photodissociation spectroscopy and dynamics of mg+ ...wlu/jcp6970.pdfsponer analysis gives the...

Photodissociation spectroscopy and dynamics of Mg+-formaldehydeW.-Y. Lu, T.-H. Wong, Y. Sheng, and P. D. Kleiber Citation: J. Chem. Phys. 117, 6970 (2002); doi: 10.1063/1.1507584 View online: http://dx.doi.org/10.1063/1.1507584 View Table of Contents: http://jcp.aip.org/resource/1/JCPSA6/v117/i15 Published by the American Institute of Physics. Additional information on J. Chem. Phys.Journal Homepage: http://jcp.aip.org/ Journal Information: http://jcp.aip.org/about/about_the_journal Top downloads: http://jcp.aip.org/features/most_downloaded Information for Authors: http://jcp.aip.org/authors

Downloaded 10 Apr 2012 to 128.112.123.27. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

http://jcp.aip.org/?ver=pdfcov

http://careers.physicstoday.org/post.cfm

http://jcp.aip.org/search?sortby=newestdate&q=&searchzone=2&searchtype=searchin&faceted=faceted&key=AIP_ALL&possible1=W.-Y. Lu&possible1zone=author&alias=&displayid=AIP&ver=pdfcov

http://jcp.aip.org/search?sortby=newestdate&q=&searchzone=2&searchtype=searchin&faceted=faceted&key=AIP_ALL&possible1=T.-H. Wong&possible1zone=author&alias=&displayid=AIP&ver=pdfcov

http://jcp.aip.org/search?sortby=newestdate&q=&searchzone=2&searchtype=searchin&faceted=faceted&key=AIP_ALL&possible1=Y. Sheng&possible1zone=author&alias=&displayid=AIP&ver=pdfcov

http://jcp.aip.org/search?sortby=newestdate&q=&searchzone=2&searchtype=searchin&faceted=faceted&key=AIP_ALL&possible1=P. D. Kleiber&possible1zone=author&alias=&displayid=AIP&ver=pdfcov


http://link.aip.org/link/doi/10.1063/1.1507584?ver=pdfcov

http://jcp.aip.org/resource/1/JCPSA6/v117/i15?ver=pdfcov

http://www.aip.org/?ver=pdfcov


http://jcp.aip.org/about/about_the_journal?ver=pdfcov

http://jcp.aip.org/features/most_downloaded?ver=pdfcov

http://jcp.aip.org/authors?ver=pdfcov

JOURNAL OF CHEMICAL PHYSICS VOLUME 117, NUMBER 15 15 OCTOBER 2002

Photodissociation spectroscopy and dynamics of Mg ¿-formaldehydeW.-Y. Lu, T.-H. Wong, Y. Sheng, and P. D. KleiberDepartment of Physics and Astronomy and Optical Science and Technology Center, University of Iowa,Iowa City, Iowa 52242

~Received 27 June 2002; accepted 26 July 2002!

We have carried out photodissociation spectroscopy studies of the bimolecular complexMg1(H2CO) in the visible and near-uv regions. The work is supported by electronic structurecalculations of the ground and low-lying excited states of the complex. Mg1-formaldehyde is boundin a C2v Mg1

uOvCH2 geometry with a theoretical bond energy ofDe9(Mg-OCH2)51.35 eV.The complex shows absorption bands that correlate with Mg1-based and formaldehyde-basedradiative transitions. The lowest-energy band is assigned asA 2A8(2B1)←X 2A1 , to an excited stateof mixed Mg1(3pp) and H2CO(p* ) orbital character. The band exhibits complex vibrationalstructure with considerable excitation of the CH2 out-of-plane wag and CvO stretch modes; thevibrational frequencies are shifted dramatically from their values in the ground state, showing theeffect of a significant weakening of the CvO bond and out-of-plane distortion of the complex.Excitation in the Mg1-based B 2A8(2B2)←X 2A1 band shows predominantly low-frequencyvibrational motions assigned to the intermolecular in-plane wag and Mg-O stretch modes. Birge–Sponer analysis gives the Mg–O bond energy in the ground state asDe951.29 eV. Partially resolvedrotational substructure clearly demonstrates a change in geometry from a linear or near linearMg-O-C (C2v) ground state to a bent (Cs) excited state@u~Mg-O-C!5139°63°#. Spectroscopicrotational constants are in very good agreement withab initio predictions for this band. TheMg1-basedD 2A1←X 2A1 band also exhibits pronounced vibrational structure including strongMg–O and CvO stretch signals, consistent with formation of a partial Mg–Os bond in this state.Mg1 is the major dissociation product through the uv–visible region. However, in theB←X, C←X, and D←X absorption bands, we also observe a substantial branching to the reactivedissociation product MgH1. The reactive branching ratio increases with photon energy through theabsorption bands, reaching a reactive quantum yield of;1

3 in the D←X band. Our results suggestthat there is no significant activation barrier to reaction above the endothermicity. ©2002American Institute of Physics.@DOI: 10.1063/1.1507584#

in

sn’’hctpeleo

ps

thn

iostou

in

-

ndtralyde-lts

east,

athe

genasmvi-

the

I. INTRODUCTION

Significant effort has been devoted to understandmetal-ion–hydrocarbon bimolecular interactions.1–5 Much ofour recent work has focused on the chemical interactionlight metal ions with small alkane and alkene hydrocarbousing photodissociation spectroscopy ‘‘half-collisionmethods.6–11These prototype reactions offer valuable insiginto the fundamental dynamics of C–H and C–C bond avation processes by metal ions. The photodissociation stroscopy of a weakly bound metal-ion–hydrocarbon compis an important tool for probing chemical interactions. Phtodissociation spectroscopy has been used by many grouprobe the interactions of both light~group-II! metal ions andtransition-metal ions with a variety of molecular ligands.11–21

The method can give quantitative information aboutstructure and bonding of the complex in both ground aexcited states, and provide an important test ofab initio elec-tronic structure calculations. In addition, photodissociatspectroscopy of a weakly bound precursor complex mimicbimolecular ‘‘half-collision’’ and can offer unique insighinto the molecular orbital interactions and nonadiabatic cpling effects that determine the reaction pathway.

Recently we used photodissociation spectroscopy to

6970021-9606/2002/117(15)/6970/12/$19.00

Downloaded 10 Apr 2012 to 128.112.123.27. Redistribution subject to AIP l

g

ofs

ti-c-x-to

ed

na

-

-

vestigate Mg1-acetaldehyde and Al1-acetaldehyde complexes in the visible and near-uv spectral regions.8–10 In eachcase the metal ion binds to the O atom in end-onM 1 – O–Cgeometry because the binding is primarily electrostatic aoxygen has a high electronegativity. We observed specbands that correlate to both metal-centered and acetaldehcentered radiative transitions in the complex. Our resusuggest that theM 1-centered 3pp(A9)←3ss(A8) andacetaldehyde-centeredp* (A9)←n(A8) transitions aremixed, resulting in substantial vibrational excitation in thcomplex and leading to broad, unresolved bands. In contrthe predominantlyM 1-centered 3pp(A8)←3ss(A8) ab-sorption bands for Mg1- and Al1-acetaldehyde each showprominent vibrational progression that can be assigned tointermolecular in-plane bend. This is consistent withab ini-tio calculations that predict a significant geometry chanfrom a near-linear to a bentM – O–C backbone on excitatioto 3pp(A8). However, detailed spectroscopic analysis wlimited by the complicated nature of the vibronic spectruand fast relaxation that resulted in broad, overlappingbronic resonance features.

In Mg1-acetaldehyde we were also able to probehigher-energy 3ps 2A8←3ss 2A8 Mg1-centered absorption

0 © 2002 American Institute of Physics

icense or copyright; see http://jcp.aip.org/about/rights_and_permissions

cytwna

ls

dco

vearoitaly

odny

-

ofouolghi

om

gw

soonth–

rem

-ino-d

asu

aydyd

thco

d towas

ies

im-intosst theedut

op---drn-ble

POby

taleenatotifi-

eler-cal-torsced

fur-de-

t tosernce

ess.nd-on

y

is

-

dge-ticO

6971J. Chem. Phys., Vol. 117, No. 15, 15 October 2002 Dynamics of Mg1-formaldehyde

band.8,9 This band shows strong activity in the low-frequenintermolecular bending and stretching modes and inhigh-frequency intramolecular acetaldehyde vibratiomodes, a CCO deformation and the CvO stretch. Whilenonreactive dissociation to Mg1 is the dominant photolysischannel throughout the spectrum, excitation in this band ashows a weak reactive branching to MgH1, MgCHO1, andMgCH3

1 products that result from C–H and C–C bonbreaking.9 Isotope substitution experiments verify that reation involves activation of the aldehydic C–H bond and none of the methyl C–H bonds. Reaction was only obserfrom the 3ps(2A8) state, although the reasons for the appent s-like orbital alignment preference for reaction are nfully understood. Simple energetics may play a role; exction to Mg1@3ps(2A8)# occurs at higher energies, possibabove a reaction barrier in the excited state.

Here we report on similar spectroscopic studiesMg1-formaldehyde. Formaldehyde is the simplest aldehymaking it an ideal system for investigating metal–carbointeractions. The relatively small size and simplicity of formaldehyde mean that accurateab initio theoretical calculationsshould be possible.22 Photodissociation spectroscopyMg1-formaldehyde also provides a revealing compariswith results from our previous work on the analogous bmore complicated acetaldehyde complex and helps to resopen issues regarding the structure and chemical bondinthe excited states, and the physical and chemical quencdynamics.8,9

We have investigated the photodissociation spectroscof Mg1(H2CO) over the spectral range from 220 to 600 nWe have also carried out supportingab initio electronicstructure calculations of the ground and low-lying enerlevels of Mg1-formaldehyde. Our experimental results shothat the Mg1-formaldehyde and -acetaldehyde photodisciation spectra are essentially similar and that the metal-ialdehyde complexes exhibit a distinct spectral characteris quite different from most other weakly bound metal-ionmolecule clusters we have investigated. Because of thetively smaller size of formaldehyde, the vibrational spectruis much less congested than for Mg1-acetaldehyde, and exhibits a more well-resolved resonance structure includpartially resolvedK-subband rotational structure. This prvides more detailed information about metal-ion–aldehybonding interactions and a more stringent test of theab initiocalculations.

II. EXPERIMENTAL ARRANGEMENT

The experimental apparatus and the application to mselected cluster photodissociation spectroscopy measments have been previously described.11 Mg1(H2CO) com-plexes were produced in the supersonic molecular beexpansion from a laser vaporization source. Formaldehvapor was obtained by heating solid paraformaldeh@(H2CO)n# to a temperature of;150 °C. The formaldehydevapor was then mixed to a concentration of;2% with Ar ina mixing cylinder. The seeded gas mix was used inpulsed gas valve at a backing pressure of 60 psi. The seharmonic of a pulsed Nd:YAG~yttrium aluminum garnet!


ol

o

-td-t-

fe,l

ntveinng

py.

y

-–at

la-

g

e

s-re-

mee

end

laser was focused onto the metal rod surface and timeoverlap the seeded gas pulse. The source mass spectrumclean, showing strong peaks only for the cluster ion familMg1(H2CO)n and MgArn

1 .Downstream from the source and molecular beam sk

mer, ion clusters were pulse-extracted and acceleratedthe flight tube of an angular reflectron time-of-flight maspectrometer. A pulsed mass gate was used to selecMg1(H2CO) parent cluster. The parent cluster was focusinto the reflectron of the time-of-flight apparatus. The outpfrom an injection-seeded Nd:YAG laser pumped tunabletical parametric oscillator~OPO! ~Spectra-Physics/QuantaRay PRO-250/MOPO SL! coupled with a Quanta Ray wavelength extension~WEX! system for frequency doubling anmixing, was time-delayed to excite the parent ion at the tuing point inside the reflectron. The OPO covers the visi~424–690 nm! and near-ir~730–1800 nm! spectral regionswith a bandwidth of,0.2 cm21. The near-uv region from212 to 345 nm was reached by frequency doubling the Ooutput. The region from 340 to 419 nm was accessedmixing the OPO output with the Nd:YAG laser fundamenat 1064 nm. This leaves a small gap in the spectrum betw419 and 424 nm. In principle, doubling the OPO outputaround 840 nm could cover this 5-nm interval. We did nmake an effort to fill this gap as it does not have any signcant effect on the spectral assignment.

Parents and daughter fragment ions were then reaccated to an off-axis microchannel plate detector in a typitandem time-of-flight arrangement. Two digital oscilloscopes, a multichannel scaler, and a set of gated integrawere used to monitor the mass spectrum and were interfato a laboratory personal computer to record the data forther analysis. The photodissociation action spectrum wastermined by normalizing the daughter signal with respecthe parent ion signal and laser power while scanning the lawavelength. Results from a series of laser power dependetests were consistent with a one-photon excitation procHowever, because of possible saturation effects for boubound transitions we cannot entirely rule out multiphoteffects in some of the observed bands.

III. ELECTRONIC STRUCTURE CALCULATIONS

Ab initio electronic structure calculations on theGAUSS-

IAN ’98 platform show Mg1-formaldehyde to be moderatelstrongly bound in an equilibrium Mg1-OCH2 C2vgeometry.23 The metal-oxide bond dissociation energyDe9(Mg-OCH2)51.51 eV at the Hartree-Fock HF/6-31111g(2d,2p) level, and the Mg1 – O equilibrium bond dis-tance isRMg–O51.991 Å. The formaldehyde moiety is relatively undistorted, with a planar OCH2 structure and a C–Obond length ofRC–O51.201 Å. These results are in gooagreement with results from earlier calculations by Partridand Bauschlicher.22 They showed that the bonding is predominantly electrostatic in nature. The strong electrostabond results in part from a polarization of charge on theatom that also slightly weakens and stretches the CvObond. We have also carried out an MP2~Moller-Plessetsecond order perturbation! optimization calculation


in

nt.

6972 J. Chem. Phys., Vol. 117, No. 15, 15 October 2002 Lu et al.

TABLE I. Properties of some low-lying electronic states of MgCH2O1 from an ab initio calculation.~Experimental values from this work are shownparentheses.!

StateaX 2A1

C2v

A 2B1b

C2v

B 2A8Cs @s(yz)→s#

D 2A1

C2v

Vertical excitationenergy~eV!

0.0 3.44~2.91! 3.63 ~3.70! 5.08 ~5.02!

Optimized structureRMg–O ~Å) 1.991 1.866 1.948 1.939/Mg–O–C ~°! 180.0~180! 180.0 144.4~139! 180.0

Vibration mode frequency~cm21! & symmetrya

n1 Mg–O–Cip bend (b2)

75 120 187~212! 47 ~85!

n2 Mg–O–Cop bend (b1)

164 262 258 186~234!

n3 Mg–O stretch (a1) 372 480 465~462! 356 ~360!n4 CH2 wag (b1) 1359 1349~640!b 1390 1343n5 CH2 rock (b2) 1387 1295 1360~989! 1318n6 CH2 scissors (a1) 1647 1609 1639 1635n7 C–O stretch (a1) 1902 1772~1240!b 1885 1869~1651!n8 CH2 s stretch (a1) 3200 3222 3199 3215n9 CH2 a stretch (b2) 3318 3351 3323 3336

aThe symmetry and mode numbering are referred to theC2v equilibrium ground-state geometry withz as the symmetry axis and the complex lying in theyzplane.

bCIS calculations predict aC2v2B1 state. Our data suggest the state is actually2A8 in Cs @s(xz)→s#. See text for a detailed discussion of this assignme

rgon

c

abt tnthththun

iela

itISrr

esedar

red

n--s aree I.

-

op-

onda-entialsi--erly

leya-

e

hea

@MP2/6-31111g(2d,2p)# and find a similarC2v complexbut with a weaker Mg1 – O bond, RMg–O52.039 Å andDe9(Mg-OCH2)51.35 eV.

The excited-state structure is complicated, with a lanumber of interacting electronic states. Radiative transitito low-lying states with both metal-centeredp←s andaldehyde-centeredp* ←n character are possible. We expethat the metal-based transitions correlating with Mg1(3p←3s) excitation should be stronger and dominate thesorption spectrum. Electrostatic arguments suggest thaMg1-based 3pp excited states will be more attractive thathe 3ss ground state since the electron density alongintermolecular axis is reduced, leading to an increase instrength of the ion-dipole attraction. On the other hand,3ps state should be less strongly bound than the grostate owing to the increase ins repulsion along the Mg–Obond axis.

To help clarify the spectral assignment we have carrout a configuration interaction with single excitation calcution @CIS/6-31111g(2d,2p)# for the low-lying doublet ex-cited states of the complex. Our previous experience wsimilar metal-ion–ligand complexes indicates that the Cmethod can be quite accurate for excited states that cospond to predominantly metal-centered excitations, whildoes a rather poor job for states with significant ligand-baexcitation character, that may involve a change in bond orand where configuration mixing and correlation effectsimportant. Results from theab initio calculations support thequalitative arguments above. The calculations find thstrong Mg1-based absorption bands. Two of the ban@3pp(2B1) and 3pp(2A8)←3ss(2A1)] are significantlyredshifted from the Mg1 atomic (3p←3s) absorption line,with vertical excitation energies of 3.44 eV~;360.9 nm! and3.63 eV ~;341.7 nm!, respectively. The 3ps(2A1)


es

t

-he

eeed

d-

h

e-itd

ere

es

←3ss(2A1) band is blueshifted with a vertical excitation eergy of 5.08 eV~;244.0 nm!. We have optimized the geometry of the three metal-based excited states and the resultdiscussed in more detail below and summarized in TablCIS calculations also predict a weaker transition (2A2

←2A1) in the near-uv region at 4.61 eV~;268.8 nm! thatcorrelates with the predominantly aldehyde-basedp* ←ntransition.

The first excited state,A 2B1 , correlates with a predominantly Mg1(3pp)-based state with the metal 3p orbitalaligned perpendicular to the plane of the molecule. Thetimized geometry for this state retainsC2v symmetry, butshows a considerable shortening in the intermolecular bto an equilibrium Mg–O bond length of 1.866 Å. The drmatic shortening of the intermolecular bond is consistwith a strong chemical interaction and formation of a partMg–O p bond in this excited state. Indeed, while the trantion is predominantly Mg1-centered, a careful orbital analysis shows that the radiative transition can be more propdescribed as excitation from the Mg1-baseds orbital into avirtual orbital of b1 symmetry that also has appreciabCvO-basedp* (b1) character. A normal-mode frequenccalculation finds three low-frequency intermolecular vibrtions: then1 Mg–O–Cintermolecular in-plane bend (b2) at120 cm21, the n2 Mg–O–C intermolecular out-of-planebend (b1) at 262 cm21, and then3 Mg–O intermolecularstretch (a1) at 480 cm21. The vibrational frequencies at thCIS level are summarized in Table I.

The second excited state,B 2A8(2B2), also correlateswith a Mg1(3pp)-based state, with the metal 3p orbitallying in the molecular plane but roughly perpendicular to tMg–O bond. Geometry optimization for this state showsdramatic structural change fromC2v to Cs geometry. The


lsoisre

r-ere

r-

tat be-eed3rge-

i

reth

apve

ntpt

ords

In

y

ce

ofrly

incu-f

ces

larhat

al

s toto

asef in-

-

re;the

tryre


atoms remain in the molecular plane but the Mg–O–Cbondangle changes from 180.0° to 144.4°. The Mg–O bond ashows slight contraction to an equilibrium bond length1.948 Å. The bonding is still primarily electrostatic in thstate-but the complex bends in order to minimize stericpulsion between the O nonbondingn orbitals and the in-plane Mg1 3pp orbital. The state is2A8 in Cs symmetry butis derived from the2B2 state of theC2v complex. The largechange in the Mg–O–Cbond angle suggests that the intemolecular in-plane bending vibration will be active in thabsorption spectrum. A frequency calculation gives the thlow-frequency intermolecular vibrations as then1 Mg–O–Cin-plane bend@a8(b2)# at 187 cm21, the n2 Mg–O–Cout-of-plane bend@a9(b1)# at 258 cm21, and then3 Mg–Ostretch@a8(a1)# at 465 cm21 ~Table I!.

The third excited state,C 2A2 , correlates predominantlywith the formaldehyde-based excited state that arises fromp* ←n excitation ~from a nonbonding oxygen-centered obital to thep* antibonding orbital on the CvO bond!. Mo-lecular orbital considerations suggest that the electrosinteractions for the ground state and excited state will nomarkedly different.8 We would expect that the formaldehydbased absorption band in the complex should be relativunshifted from its position in the bare molecule. In isolatformaldehyde thep* ←n transition origin appears near 35nm. The CIS calculation predicts a vertical excitation eneof 4.61 eV~;268.8 nm! for this band; as noted above, bcause of the limitations in the CIS method, this predictionnot expected to be especially reliable.

The fourth excited state,D 2A1 , correlates to theMg1(3ps)-based excited state with the Mg1 3p orbital ly-ing along the Mg–O intermolecular axis. The complextains C2v geometry and shows some shortening ofMg—O bond to 1.939 Å and stretching of the CvO bond to1.206 Å. The shortening of the intermolecular bond andtractive nature of this state are not expected from simelectrostatic arguments. The CIS level calculations, howeshow the formation of a partial Mg—O s-bond in this state.These predictions are in good agreement with experimedata as discussed below. A frequency calculation in the omized geometry gives the three intermolecular vibrationsbe then1 Mg–O–C in-plane bend (b2) at 47 cm21, the n2

Mg–O–C out-of-plane bend (b1) at 186 cm21, and then3

Mg–O stretch (a1) at 356 cm21 ~Table I!.

IV. RESULTS AND DISCUSSION

A. Electronic band assignment

The photodissociation action spectrum fMg1-formaldehyde shows four distinct absorption banthree redshifted and one blueshifted from the Mg1(3s→3p) resonance at 280 nm, labeled asA,B,C,D←X ~Fig.1!. All of the bands exhibit some vibrational structure.addition, the vibrational resonances of theB←X band start-ing at 29 175 cm21 also show partially resolvedK-subbandrotational structure. Band assignments are based largelthe ab initio model predictions.

The lowest-energy band starts from around 17 200 cm21

and exhibits a long series of broad vibrational resonan


of

-

e

an

tice

ly

y

s

-e

t-ler,

alti-o

,

on

s,

peaking near 24 000 cm21 and extending to;29 000 cm21.We assign this band asA 2A8(2B1)←X 2A1 , correlating pre-dominantly to the Mg1-based 3pp←3ss transition with theMg1 p orbital lying perpendicular to the molecular planethe ground-state molecule. The band position is in faigood agreement with theab initio predicted vertical excita-tion energy to theA 2B1 state of;27 700 cm21. However,based on the complicated vibrational structure observedthis band we believe the complex distorts out of the molelar plane fromC2v to Cs ~vide infra!. The predicted state o2B1 symmetry inC2v geometry then becomes2A8 in Cs .

Starting from around 29 175 cm21 we observe a newband consisting of a series of sharp vibrational resonanwith rotational substructure, extending to;31 200 cm21.This band is assignedB 2A8(2B2)←X 2A1 and correlates tothe Mg1-based 3pp←3ss transition with the Mg1 3p or-bital lying in the symmetry plane and roughly perpendicuto the Mg–O bond. The experimental data clearly show tthe molecule bends in-plane fromC2v to Cs symmetry onexcitation to this band, with2B2 resolving to 2A8 in thelower symmetry~vide infra!. The band peaks at;29 900cm21, which agrees very well with the CIS-predicted verticexcitation energy for this band of 29 250 cm21.

Toward the high-energy side of theB←X band we ob-serve the onset of a new continuum band that appearshow several broad undulations~Fig. 1!. This band starts a;31 000 cm21 ~although the origin is not obvious due toverlap withB←X) and extends to;37 000 cm21. This isclose to the position of thep* (1A9)←n(1A1) formaldehydetransition and accordingly the band is assignedC 2A9(2A2)←X 2A1 . The corresponding transition in barformaldehyde has been the subject of a large number ovestigations and has been well reviewed in literature.24–28

The formaldehydep* (1A9)←n(1A1) band is weak and extends from a band origin at 28 188 up to;43 500 cm21,showing a complicated vibrational and rotational structuvibrational analysis shows formaldehyde is nonplanar in1A9 upper electronic state, with a pyramidal structure inCs

symmetry. The Mg1-formaldehyde complex is also likely tohave aCs structure in this excited state, although geomeoptimization failed at the CIS level. The state is therefo

FIG. 1. Photodissociation spectrum of MgCH2O1 from 600 to 220 nm.


on

th

73tud

ita

her,

;oltu

.

ur

thxi

rga

oina

lonan

-

re-

al

senges,

e

d--

t.t of-erthehe

thee-

s-IS

theed

n-tial

o asHifi-

e-c--

odeb-to

-

e

-res-

luum


assigned as2A9 and is derived from the2A2 state of theplanar conformation. The CIS-predicted vertical excitatienergy for this band is 37 200 cm21; as anticipated, theagreement with the observed band position is poor fortransition.

The blueshifted band of Fig. 1, with an origin at 40 4cm21, also shows pronounced vibrational resonance struc~including activity in both low-frequency intermolecular anhigh-frequency intramolecular vibrational modes! and is as-signed asD 2A1←X 2A1 in C2v geometry, correlating to theMg1-based 3ps←3ss transition. This band position is invery good agreement with the CIS-predicted vertical exction energy of 40 980 cm21.

Mg1 is the dominant photolysis product throughout tspectral range covered in these experiments. Howevesmall reactive branching to the MgH1 product is observedthe reactive branching appears to rise as function of photsis energy through the bands reaching a maximum quanyield of ;33% in theD←X band.

B. Vibrational assignments

1. A 2A8( 2B1)]X 2A1 band

An expanded view of theA←X band is shown in Fig. 2The band extends from;17 200 to;29 000 cm21 and con-tains a long series of broad vibrational resonance feat~Table II!. These broad features are separated by;650 cm21,and each contains several sharp peaks. However, evensharp peaks are quite broad with a full width at half mamum ~FWHM! of ;70 cm21 in the 20 000–26 000-cm21

region. The FWHM becomes larger towards higher eneThis linewidth is too large to be associated with rotationbroadening given our cluster temperature ofTr;15 K ~videinfra!. The broadening in this band is also independentlaser power and is probably due to homogeneous broadenindicating a very short upper-state lifetime. The vibrationband extends over 10 000 cm21; this degree of vibrationaexcitation implies a large geometry change on excitatiThe Franck–Condon envelope is falling towards the red

FIG. 2. Expanded view of theA 2A8(2B1)←X 2A1 band. All transitions areassumed to originate from the vibrational ground state but the absovibrational numbering in the upper state is not known. The quantum nbers here are assigned relative to the first observed peak.


is

re

-

a

y-m

es

ese-

y.l

fg,

l

.d

the band origin is not obvious. The large redshift~.2.28 eV!of the band origin from the Mg1 3p←3s resonance transition indicates the interaction between Mg1 and formalde-hyde in theA state is very strong.

The vibrational structure in this band is revealing. Caful analysis of the spectrum in the 19 000–26 000-cm21 re-gion of strong absorption shows two overlapping vibrationprogressions based on mode frequencies of;650 and;1250 cm21. There are two plausible assignments for themodes. Our CIS calculations predict a large geometry chain the Mg–O bond length on excitation in this band. Thuwe should expect then3 Mg–O intermolecular stretch modto be active with a CIS predicted frequency of 480 cm21.Based on the CIS results the higher-energy mode at;1250cm21 could then be assigned either as then5 CH2 in-planerocking mode with a calculated frequency of 1295 cm21 orthe n4 CH2 out-of-plane wagging mode with a calculatefrequency of 1349 cm21. However, the observed mode frequency at;650 cm21 is substantially higher than the predicted Mg–O stretch frequency of 480 cm21. It is possiblethat the CIS calculation is simply in error by this amounThe CIS method does not adequately handle the effecmixing in some formaldehydep* -orbital character in the excited state; the effect of this mixing could result in a strongMg–O bond with a higher mode energy than predicted atCIS level. It is also difficult to understand, based on tCIS-predicted equilibrium geometry for theA 2B1 excitedstate, why there should be such strong activity in any ofhigh-frequency vibrational modes of formaldehyde. The dgree of vibrational excitation clearly points to a highly ditorted excited state complex that is not reflected in the Ccalculations.

There is, however, another possible assignment. AsCIS calculations show, the radiative transition is of mixcharacter, to an excited (b1) state with both Mg1 3p-orbitaland formaldehydep* -orbital character. Transferring electrodensity into thep* -antibonding lowest unoccupied molecular orbital ~LUMO! of formaldehyde leads to a significanweakening of the C–O bond and formation of a partMg–O p bond. As noted above, thep* ←n transition inisolated formaldehyde causes a distortion out-of-plane tpyramidal geometry.25 The resulting vibronic spectrum icomplicated and includes substantial excitation of the C2

out-of-plane wag and C–O stretch modes. If there is signcant transfer of electron density into thep* LUMO in thisband of Mg1-formaldehyde, then we expect the formaldhyde ligand to distort out-of-plane into the pyramidal struture that characterizes the1A9 excited state of bare formaldehyde. Interestingly, in the1A9 excited state of isolatedformaldehyde there is a huge change in the vibrational mfrequencies.25 In particular, because the C–O bond is sustantially weakened by the addition of electron density inthe carbonylp* antibonding LUMO, the C–O stretch frequency drops from 1746 cm21 in the ground state to 1182cm21 in the 1A9 excited state. Similarly there is a largchange in theb1 CH2 wag frequency from 1167 cm21 in theground state to 683 cm21 in 1A9. These values are remarkably close to the observed mode frequencies. With someervation, we assign the lower-energy mode at;650 cm21 to

te-


ground

are


Dow

TABLE II. Vibrational mode assignmentsa.

A 2A8(2B1)←X 2A1 B 2A8(2B2)←X 2A1 D 2A1←X 2A1

Position~cm21! Assignmentb

Position~cm21! Assignment

Position~cm21! Assignment

17 278 00(?) 29 175.0 00 40 473 00

17 959 41 29 380.7 11 40 558 11

18 522 71 29 573.5 12 40 707 21

18 650 42 29 631.4 31 40 810 31

19 205 7141 29 749.5 13 40 889 3111

19 307 43 29 836.5 3111 41 027 3121

19 768 72 29 890.0 14 41 128 32

19 892 7142 30 001.2 15 41 213 3211

20 439 7241 30 031.8 3112 41 421 33

20 559 7143 30 082.4 32 42 124 71

20 999 73 30 164.1 51 42 205 7111

21 100 7242 30 206.0 3113 42 459 7131

21 660 7341 30 284.7 3211

21 768 7243 30 343.5 3114

22 218 74 30 449.0 3115

22 309 7342 30 482.2 3212

22 415 7244 30 526.7 33

22 873 7441 30 620.4 5131

22 953 7343 30 657.0 3213

23 067 7245 30 727.6 3311

23 413 75 30 791.7 3214

23 505 7442 30 892.8 3215

23 578~?! 7344 30 927.0 3312

24 054 7541 31 073.0 5132

24 116 7443

24 238 7345

24 634 76

24 675 7542

24 750 7444

25 278 7641

25 311 7543

25 366 7445

25 791 77

25 882 7642

25 972 7446

26 391 7741

26 462 7643

26 960 78

27 557 7841

28 048 79

28 147 7842

28 635 7941

aSee Table I for mode numbering scheme. All transitions are assumed to originate from the vibrationalstate.

bThe absolute quantum numbering of the modes in theA←X band is not known. The quantum numbers hereassigned relative to the first observed peak at 17 278 cm21.

-

snotgntonbbu

278rge

yhar-dethe

e

andhs

e

then4 CH2 wag and the higher-energy mode at;1250 cm21

to then7 C–O stretch in a2A8 state of aCs complex.The vibrational structure inA←X can thus be consid

ered as a combination of only two vibrational modes. Then7

C–O stretch mode at 1250 cm21 forms a long progressionthat becomes the backbone of the absorption band. Theond mode, then4 CH2 wag, forms a new progression built oeach member of then7 series. As noted the band origin is nobvious. Because of the severe homogeneous broadeninhave not attempted to carry out isotope shift experimeFor these reasons we cannot assign absolute vibratiquantum numbers to the progressions. The quantum numfor this band given in Table II and Fig. 2 are not absolute,

nloaded 10 Apr 2012 to 128.112.123.27. Redistribution subject to AIP l

ec-

wes.al

erst

are assigned relative to the first observed peak at 17cm21. Birge–Sponer plots of each mode show rather laslopes with anharmonicity constantsvexe5661 cm21 forn7 andvexe5462 cm21 for n4 . Because of the uncertaintregarding the absolute numbering, and the rather large anmonicity we cannot easily determine the fundamental mofrequencies. Averaging over the observed spacings forseries labeled 714n and 7n41 give approximate values for thmode frequencies ofv751240 cm21 andv45640 cm21.

As noted above the vibrational resonances in this bexhibit significant homogeneous linewidths. The linewidtare consistent with an upper-state lifetime of,500 fs or,7vibrations of the Mg–O stretch~the mode that couples to th


tho

tallug

a

rla

al

-

ou

rb

ree-

.e

blehll

or

in

ap-

tedo-

iseard

-r to

–

he

ncy

tedernnictal

w-is

earf

e-

herlane

-be

notap-gp-tWef the

e offre-


dissociation coordinate!. Vibrational predissociation to anexcited state asymptote is not energetically possible inspectral region: there is not enough energy available to bbreak the Mg–O bond and reach any of the excited-sasymptotes. Dissociation must involve an electronicanonadiabatic transition to the ground-state surface throvibronic coupling~internal conversion!. In the Cs complexboth states are2A8.

These ‘‘half-collision’’ spectroscopic results suggestdynamical model for the quenching of Mg1* (3p) in a low-energy, orbitally aligned 3pp(2B1), O-end-on collision withH2CO. In this symmetry the Mg1 3p orbital is aligned per-pendicular to the molecular symmetry plane and can ovethe formaldehydep* LUMO, allowing for efficient transferof electron density into the carbonyl antibonding orbitQuenching will occur primarily by vibronic coupling(E-E,V energy transfer! through a long-lived collision complex, and result in Mg1(3s) and vibrationally hot H2COproducts with strong activity in the formaldehyde CvOstretch and CH2 wagging modes.

These conclusions are also consistent with our previstudy of the related Mg1-acetaldehyde complex.8,9 In thatcase the lowest-energy band appears as a weak and bunresolved continuum. We argued that the lack of observaspectral structure was due to an analogousE-E/V quenchingprocess accompanied by fast intramolecular vibrationallaxation ~IVR!, and resulting in a complicated and unrsolved vibrational spectrum.

2. B 2A8]X 2A1 band

An expanded view of theB←X band is shown in Fig. 3The first several peaks form an obvious progression whilhigh energy the spectrum becomes more congested~TableII !. Each resonance peak exhibits a characteristic doustructure with an intensity ratio of about 2:3. The doubsplitting is ;14 cm21 and remains roughly constant througthe entire band. The doublet splitting arises from partiaresolved rotational structure and will be described in mdetail below. First we analyze the vibrational structure.

The band origin is at 29 175 cm21 and there is an obvi-ous five-membered progression built on this origin show

FIG. 3. Expanded view ofB 2A8(2B2)←X 2A1 band. All transitions areassumed to originate from the vibrational ground state.


isthteyh

p

.

s

oadle

-

at

lett

ye

g

a fundamental frequencyve;200 cm21. The low-energymode shows a large anharmonicity and the progressionproaches an abrupt end near;30 000 cm21. Birge–Sponeranalysis gives the fundamental mode frequencyve521269 cm21 and anharmonicity parametersvexe5263 cm21

and veye521.260.4 cm21. This low-frequency vibrationis too low in energy to be the intermolecular stretch expecnear 465 cm21, and must correspond to one of the intermlecular bending modes of the complex. The progressionvery anharmonic and the apparent dissociation limit n;30 000 cm21 is too low in energy to be any of the expecteexcited-state asymptotes for the Mg1 product. Thus, the apparent dissociation limit probably corresponds to a barriesome internal motion or isomerization.

CIS calculations of theB 2A8(2B2) state equilibrium ge-ometry show a slight contraction in the Mg–O bond~from1.99 to 1.95 Å!, and a rather dramatic change in the MgO–C in-plane bending angle~from 180° to 144°!. Based onthese CIS results we expect to find strong activity in tintermolecular Mg–O stretch and in-plane Mg–O–Cbend-ing modes. We therefore assign the prominent low-frequemode to then1 intermolecular in-plane wag@a8(b2)#. TheCIS predicted frequency for this mode inB 2A8 is 187 cm21,in very good agreement with the measured value,v1

5212 cm21. We assume that the apparent barrier of;800cm21 is associated with isomerization, perhaps associawith inversion through linearity or possibly H-atom transfin the bent complex. Our previous work oMg1-acetaldehyde also showed a similar short anharmoprogression of in-plane bending vibration with a fundamenfrequency of 148 cm21.8

There are three additional progressions in the lofrequency mode evident in the spectrum. One of thesebuilt on an origin at 29 631 cm21500

01456 cm21; this pro-gression is also five-membered with an apparent end n30 500 cm21. It is likely that this corresponds to excitation oone quantum of then3 intermolecular Mg–O stretch@a8(a1)# in the complex. The measured value,v3

5456 cm21, is in very good agreement with the CIS prdicted value of 465 cm21 for B 2A8. The second and thirdquanta of the stretch mode can also be identified at higenergies and each forms a new progression with the in-pbending mode~although the last progression is not complete!. The major features of this absorption band thus canassigned to combinations of then1 intermolecular in-planebend@a8(b2)# and n3 stretch modes@a8(a1)#. The overallassignments are summarized in Table II.

There is at least one additional progression that canbe assigned in this way. The progression is built on anparent origin at 30 164 cm21 and shows a vibrational spacinof 456 cm21, the intermolecular stretch frequency. The aparent origin is at 00

01989 cm21. Such a difference cannobe solved by the combination of intermolecular modes.believe this resonance corresponds to the first quantum on5 CH2 in-plane rocking mode@a8(b2)# (v55989 cm21)with a calculated frequency of 1360 cm21. This in-planerocking motion is expected based on the geometry changthe complex. However, the agreement with the predicted


h

tho

-

fr

-enn

e

intfo

e

-io

re

uco

-lao

eom

ns

tatio

ionto

is

urhel-ta

thadta

rgeults

e-

ffer-y.ohatear

e

t

onic

r

nd

ch-


quency for this mode is not very good, even if we apply tusual ‘‘90% rule’’ to theab initio frequency.

The Mg–O stretch progression can be used to deriveMg1-formaldehyde dissociation energy. The progressionthe stretch mode itself (30

n) is rather weak but the progression with one quantum of in-plane bend (30

n101) is strong.

Birge–Sponer analysis of the data gives the fundamentalquency asv35461.961.4 cm21, with an anharmonicityconstantvexe53.2260.33 cm21. In the Morse potential approximation the excited state Mg–O bond dissociationergy is given byD05ve

2/4vexe . This leads to a dissociatioenergy for the excited state ofD08(Mg–O)52.05 eV. Giventhe Mg1 3p←3s excitation energy of 4.42 eV and thB 2A8←X 2A1 band origin at 29 175 cm21, we obtain theground-state dissociation energy ofD09(Mg–O)51.25 eV.Because the complex is bound primarily by electrostaticteractions in bothX andB states, the difference in zero-poinenergies for these two states is small. With a correctionthe zero-point energy estimated from theab initio calcula-tions, the ground-state bond energy isDe9(Mg–O)51.29 eV. This result compares very favorably with the thoretical ab initio predicted value ofDe951.35 eV at theMP2/6-31111g(2d,2p) level. Of course, the Morse potential approximation and the long Birge–Sponer extrapolatlead to significant uncertainty in the experimental result.

We now turn our attention to the rotational substructuobserved in the vibrational resonances of this band. Anpanded view of the doublet rotational structure in one sresonance is shown in Fig. 4. The rotational constantstained from theab initio calculation show largeA-axis rota-tional constants withA@B;C in both the ground and excited states. We therefore model the complex as a prosymmetric top. For a symmetric top there are two typespossible transitions, perpendicular (DK561) or parallel(DK50). For the2B2←2A1 Franck–Condon transition wexpect a perpendicular band. However, because of the geetry change, the axis switching effect could introduce soforbidden (DK50) character into the transition.29,30 Wehave estimated the axis switching angle (u r) and find that theeffect should be small (u r;4°); this is in agreement withour rotational fitting analysis that shows the observed tration is predominantlyDK561.

Figure 4 shows the result of a simulation of the rotional structure assuming a purely perpendicular transiwith ortho:para ratio 1:3, and rotational constantsA959.1,B95C950.163 cm21 and A853.75, B85C850.188 cm21.The lower energy peak corresponds roughly to the transitK951→K850 while the higher-energy peak correspondsthe nearly overlapping transitionsK950→K851 and K951→K852. Note that there is no apparentDJ substructurewithin theseDK bands. The best-fit rotational temperatureTr;15 K. The best-fit homogeneous linewidth is;1.5cm21, which is very much larger than the bandwidth of olaser system~,0.2 cm21!. The spectrum was measured at tlowest laser powers consistent with a manageable signanoise level. However, because bound–bound transitionseasily saturated we cannot rule out the possibility thatobserved linewidth may be due in part to saturation broening. Dissociation in this state might proceed through ro


e

ef

e-

-

-

r

-

n

ex-hb-

tef

m-e

i-

-n

s

o-ree--

tional coupling to the lower-lyingA state that derives fromthe 2B1 state ofC2v symmetry.

The broadening of the rotational structure gives a launcertainty on the best-fit rotational constants. The resare A959.1061.40 and B9 (C9)50.16660.022 cm21 forthe lower state, andA853.7560.30 andB8 (C8)50.18860.023 cm21 for the B state. These are in fairly good agrement with the CISab initio predictions~Table III!. While theabsolute uncertainty on each parameter is large, the diencesA82A9 andB82B9 are known much more accuratelIn particular, the difference inA constants is very sensitive tthe Mg–O–Cbond angle and these results clearly show tthe bond angle changes markedly from a linear or near-linground stateu~Mg–O–C!5180°615° to a bent excited statu~Mg–O–C!5139°63°. This result gives the first definitiveexperimental evidence that Mg1-formaldehyde has a benstructure in this excited state. Note the 139°63° value ob-tained here is quite close to theab initio result of 144.4°.Thus the rotational analysis establishes that the electrexcitation is accompanied by a large change in the Mg–O–Cintermolecular bending angle, consistent with the CISab ini-tio calculations. These spectroscopic results for theB A8state of Mg1-formaldehyde are remarkably similar to ouprevious results on the analogous Mg1-acetaldehyde3pp(A8)←3ss(A8) band and support the assignments aconclusions drawn earlier from that work.8

These results suggest a dynamical model for the quening of Mg1* (3p) in a low-energy, orbitally aligned

FIG. 4. Rotational analysis of the 102 resonance peak of theB 2A8(2B2)

←X 2A1 band. See text for fitting details.

TABLE III. Comparison of rotational constants~cm21!.

Theoretical~HF! Experimental

Ground stateX 2A1

A9 9.626 9.161.4B9 0.165 0.16660.022C9 0.162 0.16660.022

Excited stateB 2A8(2B2)A8 4.334 3.7560.30B8 0.184 0.18860.023C8 0.177 0.18860.023


a

r,e

s

6

-h

e

tra

.aa

a

tortiulusi

ni

for-

tate

e–

oficaloffor

1

124

e foredr

vi-ther--

ac-

the

ies,igh-

–

g to

to-gths.

in

the

Mg

-


3pp(2B2), O-end-on collision with H2CO: quenching willbe inefficient and occur primarily through weak rotationcoupling to the2B1 surface, followed byE-E,V energytransfer quenching on2B1 as described above. Howevethere is also evidence that this physical quenching procmay compete with a C–H bond insertion reaction to MgH1

1HCO products as discussed below.

3. C 2A9(2A2)]X 2A1 band

The C←X band correlates with thep* ←n band of bareformaldehyde. As discussed above, formaldehyde distorta pyramidal geometry in this transition with the CvO bondlength increasing from 1.21 to 1.32 Å. The CH2 wag andCvO stretch frequencies change dramatically from 11and 1746 cm21 in the ground state to 683 and 1182 cm21 inthe 1A9 excited state. In Mg1-formaldehyde this band appears as a broad continuum band that does, however, sseveral undulations separated by;600 cm21. These undula-tions are probably related to the CvO stretch mode and thCH2 out-of-plane wagging modes of aCs pyramidal com-plex as discussed above, but given the broad nature ofabsorption band, it is not possible to make a definite vibtional assignment.

4. D 2A1]X 2A1 band

An expanded view of theD←X band is shown in Fig. 5Unfortunately, the vibrational resonance peaks are broadno obvious rotational substructure is observed. There isobvious three-membered progression with a vibrational sping of ;350 cm21 built on an origin at 40 473 cm21. Thisprogression is assigned to then3 Mg–O stretch mode (a1) inD 2A1 with a calculated frequency of 356 cm21. Ab initioresults show that electronic excitation in this band leadssignificant decrease in the Mg–O bond length as a paMg–O s bond forms, so that the Mg–O stretch mode shobe active. This is consistent with the experimental spectrobserved here. Birge–Sponer analysis of this progresgives the fundamental frequency asv3536064 cm21 witha large anharmonicity constant,vexe51162 cm21. TheMg–O stretch progression is short and highly anharmoindicating a very weakly boundD 2A1 excited state. Birge–

FIG. 5. Expanded view of theD 2A1←X 2A1 band. All transitions are assumed to originate from the vibrational ground state.


l

ss

to

7

ow

he-

ndn

c-

aaldmon

c,

Sponer analysis gives a Mg–O bond dissociation energyD 2A1 of D0(Mg–O)50.36 eV in the Morse potential approximation. Again using the Mg1 3p←3s excitation en-ergy of 4.42 eV and the 00

0 origin at 40 473 cm21, we find aground-state dissociation energy ofD09(Mg–O)50.96 eV.Again, zero-point energy corrections based on theab initiofrequencies are relatively small and lead to a ground-sbond dissociation energy ofDe9(Mg–O)51.00 eV. This canbe compared with the value obtained above from the BirgSponer analysis of theB state ofDe951.29 eV, and the the-oretical bond energy ofDe951.35 eV at the MP2 level. It issurprising that the experimental bond from the analysisthe D state spectrum is so much lower than the theoretbond energy, which itself is likely to be an underestimatethe true bond energy. We do not have a good explanationthis discrepancy.

There are also two short progressions in then3 intermo-lecular stretch built on origins at 00

0185 cm21 and 000

1234 cm21. These are assigned to progressions withquantum ofn1 Mg–O–Cintermolecular in-plane bend (b2)and n2 Mg–O–C out-of-plane bend (b1), respectively. Atmuch higher energy we observe a strong peak at 42cm21 (00

011651 cm21). This peak is assigned to then7 in-tramolecular CvO stretch mode (a1) and is expected sincethe formation of a partial Mg—O bond will cause the CvObond to weaken and stretch. Note that we see no evidencany CH2 rocking or wagging motions that might be expectif the complex were to significantly distort out of planageometry.

The dissociation mechanism from this state is not obous. There is not enough energy to dissociate directly toMg1* (3p)1H2CO asymptote. Predissociation to the lowelying H2CO* (1A9)1Mg1 asymptote is energetically possible but requires distortion out ofC2v geometry since~inC2v) the upper state is2A1 and the lower state is2A2 . Itmight also be possible to couple through a vibronic intertion to the lower-lying states of2B2 or 2B1 symmetry, ordirectly to the ground-state surface since both are2A1 . How-ever, the details of this coupling are not obvious sincestates are all well separated in energy at theD-state equilib-rium geometry. Of course at these high excitation energresonance-enhanced multiphoton dissociation through hlying electronic states cannot be entirely ruled out.

C. Reaction dynamics and product branching

Mg1 is the major dissociation product through the uvvisible region. However, in theB←X, C←X, and D←Xabsorption bands, we also observe a substantial branchinthe reactive dissociation product MgH1. There are no otherdissociation products evident. In Fig. 6 we show the phodissociation product mass spectra at selected wavelenMgH1 is not observed in the lowest-energyA←X band asexpected based on the energetics. The@MgH1#/@Mg1#branching ratio is plotted as a function of photon energyFig. 7.

MgH1 is thus observed in two metal-centered and informaldehyde-centered excitation bands. MgH1 shows thesame resonance structure in the action spectrum as the1


vrsn

-u

a

nn

ho

ioes

noity.the–Hontatege-re-

tic

ont ating

c

-the

tiontions

ing

ofughs

-n.nce-ase

ion

to-

the

nedly

ar-orin

re-tan-

rblehs


nonreactive product. There is no significant difference edent in the reactive to nonreactive dissociative branchingtio due to selective excitation of different vibrational modeHowever, the branching ratio obviously increases monotocally as a function of photon energy through the nearregion. The@MgH1#/@Mg1# ratio is about 8–9% in theB←X band and increases to 10–25% in theC←X band. In thehighest-energyD←X band this ratio is close to 50% i.e.,33% total reactive yield!. Thus, in contrast to theMg1-acetaldehyde where reactive channels contribute;5%of the overall dissociation products, in Mg1-formaldehydewe see here a large branching to the reactive cha(MgH1).

Using the available bond energies the energetic thresfor the reaction

Mg11H2CO→MgH11HCO,

is E51.97 eV. Assuming a complex binding energy ofDe951.35 eV, the spectroscopic threshold for the reactshould be near 3.32 eV~;26 800 cm21!. Because we see thMgH1 reaction product in theB state at excitation energie

FIG. 6. Product mass spectra at selected wavelengths:~a! 425.44 nm;~b!334.40 nm;~c! 310 nm;~d! 247.08 nm.

FIG. 7. @MgH1#/@Mg1# product branching ratios at selected wavelengt


i-a-.i-v

el

ld

n

;29 000 cm21, it appears that reaction can proceed withappreciable activation barrier above the endothermicHowever, in the Franck–Condon excitation geometrymetal ion is nearly end-on to the O–C bond and before Cbond activation, Mg1 has to migrate to become centeredone of the C–H bonds. Depending on the electronic sthere will be an energy cost associated with this rearranment to the transition-state geometry and the increasingaction yield with photon energy probably reflects this kinebarrier to reaction.

In the D state, with an energy well above the reactithreshold, the reactive quantum yield is roughly constan;1

3. We hypothesize that reaction might occur by bendin-plane to a transition-state geometry with Mg1 centered onthe C–H bond.@This could be accomplished by vibronicoupling to the lower-lyingB 2A8(2B2) state since both areA8 in Cs geometry.# This could then be followed by a nonadiabatic bond-stretch insertion process facilitated byoverlap of the Mg1 p orbital with the C–Hs* antibondingorbital in the bent transition-state geometry. Such a reacmechanism has been used previously to explain the reacof light metal ions in theirp states with methyl C–Hsbonds.6 In this proposed mechanism nonreactive quenchcould result from a frustrated insertion reaction.

It is also useful to note that direct photodissociationbare formaldehyde is possible in the near-uv region throthree separate channels~with thermodynamic thresholds ashown!:28

H2CO1hn→H1HCO ~l<330 nm!

→H21CO ~l<361 nm!

→H1H1CO ~l<283 nm!.

It is possible that the MgH1 observed in the formaldehydebasedC←X band could result from direct photodissociatioof the formaldehyde moiety, with Mg1 acting as a spectatorHowever, this seems unlikely: we do not see any evidefor Mg1(HCO), or Mg1(CO) products that might be expected as significant products from this process. In any cthe MgH1 product observed in the metal-based excitatbands must certainly result from an intracluster reaction.

We have previously observed and investigated phochemistry in Mg1-acetaldehyde bimolecular complexes.9 Inthat case reaction was observed only for excitation inblueshifted 3ps(A8)←3ss(A8) band ~corresponding hereto D←X), suggesting as-like electronic orbital alignmentpreference for chemical quenching. However, we cautiothat this apparent orbital alignment selectivity might simpbe an energy effect: the 3ps(A8) band lies higher in energyand may be above any kinetic barrier associated with rerangement. The results we report here fMg1-formaldehyde bear this out. A reaction is observedall of the electronically excited states that lie above theaction threshold, with a reaction yield that increases substially with energy above threshold. In the Mg1-acetaldehydeexperiment we observed reaction products (MgH1,MgCHO1, and MgCH3

1) that correspond to breaking eithethe aldehydic C–H or C–C bonds with roughly compara.


anfrefo

an

uctesf

f-dhinin

y

nd

rs

e

oig

t-th

a-a–

-tae

isarne

re

-

ra

ner-

ob-

atestIS

p-nter-ionneren-

cterndp-

in

isin-

ionemhe

s aof

–

g to

ab-

innifi-ity.blyar-

datahis

alto


yields. This result showed that there were no significsteric hindrance effects on Mg1 attack of the C–C bond oacetaldehyde. Our results here are consistent with the pous work but show that the reaction yield is much higherthe Mg1-formaldehyde case.

V. CONCLUSIONS

We have carried out studies of the spectroscopychemical dynamics of Mg1-formaldehyde in the visible andnear-uv regions. The work is supported by electronic strture calculations of the ground and low-lying excited staof the complex. Probing the electronic structure providestringent test ofab initio predictions. Photodissociation othe weakly bound Mg1•OCH2 complex represents the halcollision analog of the quenching of orbitally aligneMg1* (3p) in a near-O-end-on, low-energy collision witformaldehyde and gives valuable insight into the orbitalteractions and nonadiabatic coupling effects that determthe chemical dynamics.

The complex is electrostatically bound inC2v end-onMg1

uOvCH2 geometry, with a bond dissociation energof De951.35 eV at the MP2/6-31111g(2d,2p) level.Mg1(H2CO) shows absorption bands in the visible anear-uv regions that correlate with Mg1-based andformaldehyde-based radiative transitions. The lowest-eneband is assigned asA 2A8(2B1)←X 2A1 band and correlatewith a transition that is predominantly Mg1-based3pp(2B1)←3ss(2A1). However, the complex vibronicspectrum shows very significant excitation of the CH2 out-of-plane wag and CvO stretch modes, implying that thexcited state is mixed with appreciable CvO p* antibond-ing orbital character. The CH2 wag and CvO vibrationalfrequencies in the excited state are shifted dramatically frtheir values in the ground state, showing the effect of a snificant weakening of the CvO bond, accompanied by ouof-plane distortion. The observed homogeneous linewidgive a predissociation lifetime corresponding to,7 vibra-tions in the excited state.

In contrast, excitation in the Mg1-based B 2A8(2B2)←X 2A1 band shows predominantly low-frequency vibrtional motions assigned to the intermolecular in-plane wand Mg–O stretch. Birge–Sponer analysis gives the Mgdissociation energies for the ground andB 2A8 excited statesas D0951.25 eV andD0952.05 eV, respectively. With a correction for the zero-point shift the measured ground-sbond energyDe951.29 eV is in very good agreement with ththeoretical value. ResolvedK-subband rotational structureevident in the vibrational resonances of this band and cledemonstrates a geometry change from a linear or near-liMg–O–C (C2v) ground state to a bent (Cs) excited state@u~Mg–O–C!5139°#. Spectroscopic rotational constants ain very good agreement withab initio predictions in thisband.

The D 2A1←X 2A1 band, which correlates with the predominantly Mg1-based 3ps(2A1)←3ss(A1) transition,also exhibits pronounced vibrational structure. Strong vibtional activity is observed in the Mg—O and CvO stretch


t

vi-r

d

-sa

-e

gy

m-

s

gO

te

lyar

-

modes. Birge–Sponer analysis gives the Mg–O bond egies for the ground andD 2A1 excited states asD0950.96 eV andD0850.36 eV, respectively.

These spectroscopic results are very similar to thosetained previously for the more complex Mg1-acetaldehydesystem. However, the Mg1-formaldehyde spectrum showsmore well-resolved resonance structure, allowing a betterof ab initio predictions. Overall our results show that the Cmethod works quite well for the predominantly Mg1-basedtransitions, but fails badly for the transitions that show apreciable formaldehyde-based excited-state character. Iestingly, the experimental values for the bond dissociatenergy of the ground state obtained from Birge–Spoanalysis are somewhat lower than the theoretical bondergy at both the HF and the MP2 levels~although the MP2bond energy is in better agreement!.

Mg1-aldehyde spectroscopy shows a distinct charathat is different from the spectroscopy of other weakly bouMg1-molecule complexes. Typically for complexes that suport an M 1 – O electrostatic bond@such as Mg1(H2O),Mg1(CO2), or Mg1(CH3OH)],13–15 the vibronic spectrumis dominated by a long intermolecular stretch progressionthe attractive Mg1-based excited states of 3pp-orbital char-acter. In Mg1-aldehyde complexes the vibronic spectrumdominated by considerable excitation of the high-energytramolecular aldehyde vibrational modes@in 3pp(b1) sym-metry#, or a strong intermolecular in-plane bend progress@in 3pp(b2) symmetry#. On the other hand, excitation of thstates of 3ps-orbital character typically leads to a continuuband resulting from direct dissociation of the complex to tMg1* (3p) excited state. However, in Mg1-aldehyde com-plexes this excitation leads to a bound state that showsignificant vibrational structure consistent with formationa partial Mg–Os-bond.

Mg1 is the major dissociation product through the uvvisible region. However, in theB←X, C←X, and D←Xabsorption bands, we also observe a substantial branchinthe reactive dissociation product MgH1. The reactivebranching ratio increases with photon energy through thesorption bands, reaching a reactive quantum yield of;1

3 inthe D←X band. The total reactive quantum yieldMg1-formaldehyde is significantly higher than iMg1-acetaldehyde. Our results imply that there is no signcant activation barrier to reaction above the endothermicThe increasing reaction yield with photon energy probareflects a kinetic barrier to reaction associated with rerangement into the transition-state geometry with Mg1 cen-tered on the aldehyde C–H bond. We hope that thesewill stimulate more accurate theoretical calculations of tinteresting prototype chemical system.

ACKNOWLEDGMENTS

We gratefully acknowledge the support of the NationScience Foundation for this work. We are also gratefulProfessor Jan Jensen for helpful discussions.

1K. Eller and H. Schwarz, Chem. Rev.91, 1121~1991!.2P. B. Armentrout, Acc. Chem. Res.28, 430 ~1995!; P. B. Armentrout, in


iu,r,.

, J

n,..

n,

ys

s.

at,

, J.,

G.


Selective Hydrocarbon Activation: Principles and Progress, edited by J.A. Davies, P. L. Watson, J. F. Liebman, and A. Greenberg~VCH, NewYork, 1990!.

3J. C. Weisshaar, Acc. Chem. Res.26, 213 ~1993!; J. C. Weisshaar, Adv.Chem. Phys.82, 213 ~1992!.

4P. A. M. Van Koppen, R. R. Kemper, and M. T. Bowers, inOrganometallicIon Chemistry, edited by B. S. Freiser~Kluwer Academic, Boston, 1996!.

5B. S. Freiser, Acc. Chem. Res.27, 353 ~1994!; B. S. Freiser, J. MassSpectrom.31, 703 ~1996!.

6Y. C. Cheng, J. Chen, L. Ding, T. H. Wong, P. D. Kleiber, and D. K. LJ. Chem. Phys.104, 6452~1996!; J. Chen, Y. C. Cheng, and P. D. Kleibeibid. 106, 3884~1997!; W.-Y. Lu, T.-H. Wong, and P. D. Kleiber, ChemPhys. Lett.347, 183 ~2001!.

7J. Chen, T. H. Wong, K. Montgomery, Y. C. Cheng, and P. D. KleiberChem. Phys.108, 2285 ~1998!; J. H. Holmes, P. D. Kleiber, D. A. Ols-gaard, and K. H. Yang,ibid. 112, 6583~2000!; W.-Y. Lu, P. D. Kleiber, M.A. Young, and K.-H. Yang,ibid. 115, 5823~2001!.

8W.-Y. Lu and P. D. Kleiber, J. Chem. Phys.114, 10288~2001!.9W.-Y. Lu and P. D. Kleiber, Chem. Phys. Lett.338, 291 ~2001!.

10W.-Y. Lu, M. Acar, and P. D. Kleiber, J. Chem. Phys.116, 4847~2002!.11P. D. Kleiber and J. Chen, Int. Rev. Phys. Chem.17, 1 ~1998!; P. D.

Kleiber, Adv. Met. Semicond. Clusters5, 267 ~2001!.12J. S. Pilgrim, C. S. Yeh, K. F. Willey, D. L. Robbins, and M. A. Dunca

Int. Rev. Phys. Chem.13, 231 ~1994!; M. A. Duncan, Annu. Rev. PhysChem.48, 69 ~1997!; C. S. Yeh, K. J. Willey, D. L. Robbins, and M. ADuncan, Int. J. Mass Spectrom. Ion Processes131, 307 ~1994!; M. A.Duncan,ibid. 200, 545 ~2000!.

13K. F. Willey, C. S. Yeh, D. L. Robbins, J. S. Pilgrim, and M. A. DuncaJ. Chem. Phys.97, 8886~1992!.

14C. S. Yeh, K. F. Willey, D. L. Robbins, and M. A. Duncan, J. Chem. Ph98, 1867~1993!.


.

.

15M. R. France, S. H. Pullins, and M. A. Duncan, Chem. Phys.239, 447~1998!.

16J. M. Farrar, inCluster Ions, edited by C. Y. Ng, T. Baer, and I. Powi~Wiley, New York, 1993!, p. 243; D. C. Sperry, A. J. Midey, J. I. Lee, JQian, and J. M. Farrar, J. Chem. Phys.111, 8469~1999!; J. I. Lee, D. C.Sperry, and J. M. Farrar,ibid. 114, 6180~2001!.

17D. E. Lessen, R. L. Asher, and P. J. Brucat, inAdvances in Metal andSemiconductor Clusters, edited by M. A. Duncan~JAI, Greenwich, CT,1993!, Vol. I, p. 267; T. Hayes, T. Bellert, T. Buthelezi, and P. J. BrucChem. Phys. Lett.264, 220 ~1997!.

18K. Fuke, K. Hashimoto, and S. Iwata, Adv. Chem. Phys.110, 431 ~1999!;K. Fuke, K. Hasimoto, and R. Takasu, Adv. Met. Semicond. Clusters5, 1~2001!.

19L. A. Posey, Adv. Met. Semicond. Clusters5, 145 ~2001!.20X. Yang, Y. Hu, and S. Yang, J. Phys. Chem. A104, 8496~2000!.21F. Aguirre, J. Husband, C. J. Thompson, K. L. Stringer, and R. B. Metz

Chem. Phys.116, 4071~2002!; K. P. Faherty, C. J. Thompson, F. AguirreJ. Michne, and R. B. Metz, J. Phys. Chem. A105, 10054~2001!.

22C. W. Bauschlicher, Jr. and H. Partridge, J. Phys. Chem.95, 3946~1991!;H. Partridge and C. W. Bauschlicher, Jr.,ibid. 96, 8827~1992!.

23M. J. Frisch, G. W. Trucks, H. B. Schlegelet al., GAUSSIAN 98, RevisionA.6 ~Gaussian, Pittsburgh PA, 1998!.

24G. H. Dieke and G. B. Kistiakowsky, Phys. Rev.45, 4 ~1934!.25D. C. Moule and A. D. Walsh, Chem. Rev.75, 67 ~1975!.26C. A. Cantrell, J. A. Davidson, A. H. McDaniel, R. E. Shetter, and J.

Calvert, J. Phys. Chem.94, 3902~1990!.27R. Meller and G. K. Moortgat, J. Geophys. Res.105, 7089~2000!.28G. D. Smith, L. T. Molina, and M. J. Molina, J. Phys. Chem. A106, 1233

~2002!.29J. T. Hougen and J. K. G. Watson, Can. J. Phys.43, 298 ~1965!.30S. S. Ju, P. Y. Cheng, and H. L. Dai, J. Chem. Phys.103, 2850~1995!.


photodissociation spectroscopy and dynamics of mg+ ...wlu/jcp6970.pdfsponer analysis gives the...

Documents