Structure and energetics of hydrogen-bonded networks of methanol on close packedtransition metal surfacesColin J. Murphy, Javier Carrasco, Timothy J. Lawton, Melissa L. Liriano, Ashleigh E. Baber, Emily A. Lewis,

Angelos Michaelides, and E. Charles H. Sykes

Citation: The Journal of Chemical Physics 141, 014701 (2014); doi: 10.1063/1.4882863 View online: http://dx.doi.org/10.1063/1.4882863 View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/141/1?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Insight into the description of van der Waals forces for benzene adsorption on transition metal (111) surfaces J. Chem. Phys. 140, 084704 (2014); 10.1063/1.4866175 Transition metal atoms pathways on rutile TiO2 (110) surface: Distribution of Ti3+ states and evidence ofenhanced peripheral charge accumulation J. Chem. Phys. 138, 154711 (2013); 10.1063/1.4801025 Support effects on the dissociation of hydrogen over gold clusters on ZnO(101) surface: Theoretical insights J. Chem. Phys. 137, 234704 (2012); 10.1063/1.4771905 Configurational correlations in the coverage dependent adsorption energies of oxygen atoms on late transitionmetal fcc(111) surfaces J. Chem. Phys. 134, 104709 (2011); 10.1063/1.3561287 Characterization of methoxy adsorption on some transition metals: A first principles density functional theorystudy J. Chem. Phys. 122, 044707 (2005); 10.1063/1.1839552

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

THE JOURNAL OF CHEMICAL PHYSICS 141, 014701 (2014)

Structure and energetics of hydrogen-bonded networks of methanolon close packed transition metal surfaces

Colin J. Murphy,1 Javier Carrasco,2 Timothy J. Lawton,1 Melissa L. Liriano,1

Ashleigh E. Baber,1,a) Emily A. Lewis,1 Angelos Michaelides,3,b)

and E. Charles H. Sykes1,b)

1Department of Chemistry, Tufts University, Medford, Massachusetts 02155, USA2CIC Energigune, Albert Einstein 48, 01510 Miñano (Álava), Spain3Thomas Young Centre, London Centre for Nanotechnology and Department of Chemistry,University College London, London WC1E 6BT, United Kingdom

(Received 11 March 2014; accepted 19 May 2014; published online 1 July 2014)

Methanol is a versatile chemical feedstock, fuel source, and energy storage material. Many reactionsinvolving methanol are catalyzed by transition metal surfaces, on which hydrogen-bonded methanoloverlayers form. As with water, the structure of these overlayers is expected to depend on a deli-cate balance of hydrogen bonding and adsorbate-substrate bonding. In contrast to water, however,relatively little is known about the structures methanol overlayers form and how these vary fromone substrate to another. To address this issue, herein we analyze the hydrogen bonded networksthat methanol forms as a function of coverage on three catalytically important surfaces, Au(111),Cu(111), and Pt(111), using a combination of scanning tunneling microscopy and density functionaltheory. We investigate the effect of intermolecular interactions, surface coverage, and adsorption en-ergies on molecular assembly and compare the results to more widely studied water networks on thesame surfaces. Two main factors are shown to direct the structure of methanol on the surfaces stud-ied: the surface coverage and the competition between the methanol-methanol and methanol-surfaceinteractions. Additionally, we report a new chiral form of buckled hexamer formed by surface boundmethanol that maximizes the interactions between methanol monomers by sacrificing interactionswith the surface. These results serve as a direct comparison of interaction strength, assembly, andchirality of methanol networks on Au(111), Cu(111), and Pt(111) which are catalytically relevant formethanol oxidation, steam reforming, and direct methanol fuel cells. © 2014 AIP Publishing LLC.[http://dx.doi.org/10.1063/1.4882863]

I. INTRODUCTION

Methanol and water are ubiquitous molecules in chem-istry, and the study of their interaction with surfaces isessential for a full understanding of their reactivity. Elucidat-ing the binding and assembly of these molecules at surfaceshas relevance to a number of fields, such as electrochemistry,environmental chemistry, fluid transport, and heterogeneouscatalysis.1–4 It is also a source of general insight into molecu-lar self-assembly at interfaces and the fundamental nature ofinterfacial hydrogen bonds.

Hydrogen-bonded networks of water on surfaces havebeen well studied,5–21 and scanning tunneling microscopy(STM) coupled with density functional theory (DFT) hasbeen particularly useful in determining the structure and sta-bility of the assemblies.22–26 Previous experimental work inultra-high vacuum (UHV) conditions and at cryogenic tem-peratures showed that water forms networks of various sizes,from dimers to oligomers, on metal surfaces.19 The waterhexamer is a particularly important structure, as it is con-sidered the smallest building block of ice.27 Water hexam-

a)Present address: Department of Chemistry, Brookhaven National Labora-tory, Upton, New York 11973, USA.

b)Authors to whom correspondence should be addressed. Electronic ad-dresses: angelos.michaelides@ucl.ac.uk and charles.sykes@tufts.edu.

ers consist of six water monomers arranged in a ring whereeach monomer accepts and donates a hydrogen bond. Thisring can be arranged in either a buckled or a planar manner.5

Each monomer in the planar arrangement binds at the sameheight above the surface, while the monomers in the buck-led hexamer are arranged at two discrete heights reminis-cent of the buckling found in the basal plane of ice. Theo-retical predictions of relative water hexamer stability on spe-cific surfaces have been reported; however, the experimen-tal images do not always match the predicted lowest-energystructure. While the buckled arrangement is predicted to bemore stable on Cu(111), hexamers on this surface appear pla-nar in STM images.19 Limitations with STM imaging mayproduce this apparent contradiction. It has been proposedthat the electric field from the STM tip reorients the watermolecules into a planar arrangement during the scan or theplanar hexamer image is a dynamic average of many struc-tures over the time-scale of the STM scan.9, 19, 28 The buckledhexamer of water has been imaged with STM on a hexag-onal boron nitride nanomesh on Rh(111),29 where it is pos-tulated that every second water monomer is bound weaklyto an underlying nitrogen atom through a hydrogen bond.30

Planar water hexamers have been investigated on Ru(0001)by low-energy electron diffraction, reflection adsorption IRspectroscopy and DFT, where it was found that Ru-water

0021-9606/2014/141(1)/014701/10/$30.00 © 2014 AIP Publishing LLC141, 014701-1

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

014701-2 Murphy et al. J. Chem. Phys. 141, 014701 (2014)

interactions dominate over hydrogen bonds resulting in planarhexamers.31

In contrast to the abundant work regarding adsorbed wa-ter, the adsorption of intact methanol has only recently be-gun to be investigated.32–35 Under UHV conditions, methanolmolecules adsorb on metal surfaces through the oxygen atomand can form two hydrogen bonds to other methanols viathe hydroxyl group. The resulting hydrogen-bonding betweenmonomers drives the assembly of chains and hexamers.33–35

At near monolayer coverages on Au(111) and Cu(111),methanol forms chains, which consist of hydrogen-bondedmolecules that are linearly arranged in the

√3 directions, re-

sulting in ordered domains across the whole surface. At lowercoverages methanol forms cyclic hexamers reminiscent ofwater hexamers. In contrast to water, the chiral footprint of theadsorbed methanol hexamers is apparent in STM imaging.33

Chiral planar methanol hexamers have been observed on bothAu(111) and Cu(111), while buckled surface-bound methanolhexamers have not been reported.33, 34

The relative stabilities of the various hydrogen-bondedoverlayers are dictated by the interplay of attractive molecule-molecule interactions, binding to the surface and surface cov-erage induced crowding/repulsion. It has been established forwater that the formation of either buckled or planar hexamersis directed by competition between molecule-molecule andmolecule-surface interactions that maximizes the binding en-ergy per monomer.5 The buckled hexamer produces a reducedinteraction with the surface for half of the molecules in favorof increased interactions between molecules, while the planarstructures optimize the interaction between all molecules andthe surface while sacrificing the ideal molecule-molecule hy-drogen bonding.19 The corresponding investigation for buck-led and planar methanol hexamers has not been performed.The lack of work in this area is because to date buckledmethanol hexamers have not been observed on surfaces. Pre-vious work on methanol on Au(111) and Cu(111) has in-vestigated hydrogen bonded chains and planar hexamers andshowed that the population distribution is dependent on sur-face coverage; the chain networks pack methanol monomersmore efficiently at higher coverages, while hexamers are morestable at low coverages. These networks relate to the struc-tures of solid and liquid methanol, in which the solid phasehas been shown to consist of very long hydrogen-bondedchains of methanol36 and the liquid phase is proposed to bethe dominated by hexamers.37 Methanol hydrogen bondednetwork formation on more reactive surfaces, such as Pt(111),has not been investigated.

To investigate the interplay of binding energy and sur-face coverage on network formation at the surface of close-packed transition metals, we have been engaged in joint STMand DFT studies of methanol adsorbed at various coverageson various transition metal surfaces. Here, we report resultsfor methanol on Au(111), Cu(111), and Pt(111) using a func-tional that accounts for van der Waals (vdW) interactions.These metals are important in catalysts for the oxidationand steam reforming of methanol, in direct methanol fuelcells, and as templates for self-assembly.38–41 Additionally,we provide new theoretical results for water hexamers, whichto the best of our knowledge, are the first vdW-inclusive

calculations of water hexamers adsorbed on close-packedtransition metal surfaces, and we compare these results to pre-vious experimental results for water and methanol hexamers.We report a new chiral buckled methanol hexamer on Cu(111)and Pt(111), which exists alongside the planar hexamers al-lowing us to study the STM tip induced inter-conversion ofthese structures. Our analysis presents insight into the inter-action, assembly, energetics, and chirality of methanol as afunction of coverage at the surface of catalytically relevantmetals.

II. EXPERIMENTAL

All experiments were performed with Omicron Nano-Technology variable-temperature (VT) or low-temperature(LT) ultra-high vacuum scanning tunneling microscopes. Datawere recorded with Veeco and Omicron etched W tips, aswell as mechanically cut 85:15 Pt–Ir and chemically etchedNi tips. The Ni tips were electro-chemically etched using a2 M KCl solution.42 All tips were washed in methanol be-fore being introduced to the UHV system. Once in vacuumthe tips were conditioned by applying voltage pulses whilescanning a clean surface. For methanol deposition 99.8+%ultrapure HPLC grade methanol from Alfa Aesar was used,with further purification via freeze–pump–thaw cycles. TheAu(111), Cu(111), and Pt(111) crystals were prepared withstandard Ar+ sputter (1.5 keV, 12 μA) and anneal (1000 K)cycles. The base pressure in the VT-STM setup, before cool-ing the STM stage with liquid helium via a flow cryostat,was <10−10 mbar. Pre-cooling the STM before the sam-ple was loaded decreased the pressure in the STM chamber<5 × 10−11 mbar. Annealing experiments were performed insitu in the VT-STM chamber with the STM tip held severalmm away from the surface. For LT experiments, the crys-tal was transferred in less than 5 min in vacuum <10−10

mbar to the pre-cooled STM stage after cleaning. The samplecooled from room temperature to 5 K in ∼40 min. Methanolwas deposited directly onto the 30 K (VT) or 5 K (LT) sur-face via a collimated molecular doser. Precise heating toprogressively higher temperatures induced partial desorption,rearrangement, and hence structural changes in the methanoloverlayer. The samples were then cooled to cryogenic temper-atures for imaging. The selection of heating temperatures wasguided by temperature programmed desorption experimentson Au(111), Cu(111), and Pt(111), which quantify the desorp-tion temperatures of multilayer and monolayers of methanolfrom each surface.43–45 In all of the STM scanning condi-tions, the voltage refers to the sample bias. Typical STM gapconditions for imaging methanol were between ±1.0 V and2–10 pA.

III. THEORETICAL METHODS

DFT calculations within the periodic supercell approachas implemented in the Vienna ab initio simulation package(VASP) were performed.46 The optB88-vdW functional47 wasused throughout. This is a modified version of the non-localvdW-density functional of Dion et al.48 which has been im-plemented in VASP by Klimes et al.49 and which shows

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

014701-3 Murphy et al. J. Chem. Phys. 141, 014701 (2014)

improved accuracy for a wide variety of systems.50–53 Theinner electrons were replaced by projector augmented wave(PAW) potentials,54 whereas the valence states were expandedin plane-waves with a cut-off energy of 500 eV. Adsorptioncalculations of monomers and hexamers of both water andmethanol were carried out considering (6 × 6) metal slabs cutalong the (111) direction consisting of 3 atomic layers thick-ness and separated by 18 Å of vacuum. Tests on a smaller (5× 5) unit cell show that the adsorption energy of MeOH hex-amers on Pt(111) is only 10 meV less than on a (6 × 6) unitcell; this suggests that a (6 × 6) unit cell is sufficiently largeto avoid spurious lateral adsorbate-adsorbate interactions.

In the case of adsorbed chains, a (7√

3 × √3) unit cell

was used. The metal atoms in the bottom layer were fixed tothe bulk optB88-vdW optimal positions (aCu = 3.623 Å, aAu

= 4.158 Å, and aPt = 3.978 Å), whereas all other atoms (inthe substrate and adsorbates) were allowed to relax. A relaxedbut otherwise unreconstructed (111) surface was used for allsubstrates, including Au. We used a Monkhorst-Pack k-pointsgrid of 2 × 2 × 1 and 1 × 7 × 1 for the (6 × 6) and (7

√3

× √3) unit cells, respectively. A dipole correction along the

direction perpendicular to the metal surface was applied, andgeometry optimizations were performed with a residual forcethreshold of 0.025 eV/Å. STM images were simulated usingthe Tersoff-Hamann approach,55 with a voltage of −100 mVand at a height of ∼6 Å above the metal surface.

Adsorption energies per molecule were computed asfollows:

Eads = E[m/M] − E[M] − nE[m]

n, (1)

where E[m/M] is the total energy of the adsorbed n methanolor water adstructure, E[M] is the total energy of the relaxedbare metal slab, and E[m] is the total energy of an isolatedgas phase methanol or water molecule. Favorable (exother-mic) adsorption corresponds, therefore, to negative adsorptionenergy.

In order to analyze the role of molecule-molecule andmolecule-metal interactions on adsorption, we have decom-posed Eads into these two contributions. We define themolecule-molecule contribution (essentially H-bonding inter-actions within the adsorbed molecules) to Eads as

EHBads = E[m̃] − nE[m]

n, (2)

where E [m̃] is the total energy of the methanol or water ad-structure in the absence of the metal substrate, but with allthe atoms fixed in the geometries they adopt in the adsorp-tion structure. The methanol-metal or water-metal contribu-tion, EmM

ads , is simply taken as the difference between Eads andEHB

ads . This particular energy decomposition scheme followsthe widely adopted approach used to aid understanding of hy-drogen bonded water adlayers on metal surfaces.56

IV. RESULTS AND DISCUSSION

In the following we report results from STM and DFTfor methanol on the (111) surfaces of Au, Cu, and Pt. First,we present results at near monolayer coverage (Sec. IV A),then submonolayer (Sec. IV B), and then low coverages

(Sec. IV C). This allows us to establish a qualitative “phasediagram” for adsorbed methanol on the three surfaces con-sidered. At low coverage we also discuss results fromadditional STM measurements in which interconversionsof various structures on Cu(111) were performed, includ-ing interconversions of the novel buckled chiral hexamerwe report here for the first time. We follow these results inSection V with a DFT-based analysis of the relative impor-tance of hydrogen bonding and adsorbate-substrate bondingin determining the methanol adsorption structures formed,and compare this to what we find for water adsorption on thesame surfaces.

A. Near monolayer coverage of methanol

Initial experiments focused on monolayer coverages ofmethanol on Au(111), Cu(111), and Pt(111). Between zeroand a complete monolayer, a number of methanol structuresare observed, which are outlined in Fig. 1. Monolayers andsubmonolayer amounts of methanol were prepared in the fol-lowing manner. Multilayers of methanol were deposited ontothe sample, followed by heating to the temperature indicatedin the respective figure (Figs. 2–4). This heating step des-orbed the excess methanol leaving the desired surface cov-erage. The samples were subsequently cooled to 30 K beforeimaging. At near monolayer coverages methanol forms longchains composed of many hundreds of molecules, in whicheach monomer is hydrogen bonded to two neighbors (Fig. 1A1). Overall, the methanol chains are observed to run alongthe

√3-axis of the underlying surface (Fig. 2), with Au(111)

being a special case as the local herringbone reconstruction

FIG. 1. Rows 1 and 2 show top and side-views of DFT optimized methanolstructures on a close-packed (111) metal surface. Row 3 shows the corre-sponding simulated STM images of these structures on the Cu(111) surface(tip height ≈ 6 Å, V = −100 mV). The yellow hexagonal mesh in the thirdrow indicates the underlying Cu lattice. Row 4 shows actual STM data ofmethanol on Cu(111) acquired at 5 K (−100 mV 100 pA. Scale bar = 1 nm).Column A reports data for an extended chain structure, B for a planar hex-amer, and C and D for two types of buckled hexamer. The buckled hexamer Cwas not observed experimentally by STM and so panel C4 is blank.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

014701-4 Murphy et al. J. Chem. Phys. 141, 014701 (2014)

FIG. 2. Near complete monolayers of methanol prepared on Au(111),Cu(111), and Pt(111). (a) On Au(111), methanol chains run in both com-pressed

√3 directions. The larger but fainter lines are the soliton walls of the

Au(111) surface 22 × √3 reconstruction. (b) On Cu(111), each bright line

is a pair of methanol chains running in the√

3 direction. (c) Shorter chainsare observed on Pt(111); however, unlike Au and Cu, long range orderingof the chains is not observed. Scan conditions: All images taken at 30 K.Au(111) −0.7 V, 5 pA; Cu(111) 0.1 V, 2 pA; and Pt(111) 0.1 V, 5 pA. Scalebars = 10 nm.

leads to a compression of two of the three√

3 axes of thatsurface (0.48 nm versus 0.50 nm).57, 58 On Au(111) themethanol chains orient along these compressed axes in orderto maximize the molecule-molecule interactions, and formdomains comprised of pairs of chains with a unit cell of 11× √

3. On Cu(111), methanol also forms double chains,where a staggering in the sets of chains results in Peierls-likedistortion and a unit cell of 6 × √

3.34 While short chains areobserved on Pt(111), large-scale ordering is not evident.

Turning now to DFT, our calculations predict that themethanol chains which form at high coverages have a sim-ilar structure on the three metals studied (Fig. 1(a)). Withinthe chains, individual methanol monomers bind at atop sitesof the underlying atomic lattice through one of the lone pairof electrons on the oxygen atom and are hydrogen bonded totwo neighbors in a zigzag fashion, resulting in a chain run-ning in a

√3 direction of the substrate lattice.34, 35 Our DFT

calculations predict that the adsorption energy per methanol inthe chain structure is −726 meV on Au(111), −745 meV onCu(111), and −844 meV on Pt(111) (see also Table I). Theseadsorption energies are ∼100–200 meV/methanol larger thanthe adsorption energies of the isolated methanol monomers,indicating that the chains are stable with respect to breakingup into individual adsorbed methanol molecules. The O-Ospacing may be expected to follow the metal-metal nearestneighbor distances, which in our DFT calculations are: 0.295nm for Au, 0.259 nm for Cu, and 0.281 nm for Pt. However,

FIG. 3. Samples heated to induce partial desorption of the methanol mono-layer. (a) On Au(111), methanol forms single chains and closed loops, withthe chains running predominately in the two compressed

√3 directions.

(b) On Cu(111), methanol forms double chains running in the√

3 directions.(c) On Pt(111), methanol forms short meandering chains as well as isolatedhexamers. Scan conditions: All images taken at 30 K. Au −0.8 V, 10 pA;Cu −1.0 V, 5 pA; and, Pt −0.1 V, 2 pA. Scale bars = 6 nm.

FIG. 4. Samples heated to induce near-total desorption of methanol.Methanol forms isolated hexamers after heating to 157 K on Au(111)(a) and 150 K on Cu(111) (b). After heating Pt(111) to 200 K, both buckledand planar hexamers are formed alongside disordered chains (c). Insets onAu and Cu show low coverage methanol experiments imaged at 5 K. Inset onPt shows buckled and planar hexamers performed at 30 K. Scan conditions:Au −0.8 V, 10 pA, Cu −1.0 V, 5 pA, Pt −0.1 V, 2 pA. Insets 3 × 4 nm2. Scalebars = 2 nm.

our DFT calculations show a compression of the O-O spac-ing of the linear chains on Au(111) (0.277 nm) and Pt(111)(0.271–0.274 nm), while the spacing is slightly expanded onCu(111) (0.261 nm). The O-O-O angle on Pt(111) (126.5◦)is likewise between that of Au(111) (132.5◦) and Cu(111)(113.9◦), with the larger angle producing a “straighter chain.”

B. Submonolayer coverages of methanol

Further experiments investigated submonolayer cover-ages of methanol. The samples were prepared by annealingmethanol multilayers to temperatures corresponding to par-tial desorption of the monolayer and then cooled to 30 K forimaging (Fig. 3). Desorption was evident by a reduction insurface coverage and the increased spacing between methanolchains. On Au(111), meandering closed loops comprised ofchains running predominately in the compressed

√3 direc-

tions are observed. On Cu(111), double chains are formedwith a unit cell of 7

√3 × √

3. The linear nature of the chainson both Au(111) and Cu(111) is evident. In contrast, methanolon Pt(111) displays a more disordered network, where bothisolated hexamers and short winding chains are present(Fig. 3(c)). Repulsive interactions between the methanolstructures are evident on each surface, as the size of the gapsbetween chains increases with decreasing coverage.

The structure of the methanol networks depends on thesubstrate and there is a notable lack of long range orderingon Pt(111) compared to Au(111) and Cu(111). A temptingexplanation for this is dehydrogenation of methanol by thereactive Pt(111) surface, which would lead to an absence ofavailable O-H groups and no opportunity for hydrogen bond-ing and hence a lack of driving force for network forma-tion. However, it has previously been shown that methanoladsorbs and desorbs reversibly from the terraces of Pt(111),45

with only trace amounts of methanol dehydrogenating at stepedges.59 Additionally, the short chain and hexamer structuresof methanol we observe on Pt(111) are consistent with boththe hydrogen-bonded structures predicted by DFT, and thoseobserved on Au(111) and Cu(111). These results demonstratethat dehydrogenation of the methanol by a reactive Pt(111)surface is not the cause of the lack of long range orderednetworks. Instead, we propose two potential mechanisms thatcontribute to this effect: the lower diffusion rate of methanol

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

014701-5 Murphy et al. J. Chem. Phys. 141, 014701 (2014)

TABLE I. Adsorption energies (in meV/molecule) of methanol and water hexamers and extended chains on Au(111), Cu(111), and Pt(111). The molecule-molecule (H-bonding, EHB

ads ) and molecule-metal (EmMads ) contributions to the total adsorption energy (Eads) are also provided. See the text for precise definitions

of each of these terms.

Au(111) Cu(111) Pt(111)

Eads EHBads EmM

ads Eads EHBads EmM

ads Eads EHBads EmM

ads

Methanol Monomer −436 0 −436 −554 0 −554 −733 0 −733Extended chain (A) −726 −362 −364 −745 −378 −367 −844 −368 −476Planar hexamer (B) −726 −368 −358 −727 −363 −364 −844 −302 −543Buckled hexamer (C) −657 −388 −269 −669 −386 −283 −775 −358 −417Buckled hexamer (D) −638 −383 −255 −665 −379 −286 −761 −305 −456

Water Monomer −276 0 −276 −397 0 −397 −498 0 −498Planar hexamer . . . . . . . . . . . . . . . . . . −631 −298 −333Buckled hexamer −547 333 −214 −589 −323 −266 −668 −304 −364Extended chain −541 −307 −234 −561 −294 −267 −630 −307 −323

on Pt(111) and the lower total relative energy of the chainscompared to the monomer. The barrier to diffusion ofmethanol on Pt(111) is expected to be higher than thaton Au(111) or Cu(111) since methanol monomers bindmore strongly to Pt(111) (−733 meV per methanol) than toAu(111) (−436 meV per methanol) or Cu(111) (−554 meVper methanol). This would result in a reduced diffusion rateof methanol on Pt(111) which hinders the ability of the sys-tem to sample all possible microstates and obtain the globallowest-energy structure. Additionally, the driving force toform chains is significantly lower on Pt(111) than eitherAu(111) or Cu(111). DFT calculations show that upon form-ing adsorbed chains methanol monomers are stabilized by290, 190, and 110 meV per methanol on Au(111), Cu(111),and Pt(111), respectively. (This stabilization energy is sim-ply the difference in adsorption energy between methanol ad-sorbed as monomers and adsorbed in chains.) The lower sta-bilizing effect on Pt(111) results in a decreased driving forceto form long range ordered structures. This, along with theslower methanol diffusion expected on Pt, most likely ex-plains the lack of long range ordering of hydrogen-bondedmethanol chains on Pt(111).

C. Low coverage methanol: Hexamers and shortchains and STM induced interconversions

In the next set of experiments, the samples were heatedto a temperature required to induce near total desorptionof methanol then cooled to 30 or 5 K prior to imaging(Fig. 4). The methanol remaining on the surface formedhydrogen-bonded hexamers that were highly dispersed on thesurface as seen in Fig. 4. The high degree of dispersion in-dicates a repulsive interaction between the hexamers. As wediscuss in more detail below, the hexamers consist of sixmethanol monomers arranged in a cyclic manner, in whicheach monomer donates and accepts one hydrogen bond. Thehexamer geometry allows each methanol molecule within thecluster to bind at atop positions of the underlying substrate.The hydrogen-bonded network has either a counterclockwiseor a clockwise rotation, resulting in two surface-bound enan-tiomers of the hexamers. Planar hexamers are observed onAu(111), Cu(111), and Pt(111), and a second form is observed

simultaneously on Cu(111) and Pt(111). The STM image ofthis second form resembles the buckled form of water pre-dicted by DFT and observed on the hexagonal boron nitridenanomesh on Rh(111).29

Samples with coexisting planar and buckled methanolhexamers were observed on Cu(111) and Pt(111) and offereda unique opportunity to interrogate their structure and inter-conversion which has been impossible for water to date. OnCu(111), buckled and planar hexamers and short chains wereprepared by adsorption of a small amount (0.2 Langmuir) ofmethanol on a clean Cu(111) surface at 5 K, followed by an-nealing to 20 K and cooling to 5 K (Fig. 5). On Pt(111), theywere prepared by dosing multilayers of methanol, heating to200 K, and cooling to 30 K (Fig. 4(c) and inset of same). Iso-lated methanol hexamers are observed on Pt(111) (Fig. 4(c))alongside chains, and separation between the two features in-dicates repulsive interactions between them.

By combining our STM measurements with DFT calcu-lations we have been able to determine the structures of thevarious hexamers. The DFT calculations were performed ona number of hexamer structures including buckled and planarforms. The most stable structures resulting from these calcu-lations are shown in Fig. 1. The previously discussed planarstructure (Fig. 1(b)) and one of the buckled hexamer models(Fig. 1(d)) produce simulated STM images that are consis-tent with the experimental data. One of the buckled hexamers(Fig. 1(c)) was simulated by DFT but not observed experi-mentally. In the planar hexamer, the methyl groups are ar-ranged coplanar to the surface. Each methanol binds to anatop site of the substrate and the methanol-methanol hydro-gen bond distance between neighboring molecules is equiv-alent. The structure of both types of buckled hexamers dif-fers from the planar hexamer in a number of ways. First,in the buckled hexamers there are two alternating types ofmethanol molecules, arranged so that their methyl groups areeither coplanar or perpendicular to the surface. The copla-nar methanols are ∼0.7–0.9 Å closer to the surface thanthe methanols that have the methyl groups directed awayfrom the surface. Second, in the buckled hexamers bindingto the surface is principally through the oxygen atoms of thethree lower methanol molecules, whereas in the planar hex-amer binding occurs through all six methanols. The third key

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

014701-6 Murphy et al. J. Chem. Phys. 141, 014701 (2014)

FIG. 5. Results of STM tip induced interconversion of methanol structureson Cu(111) as described in the text. A yellow arrow represents a chiralityinversion. A green arrow represents a ring-flip interconversion. A combinedgreen and yellow arrow represents an interconversion in which both a chiral-ity interconversion and ring-flip occur. The panels (a)-(c) show the progres-sion of the circled buckled hexamer to other forms; first through a ring-flip(a)-(b) and finally opening to a six-membered chain (b)-(c). Also seen is thediffusion and reorientation of other short chains. The more stable planar hex-amer at the top of panels (a)-(c) serves as frame of reference. Imaging con-ditions: All images taken at 5 K, −100 mV, 100 pA. Perturbative conditions:400 mV, 100 pA. Scale bar = 2 nm.

difference is that in the buckled hexamers the methanol-methanol separations within the hexamer (measured, e.g.,through the O-O separation between neighboring methanols)are not equivalent and alternate as one moves around the ring.As we discuss below the extent of the alteration in O-O dis-tances varies from one substrate to the next from as little as0.05 Å on Au to 0.18 Å on Pt. A similar alteration in O-O dis-tances has been reported before for buckled water hexamers.19

Comparing the two types of buckled hexamers, buckled hex-amer C (Fig. 1(c)) has each methanol molecule close to anatop site (i.e., similar sites to the planar hexamer), whereasin the second buckled hexamer (Fig. 1(d)) the methanols areclose to bridge sites. As a result the center of mass of buckledhexamer C is located above an atop site, whereas for hexamerD it is above a threefold hollow site. We think this differencein binding position on the surface is relevant to why only onetype of buckled hexamer is observed experimentally, as wediscuss below.

The adsorption energies of the various hexamers on eachmetal surface are summarized in Table I. Binding methanolmonomers together in a hexamer stabilizes the monomersand results in an increase in the adsorption energy (com-pare Eads of monomers and hexamers in Table I). On all sur-faces the planar hexamers are more stable than the buckled

hexamers by ∼60–90 meV/methanol. STM imaging supportsthis conclusion, as we observe that the planar hexamers re-main on the surface after all other structures have desorbed(Fig. 4). In order to rationalize the greater stability of theplanar configurations, we decomposed the total adsorptionenergy into hydrogen bonding between methanol molecules(EHB

ads ) and the binding of individual methanol molecules tothe metal surface(EmM

ads ) as given by Eq. (2). This revealsthat the planar hexamers offer the best trade-off between themethanol-methanol and methanol-metal interactions, leadingto the strongest overall binding. While the buckled hexamerresults in an increased methanol-methanol component com-pared to planar hexamers, the decrease in the methanol-metalinteraction is more significant, leading to an overall decreasein stability. On Pt(111), the molecule-metal contribution dom-inates the binding of both the planar and buckled hexamers,which is in line with the behavior expected of more reactivesurfaces. Interestingly, our DFT calculations indicate that un-like the O-O spacings of the chain, the O-O spacing of theplanar hexamers is almost identical regardless of the metalsubstrate: 0.268 nm on Au(111), 0.265 nm on Cu(111), and0.268 on Pt(111). In our DFT calculations the correspondingmetal-metal nearest neighbor distances are: 0.295 nm for Au,0.259 nm for Cu, and 0.281 nm for Pt. The fact that theH-bond lengths remain relatively unaltered when changingthe substrate suggests that methanol-metal interactions are aweaker function of total energy than the methanol-methanolinteractions when balancing these two contributions to theoverall stability of the cluster.

Each lobe of the planar hexamers appears identically inSTM and DFT simulated images resulting in C6 symmetry.For the buckled hexamers, the relative arrangement of per-pendicular and parallel methyl groups image as brighter anddarker lobes, respectively, and give rise to the appearanceof two interlocking triangles. The resulting buckled hexam-ers exhibit C3 symmetry. By measuring the rotation of thebrighter lobes relative to the close packed direction of the un-derlying atomic lattice, we have determined that only one ofthe buckled hexamers predicted by theory is present on thesurface (Fig. 1(d)). STM line scans show the relative heightsof the features corresponding to the perpendicular vs. planarmethyl groups on opposite sides of the hexamer (Fig. 6(a)).

The methanol hexamers are comprised of homochiralsurface-bound methanol monomers that assemble into a struc-ture which is itself rotated from the high symmetry substratedirections and hence is chiral. As supported by our DFT cal-culations, reflective symmetry dictated by the direction of thehydrogen-bonding network gives rise to two surface-boundenantiomers. The chirality of both planar and buckled hex-amers is apparent in both STM data and DFT calculations asa rotation of the hexamers as a whole from the close packedsubstrate directions. The planar hexamers on Cu(111) are ro-tated ±5◦ depending on chirality.33 DFT predicts a rotation of±3.7◦ for the buckled hexamer which is in good agreementwith the value of 3.8 ± 0.7◦ measured by STM (Fig. 6(b)).Rotation of the buckled hexamer in the plane of the surfacegives rise to structures where the apex of the brighter trian-gle of lobes corresponding to the perpendicular methyl groupspoints either up or down in the STM images.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

014701-7 Murphy et al. J. Chem. Phys. 141, 014701 (2014)

FIG. 6. Structural features of the buckled methanol hexamer on Cu(111).(a) Three separate STM line scans over opposing lobes, indicating the relativeheight of the features. Inset shows corresponding section of the hexamer eachline scan was taken over. (b) Rotational offset of the buckled hexamer fromthe high symmetry directions of the underlying Cu(111) lattice as measuredby STM. Positive angles represent counterclockwise rotation. Insets show Rand S enantiomers of the buckled hexamers.

The presence or absence of specific hydrogen bondednetworks allows us to speculate on the barriers to conver-sions between them. The coexistence of the planar hexamer,which is centered around an atop site (Fig. 1(b)), and thebuckled hexamer, which is centered at a threefold site (Fig.1(d)), implies that the observed buckled hexamer is a trappedmetastable structure with a barrier to conversion to the planarhexamer. Additionally, the absence of the buckled hexamercentered at an atop site (Fig. 1(c)) suggests a small barrieror non-activated process to conversion between this type ofbuckled hexamer and the more stable planar hexamer. Theconversion from the buckled hexamer centered at a three-fold hollow site (Fig. 1(d)) to the planar hexamer requiresmovement (diffusion) of the hexamer as a whole from beingcentered at a three-fold hollow surface site to an atop site,together with the movement and rotation of three methanolmonomers towards the surface. In contrast, the conversionfrom the buckled hexamer centered at the atop site (Fig. 1(c))to planar hexamer (Fig. 1(b)) does not require movement to anew binding site, as both are centered at atop sites. The factthat one buckled hexamer has to change adsorption sites is,

therefore, a likely source of this barrier between the buckled(Fig. 1(d)) and planar (Fig. 1(b)) hexamers. The buckled hex-amer (Fig. 1(c)) can be considered a short-lived metastablestructure that rapidly rearranges into the more energeti-cally favorable planar hexamer through an almost barrierlessprocess.

Low temperature (5 K) STM allowed us to investigatethe interconversion of these different types and chiralities ofmethanol structures as shown in Fig. 5. The stability and re-ordering of the structures on Cu(111) was investigated byrastering the STM tip over the surface at perturbative condi-tions (400 mV, 100 pA) between non-perturbative observationscans. A number of conformational changes of the hydrogen-bonded methanol networks were observed, as outlined inFigure 5. Chirality inversions of both the buckled and pla-nar hexamer were also seen (yellow arrows, Fig. 5). Thebuckled hexamer has a unique conformational interconver-sion, analogous to the ring-flip in cyclohexane, which re-sults in the upward pointing methyl groups pointing downand vice versa (green arrows, Fig. 5). This ring-flip resultsin an apparent rotation of the triangles relative to the substratelattice. The green/yellow diagonal arrows represent intercon-versions, where both a chirality and ring-flip occur. Rear-rangement and breaking of the hydrogen-bonds can occur, al-lowing the buckled hexamer to form a six-membered chain,and the six-membered chain and planar hexamers to be inter-converted (gray arrows). The formation of a buckled hexamerfrom either a planar hexamer or chain was not observed; thisimplies that the buckled hexamer is less stable than either theplanar hexamer or chain, which is supported by our theoreti-cal calculations, in which chains and planar hexamers, respec-tively, are 76 and 58 meV per methanol more energeticallyfavorable than the buckled hexamer (Fig. 1(d)), see Table I.

V. COMPARISON OF WATER ANDMETHANOL NETWORKS

Finally, before concluding, it is interesting to comparewhat we find for methanol adsorption with what is knownabout water. First, like methanol, water has been observedto form clusters (including hexamers) and chains when ad-sorbed on several metals.5 However, water also forms two-dimensional overlayer structures, such as on Pt(111) where anextended overlayer built from hexagons, pentagons, and hep-tagons has been observed.60 We found no evidence for anytwo-dimensional hydrogen bonded overlayers of methanol onPt(111) or on either of the two other surfaces consideredhere. Given that solid methanol is built from methanol chainsthe absence of ordered two-dimensional hydrogen bondedmethanol structures is not unexpected.36, 61, 62

If we focus now on comparing the methanol and waterhexamer structures adopted on the different metals we findthat this depends on a balance between molecule-moleculeand molecule-metal interactions. For methanol, the molecule-metal interactions play the dominant role. In contrast, withwater, preserving a strong hydrogen bond network pre-vails over forming water-metal bonds since the water-metalinteraction is weaker than the methanol-metal interaction(Table I). Our STM and theoretical analysis indicate that

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

014701-8 Murphy et al. J. Chem. Phys. 141, 014701 (2014)

TABLE II. Adsorption geometries (in Å) of methanol and water buckled hexamers on Au(111), Cu(111), and Pt(111). The six dO-M heights are defined as thedistance between the O atom of each monomer within the hexamer and its corresponding nearest metal atom in the surface layer. The six nearest-neighbor O-Odistances (dO-O) alternate between two characteristic values as discussed in the text and the difference between them is also provided (�O-O).

Au(111) Cu(111) Pt(111)

dO-M dO-O �O-O dO-M dO-O �O-O dO-M dO-O �O-O

Methanol hexamer (C) 2.68 2.65 0.05 2.34 2.62 0.07 2.27 2.56 0.182.68 2.65 2.35 2.62 2.27 2.592.68 2.65 2.36 2.62 2.28 2.593.36 2.70 3.26 2.69 3.20 2.763.38 2.70 3.29 2.69 3.22 2.763.42 2.70 3.32 2.69 3.24 2.77

Water hexamer 2.70 2.66 0.08 2.32 2.65 0.13 2.31 2.63 0.242.70 2.67 2.32 2.65 2.31 2.632.72 2.67 2.32 2.65 2.31 2.633.22 2.75 3.09 2.78 3.14 2.873.23 2.75 3.12 2.75 3.14 2.873.24 2.75 3.13 2.79 3.15 2.88

planar methanol hexamers are more stable on each metalstudied (Table I). Additionally, our theoretical work indicatesthat planar water hexamers are preferred on reactive metalssuch as Pt(111) (Table I), while buckled hexamers are pre-ferred on less reactive metals, which is in agreement withprevious experimental and theoretical results.29, 31 To date,buckled and planar water hexamers have not been observedin coexistence in the same experiment. Conversely, our STManalysis shows buckled and planar methanol hexamers simul-taneously on Cu(111) and Pt(111), which has allowed for theanalysis of interconversions between these hydrogen bondedstructures that has not yet been possible for water systems.

Let us focus now on the buckling in the water andmethanol hexamers. Interestingly, when we look at the struc-ture of both the buckled methanol and water hexamers we findthat in addition to the buckling there is an alteration in theO-O distances, with a set of relatively short O-O distances anda set of relatively long O-O distances (Table II). This “kekulé-like” alteration in O-O distances has been predicted before forwater, but is observed here for methanol with DFT for the firsttime. The alterations in O-O distances arise from a competi-

FIG. 7. Alteration in O-O distances within buckled hexamers (�O-O) andmonomer adsorption energy (Eads) for both water and methanol on Au(111),Cu(111), and Pt(111).

tion between specific lone pair orbitals of water and methanolthat are involved in bonding with the surface and in acceptinghydrogen bonds.63 The result is a buckled structure that canbe viewed as a hexamer built from three weakly interactingdimers. When we compare the extent of the alteration in O-Odistances we find that this is actually related to how stronglythe molecules bond to the surface. We show this in Fig. 7where we plot the extent of the alteration in O-O distances(�O-O) as a function of monomer adsorption energy for bothwater and methanol. Clearly as the strength of the interactionincreases, so too does the asymmetry in the hexamer.

VI. CONCLUSION

We have investigated coverage-dependent networks ofmethanol from near monolayer to extremely low coverage onAu(111), Cu(111), and Pt(111) with STM and DFT, showingcoverage and surface dependence on the networks formed. Inparticular, the stronger adsorbate-substrate binding of Pt(111)hinders the formation of long range ordered methanol net-works. A new structure formed by surface-bound methanol onCu(111) and Pt(111), which we assign, through STM imag-ing and DFT calculations, as a buckled hexamer of hydrogen-bonded methanol molecules has been identified. The chiralityof these structures is expressed in a rotational effect which al-lows us to assign enantiomers. This structure is similar to thebuckled water hexamers that are hypothesized to exist on lessreactive metals. We observe buckled and planar methanol hex-amers simultaneously on two separate metals, and using DFTwe determine that the planar methanol hexamer is preferredon all investigated surfaces, as it produces the optimal balancebetween hydrogen bonding and the more dominant methanol-metal interactions. In order to compare methanol and waterhexamer adsorption we did the first set of vdW-inclusive cal-culations of adsorbed water hexamers. This revealed that forthe water hexamers, hydrogen bonding interactions generallyplay a larger role, and drive the formation of buckled hex-amers, and only on more reactive metals is the molecule-metal interaction large enough to force a preference for pla-nar hexamers. These results serve as a cross comparison of

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

014701-9 Murphy et al. J. Chem. Phys. 141, 014701 (2014)

the interaction strength, assembly, and chirality of hydrogenbond networks on catalytically relevant close packed metalsurfaces.

ACKNOWLEDGMENTS

The work at Tufts was supported by the National Sci-ence Foundation (NSF) (CBET-1159882). T.J.L. acknowl-edges support from the U.S. Department of Energy (DOE)(DE-FG02-05ER15730) and E.A.L. thanks the Departmentof Education for a GAANN Fellowship. We are gratefulfor support from the FP7 Marie Curie Actions of the Euro-pean Commission, via the Career Integration Grant (GrantNo. FP7-PEOPLE-2011-CIG NanoWGS). J.C. is a Ramón yCajal fellow and Newton Alumnus supported by the Span-ish Government and The Royal Society. A.M. is supportedby the European Research Council and the Royal Societythrough a Royal Society Wolfson Research Merit Award. Weare grateful for computer time to the Barcelona Supercomput-ing Center, i2BASQUE, UCL Research Computing, the Lon-don Centre for Nanotechnology, and the UK’s national highperformance computing service HECToR (from which accesswas obtained via the UK’s Material Chemistry Consortium,EP/F067496).

1D. R. Palo, R. A. Dagle, and J. D. Holladay, “Methanol steam reformingfor hydrogen production,” Chem. Rev. 107, 3992–4021 (2007).

2K. Scott and A. Shukla, “Direct methanol fuel cells: Fundamentals, prob-lems and perspectives,” Mod. Asp. Electrochem. 40, 127–227 (2007).

3J. Sass, K. Kretzschmar, and S. Holloway, “Water adsorption on metal sur-faces: an electrochemical viewpoint,” Vacuum 31, 483–486 (1981).

4Q. Zhang, J. Zheng, A. Shevade, L. Zhang, S. H. Gehrke, G. S. Heffelfinger,and S. Jiang, “Transport diffusion of liquid water and methanol throughmembranes,” J. Chem. Phys. 117, 808 (2002).

5J. Carrasco, A. Hodgson, and A. Michaelides, “A molecular perspective ofwater at metal interfaces,” Nat. Mater. 11, 667–674 (2012).

6M. Tatarkhanov, D. F. Ogletree, F. Rose, T. Mitsui, E. Fomin, S. Maier,M. Rose, J. I. Cerdá, and M. Salmeron, “Metal- and hydrogen-bondingcompetition during water adsorption on Pd(111) and Ru(0001),” J. Am.Chem. Soc. 131, 18425–18434 (2009).

7T. Kumagai, M. Kaizu, S. Hatta, H. Okuyama, and T. Aruga, “Direct ob-servation of hydrogen-bond exchange within a single water dimer,” Phys.Rev. Lett. 100, 166101 (2008).

8H. Ogasawara, B. Brena, D. Nordlund, M. Nyberg, A. Pelmenschikov, L.Pettersson, and A. Nilsson, “Structure and bonding of water on Pt(111),”Phys. Rev. Lett. 89, 276102 (2002).

9J. S. Filhol and M. L. Bocquet, “Charge control of the water monolayer/Pdinterface,” Chem. Phys. Lett. 438, 203–207 (2007).

10H. Okuyama and I. Hamada, “Hydrogen-bond imaging and engineeringwith a scanning tunnelling microscope,” J. Phys. D 44, 464004 (2011).

11S. Meng, E. Wang, and S. Gao, “Water adsorption on metal surfaces: Ageneral picture from density functional theory studies,” Phys. Rev. B 69,195404 (2004).

12P. A. Thiel and T. E. Madey, “The interaction of water with solid surfaces:Fundamental aspects,” Surf. Sci. Rep. 7, 211–385 (1987).

13M. Henderson, “The interaction of water with solid surfaces: Fundamentalaspects revisited,” Surf. Sci. Rep. 46, 1–308 (2002).

14A. Hodgson and S. Haq, “Water adsorption and the wetting of metal sur-faces,” Surf. Sci. Rep. 64, 381–451 (2009).

15A. Verdaguer, G. M. Sacha, H. Bluhm, and M. Salmeron, “Molecular struc-ture of water at interfaces: Wetting at the nanometer scale,” Chem. Rev.106, 1478–1510 (2006).

16T. Schiros, K. J. Andersson, L. G. M. Pettersson, A. Nilsson, and H. Oga-sawara, “Chemical bonding of water to metal surfaces studied with core-level spectroscopies,” J. Electron Spectrosc. Relat. Phenom. 177, 85–98(2010).

17A. Michaelides, “Density functional theory simulations of water–metal in-terfaces: Waltzing waters, a novel 2D ice phase, and more,” Appl. Phys. A85, 415–425 (2006).

18P. J. Feibelman, “Concluding remarks,” Faraday Discuss. 141, 467 (2009).19A. Michaelides and K. Morgenstern, “Ice nanoclusters at hydrophobic

metal surfaces,” Nat. Mater. 6, 597–601 (2007).20G. Jones, S. J. Jenkins, and D. A. King, “Hydrogen bonds at metal surfaces:

Universal scaling and quantification of substrate effects,” Surf. Sci. 600,224–228 (2006).

21G. Jones and S. J. Jenkins, “Water and ammonia on Cu{110}: Comparativestructure and bonding,” Phys. Chem. Chem. Phys. 15, 4785–4798 (2013).

22K. Morgenstern and K.-H. Rieder, “Formation of the cyclic ice hexamervia excitation of vibrational molecular modes by the scanning tunnelingmicroscope,” J. Chem. Phys. 116, 5746 (2002).

23J. Lee, D. C. Sorescu, K. D. Jordan, and J. T. Yates, “Hydroxyl chain forma-tion on the Cu(110) surface: Watching water dissociation,” J. Phys. Chem.C 112, 17672–17677 (2008).

24K. Motobayashi, C. Matsumoto, Y. Kim, and M. Kawai, “Vibrational studyof water dimers on Pt(111) using a scanning tunneling microscope,” Surf.Sci. 602, 3136–3139 (2008).

25T. Mitsui, M. K. Rose, E. Fomin, D. F. Ogletree, and M. Salmeron, “Waterdiffusion and clustering on Pd(111),” Science 297, 1850–1852 (2002).

26L. Andersohn and U. Köhler, “In situ observation of water adsorptionon Si(100) with scanning tunneling microscopy,” Surf. Sci. 284, 77–90(1993).

27K. Nauta, “Formation of cyclic water hexamer in liquid helium: The small-est piece of ice,” Science 287, 293–295 (2000).

28H. Gawronski, K. Morgenstern, and K. H. Rieder, “Electronic excitation ofice monomers on Au(111) by scanning tunneling microscopy - Vibrationalspectra and induced processes,” Eur. Phys. J. D 35, 349–353 (2005).

29H. Ma, Y. Ding, M. Iannuzzi, T. Brugger, S. Berner, J. Hutter, J. Oster-walder, and T. Greber, “Chiral distortion of confined ice oligomers (N =5,6),” Langmuir 28, 15246–15250 (2012).

30Y. Ding, M. Iannuzzi, and J. Hutter, “Investigation of boron nitridenanomesh interacting with water,” J. Phys. Chem. C 115, 13685–13692(2011).

31S. Haq, C. Clay, G. Darling, G. Zimbitas, and A. Hodgson, “Growth ofintact water ice on Ru(0001) between 140 and 160K: Experiment anddensity-functional theory calculations,” Phys. Rev. B 73, 115414 (2006).

32E. M. Karp, T. L. Silbaugh, M. C. Crowe, and C. T. Campbell, “Energeticsof adsorbed methanol and methoxy on Pt(111) by microcalorimetry,” J.Am. Chem. Soc. 134, 20388–20395 (2012).

33T. J. Lawton, J. Carrasco, A. E. Baber, A. Michaelides, and E. C. H. Sykes,“Visualization of hydrogen bonding and associated chirality in methanolhexamers,” Phys. Rev. Lett. 107, 256101 (2011).

34T. J. Lawton, J. Carrasco, A. E. Baber, A. Michaelides, and E. C. H. Sykes,“Hydrogen-bonded assembly of methanol on Cu(111),” Phys. Chem.Chem. Phys. 14, 11846–11852 (2012).

35A. E. Baber, T. J. Lawton, and E. C. H. Sykes, “Hydrogen-bonded networksin surface-bound methanol,” J. Phys. Chem. C 115, 9157–9163 (2011).

36K. J. Tauer and W. N. Lipscomb, “On the crystal structures, residual en-tropy and dielectric anomaly of methanol,” Acta Crystallogr. 5, 606–612(1952).

37S. Sarkar and R. N. Joarder, Molecular clusters and correlations in liquidmethanol at room temperature,” J. Chem. Phys. 99, 2032 (1993).

38B. P. C. Hereijgers and B. M. Weckhuysen, “Selective oxidation ofmethanol to hydrogen over gold catalysts promoted by alkaline-earth-metaland lanthanum oxides,” ChemSusChem 2, 743–748 (2009).

39S. Sá, H. Silva, L. Brandão, J. M. Sousa, and A. Mendes, “Catalysts formethanol steam reforming—A review,” Appl. Catal. B Environ. 99, 43(2010).

40B. L. García and J. W. Weidner, “Review of direct methanol fuel cells,”Mod. Aspects Electrochem. 40, 229–284 (2007).

41X.-K. Gu and W.-X. Li, “First-principles study on the origin of the differentselectivities for methanol steam reforming on Cu(111) and Pd(111),” J.Phys. Chem. C 114, 21539–21547 (2010).

42M. Cavallini and F. Biscarini, “Electrochemically etched nickel tips forspin polarized scanning tunneling microscopy,” Rev. Sci. Instrum. 71, 4457(2000).

43J. Gong, D. W. Flaherty, R. A. Ojifinni, J. M. White, and C. B. Mullins,“Surface chemistry of methanol on clean and atomic oxygen pre-coveredAu(111),” J. Phys. Chem. C 112, 5501–5509 (2008).

44S. M. Johnston, A. Mulligan, V. Dhanak, and M. Kadodwala, “The structureof methanol and methoxy on Cu(111),” Surf. Sci. 530, 111–119 (2003).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23

014701-10 Murphy et al. J. Chem. Phys. 141, 014701 (2014)

45O. Skoplyak, C. A. Menning, M. A. Barteau, and J. G. Chen, “Experimentaland theoretical study of reactivity trends for methanol on Co/Pt(111) andNi/Pt(111) bimetallic surfaces,” J. Chem. Phys. 127, 114707 (2007).

46G. Kresse and J. Furthmüller, “Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set,” Phys. Rev. B 54, 11169–11186 (1996).

47J. Klimeš, D. R. Bowler, and A. Michaelides, “Chemical accuracy for thevan der Waals density functional,” J. Phys. Condens. Matter 22, 022201(2010).

48M. Dion, H. Rydberg, E. Schröder, D. C. Langreth, and B. I. Lundqvist,“Van der Waals density functional for general geometries,” Phys. Rev. Lett.92, 246401 (2004).

49J. Klimeš, D. R. Bowler, and A. Michaelides, “Van der Waals density func-tionals applied to solids,” Phys. Rev. B 83, 195131 (2011).

50J. Carrasco, W. Liu, A. Michaelides, and A. Tkatchenko, “Insight into thedescription of van der Waals forces for benzene adsorption on transitionmetal (111) surfaces,” J. Chem. Phys. 140, 084704 (2014).

51B. Santra, J. Klimes, A. Tkatchenko, D. Alfè, B. Slater, A. Michaelides, R.Car, and M. Scheffler, “On the accuracy of van der Waals inclusive density-functional theory exchange-correlation functionals for ice at ambient andhigh pressures,” J. Chem. Phys. 139, 154702 (2013).

52W. Liu, J. Carrasco, B. Santra, A. Michaelides, M. Scheffler, and A.Tkatchenko, “Benzene adsorbed on metals: Concerted effect of covalencyand van der Waals bonding,” Phys. Rev. B 86, 245405 (2012).

53W. Lew, M. C. Crowe, C. T. Campbell, J. Carrasco, and A. Michaelides,“The energy of hydroxyl coadsorbed with water on Pt(111),” J. Phys.Chem. C 115, 23008–23012 (2011).

54G. Kresse and D. Joubert, “From ultrasoft pseudopotentials to the projectoraugmented-wave method,” Phys. Rev. B 59, 1758–1775 (1999).

55J. Tersoff and D. R. Hamann, “Theory and application for the scanningtunneling microscope,” Phys. Rev. Lett. 50, 1998–2001 (1983).

56A. Michaelides, A. Alavi, and D. King, “Insight into H2O-ice adsorptionand dissociation on metal surfaces from first-principles simulations,” Phys.Rev. B 69, 113404 (2004).

57J. Barth, H. Brune, G. Ertl, and R. Behm, “Scanning tunneling microscopyobservations on the reconstructed Au(111) surface: Atomic structure, long-range superstructure, rotational domains, and surface defects.” Phys. Rev.B 42, 9307–9318 (1990).

58C. Wöll, S. Chiang, R. J. Wilson, P. H. Lippel, and C. Woll, “Determinationof atom positions at stacking-fault dislocations on Au (111) by scanningtunneling microscopy,” Phys. Rev. B 39, 7988–7991 (1989).

59K. D. Gibson and L. H. Dubois, “Step effects in the thermal decompositionof methanol on Pt(111),” Surf. Sci. 233, 59–64 (1990).

60S. Nie, P. J. Feibelman, N. C. Bartelt, and K. Thürmer, “Pentagons andheptagons in the first water layer on Pt(111),” Phys. Rev. Lett. 105, 026102(2010).

61O. Gálvez, B. Maté, B. Martín-Llorente, V. J. Herrero, and R. Escrib-ano, “Phases of solid methanol,” J. Phys. Chem. A 113, 3321–3329(2009).

62B. H. Torrie, S.-X. Weng, and B. M. Powell, “Structure of the A-phase ofsolid methanol,” Mol. Phys. 67, 575–581 (1989).

63A. Michaelides, “Simulating ice nucleation, one molecule at a time, withthe ‘DFT microscope’,” Faraday Discuss. 136, 287–297 (2007); 136,309–328 (2007).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

128.40.199.143 On: Wed, 02 Jul 2014 09:19:23