direct measurement of the energy thresholds to ... measurement of the energy thresholds to...

Direct measurement of the energy thresholds to conformationalisomerization in Tryptamine: Experiment and theory

Jasper R. Clarkson, Brian C. Dian,a! and Loïck Moriggib!

Department of Chemistry, Purdue University, West Lafayette, Indiana 47906

Albert DeFusco, Valerie McCarthy, and Kenneth D. Jordanc!,d!

Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260

Timothy S. Zwierc!,e!

Department of Chemistry, Purdue University, West Lafayette, Indiana 47906

sReceived 7 March 2005; accepted 1 April 2005; published online 7 June 2005d

The methods of stimulated emission pumping-hole filling spectroscopysSEP-HFSd and stimulatedemission pumping population transfer spectroscopysSEP-PTSd were applied to theconformation-specific study of conformational isomerization in tryptaminefTRA,3-s2-aminoethyldindoleg. These experimental methods employ stimulated emission pumping toselectively excite a fraction of the population of a single conformation of TRA to well-definedground-state vibrational levels. This produces single conformations with well-defined internalenergy, tunable over a range of energies from near the zero-point level to well above the lowestbarriers to conformational isomerization. When the SEP step overcomes a barrier to isomerization,a fraction of the excited population isomerizes to form that product. By carrying out SEP excitationearly in a supersonic expansion, these product molecules are subsequently cooled to their zero-pointvibrational levels, where they can be detected downstream with a third tunable laser that probes theground-state population of a particular product conformer via a unique ultraviolet transition usinglaser-induced fluorescence. The population transfer spectrasrecorded by tuning the SEP dump laserwhile holding the pump and probe lasers fixedd exhibit sharp onsets that directly determine theenergy thresholds for conformational isomerization in a given reactant-product conformer pair. Inthe absence of tunneling effects, the first observed transition in a givenX−Y PTS constitutes anupper bound to the energy barrier to conformational isomerization, while the last transition notobserved constitutes a lower bound. The bounds for isomerizing conformer A of tryptamine toBs688–748 cm−1d, Cs1ds860–1000 cm−1d, Cs2ds1219–1316 cm−1d, Ds1219–1282 cm−1d,Es1219–1316 cm−1d, andFs688–748 cm−1d are determined. In addition, thresholds for isomerizingfrom B to As,1562 cm−1d, B to Fs562–688 cm−1d, and out ofCs2d to Bs,747 cm−1d are alsodetermined. TheA→B andB→A transitions are used to place bounds on the relative energies ofminima B relative toA, with B lying at least 126 cm−1 aboveA. The corresponding barriers havebeen computed using both density functional and second-order many-body perturbation theorymethods in order to establish the level of theory needed to reproduce experimental results. Whilemost of the computed barriers match experiment well, the barriers for theA–F andB–F transitionsare too high by almost a factor of 2. Possible reasons for this discrepancy are discussed. ©2005American Institute of Physics. fDOI: 10.1063/1.1924454g

I. INTRODUCTION

Molecules with several torsional degrees of freedom un-dergo conformational isomerization on highly corrugated,multidimensional potential-energy surfaces containing manyminima and an even greater number of transition states sepa-rating them. When there are local minima close in energy tothe global minimum, they can acquire significant population,even when the molecule is cooled in a supersonic expansion.

Building off a foundation of conformation-specific infraredand ultraviolet spectroscopies,1–10 we have recently beenstudying the conformational isomerization dynamics of suchmolecules using a new experimental protocol developed ex-pressly for this purpose.11–14As shown schematically in Fig.1 this protocol involvessid initial cooling of the molecules ofinterest to their zero-point levels,sii d selective laser excita-tion of a single “reactant” conformation in the mixture whilethe molecules are still in a high-density region of the expan-sion where collisions are prevalent,siii d isomerization andrecooling of the molecules, followed bysivd selective detec-tion of a single “product” conformation downstream in thecollision-free region of the expansion with a probe laser.

The initial studies that used this cool-excite-cool-probescheme employed infrared excitation of a single conforma-tion via the amide NH stretch fundamentals to initiate con-

adPresent address: Department of Chemistry, University of Virginia, McCor-mick Road, Charlottesville, Virginia 22901

bdPresent address: Institute of Chemical Sciences and Engineering, ÉcolePolytechnique Féderale De Lausanne, Lausanne CH-1015, Switzerland

cdAuthors to whom correspondence should be addressed.ddElectronic mail: [email protected] mail: [email protected]

THE JOURNAL OF CHEMICAL PHYSICS122, 214311s2005d

0021-9606/2005/122~21!/214311/15/$22.50 © 2005 American Institute of Physics122, 214311-1

Downloaded 03 Aug 2006 to 128.210.142.201. Redistribution subject to AIP license or copyright, see http://jcp.aip.org/jcp/copyright.jsp

http://dx.doi.org/10.1063/1.1924454

formational excitation.11–14 Conformation-selective excita-tion and detection steps provide a means of dissecting thecomplex set of isomerization processes into individualX→Y reactant-product pairs. When the infrared laser is fixedand the probe laser tuned, one can determine which productsare formed. This scheme is called IR-UV hole-filling spec-troscopy because the “hole” created in the ground-statepopulation of the excited conformer is used to fill in popula-tion in product conformers. On the other hand, when the IRlaser is tuned while the UV probe laser is fixed to monitor agiven product conformer population, the resulting spectrumfcalled an IR-induced population transfer spectrumsIR-PTSdg records the fractional population change induced bythe infrared laser. By piecing together the results of a seriesof IR-population transfer spectra, it is possible to determinethe isomerization product quantum yields followingconformation-selective infrared excitation.

One of the nonideal aspects of the infrared excitationscheme is that absorption of one infrared photon in the hy-dride stretch fundamental regions,3500 cm−1d puts10 kcal/mol of energy into the excited conformation, wellabove many of the lowest-energy barriers to conformationalisomerization s3–5 kcal/mol typicald.12 Recently, we re-ported first results of a variation of the hole-filling methodthat employs simulated emission pumpingsSEPd rather thaninfrared excitation.14 SEP is a powerful spectroscopic tech-nique which has been utilized in a number of ways.15–17WithSEP, it is possible to tune widely over ground-state vibra-

tional energies ranging from the vibrational zero-point levelto well above the lowest-energy barriers to isomerization. Indoing so, it is possible to directly measure the energy thresh-olds to isomerization for individualX→Y reactant-productpairs. Thus, the SEP-PT spectra provide a spectroscopicmeans of placing bounds on the barrier heights to conforma-tional isomerization. This is an exciting prospect becausespectroscopy almost always characterizes the minima on thepotential-energy surface rather than the barriers.

The SEP hole-filling scheme was carried out first ontryptaminefTRA, 3-s2-aminoethyldindoleg. In the initial re-port, we demonstrated the ability of the method to measurethe barriers to conformational isomerization, even in a casewhere there are several isomers present.14 In this respect,TRA is quite a challenge because it possesses seven confor-mational isomers with significant population, even underconditions of supersonic expansion cooling.18–22As a result,there are 42 individualX→Y reactant-product pairs. Ofcourse, theX→Y andY→X pairs measure the same barrierfrom either direction. The difference in these energy thresh-olds then constitutes a measure of the energy differencesbetween the minima; that is,

EthreshsX → Yd − EthreshsY → Xd = EminsYd − EminsXd. s1d

Thus, the combined results from all 42X→Y pairs could inprinciple constitute a complete characterization of the ener-gies of the relevant stationary points on the potential-energysurface for isomerization.

FIG. 1. sad Schematic diagram of the spatial temporalarrangement of the experiment. Selective stimulatedemission pumpingsSEPd excitation of a single confor-mation is carried out early in the supersonic expansion,followed by collisional recooling either back into theoriginal minimum or into other minima followingisomerization. The changes in conformational popula-tion are detected downstream using laser-induced fluo-rescencesLIFd. In SEP hole-fillingsSEP-HFd spectros-copy, the firstspump 20 Hzd laser and secondsdump 10Hzd UV lasers are held fixed while the third UV lasersprobe 20 Hzd is scanned. In the SEP-population trans-fer sSEP-PTd, the dump laser is tuned while the probelaser is fixed at a unique vibronic transition. In eithercase, the difference in fluorescence signal with andwithout the dump laser is detected downstream.

214311-2 Clarkson et al. J. Chem. Phys. 122, 214311 ~2005!


In the present paper, we provide a complete account ofthe SEP hole-filling studies of tryptamine. In addition to amore detailed exposition of the full experimental data set, wehave now carried out an extensive set ofab initio anddensity-functional theoreticalsDFTd calculations of the tran-sition states for conformational isomerization in TRA, forcomparison with the experimental data. As we shall see, theexperimental data match well with calculations for most ofthe reactant-product pairs measured. However, the thresholdsinto conformer wellF are computed to be about a factor of 2higher than the experiment. The effect of tunneling on themeasured isomerization thresholds has been explored by us-ing tryptamine deuterated at the amino site.

The study of TRA presented here is followed up in anadjoining paper by an analogous study of conformationalisomerization in 3-indole propionic acidsIPAd and its water-containing complexes.23

II. METHODS

A. Experiment

The laser-induced fluorescencesLIFd chamber used inthese experiments has been described in detail elsewhere.11

Those aspects of the experiment that are unique to the SEPhole-filling method are described here. The three UV laserbeams enter the chamber through CaF2 windows mounted atthe Brewster’s angle before passing through a set of bafflescontaining 2310 mm slits to allow easy spatial movementof the laser beams intersecting the supersonic expansion. Thelaser beams intersect the supersonic expansion at right angleswhile the fluorescence is collected by a filtered photomulti-plier tube sPMTd sETI 9318QB with WG320 and WG305cut-off filtersd perpendicular to both the jet and the excitationbeam. TRAsAldrichd was resistively heated to 400 K in astainless-steel sample holder. The sample was entrained inhelium scommercial grade, 99.995%d at a total pressure of 7bars and expanded through a 20-Hz pulsed valvesParkerGeneral, Series 9, 2-mm orificed into a vacuum chamberpumped with a roots blowersLeybold, WS-1001d backed bytwo rotary vane mechanical pumpssSargent-Welchd to super-sonically cool the molecules into their conformational-specific zero-point levels. Total flow rates of 2310−3 bar L/s in a 1.0–1.5-ms gas pulse were used, leadingto a vacuum chamber pressure of,5310−5 bar.

Three independent Nd:YAG-syttrium aluminum garnetdpumpedsContinuum 7000 series and NY-61d doubled dyelaserssLumonics HD-500, Lambda Physik Scanmate 2E andFL3002d were employed. Foundational to all methods waslaser-induced fluorescence excitation spectroscopy, in whicha single ultraviolet lasers,0.1 mJ/pulsed was tuned throughtheS1←S0 origin region of jet-cooled tryptamine. LIF spec-troscopy was carried out at a distancesxd 12 mm down-stream from the 2-mm-diameter expansion orificesDd scor-responding tox/D=6d. SEP spectra were recorded at thissame spatial position, with the pump laserl1 fixed on theS1←S0 origin band of a specific conformer. In the dumpstep, the excited-state population was transferred to a par-ticular vibrational level in the ground statefS0svdg via stimu-lated emission induced by a second tunable lasersthe

“dump” laserd. Under optimal conditions, as much as 30%–40% of the population in theS1 state of a particular confor-mation of TRA can be driven back toS0svd. The magnitudeof this population transfer depends onsid the laser powers inboth pump and dump steps,sii d the lifetime of theS1 state,siii d the delay between pump and dump lasers,sivd the oscil-lator strength and Franck–Condon factors for the dump tran-sitions, andsvd the rate of collisional removal from the lowerlevel.

The SEP spectra were recorded using the active baselinesubtraction mode of a gated integratorsSRS 250d. The dumplasers10 Hzd was pulsed every other time the pump lasers20Hzd fires. A negative signal indicates that the dump laser hasdepleted the fluorescence signal. Typical pulse energies forthe pump and dump steps were 0.1–0.2 and 0.5–1.0 mJ, re-spectively. Focusing conditions were chosen to ensure thatthe dump laser beam was larger than the pump laser, with thepump and dump lasers focused with 50- and 70-cm focallength lenses, respectively.

In order to detect the SEP dump transition, the two laserswere delayed somewhere in the range of 2–10 ns, determinedby a compromise between maximizing the dip in the fluores-cence signal and minimizing the interference from scatteredlight from the dump laser. The gated signal was delayed asfar from the scattered light as possible while collecting suf-ficient fluorescence to detect the depletion in fluorescencesignal created by the dump laser.

The triple-resonance methods presented here share thepump-dump-probe excitation scheme with the studies ofKable and Knight24 and Burgiet al.25,26The unique aspect ofthe present work is the place of SEP excitation, sufficientlydownstream so that SEP excitation could be cleanly carriedout from the vibrational zero-point levels of the conforma-tional isomers, but in a collisional regime where sufficientcollisional cooling occurred subsequent to the SEP step torecollect the population into the zero-point levels of the vari-ous conformers prior to probing downstream. The lasers usedfor SEP excitation typically intersected the expansion atx/D=2–2.5, while the probe laser was set tox/D=6, pro-ducing a delay between pump/dump and probe of 1.8–2.0msfor optimal signal, determined by the terminal velocity of ahelium expansions1.83105 cm/sd. We estimate a collisionrate for tryptamine with helium at 43109 s−1 at the SEPexcitation point in the expansion under typical expansionconditions, falling off assx/Dd2.

Two schemes, differing in terms of which lasers weretuned and which were fixed in wavelength, were employedfor the infrared excitation.11 In SEP-population transfer spec-troscopy, the pump and probe lasers were fixed on theS1

←S0 origin transitions of two conformational isomersse.g.,A andB, respectivelyd, while the dump laser was tuned. Asits name implies, population transfersPTd spectroscopymonitors the amount of population transferred fromA to B asa function of the internal energy placed inA by the dumplaser:

A*sEintd → B. s2d

In order to selectively highlight the population transferdue to the dump laser, the pump and probe lasers operated at

214311-3 Conformational isomerization in tryptamine J. Chem. Phys. 122, 214311 ~2005!


20 Hz, while the dump laser operated at 10 Hz. The differ-ence in signal between successive probe pulses was recordedusing active baseline subtraction. By recording the LIF sig-nal size due to conformerB before and after the PT spec-trum, the ordinate can be recorded as a fractional change inthe population of the productB induced by the dump laser.

A second variation, SEP hole-filling spectroscopysHFSd,has both the pump and dump wavelengths fixed; therebyselectively exciting a particular conformational isomer to aground-state vibrational level with well-defined internal en-ergy. In HFS, the probe laser was scanned to determinewhere the population went, with gains in the probe signaloccurring whenever transitions due to energetically openproduct channels were encountered.

Since PT and HF spectroscopies require careful spatialand temporal controls of three laser beams, two laser checkswere developed to assist in setting up the experiment. Theproper timing and spatial arrangement between pump andprobe lasers could be set by pulling back the nozzle so thatcollisional refilling could not occur, and then maximizing thedip in the fluorescence signal from conformerA produced bythe pump laser fixed on the same transition. Maximizing theSEP dip was used to maximize the spatial overlap betweenpump and dump. These two checks were typically sufficientto observe some signal in the three-laser experiment. Finaloptimization of the delay and spatial overlap between pumpand dump was carried out on the three-laser signal underhole-filling conditions.

TRA deuterated at the NH2 and indole NH sitesfTRAsd3dg was prepared in order to test for the effects oftunneling involving the NH2 hydrogens on the measured en-ergy thresholds for isomerization. Because TRA is relativelyinsoluble in pure D2O, the sample was dissolved in excesstetrahydrofuran in the presence of D2O. The exchange wasallowed to occur for several hours before removal of thesolvent with a rotary evaporator. The procedure was repeatedthree times to bring the sample to near-complete deuterationat the ND2 site. The resulting purity and further experimentaldetails concerning the deuterated sample will be presented inthe Results section.

B. Calculations

The minimum-energy structures were optimized usingthe Becke3 Lee–Yang–Parr27,28 sLYPd density-functionalmethod with both the 6-31+Gsdd sRef. 29d and augmentedcorrelation consistent polarized valence double zeta30–32

saug-cc-pVDZd basis sets. Transition state structures werethen optimized at the Becke3LYP/6-31+Gsdd level and em-ploying the quadratic synchronous transitsQST3d sRef. 33dalgorithm. These calculations were followed up with theresolution of the identity second-order Møller-PlessetsRIMP2d sRefs. 34 and 35d/aug-cc-pVDZ single-point calcu-lations at the Becke3LYP/6-31+Gsdd geometries. In addi-tion, all local minima and a subset of the transition stateswere reoptimized at the RIMP2/aug-cc-pVDZ level. For se-lected minima and transition state structures, single-pointRIMP2/augmented correlation consistent polarized valencetriple zeta30–32 saug-cc-pVTZd and second-order Moller–PlessetsMP2d/aug-cc-pVDZ energies were calculated usingRIMP2/aug-cc-pVDZ geometries, and in some cases, MP2/aug-cc-pVDZ geometry optimizations were performed aswell. The results for these calculations are presented in theirentirety in supplementary material and will be referred tothroughout the text as necessary.36

The RIMP2 procedure uses a resolution of the identityoperator approach to reduce the computational effort associ-ated with MP2-level calculations. Several studies haveshown that the errors introduced in relative energies due tothe use of this approximation is generally 0.1 kcal/mol orless.34 The main advantage of the MP2 and RIMP2 proce-dures over the computationally faster Becke3LYP method isthe proper treatment of long-range dispersion interactions.6

The Becke3LYP and MP2 calculations were carried out us-ing theGAUSSIAN 03 program37 and the RIMP2 calculationswere carried out using theTURBOMOLE program.38 AllBecke3LYP calculations used the ultrafine grid and tight con-vergence criteria as implemented inGAUSSIAN 03.

As will be discussed below, there are significant differ-ences in some of the relative energies obtained from theBecke3LYP and RIMP2 calculations. To gain insight into theorigin of these differences we also carried out coupled clus-

TABLE I. Dihedral anglessdegreesd of local minima of tryptamine. For each conformer, with the exception ofG, there is a second structure differing only by exchange of the two NH2 H atoms. Results fromBecke3LYP/6−31+Gsdd calculations.

a-type enantiomers b-type enantiomers

Conformer a b c € a b c

A −76.13 −64.04 122.77 76.13 64.04 −122.77B −74.43 −62.80 −1.19 74.43 62.80 1.19H −149.79 −66.73 −121.53 149.79 66.73 121.53C2 −85.08 65.68 −125.62 85.08 −65.68 125.62J −68.60 82.57 133.11 68.60 −82.57 −133.11F −91.15 61.40 2.10 91.15 −61.40 −2.10

C1 −76.06 −179.37 −126.15 76.06 179.37 126.15E −76.51 178.72 −0.75 76.51 −178.72 0.75D −75.60 176.27 125.88 75.60 −176.27 −125.88G −179.99 180.00 0.00 179.99 −180.00 0.00I −177.92 178.73 125.48 177.92 −178.73 −125.48



ter singles, doublesstriplesd fCCSDsTdg39,40 coupled clustersingle point calculations for the various minima using the6-31+Gsdd basis set and the RIMP2 geometries.

III. CONFORMATIONAL ASSIGNMENTSAND CALCULATED STATIONARY POINTSON THE POTENTIAL-ENERGY SURFACEOF TRYPTAMINE

The present work on conformational isomerization dy-namics of TRA builds off previous studies that have assignedthe observedS1←S0 vibronic transitions to particular confor-mational isomers.3,18–22The recent high-resolution study ofthe origin bands of the isomers by Nguyenet al.22 haveprovided convincing evidence for the assignments given in

Fig. 2sad. The labels in the figure make use of a shorthandnotation for the conformers which designates the position ofthe amino group relative to the indole ring: Anti,gaucheonthe phenylsGphd, andgaucheon the pyrrolesGpyd side ofindole.3 The label in parentheses refers to the direction of theamino nitrogen lone pair relative to the indole ring.

FIG. 2. sad LIF excitation spectrum of tryptamine in the region of theS1←S0 origins. Seven conformations are observed under supersonic condi-tions. The ethylamine side chain adopts several positions relative to theindole ring as shown and labeled in the inset.sbd Tryptamine dihedral angledefinitions used in this work. Note that atom 25 is a “dummy” atom used tosimplify the motions of the amino rotation.

FIG. 3. sad Schematic PES of the nine calculated minima of tryptamine andtheir abbreviated structural designations plotted along two flexible coordi-natesC andb corresponding to an amino internal rotationsC17–N20d and aC16–C17 rotation, respectively.sbd The calculated energies of the minimaand all associated transition states interconnecting all minima about the twointernal coordinates. All energies are in kcal/mol from B3LYP/6-31+Gsdd optimizations employing the QST3 algorithm for the transition statestructures.



A. Conformational minima

The flexible ethylamine side chain present in TRA has atotal of four coordinates along which large-amplitude motioncan occurfFig. 2sbdg. These include internal rotation aboutthe C17–N20sdihedral anglecd, C16–C17sdihedral anglebd, and C15–C16sdihedral anglead bonds and inversion ofthe NH2 group. Table I summarizes the computedBecke3LYP values of these torsional angles for all minimalocated on the potential-energy surface. The geometrical pa-rameters obtained at the RIMP2 level are reported in Table Ain the supplementary material

The seven lowest-energy minimasas determined fromthe Becke3LYP calculationsd share a common configurationabout the C15–C16 bond that points the C16–C17 bond ap-proximately perpendicular to the indole ringsa= ±90°d. Theinversion of the NH2 group has a barrier high enough that itis safe to conclude that the lowest-energy conformers differprimarily in the angles of internal rotation about the C16–C17 and C17–N20 bondssb andcd. As a result, it is usefulto consider the potential-energy surface as a function ofbandc shown in Fig. 3. The threefold character of the poten-tial along each of these coordinates produces nine conforma-tional minima, seven of which are observed experimentally.The structures associated with these minima are shown inFig. 3sad, and the computed energies of the minima and tran-sition statessfirst-order saddle pointsd are given in Fig. 3sbd.The two unobserved conformers are those in which the ni-trogen lone pair points in toward the indolep cloudfGpysind /H and Gphsind /Jg, which destabilizes theseminima relative to the others. In conformerH, this leads to asubstantial distortion of the structure in which the side chainrotates to about −150.0° about the C15–C16 bond so that thenitrogen lone pair is stabilized by interaction with an aro-matic C–H group. One could argue that the distortion is solarge that structureH does not actually belong on the two-dimensionals2Dd potential-energy surface depicted in Fig. 3.

Two other minima, labeledG and I in the tables, havethe heavy atoms of the ethylamine side chain lying in orclose to the plane of the aromatic ringfFig. 4sadg. Theseminima are formed by internal rotation of each of the three

“anti” structuresE,Cs1d, and D about the C15–C16 bond.These structures highlight the possibility that pathways toisomerization could swing the ethylamine side chain fromone side of the indole plane to the other. As Fig. 4sbd shows,each of the conformational minima has a mirror image iso-mer. These enantiomersslabeled a and bd are spectroscopi-cally indistinguishable, but could play a role in isomerizationpathways that swing the ethylamine side chain from one sideof the indole ring to the other.

Finally, a combination of inversion and internal rotationof the NH2 group swaps the positions of the two hydrogens.These spectroscopically indistinguishable isomersflabeledwith and without a primes8dg are relevant to isomerizationpathways that involve these coordinates.

The relative energies of the minima computed at severallevels of theory are given in Table II. The relative energies atthe Becke3LYP/6-31+Gsdd and Becke3LYP/aug-cc-pVDZlevels of theory are in fairly good agreement. Stress is placedon the Becke3LYP/6-31+Gsdd level since the transitionstate optimizations with the Becke3LYP method were carriedout only with the same basis set. Inspection of these resultsreveals thatA,B,C2,andF are preferentially stabilized by0.6–1.8 kcal/mol relative to the other minima when goingfrom the Becke3LYP to the RIMP2 method.

For all minima the relative energies from the RIMP2/aug-cc-pVDZ and MP2/6-31+Gsdd are in fairly good agree-ment, which indicates that basis set superposition error41 isnot the origin of the discrepancies between the Becke3LYPand MP2 results. In order to determine the origin of theinconsistencies between the RIMP2 and Becke3LYP results,we also carried out CCSDsTd /6-31+Gsdd calculations atseveral of the local minima using the Becke3LYP/6-31+Gsdd and RIMP2/aug-cc-pVDZ geometries. In all cases therelative energies obtained from the CCSDsTd calculations arein good agreement with the RIMP2 results. This leads us toconclude that the stabilization ofA,B,C2,andF relative tothe other conformers is a consequence of a greater impor-tance of dispersion interactions in the former set of minima.Further support for this conclusion is provided by the resultsof the Hartree–Fock calculations, namely, the energy lower-

FIG. 4. Two minima not shown on the schematic PESwhich have the ethylamine side chain in or near theplane of the aromatic ring.sbd Each of the minima hasa mirror image isomer. ConformerB of tryptamine ishighlighted as an example. These isomers and their mir-ror images could play a role in isomerization pathways.



ing in going from the Hartree–Fock to the MP2 method isappreciable greater forA,B,C2,andF than for the otherconformers.

Figure 5 summarizes the energies of the various con-formers calculated at the different levels of theory, with con-former I being chosen as the zero of energy. This figureunderscores the close agreement between the RIMP2 andCCSDsTd results and the fact that the Becke3LYP method isgiving relative energies roughly intermediate between theHartree–Fock and CCSDsTd results. Further it is seen thatthe dispersion in the relative energies predicted by the vari-ous methods grows as one proceeds from the structures withthe chain “in the plane”sright-hand sided to the structureswith one of the amino NH bonds pointed toward the ringsystemsleft-hand sided.

B. Transition states

Searches for transition states connecting the minimawere carried out at the Becke3LYP/6-31+Gsdd level oftheory. Structural parameters associated with the transitionstates are given in Table B of the supplementary material,while their energies relative to the lowest-energy minimumsconformerAd are collected in Table III.

For the most part, the transition state energies obtainedfrom the Becke3LYP and RIMP2 methods are in fairly goodagreement, with the largest changes being in those cases forwhich one is proceeding from one of the minimaA,B,C2,or F to one of the other minima. In these cases theenergies calculated at the RIMP2 level tend to be about1 kcal/mol higher than the corresponding values calculatedat the Becke3LYP level. The predictions from theory serve asa point of comparison for the experimental results describedin Sec. IV.

IV. EXPERIMENTAL RESULTS AND ANALYSIS

A. “Upstream” LIF spectra

The hole-filling experimental protocolsSec. II Ad callsfor SEP excitation of a single conformer to take place out ofthe vibrational zero-point level of that conformer. In doingso, the initial internal energy available to the conformer forisomerization has a single, well-defined value determined bythe difference in the wavelengths of the pump and dumplasers. Much of this selective excitation is achieved by thechoice of pump wavelength, which is fixed on theS0-S1 ori-gin transition of the conformer of interest. In addition, weseek to carry out excitation at a point in the expansion where

TABLE II. Relative energiesskcal/mold of the local minima with respect toA.

Theoretical methoda

Conformer

A B C1 C2 D E F G H I J

- -HF/aug-cc-pVDZ/ /RIMP2/aug-cc-pVDZ 0.00 0.33 0.14 0.74 0.16 0.06 1.27 0.25 0.92 0.24 2.51B3LYP/6-31+Gsdd / /B3LYP/6-31+Gsdd 0.00 0.26 0.74 0.54 0.73 0.59 0.84 1.18 1.17 1.45 2.66B3LYP/aug-cc-pVDZ/ /B3LYP/aug-cc-pVDZ 0.00 0.31 0.57 0.56 0.57 0.58 0.96 0.93 1.12 1.00 2.55RIMP2/aug-cc-pVDZ/ /B3LYP/6-31+Gsdd 0.00 0.62 1.59 0.26 1.59 1.68 1.02 2.41 1.93 2.51 3.28RIMP2/aug-cc-pVDZ/ /RIMP2/aug-cc-pVDZ 0.00 0.65 1.83 0.21 1.84 1.91 0.95 2.75 2.24 2.86 3.35RIMP2/aug-cc-pVTZ/ /RIMP2/aug-cc-pVDZ 0.00 0.56 1.68 0.39 1.68 1.74 1.01 2.54 2.26 2.64 3.34MP2/6-31+Gsdd / /RIMP2/aug-cc-pVDZ 0.00 0.67 1.83 0.39 1.83 1.91 1.20 3.04 2.11 3.19 3.30CCSDsTd /6-31+Gsdd / /RIMP2/aug-cc-pVDZ 0.00 0.54 1.48 0.36 1.48 1.54 1.09 2.60 1.67 2.76 3.11

aThe theoretical method specified to the left of the double slash was used to claculate the energies and that to the right of the double slash was used to optimizethe geometries.

FIG. 5. Energies of the various conformers calcu-lated at the different levels of theory, with con-former I being chosen as the zero of energy. The resultslabeled CCSDsTd/aug-cc-pVDZ were estimatedusing: EfCCSDsTd /aug-cc-pVDZg<EfCCSDsTd /6-31+Gsddg−EfMP2/6-31+Gsddg+EfMP2/aug-cc-pVDZg.



the vast majority of the populations of the conformers havebeen collapsed into their vibrational zero-point levels. Figure6 proves that this is the case. There, a LIF excitation spec-trum of TRA recorded at the point where SEP excitationoccurssx/D=2d is compared with a downstream spectrum atthe probe positionsx/D=6d. Apart from the broader rota-

tional band contours atx/D=2, hot bands are very weak.The band at 34 911 cm−1, marked with an asterisk, is av9=1−v8=0 hot band of conformerA. By comparing the in-tensity of this hot band with the correspondingv9=0→v8=1 vibronic band of TRAsAd, we estimate that greater than95% of the population atx/D=2 is in the ground-state zero-point level.

B. Single conformation SEP spectra

SEP scans of TRAsAd ,sBd, and sCd are shown in Figs.7sad–7scd, respectively, acquired with the pump laser fixed onthe selectedS1←S0 origin in Fig. 2sad while the dump laserwas tuned over the range from 440 to 1580 cm−1 above thezero-point level in the ground state. SEP scans over this en-ergy region produced vibrationally excited TRA conformerswith energies ranging from below to well above the thresh-olds to isomerization. The SEP scan shows a dense vibronicstructure throughout this energy range, allowing narrowbounds to be placed on the isomerization barriers. The strik-ing similarities between the spectra are due to the Franck–Condon activity isolated primarily in indole ring vibrations,which are effected only slightly by the conformation of theethylamine side chain. The spectrum of bandC fFig. 7scdg isnot, in fact, a single conformation spectrum, but rather acomposite of the two unresolved transitionsCs1d andCs2d inthe origin regionfFig. 2sadg.

TABLE III. Relative energiesskcal/mold of the transition states of tryptamine.

B3LYP/6-31+Gsdd / /B3LYP/6-31+Gsdd RIMP2/aug-cc-pvDZ/ /B3LYP/6-31+Gsdd

Transition stateFrom lowest-energy

minimaFrom higher-energy

minima Relative toAFrom lower-energy

minimaFrom higher-energy

minima Relative toA

AaBa 3.02 2.76 3.02 3.03 2.40 3.03AaDa 4.11 3.38 4.11 5.05 3.47 5.05AaJa 6.34 3.68 6.34 6.16 2.88 6.16

AC2b8 6.00 5.46 6.00 5.94 5.69 5.94BaEa 3.94 3.61 4.20 4.72 3.66 5.34BaFa 4.43 3.85 4.69 4.19 3.80 4.82

C1aC2a 3.73 3.53 4.27 4.85 3.52 5.10C1aIb 1.01 0.29 1.75 1.28 0.35 2.86C2aFa 3.33 3.03 3.87 3.49 2.73 3.74C2aJa 2.64 0.51 3.18 3.44 0.41 3.70C1aDa 1.53 1.53 2.27 1.68 1.68 3.27DaEa 2.54 2.40 3.13 2.39 2.29 3.97DaIa 4.02 3.30 4.75 3.59 2.67 5.18DaJa 2.78 0.86 3.52 2.64 0.94 4.22Ea8Eb 4.12 4.12 4.71 3.66 3.66 5.33EaFa 3.62 3.37 4.21 4.32 3.66 5.34FaJa 4.38 2.56 5.22 4.87 2.60 5.88EaG 0.95 0.36 1.54 1.18 0.44 2.86GIB 2.57 2.30 3.75 2.35 2.25 4.76

HaJb8 4.90 3.42 6.08 3.96 2.61 5.89BbFa8 5.59 5.01 5.85 5.33 4.93 5.95AaHa 2.42 1.25 2.42 3.01 1.08 3.01BaHa 3.58 2.67 3.84 3.29 1.99 3.92

C2aHa 6.10 5.46 6.63 6.39 4.72 6.65C1aHa 3.33 2.89 4.07 3.56 3.22 5.15C1aEa 2.46 2.31 3.05 2.32 2.23 3.91

FIG. 6. sad LIF excitation spectra taken “upstream” atx/D=2. sbd LIFexcitation spectrum taken “downstream” atx/D=6. The asterisks* d marks ahot band of conformerA at 34911 cm−1 due to incomplete cooling.



C. SEP-PT spectra

SEP-induced population transfer spectrum obtained bycarrying out SEP excitation on TRAsAd while monitoringtransitionsB,D ,E, andF with the probe laser are shown inFigs. 8sbd–8sed, respectively. The abscissa in these scans isthe difference in the wave number between the pump anddump lasers as the dump laser is scanned, which is equiva-lent to the initial internal energy of conformerA. Positive-going signals indicate an increase in the population of theindicated conformer downstream in the expansion. A com-parison of the population transfer spectra with the SEP spec-trum in Fig. 8sad shows a sudden turn on of each spectrum ata particular threshold energy that is unique to each con-former.

The analogous scan monitoring transitionC in the LIFspectrum with the probe laserfFig. 9scdg contains contribu-tions from both theA→Cs1d andA→Cs2d conformer pairs.This spectrum has a threshold at 1000 cm−1, but it is impos-sible to know from this spectrum which conformer is respon-sible for the lowest-energy threshold, nor where the secondthreshold is to be found. Despite their overlapping origintransitions, conformersCs1d andCs2d have slightly differentvibronic transitions 413 and 422 cm−1 above theC originfFig. 9sadg, which can be used to separate out the thresholdsto isomerization for the two conformers.3 The populationtransfer spectra taken while monitoring these nonoverlappedtransitions determine the threshold forCs1d at 1316 cm−1

fFig. 9sddg, while that for Cs2d occurs at 1000 cm−1

fFig. 9sedg.These spectra illustrate several important features of the

experimental method. First, the obvious primary results to bederived from the data in Figs. 8 and 9 are the clear energythresholds for isomerizing conformerA fGpysoutdg to all six

product wells. These thresholds for isomerization are locatedat 750 sA→Bd, 1316 fA→Cs1dg, 1000 fA→Cs2dg, 1280sA→Dd, 1316 sA→Ed, and 750 cm−1 sA→Fd. Second,above threshold, the transitions in the PT spectra have rela-tive intensities commensurate with the corresponding transi-tions in the SEP spectra ofA. This argues against any mode-specific effects in the isomerization process. Third, a lowerbound on each threshold forA→X isomerization can be de-termined by identifying the last transition in the SEP spec-trum of A that should have been seen in theA→X PT spec-trum. It is likely that the upper and lower bounds on theenergy threshold are bounds on the classical barrier heightseparating the particularA→X reactant-product pair. How-ever, we need to take into account the possible effects onthese bounds of tunnelingsproducing too low an upperbound to the barrierd or a kinetic shiftsproducing too high alower boundd. These effects will be considered in the Dis-cussion section. Finally, these data raise the prospect of mea-suring similar thresholds for all 42X→Y product pairs inTRA, leading to a complete characterization of all the sta-tionary points of consequence on the potential-energy sur-face for TRA. Not surprisingly, such a complete character-ization is not possible. The spectra in Figs. 8 and 9 involveobserving the population transfer from a large populationconformersAd to small population conformerssB–Fd. Thereverse process, namely, pumping a small population con-

FIG. 7. SEP spectra of conformerssad A, sbd B, and scd C. The strikingsimilarities among the three spectra are due to the Franck–CondonsFCdfactors being localized on the aromatic ring.

FIG. 8. sad SEP spectra ofA. sb–ed SEP-PT spectra recorded by monitoringthe S1←S0 origin transition of the corresponding conformer.



former into a large population well, becomes increasinglyhard to observe as the population in the initial conformerwell decreases in size.

Despite this difficulty, the thresholds for several otherX→Y product pairs were measured. SEP-induced populationtransfer spectra pumping TRAsBd monitoring transitionsA,F, andC with the probe laser are shown in Figs. 10sbd and10scd, respectively. Both TRACs1d and Cs2d are beingprobed when monitoring transitionC in the LIF spectrumof Fig. 2sad. However, based on the lower threshold forA→Cs2d thanA→Cs1d, we tentatively assign this thresholdto theB→Cs2d process. Similarly, the threshold for pumpingconformersCs1d /Cs2d fwhetherCs1d or Cs2d is not knowngback to B was also measured at 750 cm−1, with the likelyX→Y pair being Cs2d→B. Due to signal-to-noise con-straints, only in theB→F case was it possible to establish aclear lower bound for the threshold.

The combined experimental data on the thresholds forisomerization are summarized on a 2D potential-energy sur-face in Fig. 11. The comparison with theorysFig. 3d will betaken up in the Discussion section.

D. SEP–hole-filling spectra

An interesting consequence of these three sharp thresh-olds is that they can be used to exert some control overwhich isomerization products are formed by the SEP step.This is best observed in a hole-filling spectrum, in which theSEP laser wavelengths are fixed to selectively excite a par-ticular conformer to a well-defined vibrational energy whilethe probe laser is scanned. As Fig. 12sad shows, when thedump laser is fixed at a wavelength corresponding to an en-ergy of 750 cm−1 above the zero-point level ofA, only con-formersB andF are formed as products. At 1219 cm−1 fFig.12sbdg, conformer Cs2d appears in the product spectrum,while at 1411 cm−1 fFig. 12scdg, Cs1d ,D, andE are added aswell. In Fig. 12, the population gain in the hot band ofAresults from incomplete cooling of a small fraction of thepopulation ofA during the recooling step.

Because the scans in Fig. 12 are taken with a20/10/20-Hz configuration, the SEP hole-filling signal intransitionA reflects the population difference between popu-lation reaching the ground state via fluorescence versus SEP.The fluorescing molecules produce a ground-state populationdictated by the Franck–Condon factors from the fluorescinglevel, while SEP produces a population only in a single level.When SEP is to a low-energy state, SEP will be more effi-cient than fluorescence at recooling into the reactant well,producing a gain on transitionA, as is observed in Fig. 12sad.When the dump laser reaches to levels with higher internalenergyfFigs. 12sbd and 12scdg, SEP will be more efficient atisomerization than fluorescence, leading to a depletion ontransitionA.

FIG. 9. sad An overview LIF spectrum of tryptamine with the expandedregion corresponding to vibronic transitions ofCs1d and Cs2d. Cs1d andCs2d have vibronic bands at 413 and 422 cm−1, respectively, which wereused to probe the energy thresholds free of interference from one another.The sbd SEP spectra ofA is compared to the SEP-PT spectra ofscd A→Cs1d /Cs2d, sdd A→Cs1d, andsed A→Cs2d.

FIG. 10. Thesad SEP spectrum ofB compared to the SEP-PT spectra ofsbd B→A, scd B→F, andsdd B→C.



E. Studies of deuterated TRA

Two of the four flexible coordinates in TRA involve in-ternal rotation and inversion of the NH2 group. These mo-tions are implicated in the high-resolution studies of TRA byNguyenet al.,22 which have revealed small tunneling split-tings s95 MHzd in the origin bands ofCs1d and D. In thepresent studies of isomerization dynamics, tunneling throughthe barrier could produce an energy threshold below the clas-sical barrier height. One way to test for the effects of tunnel-ing on the isomerization is to compare the isomerizationthresholds for the NH2 and ND2 isotopomers of TRA.

TRA has three exchangeable hydrogens: the two aminohydrogens and the indole NH, with the former more easilyexchanged than the latter. Despite repeated exchange cycles,complete deuteration at all three sites was never achieved,probably due to exchange back in the sample compartment.The degree of isotopic substitution was assessed using reso-nant two-photon ionization spectroscopysR2PId. Figure 13shows the R2PI spectra taken while monitoring the four masschannels of the protonated, singly, doubly, and triply deuter-ated TRA. Based on the intensities of the origin transitions,more than 90% of the sample is either doubly or triply deu-terated, with near-complete deuteration at the ND2 site an-ticipated. Furthermore, by judicious choice of the SEP pumpwavelength on the blue edge of the triply deuterated origins34 927 cm−1, marked in the figured, selective excitation ofTRAsd3d was further enhanced.

Figure 14sad shows the SEP spectrum of conformerA ofTRAsd3d with a pump wavelength set at 34 927 cm−1. Notsurprisingly, the spectrum bears a close resemblance to thespectrum of TRAsA,h3d fFigs. 14sad and 14sddg, with smallshifts in the frequencies of most bands. These similaritiesmake it possible to determine the threshold to isomerizationwith comparable accuracies on the upper and lower bounds.SEP-PT spectra were recorded for three reactant-productpairs:A→B,A→F, andB→F. The A→B pair was chosen

because conformersA andB differ primarily in the orienta-tion of the NH2 group, suggesting that tunneling might playa role in the isomerization pathway connecting them. Theother two pairs haveF as their product, and therefore repre-sent pathways where the barriers computed on the 2D poten-tial energy surfacesPESd s,1300 cm−1, Fig. 3d are signifi-cantly higher than the experiments,750 cm−1, Fig. 11d.

The SEP-PT spectrum of TRAsA,d3d→TRAsB,d3d ex-hibits a threshold to isomerizationf742 cm−1, Fig. 14sbdg thatis identical to the measured threshold in the nondeuteratedsystemf750 cm−1, Fig. 8sbdg. Therefore, under the conditionsof this experiment, tunneling does not shift the threshold forisomerization of TRAA→B.

The corresponding PT scan forA→F is shown in Fig.14scd. A comparison of this scan with the undeuterated ana-

FIG. 11. Schematic PES for TRA along two flexible internal coordinates.The arrows signify all of the conformational isomerization reactant-productpairs probed experimentally. The numbers associated with the arrows denotethe lower and upper bounds to the energy threshold.

FIG. 12. SEP-HF spectra after selective excitation of conformerA to vibra-tional levels with energies ofsad 748, sbd 1219, andscd 1411 cm−1.

FIG. 13. The mass-selected R2PI excitation spectra of deuterated tryptaminein the sad triply deuterated mass channel,sbd doubly deuterated mass chan-nel, scd singly deuterated mass channel, andsdd the undeuterated tryptamine.The S1←S0 origin transition of tryptamineA is 34 920 cm−1 and the exci-tation wavelength for the deuterated tryptamineA is marked at 34 927 toensure no interference from the other species in the LIF excitation scheme.



log fFig. 14sddg indicates that the threshold for thed3 isotopeis at 832 cm−1, 90 cm−1 above the threshold for theh3 iso-tope. The triad of bands near 750 cm−1 is clearly missingdespite its large intensity in the SEP spectrum. This result issurprising because conformersA fGpysoutdg andF fGphsupdgdiffer in both the orientation and position of the NH2 grouprelative to the indole ring. Nevertheless, there is a clear shiftin the threshold towards higher values upon deuteration, con-sistent with tunneling as the source of the shift.

The corresponding results forB→F fFig. 15g show asimilar shift, but are less clear due to a somewhat poorersignal-to-noise ratio in this case. The threshold for TRAsd3dis at 737 cm−1, with no evidence for the transition at672 cm−1 fFig. 15sddg. The corresponding band at 685 cm−1

in the spectrum of TRAsh3,B→Fd is observed, but its inten-sity is sufficiently small that its absence in the TRAsd3d scanmay be simply because it is hidden in the noisefFig. 15sbdg.

V. DISCUSSION

A. SEP as a vibrational excitation scheme

The experimental protocol employed in this work usesstimulated emission pumping to initiate conformationalisomerization in a single conformational isomer with a well-defined internal energy. SEP excitation is an important exten-sion of an earlier work that used infrared excitation in thehydride stretch region of the infrared.11–13 As noted earlier,

the principle drawback ofXH stretch infrared excitation isthat many of the lowest-energy barriers to isomerization arewell below the typicalXH stretchv=1 level s,3500 cm−1

=10 kcal/mold. As such, near-IR excitation never directlyprobes the barriers to isomerization in the threshold region.By comparison, the principle advantage of SEP excitation isthat it offers a wide tuning range that stretches from wellbelow the lowest barriers to isomerization up to energieswell in excess of these thresholds.

As a result, by observing the onset of a particularX→Y population transfer, SEP population transfer spectra pro-vide direct experimental measurements of the energy thresh-olds for isomerization in individualX→Y reactant-productpairs. Furthermore, in cases where the thresholds for bothX→Y andY→X pairs can be measured, their difference con-stitutes a measure of the relative energies of minimaX andY.Thus, in principle, step-by-step characterization of allX→Y thresholds constitutes a complete characterization of therate-limiting stationary points on the potential-energy surfacefor isomerization.

SEP excitation also offers exceptional selectivity in theexcitation step. Any vibronic transition that is not overlappedin the ultraviolet spectrum can be used to selectively excite aparticular conformation. With supersonic expansion cooling,vibronic bands typically have rotational band contours withwidths of 1–2 cm−1, so different conformers need only shift

FIG. 14. The structures indicate the conformationalisomerization and deuterated sites. The SEP spectrumof sad tryptamine is compared tosdd tryptamine-sd3d.The SEP-PT spectra ofsbd A→B and scd A→F areshown for comparison tosed Ad→Bd and sfd Ad→Fd,respectively. Only the threshold to isomerization ofAd→Fd has any indication of tunneling effectss,100 cm−1d.



the electronic origin by this amount in order to enable selec-tive excitation, a very small fraction of the excitation energys,30 000 cm−1d.

Finally, SEP offers the versatility of carrying out SEPusing vibronic bands above the origin in the pump step. Inprinciple, selection of differentS1svd levels offers a powerfulmeans of changing the Franck–Condon intensities to theground state in the dump step. These changed Franck–Condon factors could be used to narrow the bounds on thethreshold for isomerization. This potential advantage is moreone in theory than practice and, in fact, was not utilized inthe present experiment. If intramolecular vibrational redistri-bution sIVRd is fast compared to the time delay betweenpump and dump, then the oscillator strength in the dumptransition is spread over many transitions that reduce the SEPintensity in any one band considerably. In TRA, we were notable to carry out SEP effectively on bands more than200 cm−1 above theS1 origin.

The principle disadvantage of SEP in the context of thisexperimental protocol is that it turns a two-laser IRpump/UV probe experiment into a three-laser experimentsSEP pump-dump, UV probed. Furthermore, in order to usethis scheme, the molecule of interest must have an ultravioletchromophore capable of efficient SEP, with a sufficientlylong lifetime to stimulate a significant fraction of the excited-state population back down to the ground state. With thenanosecond lasers employed here, this requires excited-statelifetimes of several nanoseconds. The tryptamine conformershave excited-state lifetimes of 15 ns.18,22 In addition, themethod is not easily adapted to the quantitative measurementof product quantum yields11,13 because the SEP pump stepremoves population that is not entirely brought back to theground state by the dump step.

B. The comparison between experimentand calculations

In this section, we compare the experimentally deter-mined energy thresholds for isomerizationsFig. 10d withcomputed barrier heights, corrected for zero-point energy ef-fects sFig. 3d. This comparison is justified provided that theexperimental upper and lower bounds bracket the computedclassical barrier height. In order for the upper bound on thethreshold to be an upper bound on the classical barrierheight, tunneling must not play a significant role. The experi-mental lower bound is a firm lower bound only if the isomer-ization rate at threshold is fast compared to the vibrationalcooling rate, producing a negligible kinetic shift in thethreshold.

The measurements on TRA show no evidence for a sig-nificant kinetic shift. If vibrational cooling were competingeffectively with isomerization, then the bands near thresholdin the PT spectra should show a reduced intensityscomparedto the SEP spectrumd that depends on the cooling conditions.In TRA, the intensities of transitions above threshold in thePT spectra faithfully reflect those in the SEP spectra, consis-tent with isomerization occurring on a time scale fast com-pared to cooling. This is consistent with Rice–Ramsperger–Kassel–MarcussRRKMd estimates for the rate constants for

isomerization s.13109 s−1 at thresholdd which are fastcompared to the anticipated cooling rate by heliums2–3 cm−1/collision at a collision rate of 43109 s−1d. Amore detailed discussion of the competition between isomer-ization and collisional cooling will be taken up in the adjoin-ing paper on 3-indole propionic acidsIPAd.4 For our pur-poses here, we will use the lower bounds from the SEPmeasurements as the lower bounds on the energy barriers,assuming that the kinetic shifts are negligible.

The data on TRAsd3d sSec. IV Ed indicate that tunnelingdoes play a role in isomerization in certain regions of thepotential-energy surface; notably, in pathways leading intowell F. Recall that the threshold forA→B isomerization wasunchanged by deuteration at the amino group, while thethresholds for theA→F andB→F channels were raised bysmall but measurable amountss,100 cm−1d. Because thisshift is small, it could arise from some other effect thantunneling; perhaps due to zero-point energysZPEd shifts inthe threshold or an increased kinetic shift to the measure-ment. However, the computed shift in the barrier height dueto ZPE differences between TRAsh3d and TRAsd3d is lessthan 10 cm−1. Furthermore, the computed RRKM rate con-stants at threshold for isomerization of TRAsd3d are onlyreduced by about 20% by deuteration, and are still largecompared to the rate for cooling. Therefore, tunneling is themost likely explanation for the observed shift in theA→Fand B→F isomerization thresholds in TRAsd3d. The rela-tionship between the observed thresholds forA→F and B→F and the classical barrier height depends on the tunnelingpathway, an issue to which we shall return shortly. In othermeasuredX→Y thresholds, we assume that tunneling playsan insignificant role, and treat the observed energy thresholdas an upper bound to the classical barrier height.

1. Relative energies of the minima

A first test of the calculated potential-energy surface is tocompare the relative energies of the minima with the ob-served relative populations of the conformers in the expan-sion in the absence of SEP excitation. One must be carefulnot to put too much weight on these populations reflectingenergy differences alone, because it is possible that the cool-ing in the expansion can remove population out of thehigher-energy minima into the lower ones if the barriers arenot too great. In the case of TRA, we do not observe signifi-cant changes in these relative populations as a function ofdistance downstream in the expansionsFig. 6d, and thereforesee some merit in correlating the relative intensities of theobservedS0-S1 origins with their relative energies. The high-resolution spectra of Nguyenet al. show that the intensity oftheCs2d origin is about twice that of theCs1d origin. Sortingthese intensities from large to small yields,IsAd. IsBd. IfCs2dg. hIfCs1dg ,IsDd ,IsEd , andIsFdj, where the bandsin brackets have similar intensities. This suggests that theexperimental energy ordering is

EsAd , EsBd , EfCs2dg

, hEfCs1dg,EsDd,EsEd, andEsFdj.

This energy ordering matches reasonably well with the DFT



energies, while the RIMP2 results would predict greaterpopulations inCs2d andF than observed experimentally. It ispossible that the low experimental barriers separating theseminima from B sFig. 11d provide a pathway for removingpopulation from these minima intoB early in the expansionsx/D,2d.

2. Barrier heights

For most of the measuredA→X thresholds, the calcu-lated barrier heights, including ZPE corrections, are within acouple of hundred wave numbers of experiment. For in-stance, the experimental barriers separatingA fGpysoutdgfrom any of the three anticonformersfCs1d ,D and Eg areEthresh,1300 cm−1=3.7 kcal/mol, consistent with the com-puted barrier heights forAD andBE s,1300 cm−1 for DFTand 1600–1700 cm−1 for RIMP2d. The A→B threshold issomewhat lower experimentallys688,Ethresh,748 cm−1dthan predicted by theorysEbarrier=930 cm−1 by both meth-odsd, but still within this 200 cm−1 spread.

The glaring exceptions to this general correspondencebetween experiment and theory are theA→F and B→Fthresholds. As Fig. 3 shows, the computed barriers into wellF are all up in the 1300–1500-cm−1 range, almost a factor of2 higher than the experiment. These are the same pathwaysthat show some evidence for tunneling, and it is possible thattunneling accounts for the entire discrepancy between ex-periment and theory. However, the small shift in thresholdinduced by deuteration seems to argue against such a simpleresolution. Furthermore, sinceA and F differ along at leasttwo of the flexible coordinates, one would like to pin downthe isomerization pathway, whether tunneling or classical innature.

3. Isomerization pathways

The experimental thresholdssFig. 15d that connectA→B,A→F, andB→F are particularly low, suggesting thatthere is an efficient isomerization pathway connecting theseminima. As just noted, this pathway may involve tunneling,but if so, the tunneling penetrates some of the highest com-puted barriers on the surface, those that surround minimumF. As noted in the initial communication, the observedthresholds are consistent with a pathway fromA→Fs688,Ethresh,748 cm−1d that passes fromA→Bs688,Ethresh

,748 cm−1d and then fromB→Fs566,Ethresh,688 cm−1d.The computed potential-energy surface possesses a low-

energy trough that connects wellsA,H, andCs2d via theAHand HCs2d barriers. A possible way to resolve these differ-ences would be to simply swap the assignments of conform-ers Cs2d and F. These two conformers are both Gph con-formers, differing only in the orientation of the NH2 group.As a result, they have rotational constants which are veryclose to one another. If this swap in assignments were made,the calculations would predict anA→F barrier of 672 cm−1

based on a pathway that goes from Gpysoutd to GpysindsEbarrier=672 cm−1d and then on to Gphsoutd sEbarrier

=527 cm−1d. While such a swap is tempting, the argumentsthat Nguyenet al. used to arrive at the assignments forCs2d

andF are quite persuasive, and are consistent with previousassignments based on infrared and vibronic level data.22

With or without such a swap in assignments, it is sur-prising that tunneling would lead to a shift in theA→F andB→F thresholds, but produce no shift inA→B, since thepathways fromA or B into F necessarily involve motion ofthe entire NH2 group from the Gpy to the Gph positionswithan effective tunneling mass of,16 amud, while A→Bisomerization fGpysoutd→Gpysupdg can be accomplishedmerely by an internal rotation and/or inversion of the NH2

group s,2-amu tunneling massd. Of course, this differencein mass of the tunneling group could be compensated for bya narrower width to the barrier. Since the likely isomeriza-tion pathway for eitherA→F or B→F involves motion ofthe NH2 group over the top of the phenyl ring, one wonderswhether the interaction of the NH2 group with the indolepcloud could provide a low-energy tunneling pathway thatinverts the NH2 group as it sweeps across thep cloud. Afinal resolution of these issues will require theoretical meth-ods that can directly model the tunneling contribution to theisomerization dynamics.

Finally, even in a molecule the size of tryptamine, we arequickly losing our ability to use chemical intuition to identifythe important isomerization pathways on the potential-energysurface. The transition states identified in this work are likelyonly a subset of those needed to fully describe the isomer-ization dynamics. In particular, isomerization pathways thatinvolve swinging the ethylamine side chain from one side ofthe indole ring to the other may play a significant role in the

FIG. 15. sad SEP spectrum ofB compared toscd the SEP spectrum ofBd.The SEP-PT spectra ofsbd B→F exhibits a measured threshold almostidentical to thesdd SEP-PT ofBd→Fd. No tunneling effects are observed inthe isomerization of conformersB into F.



isomerization dynamics. The anticonfiguration supportsminimaG andI in which the ethylamine side chain is nearlyin plane. The barriers into and out of these minima are quitelow s,100 cm−1d. Analogous pathways involving motion ofthe ethylamine group from the Gpy or Gph positions aroundthe edges of the pyrrole or phenyl rings deserve further ex-ploration.

VI. CONCLUSIONS

Despite its modest size, tryptamine possesses a potential-energy surface for conformational isomerization that is achallenge to characterize via experiment and theory. Thepopulation of TRA molecules is spread over seven confor-mational minima, even with supersonic expansion cooling. Acomplete characterization of the isomerization process wouldthen require measurement of 42 independentX→Y con-former pairs, illustrating the need for new experimental toolscapable of studying the process in a conformation specificway. The present paper describes measurements of the en-ergy thresholds for isomerization using conformation-specific SEP excitation, followed by collisional cooling ofthe products prior to conformation-specific detection vialaser-induced fluorescence. The method of SEP-inducedpopulation transfer spectroscopy provides a means for mea-suring the energy thresholds for isomerization of individualX→Y conformer pairs free from interference from otherspresent in the expansion.

The application of this method to TRA has demonstratedthe power of the method, but it has also raised many unan-swered questions, providing a stimulus for future work. First,since the method uses collisions as an integral part of theexperimental scheme, isomerization is always viewed incompetition with collisional cooling. Ideally, this competi-tion can be used as a means to quantify the energy-dependentrate of isomerization for individualXsE*d→Y pairs. It ishoped that such studies can provide critical new tests ofRRKM theory, which is undergoing close scrutiny by theory,which suggests that in many circumstances, IVR into thetorsional modes may limit the rate of isomerization nearthreshold. Second, as just discussed in the preceding section,theoretical methods that can explore the isomerization path-ways more completely, and include the effects of tunneling,are needed. Third, it is important to compare the results ofthe present study with analogous studies of isomerization inthe absence of collisions, using methods such as those em-ployed by Pate and co-workers. Finally, hole-filling methodscan be applied to the study of isomerization in a much widerrange of contexts. In the adjoining paper, we will probe theeffect of a bound solvent water molecule on the barrier tointramolecular isomerization in 3-indole propionic acid.23

ACKNOWLEDGMENTS

We gratefully acknowledge support for the work fromthe National Science FoundationsGrant No. CHE-0242818d.

One of the authorssJ.R.C.d acknowledges partial supportfrom the Purdue Research Foundation for a Graduate StudentAssistantship.

1E. G. Robertson and J. P. Simons, Phys. Chem. Chem. Phys.3, 1 s2001d.2T. S. Zwier, J. Phys. Chem. A105, 8827s2001d.3J. R. Carney and T. S. Zwier, J. Phys. Chem. A104, 8677s2000d.4J. R. Carney and T. S. Zwier, Chem. Phys. Lett.341, 77 s2001d.5B. C. Dian, A. Longarte, S. Mercier, D. Evans, D. J. Wales, and T. S.Zwier, J. Chem. Phys.117, 10688s2002d.

6G. M. Florio, R. Christie, K. D. Jordan, and T. S. Zwier, J. Am. Chem.Soc. 124, 10236s2002d.

7W. Chin, M. Mons, J. P. Dognon, F. Piuzzi, B. Tardivel, and I. Dimicoli,Phys. Chem. Chem. Phys.6, 2700s2004d.

8H. Fricke, A. Gerlach, C. Unterberg, P. Rzepecki, T. Schrader, and M.Gerhards, Phys. Chem. Chem. Phys.6, 4636s2004d.

9M. Gerhards, C. Unterberg, A. Gerlach, and A. Jansen, Phys. Chem.Chem. Phys.6, 2682s2004d.

10E. Nir, I. Hunig, K. Kleinermanns, and M. S. de Vries, ChemPhysChem5,131 s2004d.

11B. C. Dian, A. Longarte, P. R. Winter, and T. S. Zwier, J. Chem. Phys.120, 133 s2004d.

12D. Evans, D. Wales, B. C. Dian, and T. S. Zwier, J. Chem. Phys.120, 148s2004d.

13B. C. Dian, G. M. Florio, J. R. Clarkson, A. Longarte, and T. S. Zwier, J.Chem. Phys.120, 9033s2004d.

14B. C. Dian, J. R. Clarkson, and T. S. Zwier, Science303, 1169s2004d.15D. J. Nesbitt, and R. W. Field, J. Phys. Chem.100, 12735s1996d.16G. W. Flynn, C. S. Parmenter, and A. M. Wodtke, J. Phys. Chem.100,

12817s1996d.17M. Silva, R. Jongsma, R. W. Field, and A. M. Wodtke, Annu. Rev. Phys.

Chem. 52, 811 s2001d.18L. L. Connell, T. C. Corcoran, P. W. Joireman, and P. M. Felker, J. Phys.

Chem. 94, 1229s1990d.19Y. D. Park, T. R. Rizzo, L. A. Peteanu, and D. H. Levy, J. Chem. Phys.

84, 6539s1986d.20L. A. Philips and D. H. Levy, J. Chem. Phys.89, 85 s1988d.21Y. R. Wu and D. H. Levy, J. Chem. Phys.91, 5278s1989d.22T. V. Nguyen, T. M. Korter, and D. W. Pratt, J. Mol. Spectrosc.sto be

publishedd.23J. R. Clarkson, E. Baquero, and T. S. Zwier, J. Chem. Phys.122, xxx

s2005d, following paper.24S. H. Kable, and A. E. W. Knight, J. Chem. Phys.93, 3151s1990d.25T. Burgi, T. Droz, and S. Leutwyler, Chem. Phys. Lett.246, 291 s1995d.26T. Burgi, T. Droz, and S. Leutwyler, J. Chem. Phys.103, 7228s1995d.27A. D. Becke, Phys. Rev. A38, 3098s1988d.28C. Lee, W. Yang, and R. G. Parr, Phys. Rev. B37, 785 s1988d.29M. J. Frisch, J. A. Pople, and J. S. Binkley, J. Chem. Phys.80, 3265

s1984d.30R. A. Kendall, T. H. Dunning, and R. J. Harrison, J. Chem. Phys.96, 6796

s1992d.31T. H. Dunning, J. Chem. Phys.90, 1007s1989d.32D. E. Woon, T. H. Dunning, and R. J. Harrison, J. Chem. Phys.96, 1481

s1993d.33C. Peng and H. B. Schegel, Isr. J. Chem.33, 449 s1993d.34M. Feyereisen, G. Fitzgerald, and A. Komornicki, Chem. Phys. Lett.208,

359 s1993d.35D. E. Bernholdt and R. J. Harrison, Chem. Phys. Lett.250, 470 s1996d.36See EPAPS Document No. E-JCPSA6-122-016523 for the complete re-

sults from the calculations including optimized geometric parameters andselected associated harmonic frequencies. This document can be reachedvia a direct link in the online article’s HTML reference section or via theEPAPS homepageshttp://www.aip.org/pubservs/epaps.htmld.

37M. J. Frisch, G. W. Trucks, H. B. Schlegelet al., GAUSSIAN 03, GaussianInc., Pittsburgh, PA, 2003.

38Quantum Chemistry Group, U. o. K., Germany, 1988.39J. Noga and R. J. Bartlett, J. Chem. Phys.86, 7041s1987d.40K. Raghavachari, G. W. Trucks, J. A. Pople, and M. Head-Gordon, Chem.

Phys. Lett. 157, 479 s1989d.41S. F. Boys and F. Bernardi, Mol. Phys.19, 553 s1970d.



direct measurement of the energy thresholds to ... measurement of the energy thresholds to...

Documents