Downloaded by guest on September 30, 2021 hssrto fgas-phase of physisorption inter- reactions. carbon-based initiate transient may the that species illustrating mediate CO CO 1, of the Fig. reactions in dissociate surface reported various first the to comprehensive of interact a overview and lack surface we Cu CO However, how of (1). understanding storage energy solar environment reaction CO same for solvent the in (H water needed with (2), is CO CO protons reduce of of To steps source 4). initial (3, the struc- on selectivity surface chemistry Cu of and role ture the understanding consider- in remain uncertainties there able studies, theoretical and experimental ous CO alcohol and hydrocarbon CO over- (2). reduce an to high ethanol) V) too (>1 requires potential and hydrogen, produces ethylene, also Cu methane, However, valu- (e.g., of formation hydrocarbons catalyze to able ability unique the with of metals, pure initiation promote to is step first steps. reaction the mech- Here activation complete and involved. a adsorption anisms of the development of the discov- understanding sus- require atomistic will that for catalyst consider a challenge We such important ering hydrocarbons. of an production catalyst a tainable scalable of and identification sustain- makes suitable a This CO thermodynamically. creating the unfavorable to inert, path highly CO because is However, clear (1). cycle a energy hydrocarbon-based able provide would chemicals T theory functional density how CO into catalysts. insights neutral reduction fresh dioxide both carbon providing to improved sites, design leads to surface suboxide Cu optimum charged CO This and toward CO. step and CO first formate chemisorbed essen- the the is as suboxide to water this converting that CO show for the we the tial and below bind K, structure 298 to suboxide at essential configuration thin is a surface of copper presence we the energies, free that and structures show the of prediction mechanical tum quan- ambient with Using interpreted catalysts. spectroscopy CO photoelectron current a X-ray best of pressure the obtaining steps surfaces, by copper initial on catalysts the tion new of discover understanding to detailed aim are electrocatalysts we current adequate, Because not fuels. liquid as such products otiue yWlimA odr I,My9 07(etfrrve aur 6 07 eiwdb hre .Cmbl n rc .Koel) E. Bruce CO and Campbell convert T. Charles to by reviewed is 2017; 26, priority January review national for (sent A 2017 9, May III, Goddard A. William by Contributed 94720 CA and 91125; 91125; CA CA Pasadena Technology, Pasadena Technology, of Institute California Photosynthesis, Artificial 94720; CA Berkeley, Laboratory, National a CO Favaro of Marco reduction in CO step chemisorbed first form to CO surfaces in Cu(111) role by critical a plays oxide Subsurface www.pnas.org/cgi/doi/10.1073/pnas.1701405114 on etrfrAtfiilPooytei,Lwec eklyNtoa aoaoy ekly A94720; CA Berkeley, Laboratory, National Berkeley Lawrence Photosynthesis, Artificial for Center Joint rvossuisuigeeto-ae pcrsoisobserved spectroscopies electron-based using studies Previous CO promising most the is (Cu) Copper 2 etcro ixd (CO con- efficiently dioxide can carbon that electrocatalysts vert new of discovery he reduction | uoiecopper suboxide a,b,c,1 2 R ersnigassanbeptwytoward pathway sustainable a representing RR, 2 )tefvrt hie hs H Thus, choice. favorite the O) a Xiao Hai , | M06L 2 euto ecin(CO reaction reduction 2 g n H and | 2 2 -CO min rsueXPS pressure ambient d,e,1 Rpout 2.Dsienumer- Despite (2). products RR 2 nolqi ul n feedstock and fuels liquid into ) n sntslciefrdesirable for selective not is and , f oeua ipyisadItgae iiaigDvso,Lwec eklyNtoa aoaoy Berkeley, Laboratory, National Berkeley Lawrence Division, Bioimaging Integrated and Biophysics Molecular 2 a Cheng Tao , 2 2 2 c ovlal yrcros a hydrocarbons, valuable to hmclSine iiin arneBree ainlLbrtr,Bree,C 94720; CA Berkeley, Laboratory, National Berkeley Lawrence Division, Sciences Chemical t7 ,weesachemi- a whereas K, 75 at oeue dobo the on adsorb molecules O euto rdcssc as such products reduction 2 nohg-au chemical high-value into 2 Rcniaeamong candidate RR 2 2 ntephysisorbed the in ntepeec of presence the in d,e 2 Ratvt and activity RR 2 | ila .GdadIII Goddard A. William , 2 2 2 nC(1)is Cu(111) on sotnthe often is O R squite is RR) electroreduc- 2 5 ) An 6). (5, 2 2 ae 1,1) u ieti iupofo h xsec fsuch species initial of the discrep- existence determining and the of uncertainties importance of These the proof indicate lacking. ancies situ is in Cu(111) direct on species but to 14), (11, evidence faces little CO is of CO there Activation chemisorbed but concept. important (3), this it surfaces support Indeed Cu CO well. preoxidized the as that on oxygen suggested surface been of has formation consequent the lower much CO at that performed (13) al. et K Eren CO 475–550 by study of CO recent range kJ· a for 93 the hand, about in Cu(100) of energy temperatures activation clean an and (finding on Torr 740 (9) about al. observed of et also was Rasmussen phenomenology by similar A kJ·mol oxygen). (with adsorbed 67 K 600 about CO and of K reaction energy 400 between activation detected of an CO be hundreds of can for adsorption Cu(110) Torr dissociative clean 1,300 first-order (pres- nearly and values a Torr higher seconds), 65 sensibly that between to showed ranging increased (12) sures is al. exposure (10), et CO the Cu(110) Nakamura reveal when However, (9), not (11). Cu(100) did Cu(111) clean K and on 250 dissociation and or K adsorption 100 between temperatures at 10 performed CO (about studies low (UHV) situ ex vacuum Previous upon ultrahigh 1B). temper- in (Fig. room observed K) to (298 substrate is (r.t.) Cu ature the physisorption of temperature no the increasing that showed (CO experiments capture electron by induced 1073/pnas.1701405114/-/DCSupplemental. at online information supporting contains article This 2 interest. 1 of conflict no declare authors University. The Princeton B.E.K., and Washington; of University C.T.C., paper. Reviewers: the wrote and data, analyzed research, formed obdfr fCO of form sorbed uhrcnrbtos ...... n ...dsge eerh per- research, designed E.J.C. and J.Y., W.A.G., T.C., H.X., M.F., contributions: Author e b.o,o [email protected]. or lbl.gov, owo orsodnemyb drse.Eal [email protected],JYano@ [email protected], Email: addressed. be may correspondence whom To work. this to equally contributed H.X. and M.F. aeil n rcs iuainCne,Clfri nttt of Institute California Center, Simulation Process and Materials ogieteftr eino mrvdcatalysts. improved of design future the water. needed guide of information to role quintessential the critical provide the findings and These suboxide copper of importance for CO step first activating essential the reveals work this functional calculations, spectros- theory density mechanical photoelectron quantum and X-ray experiments copy pressure ambient Combining Significance 2 ata rsue bten00 orad1 or revealed Torr) 10 and Torr 0.05 (between pressures partial d,e,2 2 2 a iscaieyasr nC(0)adC(1)with Cu(111) and Cu(100) on adsorb dissociatively can xoue fo e ohnrd fLnmi (L)] Langmuir of hundreds to few a [from exposures uk Yano Junko , 2,g 2 → naC ufc,i atclr ihihigthe highlighting particular, in surface, Cu a on 2 pce a eotdas nC tpe sur- stepped Cu on also reported was species CO 2 b dacdLgtSuc,Lwec Berkeley Lawrence Source, Light Advanced a tblzdb ata eaiecharge negative partial a by stabilized was g O + a,f,2 n ta .Crumlin J. Ethan and , ads weeO (where 2 2 δ − 2 ih iscaemr easily more dissociate might 7 ) h same The 8). (7, 1A) (Fig. ) www.pnas.org/lookup/suppl/doi:10. activation 2 −9 NSEryEdition Early PNAS the , −1 ads or fe relatively after Torr) mol d on etrfor Center Joint ,acrigt the to according ), tnsfrsurface for stands −1 .O h other the On ). 2 i assumed via b,2 2 pressures | f6 of 1 2 on 2

CHEMISTRY typical 1 × 1 reconstruction as shown by the low-energy electron diffraction (LEED) pattern in Fig. 2A (17). Scanning electron microscopy (SEM) measurements Fig. 2A determine that this A sputtering and annealing procedure leads to crystalline regions with tens of micrometers mean sizes. The characterized sample surface location remained unchanged throughout the APXPS experiments. The collected spectra were averaged over a beam spot size of ∼0.8 mm in diameter. Although we cannot exclude possible contributions from the presence of grain boundaries, B averaging the data over the large probed area led to an even- tual grain boundary contribution less than 1% of the overall mea- sured signal, which is below the detection limit. Therefore, their physical/chemical features were not captured in the spectra and do not constitute the focus of this study. During the APXPS experiments (Fig. 2B) performed at r.t. C (298 K), CO2 was first introduced at 0.7 Torr on the pristine metallic Cu(111) surface. For the other experimental conditions and investigated surfaces (see Table 1 and Supporting Informa- tion for further details), the CO2 partial pressure (p(CO2)) was kept at 0.35 Torr whereas the total pressure (ptot) was kept con- stant at 0.7 Torr by codosing H2O. The APXPS measurements were performed while dosing CO2 on both metallic Cu(111) D and Cux=1.5O surfaces, whereas CO2 and H2O were codosed on metallic Cu(111), Cux=1.5O, and Cux=2.5O suboxide sur- faces (18). The sample surface was clean and no evident C- or O-based contaminations were observed after the cleaning pro- Fig. 1. Overview of surface reactions of CO2 on Cu(111) under various in cedure, as shown in Fig. S1. In addition, the in situ mass anal- situ conditions. Here the g-CO2 indicates gas-phase CO2, l-CO2 indicates lin- ysis of the reactants (O2, CO2, and H2O), using a conven- ear (physisorbed) CO2, and b-CO2 indicates bent (chemisorbed) CO2.(A and tional quadrupole mass spectrometer (QMS) mounted on the B) The forms of absorbed CO2 on pristine Cu(111). (A) Both physisorbed δ− −6 analysis chamber (and operating at a partial pressure of about l-CO2 and CO2 are observed at 75 K for pressures up to 10 Torr. (B) Only −6 COδ− is observed at 298 K for pressures ranging from 10−6 Torr to 0.1 Torr. 10 Torr), did not reveal CO cross-contaminations of the gases. 2 However, Fig. S2 shows that, concomitantly with the gas dosing (C) The adsorption of CO2 when a subsurface oxide structure is deliber- ately incorporated into Cu(111) but without additional H2O. In this work we observe that a subsurface oxide coverage of about 0.08 ML is responsi- ble for stabilizing l-CO2 at 298 K and 0.7 Torr. Here Osub indicates subsurface oxygen between the top two layers of Cu. (D) The cooperative interaction of A B hv Cu <111> codosed CO2 and H2O on Cu(111) composed of 0.08 ML of subsurface oxide, leading to the first reduction step of CO2 by adsorbed H2Oads; HCOO− ind- cates adsorbed formate. e- 50 μm LEED (110 eV) LEED (110 gas phase HEA aperture HEA SEM (SE, 5 keV) (p < 1 Torr) 3λ

hv = 387 eV hv = 632 eV 2 formed while exposed to realistic gas pressures of CO and b-CO C-C Cu x O 2 C 1s l- O 1s CO2 CDl-CO H2O (13, 15). 2 C-O(H) C=C -CO3 Cu x O-Oads To advance this understanding, we investigated in detail the -CO3 HCOO H 2ads O E E x 0.5 initial steps of CO2 adsorption both alone and in the presence of H2O on Cu(111) and suboxide surfaces (Cux=1.5,2.5O) via in situ probing of the electronic structure of the surface and reac- D D tion products, using ambient pressure X-ray photoelectron spec- troscopy (APXPS) performed with soft X-rays (200–1,200 eV) C C

Cu-Osub at the solid/gas interface. These studies are complemented with b- CO2 molecular structures and binding free energies of the reaction B B C-O(H) / HCOO

products at the M06L level (16) of density functional theory normalized intensity (arb. units) normalized intensity (arb. units)

(DFT) that was optimized for molecular clusters and reaction bar- A A x 3 riers. This combination of experiments and calculations allows us to conclude that the presence of suboxide species below the 292 290 288 286 284 282 534 532 530 528 binding energy (eV) binding energy (eV) Cu surface and the presence of H2O play a crucial role in the adsorption and activation of CO2 on Cu (Fig. 1). Specifically, Fig. 2. Investigation of various Cu surfaces using APXPS. (A) LEED pattern the presence of subsurface oxygen leads to a specific interaction obtained at an electron kinetic energy of 110 eV and SEM micrograph of with gas-phase CO2 that stabilizes a physisorbed linear CO2 con- the Cu surface after sputtering and annealing cycles obtained by detect- figuration (l-CO2, Fig. 1C). In addition, H2O in the gas phase ing the secondary electrons (SE) with a kinetic energy of the primary beam (g-H2O), aided by small amounts of suboxide, drives CO2 adsorp- of 5 keV. (B) Schematic of the APXPS measurements with the highlighted tion through the transition from the linear physisorbed state to a probed volume (3λ) along the (111) direction. (C and D) C 1s and O 1s photoelectron peaks and multipeak fitting results obtained for the various bent chemisorbed species (b-CO2), which with the aid of H2O pro- − experimental conditions and investigated surfaces (at r.t., 298 K): (exper- motes the initial reduction of CO2 to formate (HCOO , Fig. 1D). imental condition A) pure CO 0.7 Torr on metallic Cu(111); (experimental The Cu surface exposing mainly the Cu(111) orientation was 2 condition B) CO2 + H2O 0.7 Torr on metallic Cu(111); (experimental condition prepared in situ from a polycrystalline sample, by repeated argon C) CO2 + H2O 0.7 Torr on Cux=2.5O; (experimental condition D) CO2 + H2O (Ar) sputtering (normal incidence, 2 keV, 45 min) and annealing 0.7 Torr on Cux=1.5O); and (experimental condition E) pure CO2 0.7 Torr on cycles in hydrogen (0.15 Torr) at 1,100 K (for 60 min), to obtain a Cux=1.5O. The experimental conditions are summarized in Table 1.

2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1701405114 Favaro et al. Downloaded by guest on September 30, 2021 Downloaded by guest on September 30, 2021 E D C 9. Vcrepnst h hteeto msinof emission photoelectron the ( to corresponds eV 293.3 o ier 279e n 8. V epciey n carbonate and respectively) eV, 288.4 and (−CO eV (287.9 linear) for 273e) chemisorbed eV), (287.3 (HCOO−) formate (denoted of presence the structures carbon oxidized BE CO higher higher adsorbed At and of (ii) fingerprints (15). car- spectral values see graphitic literature we the as on assigned based eV), be (286.3 can BE that eV), low (284.5 At species bon (i) chemical regions: see spectral main we two exhibit details) further probed the eV, 2B). (Fig. 100 layer about topmost the is from photoelectrons 1s 3λ O depth, and were escap- 1s the spectra of C energy eV photoelectron kinetic ing the 387 Because 1s of respectively. energies eV, O 632 photon and at and 1s conditions 1s APXPS O under C and acquired spectra. under- 1s area the C total for neath the by components normalized shifted respectively, chemically deconvolution, the of areas effects. state initial uino ohteC1 n spoolcrnseta(i.2 (Fig. spectra photoelectron 1s O and deconvo- 1s multipeak C by C the monitored both were on systems lution various the of the with compared shifts (8). BE state physisorbed appreciable to leading chemisorbed form adsorbate, to needed bonding CO chemical the structure 22–24). contrast, (7, electronic In configuration adsorbate gas-phase its the with leaves compared generally unchanged (7)] ener- to K binding comparable 298 sur- millielectronvolts, [surface at different few interactions a the (vdW) by of Waals gies of mediated der and Physisorption function van 1s conditions. a weak C experimental as corresponding and peaks the faces photoelectron of shifts 1s BE (19–21). O spectral difficulty CO the spectro- this chemisorbed monitor electron-based and overcomes physisorbed use APXPS between high- discriminate of to To this use difficult Our but it scopies. dynamics, makes gas adsorption pressure various of detailed allow investigations to sufficiently configuration physisorbed CO a of stabilize pressures require studies r.t. because tally per- electrochemical CO the improved toward with formance catalysts novel tailoring for a which under CO conditions chemisorbed experimental the established we steps, n CO and carbon between interplay surface. completely copper and the cannot and reactions we contaminations side Therefore, eventual eV]. 285.1 exclude at (BE) centered energy being binding corresponding [the occurred readily nations K 10 exceeding pressures (for Torr environment structure B A Surface condition Experimental experimental and APXPS structures with surface explored Cu conditions Various 1. Table aaoe al. et Favaro i.S4 Fig. h eovltdC1 pcr (see spectra 1s C deconvoluted The h dopinsaeo CO of state adsorption The experimen- steps early these probe to difficult was it Previously oudrtn o neatosbtentectls surface catalyst the between interactions how understand To and 2 nteC ufc eitiue h lcrncdniyi the in density electronic the redistributes surface Cu the on 3 ,uigceial hfe opnnssniiet the to sensitive components shifted chemically using D), 294e)(5.Fnly hr ekcnee tabout at centered peak sharp a Finally, (15). eV) (289.4 ) ). 2 eemn h ehnsso h nta CO initial the of mechanisms the determine b (λ -CO ealcC(1)CO CO Cu(111) Metallic Cu(111) Metallic steeeto enfe ah saot12nm, 1.2 about is path) free mean electron the is 2 sp Cu Cu Cu o et,adpyiobdCO physisorbed and bent), for 2 2 3 x x x tt a esaiie.Ti rvdstebasis the provides This stabilized. be can state hr eovlto ftesetaindicates spectra the of deconvolution where , =1.5 =1.5 =2.5 i.S3 Fig. CC abn(8. V,adCOH bonds C-O(H) and eV), (285.2 carbon (C-C) CO O CO O CO O 2 RR. A −6 2 and 2 2 2 or,utk fcro contami- carbon of uptake Torr), n h vrl ufc chemistry surface overall the and H + H + H + a rsue Temperature, pressure, Gas B 2 2 2 2 2 11 . 298 0.7 (1:1) O 11 . 298 0.7 (1:1) O 11 . 298 0.7 (1:1) O eot h nertdpeak integrated the reports uprigInformation Supporting Total . 298 0.7 . 298 0.7 2 k 2 B (denoted iheog to enough high T 57meV 25.7 = 2 reduction g l -CO -CO 2 we , for 2 2 is S5B in Figs. described as charac- (VB) L were band Auger valence Cu surfaces the Cu of means various by terized the limitation, evidenced this clearly overcome is and established well from been observed has levels we core BE Cu high at Finally, 25). (15, H eV adsorbed 531.8 at bonates eV; 530.8 at overlapping (HCOO−) sev- eV) = R (530.6–530.8 with and range C-(OH), species formate, spectral O-R as (such same overlap the species oxygen-based in eral by that complicated fact is identification the shift chemical pres- photoelectron the via in when hydroxylation H for of ence sites O nucleation that as noteworthy serve is can the It stabilizing 25). in 18, (15, role important an l plays suboxide of presence O ihrB,w dniyceiobdCO to lower namely chemisorbed from identify region, range, we (about this middle BE, Within technique the eV. higher 532.0 in the is and instead eV of 530.8 fall concentration between limit bonds its C-O detection ML). Cu-OH, the 0.02 surface completely below cannot likely of we most presence although and the con- Therefore, formate a fitting. up exclude of build 1s counterpart to O us spectral allowed sistent BE) 1s in C discriminated (well the C-(OH) of presence the (O O adsorbed subsurface (ii) 25–27); (15, (Cu structures surface (i) follows: as bonded O (Cu-O of low O At states regions. the adsorbed three 1s, identify into C window we spectral with BEs 1s As also O abundances. while the relative partition envelope we spectral different 1s the O for the accounting of core-level interpretation 1s the O help on analysis 2 similar (Fig. a spectra out carried we regions, face pce trt wt h orsodn scnee tB = adsorbed BE charged at negatively centered a CO 1s as adsorbed “CO labeled C they an corresponding which of the eV), (with presence 288.4 r.t. the CO sam- at observed Cu dosed species (nonoriented) authors they polycrystalline The a where ple. on reported (15), together previously al. and rately experiments et similar Deng by by inspired was work H with codosed important an by observe we decrease particular, In conditions. CO experimental adsorbed (b the configuration for adsorption only the exception change an resolution, not with spectral do eV), the 1s ∼0.15 (within O conditions and experimental 1s the C with for components shifted chemically ege l 1)hv enotie naplcytliesurface polycrystalline a on obtained been have (15) by reported al. results et The reactivity. Deng surface poten- different a can cleaning to differences lead surface tially these Overall, different procedures. by annealing inves- formed and different structures potentially gas as surface well experimental as tigated higher study, this the in by used pressures addressed be might differences component spectral group methoxy (−OCH the of development nificant CO 2 between (Fig. products r.t. reaction of ence the CO CO i.e., adsorbed eV, observe the 288.5 not in did component authors new the a however, Interestingly, work. this in the to attributed be -CO sub h ifiut ndsrmntn ewe Cu between discriminating in difficulty The odsnageterl foye ntesraeadsubsur- and surface the on oxygen of role the disentangle To ti motn ont htteB fteaforementioned the of BE the that note to important is It 2 t598e 2)(sw ics naltrscin uha such section, later a in discuss we (as (27) eV 529.8 at ) 2 δ 2 oCO to − eotn h u3 hteeto pcr.To spectra. photoelectron 3p Cu the reporting S5A, Fig. o Cu for (iii) and ); ”W eiv hi dobdCO adsorbed their believe We .” 3 E=252e)we ooigCO codosing when eV) 285.2 = BE , 2 and .Hwvr h eeto feeta uO groups Cu-OH eventual of detection the However, O. C 2 2 (H O + fteC1 Ewe CO when BE 1s C the of 2B) (Fig. eV ∼0.50 and S6 x .Teaayi efre nC1 a sdto used was 1s C on performed analysis The D). 2 H O-O .This S2). (Fig. surface Cu(111) metallic the on O . 2 ,weesDn ta.(5 bevdtesig- the observed (15) al. et Deng whereas D), 2 .I diin eosreol ekpres- weak a only observe we addition, In O. O b ads l ads -CO -CO ads o ealcC n nsbxdcCu suboxidic on and Cu metallic )on t510e n 2. V respectively eV, 529.6 and eV 531.0 at ) −CH t524e 1,25). (15, eV 532.4 at ) x 2 2 >1 ,psigfo h xouet pure to exposure the from passing ), ofiuainosre n computed and observed configuration h si etrda 3. eV 530.3 at centered is 1s O the O 3 , −CH -CO Discussion 3 ads 2 M 2 l -CO vs. CH 4,5 ruso h usurface Cu the on groups 2 2 2 2 l pcrlrgo (287.9– region spectral M 3 -() n formate and C-O(H), , NSEryEdition Early PNAS -CO 2 .O h te hand, other the On ). pce ol actually could species n H and sub 4,5 t514e;adcar- and eV; 531.4 at nmtlC (Cu- Cu metal on ) 2 n srpre in reported as and 2 2 rniinadthe and transition 0 eed nthe on depends ) n H and n H and n Cu and 2 ooe at codosed O 2 2 .These O. 2 sepa- O + where , | using f6 of 3 2 x O is 2

CHEMISTRY Table 2. DFT models of Cu(111) with various distributions of surface O atoms and of calculated O 1s BE with experimental APXPS results

Predicted δOads Method Structure and δOsub

DFT 1/4 ML Oads δOads = −2.2 eV DFT 1/4 ML Osub δOsub = −1.3 eV DFT 1/4 ML Oads + 1/4 ML Osub δOads = −0.3 eV; δOsub = −1.5 eV APXPS 0.06 ML Oads + 0.08 ML Osub δOads = −0.4 eV; δOsub = −1.6 eV

Fig. 3. Predicted structures for 1/4 ML of physisorbed l-CO2 on various Cu surfaces (Cu, light blue; C, brown; O, red, but O is marked in orange). local-density approximation (LDA)] do not account for London sub dispersion, which is usually included with empirical corrections (A–D) Top and side views of (A) pristine Cu(111), ∆G = +0.27 eV, pthresh = 33 atm; (B) Cu(111) with 1/4 ML Oads (row 1 of Table 2) ∆G = +0.21 eV, (29). However, there is no rigorous basis for the empirical vdW pthresh = 3 atm; (C) Cu(111) with 1/4 ML Osub, (row 2 of Table 2) ∆G = −0.39 correction for Cu. Instead we use the M06L version of DFT that −7 eV, pthresh = 2 × 10 Torr; and (D) Cu(111) with 1/4 ML of both Oads and includes both kinetic energy and exchange correlation functions Osub (row 3 of Table 2), ∆G = −0.13 eV, pthresh = 7 Torr. Both C and case D optimized by comparing to a large benchmark of known vdW are consistent with experiment. clusters with accurately known bonding energies (16). Further details are presented in Computational Details of DFT Calcula- tions, Dataset S1, and Tables S1 and S2. likely exposing extended grain boundaries and coexistence of different surface orientations, whereas the present study was Physisorbed CO2 on Cu(111) performed on an oriented surface. Our experimental results can Fig. 3A shows the predicted surface structure for 1/4 mono- be explained in terms of two different adsorption configurations layer (ML) equivalents (MLE) of CO2 on metallic (Osub-free) of CO2: (i) physisorbed linear CO2 (l-CO2) above 0.150 Torr Cu(111). The physisorbed l-CO2 molecules have a C-O bond (Fig. 3) stabilized by small amounts of residual Osub and (ii) distance of 1.164 A˚ compared with 1.163 A˚ in gas phase chemisorbed CO2 (b-CO2) that is formed only after adding H2O, and O-C-O angles of 179◦, with an equilibrium distance of but also requires Osub. 3.11 A˚ from the C atom of CO2 to the Cu surface, charac- For pure CO2 on pristine metallic Cu(111) (Fig. 2 C and D, teristic of weak vdW interactions. The quantum mechanical experimental condition A), we observe experimentally a weakly (QM) electronic bond energy to the surface is ∆E = −0.36 eV adsorbed l-CO2 at 298 K with a pressure of 0.7 Torr CO2. with ∆H(298 K) = −0.30 eV enthalpy of bonding (after includ- This does not agree with our DFT calculations, performed at ing ZPE and specific heat). However, the large decrease in the M06L level, including the electron correlation required for entropy from the free CO2 molecule leads to a free energy for London dispersion (vdW attraction) (16). We find an electronic physisorbed CO2 that is unfavorable by ∆G(298 K, 0.7 Torr) = binding energy of ∆E = −0.36 eV and an enthalpy of binding +0.27 eV, which would require a pressure of 33 atm to observe of ∆H(298 K) = −0.30 eV [after including zero-point energy at 298 K. (ZPE) and specific heat]; however, due to the large decrease in entropy from the free CO2 molecule, the free energy for Physisorbed CO2 with O on Cu(111) l-CO2 is uphill by ∆G(298 K, 0.7 Torr) = +0.27 eV. These The experimentally observed O 1s shifts indicate a small amount energetics would require pressures of 33 atm (∼25 M Torr) for of surface and/or subsurface adsorbed O (denoted Oads and Osub, the adsorbed l-CO2 to be observed at 298 K on pure metallic respectively) is present in our pristine Cu(111). Compared with Cu(111). This is in line with previous experimental observations reported in the literature (also Fig. 1), where only l-CO2 was observed on metallic Cu(111) surface at 298 K (7). On the other hand, our DFT calculations show that very small amounts of suboxide (one suboxide O per every four surface Cu in our calculations, but likely much smaller levels are sufficient) lead to a negative free energy of ∆G(298 K, 0.7 Torr) = −0.12 eV, which would stabilize physisorbed l-CO2 at our experimental conditions. Indeed, our experiments find evidence for small amounts (∼0.08 ML) of surface suboxide on our freshly prepared Cu(111) (Fig. 2D, experimental con- dition A). Such subsurface adsorbed O (denoted Cu-Osub) has been observed often near the Cu surface, most likely resulting from oxygen impurities in the chamber (28) or partial dissocia- tive adsorption of CO2 (13). Interestingly, even if CO2 is still in a linear configuration (similar to the gas phase), we observe experimentally that the O 1s and C 1s core-level BEs of l-CO2 shift downward by ∼4.9 eV compared with g-CO2 (Fig. S4). This important shift means that an actual interaction is taking place between the adsorbate and the surface (7, 22), although the adsorption state still resembles physisorption. Fig. 4. M06L predicted structures for chemisorbed b-CO with H O on To interpret these findings, we investigated in detail the influ- 2 2 Cu(111) with different levels of Osub. ∆G is reported for 298 K, and ence of O on the formation of l-CO2, using various levels sub p = 0.35 Torr for H2O and CO2.(A–C) Top and side views with chemical illus- of DFT calculations. These calculations are discussed in detail tration of predicted structures (A) on pristine Cu(111), ∆G = +1.07 eV; (B) on in Supporting Information. It is well known that standard DFT Cu(111) with 1/4 ML Osub, ∆G = −0.06 eV; and (C) on Cu(111) with 1/2 ML methods [e.g., generalized gradient approximation (GGA) and Osub, ∆G = +0.28 eV.

4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1701405114 Favaro et al. Downloaded by guest on September 30, 2021 Downloaded by guest on September 30, 2021 aaoe al. et Favaro l chemisorbed observe not do we case Cu the b h rdc susal norcniin,with from endothermic conditions, it making our eV, in +0.20 = unstable Torr) is product the ( O = BE a to leads tmbaigaprilcag saiie yahdoe bond- hydrogen a by (stabilized H charge to Cu partial ing surface a a bearing to bonded atom chemically is atom C ∆ experiments. our with agreement in b of H plus O 3C Fig. experiment. with consistent BE 1s 2 the O experimental the but cell, O this of the (δ shift in mental O (δ of shift BE experimental 1s O the i.S3 (Fig. spectra) 1s O and 1s addition, C both from mined Cu the to l Cu and CO for experiments our Cu b O ML 1/4 with chemisorbed that find O of levels increased Cu Cu surface surface the by two Cu stabilized one to is to atom bonded bonded chemically chemically is is atom atom C the experiments. our with agrees which eV, 1.07 = Torr) hydrogen find by we contributions stabilized one of with trans- energy surface, charge H negative Cu to bonding partial the the from accommodate ferred atoms O two (Fig. minima local several CO to physisorbed surface lead (i) calculations 4): the DFT of the chemisorbed oxide, character for the characteristics adsorption b in the change clearly showing dramatic a to experiment. with consistent both Torr, 7 3D Fig. for 3 C Fig. ing case For 3C. Fig. than .LwsN,Ncr G(06 oeigtepae:Ceia chal- Chemical planet: the Powering electrochemical the into insights New (2012) TF (2006) Jaramillo DN, Abram ER, Cave KP, DG Kuhl 2. Nocera NS, Lewis 1. -CO -CO CO lssraeO surface plus HCOOH (iv) and S7A), Fig. -CO -CO -CO -CO ads 2 (9 ,07Tr)= Torr) 0.7 K, G(298 h F acltospeitta nCu(111)O on that predict calculations DFT The oee,icesn h O the increasing However, O without Cu(111) of case the In 2 a assumed we convenience computational For hsrsl fa piu O optimum an of result This O the When CO of dosing Simultaneous euto fcro ixd nmtli oprsurfaces. copper metallic on 7059. dioxide carbon of utilization. reduction energy solar 15735. in lenges 0 b sadr odtos,Fg 3D Fig. conditions), (standard -CO n Cu and ed oB .1ad15 V eeecn ogas-phase to Referencing eV. 1.54 and 0.31 = BE to leads 2 2 × 2 2 2 2 u iha with but , eepanti ntrso h itntitrcin of interactions distinct the of terms in this explain We . sub- surface some includes that surface Cu(111) a For . n t ovrint surface to conversion its and x 2 x ed to leads l a ec ihH with react can ntcl,orDTcluain n w ae ihO with cases two find calculations DFT our cell, unit 2 =1.5 O =1.5 -CO 2 2 ed to leads ads O = ∆E i.S8 Fig. ads x sceial oddt ufc Cu surface a to bonded chemically is ecnie h he ae eotdi al 2. Table in reported cases three the consider we , ad =1.5 + tutr o07Tr f11CO 1:1 of Torr 0.7 to structure O uoiesrcue,wihsosboth shows which structures, suboxide O 2 2 Fg 4 (Fig. nsraeCu surface on nue yteCu(111)O the by induced apesr hehl f2 of threshold pressure (a O O sub sub ∆ tutr,rsetvl tertowsdeter- was ratio (the respectively structure, O .5e hra i.3D Fig. whereas eV −1.35 .3e,bticuigvbainladentropy and vibrational including but eV, −0.23 ad sub (9 ,07Tr)= Torr) 0.7 K, G(298 aigmr O more Having . ∆ l ,w n that find we 4B), (Fig. ML 1/4 to increased is oee,this However, . eot h xeietlrslso exposing of results experimental the reports -CO o O for ) = G l -CO sub O ece COOH reacted (iii) A–C), b 2 b ads -CO eraetebnigof binding the decrease /b -CO .3e ihapesr hehl of threshold pressure a with eV −0.13 .8e,wihi ntbe eethe Here unstable. is which eV, +0.28 2 2 2 of ) -CO O sub n H and lsH plus l 2 ad -CO sub + 2 ssal nyfrtecs nFg 4B Fig. in case the for only stable is of .Oreprmnsas hwthat show also experiments Our ). 2 epredict we . VfrO for eV −0.4 sub ofr omt lsOH plus formate form to susal,with unstable, is sub 2 oeaei bu .8ME For MLE. 0.08 about is coverage ai f38ad53frCu for 5.3 and 3.8 of ratio . V oddc h nature the deduce To eV. −1.6 2 rcNt cdSiUSA Sci Acad Natl Proc ntepeec fH of presence the in 2 for 2 is b ooigo h Cu the on codosing O ofiuain eosrean observe we configuration, o12M,w rdc that predict we ML, 1/2 to O sub -CO ∆ −CO b + ad b -CO sub = G rnn taldestabilizes all at none or 0 -CO chemisorbed (ii) , uldu yH by up pulled .6e,wihi stable is which eV, −0.06 sub,x 2 etr n h te O other the and center, ,teCatom C the 4A), (Fig. × = ∆G 3 ed oaQ binding QM a to leads 2 ihoeO one with ihoeO one with 2 .4e oestable more eV −2.34 u nyphysisorbed only but , (carbonate). =0.25 10 ads si gemn with agreement in is nr nio Sci Environ Energ + −7 ads 2 ( b b htsae nO an shares that i.S7B). Fig. .9e bond- eV −0.39 . (9 ,0.7 K, ∆G(298 -CO n O and -CO ∆ or,whereas Torr), n nexperi- an and 0 ads hra the whereas , (9 ,0.7 K, G(298 A × lsOH plus 2 2 b sub and sub 2 hs we Thus, . -CO sub,x nFg 4B Fig. in 0 O surface 2 2 2 n O one , 103:15729– n one and O nthis In . ad e cell per ads x x b =2.5 =2.5 .In B). This . =0.25 5:7050– -CO 2 leads but , and ads O O 2 2 , , i o Cu(111)O for mic by ietr fc fSine fc fBS fteU O ne Contract under Zwicky DOE the US on the out the of carried Caltech. by were at BES, supported supercomputer calculations of is QM Office The DOE Source Science, Research, DE-AC02-05CH11231. Light Storage of Advanced Energy Office for The Photo- Director, Center Hubs. Artificial Joint for Innovation the Energy Center of Energy of part Joint Department as the US and to the synthesis DE-SC0004993 of Award (BES), Science under Energy (DOE) Basic of Office ence, CO active ACKNOWLEDGMENTS. activating unique of for development steps rational the initial electrocatalysts. for the foundation a in ing role critical conditions. a our under favored not but possible are acid formic chemisorbed physisorbed linearly a from transition the CO favoring adsorbed needed between communication also electronic is the structure oxide H subsurface promote a to subse- of for find presence we substrate the However, products. that molecular other and activated formate to the reduction quent form to 4B), cen- (Fig. metallic expose CO to because needs layer ters, topmost the that predictions, theoretical conclude and we results (3, experimental surface our From metallic 31). pristine 6, the with show compared generally CO activity oxides improved surface perfor- for generate to catalytic treated oxides the previously lysts Cu from literature of the mance in derived empirically physisorbed stabilizing for O of initial level the modest of understanding CO detailed of a steps obtain to us enabled tions step reduction first the CO elucidates of This etc. acid, formic formate, H and atoms stabilizing surface for of configuration this that the stabilize to Cu surface bent of mixture initial the of all minimum; local stable CO no find we surface, (120 CO forcing Thus, calculations. our in with derivatives corresponding O and Cu(111) pristine but lization, bent a Cu ing some properties. redox replacing different and/or with Ni atoms or charges surface Au, the Ag, the with changing Cl of by or character energetics S as and the exam- such modify For anions favorably subsurface steps. other might reaction that hypothesize these we accelerating ple, at O aimed plus systems acid formic and OH, of from energetics H (Cu atoms face an extract = can Torr) 0.7 K, formate ∆G(298 this that the predict culations .Ko-eik ,H A, Knop-Gericke 4. CO (2012) MW Kanan CW, Li 3. sub hs eut rvd h nih htsbufc xd plays oxide subsurface that insight the provide results These observation general a explain may finding unexpected This calcula- DFT and experiments APXPS of combination This O of amounts modest of presence the However, linear inert the of Activation sur- the of character the tune to how learning that expect We ni iuxryasrto pcrsoysuyi h oteeg range. energy soft the in study spectroscopy conditions: reaction absorption 15:27–34. under x-ray oxidation situ methanol for in catalyst An copper a of phases active n rmterdcino hc Cu thick of reduction the from ing ∆ 2 Ht omfri cdpu O plus acid formic form to −OH ◦ n/rO and/or tutrsrlxit h stable the into relax structures (9 ,07Tr)=02 V[ti .3e endother- eV 0.43 is [it eV 0.26 = Torr) 0.7 K, G(298 ∼140 2 . ◦ b 2 n prpit itne(∼2 distance appropriate and ) -CO b ciainb H by activation 2 l ads -CO b -CO hmsrto noaCu a onto chemisorption O b 0 -CO vce ,ShdlNergT Schl T, Schedel-Niedrig M, avecker -CO ¨ s Cu vs. 2 2 ontdlvrsfcetsaiiain sshown as stabilization, sufficient deliver not do sub a fcetyceiobol nsc centers such on only chemisorb efficiently can sub,x ofiuain(0 ihgetceia stabi- chemical great with (30) configuration 2 2 ecin ofr omt and formate form to reactions 4B, Fig. From . hswr a upre hog h fc fSci- of Office the through supported was work This 2 2 lsH plus + ewe h o w ulyr sessential is layers Cu two top the between n pnn ptepsiiiyo forming of possibility the up opening and econclude We 4B. Fig. in reported structure + =0.5 2 n Cu and −0.05. euto tlwoeptnilo ueetoe result- electrodes Cu on overpotential low at reduction nti ae omnplt hs relative these manipulate to case) this in .O h te ad u F cal- DFT our hand, other the On ]. l 2 -CO 2 films. O ads 2 O, naC ufc.W n hta that find We surface. Cu a on O 2 0 R ti nw htC cata- Cu that known is It RR: a lo st einmodified design to us allow may l 2 tm htcmie ihH with combines that atoms b -CO . -CO mCe Soc Chem Am J l 2 -CO 2 ohv h eesr angle necessary the have to 2 oeuerqie enforc- requires molecule ads lsH plus g 20)Caatrsto of Characterisation (2001) R ogl 2 ¨ NSEryEdition Early PNAS hc ssal with stable is which , + hssrto state. physisorption )t h rsieCu pristine the to A) ˚ etr hsenables This center. 134:7231–7234. 2 2 ,fraeplus formate O, sresponsible is O l sub -CO 2 eeae a generates 2 2 n H and provid- , oabent a to o Catal Top | 2 f6 of 5 O 2 O, ad

CHEMISTRY 5. Qiao J, Liu Y, Hong F, Zhang J (2014) A review of catalysts for the electroreduction of 26. Eren B, et al. (2016) Activation of Cu(111) surface by decomposition into nanoclusters carbon dioxide to produce low-carbon fuels. Chem Soc Rev 43:631–675. driven by CO adsorption. Science 351:475–478. 6. Mistry H, et al. (2016) Highly selective plasma-activated copper catalysts for carbon 27. Bluhm H, et al. (2004) Methanol oxidation on a copper catalyst investigated dioxide reduction to ethylene. Nat Commun 7:12123. using in situ X-ray photoelectron spectroscopy. J Phys Chem B 108:14340– 7. Freund HJ, Roberts M (1996) Surface chemistry of carbon dioxide. Surf Sci Rep 25: 14347. 225–273. 28. Spitzer A, Luth¨ H (1982) The adsorption of oxygen on copper surfaces: II. Cu(111). Surf 8. Wurth W, et al. (1990) Bonding, structure, and magnetism of physisorbed and Sci 118:136–144. chemisorbed O2 on Pt(111). Phys Rev Lett 65:2426–2429. 29. Grimme S, Antony J, Ehrlich S, Krieg H (2010) A consistent and accurate ab initio 9. Rasmussen P, Taylor P, Chorkendorff I (1992) The interaction of carbon dioxide with parametrization of density functional dispersion correction (DFT-D) for the 94 ele- Cu(100). Surf Sci 269:352–359. ments H-Pu. J Chem Phys 132:154104. 10. Campbell CT, Daube KA, White J (1987) Cu/ZnOx and ZnOx /Cu(111): Model catalysts 30. Spielfiedel A, et al. (1992) Bent valence excited states of CO2. J Chem Phys 97: for methanol synthesis. Surf Sci 182:458–476. 8382–8388. 11. Pohl M, Otto A (1998) Adsorption and reaction of carbon dioxide on pure and alkali- 31. Kas R, et al. (2014) Electrochemical CO2 reduction on Cu2O-derived copper nanopar- metal promoted cold-deposited copper films. Surf Sci 406:125–137. ticles: Controlling the catalytic selectivity of hydrocarbons. Phys Chem Chem Phys 12. Nakamura J, Rodriguez JA, Campbell CT (1989) Does CO2 dissociatively adsorb on cu 16:12194–12201. surfaces? J Phys Condens Matter 1:SB149–SB160. 32. Tougaard S (1989) Practical algorithm for background subtraction. Surf Sci 216: 13. Eren B, Weatherup RS, Liakakos N, Somorjai GA, Salmeron M (2016) Dissociative car- 343–360. bon dioxide adsorption and morphological changes on Cu(100) and Cu(111) at ambi- 33. Evans S (1991) Curve synthesis and optimization procedures for X-ray photoelectron ent pressures. J Am Chem Soc 138:8207–8211. spectroscopy. Surf Interface Anal 17:85–93. 14. Fu SS, Somorjai GA (1992) Interactions of O2, CO, CO2, and D2 with the stepped 34. Munoz-Flores˜ J, Herrera-Gomez A (2012) Resolving overlapping peaks in {ARXPS} Cu(311) crystal face: Comparison to Cu(110). Surf Sci 262:68–76. data: The effect of noise and fitting method. J Electron Spectrosc Relat Phenom 15. Deng X, et al. (2008) Surface chemistry of Cu in the presence of CO2 and H2O. 184:533–541. Langmuir 24:9474–9478. 35. Barman SR, Sarma DD (1992) Investigation of the L3-M45M45 Auger spectra of Cu, 16. Zhao Y, Truhlar DG (2006) A new local density functional for main-group thermo- Cu2O and CuO. J Phys Condens Matter 4:7607–7616. chemistry, transition metal bonding, thermochemical kinetics, and noncovalent inter- 36. Pauly N, Tougaard S, Yubero F (2014) LMM Auger primary excitation spectra of cop- actions. J Chem Phys 125:194101. per. Surf Sci 630:294–299. 17. Cattelan M, et al. (2013) Microscopic view on a chemical vapor deposition route to 37. Tougaard S, Yubero F (2012) Software package to calculate the effects of the boron-doped graphene nanostructures. Chem Mater 25:1490–1495. core hole and surface excitations on XPS and AES. Surf Interface Anal 44:1114– 18. Schedel-Niedrig T, et al. (2000) Copper (sub)oxide formation: A surface sensitive char- 1118. acterization of model catalysts. Phys Chem Chem Phys 2:2407–2417. 38. Simonsen AC, Yubero F, Tougaard S (1997) Quantitative model of electron energy loss 19. Salmeron M, Schlogl¨ R (2008) Ambient pressure photoelectron spectroscopy: A new in XPS. Phys Rev B 56:1612–1619. tool for surface science and nanotechnology. Surf Sci Rep 63:169–199. 39. Yubero F, Tougaard S (2005) Quantification of plasmon excitations in core-level pho- 20. Starr DE, Liu Z, Havecker¨ M, Knop-Gericke A, Bluhm H (2013) Investigation of toemission. Phys Rev B 71:045414. solid/vapor interfaces using ambient pressure X-ray photoelectron spectroscopy. 40. Dovesi R, et al. (2014) CRYSTAL14: A program for the ab initio investigation of crys- Chem Soc Rev 42:5833–5857. talline solids. Int J Quantum Chem 114:1287–1317. 21. Grass ME, et al. (2010) New ambient pressure photoemission endstation at Advanced 41. Stevens WJ, Krauss M, Basch H, Jasien PG (1992) Relativistic compact effective poten- Light Source beamline 9.3.2. Rev Sci Inst 81:053106. tials and efficient, shared-exponent basis sets for the third-, fourth-, and fifth-row 22. Browne V, et al. (1991) Activation of carbon dioxide at bismuth, gold and copper atoms. Can J Chem 70:612–630. surfaces. Appl Surf Sci 47:375–379. 42. Favaro M, et al. (2016) A synchrotron-based spectroscopic study of the electronic 23. Sterrer M, et al. (2007) Control of the charge state of metal atoms on thin MgO films. structure of N-doped HOPG and PdY/N-doped HOPG. Surf Sci 646:132–139. Phys Rev Lett 98:096107. 43. Favaro M, et al. (2015) In situ carbon doping of TiO2 nanotubes via anodization in 24. Starr DE, Weis C, Yamamoto S, Nilsson A, Bluhm H (2009) NO2 adsorption on Ag(100) graphene oxide quantum dot containing electrolyte and carburization to TiOx Cy nan- supported MgO(100) thin films: Controlling the adsorption state with film thickness. otubes. Adv Mater Interfaces 2:1400462. J Phys Chem C 113:7355–7363. 44. Schimka L, Harl J, Kresse G (2011) Improved hybrid functional for solids: The HSEsol 25. Eren B, Heine C, Bluhm H, Somorjai GA, Salmeron M (2015) Catalyst chemical state functional. J Chem Phys 134:024116. during CO oxidation reaction on Cu(111) studied with ambient-pressure X-ray pho- 45. Heit YN, Nanda KD, Beran GJO (2016) Predicting finite-temperature properties of toelectron spectroscopy and near edge X-ray adsorption fine structure spectroscopy. crystalline carbon dioxide from first principles with quantitative accuracy. Chem Sci J Am Chem Soc 137:11186–11190. 7:246.

6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1701405114 Favaro et al. Downloaded by guest on September 30, 2021