Carbazole‐Based Tetrapodal Anchor Groups for Gold Surfaces: Synthesis and Conductance Properties

Abstract As the field of molecular‐scale electronics matures and the prospect of devices incorporating molecular wires becomes more feasible, it is necessary to progress from the simple anchor groups used in fundamental conductance studies to more elaborate anchors designed with device stability in mind. This study presents a series of oligo(phenylene‐ethynylene) wires with one tetrapodal anchor and a phenyl or pyridyl head group. The new anchors are designed to bind strongly to gold surfaces without disrupting the conductance pathway of the wires. Conductive probe atomic force microscopy (cAFM) was used to determine the conductance of self‐assembled monolayers (SAMs) of the wires in Au–SAM–Pt and Au–SAM–graphene junctions, from which the conductance per molecule was derived. For tolane‐type wires, mean conductances per molecule of up to 10−4.37 G0 (Pt) and 10−3.78 G0 (graphene) were measured, despite limited electronic coupling to the Au electrode, demonstrating the potential of this approach. Computational studies of the surface binding geometry and transport properties rationalise and support the experimental results.


Introduction
Anchor groups fulfil acritical role in materials designed to assemble on surfaces. [1] In the field of molecular-scale electronics,t hey serve as contact points between conductive molecules and electrode surfaces.T he strength of binding interactions and extent of electronic coupling between an anchor group and an electrode are important factors when designing conductive organic materials. [2] Forf undamental studies,a na nchor able to form ac onductive junction with alifetime longer than amolecular conductance experiment is adequate.H owever,w hen working towards the goal of functional devices based on conductive molecules, [2a] such as thermoelectric generators [3] and Peltier coolers, [4] strengthening the interactions between these molecules and electrode surfaces is necessary to achieve effective device lifetime and performance.Improved anchor groups can also result in more ordered surface assemblies.
Thiols,t hioethers,a nd pyridines,a mongst others,a re common anchor groups in fundamental molecular conductance studies using gold surfaces. [2] Recently,s everal new anchor groups have been designed to exhibit increased affinity for gold surfaces,with various applications. [1a-c, 5] Some examples pertinent to molecular electronics are shown in Figure S1.29 in the Supporting Information (SI). In many cases,t hese anchors are tripodal with three conventional anchor groups working together to achieve enhanced surface binding,akin to the chelate effect. Surface-p interactions can contribute additionally to this binding.T ripodal anchors often incorporate an sp 3 -hybridised carbon centre to induce the desired geometry.H igh molecular conductance is associated with conjugation, [2a] meaning that sp 3 -hybridisation is considered undesirable in conductive materials,although astudy of ap yridine-anchored tripod describes contrasting results. [5a] Positioning the sp 3 -centre outside the conductance pathway of the molecule can avoid any impact on conductance. [5b] An additional feature of larger anchors of this type is that they increase the spacing between conductive molecular backbones and thereby prevent intermolecular interactions. [5c] In this work, we present anew tetrapodal anchoring motif designed to bind strongly to gold surfaces and form ordered self-assembled monolayers (SAMs) with well-spaced conductive backbones.I nspired by tripodal systems, [5a,b,d-i] multiple anchoring points are incorporated into our design to ensure efficient surface binding. In contrast to many current approaches,weaimed to separate these additional anchoring points from the conductive backbone of the molecule and avoid sp 3 -hybridisation in the conductive pathway.W en ote that while this study focuses on the molecular electronics applications of our new anchoring motif,i tm ay also prove adaptable as au seful scaffold for other applications,f or example,g old surface-assemblies of optoelectronic materials, [6] switches [5d, 7] or polymerisation initiators. [8]

Results and Discussion
Taking the ubiquitous oligo(phenylene-ethynylene) (OPE) backbone as as tarting point, we selected the metapositions with regard to the conductive backbone as an ideal point to incorporate additional anchoring functionality.Conductance through a meta-conjugated pathway is considerably lower than the favoured para-conjugated pathway due to destructive quantum interference (QI) effects. [2b, 9] Therefore, any additional meta-functionalisation should contribute minimally to the conductance of the molecule.Inspired in part by the spirobifluorene tripods reported by the Mayor group, [5b,e] carbazole was selected as the basis of the additional anchors. This choice further disfavours any conductance through the poorly conducting meta-positions as steric factors should induce as ignificant twist between the backbone and carbazole p-systems,reducing conjugation between them. Furthermore,this twist should help to direct the conjugated backbone away from the surface.I na ddition to any interactions between its p-electrons and the surface,c arbazole can be readily functionalised in the 3-and 6-positions,a llowing convenient incorporation of additional binding functionality.
To test this design the model compound 1 was initially prepared (see SI, Section 1.6). Thes ingle crystal X-ray structure ( Figure 1a)c onfirmed the expected twist between the backbone and the carbazole units.T oi nvestigate the anchoring properties of the unit as eries of OPE2 type molecules (where 2r efers to the number of aryl rings in the backbone) were then synthesised, each with one tetrapodal unit (anchor) and one simple terminal aryl group (head). Tw o anchors were proposed, with either ac entral benzene (X = CH) or pyridine (X = N) ring, in each case functionalised with two 3,6-bis(methylthio)carbazole units,that is,four thiomethyl groups per anchoring unit. Thet hiomethyl anchor groups were preferred to acetyl-protected thiols to ensure compatibility with the synthetic route and to reduce synthetic complexity.T he use of multiple thioethers has been reported to result in strong anchoring to ag old surface,w here each thioether makes ac ontribution comparable to that expected from at hiol. [10] Fort he benzene systems,a nchoring is expected from only the functionalised carbazole subunits, whereas the pyridyl nitrogen may be able to interact additionally with the gold surface.T hree head groups were investigated, namely benzene, p-pyridine and m-pyridine.The benzene head group has no defined anchor point and should therefore interact poorly with metallic electrodes.B enzene should, however, be able to interact with graphitic surfaces through p-p overlap at appropriate contact angles.B oth pand m-pyridyl head groups should bind to metallic electrodes through the nitrogen lone pair, with the para derivative expected to show ah igher molecular conductance than the meta derivative due to QI effects.T he structures of these species are shown in Scheme 1. Our naming convention is to indicate the anchor unit (B or P)followed by the head unit (B, pP or mP), for example, BpP is the compound with abenzene-based anchor and a para-pyridyl head.
Akey building block is the thiomethyl-substituted carbazole derivative 2.A lthough it has been previously reported, limited experimental details and characterisation data are available. [11] We developed an alternative synthesis using 3,6diiodocarbazole [12] as the precursor.A fter TBDMS-protection of the N-position, thiomethyl substituents were introduced via lithiation followed by treatment with dimethyl disulfide,although optimisation was required to minimise byproduct formation and achieve practical yields (see SI, Section 1.7). Straightforward deprotection of the TBDMS group using TBAF afforded 2.Byusing nucleophilic aromatic substitution (S N Ar) reactions it was then possible to prepare the benzene-and pyridine-based scaffolds 3 and 4 (Scheme 1). Both are functionalised with an iodide group as aconvenient synthetic handle for subsequent synthetic transformations.T his is an advantage of the S N Ar route;a lternatives based on metal-catalysed coupling reactions would have ah igher potential for by-product formation and would either rely on statistical reactions or utilise al ess-reactive bromide group as the resulting synthetic handle.Adisadvantage of the S N Ar approach is that it is poorly compatible with central rings bearing electron-donating substituents,a sthese disfavour S N Ar reactions.
With the iodide species in hand, the targeted library of asymmetric molecular wires was prepared using Sonogashira protocols (Scheme 1). Forc omparison in the conductance studies,aseries of analogous OPE2 derivatives were also prepared with asimple (protected) thiol anchor (denoted with S in our naming convention) and the three head groups used in the tetrapodal series (See SI, Section 1.8). To probe the effect of changes to the length of the conductive backbone, the shorter (B(OPE1)pP)a nd longer (B(OPE3)pP)a nalogues of BpP (Scheme 1) were also prepared. They differ by the effective subtraction or addition of ap henylethynyl unit into the conductive OPE2 backbone.
Single crystal X-ray structures were obtained for model compound 1 and for BmP.F or 1,t he torsion angles between the central benzene ring and the two carbazole units are 568 8 and 608 8 (Figure 1a), whereas for BmP,which showed noticeable structural disorder in the positions of the pyridylnitrogen and the thiomethyl groups (see SI, Section 2.1), these angles are 558 8 and 398 8 ( Figure 1b). Using the SIESTA package, [13] density functional theory (DFT) simulations (see SI, Section 3.1) of model, alkyne-terminated systems in the gas phase gave corresponding angles of 388 8 and 538 8 for ab enzene (X = CH) base ( Figure 1c)a nd 248 8 and 298 8 for ap yridine (X = N) base ( Figure 1d). Ther educed torsion in the latter can be attributed to reduced steric hindrance.Hbonding interactions may also contribute (see SI, Section 2.2). In all cases,todiffering extents,the designed twist is observed between the carbazole and OPE p-systems,w hich should reduce the influence of the carbazole units on the conductive backbone.D FT was also used to simulate the assembly of these model systems on Au(111) surfaces.A ss hown in Figures 1e and 1f,t he two carbazole units splay out to lie essentially flat on the gold surface (indicative of favourable surface-p interactions), with the OPE backbone protruding at an angle.T he angle between the gold surface and the conductive backbone is 488 8 for ab enzene base and 378 8 for ap yridine base.T herefore,l ittle conjugation is expected between the conductive backbone and the carbazole units in this conformation.
Theb inding energies of these conformers to aA u(111) surface,c alculated using the counterpoise correction method, [14] are À1.86 eV (X = CH) and À1.68 eV (X = N). These values suggest that the pyridine nitrogen atom does not enhance surface binding (at least in this conformation), and in fact results in slightly weaker surface interactions than the benzene analogue.T he binding energies of some simple, conventional anchors were reported previously using the counterpoise method. [15] Comparison with these values shows that both tetrapodal anchors have considerably higher binding energies than amine (À0.30 eV), nitrile (À0.41 eV), dihydrobenzothiophene (À0.41 eV) and pyridine (À0.50 eV) anchors,a nd have enhanced binding compared to thiols (À1.51 eV). As ignificant increase in binding energy is also observed in comparison to phosphine-based tripodal anchors with three thiomethyl anchoring groups (ca. À1eV).
[5f] The adsorption energies of triptycene tripods with three thiol anchors,w hich account for loss of hydrogen upon binding to ag old surface,w ere found to be À1.62 eV (for less flexible aryl thiols) and À2.67 eV (for more flexible benzylic thiols). [5g] Thet etrapodal anchors perform comparably to the former case despite using thiomethyl anchors rather than thiols. Although it binds strongly,t he latter triptycene design includes sp 3 -carbons so is likely poorly suited to molecular electronics applications.
Thep roperties of SAMs of the tetrapodal molecules on Au(111) surfaces were assessed using ar ange of techniques. Details of the preparation of SAMs,characterisation methods and associated images can be found in the SI (Section 2.3). Atomic force microscope (AFM) imaging showed that the SAMs were densely packed films with uniform structures in the range of 0.5-1 nm. Thet hickness of the SAMs was Scheme 1. Synthesis of tetrapodalm olecular wires. Reagents and conditions: a) Cs 2 CO 3 ,DMF, 100 8 8C, 17 h; b) K 2 CO 3 ,DMSO, 70 8 8C, 2h;c )CuI, Pd(PPh 3 ) 2 Cl 2 ,T HF, DIPEA, RT,80-120 min. Fort he tetrapodalm olecular wires, the anchoring unit is coloured red, and the head unit blue (note that for B(OPE1)pP the head unit pyridine ring also forms part of the anchor group). determined using anano-scratching method. [16] Theaveraged film thickness for the OPE2 molecules was in the range of 0.6-0.65 nm, and for OPE3 derivative B(OPE3)pP it was 0.9 AE 0.04 nm. These values correspond to tilting angles of ca. 358 8 between the OPE backbones and the substrate surface,i n reasonable agreement with the DFT calculations ( Figures  1e,f). SAMs of the model compound 1 could not be prepared on gold, showing that the thiomethyl groups play ac ritical role in surface assembly.D FT simulations confirm this, showing stepwise increases in binding energy as thiomethyl groups are sequentially added to the carbazole-based anchoring motif (see SI, Section 3.1 for further discussion).
Thed ensity of molecules on the gold substrate was determined using reductive desorption and quartz crystal microbalance (QCM) measurements.R eductive desorption allowed the molecular area to be calculated based on the charge density of the desorption peak, under the assumption that four electrons correspond to desorption of as ingle molecule (i.e.o ne electron per thiomethyl anchor group). A second cycle of desorption confirmed that all molecules were desorbed from the surface in the first sweep (SI, Figure S2.07). BB and PB were investigated as representative molecules and had estimated molecular occupation areas of 265 2 and 300 2 ,respectively.This agrees well with aDFTestimated molecular footprint of 265 2 for either benzeneor pyridine-based tetrapods (see SI, Section 3.1). Thedesorption potential of benzene-based material BB was À0.57 Vvs. SCE, whereas for PB the desorption potential was À0.67 Vvs. SCE. In this case,t he pyridine-based anchoring unit appears to provide as mall additional anchoring effect compared to the benzene-based analogue.A lthough this appears to disagree with the DFT-calculated binding energies stated above,w en ote that the methods may not be directly comparable as reductive desorption relates to electrochemical stability,w hereas the binding energies relate to adsorption. QCM measurements (see SI, Section 2.5) gave estimated molecular areas around 30 %l ower than those derived from reductive desorption. AFM imaging of the QCM substrate (SI, Figure S2.09) showed that the surface was much rougher than those used in the other measurements and was therefore likely to result in underestimation of molecular areas.Asthe substrate used for reductive desorption is comparable to those used in conductance studies,w eb elieve this is the more reliable method for determining molecular area in this case.
Investigations of the conductance of molecules with multiple binding sites using single-molecule junction techniques can be challenging,a svarious junction configurations are possible. [5h] This generally results in complex conductance data from which it can be difficult to determine the contributions of different configurations.T oa void this complication, the conductance properties of the tetrapodal wires were investigated using conductive probe AFM (cAFM). Thea nalysis of SAMs of the tetrapods indicated that the molecules were binding to the gold surface with all four thiomethyl groups in the designed manner. Therefore,it was expected that the protruding head groups would be contacted with the cAFM probe and afford the desired Auanchor-head-probe junction configuration. Afurther advantage of cAFM is that measuring the conductance of SAMs should be more representative of potential large-area devices than single-molecule methods.
Electrical maps obtained via cAFM revealed that, in general, the SAMs showed very good electrical uniformity. However,the electrical uniformity of SAMs of tetrapods with pyridine bases was poorer than that of SAMs of benzenebased analogues prepared under the same conditions:t here are clearly some uncovered regions in the electrical map in the pyridine case (cf. PpP and BB,S I, Figure S2.11). The electrical conductance of the SAMs was determined through cAFM IV measurements using both Pt-and graphene-coated probes (See SI, Section 2.6). After gently approaching the surface with anew cAFM probe,the contact force was set to 2nNa nd the bias voltage was swept (typically from À1Vto + 1V)a ta tl east 20 randomly selected locations on the sample surface (Figures 2c,S2.12, S2.13 and S2.15). At least 3 IV sweeps were conducted at each location and used to calculate the differential conductance of the junction (G J ). Then umber of molecular junctions formed between the probe and the substrate was estimated using the Hertz model [17] (see SI, Section 2.6). This allowed the conductance contribution of as ingle molecule (G M )t ob ec alculated for each IV curve.Atleast 60 IV curves were measured for each molecule,g iving ad istribution of G M values (Figures 2a,b, S2.14 and S2.16). Themean G M values for each molecule are listed in Table 2.
Using either Pt-or graphene-coated probes,f or ag iven head group,t he average molecular conductance, G M ,o f ap yridine-based wire is higher than that of the equivalent benzene-based wire.H owever,t he conductance distribution of pyridine-based wires tends to be broader than that of the benzene-based analogues (Figures 2a and S2.14 in the SI). This agrees with the observation above that SAMs of pyridine-based species have lower electrical uniformity than SAMs of their benzene analogues.Given the close proximity to the Au surface,the slightly reduced steric bulk of apyridine lone pair versus ab enzene hydrogen atom could mean that more molecular conformations are possible within the Au-SAM-probe ensemble in the pyridine case.T he extent of electronic coupling between the Au surface and the conductive backbone through the pyridine nitrogen, ak nown anchor group, [18] could vary significantly with conformation, thus affording abroad conductance distribution.
G M for the tetrapodal wires is 3-5 times higher in Au-SAM-graphene junctions than in Au-SAM-Pt junctions. Stronger molecule-probe interactions are possible in the former case as p-p overlap may occur between the head groups and graphene,w hich could enhance electronic coupling and therefore increase conductance.Alternatively, p-p overlap could move the electronic contact point further down the backbone,effectively shortening the conductive pathway, thereby affording ahigher conductance. Form olecules with ag iven anchor group, G M varies with head group according to the trend pP > B > mP in all cases. The pP head group was expected to perform best with am etallic probe,d ue to interactions between the probe and the pyridine lone pair, which were anticipated to enhance electronic coupling,and constructive QI effects.Interestingly, this head group also showed the highest G M when using ag raphene-coated probe.T hese results are consistent with constructive (pP)a nd destructive (mP)Q Ie ffects occurring for both probes,e ven though in short single molecules,s uch effects are sensitive to the nature of the molecule-electrode contact and can be masked, due to parallel transport through the s-channel and p-s mixing. [19] Theh igher conductance of the B head group relative to mP suggests that p-probe interactions contribute significantly to electronic coupling for both probes,asthe former has no other binding functionality.
Thee ffect of molecular length on G M was studied using BpP and its two analogues B(OPE1)pP and B(OPE3)pP.As seen in Figures 2b-d, conductance decreases with molecular length, as expected for OPE-type molecular wires.I ndeed, ap lot of ln(G M )a gainst DFT-relaxed molecular length gives the expected linear trend (Figure 2d), allowing for estimation of the tunnelling decay factor, b. Fort his series, b = 0.42 AE 0.03 À1 ,w hich is slightly larger than the reported range for OPEs of 0.2-0.34 À1 . [2b, 18] This discrepancyc ould relate to the incorporation of the pyridine head group into the tetrapodal base of B(OPE1)pP in order to create as horter analogue of BP.T his structural change may have an additional effect on conductance and therefore distort the b-value. From the plot in Figure 2d the contact resistance of the benzene-based anchoring unit is estimated as 9.8 MW.
We further investigated these three molecules and two additional, longer analogues using DFT-based charge transport calculations with the Gollum package [20] (see SI, Section 3.2). Thel ogarithms of the calculated conductances of the OPE2 to OPE5 species follow the expected linear trend with molecular length, with b = 0.21 À1 (SI, Figure S3.02). Thec alculated conductance of B(OPE1)pP is much higher than would be expected by extrapolating this trend, indicating that it may not be arepresentative member of the OPE series. In the simulations,t he short conductive backbone of B-(OPE1)pP results in additional electronic coupling between the top electrode and the carbazole units,w hich affords additional conductance pathways and ah igher molecular conductance.W hent his additional coupling is artificially removed, the calculated conductance is lower and closer to the expected trend (SI, Figure S3.02). Thel inear trend observed in the cAFM experiments suggests that electronic coupling through the carbazole units is hindered in this case.  This could be due to roughness in the top contact or the presence of solvent molecules. Compared to thiol-anchored analogues SB, SpP and SmP, results from the Pt probe show that G M for the benzene-and pyridine-based tetrapodal wires is 5-10 times and 4-5 times lower, respectively.T his is reasonable as at hiol anchor provides strong electronic coupling to agold surface through Au-S bond formation, whereas the conductive backbones of the tetrapodal species are unable to interact directly with the surface to the same extent. To probe the nature of the electronic coupling between the tetrapods and the gold surface,w ec onducted charge transport simulations in which BB was compared to asimple,unfunctionalised OPE2, which was held in the same geometry as the OPE2 backbone of BB when relaxed in aj unction configuration (Figures 3a,b). The resulting transmission functions (Figure 3c)a re remarkably similar, suggesting that the carbazole anchoring units have negligible influence on electronic coupling and therefore on G M ,d espite providing an efficient means of surface binding. In effect, our molecular design decouples surface binding from electronic coupling.T his important observation potentially allows for investigations of unconventional or weakly binding anchor groups,which could be held in place near the surface by atetrapodal unit, allowing their electronic coupling to be probed regardless of their surface binding properties. Section 3.2 of the SI discusses additional charge-transport calculations investigating the effects of factors such as the spacing between the molecules and the top electrode,the tilt angle of the conductive backbone in the junction, and binding geometries between the head group and top contact.
Thes ingle-molecule conductance of several tripodal molecular wires has been reported in the literature,t ypically using scanning tunnelling microsocopy or mechanically controlled break-junction methods (STM-BJ and MCBJ, respectively). Any comparisons with our G M values obtained using cAFM therefore require benchmarking.T he molecular conductances of thiol-anchored species SpP and SmP have been reported using the MCBJ method. [21] Our cAFM method results in slightly lower conductances for both SpP (log G M /G 0 = À3.78 vs.log G/G 0 = À3.2 [21] )and SmP (log G M / G 0 = À4.17 vs.l og G/G 0 = À3.9 [21] ); ac omparable deviation between cAFM and MCBJ has been observed previously. [22] Such discrepancies may result from differences in the nature of the junctions or assumptions made when calculating G M from G J .Itcan be concluded, however, that it is reasonable to compare our G M values with conductances determined using MCBJ,w ith the caveat that the G M values may be slight underestimates.AsMCBJ conductances have been shown to be comparable with those obtained using STM-BJ, [15,23] it follows that comparisons with STM-BJ data are also valid.

Conclusion
Te trapodal anchor units for gold surfaces based on thiomethyl-substituted carbazole have been developed. The synthetic route is convenient and adaptable for other possible applications.T he anchor units were incorporated into OPEtype molecular wires bearing benzene, para-pyridine or metapyridine head groups,and their conductance was investigated using cAFM and charge transport calculations.The molecules form uniform SAMs on Au(111) in which the carbazole psystems and sulfur atoms bind to the surface with the conductive OPE backbones protruding at an angle of 35-508 8.T he stabilising effect of multiple thiomethyl groups was demonstrated using DFT-calculated binding energies,w hich were higher than those of conventional anchoring groups and analogues where thiomethyl groups were sequentially removed. Using cAFM, the conductance of Au-SAM-Pt and Au-SAM-graphene junctions was measured and used to determine the conductance contribution per molecule, G M , based on the contact area of the AFM probe and the area occupied by am olecule as determined by reductive desorption studies.F or OPE2 tetrapods in Au-SAM-Pt junctions, G M varied from 10 À5.20 G 0 (BmP)t o1 0 À4.37 G 0 (PpP). In all cases,tetrapods with pyridine in the base unit gave higher G M than their benzene-based analogues,b ut tended to show broader conductance distributions.F or ag iven base, G M varied with head group in the order pP > B > mP.F or Au-SAM-graphene junctions,s imilar trends were observed but G M increased by af actor of ca. 4. Charge transport calculations showed that the carbazole units play little or no role in the conductance pathway of the molecules,a nd that electronic coupling to the surface is through the benzene or pyridine ring of the anchoring unit. Such decoupling of surface binding and electronic coupling could enable the use of unconventional functional groups,which may afford strong electronic coupling but only weak physical binding,intandem with ancillary strong anchor groups with poor electronic coupling.D espite the lack of strong electronic coupling between the tetrapods and the gold surface,comparison with the literature shows that their conductance is comparable to existing tripodal systems.O ngoing research in our laboratories is exploring ways to enhance the electronic coupling,and therefore molecular conductance,o fs ystems based on at etrapodal anchor motif,w hile retaining ac onvenient synthetic approach.