J. Chem. SOC., Faraday Trans. 1, 1982, 78, 3629-3643 Solvation of Silver Ions in a Range of Binary Solvent Systems Radiation and Electron Spin Resonance Study? BY MARTYN C. R. SYMONS* AND COLIN K. ALESBURY Department of Chemistry, The University, Leicester LE1 7RH Received 2nd April, 1982 Exposure of dilute solutions of silver perchlorate in a range of pure and mixed solvents to 6oCo y-rays at 77 K gave Ago and Ag" centres with reduced silver hyperfine coupling indicating delocalisation onto solvent molecules. In certain cases this was confirmed by the detection of hyperfine coupling to ligand nuclei. Solvents used included water, methanol, cyanomethane, acetone, methyltetrahydrofuran, dimethyl- sulphoxide, dimethylformamide, ethylamine, triethylamine, pyridine, ammonia, benzene and toluene.Results for the Ago centres are interpreted in terms of primary species, which retain the solvation of the original Ag+ ions, and secondary centres, which have lost one or more of the primary solvent molecules of the cation, the limiting species being generally the unsolvated atoms. In some cases Ag:, Ag$+ and Ag:+ clusters were also detected either at 77 K or after annealing. A linear correlation between the isotropic hyperfine coupling to lo9Ag and the solvent donor number was followed by the majority of pure solvents, the most marked exceptions being pyridine and cyanomethane. Reasons for this trend and the exceptions are discussed. Other species with relatively low isotropic hyperfine coupling to lo9Ag were detected, and possible structures are suggested.We have recently shown that Ago and AgI1 centres formed from Ag+ ions in dilute solutions in cyanomethane both retain the four primary solvent molecules coordinated to Ag+ prior to irradiation, as evidenced by hyperfine coupling to 14N.l Using mixed (MeCN+ H,O) and (MeCN + MeOH) solvent systems,2 we went on to show that Ag+ ions are remarkably strongly bound to MeCN molecules, the tetrasolvate dominating to beyond the 50% mole-fraction range. That Ag+ ions are preferentially solvated by MeCN in (MeCN+H,O) systems has also been adduced by Cox et al.,3 and there is conductometric evidence that Ag+ coordinates to four MeCN molec~ies.~ Our conclusions, that in these radiolyses electron-capture by Ag(MeCN)a gives Ago(MeCN), and electron-loss gives Ag(MeCN)i+, have recently been q~estioned.~ In fact, the argument centres on the silver-atom species only.The most interesting result of exposure of solutions of AgClO, in MeCN to X-rays at 4 K was that the yield of Ago(MeCN), centres was reduced by a factor of ca. 1 /6, but that on annealing to 77 K this increased six-fold. The most probable reason for this is that at 4 K MeCN- or (MeCN); centres6 are formed, and that these transfer their electrons to Ag+ on annealing.5 However, because a silver-atom centre containing only one MeCN molecule was formed on photolysis, it was concluded that the Ag+ ions are only solvated by one MeCN molecule, the Ago(MeCN), centre being the primary centre, with Ago(MeCN), formed therefrom because of an atom solvation process. Our reasons for rejecting the conclusions of Li and Kevan will be outlined el~ewhere.~ Our major case rests on the fact that the compound Ag(MeCN), C10, gives only the centres Ago(MeCN), and AgII(MeCN), on irradiation.Herein we use our original assignments and treat the Ago(MeCN), centre as a partially desolvated unit, intermediate between Ago(MeCN), and unsolvated Ago. 7 Taken as Solvation Spectra, Part 70. 36293630 SOLVATION OF SILVER IONS EXPERIMENTAL Deuterated solvents, including [2H6]acetone (Me2CO), [2H6]dimethylsulphoxide (DMSO), [2H3]cyanomethane (MeCN), [zH,]NN-dimethylformamide (DMF) and [2H,]toluene, which were all supplied by Merck, Sharp & Dohme (Canada) Ltd, were of the highest grade available. They were used as supplied after drying and removing oxygen by freeze-thaw procedures.Solution handling, irradiation and e.s.r. spectroscopic methods were as described previ~usly.~ RESULTS AND DISCUSSION We consider first the range of silver centres formed in the pure solvents, before turning our attention to the mixed-solvent systems. No attention is given to the range of solvent radicals also detected in the g = 2 region of the spectra. PURE SOLVENTS In all cases studied, replacing the normal solvent with its perdeuterated form resulted only in line-narrowing. Most of the following discussion refers to perdeu- terated media. ACETONE As can be seen in fig. 1, the major silver centres detected at 77 K were Ago, Ag*I and Agl (data are given in tables 1-3). The Agi centre is characterised by a major triplet, the central features being partially concealed beneath intense features from solvent radicals.* The outer (MI = f 1) features appear as triplets.This is because TABLE 1.-Ago CENTRES DETECTED IN PURE- AND MIXED-SOLVENT SYSTEMS AT 77 K solvent species g,, 14N lo9Ag CD,CN (CD3)2C0 DMSO MeTHF nitromethane triethylamine 33% aq. NH, form amid e DMF toluene CD3CN + D,O CD,CN + (CD,),CO A B A B A B A B 1.997 2.001 2.003 2.003 2.003 2.002 2.001 1.998 2.001 2.001 2.001 1.995 1.996 1.999 1.998 2.000 2.000 2.001 2.003 2.002 2.000 +6 -532 -612 - 643 - 705 - 679 - 626 -601 - 578 - 636 - 597 - 661 -511 - 455 - 647 - 603 - 679 +6 -587 - 665 +7 -679 -721 - 555 a G = 10-4 T.M. C. R. SYMONS A N D C. K. ALESBURY 363 1 1 32506 A (a ) FIG. I b 1 3250G A - 2 SOG , I * ' A I Gain x 10 FIG.1.-First-derivative X-band e.s.r. spectra for AgC10, in (CD,),CO after exposure to 6oCo prays at 77 K : (a) before annealing, showing features assigned to Ago and Ag" centres, and (b) after annealing to 120 K, showing additional features assigned to A&+ centres. Central features for solvent radicals are not shown.3632 SOLVATION OF SILVER IONS TABLE 2.-Ag" CENTRES DETECTED IN PURE- AND MIXED-SOLVENT SYSTEMS AT 77 K AisoIG 14N 10SAg 14N 109Ag solvent gl g , ga, CN,CN (CD3)2C0b C DMSO nit rome thane pyridine 33% aq. NH, formamide DMF CD,CN + (CD,),CO mole fraction (CD,),CO: 0.76 C CD,CN +DMSO mole fraction DMSO: O.25-0.tl6 C 2.312 2.312 2.361 2.293 2.360 2.189 2.159 2.305 2.321 2.31 1 2.360 2.283 2.302 U 2.067 2.067 2.059 2.076 2.037 2.062 2.065 - 2.067 2.067 2.058 2.058 - 2.149 2.165 2.137 - 2.148 2.165 2.133 2.137 21 46 26 - 26 - 26 34 - 30 32 - 25 29 22 - 18 36 27 - 27 32 - 27 - - - - - - - - - - - - 32 - 24 35 - 30 - - 24 - 25 32 - 25 - - - 26.0 31.3 27.3 - 26.7 31.7 24.7 27.3 a Features not resolved; before annealing; after annealing.centres containing lo9Ag+lo9Ag, lo9Ag+lo7Ag and lo7Ag+lo7Ag give different splittings. (Both lo9Ag and lo7Ag have I = 4. They are present in nearly 50% abundance. Both have negative magnetic moments, that for lo9Ag being slightly greater in magnitude than that for lo7Ag.) These features exhibit little anisotropy. On annealing the features for Agz first narrowed and then decayed irreversibly, with features assigned to Agi+ growing-ing [fig. 1 (b)]. These features together with those for the remaining centres decayed at ca. 155 K.A range of Ago centres were detected at 77 K. Unfortunately we were unable to irradiate at 4 K so cannot judge conclusively which of these centres is the primary centre. However, as with MeCN systems, we assume that, in general, as the lo9Ag hyperfine splitting increases, so the solvent molecules originally solvating the Ag+ ions are being discarded. As stressed elsewhere, however, this is not necessarily true, since 5 s - 5 ~ admixture will result in a fall in the magnitude of the isotropic coupling and this is most likely to occur for unsymmetrical complexes.1o However, this is expected to result in g-anisotropy, with a shift to g-values < 2.00. Such shifts were not detected for the present centres.Thus the centre labelled A is probably the primary species. Centre F may not be a genuine species, since its features were always so weak that they could not be clearly resolved. The most atom-like centre, C (spin density ca. 0.99), is probably a monosolvate. Spectra for unsolvated silver atoms were not detected in this solvent. These features had all decayed at ca. 135 K. These results are interesting since they clearly establish the presence of a range of solvated silver units, but the work of Kevan and coworkers establishes that it will be necessary to work at lower temperatures before the full picture can be e~tablished.~? l1 The AgI1 spectrum changed markedly on annealing to 130 K, the parallel lines moving from 2.312 to 2.361, All(logAg) increasing from 26 to 34 G concomitantly.This probably indicates a relaxation towards the square-pyramidal arrangement of solvent molecules expected for this 4d complex. It is significant that no such change occurredM. C. R. SYMONS A N D C. K. ALESBURY 3633 TABLE 3.-sILVER CLUSTER AND OTHER CENTRES DETECTED IN PURE- AND MIXED-SOLVENT SYSTEMS AT 77 K solvent species giso Aiso(’”Ag)lG CD,CN (CD3)2C0 DMSO MeTHF ni trome thane [2H2]ethylamine pyridine 33% aq. NH, formamide DMF benzene toluene CD,CN + CD,OD CD,CN + (CD,),CO mole fraction (CD,),CO: 0-0.4 CD,CN + DMSO mole fraction DMSO: 0.25-0.8 CD,CN + MeTHF mole fraction MeTHF: ca. 0.75 DMF CD,CN + (CD,),CO mole fraction CD,CN: 0.5-0.9 1.971 1.971 1.929 1.979 1.965 1.951 1.964 1.996 1.968 1.996 - - 1.967 1.968 1.957 1.973 1.968 1.973 1.954 1.972 1.971 1.971 1.963 1.953 1.978 1.965 1.976 1.965 2.002 2.002 - 267 - 280 -112 - 273 - 129 - 138 - 148 - 225 - 157 - 200 - 130 - 100 - 290 - 267 - 137 - 194 -281 - 194 - 135 - 267 - 134 - 267 - 182 - 124 -271 - 132 - 276 - 135 - 100 - 82 a See text.on annealing above 77 K for the AgII(MeCN), centre, our conclusion being that the expected square-planar structure had already been acquired at this temperature.2 Interestingly, features for AglI at 4 K were very weak and poorly defined.5 This may reflect the inability of the tetrasolvate to change from its original configuration to the square-planar form at 4 K. The features for the 77 K centre grew in markedly on annealing, as e~pected.~ DIMETHYLSULPHOXIDE Unexpectedly, the main centre formed by electron addition at 77 K was Agi (fig.2). This probably implies some aggregation prior to irradiation since extensive mobility at 77 K is unlikely. Since DMSO is a strong ligand, as indicated by the low3634 SOLVATION OF SILVER 1 1 IONS FIG. 2.-First-derivative X-band e.s.r. spectra of AgCIO, in (CD,),SO after exposure to 6oCo y-rays at 77 K, showing Ago, Ag: and AgU centres. Central features for solvent radicals are not shown. value of AisO(lo9Ag) for the Ago centre, it is unlikely that Ag+-ClO; aggregates are present in the solutions. Only a single Ago centre was detected, in relatively low yield. The Agl* centre was well defined, and did not change on annealing. As usual, on annealing the Agi features gave way to Ag;+ features before the medium became fluid.METHYLTETRAHYDROFURAN This solvent was selected as typical of monoethers because good glasses are obtained on cooling. Two types of Ago centres were formed, as indicated in fig. 3, these species being dominant at 77 K. The spin densities of 0.90 (A) and 0.84 (B) indicate considerable delocalisation onto solvent ligands. The very low relative yield of AglI centres is expected for this solvent, which is unable to react with electrons, but traps the primary electron-loss centres efficiently.12 This is well illustrated by the fact that in the absence ofAgC10, the ejected electrons are shallowly trapped, giving a blue-black coloration to the glasses. In the presence of AgCl, the glasses become pale yellow, showing that electrons are very efficiently scavenged at 77 K.On annealing, Agi centres and later Ag;+ centres were formed, but no unsolvated silver atom features were detected. NITROMETHANE Pure MeNO, gives Me’ and ‘NO, radicals on irradiation at 77 K, neither electron-gain nor electron-loss centres bring trapped.13 Hence, as expected, both Ago and AglI centres were formed in high yield from solutions of AgClO, in this solvent. The Ago centre features were extremely broad, the large hyperfine coupling to lo9Ag ( - 661 G) indicating relatively minor delocalisation, as expected for this weakly basic solvent. The g-value of 1.995 may reflect slight admixture of 5p orbitals into the wavefunction, so that the actual spin density could be even higher than the value of ca. 92% deduced from Aiso. Agi+ centres grew-in on annealing, without any clear intermediate formation of Agz or Agi+ centres.M.C. R. SYMONS A N D C. K. ALESBURY 3635 ? 1 Gain x 100 I Sdwnt radicals FIG. 3.-First-derivative X-band e.s.r. spectra of AgClO, in MTHF after exposure to 6oCo prays at 77 K, showing features assigned to AgO (A and B), Ag; and central features for MTHF radicals. ETHYLAMINE A N D TRIETHYLAMINE The only well-defined product using EtNH, or EtND, was an Agi centre. For triethylamine solutions very broad Ago features were obtained, with a low value for Ai,,(lOgAg) (ca. -511 G). In this case no cluster centres were detected, even on annealing, nor were any AglI features detected. The contrast between these two solvents is most interesting. Triethylamine is an aprotic, strongly basic solvent, able to solvate Ag+ ions strongly, but unable to form hydrogen bonds to perchlorate ions.It might therefore be expected that ion clusters would be favoured in this solvent rather than in ethylamine, which can form hydrogen bonds to anions. Possibly ion clusters such as that in I (see later) tend to form in methylamine. Solvent-shared ion pairs do seem to be favoured in basic-protic media, especially at low temperatures. l4 PYRIDINE The major centre in this solvent was an AglI species for which the perpendicular feature showed well-defined hyperfine coupling to 14N nuclei (probably four). The very broad parallel feature was, unfortunately, unresolved in part because of overlap with broad features assigned to Agi (fig. 4). The central well-defined triplet is due to 1 -pyridyl radi~a1s.l~ These results suggest that pyridine may be a good solvent for forming electron-loss centres in preference to electron-gain centres.AQUEOUS AMMONIA Although liquid ammonia was not studied, aqueous ammonia glasses up to ca. 30% ammonia were used. Since silver halides are soluble in this medium they were used in addition to AgClO,; the initial results were independent of the anion, showing that ion-pairing or complex formation was not important. Very broad Ago features dominated the spectra at 77 K. Although hyperfine coupling to I4N is almost certainly3636 SOLVATION OF SILVER IONS n i , 32506 Gain x l o 4 Solvent radicals FIG. 4.-First-derivative X-band e.s.r. spectra of AgC10, in pyridine after exposure to 6oCo y-rays at 77 K and slight annealing, showing well-defined features for Ag" centres and the triplet features for pyridyl radicals.log 2+ I I I I I I I I I Ag, + 4 +3+2 +1 0 - 1 - 2 - 3 - 4 107 2+ I I I I I I I I I Ag, +4+3+2+1 0 - 1 - 2 - 3 - 4 FIG. 5.-First-derivative X-band e.s.r. spectra of AgF in 33% (as.) NH, after exposure to 6oCo y-rays at 77 K and annealing, showing well-defined features for Ag" centres.3637 responsible for this width, unfortunately it was not resolved. This probably means that a range of solvates containing both H,O and NH, ligands is present, each one capable of adding electrons. That ammonia ligands must be present is established by the remarkably low value for Aiso (lo9Ag) (ca. -451 G), corresponding to only ca. 64% spin density on silver. On annealing, features for Ag: became better defined, and again showed unusual width and an exceptionally low value for Aiso (- 200 G).These differences presumably arise because NH, ligands are retained in the dimer, delocalisation still being quite extensive. When the central signals, due mainly to 'NH, radicals,lG were lost on annealing, there was a concomitant growth in features assigned to an Agl* complex (fig. 5). As with pyridine, the perpendicular features were well resolved, showing hyperfine coupling to three or four 14N nuclei. Our preferred analysis is based on the presence of four ligands (fig. 5), but three equivalent 14N nuclei are equally probable. The possibility that only two NH, ligands were present can be ruled out. This complex, which is clearly a square-planar complex having a (- - - 4d$-Y2) electron configuration, is presumably formed by attack of 'NH, radicals on Agl complexes.Since these radicals are not likely to be efficient electron-acceptors, it is probable that these radicals add to the AgI, possibly remaining as ligands. Thus if, as expected,17 the major Agl complex is a linear complex, Ag(NH,)i, presumably with water molecules less strongly coordinated in the equatorial plane, ' NH, radicals would convert these into complexes having three equatorial ligands containing 14N nuclei rather than four. M. C. R. SYMONS A N D C. K. ALESBURY FORMAMIDE A N D DIMETHYLFORMAMIDE For formamide at 77 K, broad features assigned to Ago, Ag: and AglI centres were obtained. On annealing, the Agz features became better resolved and two different Ago centres were apparent.No 14N splitting was observed for these centres, but the features were broad. For [2H7]DMF, resolution was better, and several types of Ago centres were detected (fig. 6). Also a new species (a) was detected, which is discussed below. In addition A i B I 3250G FIG. 6.-First-derivative X-band e.s.r. spectra of AgC10, in [*H,]DMF after exposure to 6oCo y-rays at 77 K, showing features assigned to Ago (A and B) and Ag" centres, together with features (a) assigned to species having structure II.3638 SOLVATION OF SILVER IONS to weak unsolvated atom features, very narrow features (A) for a weakly solvated species, and intense, broad features for a strongly solvated species (B) were obtained. Species (A) grew at the expense of (B) on annealing, probably as a result of loss of primary solvent molecules.A most intriguing aspect of the spectra for B is the appearance of hyperfine splitting in the outer regions of all four features. Only 3-5 features were resolved, but high-resolution spectra reveal the presence of several further shoulders. Since the splitting of cu. 6 G is equal to that found for 14N coupling in the Ago(MeCN), complexes, it is tempting to assign this coupling to two or more 14N nuclei. This result is most unexpected since it is generally found that amides coordinate uiu oxygen. Indeed, amides are locally nearly planar at nitrogen; the nitrogen 2p-electrons, being part of a delocalised ;n-system, are not normally available for coordination to cations. We hope to study this possibility using infrared spectroscopy.BENZENE AND TOLUENE The unusually high solubility of silver salts in aromatic solvents arises because Ag+ ions form weak complexes with unsaturated systems. Indeed, well-defined com- plexes such as C,H,*AgC10,,18 are well known. This complex consists of -C,H,-Ag+-C,H,-Ag+- chains, each benzene ring being coordinated to two Ag+ ions lying above and below opposite bonds of the ring.19y20 We find that the isolated complexes are very radiation resistant, the only para- magnetic silver centre detectable after long exposure at 77 K being an Agz+ unit. However, as reported previously,21 frozen solutions do give fair yields of electron- excess centres on irradiation. The results of the present work, which was directed towards mixed-solvent systems, differ from those previously reported, for no clear reasons.We now find that solutions 1 32506 Solvent radicals Gain x 10 ( 107 Ago + ';go M I = +'h FIG. 7.-First-derivative X-band e.s.r. spectra of AgClO, in toluene after exposure to 6oCo prays at 77 K, showing features assigned to Ago centres and solvent radicals. Weak features for Agl are also seen.M. C. R. SYMONS A N D C. K. ALESBURY 3639 in dry benzene give, after irradiation, well-defined Agi+ centres with clear resolution of the four sub-components from species containing differing proportions of lo9Ag and lo7Ag. Dry toluene solutions gave an Ago centre, not previously detected in aromatic solvents (fig. 7). We suspected that the reason why different types of silver centres were formed in benzene under different conditions might be connected with phase separation, so we also studied the effect of ionizing radiation on the isolated complex (C,H, * AgC10,).In fact, the radical yield was very low, the only detectable silver species being Agi+, the MI = +E line showing the expected four features from different lo9Ag+ lo7Ag combinations. It seems probable that dilute solutions at low temperatures comprise aggregates containing three Ag+ ions and probably three benzene molecules, and these share one electron between them. Toluene gives good glasses on rapid freezing, which probably explains why monatomic centres can be formed. The high %-character for these previously unknown centres (ca. 94%) shows that delocalisation into the aromatic ring is minor.In other words the 5s(Ag) orbital is not greatly involved in the bonding for these complexes. When the silver(1)-toluene complex was isolated and irradiated, yields were again low, A&+ and A&+ being the dominating electron-excess centres. The relatively low A(lo9Ag) values for these complexes (A&+, - 163 G, corresponding to ca. 697: spin density on silver; A&+, - 122 G, also corresponding to ca. 69% on silver) suggests that the aromatic rings are still involved in these structures, and that delocalisation is facilitated relative to the monomer units. The data for the Agg+ complex are very close to those for the benzene complex, suggesting comparable structures. CORRELATION OF Ai,,(109Ag) WITH SOLVENT DONOR NUMBERS The donor number (DN) of a solvent is a widely used empirical measure of its ability to coordinate via a lone-pair of electrons.It is based upon the heat of formation of a single coordinate bond to SbC1,.22 We use it because such bonding is thought to be involved when Agl is solvated by the solvents used herein. It must be borne in mind that the correlation with Ai,,(109Ag) is only expected to hold for complexes having the same solvation number. Clearly, as the solvation number increases so the extent of electron delocalisation must increase. Hence only those species which appear to be primary species have been utilised in the correlation which is shown in fig. 8. That the correlation is reasonable for some 10 systems supports the idea that bonding into the outer vacant (for AgI) 5s and 5p orbitals dominates solvation for Ag' systems.It is interesting that data for silver nitrate in cyanomethane and cyanoethane lie close to the line, whereas those for silver perchlorate are well removed therefrom. This fits in with our suggestion that silver is coordinated to oxygen ligands from two nitrate ions.z3 In that case, the donor numbers for MeCN or EtCN are no longer appropriate. If we move these points horizontally onto the correlating line, we obtain a donor number for NO; of ca. 18, close to those for water or methanol. This seems to us to be entirely reasonable. The data for solutions in MeCN and EtCN suggest that these molecules coordinate to silver ions far more strongly than one would expect for these weakly basic solvents.In our view, this can best be explained in terms of back-donation of silver 4d (n) electrons into the otherwise vacant n* (x, y ) orbitals of the N-C group (the ' back-Chatt ' effectz4). Such bonding will reduce the silver-nitrogen bond length, thereby increasing a-overlap and delocalisation. It will also increase the positive charge on silver and the negative charge on the N-C group, thus also increasing the strength of the a-bonds.3640 450. SOLVATION OF SILVER IONS 0 EtCN MeOH MeTHFDMF DMSO HMPA I I 1 I I I I I c3 '= 0 ' 2 v 0 650 - 0 550 t ( AgClO,) 0 1 Me *, \ Triethylamine is not able to participate in such bonding, but pyridine can, and it is noteworthy that once again the A(logAg) value is far lower than predicted by the correlation. MIXED SOLVENTS WITH CYANOMETHANE MeCN + Me,CO The most remarkable result, as with the mixed MeCN + H,O and MeCN + MeOH system^,^ was that the AgO(MeCN), and Ag"(MeCN), complexes persisted up to the 0.5 mole fraction region before resolution into the nine (14N) hyperfine components was lost.In the 0.5-0.87 mole fraction region these features lost all resolution, and features for Agz centres grew in. No clear evidence for specific mixed solvates was obtained, results for the Agl* centre suggesting that damage to the Ag*(MeCN), centres may have been preferred over other mixed complexes since features for the tetracyanomethane derivative were still detectable in the 0.8 mole fraction region. Ultimately features characteristic of solutions in pure acetone were obtained in the > 0.9 mole fraction region.In most systems, on annealing, features for Agi, Agi+ and Ag:+ were detected. Other species with relatively low hyperfine coupling to lo9Ag, formed in some of these mixed-solvent systems, are discussed below. MeCN + DMSO We expected, from the correlation in fig. 8, that DMSO would displace MeCN far more readily than Me,CO because of its high donor number. This was indeed the case. Beyond ca. 0.1 mole fraction (DMSO) the 14N hyperfine features began to broaden with loss of resolution. This probably reflects absorption from AgO(DMSO),(MeCN),, AgO(DMSO),(MeCN,) and other mixed solvates, the net effect of a range of coupling constants and lines in different positions being simply to broaden out the features, as observed.M. C. R. SYMONS A N D C.K. ALESBURY 364 1 However, as with pure DMSO, these mixtures in the region > 0.8 mole fraction gave mainly Agi centres, which, on annealing, readily gave Ag;+ centres. Also, well-defined AglI centres characteristic of DMSO binding were obtained. Thus, in this mixed-solvent system there was no evidence for preferential solvation by MeCN. MeCN + MTHF In this case, preferential solvation by MeCN resembled that for acetone, the Ago(MeCN), and AgIl(MeCN), centres dominating the e.s.r. spectra up to ca. 0.6 mole fraction MTHF. From then on there was a gradual loss of resolution and increase in A(lo9Ag) for the Ago centres. Species B, detected in pure MTHF, grew in the 0.8 mole fraction region, prior to the appearance of species A. Features for AglI centres gradually diminished in the MTHF-rich region, being finally lost for the pure-solvent system.OTHER MIXED-SOLVENT SYSTEMS MTHF+ Et,N Again, as predicted from the correlation of fig. 8 and the high donor number of Et,N, the species found in pure Et,N dominated to beyond the 0.95 mole fraction MTHF range, indicating extreme preferential solvation by Et,N. C,H,+MTHF AND MeNO, On addition of 0.05 mole fraction MTHF, the Agg+ signals characteristic of benzene clusters were completely lost, and features characteristic of Ag-MTHF complexes were detected. Addition of MeNO, with a lower donor number required far greater concentrations to change the spectrum from that of pure benzene. These systems were not studied extensively, but it seems clear that the donor number is a fair measure of coordinating ability with the marked exceptions of cyanoalkanes and pyridines, and that the aromatic complexes are in no way exceptional in this regard, THE AglI CENTRES All the data obtained for AglI centres (table 2) are characteristic of d 9 complexes with a (- - - 4diz-,2) onf figuration.^^ However, in several cases, marked changes occurred on annealing involving, mainly, increases in gI1 and the magnitude of A II (lo9Ag).We suggest that these changes are the consequences of the complexes relaxing from a near-tetrahedral configuration fairly close to that of the parent Agl complexes, towards the square-planar structures that are favoured by a di2-y2 configuration. After allowing for the difference in spin-orbit coupling constants and nuclear magnetic moments, the changes are quite similar to those predicted by our correlation for copper(r1) (3d$-,2) complexes.26 OTHER PARAMAGNETIC SILVER CENTRES The cluster centres, Agi, Agz+ and Ag;+ centres have been discussed for each solvent system.A wide range of such complexes have now been studied, and the isotropic coupling constants to lo9Ag change quite markedly from one system to another. In many cases these trends resemble the trends shown in fig. 8 for Ago centres. This implies that coordination of solvent molecules is retained, as has been clearly established for the Agi+ centre detected in irradiated silver imidazole per~hlorate.~~ Other aspects of the data for these remarkable complexes will be considered later in connection with some studies on conduction-electron centres.A species previously described, probably having the limiting structure I,, was formed in many of the MeCN systems. This species has an isotropic coupling to lo9Ag3642 SOLVATION OF SILVER IONS of ca. 180 G. Yet another centre with a relatively small but isotropic coupling to lo9Ag and g-values close to 2.00 has now been detected. This species, with A(lo9Ag) = - 100 G, was formed in DMF solutions, as shown in fig. 6. There is some indication of 14N hyperfine splitting on these features, with A(14N) z 17 G, but this was never well resolved. This centre must differ from the Ago centres formed in this medium, so that simple coordination to nitrogen or oxygen is ruled out. Possibly some sort of n-bonding such as that depicted in I1 is involved.A similar species was formed in the 0.5-0.9 mole fraction (acetone) range for mixed MeCN + Me,CO systems, after annealing. This species, having IAi,,(logAg)l - 82 G UY I V (fig. 9, is thought to be a complex between Ago and acetone, since it was favoured in the high mole fraction (acetone) region, although it was not detected in pure acetone solutions. We think that the extra splitting clearly resolved in the high-field feature is due to g-value variation rather than 14N coupling but, unfortunately, we could not test this suggestion by studying the spectrum at Q-band frequencies because of low instrument sensitivity. A possible structure for this complex is shown by 111. 1 32506 FIG. 9.-First-derivative X-band e.s.r. spectra of AgCIO, in CD,CN + (CD,),CO (mole fraction 0.625) exposure to 6oCo y-rays at 77 K and annealing, showing features assigned to a species thought to complex between Ago and acetone (In).after be aM. C. R . SYMONS A N D C. K. ALESBURY 3643 Alternatively, these complexes may be formed from D,CCO(CD,) radicals [or D,cN(CD,)COD radicals for DMF] having structure IV. A species with such a structure (Ag-CH,OH) was detected in methanolic solutions, having Ai,O(lOgAg) e 130 G.2 CONCLUSIONS We conclude that most solvents bind to Ag+ via a-donation into 5s+ 5p manifold. On electron addition at low temperatures the primary solvent molecules are retained, the unpaired electron being extensively delocalised onto the ligands. The extent of delocalisation, measured by the reduction in Ai,O(lOgAg), correlates quite well with the donor number of the solvent, with the notable exceptions of cyanoalkane solvents and pyridine.In these cases back-bonding by 4d electrons is thought to be important. On annealing, solvent molecules are lost from the Ago centres, and extensive clustering usually occurs. The Ag" centres show evidence for relaxation from a near-tetrahedral to a square-planar configuration on annealing. We thank the S.E.R.C. and Kodak (Harrow) Ltd fgr a CASE Studentship (to C.K.A.) and Dr G. Farnell for helpful discussions. D. R. Brown, G. W. Eastland and M. C. R. Symons, Chem. Phys. Lett., 1979, 61, 92. C. K. Alesbury and M. C. R. Symons, J. Chem. SOC., Faraday Trans. 1, 1980, 76, 244. B. G. Cox, R. Natarajan and W. E. Waghorne, J. Chem. SOC., Faraday Trans. 1, 1979, 75, 86. A. P. Zuur and W. L. Groeneveld, Recueil, 1967, 86, 1089. A. S. W. Li and L. Kevan, J. Phys. Chem., 1981, 85, 2557. R. J. England and M. C. R. Symons, J. Chem. SOC. A, 1970, 1326. M. C . R. Symons and G. W. Eastland, unpublished. R. S. Eachus and M. C. R. Symons, J. Chem. SOC. A, 1970, 1329; 1970, 3080. * L. Shields and M. C . R. Symons, Mol. Phys., 1966, 11, 57. lo M. C. R. Symons, f. Chem. Phys., 1978, 69, 3443. l 1 L. Kevan, H. Hase and K. Kawabata, J. Chem. Phys., 1977, 66, 3834. l 2 M. C. R. Symons, Pure Appl. Chem., 1981, 53, 223. l 3 Unpublished results. l4 I. M. Strauss and M. C. R. Symons, J. Chem. SOC., Faraday Trans. I , 1978, 74, 2146. l5 H. J. Bower, J. A. McRae and M. C. R. Symons, J. Chem. SOC. A, 1968, 2696. l6 K. V. S. Rao and M. C. R. Symons, f. Chem. SOC. A, 1971, 2163. l 7 F. A. Cotton and G. Wilkinson, Adzianced Inorganic Chemistry (Wiley-Interscience, New York, 3rd edn, 1972). A. E. Hill, J. Am. Chem. SOC., 1922, 44, 1163. l9 R. E. Rundle and J. H. Goring, J. Am. Chem. SOC., 1950, 72, 5337. 2o H. G. Smith and R. E. Rundle, J. Am. Chem. SOC., 1958, 80, 5075. 21 C. E. Forbes and M. C. R. Symons, Mol. Phys., 1974, 27,467. 22 V. Guttmann, Donor-Acceptor Approach to Molecular Interactions (Plenum Press, New York, 1978). 23 D. R. Brown, T. J. V. Findlay and M. C. R. Symons, f. Chem. SOC., Faraday Trans. I , 1976,72, 1792. 24 J. Chatt, Ric. Sci. Suppl., 1958, 23, 130. 25 M. C. R. Symons, Chemical and Biochemical Aspects of Electron Spin Resonance Spectroscopy (Van Nostrand Reinhold, London, 1978). 26 M. C. R. Symons, D. X. West and J. G. Wilkinson, J. Chem. SOC., Dalton Trans., 1975, 1696. 27 G. W. Eastland, M. A. Mazid, D. R. Russell and M. C. R. Symons, J . Chem. SOC., Dalton Trans., 1980. 1682. (PAPER 2/568)