Conformational Control of Chemical Reactivity for Surface‐Confined Ru‐Porphyrins

Abstract We assess the crucial role of tetrapyrrole flexibility in the CO ligation to distinct Ru‐porphyrins supported on an atomistically well‐defined Ag(111) substrate. Our systematic real‐space visualisation and manipulation experiments with scanning tunnelling microscopy directly probe the ligation, while bond‐resolving atomic force microscopy and X‐ray standing‐wave measurements characterise the geometry, X‐ray and ultraviolet photoelectron spectroscopy the electronic structure, and temperature‐programmed desorption the binding strength. Density‐functional‐theory calculations provide additional insight into the functional interface. We unambiguously demonstrate that the substituents regulate the interfacial conformational adaptability, either promoting or obstructing the uptake of axial CO adducts.


Introduction
Fort he creation of novel materials and devices,i nspiration is frequently sought in nature.P orphyrins and other natural tetrapyrrole compounds can incorporate al arge fraction of the chemical elements in the periodic table.T heir functionality is tuned by choice of the complexed species, possible axial ligands and substituents in the macrocycle periphery.F or example,i nb iology,t he binding of small molecules to metal centres determines many vital functions. Over the past decades we have witnessed an intense interest in utilizing porphyrins on surfaces as functional building blocks with am yriad of possible applications:r anging from atomic switches to single-molecule magnets and catalysts. [1] Thes urface chemistry of cyclic tetrapyrrole compounds is therefore atopic of extended research [2] and includes the onsurface metallation [3] as well as s-block [4] and p-block [5] element functionalisation. In this context, the effect of the macrocycle substituents has been studied systematically. [6] Moreover,the reactivity of individual metal atoms on surfaces is atopical issue in single-atom catalysis, [7] whereby arrays of metalloporphyrin layers under vacuum conditions present av ersatile playground due to the coordinatively unsaturated metal centres provided by the generally favoured adsorption geometries with the macrocycle residing parallel to the substrate lattice.
Complexes of inorganic gaseous molecules with metalloporphyrins are important intermediate species in catalysis. CO, [8] NO, [8a, 9] NH 3 , [10] and H 2 O [10] have been shown to bind on metal supported metalloporphyrins and phthalocyanines axially to the metal centre.I np articular,C Oa lso exhibited an unusual cis-m-dicarbonyl ligation on top of Fe (and Co) tetraphenyl porphyrins on Ag(111) (and Cu(111)). [11] Ligation to the metal centre gives rise to the so-called structural transeffect, whereupon the metal atom is electronically and physically decoupled from the substrate. [9,10,12] Generally, as ignificant alteration of the porphyrinsr eactivity and electronic structure occurs due to the interaction with the metal surface. [13] Tu rning our attention on the topic of "switch on"functionalities of organic layers on metal surfaces,wecan find acommon approach of "decoupling" the molecule from the surface by for example,arigid tethering, [14] bulky substituents, [15] or ap latform [16] which enables a" lift-off" of the functional moiety.I nabiological environment, the macrocycle conformation can influence its functionality. [17] Here,w ew ill examine this aspect:c an we influence the function present in the free molecule (here CO binding) by the conformation of the porphyrin macrocycle ( Figure 1), hosting the metal centre,ont he surface?
ForR ut etraphenyl porphyrins (Ru-TPPs), CO is determined to have an unusually high ligation energy (1.9 eV), [18] hence can be considered as aprototypical out-of-plane ligand with the stability of acovalent attachment. Here,westudy the effect of the porphyrin surface environment on this ligation for Ru-TPP and its planarized derivatives (Ru-TPP pl )o n Ag(111). We use scanning tunnelling microscopy (STM) to find a cis-m-dicarbonyl ligation [11] stable at low temperatures (5 K) and an axial ligation at higher temperatures (200 K), which is also examined with temperature programmed desorption (TPD). In stark contrast, there is no evidence of CO binding to the planarized Ru-TPP derivatives on Ag(111) under either conditions.W ec orrelate the axial binding to conformational and electronic changes,r ationalised by density functional theory (DFT), X-ray and ultraviolet photoelectron spectroscopy (XPS and UPS) and normal incidence X-ray standing waves (NIXSW).

Imaging in Real Space
When deposited on Ag(111) at room temperature in submonolayer coverages,R u-TPP (Figure 2i nset) molecules self-assemble in as quare phase [19] described by the epitaxial   matrix  70  48 . [20] Figure 2A shows the assembly on such asurface cooled down to 5K(overview image in Figure S1). Forn egative bias voltages (%À1V), the single molecule appearance of the pristine Ru-TPP (outlined in orange) is characterised by three bright protrusions along the macrocycle and four less bright in the periphery marking the phenyl substituents.T he central bright protrusion corresponds to af illed electronic state of the Ru centre (cf.U PS below), [21] whereas the outer ones can be assigned to the protruding nwpyrroles (a-pyr) of the macrocycle. [11] Thedownward bending k tw-pyrroles (k-pyr) are not discernible.Inthe STM images of the layer we can also identify molecules with additional protrusions,l ocated on the sides of the Ru centre and perpendicular to the axis of the a-pyr (examples outlined in white and blue). Their STM appearance is virtually identical to the m-carbonyl rider ligation on Co-TPP and Fe-TPP, [11] and given as mall residual pressure of CO (cf.e xperimental section), we can confidently assign those to the analogous Ru-TPP ligation. Themolecule outlined in white can be identified as featuring a cis-m-dicarbonyl binding geometry ( Figure S2) and the example outlined in blue is characteristic of as ingle CO adsorbed in the rider mode and switching between the two adsorption sides during the STM imaging. Performing STM investigations at higher temperatures (150 K), we found solely as ingle mode of CO ligation, recognisable by uniform protrusions directly on top of the Ru centres (example outlined in green in Figure 2C,D). We attribute these to axial carbonyls. [8c] Monitoring the same area of aRu-TPP layer by STM ( Figure 2B-D), while dosing CO in situ, we note that increasing the CO exposure led to an increase in the number of protrusions until all Ru-TPP molecules became brighter, which was achieved after anominal exposure to % 2Langmuir (L) of CO.Itshould be noted that an estimate of the sticking coefficient cannot be extrapolated from the nominal value,a st he real exposure will differ due to effects such as tip shadowing. Figure 1. Models of aRutetraphenyl porphyrin (Ru-TPP, left) and ap lanarizedR u-TPP derivative (right) on Ag(111). The substituents are faded to highlight the difference in the conformation of the porphyrin macrocycles. [20] Ru, C, N, H, and Ag are shown in raspberry, grey,blue, white, and silver,respectively. TheC Ol igands can be selectively removed by STM tip manipulations at 150 Kasillustrated in the sequence of STM images in Figure 3: At the position marked by the green cross ( Figure 3A), the voltage was ramped from 1.28 Vto2.15 Vin ac onstant current mode (50 pA) while monitoring the tip height ( Figure 3B). As udden change in the vertical tip position at % 2.1 Vi ndicates desorption of the CO ligand, which is confirmed by afollow-up image revealing apristine Ru-TPP at the location of the voltage pulse ( Figure 3C). This procedure allows reliable removal of single CO ligands.O n similar systems both tunnelling current induced desorption [8c, 22] and electric field induced desorption [23] have been observed, though smaller bias voltages (< 1V)were required. Forv oltages > 1.5 V, an on-local desorption is often repor-ted. [8b,22, 24] Such ad esorption behaviour is also observed in this system with ab ias voltage of 2.0 V, when higher tunnelling currents are applied ( Figure S3).
To investigate the effect of the macrocyclic conformation on the CO ligation to the Ru centre,wehave investigated the adsorption behaviour of Ru-TPP on Ag(111) after annealing to 620 K. This process causes cyclodehydrogenation reactions between the macrocycle periphery and the phenyl substituents,l eading to af amily of four planarized Ru-TPP derivatives,R u-TPP pl . [19][20] Based on DFT calculations,t he binding energy of the most commonly occurring Ru-TPP pl 3 [19] to Ag(111) is 5.66 eV,1 .26 eV higher than that of pristine Ru-TPP.T he different derivatives can be identified by matching the characteristic outline of the structural formula (Figure 4A)t ot he STM image ( Figure 4B,E), whereas nc-AFM imaging can visualise more directly the chemical identity,a s illustrated for one of the more frequently occurring species in Figure 4C.T he resulting porphyrin macrocycle appears to exhibit asubtle bowl shape with pyrrole tilt angles of 68 8 and 88 8 (see nc-AFM and respective simulation in Figure 4C,D and Table 2) and also offers ac oordinatively unsaturated metal centre.W en ote that the surface depicted in Figure 4E has been exposed to the small amounts of CO at 5Kneeded for the tip functionalisation, however no evidence of al ateral adsorbate stabilisation was found on the Ru-TPP pl molecules by STM/nc-AFM. As the rider ligation is associated with the saddle shape deformation, [11] we would not expect this bowl configuration to permit such ligation. However,i ti sw ith some surprise that we do not observe an axial ligation at all. Theprotrusion in the centre of Ru-TPP pl observed in STM at negative bias ( Figure 4B,E,G) arises,s imilarly to Ru-TPP, from the Ru centre [21b] (cf.U PS below) and is not related to potential CO adsorption, as confirmed by the corresponding nc-AFM image ( Figure 4C). At experiments of methodical  exposure of Ru-TPP pl /Ag(111) to CO at 150 K( Figure 4F-H and Figure S4) no CO uptake by Ru-TPP pl was observed in STM data.
To corroborate the difference in adsorption behaviour clearly with the same STM tip conditions,w ep repared asample containing both Ru-TPP and Ru-TPP pl molecules on Ag(111) ( Figure 4F). In this mixed lattice exclusively the planarized derivative 1 appears ( Figure 4A). This effect is associated with the different molecular shapes of other product species that do not fit into the expressed overlayer lattice ( Figure S5). [25] The(stepwise) exposure of this layer to doses of CO at 150 Kresulted in saturating exclusively all Ru-TPP centres with CO ( Figure 4H), whereas no changes were encountered for the Ru-TPP pl species.A ni ntermediate CO coverage acquired at negative bias ( Figure 4G)highlights the difference in STM appearance between Ru-TPP,R u(CO)-TPP and Ru-TPP pl (see also Figure S5).

Binding Energy and Desorption Kinetics
To deduce informationa bout theb onds trengtho ft he axiallyligated CO on theRu-TPPlayer andtoconfirm that CO does notligatetothe Ru-TPP pl underthe same conditions,we carriedo ut systematic TPDm easurements. Aftere xposureo f thes quarep hase of Ru-TPP on Ag(111) to CO,o ur results show exclusively CO desorption ( Figure S6) in the temperature range of 200-550 K. Dosing different amounts of CO onto al ayer of Ru-TPP has no effect on the shape of the desorption curve,but only on the intensity ( Figure 5, purple), indicative of first order desorption kinetics.T he acquired spectra can be modelled by assuming apre-exponential factor of n = 10 13 s À1 and including two first-order desorption processes of equal intensities with energies of E des,1 = 0.80 eV and E des,2 = 0.84 eV (see details in Supporting Information and Figure S7). Thed ifference in binding energy of 0.04 eV could be related to Ru(CO)-TPP adsorption on both fcc and hcp hollow sites of the Ag(111) surface (see DFT model of optimised structure in Figure S8). We note that consistent with our experiments,insuch acase we would not expect ap referential occupation for the lower binding adsorption site,asnoexchange of CO between the molecules is possible at 200 Kand the desorption temperature from the Ag(111) is much lower. [26] However,w ec annot exclude am ore complex desorption behaviour as ac ause for the spectrass ignature.W hile the desorption energy is very comparable to values found for Ru(CO)-TPP on the more reactive Cu(110) surface, [8c] it is significantly smaller compared to gas-phase molecules. [18] After exposing alayer of Ru-TPP pl to CO,t here is no desorption trace of CO detected ( Figure 5, red), confirming the results from STM/AFM measurements that CO is not ligating to Ru-TPP pl .
At this stage,the following two questions arise:(1) How is the Ru-TPP affected by the CO ligation?( 2) Why are these very similar porphyrins so different in their chemical reactivity?T he following analysis will discuss the impact of the CO ligation on electronic and geometric properties of the Ru-TPP. Figure 5. Coverage dependant TPD spectra and fitting of CO desorption for m/z = 28. Different shades of purple indicate different initial CO coverages q 0 ,dosed at 200 K, on the same Ru-TPP layer.Aheating rate of 2Ks À1 was used. The red spectrum shows the same trace for Ru-TPP pl after CO exposure, confirming that CO is not ligating.

Electronic Structure
We initially investigated the XPS signature of the Ru 3d 5/2 core level as ameasure of the electronic interaction with the metal substrate ( Figure 6A). Upon ligation of CO,t he binding energy of the Ru 3d 5/2 core level shifts by 2.4 eV towards higher binding energies,i ndicating ad ecoupling of the Ru centre from the Ag substrate.The shift towards higher binding energies is in good accord with the DFT prediction (+ 1.8 eV). Note that the Ru 3d 3/2 component is coincident with the C1speak ( % 285 eV), and can be observed as asmall shoulder on the lower binding energy side for Ru-TPP and at the higher binding energy side for Ru(CO)-TPP layers. [21a] From the XPSofamultilayer of Ru(CO)-TPP on Ag(111) onecan deduce that theCOligandremains attached to theRu-TPPo nt he layers withoutd irectc ontact to theA g(111) substrate at 300 K. [27] Thebinding energy of the Ru 3d 5/2 core levelofRu(CO)-TPP directly on Ag(111) is 0.2 eV lower than that observed in the multilayer films.Such ashift is consistent with thee xpectedp olarisations creening by them etal substrate.
UP spectra further show states for both Ru-TPP (Figure 6B,b lue) and Ru-TPP pl ( Figure 6B,r ed) at binding energies of 0.4 eV and 0.9 eV,which can be correlated to the bright protrusion at negative bias voltages in the STM images (Figure 2A,B,E,G), similar to Co-TPP on Ag(111). [28] These Ru states are extinguished for Ru(CO)-TPP ( Figure 6B, purple), indicating that the interaction of Ru centres and the Ag substrate,responsible for these states,isnolonger present upon ligation ( Figure S9). [9] Further insight into the electronic changes upon the adsorption of CO is gained from DFT.F igure 6C shows that changes in the electron density upon CO adsorption are not simply restricted to the porphyrin, but also evident in the Ru-TPP/Ag(111) interface.T he charge at the interface per molecule is significantly reduced upon CO ligation (Dq Int = À0.47 e, q Int = 0.18 e), confirming the electronic decoupling of the Ru-TPP molecules from the Ag(111) surface correlated to the CO ligation and the surface trans-effect. Whilethe CO is negatively charged( q CO = À0.18 e), theR uc entreg etsm ore positively charged( Dq Ru = 0.27 e). Ac loseri nspectiono ft he orbitals tructure( Figure 6D)r eveals ad ecreased electron densityi nt he 5s orbitals of theC Oa nd ac ommensurate increasei ne lectrond ensity in the2 p orbitals,i na greement with theB lyholder modelf or chemisorbedc arbonm onoxide. [29] On theruthenium centre,adecrease of electron density in thed z 2 orbital, as well as an increase in thed zx andd yz orbitals is observed.Itisnotable that this change in theRu3delectron densityi ss imilar between, both,t he Ru-CO, andR u-Ag, whereasf or Ru-TPP pl 3 thec orrespondingD FT calculations find theelectronaccumulationtobeind z 2 ,and depletioninthe d zx ,d yz orbitals. [20] Thed epletion andg aino fe lectrons in orbitals of both Ru andCOshowaback-donationofelectrons from theRucentretothe CO ligand,which in addition to the decoupling cancontributetothe increaseinbinding energy of theR u3 d 5/2 core leveluponCOligation( Figure 6A).

Structural Determination
Our earlier structural investigation of Ru-TPP and Ru-TPP pl has shown that the adsorption height of the Ru centre differs only by 0.14 . [20] Nevertheless,t he adsorption height of the Ru centre of Ru-TPP increases upon CO ligation at 200 Kb y0 .59 from 2.59 AE 0.05 [20] to 3.18 AE 0.12 ,a s shown by NIXSW data of the Ru(CO)-TPP ( Figure 7A, Table 2). [30] Theh igh coherent fraction indicates av ery welldefined adsorption height for the molecules,c onfirming ar ather uniform geometry. [31] TheC1 sN IXSW data (Figure 7B,T able 2) show an increased average adsorption height also for the carbon atoms.T hus,w ec onclude that the nonplanarity of the Ru-TPP facilitates ac onformational change of the entire Ru-TPP upon CO ligation and enables the decoupling of the Ru centre from the Ag surface.One should note that the NIXSW measurements could only be performed on mixed layers of Ru(CO)-TPP and pristine Ru-TPP.While the Ru 3d 5/2 peaks of the two species can be clearly distinguished due to the large shift in binding energy (Figure 6A,T able 1), allowing the adsorption height for each species to be analysed individually,this is not possible for the C1 s. Thec arbon spectra, as described above,h ave to be understood as an average over all carbon atoms from both species,w hich includes additionally an egligible contribution of the Ru 3d 3/2 core level. Therefore,o nly aq ualitative comparison is meaningful. With this in mind, the XSW results are in excellent agreement with complementary DFT calculations ( Figure 7C,T able 2), which predict an increase of the Ru adsorption height of 0.69 and an increase of the average Cadsorption height of 0.11 .W epropose that the porphyrin macrocycle is lifted, while the phenyl substituents remain in  [20] (left) and Ru(CO)-TPP (right) on Ag(111). Ru, Ag, C, N, Oand Hatoms are depicted in raspberry,silver,grey,blue and white, respectively. contact with the Ag(111) substrate.T hese conformational adaptations can be interpreted as ar ather strong surface trans-effect. [10, 12a, 32] To understand the anticipated structural trans-effect for planarized Ru(CO)-TPP derivatives,w ei nvestigated aD FT geometry optimisation ( Figure S10). Here,t he trans-effect would increase the Ru adsorption height by 0.49 whereas it would leave the macrocycle mostly unaffected (Table 2). In comparison with the saddle-shaped pristine TPP,t hese deformations are smaller and show less adaptation of the macrocycle with CO ligation, which is more restricted by its adsorption to the silver surface.
It is notable that the planarized Ru(CO)-TPP derivative investigated is also as table geometry in simulation with abinding energy of the CO predicted to be smaller by 0.5 eV with respect to the pristine Ru-TPP.W ecan thus attribute the lack of experimental evidence of this species to either ahigher activation barrier associated with the decoupling of the Ru from the silver surface or to CO sticking coefficient differences of more than an order of magnitude.

Conclusion
We have studied the CO ligation on distinct Ru-porphyrins on as ilver surface by ac ombined theoretical and experimental analysis of the electronic and geometric effects of such aligation.
Rider CO-ligation at low temperatures (at 5K)and axial CO-ligation (up to % 250 K), in agreement with the Blyholder model [29] for chemisorption, were observed only for the pristine saddle-shape Ru-TPP.S TM allowed tip induced desorption of single ligands without damaging the Ru-TPP underneath, which can be used to create patterns on ananometer scale.T he large shift in binding energy of the Ru 3d 5/2 core level upon axial ligation indicates an electronic decoupling of the Ru centre from the surface and both NIXSW and DFT have confirmed significant conformational changes. While the Ru centre is affected the most, increasing in adsorption height by % 0.6 ,a ni ncrease of the adsorption height is observed for the entire molecule ( Figure 7C). From TPD we determined the desorption energy of the axial CO ligand to be 0.8 AE 0.1 eV,reduced by 1.1 eV in comparison to the CO binding strength to the free Ru-TPP.
Fort he planarized Ru-TPP derivatives,t here was no sign of CO ligation in STM, AFM and TPD measurements.With the bonding of the Ru centre to the Ag surface being similar for both investigated porphyrins,o ur results emphasize the crucial role of the flexibility of the Ru-TPP in the ligation process and the related ease of decoupling of the Ru centre from the Ag(111) surface.
Our findings with this model Ru-porphyrin/Ag(111) system are expected to be relevant for the elucidation to processes related to gas sensing, [33] and to supported single-atom catalysts (e.g. Ru-N4). Table 2: Structural parameters by DFT and NIXSW results from Ru 3d 5/2 and C1score levels for the different investigated systems. The adsorption heights are deduced from the coherent position, for the C atoms it is an average value. In parentheses, we report the DFT simulated adsorption height that would be the result of the respective NIXSW measurement for comparison.