Ilana B.
Stone‡
a,
Rachel L.
Starr‡
a,
Norah
Hoffmann‡
a,
Xiao
Wang‡
b,
Austin M.
Evans
a,
Colin
Nuckolls
a,
Tristan H.
Lambert
c,
Michael L.
Steigerwald
a,
Timothy C.
Berkelbach
*ab,
Xavier
Roy
*a and
Latha
Venkataraman
*ad
aDepartment of Chemistry, Columbia University, New York, New York 10027, USA. E-mail: tcb2112@columbia.edu; xr2114@columbia.edu; lv2117@columbia.edu
bCenter for Computational Quantum Physics, Flatiron Institute, New York, New York 10010, USA
cDepartment of Chemistry and Chemical Biology, Cornell University, Ithaca, New York 14853, USA
dDepartment of Applied Physics, Columbia University, New York, New York 10027, USA
First published on 1st September 2022
The electric fields created at solid–liquid interfaces are important in heterogeneous catalysis. Here we describe the Ullmann coupling of aryl iodides on rough gold surfaces, which we monitor in situ using the scanning tunneling microscope-based break junction (STM-BJ) and ex situ using mass spectrometry and fluorescence spectroscopy. We find that this Ullmann coupling reaction occurs only on rough gold surfaces in polar solvents, the latter of which implicates interfacial electric fields. These experimental observations are supported by density functional theory calculations that elucidate the roles of surface roughness and local electric fields on the reaction. More broadly, this touchstone study offers a facile method to access and probe in real time an increasingly prominent yet incompletely understood mode of catalysis.
In this work, we report the Ullmann coupling of aryl iodides on a Au surface submerged in solution under ambient conditions. The surface also serves as a scanning tunneling microscope-based break junction (STM-BJ) substrate, thereby enabling in situ determination of product formation by monitoring single-molecule junction conductance. STM-BJ measurements provide molecule-specific signatures (i.e. the most probable molecular conductance and molecular plateau length) that can be used to identify carbon–carbon bond formation in situ.19 Our observations are consistent with ex situ fluorescence spectroscopy and mass spectrometry measurements that confirm biaryl formation across a range of aryl iodide precursors, including those with strongly coordinating groups such as thiomethyls, pyridines, and amines. Notably, in addition to a Au surface, we find that the Ullmann coupling requires an Au surface that is rough at least on the 10–100 nm scale, as measured by AFM (ESI Fig. S1†), and a highly polar solvent. Furthermore, the reaction works primarily with aryl iodides. Finally, we probe the mechanism of the Ullmann coupling and propose that interfacial electric fields contribute to the reaction catalysis, which is distinct from the mechanisms commonly invoked for solution-based Ullmann couplings.12,16,20–24
We first characterize, in situ, the formation of the homo-coupled dimer (D1, 4,4′-diamino-quaterphenylene) when a solution of 4-iodo-4′-amino-biphenyl (I1) is subject to STM-BJ measurements from a solution deposited on a Au surface (Fig. 1a). A 0.1 mM solution of I1 in propylene carbonate (PC), a highly polar solvent with a dielectric constant of ∼64 at 25 °C,25 is placed on a rough Au STM-BJ substrate prepared by evaporating 70–100 nm of Au on a steel puck (see ESI Fig. S1† for SEM and AFM images). The reaction is monitored by performing STM-BJ measurements over a period of 18 hours using an insulated Au STM tip (see ESI† for details),26 while applying a bias voltage of 100 mV. Fig. 1b presents conductance histograms created over the measurement period showing a molecular conductance peak centered at 3.7 × 10−5G0 (G0 = 2e2/h is the conductance quantum). This conductance value is comparable to previous measurements of tetraphenyl diamine D1 (ESI Fig. S2a†).27 The 2D conductance–displacement histogram shown in Fig. 1c reveals that this conductance peak also features a molecular plateau extending to ∼0.7 nm, which is considerably longer than the length expected for a biphenyl (see Fig. S2† for additional 2D histogram).27 The longer plateau length observed in situ is consistent with that of D1,27 indicating that D1 is formed under these conditions (Fig. 1c). The lower conducting D1 forms via an Ullmann coupling of two I1 molecules, which results in a peak whose height increases over time as is seen in Fig. 1b.
We next demonstrate that the coupling reaction happens on the Au surface without performing STM-BJ measurements.28 An aliquot of a solution of I1 and TBAPF6 prepared in PC was deposited on an Au substrate. Samples of the solution were subsequently collected after 2, 5 and 10 hours and measured by fluorescence spectroscopy. If the formation of D1 required the STM measurements, we would not detect any product with this experiment through fluorescence measurements. Fig. 2a and b show fluorescence spectra measured using different substrates and conditions with those of an I1 solution that was not exposed to the Au substrate and a solution of the product D1 synthesized ex situ. These spectra are measured using 284 nm and 340 nm photoexcitation, respectively, with wavelengths chosen based on the electronic absorption spectra of I1 and D1 measured ex situ, (ESI Fig. S3a†). The spectrum reveals that after 2 hours of exposure to the Au substrate in an ionic solution, in addition to the I1 peak at 380 nm, a smaller peak appears at 455 nm corresponding to D1 (Fig. 2a, light blue). Over the course of 10–12 hours, the emission peak consistent with that of D1 increases in intensity (Fig. 2a, dark blue), indicating that dimerization is occurring when I1 is exposed to an interfacial field on a rough Au substrate. Starting with a 100 μM solution of I1, a solution was obtained that has 6.25 μM D1 after 12 hours (Fig. 2b, blue), showing that the conversion is significant.
Fig. 2 (a) Emission spectra (λex = 284 nm) of a solution of I1 placed on a rough Au-coated steel substrate over time (blue traces) overlaid with reference peaks for I1 (yellow trace) and an ex situ synthesized solution of D1 (red trace). These traces are normalized to have the same peak height at 380 nm. (b) Emission spectra (λex = 340 nm) showing a peak for D1 that appears in solution left on a rough Au-coated steel substrate (blue trace) overlaid with reference peaks for I1 (yellow trace) and D1 (red trace). Also shown are data from a solution is left on a smooth Au-coated mica substrate (green trace) and with an added suspension of Au dust (purple trace). (c) Selected ion chromatogram (LC-MS) of the I1 standard (red), the D1 standard (blue), and the solution subject to STM-BJ measurements for 12 hours (green). Masses obtained from these spectra are noted in the legend and shown in ESI Fig. S4.† (d) Molecular structures of aryl iodides studied here I1–I4. |
To elucidate the role of the Au substrate in the homocoupling of the aryl iodides, we show, using ex situ fluorescence spectroscopy, that the coupling reaction does not proceed directly on a steel substrate without Au coating (ESI Fig. S3c and d†), indicating that the Au catalyzes the reaction. Similarly, when the rough Au-coated substrate is replaced with an atomically flat Au-coated mica substrate, no dimer formation is observed over 12 hours by fluorescence spectroscopy (Fig. 2b, green trace) or after 12 hours of measurement in the STM-BJ setup (ESI Fig. S3b†). We note here that although the tip and Au-coated mica substrate are locally roughened during the STM-BJ measurement, the number of undercoordinated sites created is likely insufficient to generate a detectible amount of the product. Finally, we also observe dimer formation on an Au-coated glass substrate, which produces a rough gold layer also catalyzes the reaction (ESI Fig. S3b†). Our experiments therefore indicate that a rough Au surface is required and while also eliminating the possibility that steel on its own is catalyzing the reaction. The rough surface thus provides the active catalytic sites, similar to surface-catalyzed reactions under UHV that happen primarily at facets or defects.29–31 These results are further supported by analysis of solutions subjected to STM-BJ measurements via Liquid Chromatography Mass Spectrometry (LC-MS) spectra shown in Fig. 2c. We find that the chromatogram shows a peak at 3.6 minutes when the product D1 is present in the solution while we see a peak at 3.1 minutes when the reactant I1 is present. The reactant and product are identified by their respective masses extracted from these chromatograms and shown in ESI Fig. S4.†
We next evaluate the role of the of the solvent and ions in solution. Although the reaction proceeds in the polar solvent PC and in PC with an added electrolyte (Fig. 2b), when a solution of I1 in the nonpolar solvent trichlorobenzene is deposited on the Au substrate, no peak corresponding to D1 is observed in either STM-BJ measurements or fluorescence spectroscopy (ESI Fig. S5†), demonstrating that the solvent polarity is critical. As shown by Dawlaty and co-workers,1 molecules at the interface of a metal and solvent experience a reaction field whose magnitude is a function of the solvent dielectric constant. Specifically, they show that for a solvent with a dielectric constant over 50 (as is the case with PC), the interfacial fields can be as large as 2 V nm−1. Our experimental observations suggest that such interfacial electric fields are contributors to the catalysis of the Ullmann coupling. We note that identical results are obtained when solutions are created with an added electrolyte (such as tetrabutyl ammonium hexafluorophosphate (TBAPF6) at a 0.1 M concentration) indicating that added ions do not alter the interfacial fields significantly, consistent with previous measurements.32
Finally, when the Au surface is replaced with Au dust (diameter: 500–850 nm) suspended in a reaction mixture which contains I1 dissolved in PC with 0.1 M TBAPF6, the fluorescence spectra (Fig. 2b) indicate no dimer formation. Under these conditions, I1 is exposed to a rough surface comparable in total surface area to the Au substrate, yet no coupling reaction is observed. These results could indicate weaker reaction fields at the interface of a small metallic sphere; however, we cannot rule out other possibilities, including different surface morphologies, fewer undercoordinated sites or inhibited surface diffusion, adsorbed moieties that prevent the reactants from interacting with Au.
The observed homocoupling reaction is generalizable with respect to both the linker and molecular backbone. Three additional aryl iodides (Fig. 2d) were studied under the same conditions as I1, and all appear to homocouple. The coupling of both the thiomethyl- and pyridyl-terminated biphenyl iodides (I2 and I3, respectively) can be probed in situ through STM-BJ measurements because they are equipped with anchoring groups necessary for metallic contact. Similar to the dimerization of I1, when each aryl iodide is measured, a lower molecular conductance peak with a longer molecular plateau appears and continues to grow over time (ESI Fig. S6†). We verify the identity of these products by comparison of the 1D and 2D conductance histograms for D2 and D3 synthesized ex situ and measured independently (Fig. S6†). LC-MS spectra of the extracted solutions also confirms biaryl coupling (ESI Fig. S7†). We note that while these specific iodides were selected for their aurophilic substituents, they demonstrate that both strongly electron withdrawing and electron donating groups are tolerated in this system. Furthermore, even the larger molecule I4 homocouples on the rough Au surface despite its steric encumbrance, as verified by electronic absorption spectroscopy of the extracted solution (ESI Fig. S7†). We note that this particular coupling cannot be probed using the STM-BJ because the large twist angle between adjacent anthracene units in the product decreases its conductance below the measurable limit of our instrument.33,34
We propose a general scheme for the coupling of aryl iodides in an interfacial field (Fig. 3a). We probe this reaction scheme with mechanistic and computational studies described further below. Two aryl iodides undergo an oxidative addition forming two independent C–Au–I constructs at active sites on the rough Au substrate. If two oxidative addition events occur near one another, a small amount of surface migration may occur before a reversible reductive elimination can release the aryl iodides back into solution.35–37 Once the active Au–C bonds are close enough, a C–C bond can form between the activated aryl species, presumably through a mechanism that resembles that of thermally driven reactions on metal surfaces.16,21,38 The resulting tetraphenyl product is then released into solution, likely leaving iodine chemisorbed to the Au surface. We note that this double oxidative addition mechanism differs from the proposed mechanisms for solution-based Ullmann couplings in which the two halides have distinct roles in the coupling process: one aryl iodide free in solution is activated by the metal catalyst, thus forming an organometallic complex with which the second free aryl iodide can interact.39,40
Within this reaction scheme, oxidative addition is proposed to be the key step catalyzed by the rough Au surface and the interfacial field. We can probe this step by comparing the reactivity of aryl iodides with that of a preformed Au complex. We synthesized the Au complex Au2 (ESI Fig. S8a†), which has a preformed Au–C bond and a labile PPh3 ligand that readily dissociates in situ.41 This Au complex can be understood as an analog of the product of an oxidative addition of biphenyl iodide I2 to Au, as the dissociation of PPh3 replaces the oxidative addition step such that starting with Au2 instead of an aryl iodide bypasses the oxidative addition step. Indeed, when Au2 is measured in the junction in a non-polar solvent, a molecular conductance peak for D2 appears immediately (ESI Fig. S8b†). Furthermore, when Au2 is left on a smooth mica substrate in polar solvent, D2 is observed in the solution via fluorescence spectroscopy (ESI Fig. S8c†). Together, these results prove that when the oxidative addition step is removed from the reaction scheme altogether, neither a rough surface nor a polar solvent is necessary to the coupling. As such, we conclude that the oxidative addition is the rate limiting step in this transformation and is the key process that is assisted by the rough surface and its interfacial electric field.
For further insights into the oxidative addition process, we turn to electronic structure calculations using density functional theory (DFT); full Computational details are provided in the ESI.† For simplicity, we study the dehalogenation of iodobenzene, whose behavior is representative of other aryl halides. We emphasize that because of the idealized nature of our calculations (especially the surface morphology and absence of explicit solvent), we do not expect quantitative agreement with experimental findings, but rather qualitative insights. Using the nudged elastic band method with 6–12 images, we calculated the minimum energy reaction pathway on an Au(111) surface and on an Au(111) with a single Au adatom (Fig. 3b), as a model for surface roughness. On Au(111), we find an activation energy of 0.76 eV (17.5 kcal mol−1) and a reaction energy of −0.4 eV (−9 kcal mol−1), in good agreement with previous calculations.42 In contrast, in the presence of an Au adatom, which represents an extreme under-coordinated structure, we find a significantly reduced activation energy of 0.18 eV (3.5 kcal mol−1) and more negative reaction energy of −0.6 eV (−13 kcal mol−1).43 The reduced activation energy supports the experimental observation that a rough Au surface, where a fraction of Au atoms on step-edges are undercoordinated (though not as undercoordinated as the adatom shown in Fig. 3b) is essential for catalyzing the Ullmann coupling. The difference in activation energies can be understood by comparing the transition state geometries (Fig. 3b, inset). Compared to Au(111), the Au adatom is more readily inserted into the C–I bond.
Next, we analyze the potential impact of interfacial electric fields on the dehalogenation. In the direction perpendicular to the surface, we apply an electric field of −3 V nm−1 and +3 V nm−1, which is comparable to the interfacial field strengths previously measured1,44 and calculated45 at metal–solvent interfaces. Over this range, we find that the electric field modifies the activation energy by about 0.05–0.1 eV (∼0.5–2 kcal mol−1). We see that a positive electric field (i.e. pointing up from the surface) decreases the activation energy on a pristine surface but increases the activation energy in the presence of an adatom modestly. This contrasting behavior is explained by a difference in the relative dipole moments of the reactant and transition state structures (all dipole moments are directed away from the surface and are thus stabilized by a field pointing in the same direction): on the pristine surface, the dipole moment of the reactant is smaller than that of the transition state, and in the presence of an adatom, the reverse is true. More discussion about this behavior and its implications for a linear free energy relationship can be found in Hoffmann et al.46 These results indicate that both surface roughness and local electric fields are important in determining the kinetics of surface-mediated oxidative addition. We expect an interfacial reaction field to point in the direction of the dipole moment, which is away from the surface.1 Our calculation results show that a field in this direction could either raise or lower the activation energy, depending on the precise local morphology as indicated in Fig. 3b.
To demonstrate that the coupling reaction may involve two independent oxidative additions, we compare the reactivity of a bromide- and triflate-terminated biphenyl (B1 and T1, respectively) (Fig. 4a), which are less prone to oxidative addition than aryl iodides. Indeed, DFT calculations of the dehalogenation of bromobenzene in the presence of an Au adatom predict an activation energy of 0.5 eV (12 kcal mol−1), which is much larger than that for iodobenzene (ESI Fig. S9†). Consistent with our hypothesis of multiple oxidative additions to the Au surface, these less reactive reactants do not homocouple in the junction over the time scale studied. Although these reactants are unlikely to form two C–M–X constructs near each other, when surrounded by reactive iodides, only one such construct is necessary. Based on the hypothesis that two oxidative additions are necessary for aryl–aryl coupling, we suspect an Au-bound intermediate is more likely to be formed by a bromide than a triflate, the latter being less prone to oxidative addition, and thus B1 is more likely to cross-couple with I1 than T1 (Fig. 4a). We note that the methyl group meta to the linker on B1 and T1 is used to distinguish between the cross-coupling product CC1 and the homocoupling product D1.
When B1 and I1 are exposed to the reaction conditions described above, dimerization is observed. A molecular conductance peak for D1 appears at 3.7 × 10−5G0 in the 1D histogram after 12 hours (Fig. 4b, pink peak). After 18 hours, this peak becomes broader and asymmetric, with a shoulder on the left side (Fig. 4b, dark blue trace). The first 100 traces after the appearance of the shoulder were isolated, and two distinct peaks were resolved at 3.7 × 10−5G0 and 1.9 × 10−5G0 (Fig. 4b, light blue trace). A lower conductance peak is expected for the cross-coupled product, CC1 (Fig. 4a), due to the presence of the methyl group and consequently larger twist angle between the adjacent phenyl rings.34 We therefore assign this new conductance peak to the putative product of the cross-coupling reaction, CC1. To further validate these results, we synthesized CC1ex situ and compared its conductance value (Fig. 4b, green peak) with that of the molecule formed in situ. The alignment of the two peaks supports our hypothesis that the lower conducting peak at 1.9 × 10−5G0 can be assigned to the cross-coupled product (Fig. 4b, green trace). Finally, when we repeat these experiments with T1, no such peak corresponding to CC1 appears (ESI Fig. S10†). The order of reactivity of aryl bromides and sulfonates is highly dependent on the mechanism of the reaction, and aryl bromides are more prone to oxidative addition than aryl sulfonates.47 Although the results with I1 alone do not exclude the possibility of the aryl iodide interacting with C–M–X to form D1, these control measurements indicate strongly that the coupling follows two oxidative additions. This is because T1 should be as reactive (if not more reactive) towards that reaction T1 + C–M–X = CC1 than the aryl iodide. Yet, we do not observe any CC1 product in our mixed measurements with T1 and I1. By contrast, B1 is less reactive towards the same reaction (B1 + C–M–X = CC1), yet we see the CC1 product in the mixed measurements of I1 and B1. However, the reactivity of two oxidative additions forming the product (i.e. two C–M–X forming D1 or CC1) would be highest for I1 and lowest for T1 with B1 having an intermediate reactivity. Our experimental results suggest that in this system, the bromide is more reactive than the triflate, although it is possible that the cross-coupling product with T1 was simply never observed in the six measurements taken. The order of reactivity observed in the cross-coupling experiments is therefore consistent with our hypothesis that two independent oxidative addition steps are involved in the formation of aryl–aryl coupling products.
In this report, we have shown that the Ullmann-type coupling of aryl iodides can be catalyzed by a rough Au surface submerged in a polar solvent under ambient conditions. Through a combination of in situ STM-BJ measurements, ex situ mass spectrometry, fluorescence spectroscopy, and first-principles calculations, we have gained mechanistic insight into these transformations that occur by a double oxidative-addition mechanism. Taken together, this proof-of-principle study also suggests that STM-BJ may provide a valuable method to interrogate surface-based reactivity in real-time. More broadly, this investigation connects interfacial electric fields to chemical reactivity and reiterates the promise of electric-field catalysis that could be leveraged for a wide number of molecular transformations.
Footnotes |
† Electronic supplementary information (ESI) available. See https://doi.org/10.1039/d2sc03780g |
‡ Equal contribution. |
This journal is © The Royal Society of Chemistry 2022 |