Charge Transfer Processes in Surface-Enhanced Raman Scattering. Franck-Condon Active Vibrations of Pyridine

Transcription

1 9254 J. Phys. Chem. 1996, 100, Chrge Trnsfer Processes in Surfce-Enhnced Rmn Scttering. Frnck-Condon Active Vibrtions of Pyridine Jun Frncisco Arens, Isbel López Tocón, Jun Crlos Otero,* nd Jun Igncio Mrcos Deprtment of Physicl Chemistry, Fculty of Sciences, UniVersity of Málg, E Málg, Spin ReceiVed: December 9, 1995; In Finl Form: Mrch 1, 1996 X The surfce-enhnced Rmn scttering (SERS) spectr of pyridine hve been nlyzed on the bsis of resonnt chrge trnsfer (CT) mechnism. The most intense bnds recorded in the SERS t negtive electrode potentils correspond with the norml modes tht connect the equilibrium geometries of the neutrl molecule nd the corresponding rdicl nion. These norml modes hve been chrcterized ccording to our previously proposed method by clculting the b initio geometries of the molecule nd its nion nd then building the trnsformtion ) L -1 R, where the vector contins those vibrtions ctive in SERS-CT vi Frnck- Condon fctors, thereby providing semiquntittive selection rules for this enhncement mechnism. Introduction This work is the second in series where SERS (surfceenhnced Rmn scttering) results re systemticlly nlyzed ssuming resonnt chrge trnsfer (CT) mechnism. 1 One of the min problems in SERS is the bsence of simple rules tht llow one to recognize if the observed chnges in the reltive intensities of the bnds rise either from the presence of chrge trnsfer stte ccessible to the system under the experimentl conditions or from surfce orienttion effects (electromgnetic mechnism, EM). Moreover, contrry to the reltive simplicity of the selection rules bsed on the EM model, the CT selection rules re equivlent to those of resonnce Rmn (RR) nd re not generl, depending on the prticulr electronic structure of the metl-dsorbte system. 2-4 There re two min differences tht considerbly increse the difficulty of the nlysis of SERS spectr compred with those of resonnce Rmn. First, the presence of resonnt mechnism in Rmn cn esily be predicted from the UVvis spectrum of molecule: if the energy of the incident photon flls in the rnge of n bsorption bnd, the spectr obtined will show RR fetures. However, in the cse of SERS, these bnds re difficult to observe in the UV-vis spectrum. For instnce, in the SERS experiments in colloids the ggregtion of the prticles gives rise to new metl bsorptions overlpping with the possible CT bnds. In spite of this, this type of trnsition hs been detected for the Ag-pyridine system by vis spectroscopy 5 or electron impct techniques 6,7 in gs-solid interfces. On the other hnd, the electronic excited sttes in SERS hve chrge trnsfer chrcter nd re much less wellknown thn the excited singlets relevnt in RR. The method previously described by ourselves to nlyze SERS-CT spectr is bsed on the hypothesis tht complete electron is trnsferred from the metl to the dsorbte, in which cse the CT stte is nlogous to the rdicl nion from the moleculr point of view. 1,3 Under these conditions the process is similr to n electron energy loss experiment (EELS) under resonnce conditions (shpe resonnces). 3,8-11 This pproximtion llows the prediction of selective enhncement in SERS- CT of those norml modes relted to the differences between the equilibrium structures of the neutrl molecule nd the rdicl nion. These structures re strightforwrdly obtined from the geometry optimiztion results by employing b initio methods X Abstrct published in AdVnce ACS Abstrcts, My 1, S (95) CCC: $12.00 tht re used s nother tool in the nlysis of the spectr. This selection rule for the SERS-CT mechnism is nlogous to tht proposed by Tsuboi 12 for the A-term (Frnck-Condon fctors 13 ) in resonnce Rmn. If the enhnced vibrtions in SERS correlte the equilibrium structures of the neutrl molecule nd the rdicl nion, we will be in the position of confirming the existence of chrge trnsfer mechnism nd obtining further insight into this process. This method of nlysis hs been pplied to the SERS spectr of pyrzine on silver electrode tht show significnt enhncement of the 8, 9, nd 6 modes with respect to those intensities recorded in conventionl Rmn. This selective enhncement hs previously been nlyzed on the bsis of the EM mechnism, proposing n end-on orienttion for the molecule with respect to the surfce. 14,15 The results of our nlysis were clerly comptible with the chrge trnsfer process rther thn the previous conclusions bsed on the orienttion of the dsorbte. In order to check the vlidity of our results, we hve pplied the sme systemtic method to three RR spectr lredy published, 16 nd the observed differences between ll of them hve been stisfctorily explined. 1 The SERS of pyridine obtined t negtive electrode potentils is one of the few cses where it hs been possible to stisfctorily explin the spectrl fetures vi the CT mechnism. These results show reltive enhncement of the sme fundmentls s in pyrzine, nmely the 8, 9, nd 6 modes. The min disdvntge of the nlysis crried out by Creighton 17 is tht it is bsed on the symmetry properties of benzene, mking it impossible to explin ny difference between the SERS of pyridine or pyrzine, given tht the selection rules should be identicl in both cses. This is similr sitution to tht of the rules bsed on symmetry considertions for EELS under resonnce conditions. 11,18 In contrst, our nlysis gives semiquntittive selection rules tht should tke into ccount the different behvior in the SERS of molecules with similr electronic structures such s pyridine nd pyrzine. Given tht in pyridine generl greement exists tht the SERS fetures obtined t negtive potentils re due to CT mechnism, the im of this work is to contrst the vlidity of the nlysis process, previously employed by ourselves for the first time in the cse of pyrzine. 1 Experimentl Section The chemicl regents used cme from Aldrich nd were of the highest vilble purity. Wter ws first deionized, triply 1996 Americn Chemicl Society

2 Frnck-Condon Active Vibrtions of Pyridine J. Phys. Chem., Vol. 100, No. 22, distilled, nd then degssified before being used in recording the SERS. Pyridine ws purified t reduced pressure in Büchi GKR-51 distilltor. Rmn spectr were recorded in the cm -1 rnge with Jobin-Yvon U-1000 spectrometer fitted with cooled Hmmtsu PMT photomultiplier, using the nm exciting line from Spectr Physics 2020 Ar + lser with n effective power of 30 mw reching the SERS smple. These spectr were recorded on pure silver electrode (Metles Preciosos S.A.) tht hd been previously polished with 1.00, 0.30, nd 0.05 µm lumin (Buehler). The working electrode ws mounted in n electrochemicl cell with pltinum counter electrode nd sturted Ag/AgCl/KCl reference electrode (Princeton Applied Reserch, PAR), to which ll electrode potentils re referenced. A PAR model 173 potentiostt nd PAR model 175 progrmmer were used to control the electrode potentils. The electrode surfce ws electrochemiclly roughened in 1 M queous solution of KCl by initilly mintining the surfce potentil t V nd then subjecting it to 10 2 s pulses t V. Finlly, the SERS hve been obtined from n queous solution of 0.1 M pyridine nd 1.0 M KCl t severl surfce potentils. Clcultions Geometry Optimiztions. All of the geometry optimiztions hve been crried out by using the GAUSSIAN 90 pckge of progrms 19 t diverse theoreticl levels. For the ground stte (S 0 ; 1 A 1 ), clcultions were crried out employing nd 6-31+G* bsis sets. The rdicl nion is doublet in its electronic ground stte nd s such will be nmed the D 0 stte ( 2 B 1 ). The theoreticl description of this type of chemicl species presents two problems. In the first plce it is generlly ccepted tht diffuse functions re required in order to dequtely represent the excess of negtive chrge. This condition involves considerble increse in the clcultion time tht could mke deterrent the use of b initio clcultions s routine tool in nlyzing the SERS. In this respect, it hs been suggested tht the use of diffuse functions is not prticulrly importnt in deloclized systems like the ones discussed here. 20 On the other hnd, the results for open shell clcultions lwys show some spin contmintion. This fct cn be voided by crrying out ROHF or UNO-CAS clcultions since these methods hve lredy proved to be dequte in the study of rdicls. 21,22 In our previous work on pyrzine, complete comprtive study ws crried out between ll of these theoreticl levels employing from to G* bsis sets in order to determine the minimum theoreticl level cpble of stisfctorily explining the SERS results. The min conclusion of tht study ws tht the optimized geometries from nd clcultions with the bsis set re sufficiently ccurte to ccount for the min fetures of SERS or RR spectr. 1,23 The differences mongst the diverse theoreticl levels were found to be very smll compred to those of the SERS spectr for the sme molecule obtined by different uthors under comprble experimentl conditions. For the pyridine rdicl nion, clcultions hve been crried out using nd 6-31+G* bsis sets, the spin contmintion found in these results being moderte (on the order of ). In spite of the forementioned fcts, ROHF/ nd 3 3 UNO-CAS/ clcultions hve lso been crried out. For this ltter method, ll nturl orbitls with occuption numbers between 1.98 nd 0.02 hve been included in the ctive spce ccording to the criteri of Puly et l. 24 For comprison, the geometry of the first singlet excited stte (S 1 ) of pyridine of 1 B 1 symmetry hs lso been optimized by mens of CIS/ clcultions. Figure 1. () In-plne internl coordintes nd (b) pictoril representtion of the LUMO (B 1) of pyridine. TABLE 1: In-Plne Internl Coordintes nd Refined Scle Fctors chrcter internl coordinte scle fctor ν(c-h) R 1-5 ) R i, i ) ν ring R 6-11 ) r i, i ) δ(c-h) R ) 2-1/2 (β i - F i), i ) δ ring R 17 ) 6-1/2 (R 1 - R 2 + R 3 - R 4 + R 5 - R 6) 0.77 R 18 ) 12-1/2 (2R 1 - R 2 - R 3 + 2R 4 - R 5 - R 6) R 19 ) 4-1/2 (-R 2 + R 3 - R 5 + R 6) ν ring coupling in ortho, met, nd pr positions 0.69 Method of Anlysis. In our previous work bout pyrzine, 1 the usefulness of the following trnsformtion ws demonstrted ) L -1 R (1) insted of compring the equilibrium geometries of the two sttes directly. is vector tht vibrtionlly chrcterizes the chromophore nd represents the displcement of the potentil energy minimums of the two sttes involved in the resonnce process long ech norml coordinte. R contins the differences between the geometric prmeters of the excited sttes (D 0 in SERS-CT or S i in RR nd vibronic spectr) nd the ground stte (S 0 ) nd will be obtined from the results of the optimized geometries. In this wy, represents semiquntittive selection rule for Frnck-Condon ctive vibrtions in either SERS-CT or in electronic spectr involving D 0 or S i sttes, respectively. As in the cse of pyrzine, the vector will contin exclusively totlly symmetric vibrtions since ll of the here studied equilibrium structures belong to the C 2V point group. These mplitudes cn be used fterwrd to deduce Frnck-Condon ctive vibrtions in bsorption 25 or to clculte reltive intensities in SERS-CT or RR spectr through complex 13 or pproximte mthemticl expressions like tht proposed by Peticols et l. 26 mong others I clcd,i ) K i 2 ω i 2 In this eqution ω i is the frequency of the i norml mode nd K is simply constnt defined in such wy tht the reltive intensities of the bnds re referred to one of them which hs been rbitrrily ssigned n intensity of 100. Force Field. The L mtrix of the ground electronic stte needs to be known to crry out the trnsformtion (eq 1) for which the b initio force field t / level hs been clculted. Subsequently process nlogous to tht described elsewhere hs been followed. 1,27 Firstly, the internl coordintes hve been defined s shown in Figure 1, nd then liner combintions hve been defined ccording to the recommendtions of Puly et l. 28 (Tble 1). The b initio F mtrix in Crtesin coordintes hs been trnsformed into independent internl coordintes by using the progrm FLINT, 29 nd (2)

3 9256 J. Phys. Chem., Vol. 100, No. 22, 1996 Arens et l. TABLE 2: Frequencies (cm -1 ) of A 1 Fundmentls of Pyridine mode ν obs ν scled PED (%) 2;ν(CH) ν(ch) 20;ν(CH) ν(ch) 13;ν(CH) ν(ch) 8;ν ring ν ring;26 δ(ch) 19;δ(CH) δ(ch);28 ν ring 9;δ(CH) δ(ch);28 ν ring 18;ν ring ν ring;32 δ ring (R 17), 24 δ(ch) 12;δ ring δ ring (R 17);52 ν ring 1;ν ring ν ring;35 δ ring (R 17) 6;δ ring δ ring (R 18) Observed frequencies from the pure liquid spectrum. fterwrd modified version of the progrm QCPE# ws used to crry out the norml coordinte nlysis. This version lso permits the ppliction of the scled quntum mechnicl methodology of Puly et l. 31 in order to correct for the systemtic errors of the b initio methods. The scling process ws crried out by djusting the 19 in-plne frequencies of the pure liquid of the norml isotopomer. It ws necessry to optimize four scle fctors tht ffect the digonl elements of the F mtrix nd n extr fctor for the ν ring interctions in ortho, met nd pr positions s is the cse with benzene 27,31 or pyrzine. 1 Results nd Discussion Vibrtionl Spectr of Pyridine in the Electronic Ground Stte. The vibrtionl spectr of pyridine hve been ssigned empiriclly by Stidhm nd DiLell 32 nd lso hve been the subject of force field clcultions with similr results to those reported here by Fogrsi et l. 33 mong others. The fitted scle fctors cn be seen in Tble 1, nd the observed nd scled frequencies s well s the PED of the A 1 fundmentls of pyridine cn be found in Tble 2. The 19 nd 18 norml modes (ccording to Wilson s nomenclture 34 ) possess predominnt chrcter of δ(ch) nd ν ring, respectively, the opposite of the norml cse where it is considered tht the description of these two fundmentls is just the reverse. 34 It is to be stressed tht the 9;δ(CH) fundmentl shows considerble ring-stretching contribution. On the other hnd, the R 17 coordinte which resembles the norml mode 12;δ ring contributes 35% to the PED of the 1;ν ring vibrtion. The PED descriptions of the 8;ν ring, 9;δ(CH), nd 6;δ ring vibrtions nd to lesser extent tht of the 1;ν ring one re very similr to those found for pyrzine. 1 The reltive intensities of the Rmn spectr of the pure liquid nd the queous solution cn be seen in Tbles 3 nd 4, with the reltive intensities being normlized with respect to tht of the 1 vibrtion which hs been rbitrrily ssigned the vlue of 100. As cn be pprecited, the Rmn spectr of pyridine re chrcterized by the lrge intensities of the bnds ssigned to the 1 nd 12 modes. SERS Spectr of Pyridine. Tble 3 shows selection of diverse SERS results obtined by different uthors under diverse conditions. Although the detils of ech spectrum re different, it is possible to estblish rough clssifiction into three groups. The spectr obtined by Creighton et l. 35,36 in Au or Ag sols cn be included in the first group. These results show very similr generl fetures to those of the pure liquid or solution spectr given tht they re chrcterized by the lrge intensity of the bnds ssigned to the 1 nd 12 vibrtions. The second group contins the SERS spectr obtined on Ag electrode t -1.4 V 17 (vs SCE) or in Cu colloid employing the nm exciting line. 37 These spectr show significnt enhncement of the 8, 9, nd 6 modes similr to the spectrum obtined by Moskovits et l. in gs-solid interfce. 14 The spectr tht show these fetures will be subsequently referred to s SERS- CT. The SERS in Cu sol (752.5 nm) 37 or on Ag electrode t -0.4 V (vs SCE) 17 in Tble 3 constitutes n intermedite sitution between the two previous groups nd is cler exmple of how the resonnt condition in CT process is ffected by the position of the Fermi level vi the electrode potentils s well s by the energy of the incident photon. The SERS intensities obtined by ourselves on Ag electrode hve been summrized in Tble 4 nd show behvior similr to tht just mentioned. As the electrode potentil is mde more negtive, the chrcteristic bnds of SERS-CT become more intense (see Figure 2), lthough the excittion profiles vry from one fundmentl to nother. For instnce, the reltive intensities of the 9 nd 6 modes continuously increse while tht corresponding to the 8 mode reches mximum t or V (vs Ag/AgCl). The SERS-CT vlues in Tble 4 lso show detils tht differentite them from those of the Creighton ones. The intensity of the 9 mode reches quite n stonishing level in the spectr of Tble 4 (I rel ) 140 t -1.2 V), nd the 12 mode, with wek intensities in ll the SERS-CT spectr of Creighton, mintins n pprecible intensity in our results in greement with tht of Moskovits et l. 14 These discrepncies re typicl of this type of spectroscopy nd re due to the different nture nd morphology of the interfces which depend on the surfce-ctivting procedure nd determines different locl electric potentils. Therefore the selection rules will only be cpble of ccounting for the most outstnding fetures of these complex results. In the works of Creighton, the SERS-CT results re explined by extrpolting the Jhn-Teller effect for the rdicl nion of benzene. 3,17 This effect predicts deformtion of the moleculr structure of the rdicl nion with respect to those norml modes tht trnsform s the irreducible representtions contined in the symmetric product {E 2u E 2u } ) A 1g + E 2g, i.e., vi the 8, 9, 1, nd 6 modes, ν(ch) vibrtions being excluded. Despite tht this selection rule is strictly vlid only for the rdicl nions of both benzene or pyrzine where the relevnt symmetry is D 2h, the greement with the experimentl results is firly good. The geometric prmeters for the optimized structures of pyridine (S 0 stte) nd the differences with those of the rdicl nion ( (D 0 -S 0 )) employing diverse theoreticl levels cn be seen in Tble 5. As is the cse with pyrzine, 1 ll of the results predict deformtion quite similr to those found in benzene 21 given tht in ll of these molecules the LUMO hs shpe nlogous to tht shown in Figure 1b (LUMO: B 1 in C 2V ; B 3u in D 2h or E 2u in D 6h ). Generlly speking, the differences re lrger for the ring prmeters thn for those of the CH bonds given tht the chrge is minly deloclized in the romtic system. The vector R hs been obtined from the numericl differences (D 0 -S 0 ) by crrying out liner combintions of the geometric prmeters identicl to those shown in Tble 1. The vlues obtined from the trnsformtion (eq 1) re shown in Tble 6. Although ll of the fundmentls for pyridine possess nonzero vlues for, our procedure predicts correctly tht the 8, 9, 1, nd 6 fundmentls will be the most intense vi the SERS-CT mechnism, given tht they show vlues significntly lrger thn the remining ones. The vlues for these fundmentls with nd / clcultions re , 0.143, 0.222, nd 0.217, respectively, while those obtined with the 6-31+G* bsis set re , 0.124, 0.174, nd 0.179, respectively.

4 Frnck-Condon Active Vibrtions of Pyridine J. Phys. Chem., Vol. 100, No. 22, TABLE 3: Summry of the Reltive Intensities of A 1 Fundmentls in Rmn nd SERS Spectr of Pyridine Rmn SERS/sol SERS/electrode c mode liquid solution Au Ag Cu b Cu b Ag -0.4 V Ag -1.4 V As the reltive intensities depend on the reltive vlues of (eq 2), it is deduced tht the increse in the bsis set size does not significntly ffect the results of the nlysis. For the ske of comprison, these mplitudes hve been normlized with respect to tht of the 1 mode which hs been ssigned the vlue of 100 ( 100 included in Tble 4). The ν(ch) vibrtions show very smll vlues of in greement with the wek intensity of these bnds in the SERS-CT. The 12 mode exhibits mplitudes of or ( or 6-31+G*, respectively), tht would explin the noticeble intensity observed in our SERS-CT spectr. This vibrtion is the only one to show significnt differences between the vlues obtined by using the scled nd clculted L mtrices, nd 0.004, respectively. The remining A 1 fundmentls for pyridine show wek reltive intensities in greement with their respective smll vlues. for S 1 -S 0 Vibronic Trnsitions. The norml mode displcements in n electronic trnsition between the ground nd the first singlet excited stte of pyridine (S 1 -S 0 ; 1 B 1-1 A 1 ) re very different from those found in SERS-CT. These displcements hve been clculted from the optimized geometries of these sttes (Tble 5) nd re shown in Tble As cn be pprecited, the most intense bnds should correspond with those ssigned to the modes 9 (0.198), 18 (0.263), 12 (0.262), nd 6 (-0.436). The lrge vlues obtined for the 18 nd 12 vibrtions s well s tht of the 6 which mounts to lmost double tht clculted in SERS-CT re stressed. On the contrry, the 8 nd 1 vibrtions hve vlues tht re much less thn those obtined for SERS-CT ( nd SERS/gs-solid Ag b d < ? 1 <1 2? ? ? ? ref 35, 36 35, The intensities re mesured s the height of the bnds. The bnd intensity of the vibrtion 1 is set rbitrrily equl to 100. b Wvelength of exciting line (nm). c Electrode potentils mesured vs SCE nd nm exciting line. d From ref 14. TABLE 4: Reltive Intensities of A 1 Fundmentls Mesured in the SERS of Pyrzine nd Pyridine mode SERS (V) b 100 I c Rmn clcd solution G* 6-31+G* Pyrzine Pyridine <1 < Electrode potentils mesured vs Ag/AgCl sturted electrode. b Reltive vlues referred to tht of the 1 mode rbitrrily set equl to 100. c Reltive Rmn intensities from eq in this cse nd nd in SERS-CT, respectively, from results). These results gree quite well with the published dt for the vibronic spectr of pyridine. Both 6 nd 12 modes ply n essentil role in the bsorption 39 or the fluorescence 40 spectr, while 8 nd 1 vibrtions re not observed. Our nlysis procedure predicts clerly differentited behvior for S 1 -S 0 nd D 0 -S 0 trnsitions in spite of the fct tht in both cses the LUMO;B 1 of pyridine is involved. The configurtion with the lrgest contribution in the CIS clcultion corresponds to n electronic excittion from the moleculr orbitl 19 (A 1 ) to 22 (LUMO;B 1 ) with coefficient of Comprison with the SERS Results of Pyrzine. One of the conclusions of the previous work bout pyrzine ws tht the lrgest mplitudes cn be successfully ssocited with the strongest SERS-CT bnds, lthough numericl correltion is not simple when predictions for different norml modes re extrpolted. For exmple, in the SERS of pyrzine recorded t V the intensities of the 8, 1, nd 6 modes re 131, 100, nd 48, respectively, while the vlues re , , nd respectively ( nd /). The dependence of these intensities on the frequency (I clc from expression 2 included in Tble 4) somewht improves these predictions, lthough significnt differences still exist between the predicted nd the experimentl intensities, given tht the clculted intensities for these vibrtions re 201, 100, nd 92, respectively. For this reson the comprison with the resonnce Rmn results of pyrzine were very useful. In this work the previous results concerning the SERS of pyrzine will be used

5 9258 J. Phys. Chem., Vol. 100, No. 22, 1996 Arens et l. increse s the electrode potentil is mde more negtive, becuse there should be greter concentrtion of H + t the interfce. The observed behvior in the SERS of Figure 2 in ref 1 is just the opposite. Another fctor could be the interction between the 9 fundmentl nd n A g combintion bnd tht is recorded in the Rmn spectrum t 1233 nd 1208 cm -1, respectively. 43 A possible Fermi resonnce between these two vibrtions could be responsible for the observed dependence between the intensities of these two bnds in the SERS of pyrzine. 44 In whichever cse, it would pper cler tht there is n unexpected circumstnce tht could be the origin of the discrepncies between the SERS-CT intensities nd the vlues for this fundmentl in both molecules. The A-term in resonnce Rmn lso predicts reltive enhncement of those norml modes tht show significnt differences between their vibrtionl frequencies in the two involved electronic sttes. 13 The vibrtions tht show noticeble frequency shift exhibit simultneously lrge mode mixing in the Duschinski mtrix 45 D ) L -1 (D 0 ) L(S 0 ) (3) Figure 2. Rmn spectr of the queous solution nd SERS of pyridine on silver electrode. in the comprtive discussion tht follows nd hve lso been included in Tble 4. If we suppose tht the 8, 9, 1, nd 6 norml modes re similr in pyrzine nd pyridine, it should be possible to extrpolte their behvior in the SERS-CT of pyridine from the respective 100 vlues of both molecules. The similrity between these norml modes is supported by the respective frequencies, their PED descriptions, nd, moreover, by the contributions to the respective vlues. In Tble 7 the individul contributions of ech type of internl coordinte to the totl displcement cn be seen ( nd /). As cn be pprecited, the vlues of the 8, 9, nd 6 modes re smller for pyridine thn for pyrzine, while tht corresponding to the 1 vibrtion is greter. In spite of these discrepncies, the contributions of ech type of internl coordinte hve the sme sign nd similr height in both molecules for ll cses. The min contributions re due to the ring internl coordintes, including the 9 mode which hs predominnt contribution of δ(ch) in its PED. The 8 nd 6 modes of pyridine hve smller 100 vlues thn for pyrzine, in greement with the lower reltive intensities observed in the spectr. However, the sme cnnot be sid for the 9 vibrtion, given tht the experimentl behvior is just the opposite to tht expected from the vlues. Perhps the intensity of this bnd in pyrzine is ffected by the one recorded t bout 1200 cm The origin of this SERS bnd is not cler, hving been recently proposed tht it is due to the presence of the protonted pyrzine-h + species. 41 However this possibility seems to be unlikely given tht it is well-known tht in these interfces the ph is higher thn in the bulk of the solution. 42 Besides, it cn be expected tht the intensity of this bnd should The role of mode mixing s n lterntive or complementry origin of the displcements in the selective enhncement of the Ag vibrtions in the SERS-CT of pyrzine ws discussed in our previous work on this molecule. 1 It ws found tht this effect should hve little importnce in SERS results, given tht the lrgest D ij elements of these fundmentls re close to unity. In Tble 7, it cn lso be seen tht the results of pyridine re quite similr to those obtined for pyrzine. The D mtrix hs been obtined from the corresponding clculted L mtrices ( nd /) given tht the vibrtionl spectrum of the nion is not known, nd therefore it is not possible to predict the effect of the trnsfer of the scle fctors. The gretest difference is found for the 1 mode tht shows n pprecible mount of mode mixing with D ij vlues of nd in pyridine nd pyrzine, respectively. In order to obtin n estimtion of the frequency shift of these vibrtions in both sttes ( ν in Tble 7), the clculted nd / frequencies hve been corrected by n empiricl fctor of This procedure is routinely employed in the correction of the b initio frequencies in the ground electronic stte s well s in rdicl nions. 46 As cn be pprecited in Tble 7, the estimted frequencies in the S 0 stte reproduce the observed ones stisfctorily. The frequency shift between the two sttes hs little significnce, especilly if the pproximte chrcter of the wvenumbers is kept in mind or if one compres them with the shift of the 10;γ(CH),B 1g mode in the S 1 ; 1 B 3u stte of pyrzine (-540 cm -1 ) 39,40 which is responsible for the vibronic coupling with the S 2 ; 1 B 2u stte. This result is in greement with the respective D ij vlues close to unity. An importnt difference between the SERS results of pyridine nd pyrzine (Tble 4) is tht for the ltter molecule the fetures of SERS-CT cn be observed t much more positive potentils thn for pyridine. For instnce, t 0.00 V the 8 vibrtion lredy possesses reltive intensity of 72, while in the spectrum of pyridine recorded t V the 6 nd 9 modes show their mximum reltive intensities. These results gree with the reltive energies of the respective LUMO orbitls or the electron ffinities 10 nd with the energy differences between the D 0 nd S 0 sttes ( E in Tble 5) of both molecules, supporting the sme chrge trnsfer mechnism in both cses. The importnce of this results is tht it would llow to deduce the occurrence of the SERS-CT mechnism in n electrode experiment from the b initio energies of the rdicl nion nd the neutrl molecule. The E vlues indicte tht the rdicl

6 Frnck-Condon Active Vibrtions of Pyridine J. Phys. Chem., Vol. 100, No. 22, TABLE 5: Optimized Equilibrium Structures of the Ground Stte (S 0 ) of Pyridine nd their Differences with tht of the Rdicl Anion (D 0 ) nd the Lowest Singlet Excited Stte (S 1 ) S G* ROHF (D 0-S 0) b CAS c 6-31+G* (S 1-S 0) b CIS Bond Lengths (Å) nd Angles (deg) r 1, r 2, r 3, R 1, R 2, R R R 2, R 3, R β 1, F 1, β 2, F 2, β 3, F δch: (β-f) 1,5 δch: (β-f) 2,4 δch: (β-f) E (ev) b,d pyridine pyrzine See Figure 1 nd Tble 1 for symbols. b Differences hve been mde from clcultions employing the sme bsis set. c 3 3 UNO-CAS. d Differences between the energies of the equilibrium structures of D 0 or S 1 sttes nd S 0. TABLE 6: Norml Mode Displcement () between the Equilibrium Structures of the Ground Stte (S 0 ) nd Tht of the Anion (D 0 ) nd the Lowest Singlet Excited Stte of Pyridine (S 1 ) L -1 (S 0) mode stte S 0: stte D 0: ν clcd scled ROHF scled CAS scled 6-31+G* scled S 0: S1: CIS scled 2;ν(CH) ;ν(CH) ;ν(CH) ;ν ring ;δ(CH) ;δ(CH) ;δ(CH) ;δ ring ;ν ring ;δ ring UNO-CAS. TABLE 7: Individul Contributions to Displcements nd Vibrtionl Frequencies in the S 0 nd D 0 Sttes of Pyrzine nd Pyridine internl coordinte contributions to ν (S 0) molecule mode ν(ch) ν ring δ(ch) R 17 R 18 totl obsvd clcd ν (D 0) clcd ν b D ij c pyrzine pyridine pyrzine pyridine pyrzine pyridine pyrzine pyridine pyridine nd / clculted frequencies of S 0 nd D 0 sttes corrected by numericl fctor of b Differences between the corrected frequencies of S 0 nd D 0 sttes. c Lrgest elements of the Duschinski mixing mtrix. nion is c ev less stble in pyridine thn in pyrzine. This result grees pproximtely with the differences between the electrode potentils of the SERS-CT spectr of both molecules in Tble 4.

7 9260 J. Phys. Chem., Vol. 100, No. 22, 1996 Arens et l. Except for when the energy of the CT stte is experimentlly vilble, 5-7 this question hs lwys been quite difficult to resolve given tht the effect of dsorption on the work function of the metl nd on the position of the ffinity levels of the dsorbte is not yet well understood. 17,47 Tht discussion hs generlly focused on the energies of moleculr orbitls, while in this work it is bsed on the energy of the involved sttes. About the SERS Intensity of the 12 Vibrtion. The different intensities of this vibrtion in the spectr of Creighton et l. 17,37 nd in the ones reported here hve lredy been mentioned (see Tbles 3 nd 4). These discrepncies could be due to the different surfces used, but the possibility tht this could be due to nother effect still remins. Despite tht the vlues for this norml mode correspond very well with the noticeble intensities detected in our spectr, it should not be discrded tht in our SERS-CT n pprecible contribution from the EM mechnism in the intensities of the bnds still exists. Without tking into ccount surfce orienttion considertions, the EM mechnism will give generl nd uniform enhncement of the Rmn spectr. This effect cn be seen in the spectr obtined t 0.00 or V tht re chrcterized by the strong bnds ssigned to the 1 nd 12 modes s in the pure liquid or the solution Rmn spectr. These spectr, nd especilly those of Creighton in Ag or Au colloids, 35,36 could be referred to s SERS-EM. Therefore, it could be possible tht the enhncement of the 1 nd 12 modes in our spectr is prtilly due to tht EM mechnism. This could be the cuse of the nomlous behvior observed in the SERS intensities of the 12 mode with respect to the electrode potentil. This is the only cse long with ν(ch) vibrtions of the intensity decying smoothly when incresing the negtive electrode potentil. This hypothesis would imply tht the Creighton s surfces re prticulrly SERS-CT ctive, while those of Moskovits et l. or the ones used in this work produce contributions of both mechnisms. On the other hnd, it cn be derived tht in order to predict CT enhncement for this vibrtion comprble to the remining norml modes, much greter vlues re required thn for the remining ones. Any evidence to ccount for this nonsystemtic behvior hs not yet been found. The vlue for the vibrtion 12;δ ring is bsed fundmentlly on ring stretching contributions nd not on the R 17 internl coordinte given tht the R vlue for this coordinte is much less (R 17 ) -0.03) thn tht for the R 18 (R 18 ) in pyridine or in pyrzine). As in the previously discussed cses, pprecible mode mixing is not shown between the S 0 nd D 0 sttes (see D ij nd ν vlues in Tble 7). There is no doubt tht the SERS vlues of Tble 4 t -0.75, -1.00, or V show chrcteristic fetures of SERS-CT spectr, but it is possible tht the 1 nd 12 modes contin n EM contribution. This ssumption gin rises the problem of seprting the contributions from both mechnisms in the SERS results, question tht will remin open due to surfce orienttion effects on the EM mechnism nd to the bsence of n dequte stndrd tht llows subtrction of the EM contribution. In this respect, the ν(ch) vibrtions re in principle good cndidtes given tht they possess smll vlues nd therefore re not sensitive to the CT mechnism. Conclusions As in the cse of pyrzine, 1 our method of nlysis hs been cpble of ccounting for the min fetures in the SERS spectr of pyridine s well s those of the vibronic spectr involving the S 1 -S 0 trnsition. The results of the displcements obtined from the optimized geometries of the neutrl molecule nd the rdicl nion of pyridine re closely relted with those norml modes tht show significnt enhncement in the SERS spectr obtined t negtive electrode potentils. This reltionship depends on the nture of ech norml mode. In this respect, in the previous work on pyrzine comprtive discussion with the results of three published RR spectr ws included. In this work the results for pyrzine hve been used s reference in order to discuss the results for pyridine. In the cses where the norml modes re similr, it is possible to predict the correct behvior of the SERS-CT bnds. nd / clcultions hve been sufficient to ccount for the behvior found in the spectr. The differences in energy between the S 0 nd D 0 sttes cn explin why the CT mechnism for pyridine is observed t more negtive potentils thn for pyrzine. Given the good results obtined in the nlysis of the SERS results of pyrzine nd pyridine, it is possible to conclude tht in both cses the resonnt CT stte nd the rdicl nion should be quite similr. If this generl conclusion continues to be stisfied in future works, the SERS experiments on electrode using conventionl Rmn spectrometer could be n invluble tool in obtining informtion on rdicl nions nd chrge trnsfer processes. Acknowledgment. The uthors express their grtitude to the DGICYT for the finncil support of this work through Project PB93/0973/C02/01. References nd Notes (1) Arens, J. F.; Woolley, M. S.; Otero, J. C.; Mrcos, J. I. J. Phys. Chem. 1996, 100, (2) Moskovits, M. ReV. Mod. Phys. 1985, 57, 783. (3) Creighton, J. A. The Selection Rules for Surfce-enhnced Rmn Spectroscopy. In Spectroscopy of Surfces; Clrk, R. J. H., Hester, R. E., Eds.; Wiley: Chichester, 1988 nd references therein. (4) Otto, A.; Grbhorn, H.; Akemn, W. J. Phys.: Condens. Mtter 1992, 4, (5) Ymd, H.; Tob, K.; Nko, Y. J. Electron Spectrosc. Relt. Phenom. 1987, 45, 113. Ymd, H.; Ngk, H.; Tob, K.; Nko Y. Surf. Sci. 1987, 182, 269. (6) Schmeisser, D.; Demuth, J. E.; Avouris, Ph. Chem. Phys. Lett. 1982, 87, 324. (7) Avouris, Ph.; Demuth, J. E. J. Chem. Phys. 1981, 75, (8) Gdzuk, J. W. Annu. ReV. Phys. Chem. 1988, 39, 395. (9) Gdzuk, J. W. J. Chem. Phys. 1983, 79, (10) Nenner, I.; Schulz, G. J. J. Chem. Phys. 1975, 62, (11) Wong, S. F.; Schulz, G. J. Phys. ReV. Lett. 1975, 35, (12) Strommen, D. P.; Nkmoto, K. J. Chem. Educ. 1977, 54, 474. Hirkw, A. Y.; Tsuboi, M. Science 1975, 188, 359. (13) Clrk, R. J. H.; Dines, T. J. Angew. Chem., Int. Ed. Engl. 1986, 25, 131. (14) Moskovits, M.; DiLell, D. P.; Mynrd, K. J. Lngmuir 1988, 4, 67. (15) Muniz-Mirnd, M.; Neto, N.; Sbrn, G. J. Phys. Chem. 1988, 92, 954. (16) Suzuk, I.; Udgw, Y.; Ito, M. Chem. Phys. Lett. 1979, 64, 333. (17) Creighton J. A. Surf. Sci. 1986, 173, 665. (18) Gllup, G. A. J. Chem. Phys. 1993, 99, 827. (19) Frisch, M. J.; Hed-Gordon, M.; Trucks, G. W.; Foresmn, J. B.; Schlegel, H. B.; Rghvchri, K.; Robb, M.; Binkley, J. S.; Gonzlez, C.; Defrees, D. J.; Fox, D. J.; Whiteside, R. A.; Seeger, R.; Melius, C. F.; Bker, J.; Mrtín, R. L.; Khn, L. R.; Stewrt, J. J. P.; Topiol, X.; Pople, J. A. GAUSSIAN 90, ReVision H; Gussin, Inc.: Pittsburgh, PA, (20) Chipmn, D. M.; Prebend, M. F. J. Phys. Chem. 1986, 90, Wheeler, R. A. J. Phys. Chem. 1993, 97, 1533 nd references therein. (21) Hinde, A. L.; Poppinger, D.; Rdom, L. J. Am. Chem. Soc. 1978, 100, (22) Liu. R; Zhou, X. J. Phys. Chem. 1993, 97, (23) Geometry optimiztions for the excited singlets of pyrzine were crried out t the CIS/ level. (24) Puly, P; Hmilton, T. P. J. Chem. Phys. 1988, 88, Bofill, J. M.; Puly, P. J. Chem. Phys. 1989, 90, (25) Orlndi, G.; Zerbetto, F.; Zgierski, M. Z. Chem. ReV. 1991, 91, 867. (26) Peticols,W. L.; Strommen, D. P.; Lkshminrynn, V. J. Chem. Phys. 1980, 73, 4185 nd references therein.

8 Frnck-Condon Active Vibrtions of Pyridine J. Phys. Chem., Vol. 100, No. 22, (27) Arens, J. F.; López Tocón, I.; Otero, J. C.; Mrcos, J. I. J. Phys. Chem. 1995, 99, (28) Puly, P.; Fogrsi, G.; Png, F.; Boggs, J. E. J. Am. Chem. Soc. 1979, 101, (29) Arens, J. F.; Lopez-Nvrrete, J. T.; Mrcos, J. I.; Otero, J. C. Spectrochim. Act 1986, 42A, (30) McIntosh, D. F.; Peterson, M. R. Generl Vibrtionl Anlysis System, QCPE #576, (31) Puly, P.; Fogrsi, G.; Boggs, J. E. J. Chem. Phys. 1981, 74, (32) Stidhm, H. D.; DiLell, D. P. J. Rmn Spectrosc. 1979, 8, 180; 1980, 9, 90, 247. DiLell, D. P. J. Rmn Spectrosc. 1980, 9, 239. (33) Pongor, G.; Puly, P.; Fogrsi, G.; Boggs, J. E. J. Am. Chem. Soc. 1984, 106, Mgdó, I.; Pongor, G.; Fogrsi, G. J. Mol. Struct.: THEOCHEM 1994, 303, 243. (34) Vrsányi, G. Vibrtionl spectr of benzene derivtives; Acdemic Press: New York, (35) Creighton, J. A.; Bltchford, Ch. G.; Albrecht, M. G. J. Chem. Soc., Frdy Trns , 75, 790. (36) Creighton J. A. Surf. Sci. 1985, 158, 211. (37) Creighton J. A.; Alvrez, M. S.; Weltz, D. A.; Groff, S.; Kim, M. W. J. Phys. Chem. 1983, 87, (38) CIS/ clcultion predicts C 2V equilibrium structure for the S 1 stte of pyridine, but CIS/6-31G* force field clcultion indictes nonplnr C s structure with lower energy (Foresmn, J. B.; Hed-Gordon, M.; Pople, J. A.; Frisch, M. J. J. Phys. Chem. 1992, 96, 135). This is the only stte of pyrzine or pyridine here studied with nonplnr structure. (39) Mochizuki, Y.; Ky, K.; Ito, M. J. Chem. Phys. 1976, 65, (40) Vill, E.; Amirv, A.; Lim, E. C., J. Phys. Chem. 1988, 92, (41) Brolo, A; Irish, D. E. XIVth Interntionl Conference on Rmn Spectroscopy; Yu, N.-T., Li, X.-Y, Eds.; Wiley: Chichester, (42) See, for exmple: Moskovits, M.; Suh, J. S. J. Phys. Chem. 1984, 88, (43) Arens, J. F.; Lopez-Nvrrete, J. T.; Otero, J. C.; Mrcos, J. I.; Crdenete, A. J. Chem. Soc., Frdy Trns , 81, 405. (44) A similr Fermi resonnce effect on the fluorescence intensities of pyrzine cn be found: McDonld, D. B.; Rice, S. A. J. Chem. Phys. 1981, 74, 893. (45) Duschinski, F. Act Physicochim. URSS 1937, 7, 551. (46) See, for exmple: Tripthi, G. N. R.; Chipmn, D. M.; Miderski, C. A.; Dvis, H. F.; Fessenden, R. W.; Schuler, R. H. J. Phys. Chem. 1986, 90, (47) Otto, A.; Billmnn, J.; Eickmns, J.; Ertück, U.; Pettenkofer, C. Surf. Sci. 1984, 138, 319. JP953712Y