Quantum-chemistry-aided identification, synthesis and experimental validation ofmodel systems for conformationally controlled reaction studies: Separation of the

conformers of 2,3-dibromobuta-1,3-diene in the gas phase

Ardita Kilaj,1 Hong Gao,2 Diana Tahchieva,1 Raghunathan Ramakrishnan,3 Daniel Bachmann,1

Dennis Gillingham,1 O. Anatole von Lilienfeld,1, 4 Jochen Küpper,5, 6, 7, 8 and Stefan Willitsch1, ∗

1Department of Chemistry, University of Basel, 4056 Basel, Switzerland2Beijing National Laboratory for Molecular Sciences, Institute of Chemistry,

Chinese Academy of Sciences, Beijing 100190, China3Centre for Interdisciplinary Sciences, Tata Institute of Fundamental Research, Hyderabad 500107, India

4National Center for Computational Design and Discovery of NovelMaterials (MARVEL), University of Basel, 4056 Basel, Switzerland

5Department of Physics, Universität Hamburg, 22761 Hamburg, Germany6Department of Chemistry, Universität Hamburg, 20146 Hamburg, Germany

7Center for Free-Electron Laser Science, Deutsches Elektronen-Synchrotron DESY, 22607 Hamburg, Germany8Center for Ultrafast Imaging, Universität Hamburg, 22761 Hamburg, Germany

(Dated: June 23, 2021)

The Diels-Alder cycloaddition, in which a diene reacts with a dienophile to form a cyclic compound,counts among the most important tools in organic synthesis. Achieving a precise understanding of itsmechanistic details on the quantum level requires new experimental and theoretical methods. Here,we present an experimental approach that separates different diene conformers in a molecular beamas a prerequisite for the investigation of their individual cycloaddition reaction kinetics and dynamicsunder single-collision conditions in the gas phase. A low- and high-level quantum-chemistry-basedscreening of more than one hundred dienes identified 2,3-dibromobutadiene (DBB) as an optimalcandidate for efficient separation of its gauche and s-trans conformers by electrostatic deflection.A preparation method for DBB was developed which enabled the generation of dense molecularbeams of this compound. The theoretical predictions of the molecular properties of DBB werevalidated by the successful separation of the conformers in the molecular beam. A marked differencein photofragment ion yields of the two conformers upon femtosecond-laser pulse ionization wasobserved, pointing at a pronounced conformer-specific fragmentation dynamics of ionized DBB.Our work sets the stage for a rigorous examination of mechanistic models of cycloaddition reactionsunder controlled conditions in the gas phase.

I. INTRODUCTION

Besides polar and radical reactions, pericyclic pro-cesses are one of the three fundamental reaction typesthat form the basis of synthetic organic chemistry. Giventheir great significance in organic synthesis, rigorouslydefined mechanistic pathways are an important resourcefor reaction developers. The Diels-Alder cycloaddi-tion [1], in which a diene and a dienophile react to forma cyclic product, is a practical and widely used pericyclicreaction in organic synthesis. While the broad strokesof its mechanism are well understood, the detailed re-action manifold for any given substrate is often exten-sively discussed [2–7]. Given its mechanistic subtletiesand its importance, the Diels-Alder reaction has servedas a test-bed for establishing new types of mechanisticanalysis [8]. In case of the “canonical” concerted path-way, which involves a cyclic transition state and is widelydiscussed in the literature [2–7], only the s-cis conformerof the diene reacts to form the cyclic product – whilein a stepwise mechanism also the s-trans conformer cancontribute to the reaction. Stepwise pathways become

∗ stefan.willitsch@unibas.ch

particularly important for ionic variants of the reaction,i.e., polar cycloadditions, for which traditional conceptsfor rationalizing the mechanism such as the conservationof orbital symmetry break down [4, 7, 9, 10]. Thus, anexperimental investigation of the mechanistic impact ofindividual rotamers is clearly warranted. This, however,requires a way to probe the reactivities of the individ-ual conformers of a specific diene, a difficult task understandard liquid-phase reaction conditions.

In recent years, molecular beams have become an im-portant tool for the investigation of gas-phase chemicalreaction dynamics under highly controlled conditions[11,12]. In particular, the use of inhomogeneous electric fieldshas enabled the electrostatic deflection and thus spatialseparation of different molecular conformers and isomersaccording to their different electric dipole moments[13–17]. The combination of such a “controlled” molecularbeam with a stationary reaction target of sympatheticallycooled molecular ions in an ion trap forms a powerful toolfor studies of the kinetics and dynamics of ion-moleculereactions [12, 16–18]. Recently, this approach has en-abled the measurement of individual chemical reactivitiesof the cis and trans conformers of 3-aminophenol withtrapped Ca+ ions [16, 18] and of the two nuclear-spinisomers of water [19] toward trapped diazenylium ions

arX

iv:2

004.

0965

9v1

[ph

ysic

s.ch

em-p

h] 2

0 A

pr 2

020

2

(N2H+) in a proton-transfer reaction [17]. Consequently,molecular beams in conjunction with ion traps offer a di-rect and precise way to measure conformer-specific rateconstants and thus to investigate the reaction mechanismof polar cycloadditions. The key challenge is the identi-fication of suitable model systems amenable to a charac-terisation under these specific experimental conditions.In the present context, this means that (i) both reac-tants, the diene and the dienophile, need to be volatileenough to enable their preparation in the gas phase, (ii)the energy difference between the s-cis and s-trans con-formers of the diene needs to be small enough so thatboth can be populated in the cold environment of a su-personic molecular beam, and (iii) the difference of theirpermanent dipole moments in the molecular frame needsto be large enough to enable their efficient electrostaticseparation [11, 16].

The selection of an optimal model system is, therefore,a multi-dimensional optimization problem. Traditionally,the choice would be guided by chemical and physical in-sight. Nowadays, numerical simulations of reactions canprovide meaningful atomistic insights to support exper-imental efforts [20]. In the context of designing exper-iments, virtual screening has proven to be a powerfulapproach for suggesting compounds matching the phys-ical and chemical properties of interest. Computationalscreening has already been successfully applied in pro-tein, materials and catalytic design [21–23]. Here, weapply this approach to identify optimal dienes suitablefor controlled gas-phase polar cycloaddition reactions.

This article combines methods of theoretical, organic,and physical chemistry to lay the foundations for a subse-quent experimental characterization of conformational ef-fects in polar cycloaddition reactions. Quantum-chemicalcalculations were performed to screen the reactant spacefor a diene with physical properties optimized toward itsuse in conformer-selected reaction-dynamics experimentsin the gas phase. A synthesis for the theoretically iden-tified optimal diene, 2,3-dibromobutadiene (DBB), wasthen developed. Finally, the physical properties of thecompound were validated in a molecular-beam experi-ment separating the two conformers by electrostatic de-flection.

II. THEORETICAL AND EXPERIMENTALMETHODS

A. Theoretical screening

We applied the concept of computational screening to-wards the problem of exploring chemical space from firstprinciples [24] in order to find a polar diene for conformer-selective Diels-Alder cycloadditions. The chemical role ofpolar dienes in Diels-Alder reactions was already exploredcomputationally in various preceding studies, see, e.g.,references [10, 25]. Efficient electrostatic separation ne-cessitates a certain difference in electric dipole moments

∆µ between the s-cis and s-trans conformers of the di-ene [11, 26]. Moreover, a small energy difference ∆E be-tween the ground states of the conformers is required toensure a significant thermal population of both speciesin the molecular beam. For a successful experiment, asuitable diene has to be identified which satisfies both ofthese conditions.

High-throughput based virtual design of novel com-pounds typically starts from an initial scaffold, which caneasily be modified at multiple sites through functional-ization by substituting atoms or functional groups [27].Theoretical screening then yields the best mutated com-binations selected according to their proximity to the de-sirable physical or chemical target-property values.

In the present work, we computationally searched thechemical space of butadiene derivatives for which thes-cis and s-trans isomers exhibit maximal and mini-mal differences in dipole moments and energies, respec-tively. Substituting CH2 in positions 1 and 4 by NH,or O, and substituting the hydrogen attached to carbonin position 2 and 3 by halogens (F, Cl, Br, I), a pre-liminary density functional theory (DFT) based scan of144 candidates, not accounting for symmetrically redun-dant species, resulted in the identification of di-halogensubstituted butadiene as a promising series of candi-dates for experiments. Due to the chemical reactivityof iodine-substituted compounds, potentially hamperingsubsequent synthetic efforts, we have only included thedifluoro, dichloro, and dibromo 2,3-substituted butadi-enes for further in-depth theoretical analysis.

Torsional energy profiles were subsequently calculatedfor all three species using DFT with the double-hybridfunctional DSD-PBEP86-D3BJ[28] and a large basis set(def2-QZVPP)[29] which was previously shown to givegood performance for the prediction of torsional potentialenergy surfaces of similar molecules.[30] For the torsionalprofiles, the geometry optimizations were restricted bykeeping the torsional angle Θ = ΘH3C−CC−Y constant,imposing achirality. The entire range of 0 < Θ < 180

was scanned in steps of ∆Θ = 20. Note that due to theapplied constraints for the torsional angles throughoutthe geometry optimization, the torsional profile is sym-metric [E(360 −Θ) = E(Θ)]. Calculations were carriedout with the Gaussian09 program package [31].

B. Stark-energy and trajectory simulations

To theoretically assess the behaviour of DBB in anelectrostatic deflection experiment, Stark energies andeffective dipole moments of individual rotational statesof DBB were calculated using the CMIstark softwarepackage[32]. The calculated rotational constants anddipole moments as listed in Table II were used as in-put parameters. The Stark energies served as input pa-rameters for simulating state-specific deflection profilesof the molecular beam by the electrostatic deflector witha home-made software package based on CMIfly [11].

3

Trajectory simulations were carried out for gauche-DBB with 105 molecules for each rotational state up to amaximum rotational quantum number of Jmax = 20. Forthe unpolar s-trans-DBB, only a single quantum state,J = 0, needed to be simulated with a total number of 106

trajectories. In all cases, initial positions were uniformlysampled across the cross section of the orifice of the gasnozzle generating the molecular beam. The initial veloc-ities of the molecules were sampled from a normal dis-tribution. The velocity distribution was matched to theexperimentally determined mean longitudinal velocity of843 m/s with a longitudinal velocity spread of 10 %. Atransverse velocity spread of 4 m/s was chosen to matchthe divergence of the beam to the acceptance angle of theskimmers in the assembly. According to the theoreticalenergy difference between the ground states of gauche-and s-trans-DBB in Table II, the ratio of their thermalpopulations at room temperature is pgauche/ptrans = 0.30,taking into account the two-fold degeneracy of the gauchestructure, see Figure 2. This ratio was used to scale thesimulated deflection profiles of the two species.

In order to calculate thermally averaged deflection pro-files nσ,T (y) for each conformer (σ ∈ gauche, s-trans)at a specific rotational temperature T , we followed a simi-lar procedure as before [17, 33]. For each rotational quan-tum state |JKaKcM〉, histograms of the arrival positionsnσJKaKcM (y) normalized by the initial sample size wereextracted from the simulated trajectories. Here, J is thequantum number for overall angular momentum neglect-ing nuclear spin, i.e., for the overall rotation, Ka and Kc

are pseudo-quantum numbers for the projection of theangular momentum onto the molecular axes, and M isthe quantum number for the projection of the rotationalangular momentum onto the external-field axis. Thermalaveraging was performed using the relation

nσ,T (y) =pσNσ

Jmax∑J=0

∑Ka,Kc

J∑M=0

× (1)

×gMe−EJKaKc/kBT nσJKaKcM (y), (2)

with the partition function

Nσ =

Jmax∑J=0

∑Ka,Kc

J∑M=0

gMe−EJKaKc/kBT . (3)

Here, kB denotes the Boltzmann constant, pσ are thepopulations of the conformers at room temperature andEJKaKc are the field-free rotational energies. The de-generacy factor gM takes values gM = 1 for M = 0 andgM = 2 for M > 0. The total thermal deflection profilewas calculated from the sum of the deflection profiles ofthe gauche- and s-trans-conformers,

ntot,T(y) = ngauche,T (y) + ns−trans,T (y). (4)

C. Synthesis of DBB

Since DBB is an unstable compound which is notcommercially available, we needed to devise a synthe-sis that delivered the material in sufficiently high purityand quantity for molecular-beam experiments. Prior tothis work, Stewart Jr. et al. [34] reported a synthesisof DBB from 1,4-dihalo-2-butyne and a cuprous halidewhich formed activated halide ions present in solution.DBB was obtained through continuous distillation dur-ing the reaction. After extensive screening of poten-tial conditions, we found that the elimination reactionof 1,2,3,4-tetrabromobutane (TBB) with the stericallyhindered base 1,8-Diazabicyclo[5.4.0]undec-7-ene (DBU)primarily delivered the elimination product DBB. Themost effective conditions involved adding DBU to a solu-tion of TBB in diethyl ether under a constant stream ofnitrogen, with NaI as an additive to accelerate the substi-tution. Under these conditions, near complete conversionto DBB was observed after 1 hour of reaction time. Thepurified sample was directly used in the molecular beamapparatus. Further information on the synthesis can befound in the electronic supplementary information (ESI).

D. Experimental setup for conformer separation

A schematic of the experiment is depicted in Figure 1.Details of the experimental setup have also been de-scribed in our earlier work [16–18, 35].

A supersonic jet of DBB seeded in neon was generatedusing a pulsed gas valve and passed through two skim-mers before entering the electrostatic deflector. The re-sulting molecular beam contained a mixture of the gaucheand s-trans conformers of DBB. The inset of Figure 1depicts the inhomogeneous electric field in the deflec-tor with a cross marking the nominal molecular beamaxis. Here, the two conformers were angularly dispersedand thus spatially separated according to their differentdipole moments [11]. Behind the deflector, the molecularbeam was directed at a linear-quadrupole ion-trap (LQT)coupled to a time-of-flight mass spectrometer (TOF-MS).The entire molecular beam setup can be tilted verti-cally with respect to the TOF-MS, which allows prob-ing different regions of the dispersed molecular beam.The tilting angle thus defines a deflection coordinate y.When entering the TOF-MS, the DBB molecules wereionized by either pulsed vacuum-ultraviolet (VUV) ra-diation or femtosecond (fs) laser pulses and acceleratedonto a microchannel-plate detector (MCP) using high-voltage electrodes.

1. Molecular beam:

The molecular beam was generated from DBB vaporat room temperature seeded in neon carrier gas at 5 bar.The gas mixture was pulsed through a cantilever piezo

4

FIG. 1. Overview of the experimental setup.

valve (MassSpecpecD ACPV2, 150 µm nozzle diameter)at a repetition rate of 10 Hz and a gas pulse duration of250 µs at the LQT. The velocity in the direction of prop-agation of the resulting molecular beam was measured inthe same way as described in Kilaj et al. [17] and yieldeda value of 843(58) m/s. To determine the beam densityof DBB in the molecular beam, we first calibrated theabsolute sensitivity of the TOF-MS by loading Coulombcrystals of defined ion number in our ion trap [36]. Then,following refs. 37, 38, we measured the total ion yieldof DBB from the molecular beam using ionization with150 fs, 775 nm laser pulses. Observing the logarithmicincrease of ion yield with laser intensity [38] allowed usto estimate a DBB density of 7.8(5) × 107 cm−3. Fordeflection experiments, a voltage of 13 kV was appliedbetween the two 15.4 cm long deflector electrodes, heldat a distance of 1.4 mm, to generate the required verticalelectric field gradient [11, 17, 18, 33]. By comparison ofthe measured deflection profiles with Monte-Carlo sim-ulations [16, 17, 33], a rotational temperature of 1.0 Kcould be estimated.

2. Ion trap and TOF-MS:

The LQT is connected to a TOF-MS orthogonal to themolecular-beam propagation axis for quantitative massanalysis[35]. To enable a better ion selection in the TOF-MS, a voltage of 500 V was applied to the four end caps ofthe trap. For extraction of the ionized species, a perma-nent voltage of 4.0 kV was applied to the repeller elec-trode. A microchannel plate detector (MCP, PhotonisUSA) operating at a typical voltage of 2.3 kV was placedat the end of the flight tube.

3. Femtosecond and VUV ionization:

Non-resonant multi-photon ionization of the DBBmolecules was performed with pulses from a Ti:Sapphirefemtosecond laser (CPA 2110, Clark-MXR, Inc.) at awavelength of 775 nm and pulse duration of 150 fs fo-cused to a diameter of ∼ 30 µm in the sample. In addi-tion, a vacuum-ultraviolet (VUV) light source was usedfor soft ionization of the DBB molecules. Similar to otherwork [39–41], pulses of 118 nm light were generated usingthird-harmonic generation (THG) by focusing the thirdharmonic output beam of a Nd:YAG laser (Quantel Bril-liant, 355 nm, 5 ns) into a gas cell containing a phase-matched gas mixture of xenon and argon (ratio 1:10, totalpressure 100 mbar). The pump laser was operated at arepetition rate of 10 Hz and the pulse energy was set to25 mJ such that a UV to VUV conversion efficiency ofapproximately 10−5 was achieved. The VUV beam wasre-focused (spot size ∼ 100 µm) into the trap chamberat the center of the ion trap with a single MgF2 lens(Thorlabs, f = 200 mm) at a distance of 120 mm fromthe UV focus (spot size ∼ 15 µm). Owing to its strongerindex of refraction in the VUV, the MgF2 lens also servedas an optical element to separate the pump-laser beamfrom the VUV beam. A LiF window was used to sealoff the ultra-high-vacuum chamber housing the ion trapfrom the VUV generation chamber. In order to blockthe 355 nm pump-laser beam and prevent it from enter-ing the interaction region or damaging the UV sensitiveLiF window, a MACOR-protected pinhole was installedin front of the LiF window. The VUV detector was fab-ricated from two copper electrodes with a typical biasvoltage of about 1 kV and the VUV-induced photocur-rent was measured through the resulting voltage across

5

a 50 Ω resistor.

III. RESULTS AND DISCUSSION

A. Torsional profiles of the 2,3-dihalobutadienes

A graphical representation of the torsional profiles ofthe 2,3-difluoro-, dichloro-, and dibromobutadienes isshown in Figure 2 (a). The global minimum was found tobe the s-trans structure (at Θ = 180) in all cases. Localminima were found to be near gauche (rather than s-cis)structures at torsional angles varying from Θ = 50 to60 depending on the specific molecule.

Subsequently, differences in potential energy and ab-solute dipole moment between the local and globalminima were calculated including harmonic and an-harmonic thermal corrections of zero-point vibrationalenergy and Gibbs free energy. The DFT results forthe torsional potential and the dipole moments werein very good agreement with CCSD(T)-F12/cc-pVTZ-F12,[42, 43] and CCSD/cc-pVTZ-F12[44] (Table I) cal-culations, respectively. Furthermore, relaxation at theCCSD/cc-pVTZ-F12 level resulted in geometries identi-cal to those found by DSD-PBEP86-D3BJ/def2-QZVPPwith a root-mean-square deviation (RMSD) of 1.9 pmbetween the final geometries. These results confirm thereliability of our DFT predictions.

While exhibiting still non-negligible conformational en-ergy differences, see Table I, the large dipole moment dif-ferences among the 2,3-di[halogen]but-1,3-diene conform-ers appeared promising, motivating its selection for sub-sequent experimental investigations. For instance, thedipole-moment difference for 2,3-dibromo-1,3-butadienewas computed at the CCSD/cc-pVTZ-F12 level of the-ory (neglecting all relativistic effects) to be ∆µ = 2.11 D.

As the main result of the theoretical screening, 2,3-dibromobuta-1,3-diene (DBB) was identified as an opti-mal diene for the envisaged experiments that possessesboth a sufficiently small energy gap between the gaucheand s-trans ground states as well as a large enough differ-ence in the electric dipole moment of the two species (Fig-ure 2 (b) and Table II) . For gauche-DBB a dipole mo-ment of µ = 2.29 D was calculated at the DSD-PBEP86-D3BJ/def2-QZVPP level of theory, while the s-trans iso-mer is apolar on grounds of its inversion symmetry. Ta-ble II summarizes the calculated energy difference as wellas the absolute values of the dipole moments and the ro-tational constants for both conformers.

B. Simulations of the electrostatic deflection ofDBB

Based on the molecular properties obtained from thecomputations, we predicted trajectories of gauche- and

a)

(

)

CH2=C(F)-C(F)=CH2

CH2=C(Cl)-C(Cl)=CH2

CH2=C(Br)-C(Br)=CH2

( )Torsional angle

b)

ΔZPE

ΔG

( )Torsional angle

(

)

FIG. 2. a) Cuts through the potential energy surfaceof 2,3-dibromo-1,3-butadiene (CH2=C(Br)-C(Br)=CH2), 2,3-dichloro-1,3-butadiene (CH2=C(Cl)-C(Cl)=CH2), and 2,3-difluoro-1,3-butadiene (CH2=C(F)-C(F)=CH2) along the tor-sional coordinate Θ calculated using the DSD-PBEP86-D3BJfunctional. Due to the symmetry of the molecules, the tor-sional profiles are symmetric with respect to mirroring atΘ = 180 . The figure shows relative energies ∆E referencedto the energies of the s-trans structures at Θ = 180. b) Po-tential energy of 2,3-dibromo-1,3-butadiene as a function ofthe torsional angle Θ calculated at CCSD(T)/VTZ-F12 andDSD-PBEP86-D3BJ/def2-QZVPP levels of theory. The low-ering of the energy at the local minima due to the zero pointvibrational energy (∆ZPE) and the Gibbs free energy (∆G)is illustrated for the gauche structure. Due to its symmetry,the molecule exhibits two equivalent gauche structures.

s-trans-DBB molecules through the electrostatic deflec-

6

Method CH2=C(Br)-C(Br)=CH2 CH2=C(Cl)-C(Cl)=CH2 CH2=C(F)-C(F)=CH2

∆E (eV)CCSD(T)/cc-pVTZ-F12 0.097 0.117 0.151

DSD-PBEP86-D3BJ/def2-QZVPP 0.097 0.114 0.155DSD-PBEP86-D3BJ/def2-QZVPP + harm. therm. corr. 0.050 0.069 0.142

DSD-PBEP86-D3BJ/def2-QZVPP + anharm. therm. corr. 0.049 0.068 0.139∆µ (Debye)

CCSD/cc-pVTZ-F12 2.1072 2.2831 2.5828DSD-PBEP86-D3BJ/def2-QZVPP 2.2963 2.3837 2.5938

TABLE I. Differences in potential energy ∆E and dipole moment ∆µ between the local and global torsional minimas ofDBB including thermal corrections at T=298.15K (zero-point and Gibbs free energy) for selected 2,3-dihalogen-substitutedbutadienes.

Energy Dipole moment Rotational constants (GHz)∆E = Ecis − Etrans (eV) µ (D) Ae Be Ce

gauche-DBB 0.049 2.29 2.3526 0.8793 0.7097s-trans-DBB 0.00 4.6077 0.5997 0.5306

TABLE II. Differences in energy, dipole moments and rotational constants of gauche- and s-trans-2,3-dibromobuta-1,3-diene(DBB) calculated at the DSD-PBEP86-D3BJ/def2-QZVPP level of theory including anharmonic thermal corrections.

a)

b)

d)

c)

0 50 100 150Electric field (kV/cm)

0

1

2

Dip

ole

mom

ent

(D)

0 50 100 150

0.6

0.4

0.2

0.0

Energ

y (

meV

)

gauches-trans gauche

0 kV

13 kV

4x

FIG. 3. Simulations. Calculated Stark energies a) and effective dipole moments b) vs. electric field strength for individualrotational states with J = 0, 1, 2 of the gauche and s-trans conformers of DBB. c) Rotational state populations, summed overall levels with the same angular momentum quantum number J , for gauche and s-trans DBB at a rotational temperature of1.0 K. d) Simulated deflection profile of the gauche conformer (black line) with its different rotational-state contributions incolor. The contributions of the different J states are color-coded according the color scale indicated. The undeflected beamprofile is shown by the gray area.

tor. In Figure 3, calculated Stark energies (a) and effec-tive dipole moments (b) for rotational states with angu-lar momentum quantum numbers up to J = 20 of thegauche and s-trans conformers of DBB are shown as afunction of electric field strength. In the applied electricfields, all rotational states of the gauche conformer are

strong-field seeking with negative Stark shifts, whereasthe s-trans conformer does not exhibit a DC Stark effectbecause of its vanishing dipole moment in the molecularframe. In Figure 3 c), rotational state populations forgauche- and s-trans-DBB at a rotational temperature of1.0 K are shown. At this temperature, rotational states

7

up to J = 14 are significantly populated and can be ex-pected to contribute to the beam-deflection profiles forboth conformers.

The density profiles of the molecular beam along thedeflection coordinate at the position of intersection withthe probe laser (deflection profiles) are plotted in Fig-ure 3 d) for the gauche conformer. The color-codedcurves show the contributions from the individual rota-tional states with angular momentum up to J = 20, whilethe thick black line corresponds to the total thermallyaveraged deflection profile at a rotational temperatureof 1.0 K. For clarity, the contributions of the individualrotational states have been multiplied by a factor of 4in the figure. The inset contains the same curves withheights normalised to 1 to allow for a better compari-son. The grey area in the main plot is a simulation ofthe undeflected beam profile, at a deflector voltage of0 kV, which also corresponds to the profile of the unpo-lar s-trans conformer with the deflector voltages turnedon. The rotational states of the gauche conformer withlargest deflection are the low-angular-momentum states(small J). Consequently, significant spatial separation ofthe gauche and s-trans conformers can only be achievedexperimentally for samples with a sufficiently low rota-tional temperature [45].

C. Experimental deflection profiles

In order to measure the spatial profiles of the DBBmolecules beam emanating from the electrostatic deflec-tor, the molecules were ionized by laser pulses and ejectedinto the TOF-MS. The choice of the ionization methodturned out to be crucial. In Figure 4 a), typical TOF-MStraces obtained using fs-laser-pulse ionization (top) andVUV ionization (bottom) are shown. While fs-laser ion-ization yielded a large quantity of fragmentation productsof the parent DBB molecule, VUV ionization produced aclean mass spectrum with a single peak originating fromDBB at 212 u. The inset shows an extended mass rangearound the DBB peak, illustrating that other species orclusters with larger mass cannot be observed under thepresent experimental conditions.

Analysis of the different fragment-ion signals obtainedfrom fs-laser ionization, Figure 5 a), revealed that mostof the fragments show distinct deflection profiles. In-triguingly, the mass signal corresponding to the parentmolecule does not seem to exhibit deflection. This sig-nal could in principle be generated by the break up oflarger DBB-containing clusters which may exhibit onlyvery small dipole moments, similar to the situation ob-served in the deflection of H2O [17]. However, DBB clus-ter ions are not observed in the TOF spectra, Figure 4 a),and hence we can essentially rule out that the lack of de-flection observed for the DBB+ mass peak measured byfs ionization is due to breakup of molecular aggregates.The different deflection profiles recorded for ion signalsof the individual fragments are caused by the fs-laser-

FIG. 4. Deflection profiles. a) Representative TOF-MStrace obtained using fs-laser-pulse ionization (top) and VUVionization (bottom) of DBB. b) Molecular-beam profiles mea-sured using both ionization methods at deflector voltages of0 kV and 13 kV together with corresponding simulations. Forfs-laser-pulse ionization, the profiles for the fragment C4H+

n

are shown. Error bars represent standard errors of at leastfive independent measurements.

induced breakup of the parent DBB molecule. It is possi-ble that the electric field of the relatively long laser pulsesdrives different multiple-ionization dynamics for the po-lar gauche conformation than for to the apolar s-transconformation, thus leading to distinct conformer-specificfragmentation patterns [46]; these are further discussedon the next page.

The complexity of the observed fragmentation dynam-ics prevented us from unambiguously determining the de-flection profile of the DBB parent molecule using fs-laserionization. Therefore, we implemented soft VUV ion-ization, which is capable of ionizing DBB without frag-mentation as apparent from Figure 4 a) and the corre-sponding molecular-beam profiles in Figure 4 b). While

8

the data points measured with VUV (purple trianglesand circles, respectively) probe DBB directly, the datashown for fs-laser-pulse ionization corresponds to the ac-cumulated signal for the fragments C4H+

n (n = 0 . . . 4)(blue triangles and squares, respectively) produced underthese conditions. The experimental data points for VUVionization agree very well with the simulated thermally-averaged beam profiles, which are shown as grey dottedline (0 kV) and black solid line (13 kV). Correspondingindividual contributions from the gauche and s-trans con-formers are depicted as the blue and orange shaded ar-eas, respectively. The deflection profile at 13 kV showsa tail towards higher deflection coordinates where simu-lations indicate the presence of pure gauche-DBB. Theoverall very good agreement between the measured andsimulated deflection profiles allows us to confirm the suc-cessful separation of the DBB conformers and validatesthe accuracy of the theoretical calculations.

Further evidence for the separation of the gauche ands-trans conformers can be found in the measured frag-mentation products due to fs-laser-pulse ionization ofthe molecular beam. Figure 5 a) shows normalized pro-files of four representative fragment families Br+, C4H+

4 ,C2HnBr+, C4H4Br+ and the parent molecule DBB+ asa function of the deflection coordinate. Clearly, the tailof the profile towards large deflection coordinates, whereone expects the contribution from gauche-DBB, variesstrongly among the different fragments, with Br+ show-ing the largest and DBB+ almost zero amplitude. In theregion around y ≈ −1 mm, this behavior is inverted. Atthis location, our trajectory simulations predict a pre-dominance of s-trans-DBB. In order to quantify the im-balance of the observed fragment yields for the gaucheand s-trans conformers, we selected the data points at thelocations labeled A and B in the figure. From our simula-tions, we estimate that the populations are ps−trans ≈ 1at A and ps−trans ≈ 0 at B. We evaluate the imbal-ance between gauche and s-trans for any fragment X asthe relative difference aX = (nAX − nBX)/(nAX + nBX) withnA,BX = NA,B

X /NA,BVUV being the fragment counts NA,B

X

normalized by the total DBB beam density NA,BVUV at the

respective point as measured by VUV ionization. Theimbalance aX takes values in the range [−1,+1], corre-sponding to a strong correlation with s-trans or gaucheDBB, respectively. Figure 5 b) shows the obtained im-balance values which range from −0.9(1) for DBB+ to0.60(7) for Br+. All fragments show a tendency of in-creasing imbalance towards gauche with decreasing frag-ment size, thus suggesting that gauche DBB is morelikely to break up into smaller parts during the inter-action with the fs laser pulse. A rationalization of thisphenomenon requires further study.

IV. CONCLUSIONS

Driven by the motivation to gain a precise under-standing of the effects of molecular conformation in cy-

a)

b)

A

B

s-trans gauche

FIG. 5. Deflection curves of a molecular beam contain-ing DBB probed at different ionic fragment massesproduced by fs-laser-pulse ionization. a) Ion counts offour molecular fragments and the mass of the parent DBBmolecule vs. deflection coordinate. b) Measured imbalanceof fs-laser ionization products between the two points A andB in a). See text for details. Error bars represent standarderrors of six independent measurements.

cloaddition reactions, a quantum-chemical screening wasperformed to identify diene candidates suitable for con-former separation in a molecular-beam apparatus. As anoptimal diene, 2,3-dibromobuta-1,3-diene was found toexhibit the desired large difference in electric dipole mo-ments and small energy difference between the two con-formers. Since this particular dihalogenated diene can-not be purchased, mainly due to its intrinsic tendencyto undergo polymerization, a synthesis was developed toproduce the compound in adequate purity. Experimen-tal validation of the calculated properties was achievedby seeding DBB in a molecular beam and separating itsgauche and s-trans conformers in an electrostatic fieldgradient. A deflection profile of DBB was measured by

9

subsequent ionization and ejection into a time-of-flightmass-spectrometer. The implementation of a vacuum-ultraviolet light source achieved ionization of the parentdiene without fragmentation and therefore made it pos-sible to directly measure its deflection behavior. Spatialseparation of the two conformers was then confirmed bya close agreement of the observed deflection profiles withMonte-Carlo simulations based on the theoretical molec-ular properties. Comparison between ion yields fromVUV and non-resonant fs-laser-pulse ionization suggestdifferent fragmentation patterns for the gauche and s-trans conformers during ionization in the strong field.The polar gauche conformer showed an enhanced ten-dency to fragment in comparison with the apolar s-transconformer. The successful separation of the gauche ands-trans conformers of this tailor-made diene paves theway toward studies of conformer-selected polar cycload-dition reactions in a cold and controlled environment.

CONFLICTS OF INTEREST

The authors declare no competing financial or non-financial interests.

ACKNOWLEDGEMENTS

We thank Philipp Knöpfel, Grischa Martin and GeorgHolderried for technical support. Marco Meyer andJia Wang are acknowledged for their assistance withthe experiments. We thank Max Schwilk for fruit-ful discussions on the theoretical results. This workis supported by the Swiss National Science Founda-tion (Nr. BSCGI0_157874). O.A.v.L. acknowledges fur-ther funding from the Swiss National Science foundation(Nr. PP00P2_138932 and 407540_167186 NFP 75 BigData) and from the European Research Council (ERC-CoG grant QML). This work was partly supported by theNCCR MARVEL, funded by the Swiss National ScienceFoundation. H.G. and S.W. acknowledge support by theK.C. Wong Education Foundation.

[1] O. Diels and K. Alder, Justus Liebigs Ann. Chem. 460,98 (1928).

[2] K. N. Houk, J. González, and Y. Li, Acc. Chem. Res. 28,81 (1995).

[3] J. M. Serafimov, D. Gillingham, S. Kuster, and D. Hil-vert, J. Am. Chem. Soc 130, 7798 (2008).

[4] L. R. Domingo and J. A. Sáez, Org. Biomol. Chem. 7,3576 (2009).

[5] L. R. Domingo, J. Chil. Chem. Soc. 2615, 59 (2014).[6] M. A. F. de Souza, E. Ventura, S. A. do Monte, J. M.

Riveros, and R. L. Longo, J. Comput. Chem. 37, 701(2016).

[7] U. Rivero, O. T. Unke, M. Meuwly, and S. Willitsch, J.Chem. Phys. 151, 104301 (2019).

[8] D. A. Singleton and A. A. Thomas, J. Am. Chem. Soc117, 9357 (1995).

[9] P. J. Donoghue and O. Wiest, Chem. Eur. J. 12, 7018(2006).

[10] U. Rivero, M. Meuwly, and S. Willitsch, Chem. Phys.Let. 683, 598 (2017).

[11] Y.-P. Chang, D. A. Horke, S. Trippel, and J. Küpper,Int. Rev. Phys. Chem. 34, 557 (2015).

[12] S. Willitsch, Adv. Chem. Phys. 162, 307 (2017).[13] F. Filsinger, U. Erlekam, G. von Helden, J. Küpper, and

G. Meijer, Phys. Rev. Lett. 100, 133003 (2008).[14] F. Filsinger, J. Küpper, G. Meijer, J. L. Hansen, J. Mau-

rer, J. H. Nielsen, L. Holmegaard, and H. Stapelfeldt,Angew. Chem. Int. Ed. 48, 6900 (2009).

[15] T. Kierspel, D. A. Horke, Y.-P. Chang, and J. Küpper,Chemical Physics Letters 591, 130 (2014).

[16] Y.-P. Chang, K. Dlugolecki, J. Küpper, D. Rösch,D. Wild, and S. Willitsch, Science 342, 98 (2013).

[17] A. Kilaj, H. Gao, D. Rösch, U. Rivero, J. Küpper, andS. Willitsch, Nat. Commun. 9, 2096 (2018).

[18] D. Rösch, S. Willitsch, Y.-P. Chang, and J. Küpper, J.Chem. Phys. 140, 124202 (2014).

[19] D. A. Horke, Y.-P. Chang, K. Długołęcki, and J. Küpper,Angew. Chem. Int. Ed. 53, 11965 (2014).

[20] T. Sperger, I. A. Sanhueza, and F. Schoenebeck, Acc.Chem. Res 49, 1311 (2016).

[21] W. Notz, F. Tanaka, S.-i. Watanabe, N. S. Chowdari,J. M. Turner, R. Thayumanavan, and C. F. Barbas, J.Org. Chem. 68, 9624 (2003).

[22] S. Er, C. Suh, M. P. Marshak, and A. Aspuru-Guzik,Chem. Sci. 6, 885 (2015).

[23] A. C. Doney, B. J. Rooks, T. Lu, and S. E. Wheeler, ACSCatal. 6, 7948 (2016).

[24] O. A. von Lilienfeld, Int. J. Quantum Chem. 113, 1676(2013).

[25] M. J. S. Dewar, S. Olivella, and J. J. P. Stewart, J. Am.Chem. Soc. 108, 5771 (1986).

[26] F. Filsinger, S. Putzke, H. Haak, G. Meijer, and J. Küp-per, Phys. Rev. A 82, 052513 (2010).

[27] O. A. von Lilienfeld, Towards the Computational Designof Compounds from First Principles, vol. IX of Mathe-matical Physics Studies (Springer, 2014).

[28] S. Kozuch and J. M. L. Martin, Phys. Chem. Chem. Phys.13, 20104 (2011).

[29] F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys.7, 3297 (2005).

[30] D. N. Tahchieva, D. Bakowies, R. Ramakrishnan, andO. A. von Lilienfeld, J. Chem. Theor. Comput. 14, 4806(2018).

[31] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria,M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone,

10

B. Mennucci, G. A. Petersson, et al., Gaussian 09 Revi-sion E.01, Gaussian Inc., Wallingford CT, 2009.

[32] Y.-P. Chang, F. Filsinger, B. G. Sartakov, and J. Küpper,Comp. Phys. Comm. 185, 339 (2014).

[33] F. Filsinger, J. Küpper, G. Meijer, L. Holmegaard, J. H.Nielsen, I. Nevo, J. L. Hansen, and H. Stapelfeldt, J.Comp. Phys. 131, 064309 (2009).

[34] C. A. Stewart Jr., B. Hundred Del., and assignor to E.I. du Pont de Nemours and Co, Preparation of 2, 3-dichlorobutadiene-1,3 (1962), uS patent no. 3061653A.

[35] D. Rösch, H. Gao, A. Kilaj, and S. Willitsch, EPJ Tech.Instrum. 3, 5 (2016).

[36] P. C. Schmid, J. Greenberg, M. I. Miller, K. Loeffler, andH. J. Lewandowski, Rev. Sci. Instrum 88, 123107 (2017).

[37] S. M. Hankin, D. M. Villeneuve, P. B. Corkum, and D. M.Rayner, Phys. Rev. A 64, 013405 (2001).

[38] J. Wiese, J.-F. Olivieri, A. Trabattoni, S. Trippel, andJ. Küpper, New J. Phys 21, 083011 (2019).

[39] Y. J. Shi, S. Consta, A. K. Das, B. Mallik, D. Lacey, andR. H. Lipson, J. Chem. Phys 116, 6990 (2002).

[40] S. E. V. Bramer and M. V. Johnston, Appl. Spectrosc.46, 255 (1992).

[41] R. Steenvoorden, P. Kistemaker, A. D. Vries, L. Micha-lak, and N. Nibbering, Int. J. Mass Spectrom. 107, 475(1991).

[42] H.-J. Werner, G. Knizia, and F. R. Manby, Mol. Phys.109, 407 (2011).

[43] K. A. Peterson, T. B. Adler, and H.-J. Werner, J. Chem.Phys. 128, 084102 (2008).

[44] G. D. Purvis III and R. J. Bartlett, J. Chem. Phys. 76,1910 (1982).

[45] S. Trippel, M. Johny, T. Kierspel, J. Onvlee, H. Bieker,H. Ye, T. Mullins, L. Gumprecht, K. Długołęcki, andJ. Küpper, Rev. Sci. Inst. 89, 096110 (2018).

[46] S. Zigo, A.-T. Le, P. Timilsina, and C. A. Trallero-Herrero, Sci. Rep. 7, 42149 EP (2017).

11

Supplementary Information

S1. SYNTHESIS OF 2,3-DIBROMOBUTADIENE

Materials and methods

Diethyl ether and sodium iodide were purchased from Sigma-Aldrich. 1,8-Diazabicyclo[5.4.0]undec-7-ene was pur-chased from Alfa Aesar. 1,2,3,4-tetrabromobutane was purchased from TCI-chemicals. All reagents were used withoutfurther purification. Chloroform for NMR measurements was purchased from Cambridge Isotope Laboratories. All1H and 13C NMR spectra (Fig. S2)were recorded on a Bruker Avance III (HD)NMR instrument operated at 400 MHzand 101 MHz, respectively. Chemical shifts (δ) are reported in parts per million (ppm) relative to residual solventpeaks.

Synthesis of 2,3-dibromobutadiene

An overview of the synthesis of 2,3-dibromobutadiene is shown in Fig. S1.

Br

Br

Br

Br

Br

Br

Br

Br

4 eq. DBU2 eq. NaI

Et2O

FIG. S1. Overview of the synthesis of 2,3-dibromobutadiene.

In a 500 ml three-neck round bottom flask 1,2,3,4-tetrabromobutane (12.0 g, 32.12 mmol, 1 eq) were suspendedin diethyl ether (150 ml, extra pure, stabilized with BHT). Sodium iodide (9.64 g, 64.24 mmol, 2 eq., anhydrous> 99.5 %) was added to the suspension. A flow of nitrogen was continuously passed through the flask during theentire operation.

By making use of a dropping funnel, 1,8-Diazabicyclo[5.4.0]undec-7-ene (19.2 ml, 1.29 mol, 4 eq.) was slowly addedto the suspension. During the addition the formation of a dense yellow suspension was observed. The dropping funnelwas rinsed with diethyl ether (30 ml) and the reaction mixture was allowed to stir for 1 hour. Then, the precipitatewas filtered off over a frit and was washed with diethyl ether (3×20 ml). The organic phase was washed with saturatedammonium chloride (3× 100 ml), distilled water (1× 100 ml), and brine (1× 100 ml). The organic phase was driedwith MgSO4 and concentrated using a rotatory evaporator at 0 °C and reduced pressure to yield the product DBB(3.74 g, 55 %). The product was observed to react violently with excessive heat formation when in contact with airat room temperature. In order to prevent the sample from degradation and self-polymerization it was stored below−78 °C.

1H NMR (400 MHz, CDCI3): δ 6.43 (d, J = 1.5 Hz, 2H), 5.89 (d, J = 1.5 Hz, 2H)

13C NMR (101 MHz, CDCI3): δ 125.13, 124.75

LRMS (EI): calculated for C4H4Br2: 211.87, found: 211.85

12

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

f1(ppm

)

02000

4000

6000

8000

10000

12000

A(ddd)

4.61

B(ddd)

4.17

C(ddd)

3.97

2.10

2.25

2.00

3.953.963.963.963.983.983.993.994.014.134.154.164.164.164.184.194.194.194.594.604.604.614.614.614.624.624.624.634.64

1 HNMR(400MHz,Chloroform-d)δ4.61(ddd,J=3.8,

2.3,1.4Hz,1H),4.17(ddd,J=11.7,2.5,1.4Hz,1H),

3.97(ddd,J=11.7,2.4,1.3Hz,1H).

1

23

4Br 5

Br 6

Br 7

Br 8

H9

H 10

H 11H 12

H 13

H 14Br 1

2

34

5

Br 6

H 7H 8

H 9 H 10

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

6.5

7.0

7.5

f1(ppm

)

05000

10000

15000

20000

25000

30000

35000

40000

A(q)

6.94

B(d)

6.75

C(d)

6.43

D(ddd)

6.35

E(d)

5.89

F(tt)

5.83 G(dt)

5.69

H(dt)

5.38

0.40

0.380.301.91

0.402.00

0.17

0.30

5.375.375.375.395.395.405.635.665.665.665.675.675.675.685.705.715.715.715.825.825.825.835.835.845.845.845.865.865.865.885.885.895.895.895.906.316.316.346.346.356.356.366.376.376.386.386.386.426.426.436.436.436.446.726.756.756.766.766.796.796.806.806.836.836.946.946.946.94

1 HNMR(400MHz,Chloroform-d)δ6.94(q,J=0.7Hz,0H),6.75(d,J=2.7Hz,0H),6.43(d,J

=1.5Hz,2H),6.35(ddd,J=16.4,10.5,0.7Hz,0H),5.89(d,J=1.5Hz,2H),5.83(tt,J=7.3,

1.6Hz,0H),5.69(dt,J=16.3,0.7Hz,0H),5.38(dt,J=10.4,0.7Hz,0H).

FIG

.S2.

Analysis.

1H

NMR

analysis

ofthereactant

1,2,3,4-tetrab

romob

utan

e(bottom)an

dtheprod

uct2,3-dibrom

obuta-1,3-dien

e(top

).