly at Ecoll = 1 kcal/mol, the three-fragment product channel exhibits markedly higher reactivity compared to H abstraction at the amino and methylene groups. With increasing collision energy, the reactivity of H abstraction at the carboxyl group gradually declines, eventually reaching a plateau where the ICS stabilizes. Meanwhile, the H-abstraction reactions at the amino and methylene groups become progressively more prominent. The reaction at the CH2 group emerges as the most dominant, driven by its thermodynamic favorability, as it releases the highest amount of energy. It can be concluded that the overall profile of the cross section curves remains consistent across all glycine conformers, regardless of the initial structure from which the trajectories are initiated, with only minor differences in absolute reactivity.

Fig. 2: Site- and conformer-specific reactivity of the OH + glycine system.
figure 2

Integral cross sections as a function of collision energy for the various pathways of the OH + glycine H-abstraction reaction, with each panel specifying a different initial glycine conformer (see insets) denoted by roman numbers in increasing energy order indicating its planar (p) and non-planar (n) symmetry. Numerical values are given in Supplementary Data 2.

Full size image

Figure 3 displays the opacity functions, illustrating the reaction probability as a function of the impact parameter for the reaction of OH with the lowest-energy glycine conformer (Ip), with distinct curves representing H abstraction from the amino, methylene, and carboxyl groups. Opacity functions for all the initial glycine conformers are shown in Supplementary Fig. 2. (Interested readers may also consult with Supplementary Fig. 3 for conformer-specific scattering angle distributions.) The reaction probability curves for H abstraction at the NH2 and CH2 groups exhibit similar profiles, whereas the reactivity of the COOH group follows a distinct pattern. For NH2- and CH2-H-abstraction, the maximum impact parameter (bmax)—the critical perpendicular distance between the velocity vectors of the reactants at which the reaction can still occur—remains constant regardless of the collision energy. In contrast, for H abstraction at the COOH group, bmax decreases with increasing collision energy, highlighting its strong collision energy dependence. The energy dependence may originate from the complex nature of the COOH pathway, where the initial H abstraction is inevitably followed by the fragmentation of the dehydrogenated glycine into CO2 and CH2NH2. Unlike NH2- and CH2-abstraction, which terminate after H transfer, the H2NCH2COO• intermediate is inherently unstable and rapidly decomposes. At low collision energies, longer interaction times allow efficient energy redistribution within the system, facilitating reaction even at larger impact parameters, maintaining a high bmax. As collision energy increases, the interaction time shortens, limiting the redistribution of internal energy needed for this process, particularly at larger impact parameters. As a result, CO2 production is increasingly restricted to smaller impact parameters, leading to a decrease in bmax. For NH2- and CH2-abstraction, the reaction follows a single H-abstraction step, where success depends solely on effective approach geometry. The trends observed in the ICSs further reflect these differences. At low collision energies, COOH abstraction dominates due to prolonged interaction. As energy increases, the ICS for the complex COOH abstraction decreases, while NH2- and CH2-H-abstraction become more prominent. At high collision energies, H abstraction at the NH2- and CH2-groups dominates, with the CH2 pathway being significantly more favored due to its much higher thermodynamic driving force compared to NH2-H-abstraction.

Fig. 3: Site-specific opacity functions for the OH + glycine(Ip) system.
figure 3

Reaction probabilities as a function of the impact parameter (b) for the OH + Ip glycine conformer H-abstraction reaction at collision energies of 1, 5, 10, 15, and 20 kcal/mol. Numerical values are given in Supplementary Data 2.

Full size image

Figure 4 illustrates the initial attack angle (α) distributions of the reactants at collision energies of 10, 15, and 20 kcal/mol, providing insight into how the orientation of reactants influences reactivity and the angular preferences of reactive encounters. The left column presents the distribution for the OH radical, while the right column shows the corresponding distribution for the Ip conformer of the glycine molecule. Each row represents a different reaction pathway, distinguishing H abstraction from the NH2, CH2, and COOH groups, with the latter ultimately leading to the formation of H2O, CO2, and CH2NH2. For the OH radical, a cos(α) value of −1 signifies an approach where the O atom is directed toward the glycine molecule, whereas a cos(α) value of 1 corresponds to an approach from the H-side. The attack angle distributions remain largely isotropic across all three reaction pathways, indicating no strong preference for a specific orientation. At 10 kcal/mol, where NH2-H-abstraction exhibits low reactivity, the peak observed around cos(α) ≈ −0.2 is unlikely to reflect a mechanistic trend. In COOH-H-abstraction, a slight tendency for H-side approaches appears at the lowest collision energy, though statistical fluctuations may influence this observation. For the glycine molecule, a cos(α) value of −1 indicates an approach from the amino-side parallel to the C−C bond, while a value of 1 corresponds to an orientation where the carboxyl group faces the OH radical. When H abstraction occurs at the NH2 group, higher collision energies (15 and 20 kcal/mol) disfavor amino-side approaches parallel to the C−C bond. The lone pair on the amino-N may facilitate H-bonding with the H of the OH radical, potentially stabilizing approach geometries that lead to reactivity. For H abstraction at the CH2 group, glycine preferentially approaches with the H2NCH2 part directed toward the OH radical across all collision energies, rather than from the carboxyl-side. The spatial arrangement of the CH2-H atoms in this configuration allows for more efficient access by the OH radical, increasing the probability of successful abstraction. Additionally, H-bonding between the NH2-N and the H of the OH may further contribute to the stability of this orientation, increasing the likelihood of reactivity. For COOH-H-abstraction, glycine predominantly approaches with the carboxyl group facing the OH radical. The O atoms of the carboxyl group can engage in H-bonding with the H of the OH radical, which may enhance attraction and promote an orientation favorable for abstraction.

Fig. 4: Orientation-dependent reactivity of the OH + glycine(Ip) system.
figure 4

Normalized initial attack angle distributions of the reactant OH (left-side) and Ip glycine conformer (right-side) for the OH + glycine H-abstraction reaction at collision energies of 10, 15, and 20 kcal/mol. Each row represents a different product channel. Numerical values are given in Supplementary Data 2.

Full size image

Surprisingly, the initial attack angle distributions of glycine in COOH-H-abstraction reactions do not always follow the expected pattern. Instead of predominantly approaching the OH radical from the carboxyl side, certain conformers—specifically IIn and VIIp—exhibit a greater tendency to interact from the amino side (see Supplementary Fig. 4). Reaction dynamics simulations reveal that, despite this unexpected orientation, H abstraction still occurs at the carboxyl group. However, the process follows an indirect route: a hydrogen first migrates from the carboxyl group to the amino group, forming a biradical intermediate (H3N–CH2–COO), from which the OH radical subsequently abstracts a hydrogen. Figure 5 presents selected trajectory snapshots capturing the formation of the biradical intermediate and the final abstraction step in reactions involving the IIn and VIIp glycine conformers. An important question arises: what drives the formation of the biradical intermediate? The molecular structure of the IIn and VIIp glycine conformers (see Fig. 2) provides the key, as both adopt a geometry where the carboxyl-H is positioned near the amino-N, facilitating internal hydrogen transfer and creating favorable conditions for the emergence of the biradical pathway. Animations reveal that (22, 33, 14, 10, 7)/(26, 35, 18, 12, 9) % of the OH + Gly(IIn/VIIp) trajectories forming H2O + CO2 + CH2NH2 products proceed via a biradical structure at collision energies of (1, 5, 10, 15, 20) kcal/mol. However, as Supplementary Fig. 5 shows the fitted PES substantially underestimates the ab initio energies at the intramolecular hydrogen transfer region, indicating that the significance of the biradical pathways may be overestimated and future work is warranted to confirm this novel mechanism.

Fig. 5: Intramolecular H-transfer pathways for the OH + glycine → H2O + CO2 + CH2NH2 reaction.
figure 5

Snapshots of representative trajectories for the H-abstraction reaction of the IIn and VIIp glycine conformers with OH attacking the amino site, occurring via intramolecular H transfer and resulting in H2O, CO2, and CH2NH2 products. Cartesian coordinates for each frame are given in Supplementary Data 3.

Full size image

We have also investigated the possible indirect pathways of the H2NCHCOOH and HNCH2COOH product formations. On one hand, trajectory animations show that the former products are always obtained through direct H abstraction from the CH2 group. We have not found any H transfer pathways from CH2 to the dehydrogenated amino or carboxyl group presumably due to steric reasons. On the other hand, animations reveal that in the case of the IIIn and Vn glycine conformers the HNCH2COOH products can be obtained via an indirect, low-probability pathway, where OH abstracts a hydrogen atom from the COOH group while an intramolecular H transfer occurs from NH2 to COO as shown in Fig. 6. H transfer from NH2 (or COOH, which would lead to CO2 formation) to the dehydrogenated CH2 group is not found, as expected, because of both thermodynamic and steric reasons.

Fig. 6: Intramolecular H-transfer pathways for the OH + glycine → H2O + HNCH2COOH reaction.
figure 6

Snapshots of representative trajectories for the H-abstraction reaction of the IIIn and Vn glycine conformers with OH attacking the carboxyl site, occurring via intramolecular H transfer and resulting in H2O + HNCH2COOH products. Cartesian coordinates for each frame are given in Supplementary Data 3.

Full size image

Figure 7 illustrates the probability of forming different product conformers during NH2- and CH2-H-abstraction reactions, depending on the initial reactant conformers. The probabilities of product conformer formation during both NH2-H-abstraction and CH2-H-abstraction from different glycine conformers reveal a three-step pattern, where reactant conformers pair up based on their nearly identical likelihood of producing the same product conformer. Each pair shares the same carboxyl group spatial arrangement, emphasizing its key role in governing reaction dynamics. The observed pattern also reflects the sequential structural changes the carboxyl group undergoes during H abstraction. The first and most probable step, progressing from top to bottom, involves a rearrangement of the carboxyl-H atom, followed by a 180° rotation of the entire carboxyl group around the C–C axis, while the final step captures the simultaneous occurrence of both transformations. The magnitude of each step carries significant information, offering insights into the strength of intramolecular stabilizing forces. The observed differences indicate which structural arrangements are more favorable and which provide less stabilization for either the initial or final conformation. The magnitude of the first step, associated with the probability of rearranging the carboxyl-H spatial orientation, follows a consistent pattern in both NH2- and CH2-H-abstraction reactions. The observed trend indicates that the closed conformation of the carboxyl group, where the carboxyl-H points toward the carbonyl-O, exhibits greater stability compared to the open conformation. Product distribution data provide clear support for this conclusion, as both NH2- and CH2-H-abstraction reactions show that starting from a reactant conformer with a closed carboxyl group results in a relatively low probability of switching to an open conformation in the product. Conversely, reactant conformers with an open carboxyl group exhibit a relatively high probability of forming a product conformer in the closed conformation. The magnitude of the second step, which corresponds to product formation from reactants with the third-highest formation probability, reveals that a 180° rotation of the entire carboxyl group around the C–C axis is exceptionally rare in CH2-H-abstraction reactions, occurring in no more than 5-6% of cases. In contrast, NH2-H-abstraction follows the same trend for nearly all products, except for the formation of amino-product IV and V. The explanation lies in the nature of H abstraction and its impact on stabilizing H-bonds within the glycine molecule. During CH2-H-abstraction, hydrogen removal does not disrupt these stabilizing interactions, allowing them to maintain the integrity of the product structure. However, NH2-H-abstraction directly affects one of the H-bond centers. The departure of a hydrogen atom results in increased mobility for the remaining hydrogen, potentially introducing a repulsive interaction between the amino-N and either the carbonyl- or hydroxy-O. Analysis of the molecular interactions reveals that a stronger destabilizing effect arises when the hydroxyl-O is positioned opposite the amino-N, compared to when the carbonyl-O is in the same position. The increased repulsion in the former arrangement forces the amino-N to deviate significantly from the C-C-O plane, contributing to reduced structural stability. While the previous analysis focused on identifying patterns in the distribution of product conformers for a given product, a complementary approach provides further insights into the factors influencing selectivity in NH2- and CH2-H-abstraction reactions. Examining the probabilities of forming each possible product conformer from a specific reactant conformer allows for the identification of structural features that govern selectivity. A fundamental observation is that high selectivity in H-abstraction reactions is only possible when hydrogen abstraction occurs at the CH2 group. The Ip, IIIn, IVn, and Vn reactant conformers strongly indicate that the closed conformation of the carboxyl group serves as the driving force behind selectivity in CH2-H-abstraction. In contrast, reactions starting from IIn, VIp, VIIp, and VIIIn reactant conformers, where the glycine molecule adopts an open carboxyl conformation, exhibit lower selectivity, as the formation probabilities of two different product conformers remain nearly equal, preventing the dominance of a single product structure. The lack of high selectivity in NH2-H-abstraction can be attributed to the disruption of the hydrogen bond network that stabilizes the glycine molecule. H abstraction at the amino group increases the mobility of the remaining amino-H, facilitating interactions between the amino-N and either the hydroxyl-O or carbonyl-O, which introduce destabilizing effects and promote structural variations in the products.

Fig. 7: Conformer-specific product distributions of the OH + glycine system.
figure 7

Product distributions of the OH + glycine NH2- and CH2-H-abstraction reactions, where each histogram represents a different initial glycine conformer and the product conformer structures are displayed above the columns. The distributions are normailized over the various product conformers. Numerical values are given in Supplementary Data 2.

Full size image

Conclusions

A newly developed full-dimensional PES and QCT simulations are used to investigate the site- and conformer-specific dynamics of the OH + glycine H-abstraction reaction. The analysis focuses on three distinct pathways, targeting the amino (NH2), methylene (CH2), and carboxyl (COOH) groups. All three reaction channels are exothermic, with CH2-H-abstraction releasing the highest amount of energy, followed by NH2- and COOH-H-abstraction. Simulations reveal that COOH-H-abstraction rarely terminates at the initial dehydrogenation step, as the resulting glycine intermediate often fragments further, leading to CO2 and CH2NH2 formation. Integral cross-section analysis shows a strong dependence on collision energy. At low energies, COOH-H-abstraction dominates. Increasing the collision energy reduces the likelihood of COOH-H-abstraction, while NH2- and CH2-H-abstraction pathways become progressively more significant. At higher energies, CH2-H-abstraction prevails as the most favorable channel, primarily due to its thermodynamic advantage. The overall trends in ICS-s remain consistent across different initial glycine conformers, with only minor variations in absolute reactivity. Opacity functions provide insight into the relationship between reaction probability and impact parameter. The nearly constant maximum impact parameter (bmax) observed for NH2- and CH2-H-abstraction across all collision energies suggests that reaction efficiency is primarily governed by approach geometry. In contrast, COOH-H-abstraction displays a decreasing trend in bmax with increasing collision energy. Shorter interaction times at higher energies hinder the redistribution of internal energy necessary for CO2 elimination, leading to a progressive shift in reaction toward smaller impact parameters. Attack angle distributions provide valuable insight into how molecular orientation influences reactivity. The analysis reveals that reaction outcomes depend primarily on the initial structure of glycine rather than the approach direction of the OH radical, with the extent of this effect varying across different reactant conformers. The initial conformer-specific data also shed light on an alternative COOH-H-abstraction mechanism observed in specific cases, as simulations show that H abstraction in the case of IIn and VIIp glycine conformers is preceded by internal H transfer from the carboxyl group to the amino group, leading to the formation of a transient H3N–CH2–COO biradical intermediate. The structural arrangement of IIn and VIIp reactant conformers promotes H-migration by positioning the carboxyl-H close to the amino-N, creating favorable conditions for biradical formation. The animations further revealed that once H is abstracted from the carboxyl group, the resulting H2NCH2COO• intermediate can stabilize not only via fragmentation, but also through structural rearrangement as a minor pathway, involving an intramolecular H shift from the amino-N to the COO group, forming a species that corresponds to the product of the NH2-H-abstraction reaction. An examination of conformer-specific product distributions indicates that structural changes in the carboxyl group follow a probability-based sequence during H-abstraction reactions: an initial rearrangement of the carboxyl-H atom occurs most frequently, followed by a 180° rotation of the carboxyl group, while the simultaneous occurrence of both transformations is the least likely. The probability of large-scale structural changes, such as a complete 180° carboxyl rotation, remains low in CH2-H-abstraction but is more frequent in NH2-H-abstraction when hydrogen bonding is significantly altered. Furthermore, the closed conformation of the carboxyl group proves far more favorable than the open arrangement. Product distributions also show that CH2-H-abstraction emerges as the only pathway capable of selectivity, where specific reactant conformers predominantly yield a single product conformer. Selectivity is linked to the initial conformation of the carboxyl group, which dictates whether structural rearrangements occur during the reaction. In contrast, NH2-H-abstraction disrupts hydrogen bonding, increasing molecular flexibility and leading to a broader range of product conformers. Altogether, the presented findings not only deepen the mechanistic understanding of glycine oxidation but also provide a framework for exploring site- and conformer-specific dynamics of analogous complex processes in related biomolecular and atmospheric systems.

Methods

The computational study of the OH + glycine reaction involves the development of a full-dimensional PES and the application of QCT simulations to investigate reaction dynamics. The PES is constructed using high-level quantum chemical methods, including an explicitely-correlated version of the gold-standard coupled-cluster approach37, to provide an accurate description of the motion of the electrons. The development of the PES is carried out automatically following the generation of an initial set of geometries, which are derived from the previously-identified33,35 stationary points. This automated process, performed using the machine-learning-type Robosurfer program package38, ensures comprehensive sampling of relevant configurations, facilitating an accurate representation39 of the potential energy landscape. To unravel the reaction dynamics, the QCT method40 is employed. By treating nuclear motion classically while ensuring that the vibrational energies of the reactants correspond to their ZPE, QCT bridges the gap between accuracy and computational feasibility. By executing hundreds of thousands of simulations initiated from different glycine conformers, key reaction quantities are determined, including conformer-specific reaction probabilities, from which ICSs are derived, initial attack angle distributions of the reactants, and the relative abundances of different product conformers formed from various glycine structures. Further details on the applied methodologies and computational procedures can be found in the Supplementary Methods. Initial and final coordinates and velocities of the trajectories are given in Supplementary Data 1.

Online content

Additional computational details, T1-diagnostic values, fitting errors, relative energies of the glycine conformers, assessment of statistical uncertainties, pictorial representation of the initial structure generation, and comparison of ab initio and PES energies along a representative trajectory, as well as initial-conformer-specific-reaction probabilities, scattering angle distributions, and initial attack angle distributions for the OH + Gly(X) [X = Ip, IIn, IIIn, IVn, Vn, VIp, VIIp, VIIIn] reactions. Source data for the main figures including initial and final coordinates and velocities of the trajectories.