Dinuclear Au(I), Au(II) and Au(III) Complexes with (CF2 )n Chains: Insights into The Role of Aurophilic Interactions in the Au(I) Oxidation.

New dinuclear Au(I), Au(II) and Au(III) complexes containing (CF2 )n bridging chains were obtained. Metallomacrocycles [Au2 {μ-(CF2 )4 }{μ-diphosphine}] show an uncommon figure-eight structure, the helicity inversion barrier of which is influenced by aurophilic interactions and steric constraints imposed by the diphosphine. Halogenation of LAu(CF2 )4 AuL (L=PPh3 , PMe3 , (dppf)1/2 , (binap)1/2 ) gave [Au(II)]2 species, some of which display unprecedented folded structures with Au-Au bonds. Aurophilic interactions facilitate this oxidation process by preorganizing the starting [Au(I)]2 complexes and lowering its redox potential. The obtained [Au(II)]2 complexes undergo thermal or photochemical elimination of R3 PAuX to give Au(III) perfluorinated auracycles. Evidence of a radical mechanism for these decomposition reactions was obtained.

Introduction

The applications of gold compounds in catalysis,[ , , , , ] medicine,[ , , , ] and materials science[ , , ] have triggered an spectacular development of the Organometallic and Coordination Chemistry of gold during the last two decades.[ , , , ] One of the most captivating features of gold is aurophilicity, defined as the tendency of closed‐shell Au centers to approach at short distances. Abundant manifestations of this phenomenon have been found in the conformations and supramolecular arrays of gold compounds, mainly in crystal structures, but also in solution.[ , , , , ] Besides, these interactions influence the catalytic[ , , , , ] properties of gold complexes. Thus, complexes containing two AuI centers connected by a three‐atom bridging ligand show enhanced activity in gold‐catalyzed coupling reactions, compared to their mononuclear congeners.[ , , ] In these systems, the aurophilic interaction enables oxidative addition of YZ reagents to the AuI⋅⋅⋅AuI unit, to give [AuII‐AuII] intermediates,[ , ] which can rearrange to [AuI, AuIII] species (Scheme 1). Fast reductive elimination on the AuIII centers produces coupling products.[ , ]

Scheme 1

Sequence of oxidative addition and reductive elimination reactions leading to coupling products in dinuclear Au complexes.

AuII complexes are much less abundant than AuI or AuIII ones. Most reported examples present a Au−Au covalent bond, and are obtained by oxidative addition of halogens or similar oxidants to a AuI⋅⋅⋅AuI unit containing one or two short bridging ligands.[ , , ] Although Au−Au bond energies around 200 kJ/mol have been theoretically and experimentally determined, a very small number of stable [AuII‐AuII] complexes without supporting bridging ligands are known.[ , , , , , , ]

Despite the catalytic relevance of AuII complexes, its reactivity has been much less explored compared to that of AuI or AuIII derivatives.[ , , , ] Thermal rearrangement to mixed‐valent [AuIII, AuI] species has been commonly observed,[ , , , ] while photochemical disproportionation or reductive elimination[ , , , ] reactions have been rarely reported.

We have recently shown that complexes LAu(CF2)4AuL (L=PPh3 or PMe3) adopt a folded structure stabilized by an aurophilic interaction (Figure 1, A). The study of their conformational equilibria by NMR techniques provided clear evidence of the existence of aurophilic interactions in solution. Herein we report macrocyclic complexes containing two AuI centres bridged by (CF2)4 and diphosphine ligands (Figure 1, B), which show an interesting dynamic behaviour. In addition, we have studied the oxidation reactions of complexes LAu(CF2)4AuL (L=PPh3, PMe3, PCy3, IPr, (dppf)1/2, (binap)1/2). Unprecedented [AuII‐AuII] complexes containing a (CF2)4 chain as the only supporting bridging ligand (Figure 1, C), or with both (CF2)4 and diphosphine ligands (Figure 1, D) have been obtained. Reactivity and electrochemical studies suggest that aurophilic interactions are key for the formation of these dinuclear AuII species, which undergo unprecedented thermal or photochemical rearrangements involving the cleavage and formation of Au−C bonds.

Figure 1

Previously reported dinuclear AuI complexes (A). AuI (B) and AuII (C, D) complexes reported in this work.

Results and Discussion

Macrocyclic AuI complexes

The reaction of Ph3PAu(CF2)4AuPPh3 (1) with dppf or rac‐binap gave PPh3 and macrocyclic complexes 2 or 3, which were isolated as soluble orange or white solids, respectively (Scheme 2). In contrast, the analogous reactions with (CH2) (PPh2)2 (n=2, 4 or 5) gave slightly soluble solids with a [Au2(CF2)4(PPh2(CH2)nPPh2)] composition, as determined by elemental analysis. The solid with n=5 was partially soluble and showed characteristic 19F NMR signals associated to the Au(CF2)4Au chain, which were similar to those of 2, suggesting the presence of [Au2{μ‐(CF2)4}{μ‐PPh2(CH2)5PPh2}] (4) in the soluble fraction. Oligomeric species with larger x values, could be present in the insoluble fractions.

Scheme 2

Synthesis of macrocyclic dinuclear AuI complexes.

The crystal structure of 2 shows a macrocycle formed by two AuI centres bridged by a (CF2)4 chain and a dppf ligand (Figure 2). The ring is folded, adopting a figure‐eight shape, where the Au atoms approach at short distance. Two C−H⋅⋅⋅F interactions cooperate to stabilize this conformation. Similar Au⋅⋅⋅Au and C−H⋅⋅⋅F interactions have been observed in the crystal structure of 1.

Figure 2

ORTEP diagram (thermal ellipsoids set at 50 % probability) of the molecular structure of 2. H atoms were omitted except those involved in weak interactions (Relevant distances (Å): Au1⋅⋅⋅Au2 2.9799(2), H36‐F7 2.49, H42‐F2 2.38).

The molecule of 2 lacks any symmetry in its crystal structure, although the differences between both halves of the molecule are small. In agreement with this, at sufficiently low temperatures the 19F NMR spectrum shows seven doublets corresponding to eight inequivalent 19F nuclei, two of which are coincidentally isochronous. On increasing temperature these signals coalesce into four doublets, as expected for a folded ring with average C 2 symmetry (Figure 3). Further temperature increase produces the coalescence of the four doublets into two broad signals, due to fast inversion of the ring helicity in the NMR time scale. The rate constants for this inversion process were determined by line shape analysis of the 19F NMR spectrum at different temperatures. An activation enthalpy of 49.1 kJ/mol was calculated by Eyring analysis, which is significantly larger than the activation enthalpy determined for the helicity inversion of 1 (35.8 kJ/mol). This difference is rationalized by considering that inversion of 1 can easily take place after breaking the Au⋅⋅⋅Au and C−H⋅⋅⋅F interactions and formation of an extended species, while in 2 the distance between both Au centers is constrained by the limited conformational mobility of the macrocycle, raising the activation energy.

Figure 3

19F NMR spectra of 2 at selected temperatures (CDCl3, 564.6 MHz). Different 19F chemical environments are labelled as Fa, Fb, Fc, Fd. Interconversion between both conformers originates a/b and c/d exchange. Between 210 and 227 K the molecule adopts an unsymmetrical conformation (not shown).

Similarly, the variable‐temperature NMR spectra of 3 suggest that at low temperatures the macrocycle adopts an unsymmetrical conformation where all 19F and 31P nuclei are inequivalent (Figures S54–S56). As the temperature increases (T>211 K), fast conformational motion leads to an average C 2 symmetry. However, further temperature raise to 349 K did not produce coalescence of the observed signals, which is attributable to the intrinsic chirality of the binap ligand. Remarkably, only one diastereomer was observed despite racemic binap was used, which suggests that the configuration of the binaphthyl unit determines the helicity sense of the Au(CF2)4Au chain. The 19F NMR spectrum of 4 agrees with a cyclic structure of C 2 symmetry analogous to that of 2 (Figure S57).

AuII and AuIII complexes

Oxidation of 1 or Me3PAu(CF2)4AuPMe3 (5) with an equimolar amount of PhICl2 gave AuII complexes 6 or 7, respectively (Scheme 3), which were isolated as yellow light‐sensitive solids. Photodecomposition of 6 and 7 in solution was fast under irradiation at 402 nm, giving an almost equimolar mixture of R3PAuCl and [Au{κ2‐(CF2)4}Cl(PR3)] (R=Ph (8), Me (9)) in 2–5 minutes. Traces of octafluorocyclobutane were detected by NMR spectroscopy and MS. In the absence of light, conversion of 6 into Ph3PAuCl and 8 took more than 8 days at room temperature, or 1 h at 80 °C. Photodecomposition was also observed when solid 6 was exposed to ambient light.

Scheme 3

Oxidation of dinuclear AuI complexes and decomposition of the resulting AuII complexes.

The crystal structure of 6 shows an almost linear Cl−Au‐Au−Cl unit (Figure 4). The Au−Au distance shortens from 3.0394(3) Å in 1 to 2.5454(3) Å in 3, a value that is comparable to those found in the two reported perfluoroalkyl AuII complexes, [Au(CF3)2(py)]2 and [Au2(CF3)2{μ‐(CH2)2PPh2}] (Au−Au distances: 2.5062(9) and 2.679(1) Å, respectively). Owing to the reduction of the Au−Au distance, the P−Au‐Au−P dihedral angle is larger for 6 (96.177°) than for 1 (72.299°). The looped structure is chiral and both enantiomers are present in the unit cell. Although the molecule does not present any crystallographic symmetry element, its geometry is very close to that expected for an ideal C 2 symmetry.

Figure 4

ORTEP diagram (thermal ellipsoids set at 50 % probability, H atoms omitted) of the molecular structure of 6.

The 19F and 31P{1H} NMR spectra of 6 and 7 are very similar and suggest that the main features of the solid‐state structure of 6 are essentially maintained in solution for both complexes. Thus, the room temperature 19F NMR spectra show four sharp signals corresponding to a C 2‐symmetric species, indicating that the covalent Au−Au bond hampers the inversion of the loop. By contrast, the weak non‐covalent interactions that stabilize the folded conformation in 1 and 5 can be broken at relatively low temperatures.

The reactions of 1 or 5 with PhICl2 in a 1 : 2 molar ratio gave mainly complexes 6 or 7 instead of the expected dinuclear AuIII species resulting from dichlorination of each gold centre. The formation of the AuII complexes was fast, but subsequent reaction with PhICl2 was sluggish. Indeed, at long reaction times (24 h), unreacted PhICl2, 6 or 7, and their decomposition products (8 or 9 and LAuCl) were observed.

To get a closer insight into the influence of the auxiliary ligands in the stability of the AuII dinuclear complexes, we carried out the reactions of LAu(CF2)4AuL (L=PCy3 (10) or IPr (11)) with PhICl2. We have previously observed that bulky ligands disfavour the folded conformation of these dinuclear AuI complexes. Thus, considering the shorter Au−Au distance in the AuII complexes, we expected that an increase of the bulkiness of L should have a substantial effect in the outcome of these reactions. In line with this, no AuII complexes were detected in the reactions of 10 or 11 with an equimolar amount of PhICl2 (Scheme 4). In the reaction with 10, the main products were Cy3PAuCl and 12, likely formed by decomposition of the unstable dinuclear AuII intermediate A. Small amounts of AuIII complexes (13 and 14) were observed. By contrast, the reaction of 11 with PhICl2 led to complexes resulting mainly from dichlorination of one (15) or both (16) metal atoms, and complex 17. Remarkably, no [Au{κ2‐(CF2)4}Cl(IPr)] was detected in the reaction mixture. Compound 16 was the main product in the reaction of 11 with 2 equivalents of PhICl2. Small amounts of unidentified products were also formed in these reactions, accounting for less than 9 % of the total integral of the 19F NMR spectra of the mixtures.

Scheme 4

Reactions of AuI complexes containing bulky ligands with PhICl2.

The reaction of Ph3PAu(CF2)6AuPPh3 (18) with an equimolar amount of PhICl2 gave a mixture of AuI and AuIII species resulting from partial chlorination of the metal atoms and Au−C bonds (Scheme 5). No AuII complexes were detected. The 19F and 31P{1H} NMR spectra showed characteristic NMR signals for the Ph3PAuCF2‐ and Ph3PAu(Cl)2CF2‐ moieties, but an unambiguous identification of the components of the mixture was not possible because of extensive signal overlapping. The reaction of 18 with PhICl2 in a 1 : 2 molar ratio gave 19, which was isolated.

Scheme 5

Reaction of a AuI complex containing a (CF2)6 chain with PhICl2.

Reaction of 1 with XeF2 in CD2Cl2 gave the fluorocomplex 20 (Scheme 6). 20 is light sensitive and thermally unstable in solution at room temperature. Its decomposition produced an untractable mixture, where only [Au{κ2‐(CF2)4}F(PPh3)] (21) could be identified by NMR spectroscopy. Similarly, the reaction of 1 with Br2 gave the bromocomplex 22, which decomposed in the dark at room temperature to give [Au{κ2‐(CF2)4}Br(PPh3)] (23) and Ph3PAuBr. Owing to their instability, 20 and 22 could not be isolated. The reaction of 1 with I2 gave mainly Ph3PAuI and [Au{κ2‐(CF2)4}I(PPh3)].

Scheme 6

In situ formation of AuII fluoro‐ or bromo‐complexes.

Remarkably, complex 20 and [Au2F2(μ‐dfmt)2] (dfmt=N,N′‐bis(2,6‐dimethylphenyl)formamidinate), are the only AuII fluorocomplexes reported. The gold‐bound fluorine atoms in 20 were evidenced in its 19F NMR spectrum by a broad singlet at −186.1 ppm. The 19F NMR signals of the (CF2)4 chain and the 31P{1H} NMR signals of 20 and 22 were similar to those of 6 and 7, and agree with a similar structure of C 2 symmetry. Auracyclic complexes 21 and 23 were identified by means of their characteristic 19F and 31P{1H} NMR signals, which were compared with those of samples generated by halide substitution in [Au{κ2‐(CF2)4}I(PPh3)].

Oxidation of complexes 2 or 3 with an equimolar amount of PhICl2 gave AuII macrocyclic complexes 24 or 25 (Scheme 7), which were isolated as brown or yellow solids, respectively. Both are light‐sensitive in solution, but in the mixtures of decomposition products only the anionic complex [Au{κ2‐(CF2)4}Cl2]− could be identified by NMR spectroscopy. The oxidation of AuI dinuclear macrocyclic complexes to [AuII‐AuII] complexes has been previously observed in macrocyclic bis‐carbene complexes.[ , , ]

Scheme 7

Reactions of dinuclear macrocyclic complexes with PhICl2.

The crystal structure of 24 shows an almost linear Cl−Au‐Au−Cl unit (Figure 5), with a Au−Au distance (2.5274(3) Å) similar to that of 6 (2.5454(3) Å). The PAu(CF2)4AuP chain also adopts a looped conformation, although the PAuAuP dihedral angle (84.827°) is smaller than in 6 because of the conformational constraints imposed by the diphosphine ligand. Both enantiomers of 24 are present in the unit cell. The ferrocenediyl unit is not symmetrically disposed, making both halves of the molecule significantly different.

Figure 5

ORTEP diagram (thermal ellipsoids set at 50 % probability, H atoms omitted) of the molecular structure of 24.

As observed for the other dinuclear AuII complexes 6, 7, 20 and 22, the 19F NMR spectra of the macrocyclic AuII complexes 24 and 25 show four sharp doublets of multiplets corresponding to two diastereotopic CF2 units, in agreement with an average C2 symmetry. The δ(19F), and 2 J PF values are similar to those of the other members of the series.

Cyclic voltammetry

To gain information about the influence of the neutral ligands and the length of the (CF2)n chain on the outcome of the oxidation reactions of dinuclear AuI complexes, we carried out cyclic voltammetry measurements on complexes LAu(CF2)4AuL (L=PPh3 (1), PCy3 (10), IPr (11)) and LAu(CF2)6AuL (L=PPh3 (18), Cy3P). Mononuclear complexes LAu(n‐C4F9) (L=PPh3, IPr) were also studied for the purpose of comparison. The potential was scanned in the oxidative direction until solvent oxidation became dominant. Then, the scan direction was reversed toward negative values.

Complex 1 gave a well‐defined anodic wave at a peak potential of 1.13 V, followed by another ill‐defined wave around 1.5 V (Figure 6 and Table 1). Further anodic processes around 2 V were overlapped with the solvent oxidation wave. No corresponding reduction events were observed when the scan direction was reversed, which suggests that the observed oxidations are irreversible processes. The cyclic voltammogram of 10 was similar, showing a well‐defined oxidation wave at 1.33 V, followed another one at ca. 2.4 V. In contrast, 11 showed a weak shoulder around 1.6 V, followed by a well‐defined peak at 2.12 V. The oxidation of the other studied complexes occurred at significantly higher potential values. For instance, the first‐oxidation waves of LAu(n‐C4F9) (L=Ph3P, IPr) were observed at 2.09 and 2 V, respectively, while dinuclear complexes LAu(CF2)6AuL (L=PPh3 (18) or Cy3P) did not show any detectable oxidation event at potentials lower than 2 V. In summary, complexes 1 and 10 show significant oxidation events at potentials lower than those of dinuclear complexes with longer (CF2) chains or bulkier L ligands, whose oxidation potentials are similar to those of mononuclear complexes LAu(n‐C4F9) (L=PPh3, IPr).

Figure 6

Cyclic voltammograms of: (top) (CF2)4‐bridged; (bottom) (CF2)6‐bridged and mononuclear AuI complexes. Recorded in 0.001 M CH2Cl2 solutions, with 0.1 M NBu4PF6 as electrolyte and a 0.1 V/s scan rate.

Table 1

Electrochemical data.

Complex	Ep [V][a]
Ph3PAu(CF2)4AuPPh3 (1)	1.13, 1.5
Cy3PAu(CF2)4AuPCy3 (10)	1.33, ca. 2.4
IPrAu(CF2)4AuIPr (11)	ca. 1.6[b], 2.12
Ph3PAu(CF2)6AuPPh3 (18)	>2.0
Cy3PAu(CF2)6AuPCy3	2.5
Ph3PAu(n‐C4F9)	2.09[c]
IPrAu(n‐C4F9)	2.00

[a] Vs FeCp2 +/FeCp2. [b] Weak shoulder. [c] The observed shoulder appeared as a peak after subtraction of a blank voltammogram (see Figure S61).

Complexes 1 and 10 share in common their ability to adopt a folded structure stabilized by an aurophilic interaction. Thus, at room temperature 1 is mainly in the folded conformation and 10 is mainly in the extended one, although a significant amount of ‐10 (23 %) in exchange with ‐10 was observed at low temperature (Scheme 4). In contrast, 11, 18 and Cy3PAu(CF2)6AuPCy3 display extended conformations in their crystal structures and their NMR spectra do not show any evidence of stable folded conformers. This suggests that short Au⋅⋅⋅Au contacts could be responsible for the observed decrease in the first‐oxidation potentials of 1 and 10.

Cyclic voltammetry studies in dinuclear AuI complexes (μ‐XZ2)Au2Cl2 (X=CH2, Z=PPh2 or 3‐mesitylimidazol‐1‐yl‐2‐ylidene; X=N(i‐Pr), Z=PPh2 ) have shown that the oxidation potentials of these complexes were 0.12‐0.32 V lower than those of structurally similar mononuclear complexes. We have observed larger differences (ΔE ≥ 0.9 V) between 1 and Ph3PAu(n‐C4F9) or Ph3PAu(CF2)6AuPPh3, or between 10 and Cy3PAu(CF2)6AuPCy3. These larger differences can be attributed to a destabilization of the HOMO of the folded conformers of 1 and 10 induced by the stronger aurophilic interactions.

Mechanistic considerations

Reaction pathways leading to the observed products are outlined in Scheme 8. Chlorination of the initial dinuclear AuI complex would give a dinuclear AuII complex (B) or a mixed‐valent complex (C). Products of the type B were formed when the starting complex has a (CF2)4 chain and PPh3, PMe3, dppf or binap as neutral ligands. In contrast, complexes of the type C are formed if the gold centers are bridged by a longer chain, as in 18, or if L is a bulky ligand, as in 11. In these cases, C undergoes further chlorination to form dinuclear AuIII complexes (D). Reductive elimination with formation of a C−Cl bond from the AuIII centers would afford LAuCl and products of the type E. Complexes B evolve by photo‐ or thermally activated elimination of LAuCl to give F. Remarkably, disproportionation of B to give C was not observed, whereas oxidation of complexes B with PhICl2 was sluggish (observed for 6 and 7).

Scheme 8

Reactions leading to AuII, AuIII and mixed‐valent complexes.

An interesting case is the reaction of 10 with one equivalent of PhICl2, which gives mainly Cy3PAuCl and [Au{κ2‐(CF2)4}Cl(PCy3)] (12). Since complexes of the types LAuCl and F are typically formed by decomposition of dinuclear AuII complexes, the main reaction product should be a complex of the type B. The steric repulsions between the bulky PCy3 ligands could render this intermediate unstable, provoking fast decomposition to Cy3PAuCl and 12. Small amounts of products of the type C or D, resulting from dichlorination of the Au centres, were observed in this reaction.

We hypothesize that the close approach of the Au centers in the starting dinuclear AuI complexes is beneficial for the formation of complexes B in two ways. Firstly, it lowers the redox potential at which the first oxidation occurs, as observed for 1 and 10. Secondly, it would stabilize the initially formed [AuII, AuI] species, which could undergo a further one‐electron oxidation to give [AuII‐AuII] complexes (B). If the separation of the Au atoms is large, the [AuII, AuI] intermediates would be oxidized to [AuIII, AuI] species and then to [AuIII, AuIII] complexes.

The fast photodecomposition of complexes B and the detection of octafluorocyclobutane are indicative of a radical reaction. In agreement with this, the yield of the decomposition of 6 was reduced in the presence of TEMPO from 100 % to 77 % (photochemical) or 65 % (thermal), and the radical‐trapping diadduct (TEMPO)2(μ‐C4F8) was observed (6 % in the photodecomposition, traces in the thermal decomposition). This suggests that homolysis of the Au‐CF2 bonds takes place during the decomposition of 6. To test if the observed octafluorocyclobutane could be formed by decomposition of a complex of the type F, [Au{κ2‐(CF2)4}I(PPh3)] was heated (100 °C, toluene) or irradiated (402 or 310 nm, CD2Cl2). However, no octafluorocyclobutane was detected in these experiments.

Theoretical studies on binuclear AuII complexes suggest that the LUMO of these complexes has Au−Au and Au‐halogen antibonding character.[ , , , ] In line with this, photoexcitation of 6 could lead to weakening of the Au−Au bond facilitating its homolysis to give radical intermediates (Scheme 9), which would eliminate a molecule of Ph3PAuCl. The resulting diradical (G), could undergo an intramolecular coupling, to give auracycle 8. This would constitute the main decomposition pathway, but G could also give Ph3PAuCl and C4F8. Diadduct (TEMPO)2(μ‐(CF2)4) would be produced in the presence of TEMPO.

Scheme 9

Proposed decomposition mechanism for 6.

Conclusions

We have reported the first macrocyclic complexes containing two AuI centers linked by a perfluorocarbon chain, which adopt a rare figure‐eight conformation. The higher energetic barrier for exchange between both enantiomeric conformers of [Au2{μ‐(CF2)4}(μ‐diphosphine)] (diphosphine=dppf (2), binap (3)) respect to the acyclic analogues R3PAu(CF2)4AuPR3 (R=Ph (1) or Me (5)) is originated by conformational constraints imposed by the presence of the diphosphine ligands.

Complexes 1, 2, 3 or 5 react with halogenating agents to give dinuclear AuII complexes which present unprecedented structures. In the acyclic complexes 6 and 7, the bridging (CF2)4 chain allows the approach of both Au centres, but does not impede their separation at relatively long distances. Therefore, they constitute a new type of AuII complexes having semi‐supported AuII‐AuII bonds.

CV studies suggest a marked lowering of the oxidation potential of the dinuclear [AuI, AuI] complexes induced by the presence of aurophilic interactions. This effect, together with the presence of a (CF2)4 linker between the AuI centres and the low steric hindrance of the neutral ligands, is beneficial for the formation and stability of the [AuII‐AuII] complexes. In contrast, the presence of a longer (CF2)6 linker or bulky neutral ligands in the starting [AuI, AuI] complexes induce the formation of [AuIII, AuI] and [AuIII, AuIII] complexes.

Thermal or photochemical decomposition of complexes [(AuX)2{μ‐(CF2)4}(PR3)2] gives rise to elimination of R3PAuX and formation of AuIII complexes which are new members of a rare family of perfluorinated auracycles. Evidence of a radical mechanism for these decomposition reactions has been obtained.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.