Identifying key descriptors in surface binding: interplay of surface anchoring and intermolecular interactions for carboxylates on Au(110)† †Electronic supplementary information (ESI) available: Supporting experimental methods and supporting discussion are included in the supplementary information. See DOI: 10.1039/c7sc05313d

The relative stability of carboxylates on Au(110) was investigated as part of a comprehensive study of adsorbate binding on Group IB metals that can be used to predict and understand how to control reactivity in heterogeneous catalysis.


Procedure for determining the equilibrium constant of reversible displacement reactions in the limit of low coverage from temperature programmed reaction data
In the quantitative analysis, the consideration of the molecular fragmentation pattern, ionization cross-section, and the mass spectrometer transmission and detection coefficients is necessary. It has been shown elsewhere 1 that the number density of molecule in the ionizer, , is given as follows: (Eqn. S1) where is the total ionization cross-section of molecule , is the measured signal current for the th fragment of molecule , is the ratio of signals of the th and th fragments of molecule determined from separate In the equation above, the constants , , , and are taken from published values (Table S1).
is determined for molecule i by condensing a neat sample of molecule on / clean Au(110) and the fragmentation pattern was recorded by TPRS. The equilibrium constant of the displacement reaction (Eqn. S2) can be determined from the forward and reverse direction displacement experiments: Experimentally, the surface concentration of the two carboxylates could be quantitatively determined by the generalized formula (Eqn. S3). The selectivity fraction, i.e. the fraction of carboxylate that forms product molecule , for each adsorbate was determined by independent oxidation experiments.
where is the coverage of RCOO prior to decomposition, is the amount of product molecule produced from the decomposition of RCOO, is the selectivity fraction of RCOO that decomposes to product molecule .
Under the reaction conditions, the desorption temperature of the carboxylic acids (RCOOH, R´COOH) is much lower than the reaction temperature and the rapid pumping of the gas phase species away from the surface gives the rate of displacement as: where is the pre-exponential for the forward reaction (ML -1 ), is the molecular flux of (ML/s), is the coverage of (ML), ( ) is the activation energy of the forward displacement reaction (J/mol), is the gas constant (J/mol K), T is the temperature of the sample.
We know that at , where is the initial coverage of RCOO and at , where is the final coverage of RCOO. The initial coverage of RCOO can be = determined from the total amount of all products desorbed and is confirmed by a calibration dose of oxygen recombination. The final coverage of RCOO can be determined from the amount of product from RCOO detected. Solving the differential equation gives Eqn. S5.
where is change in coverage by the displacement process (ML), Δ is the initial coverage of (ML),  Likewise the displacement reaction was probed from the reverse direction and the same derivation produces the following: where is the activation energy of the reverse displacement reaction (J/mol), is the change in coverage by the displacement process (ML), is the initial coverage of (ML), is the pre-exponential for the reverse reaction (ML -1 ), is the molecular flux of (ML/s), ( ) is the dosing time of molecule (s).

'
( ) The equilibrium constant can be derived from and (Eqn. S7) and thus evaluated from the forward and reverse direction displacement experiments. For clarity, square brackets are used in Eqn. S7 to separate the information measured by each experiment.
Assuming that the pre-exponential factor for the forward and reverse reaction is the same, , ≅1 this simplifies to: (Eqn. S8) For temperatures above the decomposition temperature of an adsorbate the molecular flux cannot be measured directly and needs to be appropriately calibrated by a condensation experiment. At temperatures much lower than the desorption temperature, the adsorption rate of a physisorbed molecule is given as follows: where is the sticking probability.
The molecular flux can be determined by measuring the amount of molecules desorbed from a fixed molecular exposure with a time length (Eqn. S10) This can be simplified by assuming the sticking probability for the condensation is unity, ≅1 (Eqn. S11).
(Eqn. S11) The equilibrium constant can now be determined using the measureable quantities from the displacement (Forward: , , Reverse: , ) and calibration ( , ) experiments (Eqn. S12) Experimental peak deconvolution for Figure 1 The deconvoluted CO 2 signal for acetate is calculated by (Eqn. S13) where is the deconvoluted acetate signal, is the measure CH 3 signal, and is the selectivity ratio of CO 2 to CH 3 measured for isolated acetate.
( 2 ) The deconvoluted CO 2 signal for propanoate is calculated by (Eqn. S14) where is the deconvoluted propanoate signal, is the measure CF 3 signal, and is the selectivity ratio of CO 2 to CF 3 measured for isolated trifluoroacetate.
( 2 ) The sum of the deconvoluted CO 2 signal calculated for acetate and trifluoroacetate is in excellent agreement with the measured CO 2 signal which confirms that the product selectivity is coverage independent below 0.10 ML of acetate and trifluoroacetate.
Experimental results for the stability hierarchy in Table 2.
Formic acid + Acetic acid Figure S1: In separate experiments, ~1 ML of formic acid (black) and acetic acid (red) were dosed on 0.05 ML O/Au(110) at 260 K to establish their reactivity. Then acetic acid was introduced on isolated formate/Au(110) at 260 K (blue), and formic acid was introduced on isolated acetate/Au(110) at 260 K (purple).
In the competitive displacement experiments between formic acid and acetic acid, the product distribution is indicative of a majority amount of acetate on the surface, as evidenced by CO 2 and CH 3 at 580 K. The product distribution indicates that there is a minority amount of formate on the surface, as evidenced by CO 2 and HCOOH at 350 K. The displacement of formate by acetic acid and acetate by formic acid yield a similar product distribution which indicates that the two carboxylate species exist in a reversible equilibrium that favors acetate. Figure S2: In separate experiments, ~1 ML of acetic acid (black) and propanoic acid (red) were dosed on 0.05 ML O/Au(110) at 300 K to establish their reactivity. Then propanoic acid was introduced on isolated acetate/Au(110) at 300 K (blue), and acetic acid was introduced on isolated propanoate/Au(110) at 300 K (purple).

Acetic acid + Propanoic acid
In the competitive displacement experiments between acetic acid and propanoic acid, the product distribution is indicative of a majority amount of propanoate on the surface, as evidenced by CO 2 and CH 2 CH 2 at 550 K. The product distribution indicates that there is a minority amount of acetate on the surface, as evidenced by CO 2 and CH 3 at 580 K. The displacement of acetate by propanoic acid and propanoate by acetic acid yield a similar product distribution which indicates that the two carboxylate species exist in a reversible equilibrium that favors propanoate.
In the competitive displacement experiments between trifluoroacetic acid and propanoic acid, the product distribution is indicative of a majority amount of propanoate on the surface, as evidenced by CO 2 and CH 2 CH 2 at 550 K. The product distribution indicates that there is a minority amount of trifluoroacetate on the surface, as evidenced by CO 2 and CF 3 at 590 K. The displacements of trifluoroacetate by propanoic acid and propanoate by acetic acid yield a similar product distribution which indicates that the two carboxylate species exist in a reversible equilibrium that favors propanoate.

Equilibrium constant determination for acetic acid and propanoic acid
(Eqn. S12) The reaction pair of acetic acid and propanoic acid demonstrates the calculation of the equilibrium constant for a displacement experiment. The exposures of propanoic acid ( ) and acetic acid ( ) used in the displacement experiments were calibrated The surface concentration of propanoate (red data point) and acetate (black data point) for a specified dose of acetic acid as predicted using Eqn. S15 (red curve) and 11 (black curve) respectively. (C) A curve fit of the data according to Eqn. S16 (R 2 =0.956) demonstrates that the pre-factor and activation energy can be treated as constant for the displacement over the entire range of propionate coverage.
The validity of this approach used to determine K was demonstrated by the prediction of the relative surface concentration of acetate and propanoate resulting from a well-defined increase in the acetic acid exposure to adsorbed propanoate ( Figure S4). The concentration of acetate or propanoate was determined from their signature products acetic acid, m/z = 60 and 73, respectively ( Figure S4A, S4B). The surface concentration of acetate or propionate are given by Eqns. S15 and S16, respectively (Eqn. S15) where A is the pre-exponential factor (ML -1 ) and E is the activation energy of the displacement reaction. There is good agreement between the experimentally determined surface concentration of acetate and propanoate and the model curve fit (R 2 = 0.986 and 0.950, respectively) ( Figure  S4B).
The kinetic parameters of the displacement of propanoate by acetate are essentially coverage independent. By rearranging Eqn. S15, the dependence of the change in surface concentration on the activation energy and pre-exponential factor for displacement gives (Eqn. S17) (Eqn. S17) The data fits this relationship well (R 2 = 0.956) ( Figure S4C); this indicates that the preexponential factor and activation energy are essentially coverage independent over the coverage range studied.

Numerical simulations
Initial adsorption geometry explorations were performed to discriminate between monodentate top, bidentate bridge, bidentate top and chelating geometries for formate and propanoate (Figs. S5 and S6 resp.). Clearly, the preferred adsorption geometry is bidentate top for all carboxylates, similarly to what has been reported for acetate. 4 Other geometries are either unstable and therefore relax into bidentate top (arrows) or lead to considerably higher total energy, as indicated in the figures. Figure S5: Exploration of the adsorption geometry of formate. The most stable geometry is bidentate top, chelating is +0.65 eV less stable and other configurations relax into bidentate top.
Lowest-energy, 0.00 eV Figure S6: Exploration of the adsorption geometry of propanoate. The most stable geometry is bidentate top, chelating is +0.63 eV less stable, monodentate top relaxes into monodentate bridge which is +0.73 eV less stable and bidentate bridge relaxes into bidentate top.
Through a combination of strong corrugation of the surface and molecule-surface vdW interactions, a quasi-degeneracy in adsorption site emerges for isolated propanoate. The effect is significantly smaller for acetate. PBE SP calculations in the configuration given in Fig. S7 yield similar energy loss compared to the top-layer adsorption, amounting to 0.41 eV. However, the total energy i.e. including vdW contributions, is 0.12 eV higher compared to vertical for acetate. For propanoate the difference is 3 times smaller (0.04 eV). A vdW contribution of 0.37 eV is extracted for propanoate, which is close enough to compensate for the loss of energy due to suboptimal binding to Au. Acetate exhibits a lower stabilization of 0.29 eV, because it has one fewer methyl group interacting with the surface. In both cases, the molecule-Au surface distances are consistent with a stronger vdW surface-molecule interaction than in its top-layer configuration (Table S2).
Lowest-energy, 0.00 eV Figure S7: Relaxed propanoate (a) and acetate (b) geometries bonded on the second Au layer from the top, in bidentate top configuration. Acetate (propanoate) is 0.12 eV (0.04 eV) less stable than the top-row configuration.
4.0 Table S2: Distance between the C atom and the (111) microfacet in the configuration in Fig. S7. Figure S8: Relaxed configurations for propanoate on (1×2) (A-F) and on (1×1) (G-J) from initial geometry tilted with respect to the OO axis by 0-90˚ (A-D and G-J) and rotated with respect to CC axis by 45-90˚ (E-F). The respective weight of vdW and PBE contributions is represented in (K) by comparing van der Waals corrected (vdW) and PBE electronic relaxation (PBE SP ) as a function of the angle between the surface normal and the carboxyl group in the relaxed structure.

Molecular tilting and rotation for isolated propanoate
The impact of tilting the molecular plane towards the surface-with respect to the surface normal-and internal rotation -with respect to the C 0 C 1 bond-were investigated for isolated carboxylate molecules (Fig. S8). The tilted configurations are never more stable than the upright configuration. In fact, energy differences are small, below ~25 meV or k B T at 300K, in all relaxed configurations, and this observation is valid for both (1×2) and (1×1). The small energy difference over a wide range of angles indicates that the alkyl chain of an isolated molecule is highly mobile around its anchoring point at room temperature which contributes to the configurational entropy of adsorption. In contrast, internal rotation does not play an important role in the stability of the molecule on the surface.
Ionic relaxation of structures where the molecular plane has been tilted with respect to the normal surface always tends to bring the carboxyl group back into alignment with the Au center; this demonstrates the directional nature of the Au-O bond, likely because of an electronic resonance phenomenon. By comparing vdW ionic and electronic relaxed structures to PBE-only electronic relaxation (PBE SP ), as the angle of tilting increases the PBE contribution is reduced (ΔPBE SP >0) but is compensated by an increase in vdW interactions with the surface which indicates that the directional molecular orientation restricts strong interactions of the alkyl groups with the surface (Fig. S8K). For the same initial rotation angle, the relaxed molecule is systematically closer to the surface on the unreconstructed 1×1 surface which necessitates large angles for the molecule to feel the presence of the 1×2 surface.
The stability of the molecular rotations and tilting shows that the stronger binding of carboxylates compared to alkoxides minimizes the role of adsorbate-surface vdW interactions for isolated carboxylates. In contrast with alkoxides, the carboxyl group does not allow the carbon chain to freely tilt toward the surface; alkoxide leaning is responsible for stronger vdW interactions with the substrate and eventually a strong chain-length dependent surface stability.

Low Energy Electron Diffraction (LEED) results for adsorbate ordering
LEED performed on clean Au(110) confirms the (1×2) "missing row" reconstruction of the surface (Figs. S9C, S9E, S9G). 3 It has been previously shown using scanning tunneling microscopy (STM) and theoretical calculations that acetate induces a reconstruction of the Au-(1×2) to Au-(1×1) and forms a c(2×2) adsorbate ordering. 4 LEED experiments demonstrate that the characteristic LEED pattern observed for acetate (Fig. S9D), which has a c(2×2) adsorbate ordering based on STM experiments, occurs as well as for formate (Fig. S9F) and propanoate (Fig. S9H). The c(2×2) reconstruction is only visible for several seconds before beam damage affects the molecular ordering and leaves the Au-(1×1) visible. A theoretical p(1x2) and c(2x2) LEED pattern schematic is shown for reference ( Figure S9A-B). . Before annealing, bright protrusions 39 ± 4 pm taller than the background darker protrusions are observed (arrows on red line). After annealing, no bright protrusions are observed, despite a larger corrugation amplitude indicative of a better imaging resolution (black line). Images are processed with the same low-pass FFT filter to subtract high frequency noise. Profiles are averaged over 3 lines.

Interface Energy Calculation
The role of adsorbate-adsorbate interactions on the stability of acetate islands on Au(110) was previously determined using the same methodology detailed here. The change in interface energy from condensation is calculated by comparing the interface energy per carboxylate of 4 super cells containing 1 carboxylate on Au(1×2) as determined according to, (Eqn. S18) to the interface energy per carboxylate of 1 super cell with c(2×2) carboxylate layer on Au(1×1) and 3 supercells of the clean Au(1×2) as determined according to, (Eqn. S19) where E tot (clean /Au-1×2 ) is the calculated energy of the supercell of clean Au-1×2 (Fig. S11A), E tot (1/16ML /Au-1×2 ) is the calculated energy of the supercell of 1/16 ML of carboxylate on Au-1×2 (Fig. S11B), E tot (1/4ML /Au-1×1 ) is the calculated energy of the supercell of 1/4 ML of carboxylate on Au-1×1 (Fig. S11C), n Au-1×1 is the number of gold atoms in the 1×1 super cell, n Au-1×2 is the number of gold atoms in the 1×2 super cell, μ Au is the chemical potential of a gold atom and E free is the energy of a free (gas phase) carboxylate molecule. The change in interface energy per carboxylate (which quantifies the effect of condensation on carboxylate stability) can be added to the reaction energy per carboxylate (which quantifies the binding strength of isolated carboxylate) calculated previously, for comparing carboxylate stability at the condensed phase.