A combined kinetico-mechanistic and computational study on the competitive formation of seven-versus five-membered platinacycles ; the relevance of spectator halide ligands

The metalation reactions between [Pt2(4-MeC6H4)4(μ-SEt2)2] and 2-X,6-FC6H3CH[double bond, length as m-dash]NCH2CH2NMe2 (X = Br, Cl) have been studied. In all cases, seven-membered platinacycles are formed in a process that involves an initial reductive elimination from cyclometallated Pt(IV) intermediate compounds, [PtX(4-CH3C6H4)2(ArCH[double bond, length as m-dash]NCH2CH2NMe2)] (X = Br, Cl), followed by isomerization of the resulting Pt(II) complexes and a final cyclometallation step. For the process with X = Br, the final seven-membered platinacycle and two intermediates, isolated under the conditions implemented from parallel kinetic studies, have been characterized by XRD. Contrary to previous results for the parent non-fluorinated imine 2-BrC6H4CH[double bond, length as m-dash]NCH2CH2NMe2 the presence of a fluoro substituent prevents the formation of the more stable five-membered platinacycle. Temperature and pressure dependent kinetico-mechanistic and DFT studies indicate that the final cyclometallation step is strongly influenced by the nature of the spectator halido ligand, the overall reaction being much faster for X = Cl. The same DFT study conducted on the previously studied systems with imine 2-BrC6H4CH[double bond, length as m-dash]NCH2CH2NMe2 indicates that, when possible, five-membered platinacycles are kinetically preferred for X = Br, while the presence of Cl as a spectator halido ligand leads to a preferential faster formation of seven-membered analogues.

The metalation reactions between [Pt2(4-MeC6H4)4(μ-SEt2)2] and 2-X,6-FC6H3CHvNCH2CH2NMe2 34 (X = Br, Cl) have been studied. In all cases, seven-membered platinacycles are formed in a process that 35 involves an initial reductive elimination from cyclometallated PtIV intermediate compounds,  CH3C6H4)2(ArCHvNCH2CH2NMe2)] (X = Br, Cl), followed by isomerization of the resulting PtII 37 complexes and a final cyclometallation step. For the process with X = Br, the final seven-membered 38 platinacycle and two intermediates, isolated under the conditions implemented from parallel kinetic 39 studies, have been characterized by XRD. Contrary to previous results for the parent non-fluorinated 40 imine 2-BrC6H4CHvNCH2CH2NMe2 the presence of a fluoro substituent prevents the formation of the 41 more stable five-membered platinacycle. Temperature and pressure dependent kinetico-mechanistic and 42 DFT studies indicate that the final cyclometallation step is strongly influenced by the nature of the 43 spectator halido ligand, the overall reaction being much faster for X = Cl. The same DFT study 44 conducted on the previously studied systems with imine 2-BrC6H4CHvNCH2CH2NMe2 indicates that, 45 when possible, fivemembered platinacycles are kinetically preferred for X = Br, while the presence of Cl 46 as a spectator halido ligand leads to a preferential faster formation of seven-membered analogues. compounds are considered adequate models to study reductive elimination processes from d6 octahedral 54 complexes. Recent findings based on reductive elimination from platinum(IV) complexes include the 55 formation of C-C and C-halide bonds1 and a catalytic process for conversion of a C-F bond into a C-C 56 bond.2 In particular, cyclometallated platinum(IV) compounds [PtXR2(Ar′CHNCH2CH2NMe2)] or 57 [PtXR2 (Ar′CHNCH2Ar′)L] containing respectively a tridentate [C,N,N′] ligand or a bidentate [C,N] 58 and a neutral monodentate L ligand can be easily obtained from the reactions of platinum precursors 59 containing "PtR2" moieties and potentially tridentate or bidentate imine ligands, 60 Ar′CHNCH2CH2NMe2 or Ar′CHNCH2Ar′. In recent years, we have been involved in studies related to 61 the formation of platinum(II) cyclometallated compounds generated from the mentioned platinum(IV) 62 cyclometallated compounds with tridentate [C,N,N′] amino-imine ligands, or bidentate [C,N] imine 63 ligands, and a neutral monodentate ligand L.3-12 The interest in these reactions arises from the fact that 64 along the process new C-C bonds are formed via an initial reductive elimination to give a 65 noncyclometallated platinum(II) compound that, in the second step, evolves towards a cycloplatinated 66 compound. Although a common sequence operates, this process is highly versatile since both the nature 67 of the formed C-C bond and the structure of the final cyclometallated platinum(II) compound can be 68 tuned by a judicious choice of both the platinum precursor "PtR2" and the imine ligand, 69 Ar′CHNCH2CH2NMe2 or Ar′CHNCH2Ar′, used in the formation of the cyclometallated platinum (IV)  70 compound. For instance, the reaction of [Pt2Me4(μ-SMe2)2] with imine ligands Ar′CHNCH2Ar′ leads 71 to a cyclometallated platinum(IV) compound from which Caryl-Calkyl bonds are formed and the 72 subsequent metalation can lead to either five or six-membered metallacycles corresponding to C-H 73 activation at the aryl group of the imine, or at the methyl ligand of the platinum precursor, respectively 74 (Scheme 1).3,5 75 For diarylplatinum precursors, the corresponding platinum(IV) compounds lead to the formation of 76 Caryl-Caryl bonds from which either seven-membered platinacycles containing the new biaryl fragment 77 or five-membered analogues, in which the newly formed C-C bond is outside the metallacycle, can be 78 obtained.4,6,7,9-12 For the former, C-H bond activation takes place at the aryl ligand of the precursor, 79 while for the latter this process takes place at the aryl ring of the imine. A clear example (Scheme 2) is 80 obtained when cis-[Pt(C6F5)2(SEt2)2] is used, since, in this case, the ortho-fluorine substituents 81 preclude the formation of seven-membered platinacycles, due to the low reactivity of C-F bonds, and 82 therefore the reaction is directed towards the formation of five-membered analogues.8 83 A more striking result, shown in Scheme 3, was obtained in the reaction of [Pt2(4-MeC6H4)4(μ-SEt2)2] 84 with imines 2-XC6H4CHvNCH2CH2NMe2 (X = Br or Cl) since, in this case, the nature of the halide is 85 determinant: a five-membered platinacycle is obtained for X = Br, while a seven-membered platinacycle 86 is produced for X = Cl. This system has been thoroughly studied from a kinetico-mechanistic point of 87 view and although formation of a seven-membered platinacycle for X = Br was found plausible under 88 harsher conditions, the compound could not be obtained in a pure form.4 89 Since seven-membered platinacycles are a novel class of compounds with potential interest associated 90 with their cytotoxic properties,13,14 in addition to the intrinsic interest based on the formation of biaryl 91 linkages, we decided to explore novel strategies in order to analyse whether it would be possible to 92 obtain such compounds even when X = Br. In this work, the reactions of [Pt2(4-MeC6H4)4(μ-SEt2)2] 93 with imines 2-X,6-FC6H3CHvNCH2CH2NMe2 (X = Br or Cl) have been studied with the idea that the 94 fluoro substituent at the ortho position in the aryl ring of the imine ligand should prevent the C-H bond 95 activation at the imine and thus, in both cases, the reaction would be driven towards the formation of 96 seven-membered platinacycles rather than five-membered analogues. It should be noted that five-97 membered metallacycles are more stable than other ring products and generally cyclometallation 98 reactions take place with high regioselectivity to produce fivemembered rings.15-17 Kinetico-99 mechanistic and DFT studies of these types of systems should allow studying the effect of the nature of 100 the halide (Br versus Cl) in the formation of sevenmembered platinacycles. 101

RESULTS AND DISCUSSION 103 104
Preparation and characterisation of compounds 105 Initial platinum(IV) compounds [PtX(4-MeC6H4)2(2-FC6H3CHv NCH2CH2NMe2)] (5-IV-X,F, X = 106 Br and Cl; Scheme 4) were prepared in high yields, following previously established procedures, from 107 [Pt2(4-MeC6H4)4(μ-SEt2)2] and imines 2-X,6-FC6H3CHvNCH2CH2NMe2 (X = Br and Cl).18 For X 108 = Br the reaction was faster and was complete within 24 hours in toluene solution at room temperature, 109 while for X = Cl the reaction requires 48 hours under the same conditions. As expected from the lower 110 reactivity of C-F bonds, activation of this bond was not observed in either reaction. Using shorter 111 reaction times, isolation and characterisation of the coordination compound [Pt(4-MeC6H4)2(2-F,6-112 ClC6H3CHvNCH2CH2NMe2)], formed prior to the intramolecular C-Cl bond activation has also been 113 achieved; isolation of the corresponding bromo analogue has not been possible. In this case the 114 intramolecular C-Br bond activation occurred readily after coordination of the imine ligand to platinum. 115 All the isolated compounds were characterized by 1H and 19F NMR spectra, which were consistent 116 with the expected structures as well as from the data available for analogous compounds.4,9 As 117 expected, the J(Himine-Pt) values observed for the platinum(IV) compounds (45.6 and 46.0 Hz) are 118 lower than those observed for the platinum(II) compound [Pt(4-MeC6H4)2(2-F,6-119 ClC6H3CHvNCH2CH2NMe2)] (50.8 Hz). For the latter, the J(Himine-Pt) value is consistent with both 120 an E conformation of the imine moiety and the presence of an aryl ligand trans to this group.4 A set of 121 signals of very low intensity that could not be fully assigned indicated also the presence of a Z isomer in 122 the sample in small amounts (<5%). 123 When a toluene solution of compound 5-IV-Br,F was refluxed for 24 hours, the targeted seven-124 membered platinacycle [PtBr{(4-MeC6H3)(2-FC6H3)CHNCH2CH2NMe2}] (7-II-Br,F) was obtained.

125
This compound could also be obtained in a onepot process from [Pt2(4-MeC6H4)4(μ-SEt2)2] and the 126 corresponding 2-Br,6-FC6H3CHvNCH2CH2NMe2 imine under the same conditions. These results 127 indicate that the presence of an inert C-F bond in the imine ligand is an efficient strategy to drive the 128 reaction towards the formation of seven-membered platinacycles. Formation of the analogue chlorido crystal structure of this compound confirms that a biphenyl fragment involving a former para-tolyl 160 ligand and the aryl ring of the initial ligand is formed from 5-IV-Br,F in a reductive elimination process. 161 As already reported, the compounds generated on reductive elimination on platinum(IV) compounds of 162 the type 5-IV-X,Y may adopt four distinct isomeric forms (see Scheme 5); the aryl ring being trans to 163 the amine or the imine moieties and with an E or Z imine conformation.4 By monitoring changes in the 164 1H NMR spectra of II-Br,F, under the conditions suggested by the kinetic experiments detailed in the 165 next section, compound II′-Br,F, as a mixture of E and Z isomers in a proportion E : Z = 2 : 1, was 166 obtained. As previously reported,4 the values of J(Himine-Pt), which in this case are 152 Hz and 84 Hz 167 for the E and Z isomers, respectively, indicate the presence of an halide ligand trans to the imine. From 168 this mixture, XRD quality crystals were obtained and analysed; the molecular structure is shown in Fig.  169 1c and selected molecular dimensions are listed in Table 1. This compound differs from the previously 170 described intermediate in that the bromido ligand is now trans to the imino fragment, and the latter 171 displays a Z arrangement. A careful examination of this Z isomeric form of the species results in clear 172 evidence that this form cannot produce the final seven membered platinacycles for orientation reasons.

173
A similar treatment carried out on compound II-Cl,F produced rather complex 1H NMR spectra which 174 consist of mixtures of up to four possible isomers of the species plus the initial and final reaction 175 compounds (i.e. 5-IV-Cl,F and 7-II-Cl, F). From these complex mixtures, already anticipated from the 176 data collected in the next section, it was not possible to isolate any of the relevant species. 177 178 Kinetico-mechanistic studies on the formation of sevenmembered 7-II-Br,F and 7-II-Cl,F 179 metallacycles 180 The rather complex nature of both the possible reaction intermediates and the nature of the final 181 cyclometallated complexes formed in the reactions is generalised in Scheme 5. This general scheme is 182 clear both from some previous results already published,4,19 and those indicated in the previous section. 183 Given the fact that time-resolved monitoring of the processes has been found to be a perfect handle to 184 gain a better insight into the reaction mechanism, the UV-Vis monitoring of the transformation of 185 complexes 5-IV-X,F has been conducted from a kinetic perspective. 186 For complex 5-IV-Br,F the spectral changes observed on monitoring 5 × 10−4 M xylene solutions at 187 varying temperatures indicate the operation of a two-step process in the 2-24 hour range at 90 and 60 °C 188 respectively. By using the standard software indicated in the Experimental section, these changes could 189 be easily fitted to a consecutive set of two single exponentials. From the time scale, as well as from 190 parallel NMR monitoring, and the preparative procedures indicated before these processes correspond to 191 the reductive elimination from 5-IV-Br,F to II-Br,F followed by isomerisation to II′-Br,F. The follow up 192 final reaction to produce 7-II-Br,F could not be monitored due to the high temperature needed as well as 193 for its time scale. Table 2 collects the relevant kinetic and activation data derived from the plots shown 194 in Fig. 2a for the processes monitored, together with other relevant data for similar processes. The data 195 indicate that the mechanism operating for the full process perfectly parallels that found for the reactivity Br,F, requires a rather large activation enthalpy with practically no changes in entropy, which indicates a 202 transition state with a dominant breaking of the two Pt-C bonds, but keeping them organised by an 203 incipient C-C bond making. This is the behaviour expected for this type of general reductive elimination 204 reactions. As for the volume of activation (Fig. 2b), it is in line with a small compression, precisely due 205 to the new C-C bond being formed. As for the II-Br, F ⇄ II′-Br,F isomerisation reaction monitored, the 206 activation parameters agree perfectly well with those obtained for the already studied II The parallel study carried out on the 5-IV-Cl,F → 7-II-Cl,F process proved to be much more complex to 213 be monitored. As indicated in the previous section, 1H NMR monitoring of the process according to the 214 time-resolved changes obtained by UV-Vis indicated that the presence of a mixture of the four isomeric 215 forms plus the final species indicated in Scheme 5 is prevalent under all the reaction conditions. The 216 slowest process of the three step sequence observed was associated with the II-Cl,F ⇄ II′-Cl,F reaction, 217 as an increase of concentration of the II′-Cl,F form is observed at this time-scale by 1H NMR 218 monitoring; kinetics could be monitored with low methodological errors by UV-Vis. Contrarily, the 219 initial fast reductive elimination reaction proved to be the most complicated to determine kinetically due 220 to the low solubility of 5-IV-Cl,F in xylene at temperatures lower than 50 °C and the readiness of the 221 process (see Table 2). Given the fact that the outcome of the full process under the conditions studied is 222 the final 7-II-Cl,F the remaining step observed was associated with the oxidative II′-Cl,F → 7-II,Cl,F 223 reaction. The kinetic and thermal activation parameters determined for all these sets of reactions are also 224 indicated in Table 2 along with the results for the other relevant systems. Clearly the data agree very 225 well with those observed for the similar systems studied. It is thus clear that the relative ease of 226 formation of the final 7-II-X,F sevenmembered platinacycles is dictated by the presence of a X = Br or 227 X = Cl donor in the II′-X,F → 7-II,X,F reaction, while the II-X, Y ⇄ II′-X,Y isomerisation process does 228 not distinguish between the different X and Y donors on the platinum centre. 229

DFT calculations 231
In view of the data collected in Table 2 II′-X,H are lower in energy than their corresponding Z analogues (Table 3). Since the reaction leading to 238 the final products proceeds via the E isomeric form of the II′-X,H intermediates (vide infra), Z isomers 239 can be considered irrelevant to the reaction course. Furthermore, as indicated in the previous sections, 240 the distal C-H bond is too far away from the platinum(II) centre to be relevant for the oxidative addition 241 process. From the data in Table 3 it is clear that the energies of the II′-X,H intermediates are in all cases 242 lower than those of the II-X,H species, indicating that an isomerization process should be expected, as 243 observed experimentally. The isomerization transition state (TS_Isom), involving a three coordinated 244 platinum species with a dangling NMe2 group (Fig. S1 †), was also calculated and found to be around 245 140 kJ mol−1 above II-X,H (see Table 3). The geometry of the calculated TS_Isom involves a rather 246 late C-Pt-Nimine angle (180° (II-X,H) → 130° (TS_Isom) → 90° (II′-X, H)), in good agreement with 247 the kinetic activation data obtained experimentally. 248 Once the most stable II′-X,H intermediate is formed, two possible selective parallel pathways, leading to 249 the characterised metallacycles, are possible. The first one (Scheme 6, top) involves the oxidative 250 addition of the C-HA bond at the platinum followed by the reductive elimination of toluene producing 251 the five-membered platinacycle (5-II-X,H). The equivalent seven-membered platinacycle (7-II-X,H) 252 would be obtained in a similar fashion whenever the C-HB bond is activated at the metal (Scheme 6, 253 bottom). Given the fact that five-membered platinacycles are more stable than their seven-membered 254 counterparts (the calculated free energy difference being 31.8 (X = Br) and 33.4 (X = Cl) kJ mol−1, as 255 expected from simple standard considerations)17,19,20 the obtention of the larger seven-membered 256 platinacycle from the II′-Cl,H intermediate has to be due to kinetic preferences. product (see Scheme 6). In this case the energy requirements for oxidative addition (TS_CHA) and 265 reductive elimination (TS_RE1) for the smaller five-membered platinacycles are 119.6 and 141.5 kJ 266 mol−1 respectively, slightly lower than those found for the seven-membered product: 146.4 (TS_CHB) 267 and 145.1 (TS_RE2) kJ mol−1. It may be argued that the final products could also be obtained from the 268 II-X,H isomeric form, but higher barriers were obtained for these pathways both for X = Br and X = Cl. 269 Other possible pathways such as those involving the C-H activation on the tetracoordinated square 270 planar platinum centre of II′-X,H, or σ-CAM processes21 leading to the final products, were also 271 computed and found to be noncompetitive with the mechanism proposed here. 272 The results collected in Scheme 6, which are clearly in line with the experimental observations, have 273 been used to build a qualitative kinetic simulation model of product formation over time. For this 274 purpose, the relative free energy differences have been transformed into rate constants by using the 275 Eyring-Polanyi equation (i.e. k = (kbT/h) exp(−ΔG ‡/RT)), and the product evolution over time, from 276 II′-X,H, has been calculated (Fig. 3). As may be observed, at 139 °C the product distribution trend 277 matches the experimental observations: 5-II-Br,H is produced with preference to 7-II-Br,H from II′-278 Br,H, whereas the inverse (7-II-Cl,H preferably to 5-II-Cl,H) is observed for II′-Cl, H. Although the 279 time scale in Fig. 3 reasonably matches the values for X = Br (50% conversion after 24 h), for X = Cl 280 there is more than an order of magnitude difference. Nevertheless, in this high energy range, this 281 difference is easily overcome when the methodological errors involved in the DFT calculation (4-16 kJ 282 mol−1) are taken into account. 283 The validity of the mechanism in Scheme 6 has also been confirmed by its use in the formation of the 284 fluorinated compounds 7-II-Br,F and 7-II-Cl,F characterised in the present work, for which the 285 formation of the five-membered platinacycle 5-II-X,F is not possible. The calculated energy 286 requirements (Table S2 †), although very similar, are slightly lower for the X = Cl system, indicating that 287 the formation of 7-II-Cl,F should be definitively faster. In fact, the qualitative kinetic model indicates 288 that 7-II-Cl,F is obtained around four times faster than 7-II-Br,F, practically the same difference as 289 observed experimentally (Fig. S2 †). 290

CONCLUSIONS 292 293
In this work, the mechanism of formation of seven-membered platinacycles, in preference to the more 294 thermodynamically stable five-membered analogues, has been disclosed through combined kinetico-295 mechanistic and computational studies. Seven-membered platinacycles are formed as the kinetically 296 favoured products in a process which involves the reductive elimination from cyclometallated 297 platinum(IV) compounds [PtX(4-CH3C6H4)2(ArCHvNCH2CH2NMe2)] (X = Br, Cl), followed by 298 isomerization of the resulting platinum(II) compounds plus a final cyclometallation step. The results 299 indicated that the nature of the spectator halido ligand X (X = Br or Cl) is determinant in the 300 platinacycle size of reaction products. The presence of a bromido ligand slows down the formation of 301 seven-membered platinacycles in such a way that the formation of the five-membered analogue becomes 302 competitive unless the required metalation site is blocked with a fluoro substituent. Both kinetico-303 mechanistic and computational studies indicate that, contrary to previous suggestions, the isomerization 304 step is not significantly affected by the nature of the halido ligand. On the contrary, all data are 305 consistent with the fact that the final cyclometallation step is only dependent on the nature of the 306 spectator halido ligand and this step is responsible for the nature of the final products. Therefore, five-307 membered platinacycles are preferred for Br, while the presence of a Cl leads to the formation of 308 sevenmembered analogues. 309 EXPERIMENTAL 311 312

General procedures 313
Microanalyses were performed at the Centres Científics I Tecnològics (Universitat de Barcelona). 314 Electrospray mass spectra were performed at the Servei d'Espectrometria de Masses (Universitat de 315 Barcelona) using a LC/MSD-TOF spectrometer usingH2O-CH3CN 1 : 1 to introduce the sample. NMR 316 spectra were performed at the Unitat de RMN d′Alt Camp de la Universitat de Barcelona using a 317 Mercury the solution was allowed to stir at room temperature for 70 minutes. The mixture was dried over 327 Na2SO4, the solution was filtered, and the solvent was removed under vacuum to give the product.  Prismatic crystals were selected and intensity data were measured on a D8 Venture system equipped 405 with a multilayer monochromator and a Mo microfocus. The structure was solved using the Bruker 406 SHELXTL software package, and refined using SHELXL.24 All hydrogen atom positional parameters 407 were computed and refined using a riding model, with an isotropic temperature factor equal to 1.2 times 408 the equivalent temperature factor of the atom to which they are linked; further details are given in Table  409 4. 410

411
Kinetics 412 The kinetic profiles for the reactions were followed by UV-Vis spectroscopy in the full 700-300 nm 413 range on HP8452A or Cary50 instruments equipped with thermostated multicell transports. The 414 observed rate constants were derived from absorbance versus time traces at the wavelengths where a 415 maximum increase and/or decrease of absorbance were/was observed; alternatively the full spectral 416 time-resolved changes where used. For the reactions carried out at varying pressures, the previously 417 described pillbox cell and pressurising system25-28 were used and the final treatment of data was the 418 same as described before. The calculation of the observed rate constants from the absorbance versus 419 time monitoring of reactions, studied under first order concentration conditions, was carried out using 420 the SPECFIT or RecatLab software.29,30 The general kinetic technique is that previously 421 described.11,18,31 Table S1 † collects the kobs values for all the systems studied as a function of 422 starting complex, pressures and temperatures studied. All post-run fittings were carried out by using the 423 standard available commercial programs. text correspond to those obtained with the larger basis sets and can be found, along their relevant 439 thermochemical terms, in Table S3. † 440 The kinetic models have been constructed with the Copasi software49 using the deterministic (LSODA) 441 method with relative and absolute tolerance values of 10−6 and 10−12, respectively.  Table 2 Kinetic and thermal activation parameters for the two reaction steps observed for the reaction of 632 different 5-IV-X,Y leading to 7-II-X,Y according to Scheme 5 in xylene solution. * indicates a mixture 633 of II-X,Y and II'-X,Y isomers7 634 635 636 637 638 639