Of model-based pragmatism

The mechanism of methanol coupling to methyl formate over single-crystal gold catalysts has been firmly established but barely reconciled with experiments performed under practical conditions. Now, a method to close this gap has been reported, which enables the prediction of the reaction´s selectivity for a broad range of experimental conditions.

Catalytic processes contribute approximately twenty per cent to the US gross domestic product, help to create essentials such as food and clean water, remove pollution and reduce greenhouse gas emissions, and play significant roles in supplying energy; however, their design is challenging, requiring an understanding of complicated networks of reaction steps that depend on reaction conditions and potentially other catalytic variables. While many great contributions to the understanding of catalytic functions exist in the literature — very often based on single-crystal models — only a few provide thorough enough information to predict and design selective processes of practical relevance.

Now, writing in Nature Catalysis, Madix and co-workers1 provide such information. Specifically, building on over a decade’s worth of research2,3,4,5,6, they confirm the mechanism for the selective oxidation of methanol to methyl formate over nanoporous Ag0.03Au0.97 catalysts. Since garnering understanding about catalytic mechanisms involves investigating both molecular-level and bulk observable phenomena, a variety of methods of analysis is necessary. The team uses a combination of temperature programmed reaction spectroscopy (TPRS), temporal analysis of products (TAP), and catalytic reactor experiments, along with microkinetic modelling.

The TPRS experiments provided insights on the importance of a catalytic intermediate, adsorbed atomic oxygen (O*), which activates the O–H bond in methanol, forming adsorbed methoxy (CH3O*). Combining this insight with prior surface science studies about the decompositions of formate, formic acid and formaldehyde on Au surfaces7,8, the team were able to deduce the full reaction mechanism for methyl formate (HCOOCH3) formation. In the reaction mechanism, the rate-determining step is the breaking of a C–H bond in CH3O*, forming adsorbed formaldehyde, CH2O* (Fig. 1). Once formed, this species can either desorb, producing formaldehyde; couple with CH3O* to form methyl formate; or react with O* leading to CO2. In their TPRS experiments, which were performed under ultrahigh vacuum at temperatures of 120 K, selectivity toward methyl formate was controlled by the surface coverage of O*. Surface coverage is the ratio of the number of surface adsorbates to the number of surface metal atoms, with lower O* coverages — 0.05 monolayer (ML), where 1 ML equals 1 O* per surface metal atom — being highly selective for methyl formate and higher O* coverages (0.20 ML) instead favouring CO2. Catalytic reactor experiments, performed at 423 K and pressures of around 1 atm, showed high selectivity for methyl formate at O* coverages (0.10 ML) intermediate to those investigated in TPRS. While these results are seemingly in agreement, the observations from TAP experiments were not. The TAP experiments, which were performed at temperatures between 363 K and 423 K and pressures of approximately 10–5 atm CH3OH, produced primarily formaldehyde and secondarily CO2 at 423 K. Formation of methyl formate did not become significant until the temperature was decreased to 383 K, and only became dominant when the temperature was decreased further to 363 K.

Fig. 1: Schematic of the oxidative coupling of methane on Ag0.03Au0.97.

The fate of CH2O depends on the temperature and the relative surface coverages of oxygen and methoxy species. The first step in the catalytic coupling of CH3OH to HCOOCH3 is the oxygen-assisted activation of CH3OH to CH3O. Following this step is the rate-determining step, C–H bond cleavage in adsorbed CH3O, forming adsorbed CH2O. Adsorbed CH2O can either desorb or react with adsorbed O to form CO2, or couple with adsorbed CH3O to form HCOOCH3. The rates of these different processes depend on the reaction conditions. At high temperatures, desorption is favoured, and CH2O gas is produced. At lower temperatures, desorption is not favoured, leaving CH2O on the surface to react. Whether it is oxidized to CO2 or coupled with adsorbed CH3O to form C2H4O2 depends on the relative surface coverages of O and CH3O. These coverages are determined by the CH3OH pressure: high CH3OH pressures lead to higher coverages of adsorbed CH3O and thus a higher coupling rate, while lower CH3OH pressures lead to higher coverages of adsorbed O and thus higher oxidation rates.

To reconcile these results, the team employed kinetic modelling. Kinetic models use steps in the reaction mechanism along with their kinetic parameters (pre-exponential factors and activation energies) and reaction conditions to produce information such as the rate, surface coverages, and product distributions. Here, the postulated mechanism along with pre-exponential factors and activation energies derived from TPRS results allowed for the simulation of the catalytic reaction over a range of temperatures and pressures. Such simulations verified all of the experimental observations — from TPRS, TAP, and the catalytic reactor — and provided a unifying picture of the catalysis of methanol partial oxidation.

The selectivity is determined by what happens to CH2O*. One possibility is that it desorbs, producing formaldehyde. Producing methyl formate or CO2 requires the rate for the coupling with CH3O* or reacting with O* to be considerably faster than the rate of desorption. In the catalytic reactor experiments, the pressure of CH3OH is high (~1 atm), which leads to a high surface coverage of CH3O* and thus a high rate. This rate is higher than the rate of desorption and so leads to the formation of methyl formate. In the TAP experiments, the pressure of CH3OH is much lower (~10–5 bar) and thus the coverage of CH3O* is also lower. Therefore, when the TAP experiments are performed at the same temperature as the catalytic reactor experiments (423 K), CH2O* desorption is kinetically preferred over coupling. As the temperature in the TAP experiments is decreased, desorption becomes less favourable, allowing the rate of coupling to overcome the rate of desorption. In the TPRS experiments, the temperature is so low (120 K) that desorption is suppressed: the fate of CH2O* therefore depends on the surface coverage. At higher coverages of CH3O* or lower coverages of O*, coupling dominates, while at lower coverages of CH3O* or higher coverages of O*, oxidation dominates. In the pathway to methyl formate, there are therefore two selectivity-controlling species: CH3O* and O*.

Such knowledge, obtained in many cases through the analysis of model catalysts, results in the remarkable ability to optimize the selectivity of a catalytic process under realistic conditions. In this example, the catalyst must moderate the coverage of O*, enough to promote CH3OH activation but not so much as to favour CH2O* oxidation. Catalysts comprising noble metals are excellent for this purpose, as they have moderate affinities for oxygen. Perhaps a less expensive alternative could employ an oxygen buffering component such as ceria or zirconia. For the reaction conditions, high pressures of CH3OH are needed in order to generate the high coverages of CH3O* that promote coupling.

Ideally, such insights would be available for all important reactions in which catalysts play a role. Garnering such insights is certainly possible; however, it requires a variety of experimental and computational methods and the insight to deduce a complicated mechanism from seemingly disparate results.


  1. 1.

    Reece, C., Redekop, E., Karakalos, S., Friend, C. M. & Madix, R. J. Nat. Catal. (2018).

    Article  Google Scholar 

  2. 2.

    Stowers, K. J., Madix, R. J. & Friend, C. M. J. Catal. 308, 131–141 (2013).

    CAS  Article  Google Scholar 

  3. 3.

    Xu, B., Haubrich, J., Baker, T. A., Kaxiras, E. & Friend, C. M. J. Phys. Chem. C 115, 3703–3708 (2011).

    CAS  Article  Google Scholar 

  4. 4.

    Xu, B., Liu, X., Haubrich, J., Madix, R. J. & Friend, C. M. Angew. Chem. Int. Ed. 48, 4206–4209 (2009).

    CAS  Article  Google Scholar 

  5. 5.

    Xu, B., Madix, R. J. & Friend, C. M. Acc. Chem. Res. 47, 761–772 (2014).

    CAS  Article  Google Scholar 

  6. 6.

    Zugic, B. et al. Nat. Mater. 16, 558–564 (2017).

    CAS  Article  Google Scholar 

  7. 7.

    Wachs, I. E. & Madix, R. J. Appl. Surf. Sci. 5, 426–428 (1980).

    CAS  Article  Google Scholar 

  8. 8.

    Outka, D. A. & Madix, R. J. Surf. Sci. 179, 361–376 (1987).

    CAS  Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to Rachel B. Getman.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Getman, R.B. Of model-based pragmatism. Nat Catal 1, 818–819 (2018).

Download citation


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing