Au⋅⋅⋅H−C Hydrogen Bonds as Design Principle in Gold(I) Catalysis

Abstract Secondary ligand–metal interactions are decisive in many catalytic transformations. While arene–gold interactions have repeatedly been reported as critical structural feature in many high‐performance gold catalysts, we herein report that these interactions can also be replaced by Au⋅⋅⋅H−C hydrogen bonds without suffering any reduction in catalytic performance. Systematic experimental and computational studies on a series of ylide‐substituted phosphines featuring either a PPh3 (PhYPhos) or PCy3 (CyYPhos) moiety showed that the arene‐gold interaction in the aryl‐substituted compounds is efficiently compensated by the formation of Au⋅⋅⋅H−C hydrogen bonds. The strongest interaction is found with the C−H moiety next to the onium center, which due to the polarization results in remarkably strong interactions with the shortest Au⋅⋅⋅H−C hydrogen bonds reported to date. Calorimetric studies on the formation of the gold complexes further confirmed that the PhYPhos and CyYPhos ligands form similarly stable complexes. Consequently, both ligands showed the same catalytic performance in the hydroamination, hydrophenoxylation and hydrocarboxylation of alkynes, thus demonstrating that Au⋅⋅⋅H−C hydrogen bonds are equally suited for the generation of highly effective gold catalysts than gold‐arene interactions. The generality of this observation was confirmed by a comparative study between a biaryl phosphine ligand and its cyclohexyl‐substituted derivative, which again showed identical catalytic performance. These observations clearly support Au⋅⋅⋅H−C hydrogen bonds as fundamental secondary interactions in gold catalysts, thus further increasing the number of design elements that can be used for future catalyst construction.


Introduction
Gold catalysis has undergone arapid development in the past two decades. [1]As is the case for numerous other metalcatalysed transformations,t his success story is oftentimes associated with the development of ligands and the tailoring of their properties to meet the specific requirements of the metal and targeted reaction.In gold(I) catalysis,acationic LAu + complex is usually the catalytically active species. [2]In order to stabilize these species and to generate highly active and efficient catalysts,s econdary ligand metal interactions have been reported as being important in ligand design and have fuelled numerous advances in the field of homogenous catalysis.Particularly gold-arene interactions have repeatedly been described to being beneficial in gold(I) catalysis. [3]This was,for example,illustrated in the Buchwald biarylphosphine ligands (Figure 1), in which the lateral arene ring is involved in bonding to the metal center. [4]This design principle was adopted in several other phosphines and N-heterocyclic carbenes, [5] such as in the imidazo [1,5-a]pyridin-3-ylidene platform first described by Lassaletta and Glorius [6] or in Alcarazos N-arylpiperidinophosphines. [7]ecently,wereported on transition metal catalysts based on ylide-functionalized phosphines (YPhos). [8]In gold(I)catalyzed transformations with moderately (A) [8a] as well as highly electron-rich (B and C) [9] YPhos systems exceptionally high turnover numbers were observed.All of these YPhos catalysts so far have relied on triphenyl phosphonium groups which likewise foster arene-gold interactions,t hus contributing to the stability and high catalytic performance of the corresponding LAu(I) + species.
Theu nderstanding of such secondary ligand metal interactions is important for future ligand design and for the development of high-performance catalysts.T hus,w ew ondered if these arene-gold interactions were indeed necessary or if equally high activities can also be achieved without such stabilizing interactions.I fa rene-gold interactions could be omitted or replaced by other interactions,t his would lead to ap aradigm shift in ligand design and would significantly broaden the structural scope and thus facilitate the synthesis of efficient ligands in the future.W ee nvisioned that the YPhos ligands would be an ideal ligand platform to systematically address the importance of arene-gold interactions for catalysis.T he PPh 3 moiety in the YPhos ligands can easily be replaced by at ricyclohexyl phosphonium group ( Cy YPhos), thus preventing the interaction between the metal and the phenyl group without changing the overall ligand architecture.H erein, we show that indeed the often-invoked arenegold interactions are not necessary,b ut can be replaced by stabilizing hydrogen bonds,w hich are equally suited in generating highly active catalysts.

Ligand Synthesis
To probe the importance of supporting interactions between gold and the phosphonium moiety,w ep resent ad etailed study of the performance of the three PCy 3substituted ligands Cy Y S PCy 2 (L1), Cy Y oTol PCy 2 (L2)a nd Cy Y Mes PCy 2 (L3)a sc ongeners to A, B and C.L igands L2 and L3 have previously been designed for the selective Pdcatalysed monoarylation of small primary amines. [10]Additionally,t he iso-propyl-derivative of L1, Cy Y S PiPr 2 (L4)w as synthesized to evaluate the influence of the lower steric bulk of the smaller alkyl group on the catalytic ability.L1 and L4 were prepared on gram scale as white solids by reaction of the metalated ylide Cy Y S -Li with Cy 2 PCl and iPr 2 PCl in good yields of 75 %and 70 %, respectively. [11]1 features two sets of doublets in the 31 P{ 1 H}-NMR spectrum at 31.7 and À7.3 ppm, with ac oupling constant of 2 J PP = 106.8Hz, while L4 displays resonances at 31.5 and 1.5 ppm with as lightly smaller coupling constant of 2 J PP = 105.7 Hz.In addition to the obtained new ligands, L1 and L4, we targeted ac ompletely arene-free Cy YPhos derivative,t o eliminate any possible metal-arene interaction.Therefore,we tackled the synthesis of Cy Y SF -PCy 2 (L5), an analogue of 1 in which the p-tolyl motif is replaced by aperfluorobutyl chain.To access this ligand, the metalated ylide Cy Y SF -Li first had to be synthesized.By analogy to the procedure reported for Cy Y S -Li, [11] the protonated precursor Cy Y SF -H was prepared in ao ne-pot reaction from the easily accessible phosphonium salt [Cy 3 P-CH 3 ]I and the commercially available perfluorobutanesulfonyl fluoride in the presence of two equiv of KHMDS in av ery good yield of 83 %( Scheme 2).Cy Y SF -H was isolated as ap ale-yellow powder and fully characterized (see SI for details).
Deprotonation of Cy Y SF -H with n-butyllithium afforded the metalated ylide Cy Y SF -Li which was used in situ and directly reacted with PCy 2 Cl to yield Cy Y SF -PCy 2 (L5)a s ac olorless solid in 72 %y ield.With d P = 32.3 and À0.7 ppm, its 31 P{ 1 H} NMR signals are in the same range as observed for L1 and L4,while with 2 J PP = 95.0Hz, the coupling constant is significantly smaller.I mportant NMR spectroscopic and crystal structure parameters are given in Table 1.XRD analyses confirm the expected connectivity and show the typical arrangement of PCy 3 substituted YPhos ligands,were the alkyl groups attached to the phosphorus atom point away from the phosphonium group to minimize the steric pressure. [12]heme 1. Synthesis of Cy Y S -PR 2 by reaction of the metalated ylide Cy Y S -Li with dialkyl chlorophosphines (PR 2 Cl with R = iPr or Cy).To evaluate the ligand electronic properties,wecalculated the Tolman electronic parameter (TEP) by analyzing the carbonyl stretching frequencies of the corresponding [Rh-( Cy YPhos)(acac)(CO)] complexes (Table 2).As expected, Cy Y S -PCy 2 (L1), with aT EP calc of 2057.0 cm À1 ,i sm ore electron-releasing than its isopropyl-substituted congener L4 (TEP calc 2058.7 cm À1 ).Furthermore,t he stabilizing effect of the sulfonyl moiety becomes evident, by comparison with aryl substituted YPhos ligands L2 and L3,w hich are significantly more electron-donating than L1 and L4.I nterestingly, Cy Y SF -PCy 2 (L5)with aTEP calc value of 2059.8 cm À1 is even less electron-releasing than the simple alkyl phosphine PCy 3 (TEP calc 2058.1 cm À1 ).This result clearly demonstrates the further increased electron-withdrawing nature of the perfluorobutylsulfonyl moiety and thus shows how easily the electronics of YPhos ligands can be tuned via backbone modification.

Synthesis and Structures of the Gold Complexes
With the novel ligands in hand, we next prepared [ which matches well with the experimental observations.Such ad ownfield shift has also recently been reported for other Au•••H À Ci nteractions,b ut has controversially been discussed. [14]In the case of complex P1,o nly ab roadening and no splitting of the PCH signal was observed thus suggesting aweaker gold hydrogen interaction.Thee xistence of Au•••HÀXh ydrogen bonds has controversially been discussed in the literature, [15,16] but were experimentally and computationally proven in recent studies. [17]Fore xample,t he groups of Bourissou, [18] Berger and Monkowius [19] as well as Ruliz ˇek [20] demonstrated that N + -H ammonium or pyridinium groups are suitable donors for the formation of NÀH•••Auhydrogen bonds.Other strong donors such as O À H, F À H, NH 3 and HCN were also found to form close-contact interactions with gold. [21]Au•••H À Cinteractions have very recently been reported, but have never been discussed as structural motif for ligand design in catalysis. [22]espite numerous early discussions,itisnow well established that the Au atom in these Au•••HÀXi nteractions acts as an electron donor and the XÀHm oiety as an acceptor,w hich is in contrast to "classical" agostic interactions which rely on the donation of electron density into an empty orbital at the metal center. [23]Accordingly,t hese interactions were named anagostic interactions or "contra-electrostatic" hydrogen bonds. [24]Thel ater term emphasises the relation to classical H-bonds,w hich however differ in the charge of the Ca nd H  1958.7 2058.1 Ph Y S PCy 2 (A) [b] 1953.5 2055.1 Ph Y oTol PCy 2 (B) [c] 1947.5 2051.7 Cy Y oTol PCy 2 (L2) [d] 1941.atom upon approximation of the donor (gold versus the Hbond acceptor).24a] Further proof of the Au•••HÀCinteractions in the YPhosgold complexes is manifested in their molecular structures (Figure 3).Indeed, with 2.39(4) and 2.38(5) in P1 and P5 respectively,the Au•••H À Cdistances are remarkably short and significantly shorter than their sum of Va nder Waals radii (2.86 ) [25] as well as hydrogen bonds previously reported. [22]16a] To the best of our knowledge the Au-H9 distance of 2.38(5) in P5 is the shortest distance reported to date for aA u•••HÀCh ydrogen bond. [26]Presumably,t his interaction is stronger than those found with other CÀHm oieties due to the stronger polarization of this entity next the positively charged phosphorus centre.
Of course,one could argue that the interaction might also be caused by steric pressure within the molecule.H owever, the latter should be less critical in the structure of the complex with iPr-substituted phosphine L4.H ere,t he smaller alkyl groups should give rise to amore flexible ylide structure and allow aw idening of the P-C-P angle,w hich was found to be decisive for the approach of the phosphonium group to the metal center and hence for secondary metal ligand interactions. [27]Unfortunately,n oc rystals of P4 of sufficient quality could be obtained to allow for the direct location of the hydrogen atom in the electron density map.H owever,t he observed AuÀCd istance clearly indicates that also short interactions between gold and the PCy 3 unit are present in P4 (Table 3).It is interesting to note,t hat P2 and P3 with aryl groups in the ylide-backbone showed considerably longer C À H•••Aud istances.N onetheless,t hey are shorter than the sum of Va n-der Waals radii, thus suggesting that weak secondary ligand metal interactions are still present in these complexes.
These differences in the C À H•••Audistances can be explained by the different bulk of the sulfonyl and the aryl groups.Whereas the flexible sulfonyl group allows the complexes P1, P4 and P5 to adopt the preferred geometry with aplanar Au-P-C-P unit, [28] the rigid tolyl and mesityl substituents enforce ad eformation, which results in the rotation of the PCy 2 moiety and hence the coordination of AuCl "outside" the center of the pocket formed by the PCy 3 and PCy 2 units (see Figure S36).This also results in as lightly reduced steric pressure of the aryl-substituted ligands directed towards the metal center as measured by the percent buried volume (V bur %) (Table 3). [29]While all ligands are highly sterically demanding,covering more than half of the sphere defined in the model, it is slightly lower for the YPhos ligands L2 and L3 with an aryl group in the ylidic backbone.With 54.8 V bur % L1 is slightly bulkier than L4 (53.2 V bur %) and L5 (52.5 V bur %), but similar demanding than the PPh 3 substituted analogue A. This comparison between L2 and L3 and the sulfonylsubstituted ligands shows that the hydrogen bond is easily affected by steric effects.

Calorimetric Studies
In order to gain further insights into the relative stability of cyclohexyl-vs.1kcal mol À1 ,b oth ligands possess similar binding energies to the metal center.Thee nthalpy values,w ew ill emphasize, reflect all interactions with the metal (s,p and secondary interactions).TheT EP value for A,2 055.1 cm À1 ,( vs. 2057.0cm À1 for L1)i ndicates that A is ab etter donor ligand just on an electronic basis.T he infrared data suggest that more negative enthalpies of ligand substitution should be expected for the reaction in Scheme 4i nvolving A,a nd experimental results validate this expectation.

Catalytic Performance
Next, we turned our attention to the catalytic activity of the [Au( Cy YPhos)Cl] complexes.W es elected the hydroamination of phenylacetylene with aniline as first test reaction since this would permit ac omparison with the previously reported PPh 3 -substituted YPhos ligands.F urthermore,t he effectiveness of gold catalysts in this reaction is severely influenced by the stability of the cationic gold species towards reduction to gold(0), which in turn is affected by secondary metal-ligand interactions,t hus making it an ideal test reaction. [30]To this end, the [Au(YPhos)Cl] complexes of A, B and C as well as their direct PCy Additionally,w ea lso tested P4 to investigate the effect of the size of the alkyl groups attached to the phosphorus (III) atom on the catalytic activity.Furthermore,w ea lso examined complex [Au( Cy Y SF PCy 2 )Cl] (P5)w ith the perfluorinated "aryl-free" ligand L5,w hich excludes the presence of any arene-gold interactions and hence gives direct information about the importance of these secondary interactions in the YPhos ligands.
Figure 4s hows the conversion time plots of the comparison between the Ph YPhos and Cy YPhos ligands.N ote that without the addition of NaBAr F no conversion was observed.Thesame holds true for using NaBAr F without addition of any gold complex.To our delight, the cyclohexyl-substituted complexes proved to be highly active catalysts (Figure 4, left), which performed considerably better than simple PPh 3 or PCy 3 (see the SI for further details).Most importantly,t hey were equally efficient as their PPh 3 -substituted analogues, (e.g.A and P1)o ro nly slightly less active (B/C vs. P2/P3).Overall P1 and P3 performed superbly,leading to almost full conversion to the imine after approx.5h of reaction time.Only the tolyl-substituted catalyst is slightly less effective.Changing the phosphine alkyl groups from cyclohexyl in P1 to isopropyl in P4,l ed only to am oderate drop in activity (Figure 4, right).Most interestingly,t he completely arenefree Cy YPhos complex P5 also gave nearly full conversion after 24 h, thus demonstrating that the absence of arene-metal interaction also leads to highly active gold catalysts.T he slightly lower activity of P5 compared to its tosylate analogue P1 can be attributed to the more electron-withdrawing nature of the C 4 F 9 -substituents.M ost importantly,t he PCy 3 -substituted YPhos ligands also preserved their high activity at lower catalyst loadings of only 0.05 mol %( Figure 4, right, dashed lines).Again, P1 performed equally well compared to A.This is an important finding,since ligand design has so far focussed on the introduction of aryl substituents to incorporate arene gold interactions for stabilizing the catalytically active Au I+ species.O ur results clearly demonstrate that the active species with Ph YPhos and Cy YPhos are equally active thus suggesting that Au•••HÀCi nteractions are equally suited for the generation of highly active gold catalysts,w hich even operate at very low catalyst loadings.T his observation points to new possibilities for ligand design in gold catalysis.To further prove the comparable performance of Ph YPhos and Cy YPhos we synthesized well-defined, cationic digold complexes bearing these ligands and examined their activity in the hydrophenoxylation and in the hydrocarboxylation of diphenylacetylene.I no rder to gain access to digold hydroxides,w ec hose to follow an ew route which utilizes Au-aryl complexes as precursors to cationic complexes (Scheme 5). [31]herefore,a fter synthesizing the corresponding Au-aryl complexes 1a and 1b,a ddition of acid to their acetonitrile suspensions,f ollowed by evaporation of the solvent and extractions with DCM/H 2 Oled to the desired digold hydroxide complexes 2a and 2b bearing the two YPhos ligands. [32,33]hec omplexes 1a, 1b and 2b were also characterized by XRD analysis.Most interestingly, 1balso exhibited short AuÀ Hinteractions (< 2.5 ), [34] thus demonstrating that hydrogen bonds are not limited to the AuCl complexes.
Thec omplexes were evaluated in terms of their catalytic activity in the hydrophenoxylation and hydrocarboxylation of diphenylacetylene (Scheme 6). [35,36] n the hydrophenoxylation to 3,b oth complexes displayed lower activity than the state of the art digold complex bearing the IPr ligand (IPr = N,N'-bis[2,6-(di-isopropyl)phenyl]imidazol-2-ylidene). [31]owever,their overall performance was essentially the same, irrespective of the structure of the R Y S Phos (R = Ph or Cy) ligand.In hydrocarboxylation to 4,t heir activity was higher when compared to that in hydrophenoxylation. [31]Again, the same trend is observed with both complexes displaying comparable catalytic activities.I ts eems that the bulky, electron-donating YPhos ligands can participate in dual gold catalysis as well, albeit leading to decreased activity in comparison with IPr.This is still significant, considering that both the ligand and the counterion (and their specific combination) are targets for optimization as they markedly affect the outcomes of these reactions.Ofn ote,a ctivation of the corresponding AuCl complexes with NaBAr F did not lead to product formation in either reaction shown in Scheme 6, under various conditions (inert atmosphere,u nder air, premixing of the gold complex and the chloride abstractor).Synthesis and evaluation of other well-defined, cationic complexes of this kind will be pursued further.

Computational Studies
To further evaluate the stability and the nature of the secondary interactions between gold and the different Cy YPhos ligands we performed computational studies on the PW6B95D3/def2tzvp (MWB60 for Au)l evel of theory.W e were particularly interested in answering the following questions:i )A re C À H••Aui nteractions the most favored interactions or can the aryl or the sulfonyl groups in the ylide backbone of the Cy Y S -substituted YPhos ligands also bind and thus stabilize the metal center?ii) Are the short contacts between the CH protons of the PCy 3 moiety and the gold atom in P1 and P5 observed by XRD analysis present in the gas phase and solution structures and iii)what is the nature of these C(sp 3 )-H•••Aui nteractions?
Geometry optimization, conformers.T oc onfirm that the conformer of P1 observed in the crystal structure (C1)isalso the preferred conformation in the cationic gold complexes as eries of structures were optimised (Table 4).Local energy minima were found for the conformers C2,inwhich the ylidic substituent is rotated by % 1608 8 about the P2 À C1 bond and the gold atom is coordinated by the sulfonyl group,a nd C3,i n which the ylidic substituent is rotated by % 1808 8 and an arenegold interaction can be observed.Energy optimization showed that conformer C1 with the experimentally observed C(sp 3 )ÀH•••Aui nteraction is thermodynamically preferred over the structures exhibiting an S=O••Au( DDG = 72 kJ mol À1 )o ra rene-Aui nteraction (47.6 kJ mol À1 ).This preference is even more pronounced in P5 with the perfluorobutyl group.Itisnoteworthy that also for P3 with the mesityl substituent in the ylide backbone the conformer with the PCy 3 moiety oriented towards gold is preferred over a C3 analogue with am esityl-Aui nteraction (DDG = 65 kJ mol À1 ,s ee SI).

Hence,t he calculations clearly confirm the favourable C
Interestingly,t he preference of the hydrogen bonded conformer C1 is also observed for the cationic gold complexes,w hich are more important in the catalysis.T his also holds true for ac ationic gold complex with additional coordination of phenylacetylene (pa) including aniline as solvent (PCM model), which reflects the reaction conditions during catalysis.A ss hown in Table 4t he preference of the hydrogen bonded conformer is still significant for these complexes [LAu(pa)] + ,a lbeit being slightly less pronounced than in the neutral LAuCl complexes due to the decreased electron density at gold and the resulting weaker electron donation from Au to the C À Hbond (vide infra).To obtain an estimate for the strength of the CÀH••Aui nteractions we calculated aconformer of C1 in which the cyclohexyl group is rotated about the PÀCb ond to prevent any CÀH••Au interaction.This conformer revealed to be energetically disfavored over C1 by 65 kJ mol À1 . [37]It is also noteworthy, that all attempts to replace the C À H••Aui nteraction by explicit coordination of aniline failed, always resulting in the dissociation of the amine during energy optimization.This clearly underpins the favorable hydrogen bonding.
Bonding analysis.Having established that the C(sp 3 )À H••Aui nteractions give rise to the thermodynamically most favoured structures,w en ext turned our attention towards studying the nature of this bonding interaction.To this end, natural bond orbital (NBO), quantum theory of atoms-inmolecules (QTAIM) and noncovalent interaction (NCI) analyses on the gold complexes P1 and A were performed (including solvent effects:P CM for aniline).Overall, the computational studies in unison confirm the presence of attractive C À Hg old interactions.T he QTAIM studies show several bond critical points (BCP) in each of the gold complexes between the Au and the hydrogen atoms of CH and CH 2 in the PCy 3 and PCy 2 moieties.T he BCPsw ith the highest electron density 1(r)and Laplacian r 2 1(r)are always found between the Au atom and the PCH group in the PCy 3 moiety and range between 1(r) = 0.015-0.024ebohr À3 and r 2 1(r) = 0.0432-0.0646ebohr À5 (Table 5). [38]These values are smaller than those reported for an Au•••HÀNhydrogen bond (1(r) = 0.033 ebohr À3 and r 2 1(r) = 0.08 ebohr À5 ), [20] but higher than those of the Au•••HÀChydrogen bond in agold cluster with ap henylene-bridged diphosphine ligand (1(r) = 0.016 ebohr À3 and r 2 1(r) = 0.037 ebohr À5 ). [22]TheN CI plots show ablue isosurface around these BCPs, indicating astrong attractive interaction (Figure 5).Furthermore,the calculated CÀHb onds involved in the hydrogen bond are slightly elongated compared to the non-interacting CÀHbonds of the PCy 3 moiety (e.g.f rom 1.0949 to 1.1004 in P1).NBO analysis also shows ahigher occupancyofthe s*(C À H) orbital of 0.030 et o0 .040ew hile the occupancyo ft he binding s orbital is not significantly decreased.Furthermore,the NBO analysis (second-order perturbation theory) revealed an orbital contribution to the Au•••HÀCh ydrogen bond, which consists of three donor-acceptor interactions between occupied orbitals at gold (two d-orbitals as well as s Au-P )a nd the s*(C À H) orbital.These contributions amount to approx.DE(2) = 14.4 kJ mol À1 in P1,which is weaker than the Au•••H À Nhydrogen bond reported by Bourissou and co-workers,but significant (see chapter 4.4 in the SI).Overall, the data clearly AE 0.0 + 61.5 [b] + 82.9 [c] L5-Au(pa) AE 0.0 [a] + 68.8 [a,b] + 105.8 [a,c] [a] including solvent effects using apolarizable continuum model with aniline as solvent.
[c] 908 8 rotation around P2-C1.No additionals econdary ligand gold interaction.argue for weak hydrogen bonds with the Au atom acting as electron donor and the C À Hm oiety as an acceptor. [24]Most interestingly,t he calculations indicate that the strongest Au•••H À Ci nteraction is present in complex P5 (e.g.DE(2) = 18.8 kJ mol À1 )s ee SI for details), thus being in line with the data obtained from XRD and VT NMR studies (see above).This corroborates with the calculated positive charges at the phosphorus atom in the phosphonium group,which is slightly higher in P5 (q P =+1.63) than in P1 (q P =+1.62).
In the past, the observation of bond critical points and deshielding effects have been controversially discussed and also explained by steric compression within acomplex rather than Au-H hydrogen bonding. [14,39] o further elaborate on this question, we calculated the C À Hv ibration in P5.While steric compression should result in as tiffer C À Hv ibration, hydrogen bonding should weaken and thus soften the C À H vibrational mode.I ndeed, calculations show the smallest value of the v(CÀH) vibrations in the PCy 3 moiety for the CÀ Hbond interacting with the gold center.This vibration in P5 is red-shifted by 82 cm À1 compared to the free ligand L5,thus supporting the presence of aweak hydrogen bond.This is also consistent with the variation in the C À Hb ond length upon complexation, with as mall elongation, 0.006 ,o nly for the PCÀHb ond involved in the Au•••HÀCi nteraction (see Table 34  We further analysed the corresponding cationic gold complexes with phenylacetylene as additional ligand, which are considered to be the catalytically active species.M ost importantly,i tw as found that the C(sp 3 )H•••Aui nteractions still persist, albeit with slightly lower values for 1(r), r 2 1(r) and alower occupancyofthe CÀH s*orbitals.T his is well in line with the weaker donor capacity of the cationic gold centre which ultimately leads to aw eaker hydrogen bond.This was already indicated by the relative energies of the different conformers (Table 4).It is noteworthy that the 1(r)a nd r 2 1(r)v alues in the cations-albeit being lower than in the neutral complexes-are still higher than those calculated for the arene-Auinteraction in A.Wealso would like to point out that although the arene-gold and C(sp 3 )H•••Aui nteractions are both stabilizing effects,t he electrons in both interactions flows in opposite directions.Whereas the metal center in the arene-gold interaction acts as acceptor,itiselectron donor in the hydrogen bond.These characteristics should also influence the properties of the metal and thus the catalytic activity. [40]oof of concept Having established that gold-hydrogen interactions are equally well suited for highly efficient YPhos based gold(I) catalysis,w en ext wanted to explore the generality of the concept.Au-complexes supported by Buchwald-type biarylphosphines have shown remarkable performance in gold(I) catalysis.W ew ondered if substitution of the biaryl moiety with ap henyl-2-cyclohexyl group would enable gold-hydrogen interactions and thus eventually lead to similar high catalytic performance.T ot his end, we chose Cy JohnPhos as the parent Buchwald-type phosphine and synthesized its phenyl-2-cyclohexyl analogue Cy-Cy JohnPhos (L6)a ccording to ap reviously reported procedure. [41]Thec orresponding gold complex (P6)was prepared from the free phosphine and [Au(tht)Cl] and could be isolated as colourless solid in quantitative yield (Figure 6).Strikingly,e lucidation of the molecular structure of P6 revealed-similar to the YPhos ligands-a relatively short Au•••H À Cd istance of 2.77(4) , thus indicating the presence of agold-hydrogen interaction in the solid-state structure.Broad signals between 48-25 ppm in the 31 P{ 1 H} NMR spectrum and at d H = 3.91 ppm in the  1 HNMR spectrum for the CH-proton of the phenyl-bound cyclohexyl group also indicate the existence of this interaction in solution.Furthermore,t he lower buried volume of L6 (%V bur = 38.1)compared to the YPhos ligands suggests that this interaction is not only enforced by steric bulk and hindered conformational changes.T he presence of the CH•••Auh ydrogen bond is also confirmed by DFT studies.Interestingly,t he energy-optimized structure of P6 shows ashorter CH•••Audistance (2.44 )than found in the crystal.This is probably caused by packing effects in the crystal structure resulting in ar ather large Au-P-C-CCy dihedral angle of 36.3(1)8 8 (c.f.9 .4 in the calculated structure) due to arene-cyclohexyl interactions in neighboring molecules (see SI). QT/AIM as well as NBO analyses show similar values than those obtained for the YPhos ligands,f or instance an electron density of 1(r) = 0.0204 ebohr À3 and aL aplacian of r 2 1(r) = 0.0545 ebohr À5 at the bond critical point in P6.
Having established the presence of the CH•••Auhydrogen bond in P6,w en ext compared the performance of Cy JohnPhos•AuCl and P6 in the hydroamination of phenylacetylene with aniline under exact same conditions (0.1 mol %, 50 8 8C) as applied above.B ased on the results obtained with the YPhos ligands (see above), we expected similar catalytic performance of both complexes.Indeed, both ligands performed equally well (Table 6), giving full conversion to the imine after approx.5h reaction time.T his observation impressively confirms that arene-gold interactions are no prerequisite for the design of efficient gold catalysts,but that Au•••H À Cbonds are equally suited as design principle in ligands other than ylide-substituted phosphines.Most likely,i ti sa lso transferable to any other donor ligand, for instance carbenes by replacement of the pending aryl substituent in the Glorius and Lassaletta ligands (Figure 1) with ac yclohexyl group or other alkyl moieties.

Conclusion
In conclusion, we have performed as ystematic study on PPh 3 and PCy 3 -substituted YPhos ligands to elucidate the importance of different secondary ligand metal interactions for catalysis.Whereas earlier investigations have demonstrated that the PPh 3 group is involved in gold-arene interactions, NMR spectroscopic studies as well as XRD analyses of the cyclohexyl-substituted ligands, Cy YPhos,r evealed the presence of remarkably strong Au•••HÀCh ydrogen bonds, amongst the shortest Au-H interaction reported to date.
Computational studies further confirmed the bonding interaction between the gold center and the PCy 3 moiety and clearly showed that the metal acts as electron donor.Adirect comparison of the stability of the gold complexes of aP h 3 Psubstituted ligand and its PCy 3 analogue by calorimetric studies showed that both ligands bind similarly strong to the metal.Strikingly,t he PCy 3 -substituted ligands delivered highly potent gold catalysts,which showed equal performance to their phenyl-substituted analogues.T he generality of the ability of hydrogen bonds to support stable gold catalysts was demonstrated by means of acyclohexyl substituted derivative of the widely used biaryl phosphines.A lso for this class of phosphine ligands an identical catalytic performance was observed for the biaryl and the cyclohexyl-substituted system.
Overall, these observations clearly demonstrate that goldarene interactions are no prerequisite for the design of highly effective gold catalysts,w hich has often been assumed in the literature,b ut can be replaced by hydrogen bonds.T hus,n ot only flanking arene substituents but also hydrogen bond donors may be introduced as stabilizing moieties in the ligand structures thus further expanding the tools available for future ligand design in gold catalysis.

Scheme 2 .
Scheme 2. Synthesis of the ylide precursor Cy Y SF -H from perfluorobutyl sulfonyl fluoride, KHMDS and the phosphoniumsalt [Cy 3 P-Me]I and subsequent reaction with nBuLi and PCy 2 Cl to yield Cy Y SF -PCy 2 (L5).

Scheme 3 .
Scheme 3. Preparation of [Au( Cy Y S PR 2 )Cl] complexes(with R = iPr and Cy)f rom the reaction of L1, L4 and L5 with [Au(tht)Cl] in THF.
phenyl-substituted YPhos ligands bound to gold, we initiated as olution calorimetry study focusing on ligand substitution of the labile dimetylsulfide (DMS) ligand in the common gold synthon, [Au(DMS)Cl].R esults are presented in Scheme 4. Theb atch solution calorimetry results clearly show the very similar enthalpy of reaction values (À19.6 AE 0.4 (for A) and À18.3 AE 0.4 (for L1)k cal mol À1 )o btained for this simple ligand substitution.These data clearly indicate,t hat within

Table 4 :
Energies [kJ mol À1 ]o fthe different possible conformers of P1 and P5 and their cations with and without additional coordination of phenylacetylene (pa)r elative to the energy of conformer C1.Complex DDG [kJ mol À1 ] DDG [kJ mol À1 ] DDG [kJ mol À1 ] (pa) + (R = pTol) AE 0.0 [a]+ 29.4[a,b]   + 19.0[a,c] in the SI).To compare the observed C(sp 3 )ÀH•••Aui nteraction with arene•••Aui nteractions,Q TAIM and NCI analysis were also performed for the YPhos-AuCl complex A with the PPh 3 substituted ligand.Thec alculations revealed an arene•••Au and aC(sp 3 )H•••Auinteraction between the PPh 3 phenyl and the PCy 2 group,r espectively,a nd the gold center.At the corresponding BCPsanelectron density of 1(r) = 0.0141 and 0.0121 ebohr À3 is found, which-in contrast to our initial expectation-is considerably lower than the electron density observed at the BCP for the C(sp 3 )H•••Aui nteraction in P1.Furthermore,the NCI scatterplot (see SI) of P1 clearly shows the additional spike at sign(l 2 )1 % 0.022, representing the strongly attractive C(sp 3 )H•••Auinteraction, while the attractive interactions in A only extend to sign(l 2 )1 % 0.018 further confirming the weaker nature of the arene•••Aua nd C-(sp 2 )H•••Auinteractions.
J PP = 106.8Hz) in L1 to 40.7 and 29.9 ppm ( 2 J PP = 35.6Hz)forP1.I nterestingly,i nc ontrast to the phosphine precursors,t he protons of the tertiary carbon atoms in the phosphonium group appeared as as lightly broadened signal in the 1 H-NMR spectrum of complexes P1 and P4.T he corresponding carbon atom shows as imilar behavior in the 13 C{ 1 H} NMR spectrum and appears as ab road doublet at around 36.9 ppm.This broadening of the PCH protons of the PCy 3 group is even more pronounced for complex P5 and could be caused by an attractive Au•••H-C(sp 3 )i nteraction.Indeed, VT-NMR studies of as olution of P5 in DCM-d 2 showed asplitting of the signal with one proton signal being downfield shifted from 3.21 ppm to 4.53 ppm at À80 8 8C(see FigureS27 and S28).Thetwo signals integrate in a2 :1 ratio,s uggesting that the downfield shifted signal corresponds to one proton interacting with the metal center.DFT calculations (see SI for details) confirm adownfield shift of the PCH proton interacting with gold by 1.24 ppm (5.18 ppm versus 3.89 and 3.99 ppm for the PCH signals),

Table 2 :
Comparison of the TEP values for different alkylphosphines and YPhos ligands.Ligand n Co Rh [cm À1 ]T EP calc.[cm À1 ] [a]

Table 3 :
31P{ 1 H} NMR data, V bur %and important crystal structure bond lengths and angles for the novel Cy YPhos complexes.

Table 5 :
Results of the computational studies on the secondary ligand gold interactions in the P1, A, P3 and P5.For further details, see the Supporting Information.Calculationswere performed on the PW6B95D3/def2tzvp (MWB60f or Au) level of theory including aPCM model with aniline as solvent.

Table 6 :
Results of the hydroamination of phenyacetylenewith aniline catalyzed by the gold complexes with L6 and CyJohnPhos.Reaction conditions: 0.1 mol %Y Phos ligand and 0.1 mol %NaBAr F ,n eat, 50 8 8C.