Electrostatic catalysis of a Diels–Alder reaction

It is often thought that the ability to control reaction rates with an applied electrical potential gradient is unique to redox systems. However, recent theoretical studies suggest that oriented electric fields could affect the outcomes of a range of chemical reactions, regardless of whether a redox system is involved. This possibility arises because many formally covalent species can be stabilized via minor charge-separated resonance contributors. When an applied electric field is aligned in such a way as to electrostatically stabilize one of these minor forms, the degree of resonance increases, resulting in the overall stabilization of the molecule or transition state. This means that it should be possible to manipulate the kinetics and thermodynamics of non-redox processes using an external electric field, as long as the orientation of the approaching reactants with respect to the field stimulus can be controlled. Here, we provide experimental evidence that the formation of carbon–carbon bonds is accelerated by an electric field. We have designed a surface model system to probe the Diels–Alder reaction, and coupled it with a scanning tunnelling microscopy break-junction approach. This technique, performed at the single-molecule level, is perfectly suited to deliver an electric-field stimulus across approaching reactants. We find a fivefold increase in the frequency of formation of single-molecule junctions, resulting from the reaction that occurs when the electric field is present and aligned so as to favour electron flow from the dienophile to the diene. Our results are qualitatively consistent with those predicted by quantum-chemical calculations in a theoretical model of this system, and herald a new approach to chemical catalysis.

Electrostatic catalysis is the least developed form of catalysis in syn thetic chemistry (even though it is widely harnessed by enzymes [13][14][15][16] ). This is because electrostatic effects are strongly directional and are effectively quenched in polar media. Enzymes overcome these prob lems by using a lowpolarity active site, in which the substrate binds in a precisely oriented manner; one or more charged residues within this site can then create an oriented local electric field that can catalyse the reaction. In synthetic chemistry, one can mimic this process to some extent by using charged functional groups on the substrate or catalyst; however, balancing the need for low solvent polarity with the limited solubility of charged residues in nonpolar solvents leads to compro mises that weaken the catalytic effect. For example, aminoxyl radicals (R 1 R 2 NO • ) are stabilized via resonance with R 1 R 2 N +• O − . Thus, when a (remote) negatively charged functional group is placed on the lefthand side of the N-O • bond, the electrostatic stabilization of this minor con tributor leads to further stabilization of the species 2,3 . This stabilization, which has been verified experimentally, promotes dissociation of the R 1 R 2 NO-H and R 1 R 2 NO-R bonds by as much as 20 kJ mol −1 in the gas phase 2 ; however, the effect, while still of practical significance as a 'pH switch' of radical stability, is effectively halved in energetic terms in lowpolarity solvents such as dichloromethane 17 , and essentially quenched in polar solvents 18 .
If one could use external electric fields instead of charged chemical species as the 'catalyst' , one could manipulate a much broader range of reactions, conveniently altering both reactivity and selectivity in a tunable manner that can be predicted by theory. However, to probe this concept experimentally, one must develop a method of controlling the orientation of the external electric field (EEF) with respect to the reaction centre. Previously, EEFs have been used to guide the selec tivity of isomerization reactions in which polar intermediates or tran sition states are involved 19,20 ; but controlling the orientation of the EEF as two molecules collide in bimolecular reactions adds another dimension to the problem. Here we show that this can be achieved by combining surface chemistry procedures with stateoftheart single molecule techniques that are based on scanning tunnelling microscopy (STM). STMbased singlemolecule electrical meas urements can reveal information on chemical coupling averaged over thousands of collisions. This gives us the ability to control the dynamics of the approaching reactants and deliver the field stimulus upon collision. Using this approach, we show that a simple, textbook bimolecular carbon-carbon bondforming reaction, the Diels-Alder reaction-involving reagents of ostensibly negligible polarity-can be accelerated by an oriented EEF.
Diels-Alder reactions-which involve a conjugated diene and a substituted alkene (the 'dienophile')-constitute a major family of chemical processes that are used in the preparation of fine chemicals 21 . Our choice of these reactions was inspired by theoretical predic tions by Shaik and colleagues 1 , who suggested that the barrier heights for certain Diels-Alder reactions can be lowered substan tially when an electric field is oriented appropriately. Here we use a Letter reSeArCH surfacetethered furan derivative as the diene, and a norbornylogous bridge with a terminal double bond as a nonpolar dienophile ((±) NB, tricyclo[4.2.1.0 2,5 ]non7ene3,4dimethanethiol; only the (1(R),2(R),3(R),4(R),5(S),6(S))-enantiomeric form is shown in Fig. 1a). Norbornylogous bridges are conformationally rigid molecules, and have been extensively used as electrical conduits for probing how geometrical and structural factors influence chemical and electro chemical phenomena [22][23][24] . The NB in Fig. 1a is a short, rigid, nonpolar dienophile with two CH 2 SH groups (feet) in trans-stereochemistry; the presence of these feet allows unambiguous orientation of the distal double bond (dienophile) when assembled on flat gold surfaces 23,25 . The rigidity of NB helps us to position and align the dienophile with respect to the EEF when the diene part of the system is brought nearby. The molecular length of the dienophile is kept to a minimum (five sigma bonds) in order to maintain the Diels-Alder product within the conductance limit of our system.
To ascertain whether this Diels-Alder reaction is sensitive to the presence of an oriented EEF, we used quantum chemistry to study the field effect on the reaction barrier. This reaction has four structurally distinct Diels-Alder products; results for the reaction with the lowest barrier (exo-syn) are shown in Fig. 2 (see Supplementary Information for full results). There are two diastereoisomers for each of the four products, with the substituents of the furan being located either on the left of the molecule (see the blue diastereomer in Fig. 2a) or on the right (the red diastereomer).
The positioning of these substituents leads to different interactions with the CH 2 SH groups at the opposite end of the NB, resulting in slightly different energies when no EEF is present and very different responses to the applied field. Experimentally, the NB is known to sit at Figure 1 | Electrostatic catalysis of a Diels-Alder reaction. a, We studied the effects of an external electric field on the reaction rate by using single molecule STMBJ conductance measurements, which provide the oriented electricfield stimulus and also count reaction events. The diene (a furan) is attached to the STM tip via a thiol group ('S'); the dienophile (a norbornylogous bridge) is attached in a known orientation 26 to a flat gold surface via two thiols. Four structurally distinct products may be formed, each having two diastereoisomers; the kinetically favoured product is shown here. ΔV is the voltage difference between the tip and surface electrodes. b, Possible resonance structures of the transition state. When an electric field is present, minor contributors I or III may be stabilized enough to undergo resonance with II, lowering the reaction barrier. The vertical arrows show the field direction most likely to stabilize I or III, with I expected to experience greater stabilization at a given field magnitude. These were the kinetically favoured products; the six other possible products had much higher reaction barriers over the experimental range of field strengths. In the blue diastereomer, the substituents of the furan are located on the left of the molecule; in the red diastereomer, these substituents are located on the right. b, The coordinate axes used to orient the field with respect to the molecule. The z axis lies along the double bond of the dienophile, while the x axis is directed along the NB backbone. c, The scenario being modelled, showing the NB bridge sitting in the experimentally determined orientation with respect to the surface of the STM plate and to the electric field lines, which are passing through the reaction centre at an oblique angle to the NB double bond. d, The predicted effects of the strength and direction of the external electric field (EEF) on the reactionbarrier height (ΔE ‡ ) for formation of the two exo-syn diastereoisomers in a (see Supplementary Information). The formation of the blue structure is quite insensitive to the EEF over the experimental range of field strengths, while the formation of the red structure shows strong field sensitivity. The diagrams in the bottom right corner show the possible directions of the electric field (where E surface denotes the voltage applied between the surface and the tip). Letter reSeArCH an angle to the gold surface 26 , tilted by 30° along the y axis and 25° along the z axis (Fig. 2c). For the exo-syn product, the 30° yaxis tilt means that the field lines are oriented roughly along the average vector of the forming bonds. However, the zaxis tilt means that the furan SH group sits above the forming bonds for the blue diastereoisomer (Fig. 2c), but to the side of the molecule for the red diastereoisomer. As a result, the blue structure is predicted to be quite insensitive to the EEF over the experimental range of field strengths, while the red structure should show strong field sensitivity (see Supplementary Information, section 3). Specifically, it is predicted that, for negative bias, the barrier to for mation of the red isomer will decrease with increasing field strength; for positive bias, it is predicted to increase (Fig. 2d). This trend occurs because the negatively biased EEF can stabilize resonance contributor I (Fig. 1b); a positively biased field will destabilize it. In principle, a pos itively biased field should also lower the barrier height by stabilizing resonance contributor III. However, configuration III has much less inherent stability than contributor I, as the electronegative oxygen prefers to bear a negative rather than a positive charge. As a result, this configuration contributes only at strong positive fields, outside of the experimental range (see Supplementary Figs 3 -8). Within the experi mental range, our calculations predict that the frequency of adduct for mation should increase systematically as the strength of the negatively biased field increases, up to a factor of 1.5 at −0.75 V, while remaining relatively independent of EEF strength for positively biased fields.
To test these predictions experimentally, we attached the NB to the surface of a flat gold electrode, and the furan to the STM gold tip. We then undertook a series of STM breakjunction approach (BJ) experiments known as 'blinking' 7,27 . The blinking technique detects the formation of molecular bridges between an STM tip and a sub strate electrode-while they are fixed at a specific electrode-electrode distance-by imposing an initial setpoint tunnelling current ( Fig. 3 and Supplementary Information, section 2). After the setpoint current is reached, the feedback loop is turned off and the current is monitored. Current jumps (blinks) appear when a molecular bridge spans the gap between the electrodes (Fig. 3c). We detected blinks in conductance of magnitude (1.5-2)× 10 −6 G 0 when the tip and substrate were separated by a distance that allows the Diels-Alder reaction to occur (about 1 nm). These junctions formed only when both reactants were present. When either reactant was removed from its respective electrode, or when their saturated analogues were used (2methyl3tetrahydrofuranthiol on the tip, or a hydrogenated version of NB on the surface-a system that is structurally identical but which lacks the diene-dienophile character), there was no evidence of molecularbridge formation ( Supplementary Information, section 2). Hence, the junctions are formed through the Diels-Alder reaction. At positive voltage biases (sur face positive), the frequency of molecularbridge formation is constant (five blinks per hour) over a wide range of applied biases. In contrast, at negative biases the frequency is clearly affected by the strength of the field, and increases from five blinks per hour at a bias of −0.05 V, to 25 blinks per hour at a bias of −0.75 V (Fig. 4). These trends are in com plete qualitative agreement with the theoretical predictions in Fig. 2d. Quantitatively, there are differences, which may relate to the differ ence in the realms being studied experimentally and computationally (singlemolecule reaction rates versus bulk reaction rates), and/or the use of precomplexes in calculating field effects on barrier heights (see Supplementary Information).
As further validation that the experiment is detecting the forma tion of carbon-carbon bonds, we note that the average lifetime of the blinks was 0.4 s (Fig. 3d), with poor dependence on the electric field magnitude for positive biases up to 0.75 V ( Supplementary  Information, section 2). This lifetime is around the same as that observed for standard singlemolecule wires that are thiolated at both ends 6,24 . A bias value higher than +0.75 V or lower than −0.75 V led to a drop in the lifetime of the junctions, owing to the instability of the gold-sulfur contacts. Hence, we kept the upper limit of the bias within the range +0.75 V to −0.75 V, to allow us to compare different biases while maintaining similar junction stabilities. We also confirmed the formation of mechanically stable carbon-carbon bonds by collecting pulling curves during the blinking events ( Supplementary Figs 2-6), or by performing pushing/pulling cycles ( Supplementary Figs 2-7). Pulling curves collected over the blinks showed a plateau with an average pulling length of 0.2 nm, which corresponds to the stretching of the singlemolecule bridge. When the pulling was exerted over the tunnelling background or on random noise ( Supplementary Figs 2-6), a clean exponential decay was observed, testifying that the above blinks are attributable to a stable molecular junction, rather than resulting from the migration of gold atoms or from fluctuations in molecular conformations 28,29 .
We further confirmed that robust molecular junctions were formed using an STMBJ approach 5,30 referred to as tapping. Here, the furanmodified tip was repeatedly driven into and out of contact with the NB-modified substrate ( Supplementary Figs 2-7): when the reac tants were mechanically brought together, molecular junctions formed; when the tip was pulled away, the junctions broke. During the pull ing portion of the cycles, we detected plateaus in the currentversus distance curves of the same conductance magnitude as that observed in the blinking experiments ((1.5-2)× 10 −6 G 0 ; Supplementary Information, section 2 and Supplementary Figs 2-7), further support ing the idea that stable molecular wires form when the two reactants are brought together under an electric field. Moreover, changes to the mag nitude and direction of the field applied across the reactants affected the rate of the Diels-Alder reaction in a similar manner to that found via

Letter reSeArCH
blinking. Tapping data show that product formation increases by 4.4 fold, from 4.2% (252 product molecules out of 6,000 attempts) to 18.6% (1,116 product molecules out of 6,000 attempts) when the surface bias is changed from −0.05 V to −0.75 V ( Supplementary Figs 2-7).
We have presented the first (to our knowledge) experimental evidence of a nonredox, bondforming process being acceler ated by an oriented EEF. Our experimental results are qualitatively consistent with theoretical calculations, and result from the abil ity of the electric field to electrostatically stabilize a minor charge separated resonance contributor of the transition state. This ability to manipulate chemical reactions with electric fields offers proofofprinciple for a change in our approach to heterogeneous catalysis.  To keep the distance between the surface and the tip constant across the bias range, we used the same setpoint current and changed the bias while the STM feedback was turned off. We performed blinking experiments over periods of one hour. At the end of each onehour period, we changed the furanmodified STM tip and its lateral position with respect to the surface to compensate for the loss of reactants. We repeated this procedure eight times, giving each bias point (magnitude and direction) a total experimental time of eight hours. We changed the chronology of the selected biases randomly for each repeat. Error bars represent the standard deviation from the eight (onehour) intervals. The dashed lines are included for visual guidance.