Structural and electronic properties of iron monoxide clusters FenO ...

Structural and electronic properties of iron monoxide clusters FenO ...

JOURNAL OF CHEMICAL PHYSICS VOLUME 119, NUMBER 21 1 DECEMBER 2003Structural and electronic properties of iron monoxide clusters Fe n Oand Fe n O À „nÄ2–6…: A combined photoelectron spectroscopyand density functional theory studyGennady L. Gutsev a) and Charles W. Bauschlicher, Jr.Mail Stop 230-3, NASA Ames Research Center, Moffett Field, California 94035Hua-Jin Zhai and Lai-Sheng Wang b)Department of Physics, Washington State University, Richland, Washington 99352and W. R. Wiley Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory,Richland, Washington 99352Received 30 June 2003; accepted 4 September 2003We report a combined anion photoelectron spectroscopy and density functional theory DFT studyon a series of iron monoxide clusters, Fe n O(n2 – 6). Well-resolved photoelectron spectra wereobtained for Fe n O at variable detachment energies, allowing the ground state and numerouslow-lying excited states of Fe n O to be observed. Sharp threshold photoelectron features wereobtained for each species, which suggest rather small geometry changes between the anion andneutral ground states for the monoxide clusters and allows the electron affinities of the neutralclusters to be measured accurately. Extensive DFT calculations using the generalized gradientapproximation were carried out for both Fe n O and Fe n O . Optimized geometries of the ground andlowest excited states of both the anion and neutral species are reported along with the ground-statevibrational frequencies and fragmentation energies. Theoretical electron affinities were comparedwith the experimental measurements to verify the ground states of the iron monoxide clustersobtained from the DFT calculations. © 2003 American Institute of Physics.DOI: 10.1063/1.1621856I. INTRODUCTIONThe interaction between iron and oxygen is relevant tounderstand important chemical and biochemical processes,such as corrosion and oxygen transport in biological systems.Iron monoxide clusters represent ideal model systems, whichmay help address these fundamental processes at the atomiclevel. However, the interaction of atomic oxygen with neutraland charged iron clusters is not yet well understood.Experimental spectroscopic data are known for the smallestFeO (r e 1.616 Å, 1 e 880 cm 1 , 2 4.170.08 D, 3 D o4.70.2 eV 4 and FeO (D o 3.530.06 eV 5 . For largerFe n O species, Fe n – O binding energies were obtained fromcollision induced dissociation CID measurements 6 for n2 – 17. Preliminary photodetachment photoelectron spectraof oxygen-chemisorbed iron cluster anions Fe n O (n1 – 16), as well as several O-rich iron clusters, were reportedby one of us L.S.W.. 7–10 Theoretical studies 11–14have been performed for FeO and its ions at the ab initio anddensity functional theory DFT levels. For the larger species,the only computations that we are aware of are for theneutral Fe n O(n2 – 6) using DFT within a local spin densityapproximation LSDA. 8The complicated electronic structure of transition metalsystems, and iron clusters in particular, has posed considerableexperimental and theoretical challenges. The aim of thepresent work is to obtain well-resolved photoelectron spectraa Electronic mail: ggutsev@mail.arc.nasa.govb Electronic mail: Fe n O at various photon energies in order to better evaluatethe electron affinities EA, to resolve low-lying excitedstates of the corresponding neutral clusters, and to compareour experimental findings with the results of our extensivecomputations performed using density functional theory withthe generalized gradient approximation DFT-GGA. We alsopresent optimized geometries of the ground- and lowest excitedstates of the Fe n O and Fe n O species (n2 – 6), alongwith their ground-state vibrational frequencies and fragmentationenergies. Such a combined experimental and computationalstudy allows us to assign the ground states of theanionic and neutral iron monoxide clusters, which all havehigh spins. There are a large number of low-lying excitedstates for both the neutrals and anions.II. EXPERIMENTAL AND COMPUTATIONAL DETAILSA. Experimental methodsThe experiments were carried out using a magneticbottle-typephotoelectron spectroscopy PES apparatusequipped with a laser vaporization supersonic cluster source,details of which have been described previously. 15,16 Briefly,the Fe n O cluster anions were produced by laser vaporizationof a pure iron target in the presence of a helium carriergas, and were analyzed using a time-of-flight mass spectrometer.The oxygen was either from residual oxygen in thesource chamber/target surface or from N 2 O(1%) seeded inthe helium carrier gas. The Fe n O (n2 – 6) species wereeach mass selected and decelerated before being photodetached.Four detachment photon energies were used in the0021-9606/2003/119(21)/11135/11/$20.00 11135© 2003 American Institute of PhysicsDownloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

11136 J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Gutsev et al.FIG. 1. Photoelectron spectra of Fe 2 O at a 532 nm 2.331 eV; b 355nm 3.496 eV; c 266 nm 4.661 eV; and d 193 nm 6.424 eV.FIG. 2. Photoelectron spectra of Fe 3 O at a 532 nm; b 355 nm; c 266nm; and d 193 nm.current study: 532 nm 2.331 eV, 355 nm 3.496 eV, 266nm 4.661 eV, and 193 nm 6.424 eV. Photoelectrons werecollected at nearly 100% efficiency by the magnetic bottleand analyzed in a 3.5 m long electron flight tube. The photoelectronspectra were calibrated using the known spectrumof Rh , and the energy resolution of the apparatus wasEk/Ek2.5%, that is, 25 meV for 1 eV electrons.B. Theoretical proceduresThe GAUSSIAN 98 program 17 was used for all the calculations.We used the basis sets denoted as 6-311G* in theGaussian program, namely (15s11p6d1 f )/10s7p4d1 f forFe Refs. 18–20 and (12s6p1d)/5s4p1d for O. 21 Thefollowing exchange-correlation functionals were tested onFe 2 O and Fe 2 O : BLYP Becke’s exchange 22 and Lee–Yang–Parr’s correlation 23 , BP86 Becke’s exchange 22 andPerdew’s correlation 24 , BPW91 Becke’s exchange 22 andPerdew-Wang’s correlation 25 , BPBE Becke’s exchange 22and Perdew–Burke–Ernzerhof’s correlation 26 , and hybridB3LYP. 27,28 The BPW91 functional was used in the calculationsfor n3. Assignment of the spectroscopic states ofFe 2 O and Fe 2 O was based on the symmetry of the Slaterdeterminants built using the one-electron DFT orbitals hybridor Kohn–Sham ones. Optimizations for n3 were performedwithout imposing symmetry constraints, and the optimizedstates are marked with the number of unpairedelectrons 2S, where S is the total spin.The geometry of each species was optimized and a subsequentharmonic frequency calculation was performed usinganalytical second derivatives in order to confirm that theoptimized geometry corresponded to a minimum. For theneutral species, the initial state had the same number of unpairedelectrons as the corresponding bare Fe n cluster. 29States with fewer and greater numbers of open shells werestudied until we were confident that we had found the groundstate. The starting numbers of unpaired electrons of theFe n O anions were chosen as the numbers of unpaired electronsin the corresponding neutrals increased by 1, and thenstates with fewer and greater numbers of open shells werestudied. Generally, three to five optimizations were performedin each case. Our reported electron affinities, EAs,are computed as the differences in total energies of eachspecies at its equilibrium geometry, and therefore correspondto adiabatic values.Downloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Iron monoxide clusters11137FIG. 3. Photoelectron spectra of Fe 4 O at a 532 nm; b 355 nm; c 266nm; and d 193 nm.FIG. 4. Photoelectron spectra of Fe 5 O at a 532 nm; b 355 nm; c 266nm; and d 193 nm.III. EXPERIMENTAL RESULTSThe photoelectron spectra for Fe n O (n2–6) areshown in Figs. 1–5, respectively, at four photon energies:532, 355, 266, and 193 nm. These data are significantly improvedin comparison to our previous reports in severalrespects. 7,8,10 First, each species was studied at four differentphoton energies in the present work, whereas only the 355and 266 nm photons were used previously. The low photonenergy spectra at 532 nm allowed the low binding energyfeatures to be better resolved, whereas the high photon energyspectra at 193 nm provide information about high-lyingexcited states of the neutral clusters. Second and more important,efforts were made to control the temperatures of theclusters by tuning the firing timing of the vaporization laserrelative to the carrier gas, and choosing the later part of thecluster beam where the clusters have a larger residence timein the nozzle and tend to be colder for photodetachment. 30,31The colder clusters give rise to sharper and better-resolvedspectral features, which would have been smeared out otherwiseeven with good instrumental resolution. 32–34Numerous discrete and well-resolved electronic transitionswere observed in the PES spectra of Fe n O (n2 – 6), as shown in Figs. 1–5. The PES features representthe electronic transitions from the ground-state Fe n O anionsto the ground- and excited states of the correspondingFe n O neutrals. Basically, the lower binding energy portionbelow 2.5 eV of the PES spectra for each Fe n O specieswas better resolved with several well-separated sharp features.In particular, a sharp threshold peak was observed ineach spectrum. Each feature is likely to represent a singleelectronic state of a neutral, which indicates that the energylevels near the Fermi level are not too congested. The narrowbandwidths of the low-energy peaks and the threshold peak,in particular, allows the suggestion that structural changesbetween the Fe n O anions and their neutral counterparts aresmall. In contrast, the higher binding energy portion of thePES spectra for each Fe n O species appeared to be extremelycongested, and only few prominent features could beidentified from the apparently continuous electron signals.The observed vertical detachment energies VDE for all thelabeled features X–F in Figs. 1–5 are given in Table I.We note that the temperature for the cluster anions wasexpected to be at around or slightly below roomtemperature 32 under our current experimental conditions.Consequently, hot band transitions were largely suppressed,as indicated by the disappearance or decrease of low bindingDownloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

11138 J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Gutsev et al.TABLE I. Adiabatic and vertical detachment energies ADEs and VDEs ofFe n O derived from photoelectron spectra. All energies are in eV.Observed feature ADE a,b VDE bFIG. 5. Photoelectron spectra of Fe 6 O at a 532 nm; b 355 nm; c 266nm; and d 193 tails and the appearance of sharp onset of the thresholdPES feature X. Nevertheless, a substantial low bindingenergy tail was observed for the spectra of Fe 5 O Fig. 4,showing that the relatively large iron monoxide species werestill hot under our current source conditions. Furthermore,the intense and increasing tails at the higher binding energyside of the PES spectra of Fe 5 O and Fe 6 O are characteristicof thermionic emission from hot clusters. The absenceof such high binding energy tails in the spectra of Fe 2 O ,Fe 3 O , and Fe 4 O implies that these smaller monoxide specieswere relatively cold.The well-resolved sharp threshold features X in Figs.1–5 provide important information for the ground states ofboth the anions and the neutrals and facilitate comparisonwith the results of calculations. The sharp threshold peakssuggested that there is little geometry change between theground states of the anions and the neutral clusters and thatthe extra electron in the anions occupies a nonbonding orbital.The peak maximum in each spectrum yielded the VDEfor the ground-state transitions of Fe n O , as listed in Table I.The sharp onset of the threshold peak X allowed a fairlyaccurate determination of the adiabatic detachment energiesADE for the ground-state transitions, i.e., the EAs of theFe 2 O X 1.602 1.642A2.042B2.252C2.525D3.305E3.605F3.925G 4.3Fe 3 O X 1.442 1.442A1.752B1.942C2.203D2.332E3.184F3.845Fe 4 O X 1.702 1.702A1.852B1.973C2.165D2.563E2.955F 4.3Fe 5 O X 1.852 1.872A2.002B2.205C2.493D 2.9E 3.3Fe 6 O X 1.702 1.732A1.963B2.323C2.433D 2.8E 3.2F3.786a Also represent the adiabatic electron affinities of the corresponding neutralFe n O species.b Numbers in parentheses represent the experimental uncertainties in the lastdigit.corresponding Fe n O neutral clusters. Since no vibrationalstructures were resolved, we evaluated the ADE by drawinga straight line at the leading edge of the feature X and thenadding the instrumental resolution to the intersections withthe binding energy axis. Although this is an approximateprocedure, it yields consistent ADEs from spectra taken atdifferent photon energies, due to the relative sharp onset ofthe spectra. All the ADEs were determined from the 532 nmspectra, which yielded the most accurate values because of abetter spectral resolution. The ADEs obtained in this way arealso given in Table I.IV. THEORETICAL RESULTSA. Structures and frequenciesIn agreement with the previous discussion, 35 both Fe 2 Oand Fe 2 O possess a number of closely spaced states foreach spin multiplicity. We performed an extensive searchwithin linear Fe–O–Fe, Fe–Fe–O, and angular symmetricand nonsymmetric Fe–O–Fe configurations. The groundDownloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Iron monoxide clusters11139TABLE II. Results of computations on the ground states of Fe 2 O( 7 A 2 ) and Fe 2 O ( 8 A 2 ) using differentmethods.B3LYP BLYP BPW91 BP86 BPBEFe 2 O 7 A 2 7 AR e O–Fe), Å 1.799 1.804 1.789 1.787 1.789FeOFe° 70.64 71.70 71.58 71.55 71.55 (b 2 ), cm 1 224 268 286 287 287 (a 1 ) 351 329 343 343 343 (a 1 ) 731 715 730 734 730, Debye 3.199 2.862 2.805 2.780 2.800EA, eV a 1.39 1.21 1.30 1.44 1.28Fe 2 O 8 A 2 8 AR e (O–Fe), Å 1.848 1.849 1.833 1.831 1.832FeOFe° 68.4 69.76 69.50 69.53 69.53 (b 2 ), cm 1 144 266 273 281 274 (a 1 ) 330 302 317 315 317 (a 1 ) 666 647 667 670 668a Experimental value this work is 1.600.02 eV.states of Fe 2 O and Fe 2 O were found to be 7 A 2 and 8 A 2 ,respectively. We should note that the 7 A 1 state of Fe 2 O hasessentially the same geometry as the ground 7 A 2 state, and iscomputed to be only 0.06 eV above. Thus, we cannot ruleout the 7 A 1 as the ground state of Fe 2 O. However, the 7 A 1state is not related to the 8 A 2 state of the anion by a oneelectrondetachment process, and therefore would not be observedin the experiment. The Mulliken populations showthat the charge on the oxygen increases from 0.39e inFe 2 Oto0.54e in Fe 2 O . That is, the extra electron addsmostly into the Fe 4s orbitals, which were depleted by donationto the oxygen in the neutral clusters.Optimizations performed without imposing C 2v symmetryconstraints yielded 7 A and 8 A states with identical totalenergies and bond lengths as those for the 7 A 2 and 8 A 2states, respectively. As shown in Table II, all five methodstested provide similar geometrical structures and vibrationalfrequencies except for the B3LYP, which predicts the bendingfrequency of the Fe 2 O anion to be too low. All the EAscomputed at the different levels of theory are smaller thanthe experimental value, with the largest deviation 0.4 eVseen at the BLYP level Table II. The Perdew’s correlationfunctionals PW91, P86, and PBE provide nearly identicalgeometries and vibrational frequencies. The best EA withrespect to the experiment was obtained at the BP86 level.However, this method overestimated the EAs of pure ironclusters. 29The structures of Fe 2 O and Fe 2 O with different spinmultiplicities were optimized at the BPW91 level and arepresented in Fig. 6. One can see a rather striking peculiarity:the ferromagnetic high-spin ground states are very close intotal energy to the low-spin antiferromagnetic states. The latterpossesses two nonequivalent Fe atoms. By imposing C 2vsymmetry constraints, one can obtain a nonmagnetic singletof Fe 2 O, but this nonmagnetic 1 A 1 state r e (O–Fe)1.69 Å, r e (Fe–Fe)2.12 Å] is 2.27 eV above the antifer-FIG. 6. Lowest energy states of Fe 2 Oand Fe 2 O . Bond lengths are in Å andlocal magnetic moments in bold ateach atom are in bohr magnetons. Totalenergies of the correspondingground states are taken as zero forboth series.Downloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

11140 J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Gutsev et al.FIG. 7. Lowest energy states of Fe 3 O and Fe 3 O .Seethe caption of Fig. 6. The total numbers of unpairedelectrons are labeled as 2S.romagnetic 1 A state. For the doublet Fe 2 O anion, imposingC 2v constraints leads to a state that is unbound.Figures 7–10 display optimized ground-state structuresof neutral Fe n O(n3 – 6) and their anions, along with thelow-lying states possessing the number of unpaired electronsadjacent to those in the corresponding ground states. Oxygenis twofold coordinated in species with n3 and 4, while it isthreefold coordinated in the larger clusters. This is in agreementwith the previous LSDA results, 8 except that the latterpredicted that a twofold edge-on oxygen adsorption is morestable for Fe 5 O. We found the lowest energy edge-on structurefor Fe 6 O to correspond to a 2S17 state, which isabove the 2S19 ground state Fig. 10 by 0.29 eV.Small iron clusters are highly magnetic with numerousunpaired electrons. Excluding Fe 4 O, the number of unpairedelectrons in ground states of Fe n and Fe n O are the same. TheFIG. 8. Lowest energy states of Fe 4 Oand Fe 4 O . See the captions in Figs. 6and 7.Downloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Iron monoxide clusters11141FIG. 9. Lowest energy states of Fe 5 O and Fe 5 O .Seethe captions in Figs. 6 and 7.difference between the total energies of the Fe 4 O states with2S12 and 2S14 is only 0.011 eV, which seems to berelated to a small separation between the 2S12 and 2S14 states of bare Fe 4 its 2S14 state is lower by 0.08 eVas computed 29 at the same BPW91/6-311G* level.Salahub and Chrétien 36 found a higher separation of 0.36 eVat the PP86 level. Thus, oxidation of the iron clusters, atleast in the range of 2nZ6 considered here at theBPW91/6-311G* level of theory, appears not to quench themagnetic moments of the bare iron clusters.Excluding Fe 4 O, the number of unpaired electrons in theanion ground states differs by only 1 with respect to thenumber of unpaired electrons in the ground state of the correspondingneutrals. The ground state of Fe 4 O has 2S15 and is well separated from the states with 2S13and 2S17 Fig. 8, while the neutral state of Fe 4 O withFIG. 10. Lowest energy states of Fe 6 O and Fe 6 O . Seethe captions in Figs. 6 and 7.Downloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

11142 J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Gutsev et al.TABLE III. BPW91 harmonic vibrational frequencies in cm 1 and relative intensities of Fe 3 O, Fe 4 O, Fe 5 O,Fe 6 O, and their anions. Absolute intensities KM/Mole: Fe 3 O:68.8, Fe 4 O:90.3, Fe 5 O:64.2, Fe 6 O:57.7;Fe 3 O :107.6, Fe 4 O :126.6, Fe 5 O :62.5, Fe 6 O :56.9.Freq. Fe 3 O Fe 3 O Fe 4 O Fe 4 O Fe 5 O Fe 5 O Fe 6 O Fe 6 O 1 1150.14 1430.01 1080.01 990.04 950.01 1050.01 1060.00 1050.01 2 2200.03 2210.00 1150.01 1610.00 1020.01 1430.01 1120.00 1140.00 3 2320.37 2300.24 1740.01 1670.00 1560.02 1590.00 1530.00 1380.00 4 3360.00 3470.00 1910.02 2020.00 1650.02 1730.02 1530.00 1390.00 5 4890.17 4060.37 2350.02 2310.04 1800.02 1850.03 1720.00 1710.01 6 6501.00 6131.00 2380.01 2330.02 2070.03 2080.02 1850.01 1930.00 7 3530.00 3430.00 2120.09 2270.00 2220.00 2060.00 8 4310.00 3820.11 2350.19 2440.07 2220.00 2170.00 9 6391.00 6121.00 2970.01 2850.02 2310.05 2350.01 10 3260.02 3340.01 2850.08 2800.00 11 3720.16 3800.10 2850.08 2990.01 12 5621.00 5371.00 3300.00 3210.04 13 3490.01 3350.01 14 3520.01 3720.00 15 5311.00 5241.002S14 is slightly above the state with 2S12. Since theanion and neutral tend to differ by only one unpaired electron,we would not be surprised if the ground state of Fe 4 Oisin fact a 2S14 state if a higher-level theory is applied.Harmonic vibrational frequencies for the ground statesof Fe n O and Fe n O are presented in Table III. The highestfrequency mode is associated with the Fe n – O stretching. TheFe n – O bending modes depend on the shape of the correspondingmetal cluster. For example, the second highest frequencymode in Fe 4 O corresponds to the vibration of the Oparallel to the upper two Fe atoms, while the lowest frequencymode is the vibration of the O perpendicular to theTABLE IV. Fragmentation energies in eV of neutral and negatively charged iron monoxide clusters.NeutralAnionChannel BPW91 Channel BPW91Fe 2 O → Fe 2 O 5.62 Fe 2 O → Fe 2 O 5.96→ FeOFe 2.60 Fe 2 O → Fe 2 O 5.31Fe 2 O → FeO Fe 2.62Fe 3 → Fe 2 Fe a 2.32 1.91 Fe 3 → Fe 2 Fe 2.84Fe 3 O → Fe 3 O 6.29 Fe 3 O → Fe 3 O 6.17→ Fe 2 OFe 2.99 → Fe 2 O Fe 3.05→ FeOFe 2 3.40 → FeO Fe 2 3.47→ FeOFe 2 3.78Fe 4 → Fe 3 Fe a 3.06 2.19 Fe 4 → Fe 3 Fe 3.34Fe 4 O → Fe 4 O 6.01 Fe 4 O → Fe 4 O 5.84→ Fe 3 OFe 2.70 → Fe 3 O Fe 3.01→ Fe 2 OFe 2 3.56 → Fe 2 O Fe 2 3.86→ FeOFe 3 3.82 → FeO Fe 3 4.16Fe 5 → Fe 4 Fe a 3.26 2.25 Fe 5 → Fe 4 Fe 3.37Fe 5 O → Fe 5 O 5.80 Fe 5 O → Fe 5 O 5.59→ Fe 4 OFe 3.09 → Fe 4 O Fe 3.12→ Fe 3 OFe 2 3.66 → Fe 3 O Fe 2 3.97→ Fe 2 OFe 3 4.33 → Fe 2 O Fe 3 4.67→ FeOFe 4 3.88 → FeO Fe 4 4.28→ Fe 5 O 5.80Fe 6 → Fe 5 Fe a 3.74 3.17 Fe 6 → Fe 5 Fe 3.51Fe 6 O → Fe 6 O 5.70 Fe 6 O → Fe 6 O 5.67→ Fe 5 OFe 3.64 → Fe 5 O Fe 3.54→ Fe 4 OFe 2 4.54 → Fe 4 O Fe 2 4.47→ Fe 3 OFe 3 4.98 → Fe 3 O Fe 3 5.17→ Fe 2 OFe 4 4.93 → Fe 2 O Fe 4 5.18→ FeOFe 5 4.21 → FeO Fe 5 4.50→ Fe 6 O 5.59Fe 7 → Fe 6 Fe b 3.11a Bond strengths of pure iron clusters: BPW91 values in italic are from Ref. 29, experimental data in bold arefrom Ref. 6.b Experimental data from Ref. 6.Downloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Iron monoxide clusters11143TABLE V. Experimental and BPW91 adiabatic electron affinities EA, in eV of bare iron clusters and iron monoxide clusters along with the BPW91 verticaldetachment energies VDE, in eV from the Fe n O ground states to the low and high spin states of the corresponding neutrals.Fe Fe 2 Fe 3 Fe 4 Fe 5 Fe 6EA BPW91 a 0.63 0.94 1.47 1.76 1.84 1.61Expt. b 0.1510.003 0.9020.008 1.450.08 1.800.05 1.750.05 1.610.05FeO Fe 2 O Fe 3 O Fe 4 O Fe 5 O Fe 6 OEA BPW91 c 1.26 1.30 1.34 1.60 1.63 1.53Expt. d 1.500.02 1.600.02 1.440.02 1.700.02 1.850.02 1.700.02FeO Fe 2 O Fe 3 O Fe 4 O Fe 5 O Fe 6 OVDE Low spin 2.45 1.32 1.36 1.67 1.68 1.93High spin 1.26 1.68 1.91 2.30 1.77 1.65Expt. 1.500.02 1.640.02 1.440.02 1.700.02 1.870.02 1.730.02a See Ref. 29.b The EA of Fe and the EA of Fe 2 are from Ref. 41. The rest are unpublished data from the Wang group, which are slightly improved over those in Ref. 33.c This work, the EA of FeO is from Ref. 13.d This work the PES data for FeO not shown. The EA of FeO obtained by Engelking and Lineberger Ref. 40 is 1.4920.020 eV.upper two Fe atoms. For Fe 6 O, the oxygen interacts withthree Fe atoms, and the bending frequencies corresponds tothe second and third highest frequency modes. Compared tothe vibrational frequencies of the bare iron clusters, 29 theyare generally larger. For example, the vibrational frequenciesof Fe 3 are 62, 231, and 353 cm 1 , while Fe 3 O has the lowestand highest frequencies of 115 and 650 cm 1 , respectively.While not relevant to the photodetachment experiment, wealso report the infrared intensities to aid the interpretation ofany future matrix experiments. The infrared intensities of theiron monoxide clusters are larger than those of bare ironclusters. For each molecule, most of the intensity falls intothe highest frequency mode.B. Thermodynamic stabilityFragmentation energies for different channels of the neutraland charged Fe n O species were computed as differencesin total energies of a species and its decay products correctedfor the ZPEs, as presented in Table IV. For comparison, weincluded the Fe-abstraction energies from the bare iron clustersthat were previously computed 29 and were also obtained 6from the CID measurements. The Fe n – O binding energieswere found to be much higher than the Fe n1 – Fe energies. Acomparison of the computed and experimental Fe n1 –Febond strengths shows that the BPW91 approach, as well asother pure DFT methods, tends to overestimate the bondstrengths by about 1 eV. For comparison, the bond strengthin the ground-state O 2 dimer computed at the sameBPW91/6-311G* level is 5.72 eV; the experimentalvalue 37 is 5.12 eV. The bonding of the O atom to iron clustersis of a similar strength and is nearly independent of the clustersize. The charged Fe n O clusters also possess highFe n – O binding energies, owing to similar EAs of the Fe nand Fe n O clusters, as will be discussed below.V. COMPARISON OF EXPERIMENTALAND THEORETICAL RESULTSThe three lowest states of Fe 2 O that are connected to theground 8 A 2 state of the anion are the 7 A 2 ground state andthe 9 A 2 and 7 B 1 excited states, which are 0.32 and 0.78 eVabove the 7 A 2 state, respectively. Detachment to form the7 A 2 state is associated with the peak labeled X in Table I andFig. 1. On the basis of our computed excitation energies, the9 A 2 state probably corresponds to peak A, while the formationof the 7 B 1 state is associated with either B or C. Weperformed a Franck–Condon analysis 38,39 for the 7 A 2 , 9 A 2 ,and 7 B 1 states, and the shape and width of the primary detachmentpeak are similar for all three states. This is differentfrom what is observed in experiment. Thus, the intensity ofthe peaks is likely due to the character of the orbital fromwhich the electron is detached. For all of the systems, thehighest occupied spin orbital of the anion is mostly ofFe 4s character, while the highest occupied spin orbital is amixture of Fe 4s and 3d. There is only a small oxygen componentin either the highest occupied or orbital. Theother open-shell orbitals tend to have a smaller 4s componentthan the highest occupied orbital, but a bigger 4scomponent than the highest occupied orbital. Because 4sdetachments have larger cross sections than 3d detachments,we expect that detaching an electron to form the low-spinneutral should have a larger intensity than detaching a electron to form the high-spin neutral. For all systems the Xpeak is one of the most intense at the lowest detachmentenergy 532 nm, consistent with detachment from an orbitalwith the mostly Fe 4s character and rather similar geometriesfor the neutral and its anion. While this is consistent with theBPW91 results for Fe 2 O to Fe 5 O , the calculations predicta detachment of a electron for Fe 6 O . However, the 2S19 and 21 states of Fe 6 O and the 2S20 and 18 states ofFe 6 O are close in energy, and therefore it is possible that theordering of the low-lying states of either the anion or neutralis incorrect.Experimental and theoretical EAs of the Fe n O clustersare compared in Table V and Fig. 11. We also include resultsfor FeO, and we note that our experimental value is veryclose to that of Engelking and Lineberger. 40 For comparison,we also give experimental 33,41 and theoretical 29 EAs of thebare Fe n clusters. Three observations can be made immediately.First, the difference between the adiabatic and verticaldetachment energies is small in both theory and experiment.Second, the size-dependent trend in Fe n O is reasonably re-Downloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

11144 J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Gutsev et al.produced by the BPW91. Third, upon oxidation by theatomic oxygen, the EAs increase markedly only for thesmallest species (n1 and 2, whereas for larger species(n3 – 6) oxidation only slightly changes the EAs. The EAtrend of Fe n OvsFe n may be understood from the anion-toneutralstructural changes of the iron monoxide. As can beseen in Figs. 7–10, for the n3 – 6 monoxides, the majorstructural changes between the anions and neutrals take placein the Fe–Fe bond lengths, whereas the Fe–O bond lengthschange very little. This is consistent with the metal-basednature of the anion HOMOs and explains why the oxidizationof atomic oxygen does not lead to appreciable increaseof the EAs for these relatively large monoxide species Fig.11. On the other hand, photodetachment from Fe 2 O inducesprimarily changes in the Fe–O bond length withoutsubstantial change of the Fe–Fe bond length. An inspectionof the anion HOMO shows that it is antibonding betweenoxygen and the Fe atoms, and weakly bonding between thetwo Fe atoms. That is why oxidation of Fe 2 leads to anapparent increase in the EA (Fe 2 OvsFe 2 , Fig. 11.It should be pointed out that while the theoretical EAsagree well with the experimental data for the bare Fe 2 –Fe 6clusters, the EAs are less accurately reproduced for the monoxidespecies Fe n O(n2 – 6). The BPW91 method underestimatesthe EAs by 0.1–0.2 eV for most monoxide species;the largest difference between theory and experiment reaches0.30 eV for Fe 2 O. Increasing the basis set for oxygen to6-311G(3df) did not result in any appreciable changein the EA of Fe 2 O. However, such discrepancies are stillwithin the typical error bars of the DFT methods and areacceptable. For the atomic EA, the overestimation is probablyrelated to an incorrect description of the ground-state4s 2 3d 6 configuration of the Fe atom, where the pure DFTmethods are biased 42 toward a mixed 4s 2 3d 6 occupation.Good agreement between our theoretical and experimentalEAs lends credence to the current DFT calculations ofboth the monoxide anions and the neutrals. It is safe, ingeneral, to assign the threshold PES feature X in Figs. 1–5to the transition from the anion ground state to the neutralground state identified in the DFT calculations Figs. 6–10,respectively. These ground-state transitions are well characterizedby one-electron detachment processes, as indicatedby the spin states. Therefore, we would not be surprised tofind that the true ground state of Fe 4 O is the 2S14 stateFig. 8, even though it is computed to be slightly higher inenergy by 0.011 eV than the 2S12 state, because theground state of Fe 4 O is clearly the state with 2S15.FIG. 11. Comparison of the experimental filled dots and BPW91 emptydots adiabatic electron affinities in eV of Fe n O obtained in the presentwork with previous experimental empty squares data for the bare Fe nclusters.VI. SUMMARYIn summary, we report a combined anion photoelectronspectroscopy and density functional theory study on a seriesof iron monoxide clusters, Fe n O(n2 – 6). Well-resolvedphotoelectron spectra were obtained for Fe n O at severalphoton energies, allowing the ground- and numerous lowlyingexcited states of Fe n O to be observed. The ground-stateadiabatic and vertical detachment energies were determinedmore accurately from the well-resolved spectra obtained atlow photon energies. Sharp threshold features were observedfor all the Fe n O species, which indicates rather small geometrychanges between the monoxide anions and the correspondingneutrals upon photodetachment. Extensive densityfunctional theory calculations with generalized gradient approximationwere carried out for both Fe n O and Fe n O withdifferent spin states. Optimized geometries of the groundandlowest excited states of both anion and neutral speciesare reported, along with their ground-state vibrational frequenciesand fragmentation energies. Theoretical adiabaticdetachment energies are compared with the experimentalmeasurements, and good agreement is achieved.ACKNOWLEDGMENTSThe theoretical work done in California was supportedfrom NASA Ames Research Center through Contract NAS2-99092 to Eloret Corporation to G.L.G. The experimentalwork done in Washington was supported by the NationalScience Foundation CHE-9817811 and was performed atthe W. R. Wiley Environmental Molecular Sciences Laboratory,a national scientific user facility sponsored by DOE’sOffice of Biological and Environmental Research and locatedat the Pacific Northwest National Laboratory, operatedfor DOE by Battelle. The Franck–Condon simulations wereperformed using FCFGAUS, and these authors would like tothank K. M. Ervin for making this program available tothem.1 A. S. C. Cheung, N. Lee, A. M. Lyyra, and A. J. Merer, J. Mol. Spectrosc.95, 2131982.2 D. W. Green, G. T. Reedy, and J. G. Kay, J. Mol. Spectrosc. 78, 2571979.3 T. C. Steimle, D. F. Nachman, J. E. Shirley, and A. J. Merer, J. Chem.Phys. 90, 5360 1989.4 A. J. Merer, Annu. Rev. Phys. Chem. 40, 4071989.5 E. R. Fisher, J. L. Elkind, D. E. Clemmer, R. Georgiadis, S. K. Loh, N.Aristov, L. S. Sunderlin, and P. B. Armentrout, J. Chem. Phys. 93, 26761990.Downloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

J. Chem. Phys., Vol. 119, No. 21, 1 December 2003 Iron monoxide clusters111456 P. B. Armentrout, Annu. Rev. Phys. Chem. 52, 4232001.7 L. S. Wang, H. Wu, and S. R. Desai, Phys. Rev. Lett. 76, 4853 1996.8 L. S. Wang, J. Fan, and L. Lou, Surf. Rev. Lett. 3, 695 1996.9 J. Fan and L. S. Wang, J. Chem. Phys. 102, 87141995.10 H. Wu, S. R. Desai, and L. S. Wang, J. Am. Chem. Soc. 118, 5296 1996.11 C. W. Bauschlicher, Jr. and P. Maitre, Theor. Chim. Acta 90, 1891995.12 J. F. Harrison, Chem. Rev. Washington, D.C. 100, 679 2000.13 G. L. Gutsev, B. K. Rao, and P. Jena, J. Phys. Chem. A 104, 5374 2000.14 Y. Nakao, K. Hirao, and T. Taketsugu, J. Chem. Phys. 114, 7935 2001.15 L. S. Wang, H. S. Cheng, and J. Fan, J. Chem. Phys. 102, 9480 1995.16 L. S. Wang and H. Wu, in Advances in Metal and Semiconductor Clusters,Vol. 4, Cluster Materials, edited by M. A. Duncan JAI, Greenwich,1998, pp. 299–343.17 M. J. Frisch, G. W. Trucks, H. B. Schlegel et al., GAUSSIAN 98, RevisionA.11, Gaussian, Inc., Pittsburgh, PA, 2001.18 A. J. H. Wachters, J. Chem. Phys. 52, 1033 1970.19 P. J. Hay, J. Chem. Phys. 66, 4377 1977.20 K. Raghavachari and G. W. J. Trucks, J. Chem. Phys. 91, 1062 1989.21 M. J. Frisch, J. A. Pople, and J. S. Binkley, J. Chem. Phys. 80, 32651984.22 A. D. Becke, Phys. Rev. A 38, 30981988.23 C. Lee, W. Yang, and R. G. Parr, Phys. Rev. B 37, 7851988.24 J. P. Perdew, Phys. Rev. B 33, 8822 1986.25 J. P. Perdew and Y. Wang, Phys. Rev. B 45, 13244 1992.26 J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 38651996.27 A. D. Becke, J. Chem. Phys. 98, 56481993.28 P. J. Stevens, F. J. Devlin, C. F. Chablowski, and M. J. Frisch, J. Phys.Chem. 98, 11623 1994.29 G. L. Gutsev and C. W. Bauschlicher, Jr., J. Phys. Chem. A 105, 108222001.30 X. Li, H. Wu, X. B. Wang, and L. S. Wang, Phys. Rev. Lett. 81, 19091998.31 L. S. Wang and X. Li, in Clusters and Nanostructure Interfaces, edited byP. Jena, S. N. Khanna, and B. K. Rao World Scientific, Singapore, 2000,pp. 293–300.32 J. Akola, M. Manninen, H. Hakkinen, U. Landman, X. Li, and L. S. Wang,Phys. Rev. B 60, 11297 1999.33 L. S. Wang, X. Li, and H. F. Zhang, Chem. Phys. 262, 532000.34 H. J. Zhai, L. S. Wang, A. N. Alexandrova, and A. I. Boldyrev, J. Chem.Phys. 117, 7917 2002.35 W. Grochala and R. Hoffmann, J. Phys. Chem. A 104, 9740 2000.36 S. Chrétien and D. R. Salahub, Phys. Rev. B 66, 155425 2002.37 K. P. Huber and G. Herzberg, Constants of Diatomic Molecule Van NostrandReinhold, New York, 1979.38 P. Chen in Unimolecular and Bimolecular Reactions Dynamics, edited byC.-Y. Ng, T. Baer, and I. Powis Wiley, Chichester, 1994 p. 371.39 K. M. Ervin, T. M. Ramond, G. E. Davico, R. L. Schwartz, S. M. Casey,and W. C. Lineberger, J. Phys. Chem. A 105, 10822 2001.40 P. C. Engelking and W. C. Lineberger, J. Chem. Phys. 66, 5054 1977.41 D. G. Leopold and W. C. Lineberger, J. Chem. Phys. 85, 511986.42 C. W. Bauschlicher, Jr. and G. L. Gutsev, Theor. Chem. Acc. 108, 272002.Downloaded 25 Mar 2007 to Redistribution subject to AIP license or copyright, see

More magazines by this user
Similar magazines