Introduction

Silver coordination in non-aqueous solutions received much interest due to a number of applications ranging from nanomaterial synthesis [1,2,3], metallic coatings depositions [4, 5], catalysis [6,7,8], gas separations [9, 10] and electrochemistry [11, 12].

Polyamine N-donor ligands have been shown to be good chelators for silver(I) and other transition or f-group cations, both in aqueous [13,14,15,16,17,18,19] and non-aqueous solvents [20,21,22,23,24,25,26]. In the latter works, the structure and donor properties of the ligand, metal charge, solvation/desolvation processes of the species were analyzed considering the whole thermodynamic parameters of complex formation (ΔG°, ΔH° and ΔS°).

The definition of metal speciation is of paramount importance in water for biomedical, technological and environmental applications [27,28,29,30,31,32]. However, the understanding of the thermodynamics of solvation and complex formation of metal ions in organic solvents is important for the development of liquid–liquid extraction processes where auxiliary organic complexing agents are often necessary to obtain high selectivity [33,34,35].

The N-donors affinity for silver(I) was previously investigated in dimethyl sulfoxide (DMSO), dimethylformamide (DMF) and propylene carbonate (PC) [21], and more recently, in acetonitrile (AN), where the thermodynamics of complex formation with some monoamines were obtained [24, 36, 37]. Less data are available about the coordination of sp2 hybridized nitrogen donors, like pyridines, in AN and the solvent influence on the species involved, despite the coordination chemistry of silver with polypyridyl ligands, in many other solvents, is very rich because of the flexible coordination sphere of this ion and its ability to form, both in solution and in solid state a variety of architectures [38,39,40,41,42].

Pyridine-containing ligands also constitute an interesting comparison with amines, as the nitrogen atom is formally tertiary and additional rigidity is provided by the aromatic structure. Due to structural rigidity and lower σ-donor ability than amines [21], a much weaker strength of interaction with metal ions was shown for several transition metal ions and polypyridines in DMSO [43].

In this work, we report the results of a study on the coordination of silver(I) by a series of pyridines in AN to investigate both the effect of steric constraints and of the electron donating/withdrawing properties of the substituents. The pyridines considered are reported in Scheme 1, together with their names and abbreviations used in this work. Data with some non-aromatic cyclic amines (Scheme 2), which are stronger σ-donors than pyridines and rigid at the same time, are also included for comparison. Thermodynamic data obtained for these ligands are compared with those available in water and DMSO which differ from AN for their dielectric constants (ε) and donor numbers (DN) [44], with a particular focus on the influence of their acid–base properties on the stability of silver(I) complex formation in the different solvents.

Scheme 1
scheme 1

Structure of the aromatic amines discussed in this work

Full size image
Scheme 2
scheme 2

Structure of the heterocyclic amines discussed in this work

Full size image

Experimental

Materials

Anhydrous silver perchlorate was obtained by drying AgClO4·H2O (Aldrich, Germany) at 323.15 K under vacuum, the ligands (Aldrich, Germany > 98%) were purified by fractional distillation. AN (Fluka, Germany > 99%) and the background salt (tetraethylammonium perchlorate, NEt4ClO4), used to maintain the required 0.1 mol dm–3 ionic strength, were purified and dried according to the described procedures [45].

Weighted amounts of the reagents were dissolved in anhydrous AN to prepare the metal and ligand solutions, with the ionic medium adjusted to 0.1 mol dm–3 with NEt4ClO4. All the solutions were prepared afresh before use in a MB-150 Braun glove box under atmosphere of dry nitrogen. The water content in the solutions, typically 3—7 ppm, was determined by a Metrohm 684 KF Coulometer.

Potentiometry

All measurements were conducted in a cell maintained at 298.15 ± 0.1 K inside a dry box to guarantee the absence of moisture during the measurements. The cell electromotive force (e.m.f.) was measured by means of an Amel 338 pHmeter equipped with a silver Metrohm 6.1248.010 electrode, as a working electrode, and a Metrohm 6.0718.000 reference silver electrode. The Nernstian response of the ion-selective electrode was checked in the 10–6 < [Ag+] < 10–1 mol dm–3 concentration range. Potentiometric experiments were performed by titrating silver perchlorate (2.0 < C°Ag+  < 15.0 mmol dm–3 in Vcell = 20 mL) with solutions of the ligands (50.0 < C°L < 200 mmol dm–3 in Vburette = 5 mL). Titrations were performed with at least three different initial silver(I) ion concentrations, and some of them were carried out in duplicate to verify the reproducibility of the system which was better than 0.4 mV.

The formation constants of the complexes were obtained by using the computer program Hyperquad [46].

Calorimetry

A Tronac model 87–558 precision calorimeter was employed to measure the heats of reaction, with the measure-cell modified in order to be assembled inside the dry box. The response of the calorimeter was checked by titration of tris (hydroxymethyl) aminomethane (THAM) with a standard solution of HCl in water. The experimental value of the standard enthalpy of neutralization of THAM was found to be ΔH° =  − 47.58 ± 0.20 kJ mol–1, in good agreement with the published value of − 47.53 ± 0.13 kJ mol–1 [47]. Calorimetric titrations were performed at 298.15 ± 0.1 K by adding known volumes of ligand (Vburette = 5 mL) to silver(I) solutions (Vcell = 20 mL) with concentrations higher than those used for potentiometry.

In the case of py, 3-cl-py and 4-me-morp, which formed the weakest complexes among those investigated, the concentration range of metal ion was also extended up to 40 mmol dm–3 (and also 80 mmol dm–3 for 3-cl-py) to obtain a more relevant heat signal in the direct titrations. To reach higher ligand-to-metal ratios, also “reverse” calorimetric titrations, were run for all the systems, where the metal ion solution was placed in the burette and the ligand solution in the cell. For the more critical systems, i.e., 3-cl-py, 4-me-morp and py, the experiments were designed to maximize the formation of ML (3-cl-py), or superior ML2 (4-me-morp) or ML3 species (py). For the reverse titrations, 100—280 mmol dm–3 solutions of the silver(I) ion (Vburette = 5 mL) were added to 40—140 mmol dm–3 solutions of the ligand (Vcell = 20 mL). The heats of dilution of the reactants, determined in separate runs, were negligible. The neat reaction heats were used as input data for the cEST program [48] for the calculation of the standard formation enthalpies and Hyss for the speciation calculation [49]. The simultaneous fitting of the stability constants and formation enthalpies of the data of direct and reverse titrations (see results) allowed to determine accurate values of the stability constants and enthalpy values, in agreement with potentiometric data.

Results and discussion

The fitting of the potentiometric and calorimetric data was done considering the following complex formation equilibrium:

$${\text{Ag}}^{ + } + j{\text{L}} \rightleftharpoons {\text{AgL}}_{{\text{j}}}^{ + }$$
(1)

where L corresponds to the ligands shown in Tables 1 and 2. The values of j in (1), considered in the speciation models, range from 1 to 3 for all the pyridines (except 3-cl-py which only forms ML species) and pipd, while j = 1, 2 for L = morp and 4-me-morp. The formation constants and the corresponding thermodynamic parameters (ΔGj°, ΔHj° and TΔSj°) are reported in Tables 1 and 2 for pyridines and cyclic amines respectively, together with data available in other solvents [18, 50].

Table 1 Overall stability constants and thermodynamic functions (kJ mol−1) for the complex formation between silver(I) and aromatic amines in AN at 298.15 K and I = 0.1 mol dm−3 as defined in reaction (1)
Full size table
Table 2 Overall stability constants and thermodynamic functions (kJ mol−1) of silver(I) complexes with heterocyclic amines in AN at 298.15 K and I = 0.1 mol dm−3; estimated three standard deviations in parentheses
Full size table

Potentiometric titration data (Δe.m.f. vs. CL/CAg+) are plotted in Fig. 1a-c for silver(I) with py, 4-me-py (representatives of substituted pyridines) and with the cyclic pipd amine.

Fig. 1
figure 1

Plot of the observed (symbol) and calculated (solid line) Δe.m.f. values obtained for silver(I) titrations in AN versus the total ligand-to-metal (CL/CAg+) ratio with: (a) py (C°Ag+  = 8.76 (open square) and 12.14 mmol dm–3 (open triangle)), (b) 4-me-py (C°Ag+  = 4.53 (open square) and 9.36 mmol dm–3 (open triangle)) and (c) pipd (C°Ag+  = 5.68 (open square) and 13.87 mmol dm–3 (open triangle))

Full size image

From the analysis of the titration data (Fig. 1a, b), it can be seen that py and 4-me-py does not show a sharp variation in e.m.f values when the CL/CAg+ ratio increases. This is an indication that complex(es) of low stability(es) are formed. All speciation models tested for potentiometric data in the experimental conditions with highest CL/CAg+ did not reveal the formation of AgL3 species (no improvement of the quality of fitting) both for the other mono- and di-substituted pyridines reported in Table 1. Therefore, only the values of logβj with j = 1, 2, were obtained by potentiometry. For the system silver(I)-3-cl-py, from the potentiometric titrations (Fig. S1) only the AgL species was detected, which reached a maximum of about 21% of presence in our experimental conditions (Fig. S2). Calorimetric experiments, run to obtain enthalpy values, were also designed to check the possible formation of a AgL3 species (“reverse” titrations, see Experimental) by exploring very high CL/CAg+ ratios. This approach allowed to confirm the presence of AgL3 species for all systems except 3-cl-py (Table 1, vide infra). In Fig. S3, it is possible to appreciate how, in the conditions of the potentiometric titrations, the formation of AgL3 species was too small to be detected.

The results of the calorimetric titrations are shown in Fig. 2a-c, in the form of Δhv (total reaction heat per mole of metal ion) versus CL/CAg+, for the silver(I)-py, -4-me-py and -pipd systems.

Fig. 2
figure 2

Plot of the observed (symbol) and calculated (solid line) total molar enthalpy changes, Δhv, as a function of CL/CAg+ for silver(I) titrations in AN. (a) py (C°Ag+  = 28.04 (filled circle) and 39.03 mmol dm–3 (filled triangle)), (b) 4-me-py (C°Ag+  = 31.34 (filled circle) and 38.71 mmol dm–3 (filled triangle)) and (c) pipd (C°Ag+  = 19.70 (filled circle) and 29.78 mmol dm–3 (filled triangle)). On the right, reverse titrations are shown: (a’) py (C°L = 64.23 (open circle) and 92.51 mmol dm–3 (open triangle)), (b’) 4-me-py (C°L = 60.03 (circle circle) and 85.47 mmol dm–3 (open triangle)) and (c’) pipd (C°L = 41.72 (open circle) and 58.73 mmol dm–3 (open triangle))

Full size image

The experimental data for the silver(I)-py and -4-me-py systems (Fig. 2a,b) agree with what expected on the basis of the potentiometric study. The experimental points are separated for different concentrations of initial silver(I) ion, according to the formation of species of relative low stability. The reverse titrations (Fig. 2a’ and b’) demonstrate that the separation persists even at higher CL/CAg+ ratios. All the experimental data obtained with direct and reverse titrations were analyzed to obtain logβ3 and ΔH°3 values. In this procedure, the logβj (j = 1, 2), obtained independently by the potentiometric data treatment, were maintained as fixed parameters, while the complexation enthalpies (ΔH °j , j = 1—3) and the logβ3 were adjusted to obtain the best fit of data (full lines in the plots). This approach was followed also for the other pyridine systems (except for 3-cl-py, for which only logβ1 and ΔH °1 were simultaneously fit) and confirmed the existence of three AgLj species. It should be underlined that calorimetric titrations for 3-cl-py reveal low heat exchange, and for this reason, reverse calorimetric titrations were also run in a wider range of concentrations (up to 140 mmol dm–3 of the ligand in cell) to enhance the calorimetric response (Fig. S4).

In the case of cyclic ligands, pipd (Fig. 1c) reached the highest Δe.m.f. values both with respect to pyridinic and other substituted cyclic- amines, morp and 4-me-morp, which indicates the formation of more stable complexes than with all pyridine ligands and with morp and 4-me-morp.

The full lines, calculated with the stability constants listed in Tables 1 and 2, are in good agreement with the experimental values.

For silver(I)-pipd system, the overlap of the curves up to CL/CAg+  = 2.0 indicate the presence of two complexes of similar stabilities (logK1 = 3.45, logK2 = 3.25) followed by the formation of a less stable AgL3 species (logK3 = 0.61, Table 2). In the latter case, a calorimetric data fitting procedure similar to that described above for the family of pyridines was followed to verify the formation of the AgL3 species for all the cyclic amines. Only AgL and AgL2 species were detected for morp and 4-me-morp. The excellent fit between calculated and experimental values in Fig. 2 confirms the mutual consistency between the potentiometric and calorimetric measurements.

An improvement of the goodness of fit was obtained when AgL3 species were included in the models employed in the calorimetric data treatment, in most of the systems investigated (Table 1 and 2). Previous studies carried out in aqueous solution and DMSO [50] showed that these ligands were able to form only the AgL and AgL2 species (Table 1 and 2), while the formation of the AgL3 species is confirmed in this work in AN thanks to the use of reverse calorimetric titrations, which allow to reach high CL/CAg+ ratios and thus a significant percent formation (relative to total Ag) of AgL3 (up to 10% for py, up to 13% and 21%, respectively, for the other mono- and di-substituted pyridines and up to 23% for pipd, Fig. S5). By using the same calorimetric technique, the AgL3 species were revealed also for other aliphatic monoamines recently investigated in AN [33].

The formation of all species is enthalpy stabilized in AN, while the entropy term is always negative, which is characteristic for the formation of metal–ligand covalent bonds and results from the relatively weak solvation of the ligands involved in complex formation and the absence of charge neutralization [20, 21, 51].

The logK1 values for the aromatic N-donors in AN (Table 1) are lower than those previously reported for primary or secondary monoamines [24]. For example, for n-propylamine (n-pr), logK1 was found to be 3.54 with an enthalpy change (ΔH °1 ) of − 27.1 kJ mol−1. Additionally, the logK1 values for these aromatic N-donors are also lower than those obtained for cyclic secondary pipd and morp. This decrease in logK1 values can be attributed to the reduced σ-donor ability exhibited by these aromatic N-donors. However, the logK1 values for the pyridines in Table 1 are closer to the stability observed for 4-me-morp (Table 2) and tertiary monoamines in general [21, 24]. For instance, the logK1 value for tripropylamine (tri-pr) is 1.66 [24] with an enthalpy change (ΔH °1 ) of − 23.2 kJ mol−1. This suggests that the silver(I) complexes with aromatic N-donors present stabilities similar to tertiary monoamines due to their limited σ-donor strength in solution, which is much higher for primary and secondary amines [42]

In addition, the logK1 for py is increasing in the considered solvents in the order: DMSO < AN < water. This trend obtained for py is opposite to that obtained for ligands containing primary or secondary amine groups and can be explained by the prevalence of the metal over ligand solvation in DMSO for those who are H-bond acceptors only [21, 43, 52]. Similarly, the formation constants (and the enthalpy values) for all the pyridinic ligands considered here, follow the trend expected on the basis of metal ion solvation (Table 1). The thermodynamic parameters for the metal ion transfer (ΔG °tr , ΔH °tr, TΔS °tr ) from AN to water and DMSO [53] are reported in Table 3, together with the dielectric constants (ε) and donor numbers (DN) of the solvents [44]. The ΔG °tr and ΔH °tr, data show that the solvation strength for silver(I) increases in the order water < AN < DMSO. The enthalpy of solvation of silver(I) is more negative in the aprotic solvents compared to the protic one like water (Table 3). This can be due to the intrinsically stronger Ag-solvent bond with DMSO and AN than water, as shown by gas phase experiments [54] and also to the different solvation structure of the metal ion. In fact Ag(I) is solvated by an average of 4 solvent molecules [55, 56] in DMSO and AN, while a quasi-linear (“2 + 2”) coordination has been shown in water [32].

Table 3 Dielectric constants (ε) and donor numbers (DN) of some organic solvents with the free energies, enthalpies and entropies of transfer (kJ mol−1) of silver(I) ion from AN to water and DMSO solvent (ΔG °tr(AN→solv) , ΔH °tr(AN→solv) and TΔS °tr(AN→solv), respectively) at 298.15 K
Full size table

As a result, the stability of metal-pyridines complexes follows a trend where the complexes are less stable in the organic solvents than in water due to the differing solvation energies of the silver(I) ion.

As pKa values are key parameters defining the acid–base behavior of the ligands in water [57], AN and DMSO [61], these values are documented in Table 4 and referred to the reaction LH+  + S ⇄ L + SH+ (L = ligand, S = solvent). The analysis of these data should be useful to investigate the general trend in stability constants and thermodynamic parameters, not only for py, but also for all the other mono- (4-me-py, 3-me-py and 2-me-py), di-substituted pyridines (2,4-dime-py, 2,3-dime-py and 2,2-dime-py) and cyclic (pipd, morp and 4-me-morp) amines.

Table 4 pKa data for amines in different media referred to reaction: LH+  + S ⇄ L + SH+
Full size table

In general, it is evident that LH+ are much weaker acids in AN than in water (and in DMSO) as the difference of the pKa values (ΔpKa) is systematically larger by about 7 (or 9) pH units compared to the corresponding values for water (or DMSO), mainly due to the lower dielectric constant and the weaker basicity of AN, which result in smaller free energy of solvation of LH+ (or higher solvation of L) in AN than in DMSO and water [61, 70].

DMSO is known for its ability to be a strong hydrogen bond acceptor [71]. In DMSO, primary amines act as stronger bases compared to water, while pipd is slightly weaker as a base in DMSO than in water (Table 4). Additionally, pyridine-containing ligands exhibit significantly weaker basicity in DMSO as compared to water [72, 73].

The values of logK1 versus the corresponding pKa for the ligands studied in this work in the three solvents (Table 1 and 2) are reported in Fig. 3. Also, the data (taken from ref [24]) relative to the aliphatic secondary monoamine dibutylamine (di-but) are inserted in Fig. 3, for comparison with the secondary cyclic pipd. For di-but data of logK1 and pKa are available in AN and DMSO (logK1 = 3.14, 2.66, respectively [24]; for pKa, see Table 4).

Fig. 3
figure 3

LogK1 as a function of pKa for the studied amines in the different solvents (AN, DMSO and water)

Full size image

In general, it can be observed that logK1 values increase with increasing in pKa, in all the solvents considered, in line with an increase in basicity of the ligands, with a little discontinuity on going from the family of pyridines to the amines in water and AN. However, if the pyridines and cyclic amines are considered separately, a nearly linear correlation between pKa and logK1 can be appreciated (see below, Fig. S6).

As far as the pyridines are considered, the decrease in pKa value in the aprotic AN is particularly evident going from py to 3-cl-py (ΔpKa AN py→3-cl-py = 2.97) and parallels the strong decrease already observed in water (ΔpKa water py→3-cl-py = 2.25), which was mainly ascribed to the inductive effect of chloride [67]. Also, the stability of silver(I)-3-Ccl-py complex is particularly low as compared to -py and certainly related to this inductive effect of the chlorine, which decreases the donor properties of the N atom. In agreement with that effect, also the obtained ΔH°1 value is rather low as compared to py.

On the contrary, when a methyl electron-donor group is inserted, the basicity increases, the pKa passing from 12.53 for py to 13.28 for 2-me-py in AN. In water a similar increase of 0.7 pKa units is observed. The inductive effect of the electro-donating group is more effective when present in α or γ position [74] and this is also reflected in the logβ1 value for 2-me-py and 4-me-py which are higher not only with respect to py but also to 3-me-py. The values of logK1 and ΔH°1 obtained in this work in AN, together with those for di-but [24], as a function of pKa are shown in Fig. 4.

Fig. 4
figure 4

Values of logK1 (circles) and ΔH°1 (empty circles) as a function of pKa in AN

Full size image

It can be clearly noted, that both ΔH°1 for pyridinic ligands (Table 1) and logK1 (Table 1) increase along the pKa, and this is true also for the cyclic amines (Table 2): an exception is found for 2,6-dime-py, which shows a slightly lower stability, likely due to steric hindrance of the two methyl groups adjacent to the N donor.

In general, within a given family of ligands, a good correlation between logK1 and pKa is obtained (pyridines, R2 = 0.968 and for cyclic amines R2 = 0.988), whereas the linearity is less evident when all the ligands in Fig. 4 are considered (R2 = 0.825, Fig. S6-7). The linear relationship is even better for the − ΔH °1 vs. pKa plot (R2 = 0.937), when considering all the ligands, both pyridines and cyclic amines, as can be seen from the interpolation and the relative equation reported in Figure S8. This result supports the strong correlation between the strength of Ag–N bond, mainly reflected in − ΔH °1 value, and the base σ-donor ability of the N atom.

Moreover, it can be observed that: (i) the cyclic secondary amine pipd shows stability and ΔH °1 very close to that of secondary linear ones (for di-but logK1 = 3.14, ΔH °1  =  − 27.9 kJ mol−1), thus indicating that there are no remarkable influences on complexation, due to the rigidity introduced by the cyclization of the ligand; (ii) the stability of silver(I)-morp complex is lower than that of pipd, due to a decreased favorable enthalpy term, owing to the electro-attractive properties of the oxygen atom in the cycle which decreases the donor ability of the N atom [75]; (iii) when the methyl group is present on N atom, in 4-me-morp, the stability of the complex is much decreased: the σ-donor properties of the tertiary nitrogen atom are lowered and the complex solvation energy can be reduced due to the increased size of the species (the larger complex size can provide a greater surface area for interactions with the surrounding solvent molecules, leading to a reduction in solvation energy in some cases) [52, 75]. This agrees with the lower stability and enthalpy of the complex formed and with a more unfavorable entropy effect with respect to morp.

The second stepwise stability constant for the formation of silver(I) with the secondary pipd and morp (Table 2) K2, are similar to K1 in AN. The complete set of thermodynamic functions in AN for the reactions concerning the second complexation step reveals that this stability trend results from a balance between the favorable enthalpy of reaction and unfavorable entropy. The stepwise enthalpy (− ΔH °K2  >  − ΔH °K1 ; that is − 32.5 vs − 25.5 for pipd and − 30.1 vs − 23.4 kJ mol−1 for morp) and entropy trends (− ΔS °K2  >  − ΔS °K1 , that is − 8.8 vs − 22.0 for pipd and − 7.9 vs − 18.0 kJ mol−1 for morp) can be interpreted as reflecting a greater desolvation occurring at the first step of complexation, which makes ΔH °K1 less negative than ΔH °K2 , in spite of the certainly greater metal–ligand interaction in the first complexation step. This is in line with the entropy values which, in the second step, are more negative than in the first step. This trend (K2 similar to K1 and − ΔH °K2  >  − ΔH °K1 ) is present also in DMSO for the aliphatic pipd and morp.

For the pyridinic ligands the trend in the values of ΔH °K1 and ΔS °K1 with respect to ΔH °K2 and ΔS °K2 is similar, but not so remarkable as for secondary amines pipd and morp. As a matter of fact, − ΔH °K1 values are lower, in the range 15.5—18.3 kJ mol−1 for the first step and − ΔH °K2 is in the range 15.7—18.3 kJ mol−1, whereas the decrease in ΔS °K2 are more significant with respect to ΔS °K1. Thus, the result of the balance between enthalpy and entropy terms always results in K2 < K1.

Finally, when the AgL3 species is formed, a low stability is found (Table 1 and 2). The enthalpy associated to the third formation step is close to that found for the other species, whereas the ΔS °K3 is much more negative. This agrees with the fact that a strong Ag–N bond is always formed, but unfavorable entropy changes occur, which almost completely compensate the complexation enthalpy.

Conclusions

The complex formation between silver(I) ion and a series of N-donors in AN is characterized by high exothermic enthalpy changes indicating formation of relatively strong Ag–N bonds, while the unfavorable entropy is diagnostic of a decrease in disorder of the system. Carefully designed calorimetric experiments allowed to obtain both stability constants and enthalpy terms and highlighted the presence of species formed at high metal-to-ligand ratios, i.e., AgL3 in this work.

The logK1 values (and the negative enthalpy values) for pyridine ligands is increasing in the order: DMSO < AN < water. This stands in contrast to the trend observed for ligands containing primary or secondary amine groups, which are H-bond solvated in DMSO and water and is explained by the prevalence of metal ion solvation on ligand solvation, in the case of pyridine ligands.

This study shows that logK1 values increase with increasing pKa values for all considered solvents, indicating that higher pKa values correspond to greater ligand basicity and in general more enthalpy stabilized complexes. The behavior of the pyridine family in AN, closely resembles that observed in water. Finally, pipd exhibits stability and ΔH°1 close to linear secondary amines, suggesting minimal impact on complexation from ligand cyclization.