Chemical reactions of conformationally selected molecules in a beam with Coulomb-crystallized ions Daniel Rösch,1 Stefan Willitsch,1,a) Yuan-Pin Chang,2 and Jochen Küpper2,3,4,b) 1)Department of Chemistry, University of Basel, Klingelbergstrasse 80, 4056 Basel, Switzerland 2)Center for Free-Electron Laser Science, DESY, Notkestrasse 85, 22607 Hamburg, Germany 3)Department of Physics, University of Hamburg, Luruper Chaussee 149, 22761 Hamburg, Germany 4)The Hamburg Center for Ultrafast Imaging, University of Hamburg, Luruper Chaussee 149, 22761 Hamburg, Germany (Dated: 27 January 2014) Many molecules exhibit multiple conformers that often easily interconvert under thermal conditions. There- fore,singleconformationsaredifficulttoisolatewhichrendersthestudyoftheirdistinctchemicalreactivities challenging. We have recently reported a new experimental method for the characterization of conformer- specific effects in chemical reactions [Y. P. Chang et al., Science 342, 98 (2013)]. Different conformers are spatially separated using inhomogeneous electric fields and reacted with a Coulomb crystal of cold, spa- tially localized ions in a trap. As a first application, we studied reactions between the two conformers of 4 1 3-aminophenol and Ca+. We observed a twofold larger rate constant for the cis compared to the trans con- 0 former which was rationalized in terms of the differences in the long-range ion-molecule interactions. The 2 presentarticleprovidesadetaileddescriptionofthenewmethodandafullaccountoftheexperimentalresults as well as the accompanying theoretical calculations. n a J PACS numbers: 82.20.-w, 34.50.Lf, 33.15.-e, 82.30.Fi 4 2 I. INTRODUCTION organic molecules.17–19 In these studies, individual con- ] formers of cations were selectively generated capitalizing h on the different ionization energies of their parent neu- p Many molecules possess multiple conformers (rota- trals. In addition, photodissociation studies on neutral - tional structural isomers) that interconvert with low en- m conformers of small organic molecules have been carried ergy barriers through hindered rotations about single out.20–22 Inthesestudies,individualconformerswerenot e covalent bonds. It is well known that different confor- h mations can exhibit distinct chemical reactivities.1–3 To separatedpriortophotoexcitation. Consequently,theex- c perimental results contained contributions from all pop- gain a comprehensive understanding of the chemical be- . s havior of molecules, it is thus necessary to investigate ulated conformations. These were disentangled through c Rydbergtaggingmethodswhichallowedforresolvingthe their conformer-specific chemistry. To this end, individ- i s ualmolecularconformationsneedtobeisolatedandchar- smallenergydifferencebetweenconformersmanifestedin y the kinetic energy release of H photofragments. acterized. Moreover,thecapabilitytomanipulateconfor- h p mational distributions provides means to influence reac- For bimolecular reactions in the gas phase only few [ tivitiesandproducts,addingtotherepertoireofmethods investigations of conformational effects have been re- to control chemical processes. ported so far. Taatjes et al.23 observed conformer- 1 When studies of conformational effects are carried dependentreactivitiesfromthesimplestCriegeeinterme- v 7 out at low temperatures in the gas phase, the ther- diate (CH3OO) which plays a key role in ozonolysis re- 4 mal interconversion of conformations is suppressed. In actions. In a cryogenic matrix, Khriachtchev et al.24 ob- 3 recent years, significant progress toward the spectro- served distinct conformer-dependent products from two 6 scopic characterization of specific conformations has formic acid conformers reacting with oxygen atoms. 1. been achieved.4–13 For instance, studies of photoinduced Several techniques for the manipulation of conforma- 0 ground-state isomerization reactions mapped out inter- tional distributions of molecules in the gas-phase have 4 conversion pathways between conformations, i.e., min- been reported. Ion mobility allows the separation of 1 ima on the corrugated potential energy surfaces.14–16 molecules according to their shape and thus serves as : v Investigations of the conformational landscape of small akeytechniquetoseparateclassesofconformersoflarge Xi molecules, e.g., poly-aminoacids, helped to understand charged molecules.25 For neutral molecules, the spatial the folding motives in peptides.10 separation of specific conformers has been achieved us- r a Unimolecular conformer-specific dynamics have been ingelectricfields.26–28 Whileallconformersofamolecule investigated in the photodissociation of cations of small havethesamemass,theyoftendifferbytheirdipolemo- ments. Different dipole moments lead to different Stark shifts of rotational energy levels in electric fields. Thus, in an inhomogeneous electric field different forces act on a)Electronicmail: [email protected] the different conformers, which can be used for the ma- b)Electronicmail: [email protected] nipulation of their translational motion. 2 Recently,wehaveadaptedthistechniquetoinvestigate Two skimmers with diameters of 2 mm and 1 mm were the chemical reactivities, i.e., rate constants, of specific placed 15 cm and 27 cm downstream from the nozzle, conformers in the prototypical bimolecular reactions of respectively. After skimming, the collimated molecular 3-aminophenol (3AP) and Ca+.29 3AP has two stable beam entered the 15 cm long electrostatic deflector.39–42 conformers (denoted cis and trans) which differ in the A cut through the electrodes of the deflector including a orientation of the OH group and have significantly dif- contourplotofthegeneratedelectricfieldisshowninFig- ferent dipole moments (2.33 D and 0.77 D for the cis- ure 1. The vertical gap between the deflector electrodes and trans-species, respectively). 3AP was entrained in perpendicular to the molecular beam axis was 1.4 mm. a molecular beam and spatially separated using an elec- The shape of the electrodes was designed to generate a trostatic deflector. The dispersed molecular beam was strong inhomogeneous electric field with a nearly con- directed at a stationary reaction target consisting of a stant gradient along the y axis.39,40 The molecular beam Coulomb crystal of Ca+ ions, i.e., an ordered structure passed a third skimmer with a diameter of 1.5 mm for oftranslationallycoldionsatatemperatureofafewmil- differentialpumpingintoachamberpumpedbya345l/s likelvins in a trap.30 Singly ionized Ca ions were chosen turbomolecular pump. Subsequently, the beam entered as a co-reactant because they can easily be Coulomb- thereactionchamber,pumpedbya550l/sturbomolecu- crystallized by laser cooling. Ca+ is also known for its larpump,throughanotherdifferentialpumpingaperture reactivity with organic molecules acting as a catalyst for formed by a 3.5 cm long, 10-mm-diameter tube. Typi- the activation of inert chemical bonds31,32 such as C-F cal pressures during the experiments were 7·10−6 mbar, and C-O.33–35 9·10−8 mbar, and 2·10−9 mbar in the source, deflector In the present article, we give a detailed account of and reaction chambers. our methods and results on the conformer-specific reac- tivities of 3AP with Ca+. The outline of the paper is as follows: section II and section III describe the exper- B. Ion trap setup imental setup and theoretical procedures employed. In section IV, we present a characterization of the electro- In the reaction chamber, Coulomb crystals of laser- static deflection of the conformers, their reaction pro- cooled Ca+ ions were generated and trapped in a lin- files, conformer-specific reaction rate constants as well ear radiofrequency (RF) ion trap.30,37 Ca+ ions were as mass spectra of the reaction products. An analysis produced by non-resonant multi-photon ionization of a of the results based on theoretical calculations follows in beam of Ca atoms evaporated from an oven and passing section V. through the center of the ion trap,30,43 see Figure 1(a). Theiontrapconsistedoffoursegmentedcylindricalelec- trodeswitharadiusr =4.0mmarrangedinaquadrupo- II. EXPERIMENTAL SETUP larconfiguration. Toconfinetheionsintheplaneperpen- diculartothetrapsymmetryaxis, RFvoltageswitham- The experimental setup consists of two main parts: a plitudesV =350VandfrequenciesΩ=2π×3.1MHz 0,RF molecular beam deflection apparatus for the separation wereappliedwithoppositepolaritiesacrossadjacentelec- of3APconformersandaniontrapapparatusforthegen- trodes. To confine the ions along the axis, static volt- eration and storage of Coulomb crystals of laser cooled ages in the range of V = 1–10 V were applied to the end Ca+ ions. The individual experimental procedures have endcap electrodes. The atomic beam was ionized using beenreportedpreviously.27,29,30,36,37 Inthefollowing,we the third harmonic (355 nm) of a Nd:YAG laser close focus on the details of the combined apparatus and the to the center of the ion trap. The Ca+ ions were laser- methodologyforconformer-specificreactionexperiments. cooled with beams produced by two external cavity en- hanced diode lasers operating at wavelengths of 397 nm and 866 nm to pump the (4s) 2S → (4p) 2P and 1/2 1/2 A. Conformer deflection setup (3d) 2D → (4p) 2P transitions, respectively,30 see 3/2 1/2 Figure 1(c). The frequencies of the two laser beams The molecular beam machine for conformer deflection weresimultaneouslymonitoredusinganautomatedfiber- consisted of a series of differentially pumped vacuum switcher coupled to a wavemeter and stabilized by a chambers. The source chamber housing a pulsed valve computer-controlled voltage feedback loop. The result- waspumpedbytwo1650l/sturbomolecularpumps. The inglaserlinewidthswereontheorderofafewMHz. The deflector chamber containing the electrostatic deflector laser powers employed were about 600 µW and 200 µW waspumpedbya500l/sturbomolecularpump,asshown for the 397 nm and 866 nm beams, respectively. Upon in Figure 1(a). A solid sample of 3AP (Sigma-Aldrich, lasercooling,theionslocalizedinspaceandformedthree 98 %) was placed in a reservoir cartridge and vaporized dimensional spheroidal Coulomb crystals30,37 with a ra- at 145 ◦C inside a high-temperature Even-Lavie valve.38 dius r ≈ 200 µm and a width z ≈ 550 µm typically Thevalvewasoperatedatabackingpressureof35barof consisting of ∼700 ions. The secular kinetic energy of neonatarepetitionrateof600Hz. Thetypicalrotational the laser cooled ions amounted to E ≈ k ·10 mK. sec B temperature of 3AP in our experiments was about 1 K. Two-dimensionalcutsofthecentralplaneoftheCoulomb 3 (a) 30 cm 15 cm 82 cm x z turbo pump y turbo pump valve Ca oven D imaging system C C skimmers deflector ion trap turbo pump turbo pump turbo pump 355 nm (b) E / kV cm-1 (c)(4p) 2P 33 115050 (4p) 2P3/2 22 +10 kV 1/2 m 397 nm m11 110000 866 nm /00 866 nm y −1−1 ground 5500 397 nm (3d) 2D 5/2 −2−2−2 −1 0 1 2 (3d) 2D −2 −1 0 1 2 3/2 x / mm (4s) 2S1/2 FIG. 1. (a) Schematic top view of the experimental setup for studying conformer-selected chemical reactions. See text for details. (b) Electric field strength E along a cut through the electrostatic deflector. (c) Diagram of energy levels accessed during Doppler laser cooling of 40Ca+. crystals were imaged by collecting a solid angle of the as a function of time was determined from the crystal atomic fluorescence generated during laser cooling using volumes.44 Note that the 3AP molecules in the reaction an enhanced CCD camera coupled to a microscope with volume were replenished with each gas pulse. Therefore, ten-fold magnification. their number density was essentially constant during the measurement time and the decrease of the number N(t) of Ca+ ions in the crystal as a function of time t fol- lowed pseudo-first-order kinetics. Pseudo-first-order re- C. Reaction rate measurements action rate constants k were determined at specific de- 1 flection voltages and deflection coordinates y according The first step in each reaction experiment consisted of to the rate law the formation of a Coulomb crystal. Subsequently, the molecular beam valve was switched on to admit pulse N(y,t) ln =−k (y)t. (1) trains of deflected 3AP molecules to collide and react N(y,t=0) 1 with the spatially localized ions. Different parts of the deflectedmolecularbeamweredirectedatthestationary The deflection coordinate y is defined as the offset of the Coulomb crystal reaction target by tilting the molecular deflected from the nominally undeflected beam at the beamsetup. Ionsthatreactedwith3APformedproduct position of the Coulomb crystal. All measurements were ions which remained trapped, but were not laser cooled performedwiththesamepoweranddetuningofthecool- and, therefore, did not fluoresce.44 These product ions ing laser from resonance to ensure a constant and well- were sympathetically cooled by the remaining Ca+ ions defined population of all three electronic levels of Ca+ to form a dark shell around the crystal. The progress accessedduringlasercooling(seeFigure1(c)). Thepop- of the reaction was monitored by observing the shrink- ulationsoftherelevantCa+ statesweredeterminedfrom ing of the bright fluorescing Ca+ core of the Coulomb a calibrated eight-level optical Bloch equation treatment crystals as a function of time. Images of the crystals including the effects of magnetic fields.45 were recorded every 30 s with a camera shutter time of Reactions with residual background H gas in the ion 2 0.4 s over reaction times of typically 8 to 15 min. From trap chamber also contributed to a removal of Ca+ ions the recorded images, the number of unreacted Ca+ ions from the trap. The corresponding loss rates were mea- 4 suredforeachsetofexperimentsfollowingthesamepro- III. THEORETICAL AND COMPUTATIONAL METHODS cedures described as above but without admitting the molecularbeam. Theresultingvaluesforthebackground A. DFT calculations of reaction paths on the ground-state lossratesweresubtractedfromthemeasuredratesinthe potential energy surface actualreactionexperiments. Wenotethatcollisionswith the Ne carrier gas of the molecular beam did not lead to any observable loss of Ca+ ions from the trap, as con- Short-range ion-molecule interactions were investi- firmed by control experiments with pure Ne beams. gated computationally using density functional the- ory (DFT) calculations. Stationary points along reac- tion paths to two possible products were computed at the DFT MPW1K/cc-pVTZ level of theory, using the Gaussian 09 software suite.48–50 Transition state struc- tures were calculated by a quadratic-singular-transit ap- proach (QST-3),51,52 from energy-minimized Ca+-3AP D. Mass spectrometry of trapped ions andCa+-product-radicalcomplexesandaninitialtransi- tionstateguess. Toverifyconvergencetoasaddlepoint, theresultingtransitionstatestructurewasdistortedand The ionic reaction products were analyzed using resubmitted as a starting point in a new QST-3 calcula- resonant-excitation (RE) mass spectrometry of the tion for the transition state search. Basis set superposi- Coulomb crystals.37,45 Here, the motion of specific ion tion errors were corrected using the counterpoise routine specieswasresonantlyexcitedbyscanningthefrequency provided in Gaussian 09. of an additional RF drive voltage (0.2–0.3 V) applied to one of the trap electrodes. When the RF field was resonant with the motional frequency of a trapped ion species, the Coulomb crystal heated up. This lead to a dislocation of the Ca+ ions from their equilibrium posi- B. Adiabatic capture theory tion. RE mass spectra were recorded by slowly scanning theexcitationfrequencywhilemonitoringtheincreaseof Long-range ion-molecule capture kinetics were mod- thefluorescenceyieldinaregionclosetobutoutsidethe eled using the adiabatic capture theory developed by normal extent of the Coulomb crystal. RE mass spectra Clary and co-workers.53,54 The long-range interaction of multi-component crystals generally show broad peaks potential V between an ion and a polar molecule was that are shifted with respect to single-species crystals.46 approximated by the sum of the dominant charge- The exact intensity and position of the features depends permanent dipole55 and charge-induced dipole interac- on the scan speed, the drive amplitude, the scan direc- tions tion and the crystal composition. Therefore, RE mass spectrometry only allows an approximate determination qµ cosβ q2α V(R,β)=− D − , (2) ofthemassesofthespeciespresentinamulti-component R2 2R4 Coulomb crystal.45 whereRisthedistancebetweentheionandthecenterof mass of the 3AP molecule, µ is its permanent electric D dipole moment, β the orientation angle of the molecular dipole moment with the ion-molecule axis, q the charge of the ion and α the scalar polarizability of 3AP. Using E. Molecular beam profile measurements the methods described in ref. 54, centrifugally corrected androtationallyadiabaticpotentialenergycurvesforthe system Ca+ + cis-/trans-3AP were calculated for 3AP Spatial deflection profiles of 3AP were recorded in a rotational states with quantum numbers ranging from time-of-flight (TOF) mass spectrometer that replaced j =0 to j =100 for R between 2 and 48 a . The dipole 0 the ion-trap apparatus. 3AP molecules were ionized moments of 3AP were taken from ref. 56, the isotropic via resonance-enhanced two-photon ionization (R2PI) polarizabilities were calculated at the DFT B3LYP/aug- by a frequency-doubled pulsed dye laser pumped by a cc-pVTZ level of theory. Rotational-state-specific reac- Nd:YAG laser with a repetition rate of 20 Hz. Pulses of tioncrosssectionsforj =0uptoj =15werecalculated 10 ns duration with an energy of approximately 0.4 mJ from a summation over all partial waves for which the were focused to a spot size of 240 µm in the interac- maximum of the centrifugally corrected potential energy tion volume. The molecular ions were mass-selectively curve did not exceed the experimental collision energy. detected by their arrival time on a multi-channel-plate Effective capture rate constants were calculated by mul- (MCP) detector. cis- and trans-3AP were differenti- tiplyingthestate-specificcrosssectionswiththevelocity ated through their distinct excitation wavenumbers of andtherelevantstatepopulationsattherotationaltem- 34109 cm−1 and 34467 cm−1, respectively.47 perature of 3AP in the molecular beam. 5 1 approach, andevery individualmoleculewaspropagated (b) -1m 0 2 through a simulated beamline that includes all mechani- nergy W / c---321 µ / Deff10 ccouafllTamahtpoeedleesrcptfuruaolrtemeisasltaodhtfeetflahseiecnrtogeitlxoaenp-tqeiporurniamoanfiletlnuettmeIam(l-yspst,eeaTtrtuareotptud.)refefloerTcartoinotnewnpasresomficblaelles- E-4 (a) Is(y) using -5 -1 20 60 100 140 20 60 100 140 E / kV/cm E / kV/cm 1 (cid:88)N I(y,T )= w (T )I (y). (3) FIG. 2. (a) Stark energies W for the lowest rotational quan- rot w s rot s s=1 tum states j = 0–2 of (blue) cis- and (red) trans-3AP as a function of the electric field strength E. (b) Effective dipole Here,N isthenumberofquantumstatesincludedinthe momentsµeffforj =0−2ofcis-(blue)andtrans-(red)3AP. simulation and ws(Trot)=gMgnse(W0−Ws)/(kBTrot) is the population weight for a given quantum state. W is the 0 field-free energy of the ground state and W the field- s C. Molecular-dynamics simulations of Coulomb crystals free energy of state s. g = 1 for M = 0 and g = 2 M M otherwise. g accountsfornuclearspinstatisticalweight ns Fluorescence images and RE mass spectra of the ofthecurrentstatewithgns =1forallrotationalstatesof multi-component Coulomb crystals were simulated us- (cid:80)N 3AP. Thenormalizationconstantisgivenbyw = w . ingmoleculardynamics(MD)methods. MDsimulations s s=1 wereperformedusingamodifiedversionoftheProtomol program package.57 Fluorescence images were simulated from ion trajectories calculated by solving the classical IV. EXPERIMENTAL RESULTS three-dimensional equations of motion of the ions in the trap under the influence of laser cooling.37,58 To mini- A. Deflection curves of 3AP mizecomputertime,anisotropicfrictionforcetoemulate laser cooling and the pseudopotential approximation for Thedensityofeachconformerinthedeflectedanddis- theiontrapwasused.30 REmassspectraweresimulated persed molecular beam was measured by recording the following the methods described in Ref. 45, 46. Briefly, numberofR2PI-ionizedcis andtrans conformersof3AP the Coulomb crystals were offset from the central trap as a function of the deflection coordinate y. The mea- axis by 20 µm at the beginning of the simulation and suredconformer-selectivedeflectionprofilesareshownin allowed to relax. The Fourier transform of the time- Figure 3, in which each data point represents the sig- dependent total kinetic energy of the ions yielded the nal averaged over 1000 laser shots. When high voltages frequency spectrum of the Coulomb crystal which is also were applied to the deflector, both conformers were de- the experimental observable. The frequencies obtained flected upwards. The deflection was considerably larger by this method were calibrated using a comparison of a for the more polar cis-3AP. For instance, for a deflec- measured and calculated RE mass spectrum of a pure tor voltage of 7.5 kV above y = 6 mm a pure sample of Ca+ crystal. A correction factor of 0.92 was applied to cis conformers was obtained (see Figure 3(a)). The in- the calculated frequencies of each spectrum to achieve sets in Figure 3 show that the fraction of cis-3AP in the optimal agreement with the experiment. probed sample can be continuously tuned as a function ofy. Atheightsabovethecut-offofthetrans-3APbeam profile,thedensityofthecis conformersisstillcompara- D. Monte-Carlo simulations of molecular beam profiles ble to its density in the free jet, i.e., it is only decreased to one fourth. When increasing the voltages to 13 kV The simulation of spatial deflection profiles has been (Figure 3(d)), cis-3AP was deflected so strongly that it described in detail previously.36,59 Briefly, the electric was essentially depleted from the detection region. As a fieldE anditsgradient(∇(cid:126)E)werecalculatedusingfinite consequence, an almost clean sample of trans-3AP was element methods implemented in the COMSOL Multi- obtained. physics program. Stark energy curves W(E) of the 3AP Monte Carlo simulations of the deflection curves are quantum states and their effective dipole moments µ shown as solid lines in Figure 3. The simulations at an eff were calculated using CMIstark60 (see Figure 2). From initialrotationaltemperatureof1.1Kagreewellwiththe the electric fields and Stark energy curves, the molecu- experimental profiles. In particular, the fractional inten- lar beam deflection profiles were calculated with libcold- sities plotted in the insets were reproduced by the simu- mol.36 Trajectories for molecules in individual rotational lations (solid lines). The simulated profiles and the mea- quantum states were obtained by numerical integration sured population ratios of the two conformers were used ofthe3DequationsofmotionusingaRunge-Kuttaalgo- for fitting conformer-specific rate constants from mea- rithm. Theinitialconditionsaccordingtotheparameters sured reaction rate profiles as described in the following of the molecular beam were sampled by a Monte-Carlo sections. 6 1.0 1.0 Ion signal [arb. u.]321(a) Vdefl = 5 kVPurity cis01 0y / m4m 8 321 (b) Vdefl = 7.5Purity cis k01V 0y / m4m 8 Ion signal [arb. u.]0000....2468(a) jjjjj ===== 02468 0000....2468(b) jjjjj ===== 02468 0 0 0.0 0.0 –2 0 2 4 6 8 –2 0 2 4 6 8 –2 0 2 4 6 8 10 –2 0 2 4 6 8 10 0.2 0.2 Ion signal [arb. u.]321 (c) Vdefl = 10 kV 321 (d) Vdefl = 13 Purity transkV01 0y / m4m 8 Ion signal [arb. u.]0.1(c) τττττ ===== --01221 0.1(d) τττττ ===== --01221 0 0 0.0 0.0 –2 0 2 4 6 8 –2 0 2 4 6 8 –2 0 2 4 6 8 10 –2 0 2 4 6 8 10 Deflection coordinate y / mm Deflection coordinate y / mm FIG. 3. Density profiles of the deflected beam of cis (blue) FIG. 4. Deflection profile simulations at V = 7.5 kV for defl andtrans (red)3APatdeflectorvoltagesV =(a)5kV,(b) evenj =0toj =8forspecificrotationalstatesjof(a)trans- defl 7.5 kV, (c) 10 kV and (d) 13 kV. These data were measured and (b) cis-3AP, and for specific asymmetric-top quantum byconformer-specificmulti-photonionizationasafunctionof numbers τ at j = 2 of (c) trans- and (d) cis-3AP. See text the molecular-beam deflection coordinate y. Solid lines: cor- for details. respondingMonte-Carlotrajectorysimulations. Intheinsets, purities of (b) cis and (d) trans conformers are given, as ob- tained by dividing the relevant conformer density profile by mined. The ratio of the experimentally determined first thesumofthecis andtrans profiles. Errorbarsindicatethe order rate constant to the known second order rate con- statistical 95% confidence interval of the data points. stant was equivalent to the time-averaged number den- sity n . For k =4.30(4)×10−3 s−1, measured with a avg 1 beam of 50 mbar of N O seeded in 30 bar of Ne we ob- 2 ForthesimulateddeflectionprofilesshowninFigure3, tainedn (N O)=3.84(62)×107 cm−3. Assumingthat avg 2 different rotational states j have different spatial distri- the number densities in the beam were proportional to butions according to their different effective dipole mo- thepartialpressuresbeforeexpansion(for3APat145◦C ments µeff. Figure 4 shows the simulated deflection pro- approximately 10 mbar [62]), the 3AP density was esti- files of individual states ranging from j = 0 to j = 8. mated to be n =7.7(12)×106 cm−3.63 avg Foreachprofile, thecontributionofallj statesweighed τ with their relative thermal populations at the rotational temperature of 1.1 K and their statistical weights were C. Reaction profiles and conformer-specific rate constants of included. Thus,theareaunderneatheachjprofileinFig- Ca+ +3AP ure 4 represents the relative thermal population of each j manifold at 1.1 K as well as their relative contribu- In Figure 5, the experimentally determined pseudo- tion to the reactions. Comparing the deflection profiles first-order rate constants k are shown as a func- of individual j states for the two conformers, profiles of 1,total tion of deflection coordinate y for four deflector voltages lowj ofcis-3APexhibitsignificantlystrongerspatialde- V =5, 7.5, 10 and 13 kV. Each data point in Figure 5 flection than those of trans-3AP. However, for high j defl represents the mean of at least four individual reaction states both conformers show similar deflection patterns measurements. The measured rate constants k (y) demonstrating the quickly vanishing dipole moment for 1,total reflectboth,thedensitydistributionsofconformersinthe rotationally excited species and the need for very cold deflected molecular beam n and the conformer- molecular beams.28 trans/cis specific second-order rate constants k of the re- 2,trans/cis action: B. Number density of 3AP k (y)=k n (y)+k n (y) (4) 1,total 2,cis cis 2,trans trans The absolute number density of 3AP in the molecu- n (y)andn (y)weredeterminedasdescribedinsec- trans cis lar beam was derived by calibration against the reac- tions IVA and IVB. k and k were determined 2,cis 2,trans tion N O+Ca+ → CaO++N . From measurements of from a global fit of (4) to the reaction-rate profiles in 2 2 thepseudo-first-orderrateconstantsforthisreactionand Figure5. Thefityieldedtheconformer-specificratecon- the reported value for the second-order-rate constant,61 stantsk =2.3(9)×10−10 cm3s−1, k =1.1(4)× 2,cis 2,trans the density of N O molecules in the beam was deter- 10−10cm3s−1andtheratiok /k =2.1(5)within 2 2,cis 2,trans 7 0.8 0.8 3.0 u.] (a) Vdefl = 5 kV (b) Vdefl = 7.5 kV ate [arb. 00..46 00..46 3-1 ms2.0 ction r0.2 0.2 -10 0c a 1 Re / 21.0 0.0 0.0 k –2 0 2 4 6 8 10 –2 0 2 4 6 8 10 0.8 0.8 u.] (c) Vdefl = 10 kV (d) Vdefl = 13 kV 0.0 arb. 0.6 0.6 0.02 0.04Ca+ (4p)0 s.t0a6te popula0ti.o0n8 fraction0.10 0.12 e [ at0.4 0.4 n r FIG. 6. Conformer-averaged bimolecular rate constant k2 as ctio0.2 0.2 a function of the population in the Ca+ (4p) state. Error ea bars represent the statistical 95% confidence interval. The R line represents a linear regression to the data. 0.0 0.0 –2 0 2 4 6 8 10 –2 0 2 4 6 8 10 Deflection coordinate y / mm FIG.5. Reactionprofiles(symbols)andtheirfits(lines)at(a) rate constant for reactions out of the excited (4p) state 5kV,(b)7.5kV,(c)10kVand(d)13kV.Thesolidblacklines was found to be two to three orders of magnitude larger represent the calculated total contributions of both conform- thantheratecoefficientsforreactionsoutofthe(4s)and ers. Dashed lines represent individual contributions of the (3d) states. In the experiments reported in this paper, a trans (red) and cis (blue) conformers. Error bars indicate detuning of 47-56 MHz was used, yielding a population the statistical 95% confidence interval of the data points. of the (4p) level of 5−10%. Because of its large rate constant, this channel dom- inates the reaction rates observed in the experiment a 95% confidence interval. This fit also yielded the ro- and the contribution of the other channels can be ne- tational temperature of the molecules in the beam to be glected in good approximation. Thus, one can assume 1.1 K. that the conformer-specific rate constants determined in section IVC only reflect reactions with Ca+ (4p). Scaled to a state population of 100%, the conformer- D. Variation of Ca+ electronic state populations specific second-order rate-constants k = 3.2(13) × 2,cis 10−9 cm3s−1 and k = 1.5(6)×10−9 cm3s−1 for 2,trans As described in section IIB, Ca+ ions were constantly thereactionofcis-3APandtrans-3AP,respectively,with excited during laser cooling so that collisions occurred Ca+ (4p) were obtained. between 3AP and Ca+ in the (4s) 2S , (3d) 2D and 1/2 3/2 (4p) 2P states. To study the effect of the electronic 1/2 excitation of Ca+ on the reaction rates, we varied the E. Mass spectra of reaction products Ca+ state populations by changing the detuning of the cooling laser beam from the (4s)→(4p) resonance while Figure 7(a) and (b) show RE mass spectra before and optimizing the (3d)→(4p) repumping laser detuning to after a typical reaction, respectively. In the spectrum achieve the best cooling conditions. Figure 6 shows the of the pure Ca+ Coulomb crystal, Figure 7(a), a single measured rate constants as a function of the (4p) state peakatanexcitationfrequencyof140kHzwasobserved. population using an undeflected molecular beam of 3AP ThisfeaturewasalsopresentintheREmassspectrumof molecules. From this set of measurements, the state- themulticomponentcrystalafterthereactionandcould specific rate constants k2 listed in Table I were derived unambiguouslybeassignedtotheexcitationofCa+ ions following the procedures outlined in Refs. 43, 45. The withamassof40u. ThespectrumoftheCoulombcrys- tal after reaction showed two additional strong peaks at 120 and 165 kHz. The feature at lower frequency was TABLEI.Bimolecularrateconstantsk forreactionsofCa+ assigned to product ions. MD simulations for crystals 2 composed of 350 Ca+ and 325 heavier ions indicate that initsrelevantelectronicstateswith3APmoleculesinanun- deflected beam. the product ion mass is in the range of 50 to 60 u, sug- gesting that the reaction products are CaOH+ (57 u) or Reaction channel k2 / cm3 s−1 CaNH+2 (56 u). As discussed in detail in Refs.,45,46 our approximate simulation approach cannot be expected to Ca+ (4s) 2S + 3AP 2.66(44)×10−11 1/2 perfectly reproduce the observed peak positions and in- Ca+ (3d) 2D + 3AP 2.69(45)×10−12 3/2 tensities in the spectra as it is not a faithful representa- Ca+ (4p) 2P1/2 + 3AP 1.91(32)×10−9 tion of the complex processes leading to the signal mea- 8 s] (a) (b) tialenergysurfacesoftheexcitedchannelsisbeyondthe nit0.8 0.8 scope of the present work, the results of the DFT calcu- u b. lations for the ground-state surface can nonetheless give ar0.4 0.4 valuableinsightsintopossibletransition-state(TS)struc- sity [ 0 0 turesandreactionpathways. Figure8showsaschematic en 60 u 50 u potential energy diagram of stationary points and TS nal int--00..48 --00..48 CaOH+ ofofrtmhethCeai+o(n4-sm)o+lecciusl/etrcaonms-p3lAexPCr1eaicntitohne.eTnthreanrecaeccthaannts- sig (57 u) nelinwhichCa+isboundabovethearomaticring. From 20 60 100 140 180 20 60 100 140 180 C1, the reaction proceeds either by abstraction of OH or frequency / kHz NH , yielding the ionic products CaOH+ and CaNH+, 2 2 FIG.7. Resonance-excitationmassspectra(uppertraces)and respectively. For the pathway leading to CaOH+ + 3- theirmoleculardynamicssimulations(lowerinvertedtraces). aminophenyl radical, the reaction proceeds through TS- (a) Top: experimental spectrum of a pure Ca+ crystal. Bot- CaOH. This TS was found to be identical for both con- tom, dashed line: corresponding simulation using a crystal formers of 3AP as the OH group is displaced from the with 675 ons. Solid line: simulated spectrum scaled by 0.92 aromatic ring towards the Ca+ ion. The resulting prod- along the frequency axis to match the experiment. (b) Top: ucts are identical for both conformers. This pathway is experimentalspectraafterareactiontimeof8minwith3AP. calculated to be exothermic by ≈0.3 eV (2320 cm−1). Bottom: scaledsimulatedspectraofcrystalscomposedof350 Ca+ and 325 heavy ions with mass 50 u (black), 57 u (red), and 60 u (green). For the second pathway, leading to CaNH+ and 2 cis/trans-3-hydroxyphenyl radical, a TS structure TS- CaNH , analogous to TS-CaOH, was found. The Ca+ 2 suredintheexperiments(seesectionIIIC). Nonetheless, ioniscoordinatedabovethearomaticringandtheamino the MD simulations serve as a useful guide for the inter- group is displaced out of the plane towards the ion. Two pretation of the mass spectra. differentTS-structuresforthetwoconformersof3APex- Based on the analysis of the ion trajectories obtained ist. They differ by the orientation of the OH-group with in the MD simulations, the peak at ≈ 165 kHz was respect to the amino group. Their energy difference was assigned to a high-frequency excitation of Ca+ ions in calculated to be 76 cm−1. The conformational depen- the combined potential of the trapping fields and the dence is preserved throughout the product channel, but product ions. The weak broad signal in the range from the energy difference between the two conformeric path- 60–100 kHz is indicative of the presence of even higher ways is very small. The pathway leading to CaNH+ is 2 masses,possiblyarisingfromconsecutivereactionsofthe calculated to be endothermic by ≈ 0.7 eV (5646 cm−1). primaryproductionswith3APfromthemolecularbeam. Under the present conditions, this second pathway is ex- pectedtobethermodynamicallyaccessibleonlyforreac- tions with Ca+ in the excited (4p) and (3d) states. V. REACTION MECHANISMS AND KINETICS The present DFT calculations predict that Ca+ in its A. Reaction pathways on the ground-state potential energy groundstatereactswith3APtoCaOH+viaasubmerged surface transition state, TS-CaOH in Figure 8, that is lower in energy than the reagents. The energy profile along the AccordingtoTableI, theCa+ (4p)staterateconstant reaction coordinates is reminiscent of the situation in re- is at least two orders of magnitude larger than those in latedabstractionreactions,e.g.,Ca++CH F33,44orthe 3 the(4s)and(3d)states,suggestingdifferentreactiondy- reactionsofOatomswithalkenes.66,68Thepresentcalcu- namics for these channels. The large values for the rate lationspredicttheTSstructuretobe1.2eVlowerinen- constantsobtainedforreactionsintheCa+ (4p)stateare ergy than the reactants. Comparing with the kinetics in indicative of a capture process.64,65 In this case, the re- similartypeofreactionsasobservedinRefs.33,66,68,it actionrateislimitedbytherateofformationofthereac- is difficult to see how the low-lying barrier in the present tioncomplex. Afterward,thereactionproceedswithnear case can lead to a rate constant for the Ca+ (4s) + 3AP unitefficiency. Thekineticsofthereactionarethensolely channel about two orders of magnitude lower than the controlledbylong-rangeintermolecularinteractions. For capture limit observed (see section IVD). Possible rea- reactions with Ca+ (4s) and (3d), however, the signif- sons for this discrepancy could be an underestimation of icantly smaller rate constants compared to the capture the barrier height by the current DFT approach or the limit are indicative of the existence of barriers on the existence of additional barriers or dynamic constraints reaction path which limit the reaction rates.66,67 In the that have not been accounted for. Overall, the present next paragraphs, the possible roles of the different 3AP calculations give no indication of a short-range reaction conformations in these two types of situations are dis- mechanismthatcouldexplainthefactorof2differencein cussed. theobservedrateconstantsforthereactionofcis/trans- While a high-level ab initio calculation for the poten- 3AP with Ca+ in its ground state. 9 CaNH+ + trans-3-hydroxyphenyl radical 2 Ca+ + cis-3-AP 5859 cm-1 5836 cm-1 trans-C2-CaNH 2 135 cm-1 trans-TS-CaNH2 y 0 cm-1 g CaNH+ + cis-3-hydroxyphenyl radical 2 er -3712 cm-1 -2317 cm-1 n - 3788 cm-1 -4693 cm-1 E - 4773 cm-1 Ca+ + trans-3-AP cis-TS-CaNH cis-C1 2 -9310 cm-1 CaOH+ + 3-aminophenyl radical cis-C2-CaNH 2 -14766 cm-1 -14 874 cm-1 TS-CaOH -1 5196 cm-1 trans-C1 C2-CaOH FIG. 8. Schematic energy diagram of stationary points and transition states on the Ca+(4s)+cis/trans-3AP potential energy surface. See text for details. 1-mc ch / E ygre110550000000(a) J=C3J0am0+a x+c=o 4clli1iss7-i3oAn Penergy 115050000000(b) JJ==13C50aJ00+m a+x =ctr3oa4llni2ssi-o3nA ePnergy 3-1nstant k / cms433...050x10-9 ne laitnetoP-1-050000 500J1=010J50=001500 2000 2500 -1-050000 500 10J0=001500 2000 2500 pture rate co221...505 ctriasn-3sA-3PAP Ion-molecule distance R / pm Ca 500 900 1300 1700 Molecular beam velocity v / ms-1 FIG.9. Centrifugallycorrectedlong-rangeinteractionpoten- FIG.10. Capturerateconstantsofcis-andtrans-3APinthe tials for (a) cis- and (b) trans-3AP in the rotational ground rotational quantum state j =0 as a function of the collision state j =0. From Ref. [29]. Reprinted with permission from velocity. Theexperimentalcollisionvelocityamountedtov= AAAS. 900 ms−1. B. Capture dynamics in the Ca+ (4p)+3AP excited channel conformers, respectively. Thus, in a classical picture, a Figure9showscentrifugallycorrectedadiabaticpoten- larger impact parameter bmax =Jmax/µv results for the tial energy curves for the reaction of Ca+ (4p) with cis cis-conformer, with µ being the reduced mass and v the and trans-3AP in their j = 0 rotational states. In the collision velocity so that a larger reaction cross section case of cis-3AP, the centrifugal barrier is more strongly σ =πb2max is obtained for the cis compared to the trans suppressed and reactive collisions proceed up to larger species. maximum values J of the total angular momentum. As shown in Figure 10, the calculated capture rate max For the collision energy of the present study (0.123 eV), constants depend on the collision energy. Moreover, the we find J = 417 and 342 for the cis- and trans- monotonicincreaseofrateconstantswithdecreasingcol- max 10 -1s x10-9 -1s x10-9 of 3AP with electronically excited Ca+ (4p). The reac- 3m2.9 (a) 3m (b) j = 0 tion rates of 3AP with Ca+ in its (4s) 2S ground and c c2.85 1/2 k / 2.8 k / (3d) 2D3/2 excited states were found to be two to three nt nt 2.80 orders of magnitude smaller. CaOH+ and CaNH+ were a2.7 a 2 nst nst j = 1 identified as the likely reaction products by resonant- co2.6 co2.75 excitation mass spectrometry. e rate 2.5 e rate 2.70 jj == 32 tioTnhceharnanteelcwoansstfaonutndobtsoerbveedclofoser ttohethCeac+apt(u4pre) lrimeaict-. ptur2.4 ptur2.65 j = 4 j = 5 The difference in the reactivities of the two conformers a 0 2 4 6 8 10 12 14 a -5-4 -3 -2 -1 0 1 2 3 4 5 couldberationalizedintermsofadiabatic-capturetheory C j C τ τ=0 in very good agreement with the experimental findings. Within the capture picture, the increased reaction rate FIG. 11. State-dependent capture rate constants k for cis- forthecis conformercomparedtothetrans speciesisex- 3-aminophenol. (a) Dependence on rotational state j(τ =0) plained by the stronger ion-dipole long-range interaction for j = 0 to j = 15. (b) Dependence on the asymmetric-top whichresultsinalargercapturecrosssection. Thesmall quantum number τ for j = 0 to j = 5. Lines are drawn to reaction rates with Ca+ in its (4s) and (3d) states indi- guide the eye. See text for details. catetheexistenceofdynamicbottlenecksalongthereac- tionpath. PreliminaryDFTcalculationsforthereaction lision energy is more prominent for the cis species with of 3AP with Ca+ on the ground state potential energy the larger dipole moment. Therefore, the ratio of the re- surface enabled a first characterization of possible reac- activity between the two conformers becomes larger for tion pathways. However, more extensive computations smaller collisional energies. We also investigated the de- are necessary to elucidate the short-range dynamics in pendence of the capture rate constants on the rotational all three reaction channels probed in the present experi- state of 3AP. As exemplified for the cis conformer at ments. the experimental collision energy in Figure 11 (a), the We expect that the present technique of combining capture rates slightly decrease as j increases. Since the electrostatic conformer selection with highly sensitive rotational temperature is about 1.1 K, about 90 % of Coulomb-crystal methods will enable the study of con- the 3AP population is confined to the rotational states formational effects in a range of ion-molecule reactions. j =1−5forwhichtherelativedifferenceoftheratecon- Electrostaticconformerseparationisapplicabletoavari- stants is < 10 %. Figure 11 (b) shows the dependence etyofpolarmoleculesaslongastheirconformersexhibit of the capture rate constant for the cis conformer on the appreciably different dipole moments. More advanced asymmetric-top quantum number τ. The dependence is techniques for the separation of molecular species and only weak over the range of states j =0−5. individual quantum states using electric field manipula- By averaging over capture rate constants of all popu- tion have been reported and could be implemented in lated rotational states of 3AP at 1.1 K and for a colli- the current methodology.26,69–72 For the ionic reaction sionenergyof0.123eV,oneobtainstheeffectivecapture partners, the generation of Coulomb crystals of sympa- rate constants k =2.7×10−9 cm3s−1 and k = thetically cooled ions allows the study of a wide range 2,cis 2,trans 1.8×10−9 cm3s−1 and their ratio k /k = 1.5. of atomic and molecular ionic species.30 Moreover, the 2,cis 2,trans These values are in good agreement with the experimen- preparation of Coulomb crystals with molecular ions in tally observed second-order rate constants for the reac- selected internal quantum states has been recently ac- tionwithCa+ intheexcited(4p)state, seesectionIVD. complished73,74 so that simultaneous studies of confor- mational and state-specific effects are now within reach for a wide range of ion-molecule reactions. VI. SUMMARY AND CONCLUSIONS We have presented a new method for the character- ACKNOWLEDGMENTS ization of conformer-specific chemical reactivities. In a proof-of-concept study, the two conformers of 3AP were This work has been supported by the Swiss National spatiallyseparatedinamolecularbeamusingtheelectro- ScienceFoundationgrant.nr.PP00P2_140834,theUni- static deflector. Subsequently, the separated conformers versityofBasel,andtheexcellencecluster“TheHamburg reactedwithastationarytargetofCoulomb-crystallized, Center for Ultrafast Imaging – Structure, Dynamics and laser-cooled Ca+ ions. Second-order rate constants for Control of Matter at the Atomic Scale” of the Deutsche thereactionsoftheindividualconformerswithCa+ were Forschungsgemeinschaft. obtained. The reaction rate for the cis conformer was found to be a factor of two larger than that for the trans 1D. H. R. Barton, “The stereochemistry of cyclohexane deriva- conformer. A detailed analysis of the rate constants for tives,” J.Chem.Soc.1953,1027(1953). the individual electronic states of Ca+ showed that the 2H.C.Dunathan,“Conformationandreactionspecificityinpyri- observed reaction rates are dominated by the reaction doxalphosphateenzymes,” PNAS55,712–716(1966).