Reactions of Benzylsilicon Pyridine-2-olate BnSi(pyO)₃ and Selected Electrophiles—PhCHO, CuCl, and AgOTos

Abstract

In dynamic equilibria, benzylsilicon pyridine-2-olate (BnSi(pyO)₃, L) and benzaldehyde react with addition of the Si(pyO) moieties to the PhHC=O carbonyl group and formation of compounds BnSi(–O–C(H,Ph)–N^(2-pyridone))_x(pyO)_3−x (L’, L’’, and L’’’, for x = 1, 2, 3, respectively). Addition of CuCl to a solution containing L, L’, L’’, and L’’’ results in the formation of BnSi(pyO)₃CuCl (LCuCl), shifting the equilibrium towards L with liberation of benzaldehyde. In THF as a solvent, the reaction of L in the presence of excess CuCl affords the complex LCuClCuCl. Upon dissolving in chloroform, it transforms into LCuCl with precipitation of CuCl. The solid state structure of LCuClCuCl features both the monomeric complex with CuClCuCl pattern and a dimer thereof with CuClCu(Cl)₂CuClCu pattern and a central Cu₂Cl₂ four-membered ring. This dimer of LCuClCuCl is the first crystallographically characterized representative of this Cu(I)-only Cu₄Cl₄ motif. The reaction of LCuCl and silver tosylate (AgOTos) in THF affords LCuOTos with precipitation of AgCl, whereas LAgOTos was obtained from L and AgOTos. In the crystal structure, LAgOTos features tetracoordinate Ag(I) in a distorted tetrahedral AgN₃O coordination sphere and a short Ag···Si trans-annular contact (3.3245(7) Å). ¹⁰⁹Ag NMR spectroscopy indicates a change in the coordination in solution, with δ ¹⁰⁹Ag = +551 and +419 ppm in the solid and in CDCl₃ solution, respectively. In combination, ²⁹Si NMR spectroscopy indicates changes in the Si coordination sphere, with δ ²⁹Si = −74.2 and −66.5 ppm in the solid and in CDCl₃ solution, respectively. Conversion of LAgOTos with tetraethylammonium chloride results in the precipitation of AgCl with release of L.

1. Introduction

Silicon compounds with penta- or hexacoordinate Si atom have attracted researchers’ interest for many decades [1,2,3,4,5]. More than merely seeing the silicon atom’s higher coordination number as nothing but a scientific curiosity, the exploration of substitution reactions at Si (usually along associative mechanisms [6]) and molecular dynamics associated therewith (e.g., Berry pseudorotation [7] about the pentacoordinate central atom) are closely bound to Si hypercoordination. Other aspects include the exploration of Si–C bond activation associated with the enhanced Si coordination number [8], and theoretical and experimental investigations, which address the general understanding of bonding situations about hypercoordinate Si atoms [9,10,11]. Moreover, transition metal atoms in the Si coordination sphere may also enhance the Si coordination number in the sense of a “simple Lewis base”, i.e., as a lone pair donor toward Si [12,13,14,15,16,17], and they may support Si–X (e.g., X = H) bond activation through agostic interactions, eventually leading to oxidative addition reactions as entrance into catalytic cycles [18,19]. In some cases, Si-hypercoordination itself provides sufficient activation for Si–C bond cleavage and hydrocarbyl transfer to an electrophilic reagent, such as allyl shift from Si to an Si-coordinated aldehyde [20,21], imine [22], quinoline derivative [23], or diazo group [24]. Ring opening of a silacyclobutane upon Si-hexacoordination is another example [25], and Si–H bond cleavage and H transfer to an adjacent electrophile [26,27] as well as a related Si–CN rearrangement may occur in a similar manner [28]. Even metal-free base-supported aryl transfer reactions with an aryl silane were reported [29], and UV irradiation may also initiate Si–C bond cleavage and alkyl, alkenyl, or aryl transfer to an Si-coordinated electrophile [30,31]. Various other reactions may, sometimes in addition to Si-hypercoordination, require the action of an electron rich d-element catalyst. Examples are d¹⁰-metal catalyzed cross-coupling reactions [32,33,34], allyl transfer reactions [35], and alkylation reactions [36,37]. Among the d¹⁰-metal catalysts used, Cu(I) compounds are very prominent in different roles such as redox-active species in radical mechanisms [33] and soft electrophiles for the formation of hydrocarbyl cuprates(I) as reactive intermediates [36].

In the context of our investigations of hetero-dinuclear complexes, we already reported Cu(I) and Cu(II) complexes of the types RSi(pyO)₃CuCl and RSi(pyO)₄CuCl (R = different hydrocarbyl groups), respectively, in which silicon and copper are held in close proximity by pyridine-2-olate as a buttressing moiety [38,39]. For these complexes, crystallographic analyses confirmed Si-Cu distances that are markedly shorter than the sum of the van der Waals radii, and computational analyses confirmed weak donation of electron density from Cu(I) toward Si in the complexes of the type RSi(pyO)₃CuCl. As these compounds combine Si-hypercoordination and spatial proximity of Cu(I) and Si as well as the presence of an Si–C bond, the current study aimed to explore the reactions of this system when exposed to selected electrophiles (Scheme 1). In the role of an organic electrophile, benzaldehyde was chosen as a representative carbonyl compound. As Si-bound allyl groups may easily undergo rearrangement reactions upon Si-hypercoordination, even in the absence of metal catalysts, the benzylsilicon compound BnSi(pyO)₃ (L) was chosen as the silane component of our study.

2. Results and Discussion

2.1. Reaction of Benzyltris(2-pyridyloxy)silane (Ligand L) and Benzaldehyde

At the beginning of this study, Si–C activation via Si-hypercoordination in the absence of any d-metal component was probed. As a set of NMR scale experiments, three samples of silane L (ca. 1 mmol of silane L, dissolved in 2 mL of CDCl₃, for details see Supplementary Materials Figures S1–S10) were mixed with benzaldehyde in molar ratio L:PhCHO = 1:1, 1:3, and 1:10. From the resultant clear solutions, ¹H, ¹³C, and ²⁹Si NMR spectra were recorded. The latter (Figure 1) clearly indicate the formation of new silicon compounds. Whereas the presence of some starting material L is reflected by a ²⁹Si NMR signal at −54.8 (cf. −54.7 reported for a CDCl₃ solution of this compound [39]), the systematic development of six new signals in the range between −50 and −53 ppm pointed at the formation of new compounds with BnSi(OR)₃ motif (R = variable substituent). With increasing benzaldehyde concentration, the systematic mutual gain vs. loss of signal intensities is speaking for the formation of three principle groups of compounds. The same pattern is reflected by the benzylic CH₂ ¹³C NMR signals (Figure 2). In the corresponding ¹³C NMR spectra (cf. Figure S7), the emerging of new signals in the range between 76 and 79 ppm indicates the formation of compounds with sp³ hybridized C atoms that carry electron withdrawing substituents. Moreover, upon dilution with CDCl₃, the signals at higher field (including the signal of the starting material L) gain intensity at the expense of intensity of the signals at lower field, thus pointing at the reversal of the reactions in dynamic equilibria. ²⁹Si NMR spectra recorded from solutions of L and benzaldehyde (molar ratio 1:3, same concentrations) in other solvents (acetonitrile-d3, THF-d8, toluene-d8) convey essentially the same picture of the shift range, number, and relative intensities of the signals of the new products (cf. Figures S13 and S14).

These NMR spectroscopic observations point at addition reactions of benzaldehyde and the Si-bound 2-pyridyloxy groups with formation of the groups L’, L’’, and L’’’ of stereoisomers of products of addition of one, two, and three equivalents of benzaldehyde, respectively (Scheme 2). The number of (groups of) diastereomers (1, 1, 3, 2 for L, L’, L’’, L’’’, respectively) equals the number of ²⁹Si NMR signals and ¹³C(CH₂) NMR signals per “group” in the corresponding sets of spectra. Moreover, ¹H-¹⁵N HMBC NMR spectra underline the formation of groups with N-alkylated pyridine N atoms. The spectrum of the starting material L (Figure S11) exhibits a ¹⁵N signal at δ 279.3 ppm and particularly pronounced correlation with position H⁵ of the pyridine-2-olate group. A spectrum of a solution of L and benzaldehyde in 1:3 molar ratio (Figure S12) exhibits signals in the same range (276–280 ppm), speaking for the presence of compounds with pyO groups in the product mixture, and a broad set of signals in the range 191–194 ppm, which exhibit correlations with a group of ¹H signals that can be assigned to H⁵ positions of the new moieties formed. Such an upfield shift in pyridine N atoms’ ¹⁵N NMR signals is typical upon N-functionalization (e.g., protonation [40] or N-alkylation [41]).

With respect to the molecular structures of silicon pyridine-2-olates, which generally feature Si–O bound 2-pyridyloxy groups with dangling pyridine N atoms [42,43], the pyridine N atoms of compound L represent rather soft Lewis bases. They are suitable to attack the aldehyde’s carbonyl C atom, and the α-heteromethanolate formed (a rather hard Lewis base) can be stabilized by the available hard Lewis acid, i.e., by forming a new O–Si bond. This principle of Lewis acid supported addition of a pyridine and an aldehyde has precedents in the literature. Scheme 3 shows two examples, the first of which (Scheme 3a) involves a 2-pyridyloxy moiety [44]. Even though the authors have drawn a zwitterionic form of the product, its structure can be related to those of the silicon compounds shown in Scheme 2, i.e., an N-alkylated 2-pyridone derivative. The crystal structure of this product reveals a shorter (1.470(4) Å, CH₂O–B) and a longer (1.574(4) Å, CO–B) B–O bond, and we interpret this as a result of contributions of the 2-pyridone form to the resonance structures (contributions of an O···B dative bond). The second example (Scheme 3b) illustrates this principle for a simple pyridine [45]. Anders et al. reported various further examples, including BF₃ and SbCl₅ as Lewis acids, and their studies involved the O-silylated compound [(Py)(p-Tol)HC-O-SiMe₃] [O₃SCF₃] [46].

2.2. Reaction of Benzyltris(2-pyridyloxy)silane (L), Benzaldehyde and CuCl

To explore whether compounds L’, L’’, or L’’’ can be used for complexation of CuCl or Cu-supported benzyl transfer to the carbonyl C atom of benzaldehyde, a solution of L (0.44 mmol) and benzaldehyde (0.46 mmol) in CDCl₃ (0.7 mL) was prepared, followed by the addition of CuCl (0.44 mmol) and, upon its dissolution, recording of ¹H, ¹³C, and ²⁹Si NMR spectra (cf. Figures S15 and S16). They indicated the formation of LCuCl, while benzaldehyde (previously in equilibrium with, in particular, L’ and L’’, cf. Figure 1a) coexisted in solution, and there was no sign of the formation of other Si-complexes with BnSiO₃ or SiO₄ coordination spheres. As excess CuCl remains insoluble in CDCl₃ (vide infra), repetition of this experiment with additional CuCl was carried out in THF-d8. The only slightly shifted ²⁹Si NMR signal (δ ²⁹Si −72.1 for LCuCl in CDCl₃ [39], −71.1 observed with the THF-d8 solution of this experiment), the general retention of the overall ¹H and ¹³C NMR signal patterns comparable to those of LCuCl and benzaldehyde, and the absence of new signals indicative of an OC(Ph,H,CH₂Ph) moiety showed that the Si-bound benzyl did not undergo a transfer reaction under these conditions (cf. Figures S17 and S18). Nonetheless, an additional experiment answered the question as to why two equivalents of CuCl could be dissolved in a THF solution of L.

When treated with excess CuCl in THF as a solvent, L proved capable of dissolving up to two equivalents of copper(I) chloride, and greater excess remained practically insoluble. From such a solution, the product LCuClCuCl crystallized (Figure 3,

Table A1. Crystallographic data from data collection and refinement for LCuClCuCl and [Cu(HpyO)₆](OTos)₂.

Parameter	LCuClCuCl	[Cu(HpyO)6](OTos)2
Formula	C22H19Cl2Cu2N3O3Si	C44H44CuN6O12S2
Mr	599.47	976.51
T(K)	180(2)	180(2)
λ(Å)	0.71073	0.71073
Crystal system	triclinic	monoclinic
Space group	$P\overline{1}$	P21/c
a(Å)	9.8896(3)	9.5568(4)
b(Å)	14.2302(4)	23.8922(13)
c(Å)	17.4646(4)	9.8426(4)
α(°)	89.113(2)	90
β(°)	88.283(2)	93.669(3)
γ(°)	76.202(2)	90
V(Å3)	2385.72(11)	2242.78(18)
Z	4	2
ρcalc(g·cm−1)	1.67	1.45
μMoKα (mm−1)	2.1	0.7
F(000)	1208	1014
θmax(°), Rint	26.0, 0.0348	25.0, 0.0922
Completeness	99.9%	99.9%
Reflns collected	35,772	27,835
Reflns unique	9385	3951
Restraints	12	10
Parameters	605	332
GoF	1.027	1.028
R1, wR2 [I > 2σ(I)]	0.0277, 0.0657	0.0493, 0.1209
R1, wR2 (all data)	0.0367, 0.0689	0.0777, 0.1321
Largest peak/hole (e·Å−3)	0.37, −0.35	0.34, −0.32

). The crystallographic asymmetric unit features two independent molecules of LCuClCuCl, one of which resembles one half of a dimer with an inversion symmetric CuClCu(Cl)₂CuClCu moiety in its predominant site occupancy, whereas the other one is a simple monomer with a chain-like CuClCuCl moiety (Figure 3a and Figure 3c, respectively). In both cases, the additional CuCl equivalent is formally bound to the Cl atom of the motif of the previously reported complex LCuCl with only marginal indication of Cl abstraction therefrom. Whereas in LCuCl [39] the Cu–Cl bond length is 2.334(1) Å, in LCuClCuCl the corresponding bond lengths Cu1–Cl1 (2.455(1) Å) and Cu3–Cl3 (2.437(1) Å) are ca. 0.1 Å longer. Interestingly, lengthening of the Cu–Cl bond is accompanied by significant shortening of the Si···Cu trans-annular distance (3.227(1) Å in LCuCl [39], Si1–Cu1 3.138(1) Å, and Si2–Cu3 3.130(1) Å in LCuClCuCl). Both effects, the Cu–Cl bond lengthening and the closer Si···Cu approach, result from a predominant formal shift of the Cu atom into the Si(pyO)₃Cu cage, which is indicated by the pronounced widening of the N-Cu-N angles (in LCuClCuCl Σ(NCuN) = 343.7 deg and 343.5 deg for molecules 1 and 2, respectively, vs. 336.2 deg in LCuCl [39]) and less widening of the sum of O-Si-O angles (in LCuClCuCl Σ(OSiO) = 337.4 deg and 335.3 deg for molecules 1 and 2, respectively, vs. 332.7 deg in LCuCl [39]). The pronounced change in the out-of-plane(N₃) location of the Cu atom (0.591(2) Å in LCuCl vs. 0.477(1) and 0.481(1) Å in LCuClCuCl) vs. the less pronounced change in the out-of-plane(O₃) location of the Si atom (0.504(2) Å in LCuCl vs. 0.455(1) and 0.477(1) Å in LCuClCuCl) conveys the same picture.

The Cu-Cl-Cu-Cl motif shown in Figure 3b,c has literature precedence of some other crystallographically characterized Cu(I) complexes: Cu2-A [47], Cu2-B [48], and Cu2-C [49] (Figure 4). Moreover, complexes with a related Cu^IIClCu^ICl motif were reported, e.g., compound Cu2-D [50].

Table 1. Selected bond lengths (Å) and angles (deg) of the monomer of LCuClCuCl (as shown in Figure 3c) and complexes Cu2-A [47], Cu2-B [48], Cu2-C [49], and Cu2-D [50]. The assignment of bonds and angles is shown in Figure 4.

	LCuClCuCl	Cu2-A	Cu2-B	Cu2-C	Cu2-D
a	2.4366(6)	2.63	2.54	2.43	2.81
b	2.1115(7)	2.10	2.13	2.11	2.13
c	2.0786(8)	2.09	2.09	2.10	2.11
α	109.00(3)	101.6	105.8	118.4	92.5
β	177.35(3)	175.7	160.5	177.7	176.8

lists the bond lengths and angles (indicated as a, b, c, α, β in Figure 4, right) of these compounds for comparison with the related features of monomeric LCuClCuCl (Figure 3c).

Compound Cu2-C, which also features a N₃Cu-Cl-Cu-Cl motif, exhibits very similar bond lengths a, b, c as the monomer of LCuClCuCl. In contrast, the N₂(alkene)Cu-Cl-Cu-Cl motifs of Cu2-A and -B exhibit markedly longer Cu–Cl bonds a (by 0.2 and 0.1 Å, respectively). In Cu^IIClCu^ICl complex Cu2-D, this distance is even longer, by 0.4 Å, and it can be interpreted as a weak end-on coordination of a linear [CuCl₂]⁻ anion in apical position of an otherwise square-planar Cu^IIN₃Cl coordination sphere. Regardless of the strength of binding of the formal anion moiety [CuCl₂]⁻ to Cu(I) or Cu(II) via bond a at angle α, its bond lengths b and c are very similar for all compounds listed in Table 1, and the angle β is close to linearity in most cases. The deviation of the latter in case of Cu2-B could be a result of a disorder associated with the relevant Cu site. This disorder is related to the disorder encountered with compound LCuClCuCl as shown in Figure 3a,b, where a pair of molecules of a monomeric form with rather linear terminal ClCuCl moieties and a dimeric form with ClCu(Cl)₂CuCl bridging moiety are found on the same site. Whereas in the case of LCuClCuCl, site occupancy is dominated by the dimer (ca. 89%), in the case of Cu2-B [48] it represents the minor part (ca. 21%). Nonetheless, it represents the only crystallographically characterized precedence of a Cu(I)-only CuClCu(Cl)₂CuClCu motif. Further examples were reported for mixed-valence complexes with Cu^IIClCu^I(Cl)₂Cu^IClCu^II pattern, i.e., Cu4-A [51], Cu4-B [52], and Cu4-C [53] (Figure 5). For comparison,

Table 2. Selected bond lengths (Å) and angles (deg) of the dimer of LCuClCuCl (as shown in Figure 3a), the dimeric disorder part in the crystal structure of Cu2-B [46], and the mixed-valence complexes Cu4-A [51], Cu4-B [52], and Cu4-C [53]. The assignment of bonds and angles is shown in Figure 5.

	LCuClCuCl	Cu2-B	Cu4-A	Cu4-B	Cu4-C
a	2.4545(6)	2.54	2.49	2.65	2.65
b	2.2080(8)	2.22	2.19	2.14	2.15
c	2.2739(10)	2.71	2.30	2.38	2.47
d	2.3691(16)	2.14	2.31	2.24	2.21
α	131.88(4)	98.7	110.2	111.8	108.5
γ	77.85(3)	77.8	75.1	79.0	79.4
δ	148.30(7)	111.7	137.9	164.9	164.5

lists the bond lengths and selected angles of this motif for compounds LCuClCuCl (the part shown in Figure 3a), Cu2-C (disorder part of minor occupancy), Cu4-A, Cu4-B, and Cu4-C.

Whereas in the crystal structure of LCuClCuCl for molecular site 1 (cf. Figure 3a,b) the dimer clearly dominates the disorder, the related arrangement in the structure of compound Cu2-B [48] is less favorable, concluded from the minor occupancy of this part in the structural disorder. The Cl···Cl separations of the central Cu₂Cl₂ moiety (3.61 Å in LCuClCuCl vs. 3.80 Å in Cu2-C) provide an answer as to why the positioning of copper atom Cu2 in a bridging position is less favorable in the latter and results in a Cu₂Cl₂ motif with a short Cu–Cl bond (2.14 Å) and a markedly longer Cu···Cl contact (2.71 Å). In contrast, in LCuClCuCl and in the mixed-valence copper complexes Cu4-A, -B, -C, the lengths of the Cu–Cl bonds in the central four-membered cycle are similar to one another (in Cu4-A) or deviate to a minor extent (by up to 0.26 Å in Cu4-C). Interestingly, in the bond length patterns of the Cu-Cl-Cu(Cl)₂Cu-Cl-Cu moiety, compounds LCuClCuCl and mixed-valence Cu^IICu^I-complex Cu4-A exhibit the greatest similarity, irrespective of the different oxidation number of the terminal Cu atoms. One common feature of all complexes listed in Table 2 is the similarly sharp Cu-Cl-Cu angle γ of the four-membered cycle, which spans the range 75.1–79.4 deg. The Cu-Cl-Cu bond angle α at the two-coordinate Cl atom and the torsion angle δ, however, span very broad ranges. They represent very flexible features of the Cu-Cl-Cu(Cl)₂Cu-Cl-Cu motif that can adopt to optimize packing interactions for crystallization.

2.3. Syntheses and Molecular Structures of the Tosylates LCuOTos and LAgOTos

As the addition of a second equivalent of CuCl to LCuCl (i.e., the formation of LCuClCuCl) points at a potential way of Si–C activation (Cu–Cl bond lengthening, closer approach of Cu trans to the Si–C bond), the next experiment aimed at replacing chloride by a hard anion X⁻ with formation of a more ionic Cu–X bond. For this purpose, LCuCl was treated with one equivalent of silver p-toluenesulfonate (AgOTos) in THF for replacing Cl by OTos with formation of LCuOTos and AgCl (Scheme 4). In this context, we also probed the accessibility of the corresponding silver complex LAgOTos and its conversion into the corresponding chloro complex “LAgCl”. Whereas the tosylato complexes LMOTos (M = Cu, Ag) were accessible along the routes shown in Scheme 4, treatment of a CDCl₃ solution of LAgOTos with one equivalent of tetraethylammonium chloride led to the precipitation of AgCl with liberation of silane L, as confirmed via NMR spectroscopy of the resultant solution. The latter experiment shows that the coexistence of L and AgCl is thermodynamically more favorable than the formation of a complex “LAgCl” under these conditions and that the initial complexation of the Ag atom in LAgOTos does not pose a sufficient kinetic barrier to prevent the release of AgCl under these conditions.

The new products LCuOTos and LAgOTos crystallized from THF solution upon slow addition of pentane (diffusion via gas phase). Whereas the silver complex LAgOTos crystallized within few days, the first crystallization experiment of LCuOTos afforded an oily product, which eventually crystallized over the course of some weeks and afforded crystals of two different modifications of LCuOTos in combination with a few crystals of the decomposition product [Cu(HpyO)₆](OTos)₂ (cf. Appendix A and Table A1) that had formed in the interim. Compounds LCuOTos and LAgOTos were characterized by single-crystal X-ray diffraction analyses, which confirmed the expected molecular structures (Figure 6,

Table A2. Crystallographic data from data collection and refinement for LCuOTos modification 1, LCuOTos modification 2, and LAgOTos.

Parameter	LCuOTos Mod1 1	LCuOTos Mod2 1	LAgOTos
Formula	C29H26CuN3O6SSi	C29H26CuN3O6SSi	C29H26AgN3O6SSi
Mr	636.22	636.22	680.55
T(K)	180(2)	180(2)	180(2)
λ(Å)	0.71073	0.71073	0.71073
Crystal system	monoclinic	monoclinic	monoclinic
Space group	P21/n	P21/n	P21/n
a(Å)	10.6762(5)	10.4695(6)	10.6128(2)
b(Å)	9.1920(5)	28.7746(14)	28.4084(4)
c(Å)	28.9364(18)	10.5738(7)	10.6043(2)
β(°)	91.449(5)	117.852(4)	116.952(2)
V(Å3)	2838.8(3)	2816.4(3)	2849.87(10)
Z	4	4	4
ρcalc(g·cm−1)	1.49	1.50	1.59
μMoKα (mm−1)	0.9	0.9	0.9
F(000)	1312	1312	1384
θmax(°), Rint	26.0, 0.0907	25.0, 0.0793	28.0, 0.0354
Completeness	99.9%	99.8%	100%
Reflns collected	25,592	19,178	68,380
Reflns unique	5585	4958	6858
Restraints	45	0	13
Parameters	400	371	394
GoF	0.995	1.027	1.064
R1, wR2 [I > 2σ(I)]	0.0410, 0.0980	0.0615, 0.1577	0.0225, 0.0579
R1, wR2 (all data)	0.0813, 0.1083	0.1004, 0.1777	0.0263, 0.0594
Largest peak/hole (e·Å−3)	0.46, −0.44	1.80, −0.41 2	0.42, −0.36

¹ The crystals were obtained from an initial synthesis batch, which was treated as outlined in the synthesis part (product solution connected to a Schlenk tube with n-pentane), and it took some weeks for spontaneous crystallization to commence. One well-shaped isolated crystal formed in an isolated position in the upper parts of the solution, and several multi-crystalline lumps formed in various parts of the product container. The former was used for the first attempt of single-crystal X-ray diffraction analysis, and it proved to be one modification of the target product (LCuOTos mod1). From the latter, small pieces were extracted, some of which were single-crystalline, and their single-crystal X-ray diffraction analysis confirmed their identity as another modification of the target product (LCuOTos mod2). As the latter did not suffer any obvious disorders, has a higher density, and formed multiple seed crystals, we interpret modification 2 as the more stable modification. ² The high residual electron density peaks are, most likely, the result of a high density of stacking faults in the crystal. The crystal growth (multi-crystalline lumps), streaking in the diffraction pattern, and the arrangement of residual electron density peaks in “ghost molecules”-like manner are in support of problems related to faulty crystal growth. The rather small size and the quite uniform dimensions of the crystal (0.20 × 0.19 × 0.17 mm³) rule out absorption problems as the major cause. (A test refinement of the S, Cu, Si sites and their “ghost molecules” counterparts with a site occupancy variable indicated a total “site occupancy” of 6% for the latter).

).

From compound LCuOTos, we obtained a major crystalline product, which consisted of multi-crystalline lumps, and the batch contained a single crystal in an isolated position. X-ray diffraction analyses confirmed that they belong to two different modifications of this compound. Whereas the modification represented by the isolated single crystal (modification 1, Figure 6a) has a slightly greater unit cell volume (by 0.8%) and exhibits some disorder of the tosylate group, the bulk product (modification 2, Figure 6b) formed small multi-crystals of poor diffraction power, which suffer from effects of stacking faults, but they exhibit a more compact unit cell and no disorder (cf. Table A2). In principle, the molecular shapes resemble the motif of LCuCl, i.e., silane L binding to Cu in a tripodal tridentate fashion, in combination with the Cu-bound anion (in this case tosylate) furnishing an almost tetrahedral Cu coordination sphere (cf. Figure 7a,b). The distortion of the latter is mainly determined by the formal approach of Cu toward the Si atom, which results in Cu···Si distances of 3.1177(9) and 3.0781(15) Å in modifications 1 and 2, respectively, associated with a widening of the three N-Cu-N angles (Σ(NCuN) = 343.4 and 345.9 deg, respectively). The out-of-plane(N₃) location of the Cu atom (0.484(2) Å in modification 1 and 0.438(3) Å in modification 2) and the out-of-plane(O₃) location of the Si atom (0.471(2) Å in modification 1 and 0.450(3) in modification 2) again point at the formal motion of the Cu atom toward the Si(pyO)₃ cage as the predominant contribution to the shortening of the Si-Cu distance.

In the corresponding silver complex LAgOTos, the d¹⁰-metal coordination sphere is more distorted (Figure 7c). It exhibits two particularly small angles on opposing edges (O4-Ag1-N2 and N1-Ag1-N3), indicating a stretching of the tetrahedron in this direction. Moreover, the sum of N-Ag-N angles (327.9 deg) is markedly smaller than the corresponding values in the aforementioned copper complexes, which points at a displacement of the Ag atom out of this tripodal base. In fact, the out-of-plane(O₃) location of the Si atom in LAgOTos (0.495(2) Å) is similar to the corresponding value in LCuCl 0.504(2) Å), but the out-of-plane(N₃) location of the Ag atom (0.757(1) Å) is more pronounced than the corresponding displacement of the Cu atom in LCuCl (0.591(2) Å) and thus the cause of the slightly longer M-Si distance in LAgOTos (3.3245(7) Å). Even though this distance is ca. 0.25 Å longer than the Cu-Si distance in LCuOTos modification 2, it corresponds to a similarly close M-Si-approach with respect to the sum of the van der Waals radii (88% for LCuOTos modification 2, 87% for LAgOTos) [54].

The polarization of the Cu–Cl bond of LCuCl by additional CuCl (in LCuClCuCl) or its substitution by a more polar Cu–OTos bond cause average shortening of the Cu–N bonds (ΣCu–N = 6.16, 6.06 (both CuN₃ sites), 6.08, and 6.04 Å in LCuCl, LCuClCuCl, LCuOTos modification 1, and modification 2, respectively). The Si–O bond lengths, however, change only marginally (ΣSi–O = 4.89; 4.86 and 4.88; 4.88, 4.87, and 4.86 Å in LCuCl, LCuClCuCl, LCuOTos modification 1, modification 2, and LAgOTos, respectively). Thus, from the perspective of bond lengths and other molecular metric parameters, the Si coordination spheres of the herein reported complexes are hardly influenced by the changes in the group 11 metal or its coordination sphere. Of note, the Cu-Si and Ag-Si distances observed with the herein reported compounds are among the shortest contacts encountered with such compounds that feature three-atomic bridges between Si and the coinage metal atom. Examples of other crystallographically characterized compounds with short M···Si contacts (M = Cu, Ag) are shown in Figure 8. A shorter contact was observed with CuSi-A [55], whereas in phosphane bridged complexes CuSi-B [56] and AgSi-B [56] the corresponding atom separations are longer. We attribute the markedly short Cu···Si contact in CuSi-A to the combination of the rather rigid Cu-N-bound ligand and the low Cu-coordination number. Introduction of an additional mono-atomic bridge (like a bridging H atom, in CuSi-C [57]) or the use of di-atomic bridges (like in CuSi-D [58], AgSi-D [58] and CuSi-E [59]) forces the metal atom closer to the Si atom. In complexes with di-atomic bridges, rather rigid Cu-N-bound bridging ligands may also support shorter Cu···Si contacts, e.g., in CuSi-F [60] and CuSi-G [61]. Among the examples shown in Figure 8, compound CuSi-D [58] is particularly noteworthy as it exhibits a surprisingly short Cu-Si distance, which is the result of a special bonding interaction, i.e., electron pair donation from the Si–Si σ-bond toward the Cu⁺ cation.

2.4. NMR Spectroscopic Characterization of LCuClCuCl, LCuOTos, and LAgOTos

In their ¹H and ¹³C NMR spectra, the compounds LCuClCuCl (in THF-d8), LCuOTos (in CDCl₃), and LAgOTos (in CDCl₃) produce the expected sets of signals (cf. Figures S19, S20, S22–S24, S26–S28, respectively, in the Supplementary Materials), which are consistent with the patterns observed for LCuCl and related complexes of the type RSi(pyO)₃CuCl [39], i.e., the spectra exhibit one set of ¹H and ¹³C signals for the three chemically equivalent pyO moieties and the set of signals for the Si-bound benzyl group. In addition, the expected signals of the tosylate group in LMOTos (M = Cu,Ag) are present. Noteworthy for the ¹H and ¹³C spectra is the observation that only in the case of LAgOTos the signals associated with the pyO groups are markedly broadened in a way comparable to the broadening observed with the corresponding signals of LCuCl and related complexes of the type RSi(pyO)₃CuCl [39].

Within the series of the herein reported compounds, all of which basically feature a BnSi(O-py)₃···M (M = Cu, Ag) coordination sphere, ²⁹Si NMR spectroscopy is a helpful tool to elucidate the effects of the metal M in the Si coordination sphere in the individual complexes. In our previous study [39], we already reported for L and LCuCl that the presence of Cu in the Si coordination sphere causes an upfield shift in the ²⁹Si NMR signal of the ligand L (δ = −54.7 ppm in CDCl₃) to δ = −72.1 and −67.6 ppm for LCuCl in CDCl₃ and in the solid state, respectively. A computational analysis of ²⁹Si NMR shifts for a related ligand (PhSi(pyO)₃) and its LiCl complex (PhSi(pyO)₃LiCl) indicated that the mere coordination of the pyO nitrogen atoms to a cation (Li⁺) may cause such a marked upfield shift Δδ of −18 ppm. This determines the view point of the following interpretation. Comparison of some molecular features derived from crystal structure analyses of LCuCl, LCuClCuCl, and LCuOTos (vide supra) has revealed some shortening of the Cu···Si separations along this series, a formal approach of Cu toward Si, and an average shortening of the Cu–N bond lengths in combination with the development of a more polar bond (lengthening of the Cu–Cl bond or replacing it by a Cu–O bond) trans to Cu···Si. The latter should lower the Cu atom’s capability of acting as a lone pair donor toward Si. Nonetheless, similar or enhanced upfield shifts in the ²⁹Si NMR signals are also observed for LCuClCuCl (δ_iso = −71.4 and −75.5 ppm in the solid state, cf. Figure 9, and −71.1 ppm in THF-d8 solution) and LCuOTos modification 2 (δ_iso = −75.5 ppm in the solid state, δ = −74.4 ppm in CDCl₃ solution). Thus, we interpret the structural effects on the ²⁹Si NMR shifts in these compounds predominantly as the results of enhanced coordination of the pyO nitrogen atoms to the more positively polarized Cu atoms rather than the effects of enhanced Cu⟶Si lone pair donation. In addition to the more upfield shifted signals relative to that of LCuCl in the solid state, the ²⁹Si CP/MAS spectrum of LCuClCuCl bears another noteworthy feature. The two signals of the two crystallographically independent Si sites exhibit the same intensity (by integration) but different line width (cf. inset in Figure 9). As one of the two molecular sites is affected by a disorder (cf. Figure 3a,b), we assign this signal to the Si site labeled Si1 and the enhanced width of the signal to slightly different Si1 coordination spheres arising from the different disorder positions.

In contrast to LCuOTos, which exhibits similar ²⁹Si chemical shifts in the solid state and in CDCl₃ solution (Figure 10a,b), the corresponding silver complex LAgOTos is more responsive to this change in state (Figure 10c,d). In the ²⁹Si CP/MAS NMR spectrum, it gives rise to a signal at δ_iso = −74.2 ppm, quite similar to its Cu-analog. In CDCl₃ solution, however, the signal emerges at δ = −66.5 ppm. In terms of the interpretation of the other ²⁹Si NMR shifts, this downfield shift would point at weaker N–Ag coordination in solution, perhaps accompanied by an enhanced average Ag···Si separation. NMR data of other nuclei of LAgOTos (e.g., the aforementioned signal broadening observed in ¹H and ¹³C spectra, and the ¹⁰⁹Ag NMR data, vide infra) add support to this picture.

¹⁰⁹Ag NMR spectroscopy is a useful tool to gain information about the bonding situation of this metal in various compounds, because the spin ½ nucleus basically allows for acquisition of spectra, which in case of their light coinage metal (⁶³Cu) congeners are hampered by the features of the quadrupolar nucleus. Thus, for example, compounds such as silver halides were studied ¹⁰⁹Ag NMR spectroscopically in the solid state some decades ago [62], and in very recent investigations of silver(I) complexes of N-heterocyclic carbenes, ¹⁰⁹Ag NMR spectroscopy was shown to be a useful probe for interactions with (and Brønsted acidities of) various surfaces [63]. For compound LAgOTos, the ¹⁰⁹Ag NMR spectra (Figure 11), recorded from the solid and the CDCl₃ solution, provide some information about changes in the Ag coordination sphere upon dissolving. The ¹⁰⁹Ag NMR signal of solid LAgOTos (δ_iso = 551 ppm) is, in principle, well in accordance with tetracoordinate silver(I). This shift falls in the shift range of compounds such as silver sulfite (δ_iso = 466.1 and 409.5 ppm for the two crystallographically independent Ag sites, which feature AgO₄ and AgSO₃ coordination spheres) or tetramethylammonium dichloroargentate(I), which features chains of AgCl₄ tetrahedra (δ_iso = 649.5 ppm) [64]. Moreover, the absence of larger spinning side bands in the CP/MAS spectrum recorded at υ_rot = 6 kHz (corresponding to a distance of ±185 ppm from the isotropic shift signal) is speaking for a rather narrow span of the chemical shift tensor, also in accordance with the rather tetrahedral coordination sphere observed in the crystal structure of LAgOTos. In CDCl₃ solution, the ¹⁰⁹Ag NMR signal is shifted to higher field (δ = 419 ppm). Even though this chemical shift is still representative of tetracoordinate Ag, the noticeable shift difference to the solid state (a Δδ of −132 ppm) is speaking for changes in the coordination sphere. The observations of solid- vs. solution-state ²⁹Si and ¹⁰⁹Ag NMR spectroscopy, in combination with the findings of ¹H and ¹³C NMR spectra of CDCl₃ solutions of LAgOTos (broad signals associated with the atoms of the pyO moieties, cf. Figures S26–S28 in the Supplementary Materials), point at partial release of N–Ag coordination in solution, Ag-coordinated and dangling pyO arms of L being in exchange. In principle, Ag-O coordination by additional tosylate O atoms may be a reason for the retention of the Ag coordination number (An example of O₂Ag chelating coordination of tosylate within an AgN₂O₂ coordination sphere with two pyridine N atoms has been reported by Zhang et al. [65]).

3. Materials and Methods

3.1. General Considerations

BnSi(pyO)₃ (L), CuCl, and LCuCl were available from previous studies [39]. AgOTos (Fluka, Memmingen, Germany, >96%) was used as received without further purification. Benzaldehyde (Sigma-Aldrich, Steinheim, Germany) was distilled under argon atmosphere and stored in a Schlenk flask under dry argon prior to use. THF (VWR, Darmstadt, Germany, 99.7%) was stored over Na-wire. CDCl₃ (Armar Chemicals, Döttingen, Switzerland, 99.8%), CHCl₃ (Honeywell, Offenbach am Main, Germany, >99.5%), THF-d8 (Deutero, Kastellaun, Germany, 99.8%), acetonitrile-d3 (Deutero, Kastellaun, Germany, 99.8%), toluene-d8 (Armar Isotopes, Leipzig, Germany, 99.5%), and n-pentane (VWR, Darmstadt, Germany, 99.8%) were stored over activated molecular sieves (3 Å) for at least 7 days and used without further purification. All reactions were carried out under argon atmosphere utilizing standard Schlenk techniques. For syringe filtration, PTFE syringe filters of 15 mm diameter and 0.2 μm pore size (Roth, Karlsruhe, Germany) were used.

Solution NMR spectra (¹H, ¹³C, ¹⁵N, ²⁹Si) were recorded on a Bruker Nanobay 400 MHz spectrometer. ¹H, ¹³C, and ²⁹Si chemical shifts are reported relative to SiMe₄ (0 ppm) as internal reference, ¹⁵N chemical shifts are reported relative to ammonia. The solution state ¹⁰⁹Ag NMR measurement was carried out on a Bruker Biospin 500 MHz AV III instrument using a 5 mm BBFO probe (SF(¹⁰⁹Ag) = 23.27 MHz). ¹⁰⁹Ag chemical shifts are referenced to saturated AgNO₃ in D₂O (0 ppm) [66]. ²⁹Si CP/MAS NMR spectra were recorded on a Bruker Avance 400 WB spectrometer using 4 mm zirconia (ZrO₂) rotors and an MAS frequency of υ_rot = 5 kHz. The ²⁹Si chemical shift is reported relative to SiMe₄ (0 ppm) and was referenced externally for ²⁹Si to octakistrimethylsiloxyoctasilsesquioxane Q₈M₈ (most upfield signal of Q⁴ groups at δ_iso = −109 ppm). The ¹⁰⁹Ag CP/MAS NMR spectrum was measured at a Bruker Biospin 700 MHz Avance NEO instrument equipped with 3.2 mm low gamma DVT CP/MAS probe (SF(¹⁰⁹Ag) = 32.59 MHz). A thin walled ZrO₂ rotor was used, the applied contact time was 30 ms at a spinning speed of υ_rot = 6 kHz. The cross-polarization conditions were optimized using silver acetate and verified with silver tosylate [67,68].

For single-crystal X-ray diffraction analyses, crystals were selected under an inert oil and mounted on a glass capillary (which was coated with silicone grease). Diffraction data were collected on a Stoe IPDS-2/2T diffractometer (STOE, Darmstadt, Germany) using Mo Kα-radiation. Data integration and absorption correction were performed with the STOE software XArea (version 2.3) and XShape (version 2.25), respectively. The structures were solved by direct methods using SHELXT [69] and refined with the full-matrix least-squares methods of F² against all reflections with SHELXL-2019/3 [70,71,72]. All non-hydrogen atoms were anisotropically refined, hydrogen atoms were isotropically refined in idealized position (riding model). For details of data collection and refinement see Appendix B, Table A1 and Table A2. Graphics of molecular structures were generated with ORTEP-3 [73,74], POV-Ray 3.7 [75], and MERCURY [76]. CCDC 2512147 (LCuClCuCl), 2512148 (LCuOTos mod1), 2512149 ([Cu(HpyO)₆](OTos)₂), 2512150 (LCuOTos mod2), and 2512151 (LAgOTos) contain the supplementary crystal data for this article. These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via https://www.ccdc.cam.ac.uk/structures/ (accessed on 30 November 2025).

3.2. Syntheses and Characterization

• Compound LCuClCuCl

A Schlenk flask was charged with magnetic stirring bar, compound L (0.473 g, 1.18 mmol), and CuCl (0.293 g, 2.96 mmol), evacuated and set under argon atmosphere. After adding THF (3 mL), the mixture was stirred at room temperature for 1 h. The solution was separated from the remaining solid material via syringe with a syringe filter. As storage of the product solution at 5 °C for 3 d and at −24 °C for another 3 d did not afford any solid product, the solution was connected to a second Schlenk flask containing n-pentane (2 mL) for slow diffusion via gas phase. Crystals of the product, compound LCuClCuCl, formed within 24 h. Thereafter, the supernatant was removed by decantation, and the crystals were dried in vacuum. Yield: 0.402 g (0.67 mmol, 57%). ¹H NMR (THF-d8): δ (ppm) 8.87 (d 4.2 Hz, 3H, H⁶), 7.79 (m, 3H, H⁴), 7.44 (d 7.4 Hz, 2H Ph-o), 7.31 (m, 2H, Ph-m), 7.15 (t 7.5 Hz, 1H Ph-p), 7.10 (m, 3H, H⁵), 6.95 (d 8.2 Hz, 3H, H³), 2.83 (s, 2H, CH₂); ¹³C{¹H} NMR (THF-d8): δ (ppm) 158.7 (C²), 149.2 (C⁶), 141.9 (C⁴), 136.1 (Ph-i), 130.0, 129.4 (Ph-o/m), 126.3 (Ph-p), 120.9 (C⁵), 115.5 (C³), 22.0 (CH₂); ²⁹Si{¹H} NMR (THF-d8): δ (ppm) −71.1; (CP/MAS): δ_iso (ppm) −71.4, −75.5.

• Compound LCuOTos

Compound LCuCl (0.335 g, 0.67 mmol) was filled into a Schlenk flask with magnetic stirring bar under dry argon atmosphere. Upon addition of solid AgOTos (0.163 g, 0.67 mmol), the Schlenk flask was evacuated and set under argon atmosphere again, whereupon chloroform (1 mL) was added, and the mixture was stirred for 5 min at room temperature. During that time, the crystals of the starting materials dissolved, and a fine white precipitate formed. Thereafter, the solution was separated from the fine solid material via syringe with a syringe filter and transferred into another Schlenk tube, which was then charged with a seed crystal from an earlier batch (positioned slightly above the level of the solution) before it was connected to another Schlenk flask containing n-pentane (2 mL) for slow diffusion via gas phase. Crystals of the product, compound LCuOTos, formed within 72 h at room temperature. Thereafter, the supernatant was removed by decantation, and the crystals were washed two times with a mixture of CHCl₃ and n-pentane (2 × 0.2 mL of CHCl₃:n-pentane 2:1) and dried in vacuum. Yield: 0.092 g (0.15 mmol, 22%). ¹H NMR (CDCl₃): δ (ppm) 8.89 (d 4.2 Hz, 3H, H⁶), 8.35 (br., 2H, tosylate-o), 7.66 (m, 3H, H⁴), 7.2–7.4 (mm, 7H, Ph and tosylate-m), 7.04 (m, 3H, H⁵), 6.79 (d 8.1 Hz, 3H, H³), 2.72 (s, 2H, CH₂), 2.43 (s, 3H, CH₃); ¹³C{¹H} NMR (CDCl₃): δ (ppm) 157.3 (C²), 149.0 (C⁶), 141.6, 140.7 (tosylate), 140.0 (C⁴), 135.4 (Ph-i), 129.1 (tosylate CH), 128.9, 128.7 (Ph-o/m), 127.0 (tosylate CH), 125.7 (Ph-p), 120.1 (C⁵), 113.9 (C³), 22.1, 21.3 (CH₂, CH₃); ²⁹Si{¹H} NMR (CDCl₃): δ (ppm) −74.4; (CP/MAS): δ_iso (ppm) −75.5.

• Compound LAgOTos

Compound L (0.431 g, 1.07 mmol) was filled into a Schlenk flask with magnetic stirring bar under dry argon atmosphere, and solid AgOTos (0.301 g, 1.08 mmol) was added, whereupon the flask was evacuated and set under argon atmosphere again. After adding THF (2 mL), the mixture was stirred for 1 h at room temperature to afford an almost clear solution. The solution was separated from the turbidity via syringe with a syringe filter and transferred into a new Schlenk flask, which was then connected to another Schlenk flask containing n-pentane (2 mL) for slow diffusion via gas phase. The crystalline solid product, compound LAgOTos, formed within 24 h at room temperature. Thereafter, the supernatant was removed via decantation, and the crystals were washed with a mixture of THF and n-pentane (2 × 1.5 mL of THF–n-pentane 2:1) and dried in vacuum. Yield: 0.605 g (0.89 mmol, 83%). ¹H NMR (CDCl₃): δ (ppm) 8.36 (br., 3H, H⁶), 7.92 (d 8.1 Hz, 2H, tosylate-o), 7.63 (m, 3H, H⁴), 7.10–7.30 (mm, 7H, Ph and tosylate-m), 6.96 (br., 3H, H⁵), 6.72 (d 8.3 Hz, 3H, H³), 2.80 (s, 2H, CH₂), 2.33 (s, 3H, CH₃); ¹³C{¹H} NMR (CDCl₃): δ (ppm) 158.6 (C²), 149.1 (C⁶), 142.6 (tosylate), 141.1 (C⁴), 139.6 (tosylate), 134.4 (Ph-i), 128.8, 128.7 (2×) (Ph-o/m, tosylate CH), 126.2 (tosylate CH), 125.8 (Ph-p), 119.8 (C⁵), 114.3 (C³), 21.7, 21.3 (CH_2, CH₃); ²⁹Si{¹H} NMR (CDCl₃): δ (ppm) −66.5; (CP/MAS): δ_iso (ppm) −74.2; ¹⁰⁹Ag{¹H} NMR (CDCl₃): δ (ppm) 419.0; (CP/MAS): δ_iso (ppm) 550.5.

4. Conclusions

Under the conditions applied (moderately polar solvents such as chloroform and THF, room temperature), the Si–C bond of benzylsilane BnSi(pyO)₃ (L) resists cleavage and transfer to an electrophile (benzaldehyde), both in the absence and presence of CuCl. Instead, in the former case, in a dynamic equilibrium the aldehyde group undergoes addition of the Si-bound pyridine-2-olate with binding of the pyridine N-atom to the carbonyl C atom and formation of an Si–O bond to the alcoholate thus formed. In the latter case, the formation of complexes such as LCuCl (in chloroform) and LCuClCuCl (in THF and in the presence of excess CuCl) is preferred. Also, replacing chloride by tosylate (formation of complex LCuOTos upon treating LCuCl with AgOTos) did not lead to Si–C cleavage in the presence of benzaldehyde. Nonetheless, in LCuClCuCl and LCuOTos the Cu atoms are closer to Si than in the parent complex LCuCl, which may indicate at a potential way of activating the trans-disposed Si–C bond. In contrast to the copper(I) complexes, silver(I) is a less favored electrophile in the tripodal ligand L. LAgOTos was obtained from the reaction of L and AgOTos, but treatment with chloride (Et₄NCl) caused precipitation of AgCl with liberation of L. Moreover, solution NMR spectroscopic investigations of LCuOTos and LAgOTos showed that the latter is more prone to changing the metal coordination sphere in solution. In the solid complexes LCuOTos and LAgOTos, however, Cu(I) and Ag(I) approach the Si atom in a similar manner, i.e., establishing contacts which are markedly shorter than the sum of the van der Waals radii (88% and 87%, respectively). In addition to yielding a new complex with very short nonbonding Si···Ag contact, the structural features of LAgOTos motivate us to include silver(I) complexes in future studies of Si–C bond activation.