Oxalic acid dihydrate (OAD) (Reakhim, 98%) and α-glycine (Sigma-Aldrich, 98%) were used as purchased. γ-Glycine was obtained from α-glycine (Sigma, 98%) by exposure in wet ammoniac vapor during a long time as described in (Boldyreva et al., 2003). Anhydrous oxalic acid (OA) was obtained by heating the OAD on the hot plate at 145°С for 3 h. Analysis of the powder diffraction pattern of the heated powder showed the presence of the anhydrous α-OA as the main product with an additional small amount of β-OA and trace amounts of OAD (Figure 1). OAD might be difficult to remove by heating or it might be formed due to the sorption of moisture from the air when powder is cooled. However, those impurities did not affect mechanochemical experiment. Bis-glycinium oxalate (G₂O) was obtained by slow evaporation of a water solution of α-glycine and OAD in a molar ratio of 2:1 respectively.

The rate of a mechanochemical process at the initial stage is well-known to depend directly on the number and the area of contacts between particles (Boldyrev, 2006), therefore, before mixing and mechanical treatment the powders of all reagents were сlassified using Retsch sieves with cell diameters of 400 μm and 500 μm, so that the samples with particle sizes of only 400–500 μm were used for experiments. Samples were weighed using an OHAUS Analytical Plus analytical balance. Each mixture of reagents was mixed mechanically in a cylindrical container with one rotation axis for 1 minute to achieve a certain universal degree of mixing homogeneity before mechanical treatment. During described procedures no reaction was observed, which was controlled by Powder X-Ray Diffraction (PXRD).

Mechanical treatment was performed using a vibrational mill NARVA Vibrator DDR-GM9458 with a fixed processing frequency of 50 Hz (Figure 2), time of treatment was monitored by a stopwatch. Before treatment the insides of a steel milling jar, the surface of a grinding body (steel ball with a diameter of 9.5 mm and mass of 3.5 g) and tweezers, which were used for placing a ball into the jar were treated with ethanol and dried thoroughly in order to get rid of any contamination. Most experiments were carried out at a relative air humidity of 35 or 50%–60%. We did not observe any noticeable warming of the milling jar during the short treatment time that was needed to complete the transformations (not exceeding 3 min).

Phase composition after mechanical treatment was determined by powder X-ray diffraction using a multifunctional X-Ray diffractometer STOE STADI MP (Cu Kα1 radiation, λ = 1.54060Å, curved Ge (111) monochromator and a semiconductor linear detector Dectris MYTHEN 1K in a transmission geometry). The recording time of one diffraction pattern in the 2θ range from 5° to 60° was about 40 min. Full-profile analysis of diffraction patterns was carried out using the Rietveld method implemented in the GSAS-II program (Toby and Von Dreele, 2013). (see also Supplementary Information file)

To conduct the series of experiments with different reagents sample powders were selected each time as follows: an equimolar mixture of A (particle size 400–500 μm) and B (particle size 400–500 μm) were taken in such ratios as to keep the total mass equal to 40 mg. Approximate masses of reagents are listed in Table 1.

A series of independent experiments with multiple samples of an equimolar mixture of glycine with oxalic acid dihydrate was used for each kinetic curve. For each sample, the mechanical treatment was stopped at a different fixed time of continuous mechanochemical treatment (60 s, 90 s, 95 s, 96 s, 97 s, 100 s, 105 s, etc. in the α-glycine + OAD experiment 30 s, 40 s, 60 s, 140 s, 210 s in the G₂O + OAD experiment 35 s, 38 s, 40 s, 42 s, 50 s, 60 s, 65 s, 75 s, 90 s, 100 s, 110 s, 120 s, 135 s, 160 s, 180 s in the γ-glycine + OAD experiment, so that the treatment was interrupted and the jar opened only once. After mechanical treatment, the whole treated sample was placed quickly in the sample holder and analysed by PXRD (the time from the extraction of the sample and the start of a diffraction experiment not exceeding a few minutes). 40 mg of powder was a sufficient amount of powder to fill our sample holder. Trace amounts of powder were left mostly on the milling ball. For a comparison, we collected probes from different parts of the jar during the trial experiment, and they all had the same phase composition, proving that the sample was homogeneous.

The main difference is that no pronounced induction period was observed in previous experiments. Induction period may indicate that some “invisible” processes in the system occur, before a transformation becomes observable. These processes may be quite different from each other. In principle, an induction period may be related to mixing of reactants improving their contacts (Lapshin et al., 2021), but this option is excluded in our experiments, since the reactants were already mixed carefully before being put into the reaction jar. Another option is that the particle size is reduced during the induction period, so that the transformation rate increases sharply only after the particles of reactants become small enough (Lampronti et al., 2021). This option is in principle possible, but it could not be supported experimentally, since no broadening of X-ray diffraction patterns was observed for the samples subjected to mechanical treatment. The diffraction patterns of the mixture of initial reagents at treatment times of 25, 40, 60 and 95 s are completely identical. At the same time, a noticeable broadening of X-ray diffraction patterns could be observed only if the size of particles was reduced from original 400–500 nm to 100. There was no evidence of disorder, a decrease in coherent scattering regions, or an increase in background intensity during the induction period.

The existence of the induction period is in general a common feature of many solid-state transformations and is usually related with reaction starting at selected “active sites” and then self-accelerating due to positive feed-back (autocatalysis) (Boldyrev et al., 1979). With the development of techniques for time-resolved in situ (TRIS) monitoring of mechanochemical reactions, there is mounting evidence that typical mechanochemical reactions also often have induction periods (Hutchings et al., 2017; Linberg et al., 2023). It was supposed to result from a need to mechanically activate the material prior to transformation (Linberg et al., 2023), so that there is no real conceptual “gap” between “preliminary mechanical activation” and a “mechanochemical reaction” (Boldyreva, 2023; Boldyrev, 1996).

Mechanical treatment in a vibrational ball mill is not truly continuous, but consists of repeated mechanical impacts. Various processes may occur either during a mechanical impact, or between the impacts (Boldyreva, 2023; Boldyrev, 1972; Boldyrev, 2006). They may be related to the phenomena occurring in the solid state (accumulation of defects, fragmentation, seeding of a new phase - usually during an impact; secondary rearrangement of defects, polymorphic transformation as a result of seed growing–already after an impact), or in a fluid state (dissolution, recrystallization, chemical reaction – already after an impact). For inorganic compounds, activation during the induction period is known to occur usually through the formation of defects in the solid. The periodic dynamic mechanical stresses imposed by ball milling drive the solid away from its thermodynamic equilibrium state into a metastable (activated) state (Boldyreva, 2023; Boldyrev, 1972; Boldyrev, 2006). For organic compounds other options may be more likely (Boldyreva, 2013). In a study of a mechanochemical Knoevenagel condensation of vanillin and barbituric acid, the origin of the induction period was found to be a feedback cycle involving both chemical and mechanical factors and non-obvious and dynamic rheological changes in the reaction mixture. During the reaction the physical form of the reaction mixture changed from a powder to a cohesive rubber-like state, and this resulted in the observed reaction rate increase after the induction period (Hutchings et al., 2017). In a recent study (Linberg et al., 2023), a very long (hours) induction period of the formation of the final product of a mechanochemical synthesis of 1: 1 cocrystal of carbamazepine and isonicotinamide was shown to be caused by a delayed mechanically driven solid-state polymorphic transformation of one polymorph of this cocrystal into another. In our case, however, no polymorphic transitions were detected neither for the reactants, nor for the final (GO) and intermediate (G₂O) products.

In order to elucidate the nature of the processes that account for the induction period of a mechanochemical transformation, it is helpful to consider the effect of impact frequency on the induction period. A comparison of the results of mechanical treatment with different impact frequencies in this study and in the two previous ones (Michalchuk et al., 2017; Tumanov et al., 2011) (Table 2) may suggest that the processes that are important for the reaction between Gly and OAD take place between the impacts: induction period was observed, if the impact frequency was high (experiments in this work), i.e., there was not sufficient time between the impacts to let the processes that are important for the reaction trigger the reaction immediately.

One of the reactants, OAD, is a crystal hydrate. The mole amount of crystal water (2 mol of water per 1 mol of oxalic acid) is comparable with that of the oxalic acid and glycine that participate in the mechanochemical reaction to form an intermediate product with 2:1 glycine:oxalic acid ratio (G₂O) and the final product with the 1:1 glycine:oxalic acid ratio (GO). It is known that such a significant amount of crystal water may have a significant effect on a mechanochemical transformation, comparable with that of liquid water added to the system in a liquid-assisted grinding (LAG) process (Losev and Boldyreva, 2014). We have supposed that water may also play an important role in the “Gly + OAD” system, even though the products are not crystal hydrates. If a mechanical impact helps to release water from OAD, then the molecular salts in the “glycine-oxalic acid” system may form not in the solid state, but in water solution. When the system is mechanically treated at high frequencies (50 Hz), the fluid phase has no sufficient time to be formed in between two subsequent mechanical pulses. Therefore, the reaction is not observed, until a certain critical amount of water molecules has been released that can facilitate the interaction between glycine and oxalic acid in a fluid, or a fluidized state. Subsequently, with continued application of mechanical load, the reaction proceeds like an avalanche with an almost complete transition of the reagents into the final GO product of the reaction in 2–3 s.

We have carried out additional experiments to test this hypothesis.

According to the previously published data, the treatment of a mixture of α-glycine with anhydrous oxalic acid does not result in any reaction (Losev et al., 2013), supporting the hypothesis of the key influence of the presence of water molecules in the system on the course of the mechanochemical reaction. In this work, ball milling of a mixture of α-glycine with anhydrous oxalic acid for 3 min also did not give any products. The reaction could be observed, however, if the air was humid. It is worth mentioning, that also the induction period of the reaction of Gly with OAD shortened with increasing air humidity: in spring (50–60 rel.% humidity) it was shorter as compared to the experiments performed in winter (about 35 rel.% humidity) (Figure 6). Such seasonable irreproducibility of mechanochemical reactions related to the effect of air humidity on the process has been documented earlier for other systems (Tumanov et al., 2017).

In the next series of experiments, mixtures of glycine and OAD were preliminary mechanically treated for 25 s and then stored in a closed jar. After 2 h of storage the mixture contained only the starting reagents (glycine and OAD), as was shown by X-ray powder diffraction. A complete conversion of the reagents into the final product, GO, was observed, when the storage period was increased to a week. This series of experiments supports the hypothesis that the reaction between Gly and OAD is initiated through the release of hydration water molecules from the OAD structure. Water facilitates this reaction if present as a fluid–sorbed from the air or formed on dehydration of OAD. One can suppose that it is the release of the sufficient amount of water that occurs during the induction period of the mechanochemical synthesis between the mechanical impacts. The impacts themselves destabilize the crystal structure of OAD and facilitate the dehydration. Remarkably, mechanical treatment of OAD alone does not give anhydrous oxalic acid in any detectable amount (Losev and Boldyreva, 2014; Losev et al., 2013). Apparently, an equilibrium between the hydrated and the anhydrous forms is shifted very much to the hydrated one, unless there is a chemical substance other than water, that can react with the anhydrous oxalic acid, to form a salt. In this work this “other substance” is glycine. One can also suppose that the rate of the water release depends on the number of defects in the crystal structure of OAD, which are formed at the stage of preliminary mechanical activation during the induction period.

Another interesting feature of the mechanochemical transformations in the “Gly + OAD” system is that first G₂O is formed, which then disappears from the reaction mixture to be substituted completely by the final product, GO. This sequence of transformations was observed in all the experiments, also in this work. The details of kinetics, however, differed from experiment to experiment, depending on the protocol of mechanical treatment. We tried to rationalize these observations taking into account the possible role of water in the process.

Analysis of the product composition at different times of mechanical treatment in this work shows the formation of bis-glycinium oxalate (G₂O) in the system at a maximum concentration (about 12 wt.%) immediately after the end of the induction period (simultaneously with an increase in the content of the final GO product) and a subsequent drop in its content simultaneously with an increase in content of the final reaction product in the form of GO (Figure 4). This observation has a reasonable explanation from the point of view of the difference in aqueous solubility of α-glycine (∼0.25 g/mL in H₂O at 25°C) and oxalic acid (∼0.12 g/mL in H₂O at 25°C). The moment the transformation begins, the water released from crystalline OAD is more supersaturated with respect to glycine than with oxalic acid (the maximum possible volume of added water, obtained from the calculation of the complete decomposition of 25 mg of oxalic acid dihydrate, is 7.2 μL). A stoichiometric glycine predominance in a solution leads to partial crystallization and growth of G₂O despite the equimolar composition of the initial mixture of reagents. Subsequently, the G₂O formed initially reacts with the remaining oxalic acid molecules, presumably also with water solution involved, to form more GO.

The peak G₂O concentration observed in this work is about 5 mol.% (or 12 wt.%), which is at the level of the G₂O content obtained as a result of in situ experiments at the synchrotron radiation source at processing frequencies of 25–30 Hz (Michalchuk et al., 2017) and differs significantly from the G₂O concentration determined in an ex situ experiment on a model setup at a processing frequency of 1.67 Hz (Tumanov et al., 2011). It was suggested (Michalchuk et al., 2017), that such a difference may be associated with the predominant formation of G₂O in the end parts of the mill container, because the formation of G₂O is facilitated by the direct impact, as well as by the difficulties of its detection when positioning the synchrotron radiation beam in the area of the walls (Michalchuk et al., 2017). This model also provides an explanation for the significant G₂O content (about 40 mol%) at the initial processing times when using a model apparatus with a falling weight (Tumanov et al., 2011). The results obtained in the present work do not, however, support this hypothesis, since in the case of the NARVA Vibrator DDR-GM9458 mill the entire volume of the substance is subjected primarily to impacts due to the vertical direction of displacement of the jar and grinding body during operation. The results of this work can suggest that the initial content of G₂O is likely to be determined by the frequency of treatment, namely by the relaxation time of the system between impact pulses. A longer relaxation time promotes the release of more crystal hydrate water, which leads to the crystallization predominantly of G₂O due to the difference in solubility of glycine and OAD.

In order to gain more understanding of the kinetics of the accumulation and disappearance of G₂O and GO in the “Gly + OAD” system on mechanical treatment, additional experiments were carried out. During those experiments a mixture of OAD with G₂O was mechanically treated. G₂O was obtained by crystallization from an aqueous solution at a molar ratio of glycine to oxalic acid of 2:1. Powder diffraction patterns of several chosen experiments are shown on Figure 7.

Complete conversion of the reagents into the product (GO) took approximately the same total time as when OAD was milled in a mixture with α-Gly (Figure 4). However, the induction period was significantly shorter (about 35 s), as compared to 95 s when α-glycine was a starting reagent. The reaction progressed more smoothly, so that the amount of GO did not increase abruptly immediately after the induction period (Figure 8). One can relate a longer induction period of the mechanochemical reaction with α-Gly with a higher lattice energy of α-glycine as compared with G₂O, that makes it more difficult to break the intermolecular hydrogen bonds and dissolve α-glycine in water as compared with its salt. Lattice energies often correlate with melting temperatures and solubility values (Perlovich, 2017; Perlovich, 2020). A comparison of the melting temperatures of α-glycine and G₂O shows a difference of almost 100°C [250°C for α-glycine (Boldyreva et al., 2003; Perlovich et al., 2001) and 155°C for G₂O (Chitra and Choudhury, 2007; Fleck and Petrosyan, 2014; Uma Devi et al., 2008)], which indicates a higher energy of intermolecular interactions, mainly hydrogen bonds, stabilizing the structure of glycine as compared to its molecular salt.

A question that remained unanswered in the previous studies of the mechanochemical reactions in the “Gly + OAD” system was if solid GO, the final product, forms from solid G₂O as an intermediate phase resulting from a solid-state transformation (decomposition, elimination of half of glycine cations), or the two glycine salts with different stoichiometries are formed independently during parallel reactions, and G₂O content decreases since this salt, first formed faster, then dissolves and recrystallizes into a more stable GO form. The second option requires the presence of water. In fact, no reaction between G₂O and anhydrous oxalic acid could be observed in this work after a continuous mechanical treatment for 3 min. The G₂O alone is stable with respect to mechanical treatment and did not transform into GO when ball-milled for 3.5 min, either dry, or in the presence of an equimolar amount of distilled water.

Glycine is prone to polymorphism (Boldyreva, 2021). In previous publications different polymorphs were treated in different mechanical activators (Table 1), and this complicates a comparison of the results. Therefore, in this work we have carried out a comparative study of α– and γ-glycine in a NARVA Vibrator DDR-GM9458 vibrational mill under identical conditions. Kinetics of the transformations with γ-glycine (Figure 9) was qualitatively similar to that with α–glycine (Figure 4): it also was characterized by a pronounced induction period, but this induction period was shorter, than in the case of the reaction with α–glycine. One can relate this difference with the difference in the interactions of the two glycine polymorphs with water, in particular, with the kinetics of their dissolution (Parks et al., 2017). The choice of the starting polymorph seemed to be less important for the existence of the induction period, than the choice of the instrument for mechanical treatment, in particular, of the frequency of mechanical impacts.