You are on page 1of 7

Journal of Catalysis 328 (2015) 36–42

Contents lists available at ScienceDirect

Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat

From the Sabatier principle to a predictive theory of transition-metal


heterogeneous catalysis q
Andrew J. Medford, Aleksandra Vojvodic ⇑, Jens S. Hummelshøj, Johannes Voss, Frank Abild-Pedersen,
Felix Studt, Thomas Bligaard, Anders Nilsson, Jens K. Nørskov ⇑
SUNCAT Center for Interface Science and Catalysis, Department of Chemical Engineering, Stanford University, Stanford, CA 94305, USA
SLAC National Accelerator Laboratory, 2575 Sand Hill Road, CA 94025, USA

a r t i c l e i n f o a b s t r a c t

Article history: We discuss three concepts that have made it possible to develop a quantitative understanding of trends
Received 19 November 2014 in transition-metal catalysis: scaling relations, activity maps, and the d-band model. Scaling relations are
Revised 22 December 2014 correlations between surface bond energies of different adsorbed species including transition states; they
Accepted 29 December 2014
open the possibility of mapping the many parameters determining the rate of a full catalytic reaction
onto a few descriptors. The resulting activity map can be viewed as a quantitative implementation of
the classical Sabatier principle, which states that there is an optimum ‘‘bond strength’’ defining the best
Keywords:
catalyst for a given reaction. In the modern version, the scaling relations determine the relevant ‘‘bond
Heterogeneous catalysis
Transition metals
strengths’’ and the fact that these descriptors can be measured or calculated makes it a quantitative the-
Theory ory of catalysis that can be tested experimentally by making specific predictions of new catalysts. The
Computational catalysis quantitative aspect of the model therefore provides new possibilities in catalyst design. Finally, the d-
DFT band model provides an understanding of the scaling relations and variations in catalytic activity in terms
Sabatier principle of the electronic structure of the transition-metal surface.
Scaling relation Ó 2015 Published by Elsevier Inc.
Descriptor

1. Introduction Since the early part of the 20th century, the Sabatier principle
has provided a conceptual framework for our thinking about the
Understanding how solid surfaces work as catalysts is attracting optimum catalyst [2,3]. The idea is that the best catalysts should
renewed interest. Not only is the majority of the chemical industry bind atoms and molecules with an intermediate strength: not too
relying on catalysis, but the search for renewable energy solutions weakly in order to be able to activate the reactants, and not too
and for sustainable production processes also requires completely strongly to be able to desorb the products. This leads to a vol-
new chemical processes and catalysts [1]. The hope is that through cano-type relationship between activity and bond strength as illus-
a fundamental understanding of basic concepts in heterogeneous trated in Fig. 1. This picture has been very successful in providing a
catalysis, we can develop new catalyst design strategies. The chal- general qualitative understanding of why there might be an opti-
lenge is to identify descriptors of catalytic activity and discover mum catalyst. As a theory of heterogeneous catalysis [3], it has
how they determine the activity and selectivity of a catalytic mate- the major shortcoming that it is not quantitative. This means a lack
rial. This will allow an understanding of trends from one catalyst to of predictive power, and hence, that it cannot be tested experimen-
the next, and it will open the possibility to design catalysts by tally. It is not clear which quantity should be used to characterize
searching for materials with descriptors close to the optimum the ‘‘bond strength’’ between the relevant intermediate and the
value that corresponds to the highest activity. catalyst, or what value of the ‘‘bond strength’’ corresponds to the
optimum catalyst. For these reasons, the Sabatier principle cannot
be used as a basis for catalyst design rules.
q
Paper submitted in recognition of the enormous influence Haldor Topsøe has The advent of computational electronic structure methods that
had on the advancement of the science of heterogeneous catalysis and his are efficient enough to treat reactions on metal surfaces and accu-
continued interest and support of the development of a predictive theory of the rate enough to have predictive power has significantly changed the
phenomenon. availability of systematic data for adsorption energies and activa-
⇑ Corresponding authors at: SLAC National Accelerator Laboratory, 2575 Sand Hill
Road, CA 94025, USA.
tion energies for surface chemical reactions. An enormous amount
E-mail addresses: alevoj@stanford.edu (A. Vojvodic), norskov@stanford.edu of effort has gone into establishing methods and data describing
(J.K. Nørskov). transition-metal surfaces [4–21]; this has enabled a priori

http://dx.doi.org/10.1016/j.jcat.2014.12.033
0021-9517/Ó 2015 Published by Elsevier Inc.
A.J. Medford et al. / Journal of Catalysis 328 (2015) 36–42 37

identification of good descriptors for a given catalytic reaction, and


it has made it possible to map out optimum regions of descriptor
space to help discover leads for new catalysts. In the present per-
spective, we briefly highlight some of these new developments.

2. Quantifying the bond strength

One of the first challenges toward a quantified heterogeneous


theory of catalysis is determining a suitable quantity, or ‘‘descrip-
tor’’, to represent the ‘‘bond strength’’ of interest. The early lit-
erature often used bulk properties such as the heat of formation
of the metal oxide [22] or suitable compound [23], or the number
of d-orbital electrons [24]. As noted already by Sabatier, the rele-
vant bond strengths should of course be related to bonds at the Fig. 1. Schematic representation of the qualitative Sabatier principle.
surface and not in the bulk or in molecules [3]; however, prior to
the recent advances in computational chemistry, bulk formation
energies were the simplest and most reliably determined quanti- Let us also assume that the first step, N2 dissociation, is rate lim-
ties to use. An example using the oxide heat of formation to iting. Both experiments and theory show that this is the case for
describe ammonia synthesis activity is shown in Fig. 2a. It can be the most active catalysts [26–28]. In this simple model, the rate
seen that while there is some correlation, the ‘‘volcano relation- is determined by two variables that are in principle independent:
ship’’ is far from convincing. the transition-state energy (activation barrier) for N2 dissociation
The volcano plot is much improved when the actual nitrogen (EN–N) (determining the rate of dissociation) and the adsorption
adsorption energies on the relevant metal surfaces are used as energy of nitrogen (EN) (determining the coverage on the surface).
the descriptor (see Fig. 2b). Here the same experimental rates are Fig. 3a shows the rate calculated with this mechanism and a mean-
plotted against the nitrogen adsorption energies as determined field kinetic model as a function of these two variables. Values of
by density functional theory (DFT) calculations [25]. (EN, EN–N) of stepped transition-metal surfaces are overlaid on
The improved relationship of the volcano plot in Fig. 2b indi- the plot. It can be seen that there is a clear linear scaling relation
cates that the adsorption energy of N⁄ at the surface site provides between the two variables. This means that a single variable is
a much better quantitative description of the ‘‘bond strength’’ con- enough to define the catalyst in terms of the ammonia synthesis
cept of the Sabatier principle than the oxide heat of formation. This rates, thus explaining why the experimental data in Fig. 2b and
is intuitive, since the N⁄ adsorption energy is a surface property, the model data in Fig. 3b can be ordered very well using the nitro-
and the reaction is mediated by the surface. Furthermore, the N⁄ gen adsorption energy as the descriptor.
adsorption energy probes the interactions between the transition The approach outlined above for a simple reaction can be gen-
metal and nitrogen (rather than oxygen), which is present in all eralized to much more complex reaction networks. Consider for
intermediates of the ammonia synthesis reaction. example the hydrogenation of CO to form methane, methanol, or
ethanol. Even a very simple model for this reaction network con-
3. Dimensional reduction: scaling relations tains 52 elementary reactions (a subset of which are shown in
Fig. 4, see Ref. 30 for more details) and therefore, in principle,
While nitrogen is undoubtedly a key intermediate for ammonia has approximately 104 independent energy variables (one reaction
synthesis, there are also several other intermediates and transition energy and one activation energy for each elementary reaction
states that play a role in the reaction. In order for the bond strength step) characterizing the reaction. Understanding trends in a space
to be a valid descriptor, it must describe the interaction between with more than 3 dimensions is a daunting task, leaving very little
the surface and all relevant intermediates and transition states. space for intuition or for developing catalyst design principles. A
To illustrate why a single descriptor is sufficient, consider the fol- very considerable reduction in complexity is needed.
lowing very simple mechanism for ammonia synthesis: The solution to this problem lies in the generality of the scaling
relations. Fig. 4 shows that the intermediates and transition states
N2 ðgÞ þ 2 ! N—N þ ! 2N of many different classes of elementary steps can be determined
using only a linear combination of the carbon and oxygen
N þ 3=2H2 ðgÞ ! NH3 ðgÞþ adsorption energies. In this case, there are two descriptors since

Fig. 2. (a) Volcano plot for the ammonia synthesis reaction over K-promoted transition-metal catalysts at 523 K and 0.8 bar using oxide heat of formation as a bond strength
descriptor (adapted from Ref. [22]). (b) Volcano plot of the same rates with the nitrogen adsorption energy at stepped metal surfaces as determined by DFT calculations using
the RPBE functional (from Ref. [25]) as a descriptor.
38 A.J. Medford et al. / Journal of Catalysis 328 (2015) 36–42

Fig. 3. (a) Potential energy diagram for simplified ammonia synthesis reaction, along with active site motif for the (2 1 1) surfaces with the step as the active site. (b) Ammonia
synthesis rate as a function of nitrogen adsorption energy and N2 dissociation barrier with energetics for FCC/HCP metal step sites and scaling line. Shaded area shows the
theoretical limit since the activation barrier (Ea) must always be positive (c) Rate shown as a function of the nitrogen adsorption energy, utilizing that EN–N is a linear function
of EN. The red point marked ‘‘CoMo’’ denotes the expected rate on a mixed site containing both Co and Mo based on interpolation between the two nitrogen adsorption
energies [29]. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 4. Scaling relations for 26 intermediates and 16 transition states in the conversion of CO and H2 to ethanol (left) as a function of carbon and oxygen adsorption energies
on transition-metal (2 1 1) surfaces. Tick marks are 1 eV apart; scaling parameters and more details can be found in Ref. [30]. Classes of elementary reactions included in the
analysis along with color code are shown (top right) along with the error between scaled and DFT energies for all scaling relations (bottom right). (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of this article.)

the bonding at the surface takes place through both C and O atoms, define a mapping of catalytic activity onto descriptor space. This
and the binding energy of these two atoms is relatively indepen- mapping, which is a generalization of the volcano relations of
dent. The bond energy of H to the surface scales well with that of Figs. 2 and 3b, we denote activity maps. For CO hydrogenation,
C and therefore does not need to be considered an independent the activity maps for different products as a function of the C
variable. By using the concept of scaling relations, the effective and O adsorption energies are shown in Fig. 5a–c. In such a map,
dimensionality of the problem has therefore been reduced to only an active site on a catalyst surface is defined by the values of the
2, leading to a much more manageable problem in terms of both two descriptors, and in the figure, the values for (2 1 1) step con-
optimization and understanding. figurations of a number of relevant transition metals are included.
The maps provide a means to rationalize the known trends about
4. Activity maps: understanding trends in catalytic activity and transition-metal catalysts for these reactions: Cu is the best ele-
selectivity mental metal catalyst for methanol synthesis [31], Ru, Co, Ni, and
Rh are the best for methanation [32], and no good elemental metal
The scaling relations define the descriptors for a given reaction, catalyst is known for ethanol synthesis (rhodium has been shown
and by combining the scaling relations with a kinetic model we can to catalyze the synthesis of ethanol with reasonable selectivity in
A.J. Medford et al. / Journal of Catalysis 328 (2015) 36–42 39

Fig. 5. Rates and selectivities for the formation of methane, methanol, and ethanol as a function of carbon and oxygen adsorption energies. The rates and selectivities are
calculated using a microkinetic model, which utilizes the scaling relations of Fig. 4 to reduce the complexity. The reaction conditions are as follows: T = 593 K, PCO = 30 bar,
PH2 = 60 bar. (a) The rate of methane formation, (b) rate of methanol formation, (c) rate of ethanol formation, and (d) selectivity map for the three different products along
with carbon and oxygen binding energies of A3B transition-metal alloys (indicated as white x’s). The inset of (d) shows several alloys that were identified as promising ethanol
catalysts. See Ref. [30] for more details.

the presence of specific promoters or supports, but is often selec- One important question is to understand the physical basis for
tive to methane, especially on inert supports [33]). these relationships and why they are so general. Electronic struc-
Knowing the rates of different parallel reaction pathways lead- ture calculations and spectroscopic measurements have been used
ing to different products also allows us to map out the selectivity of to provide detailed answers to this question. Fig. 6a shows the cal-
different catalysts as a function of descriptor values. This is illus- culated variations in the adsorption energy of nitrogen, EN, and in
trated in the selectivity map in Fig. 5d for the production of the energy of the transition-state energy, EN–N, as a function of a
methane, methanol, and ethanol over stepped metal surfaces. parameter describing the electronic structure of the metal surfaces.
Selectivity toward higher alcohols (modeled by ethanol) is only The parameter chosen is the upper band edge of the metal d band
found in a narrow strip of parameter values, far from elemental projected onto the surface atom(s) to which the adsorbates bond
metal values. [42,43]. In the d-band model, this parameter (or the d-band center,
in cases where the coupling matrix elements are roughly constant
[44]) correlates with the hybridization energy. The reason is that
5. The electronic structure factor in transition-metal catalysis: the interaction strength is related to the filling of the anti-bonding
the d-band model states formed between the metal d states and the adsorbate valence
states. The higher the d states are in energy relative to the highest
Scaling relations between activation energies and reaction ener- occupied states at the Fermi energy, the higher in energy the anti-
gies have been known in various subfields of chemistry for decades bonding states are, the more empty they are, and the stronger the
as Brønsted relations [34], Evans–Polanyi relations [35,36], or the interaction becomes. This picture has been confirmed by direct
Hammett equation [37]. The adsorption energy scaling relations spectroscopic measurements, as illustrated in Fig. 6.
represent a significant extension of this concept since they cover Since both EN–N and EN depend linearly on a parameter describ-
not only activation energies but also energies of intermediates, ing the surface electronic structure, they also depend linearly on
and since they relate these energies to any suitable adsorption each other. The explanation of the scaling relations is therefore
energy rather than just the reaction energy [38]. Establishing the that both transition-state and adsorption energies of different
scaling relations relies on access to adsorption energies and transi- intermediates depend on the surface electronic structure in similar
tion-state energies for many adsorbates on a statistically large ways. Different adsorbate states hybridize with the surface d elec-
enough number of surfaces. Systematic data did not become avail- trons to form bonding and anti-bonding states. The strength of the
able until theoretical methods for calculating such quantities interaction depends critically on the degree of filling of the anti-
became accurate and fast enough to treat the complex systems of bonding states, which in turn depends on the position of the d
interest in heterogeneous catalysis [38]. Benchmarking the compu- states relative to the Fermi level. This picture has been confirmed
tational methods against measurements is a very important part of in detailed X-ray spectroscopy experiments probing the filled and
establishing such a procedure, and it has been shown that modern empty states in adsorbate systems, see inset in Fig. 6 [43,45].
theoretical methods provide reasonably accurate adsorption ener-
gies [39,40]. Most recently, it was shown that the scaling relations
and the position of the maximum in activity maps are quite insen- 6. Catalyst design strategies
sitive to errors in underlying electronic structure treatment for
exchange and correlation effects, a result that helps understand Perhaps the most important consequence of a quantitative
the success of the approach outlined above [41]. model of catalytic activity and selectivity is the possibility of using
40 A.J. Medford et al. / Journal of Catalysis 328 (2015) 36–42

(a) 8 (b) 8
XES Ag
XAS

Nitrogenadsorptionenergy [eV]

N2 dissociationbarrier [eV]
6 N/Ni 6

Cu
4 4
Ag Pd
Pt
2 Cu 2 Ni Co
Pd Pt Fe
Ni Co Rh
0 0 Ru
XES Fe
Rh Ru Re
XAS
Re
2 N/Cu 2

4 4
2 1 0 1 2 3 4 5 2 1 0 1 2 3 4 5
εdT [eV] εdT [eV]

Fig. 6. Nitrogen adsorption energy (a) and N2 transition-state potential energy (b) as a function of the upper band edge of the d projected density of states of the metal surface
atoms. The insets in (a) show experimental X-ray absorption and emission spectra for N adsorbed on the (1 1 1) surface of Cu and Ni measuring the filled (indicated by shaded
areas) and empty N-induced states, respectively. The emptying of the anti-bonding N-2p/metal-3d states on going from Cu to Ni is clearly seen, explaining why Ni binds N
considerably stronger [43,45].

activity and selectivity maps to aid the search for new catalysts, The scaling relations, here presented for the stepped (2 1 1) met-
either for known reactions or for new reactions and process al surfaces, provide some very stringent restrictions on the possi-
conditions. bilities of finding good catalysts. Take the example of ammonia
The reduction in the number of parameters characterizing a cat- synthesis to illustrate this. It would be desirable to find an ammo-
alyst system for a given reaction offered by the scaling relations nia synthesis catalyst that has a lower barrier for N2 dissociation
allows for identification of a few descriptors and enables one to and does not bind N atoms stronger at the same time. Fig. 3a shows
identify the optimum values. This accrued knowledge opens the how the scaling relation effectively makes that impossible for the
possibility for a more rigorous route to catalyst design. An early elemental (2 1 1) stepped transition-metal surfaces. Given the scal-
example of such an approach was the suggestion based on a vol- ing relations, finding a good catalyst becomes a question of identi-
cano relation similar to Fig. 2b that a mixed Co–Mo catalyst should fying the best compromise between different competing factors
be close to optimum as a catalyst for ammonia synthesis [29]. The (e.g., N2 dissociation and NH3 desorption).
idea is quite intuitive from Fig. 2b or 3b: having N adsorption ener- An alternative strategy is to find ways in which to circumvent
gies somewhere between Co, which binds N too weakly, and Mo, the scaling relations, that is, find ways to stabilize the adsorbed
which binds N too strongly, should give a close to optimum cata- state of one of the intermediates or transition state without stabi-
lyst. The intermediate bond energy is borne out by detailed calcu- lizing others. Fig. 3a indicates that there is ample room for finding
lations of different model systems of the Co–Mo active site. These a better ammonia synthesis catalyst if we could find effective ways
Co–Mo mixed metal catalysts were shown experimentally to give of circumventing the scaling relations, or find active site motifs
activities comparable to Ru [29]. Such an intuition could not have that have a lower-lying scaling relation than the (2 1 1) stepped
been obtained from Fig. 2a. metal surfaces [42]. For example, if the close-packed surface is con-
More generally, calculation (or measurement) of descriptor val- sidered as the active site, the scaling relation will change, as shown
ues for a range of possible catalysts can be used to screen for new in Fig. 7a; however, this shift leads to a significant increase in the
catalysts. Fig. 5d includes a screening of clean (2 1 1) stepped metal transition-state energy, and thus a decrease of many orders of
sites for higher alcohol synthesis (the effect of additives is not magnitude in the rate as seen in Fig. 7b. Another possible strategy
included here). The calculations point to a series of potentially is to add alkali metals as promoters. As seen in Fig. 7, a potassium
interesting catalyst candidates. The prediction that unpromoted promoter stabilizes the transition state for N2 dissociation without
Pt–Co and, in particular, Cu–Co should be good candidates with a affecting the N adsorption energy significantly [48], leading to a
high selectivity has been verified experimentally [46,47]. different scaling relation and an order of magnitude increase in

Fig. 7. (a) Ammonia synthesis rate as a function of nitrogen adsorption energy and N2 dissociation barrier, similar to Fig. 3b. Scaling relation lines correspond to metal close-
packed (1 1 1) surfaces (red), (2 1 1) steps (black), and (2 1 1) K-promoted steps (blue). (b) Volcano plots corresponding to each scaling relation line. The kinetic model and
other conditions are the same as in Fig. 3. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
A.J. Medford et al. / Journal of Catalysis 328 (2015) 36–42 41

rate. Designing active site motifs with a three-dimensional charac- [8] F. Illas, N. López, J.M. Ricart, A. Clotet, J.C. Conesa, M. Fernández-García,
Interaction of CO and NO with PdCu(1 1 1) surfaces, J. Phys. Chem. B 102 (1998)
ter that can distinguish between various intermediates and transi-
8017–8023, http://dx.doi.org/10.1021/jp982118w.
tion-state structures is a very important path toward engineering [9] A. Eichler, J. Hafner, A. Groß, M. Scheffler, Trends in the chemical reactivity of
catalyst structures that can further escape the scaling relations; surfaces studied by ab initio quantum-dynamics calculations, Phys. Rev. B 59
biological catalysts exploit this strategy extensively. Two recent (1999) 13297–13300, http://dx.doi.org/10.1103/PhysRevB.59.13297.
[10] M.W. Wu, H. Metiu, The effect of strain on the adsorption of CO on Pd(1 0 0), J.
studies show possibilities of escaping the scaling relations by Chem. Phys. 113 (2000) 1177, http://dx.doi.org/10.1063/1.481896.
exploiting the strategy of metal doping [49] and the importance [11] A. Michaelides, P. Hu, A density functional theory study of CH2 and H
of 3D confinement in oxides [50]. adsorption on Ni(1 1 1), J. Chem. Phys. 112 (2000) 6006, http://dx.doi.org/
10.1063/1.481173.
The quantitative theory outlined above for thermal heteroge- [12] K. Reuter, C. Stampfl, M. Verónica Ganduglia-Pirovano, M. Scheffler, Atomistic
neous catalysis has been extended to a predictive theory of hetero- description of oxide formation on metal surfaces: the example of ruthenium,
geneous electrochemical systems by taking the electrochemical Chem. Phys. Lett. 352 (2002) 311–317, http://dx.doi.org/10.1016/S0009-
2614(01)01472-5.
potential and pH effects into account when modeling the thermo- [13] T. Shubina, M.T. Koper, Quantum-chemical calculations of CO and OH
dynamics of electrocatalytic reactions on surfaces [51,52]. This the- interacting with bimetallic surfaces, Electrochim. Acta 47 (2002) 3621–3628,
ory has formed the basis for computational searches for new http://dx.doi.org/10.1016/S0013-4686(02)00332-8.
[14] J.R. Kitchin, J.K. Nørskov, M.A. Barteau, J.G. Chen, Modification of the surface
electrode materials [16,53,54]. electronic and chemical properties of Pt(1 1 1) by subsurface 3d transition
We have briefly outlined elements of a consistent quantitative metals, J. Chem. Phys. 120 (2004) 10240–10246, http://dx.doi.org/10.1063/
theory of transition-metal surface catalysis. There are many chal- 1.1737365.
[15] S. Kandoi, A.A. Gokhale, L.C. Grabow, J.A. Dumesic, M. Mavrikakis, Why Au and
lenges ahead before we have a complete theory of heterogeneous
Cu are more selective than Pt for preferential oxidation of CO at low
catalysis. Outstanding issues include an understanding of finite temperature, Catal. Lett. 93 (2004) 93–100, http://dx.doi.org/10.1023/
size effects, the role of supports, and a proper description of the B:CATL.0000016955.66476.44.
surface chemistry of transition-metal compounds. All of this [16] X. Lin, W.F. Schneider, B.L. Trout, Chemistry of sulfur oxides on transition
metals. III. Oxidation of SO2 and self-diffusion of O, SO2, and SO3 on Pt(1 1 1), J.
requires new theoretical methods, increased accuracy, better Phys. Chem. B 108 (2004) 13329–13340, http://dx.doi.org/10.1021/jp048507.
kinetic models including the ability to treat complex systems and [17] J. Greeley, T.F. Jaramillo, J. Bonde, I.B. Chorkendorff, J.K. Nørskov,
reaction networks, and better methods for storage, retrieval, shar- Computational high-throughput screening of electrocatalytic materials for
hydrogen evolution, Nat. Mater. 5 (2006) 909–913, http://dx.doi.org/10.1038/
ing, and mining of data and information [55,56]. Given the devel- nmat1752.
opments in the last decade this is not outside our reach. [18] A.U. Nilekar, Y. Xu, J. Zhang, M.B. Vukmirovic, K. Sasaki, R.R. Adzic, et al.,
Bimetallic and ternary alloys for improved oxygen reduction catalysis, Top.
Catal. 46 (2007) 276–284, http://dx.doi.org/10.1007/s11244-007-9001-z.
Author contributions [19] H. Grönbeck, A. Hellman, A. Gavrin, Structural, energetic, and vibrational
properties of NO(x) adsorption on Ag(n), n = 1–8, J. Phys. Chem. A 111 (2007)
6062–6067, http://dx.doi.org/10.1021/jp071117d.
The manuscript was written through contributions of all [20] R.A. van Santen, M. Neurock, S.G. Shetty, Reactivity theory of transition-metal
authors. All authors have given approval to the final version of surfaces: a Brønsted–Evans–Polanyi linear activation energy-free-energy
the manuscript. analysis, Chem. Rev. 110 (2010) 2005–2048, http://dx.doi.org/10.1021/
cr9001808.
[21] H.A. Hansen, J.B. Varley, A.A. Peterson, J.K. Nørskov, Understanding trends in
Funding sources the electrocatalytic activity of metals and enzymes for CO2 reduction to CO, J.
Phys. Chem. Lett. 4 (2013) 388–392, http://dx.doi.org/10.1021/jz3021155.
[22] K. Aika, A. Ozaki, Catalysts for synthesis of ammonia, in: M. Boudart, J.R.
U.S. Department of Energy Office of Basic Energy Science. Anderson (Eds.), Catal. Sci. Technol., vol. 1, Springer, Berlin, 1981.
U.S. Department of Defense National Defense Science and [23] W.J.M. Rootsaert, W.M.H. Sachtler, Interaction of formic acid vapour with
tungsten, Z. Phys. Chem. 26 (1960) 16–26, http://dx.doi.org/10.1524/
Engineering Graduate Fellowship (A.J.M). zpch.1960.26.1_2.016.
[24] G.C. Bond, Principles of Catalysis, second Revised ed., The Chemical Society,
1972.
Acknowledgments
[25] A. Vojvodic, A.J. Medford, F. Studt, F. Abild-Pedersen, T.S. Khan, T. Bligaard,
et al., Exploring the limits: a low-pressure, low-temperature Haber-Bosch
Support from the U.S. Department of Energy Office of Basic process, Chem. Phys. Lett. 598 (2014) 108–112, http://dx.doi.org/10.1016/
Energy Science to the SUNCAT Center for Interface Science and Cat- j.cplett.2014.03.003.
[26] R. Schlögl, Catalytic synthesis of ammonia-a ‘‘never-ending story’’?, Angew
alysis is gratefully acknowledged. A.J.M. is grateful for support by Chem. Int. Ed. Engl. 42 (2003) 2004–2008, http://dx.doi.org/10.1002/
the U.S. Department of Defense through the National Defense anie.200301553.
Science and Engineering Graduate Fellowship Program. [27] G. Ertl, Surface science and catalysis—studies on the mechanism of ammonia
synthesis: the P. H. Emmett Award address, Catal. Rev. 21 (1980) 201–223,
http://dx.doi.org/10.1080/03602458008067533.
References [28] A. Logadottir, T.H. Rod, J.K. Norskov, B. Hammer, S. Dahl, C.J.H. Jacobsen, The
Brønsted–Evans–Polanyi relation and the volcano plot for ammonia synthesis
over transition metal catalysts, J. Catal. 197 (2001) 229–231, http://dx.doi.org/
[1] N.R.C. Committee on the Review of the Basic Energy Sciences Catalysis Science
10.1006/jcat.2000.3087.
Program, Catalysis for Energy: Fundamental Science and Long-Term Impacts of
[29] C.J.H. Jacobsen, S. Dahl, B.S. Clausen, S. Bahn, A. Logadottir, J.K. Nørskov,
the U.S. Department of Energy Basic Energy Science Catalysis Science Program,
Catalyst design by interpolation in the periodic table: bimetallic ammonia
The National Academies Press, 2009.
synthesis catalysts, J. Am. Chem. Soc. 123 (2001) 8404–8405, http://dx.doi.org/
[2] P. Sabatier, La catalyse en chimie organique, Librairie Polytechnique, Paris et
10.1021/ja010963d.
Liège, 1920.
[30] A.J. Medford, A.C. Lausche, F. Abild-Pedersen, B. Temel, N.C. Schjødt, J.K.
[3] M. Che, Nobel Prize in chemistry 1912 to Sabatier: organic chemistry or
Nørskov, et al., Activity and selectivity trends in synthesis gas conversion to
catalysis?, Catal Today 218–219 (2013) (1912) 162–171.
higher alcohols, Top. Catal. 57 (2014) 135–142, http://dx.doi.org/10.1007/
[4] B. Hammer, K. Jacobsen, J. Nørskov, Dissociation path for H2 on Al(1 1 0), Phys.
s11244-013-0169-0.
Rev. Lett. 69 (1992) 1971–1974, http://dx.doi.org/10.1103/PhysRevLett.
[31] F. Studt, F. Abild-Pedersen, Q. Wu, A.D. Jensen, B. Temel, J.-D. Grunwaldt, et al.,
69.1971.
CO hydrogenation to methanol on Cu–Ni catalysts: theory and experiment, J.
[5] G. Wiesenekker, G.J. Kroes, E.J. Baerends, R.C. Mowrey, Dissociation of H2 on
Catal. 293 (2012) 51–60, http://dx.doi.org/10.1016/j.jcat.2012.06.004.
Cu(1 0 0): dynamics on a new two-dimensional potential energy surface, J.
[32] M.P. Andersson, T. Bligaard, A. Kustov, K.E. Larsen, J. Greeley, T. Johannessen,
Chem. Phys. 102 (1995) 3873, http://dx.doi.org/10.1063/1.468547.
et al., Toward computational screening in heterogeneous catalysis: pareto-
[6] J.-F. Paul, P. Sautet, Density-functional periodic study of the adsorption of
optimal methanation catalysts, J. Catal. 239 (2006) 501–506, http://dx.doi.org/
hydrogen on a palladium (1 1 1) surface, Phys. Rev. B 53 (1996) 8015–8027,
10.1016/j.jcat.2006.02.016.
http://dx.doi.org/10.1103/PhysRevB.53.8015.
[33] M. Haider, M. Gogate, R. Davis, Fe-promotion of supported Rh catalysts for
[7] A. Eichler, J. Hafner, G. Kresse, Reaction path for the dissociative adsorption of
direct conversion of syngas to ethanol, J. Catal. 261 (2009) 9–16, http://
hydrogen on the (1 0 0) surfaces of face-centered-cubic transition metals, Surf.
dx.doi.org/10.1016/j.jcat.2008.10.013.
Rev. Lett. 04 (1997) 1347–1351, http://dx.doi.org/10.1142/
S0218625X97001796.
42 A.J. Medford et al. / Journal of Catalysis 328 (2015) 36–42

[34] J.N. Brønsted, K.J. Pedersen, The catalytic disintegration of nitramide and its [46] J.M. Christensen, A.J. Medford, F. Studt, A.D. Jensen, High pressure CO
physical-chemical relevance, Z. Phys. Chem., Stöchiomet. Verwandtschaftsleh. hydrogenation over bimetallic Pt–Co catalysts, Catal. Lett. 144 (2014) 777–
108 (1924) 185–235. 782, http://dx.doi.org/10.1007/s10562-014-1220-x.
[35] R.P. Bell, The theory of reactions involving proton transfers, Proc. R. Soc. A: [47] G. Prieto, S. Beijer, M.L. Smith, M. He, Y. Au, Z. Wang, et al., Design and
Math. Phys. Eng. Sci. 154 (1936) 414–429, http://dx.doi.org/10.1098/ synthesis of copper-cobalt catalysts for the selective conversion of synthesis
rspa.1936.0060. gas to ethanol and higher alcohols, Angew. Chem. Int. Ed. Engl. 53 (2014)
[36] M.G. Evans, M. Polanyi, Further considerations on the thermodynamics of 6397–6401, http://dx.doi.org/10.1002/anie.201402680.
chemical equilibria and reaction rates, Trans. Faraday Soc. 32 (1936) 1333, [48] S. Dahl, A. Logadottir, C.J.H. Jacobsen, J.K. Nørskov, Electronic factors in
http://dx.doi.org/10.1039/tf9363201333. catalysis: the volcano curve and the effect of promotion in catalytic ammonia
[37] L.P. Hammett, M.A. Paul, The relation between the rates of some acid catalyzed synthesis, Appl. Catal. A: Gen. 222 (2001) 19–29, http://dx.doi.org/10.1016/
reactions and the acidity function, H 0, J. Am. Chem. Soc. 56 (1934) 830–832, S0926-860X(01)00826-2.
http://dx.doi.org/10.1021/ja01319a018. [49] N.B. Bendtsen, V. Petrykin, P. Krtil, J. Rossmeisl, Beyond the volcano limitations
[38] F. Abild-Pedersen, J. Greeley, F. Studt, J. Rossmeisl, T. Munter, P. Moses, et al., in electrocatalysis – oxygen evolution reaction, Phys. Chem. Chem. Phys. 16
Scaling properties of adsorption energies for hydrogen-containing molecules (2014) 13682–13688, http://dx.doi.org/10.1039/C4CP00571F.
on transition-metal surfaces, Phys. Rev. Lett. 99 (2007) 016105, http:// [50] A.D. Doyle, J.H. Montoya, A. Vojvodic, Improving oxygen electrochemistry
dx.doi.org/10.1103/PhysRevLett. 99.016105. through nanoscopic confinement, Chem. Catal. Chem. (2014), http://dxdoi.org/
[39] B. Hammer, L. Hansen, J. Nørskov, Improved adsorption energetics within 10.1002/cctc.201402864.
density-functional theory using revised Perdew-Burke-Ernzerhof functionals, [51] J.K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J.R. Kitchin, T. Bligaard,
Phys. Rev. B 59 (1999) 7413–7421, http://dx.doi.org/10.1103/ et al., Origin of the overpotential for oxygen reduction at a fuel-cell cathode, J.
PhysRevB.59.7413. Phys. Chem. B 108 (2004) 17886–17892, http://dx.doi.org/10.1021/jp047349j.
[40] J. Wellendorff, K.T. Lundgaard, A. Møgelhøj, V. Petzold, D.D. Landis, J.K. [52] C. Taylor, S. Wasileski, J.-S. Filhol, M. Neurock, First principles reaction
Nørskov, et al., Density functionals for surface science. Exchange-correlation modeling of the electrochemical interface: Consideration and calculation of a
model development with Bayesian error estimation, Phys. Rev. B 85 (2012) tunable surface potential from atomic and electronic structure, Phys. Rev. B 73
235149, http://dx.doi.org/10.1103/PhysRevB.85.235149. (2006) 165402, http://dx.doi.org/10.1103/PhysRevB.73.165402.
[41] A.J. Medford, J. Wellendorff, A. Vojvodic, F. Studt, F. Abild-Pedersen, K.W. [53] B. Hinnemann, P.G. Moses, J. Bonde, K.P. Jørgensen, J.H. Nielsen, S. Horch, et al.,
Jacobsen, et al., Assessing the reliability of calculated catalytic ammonia Biomimetic hydrogen evolution: MoS2 nanoparticles as catalyst for hydrogen
synthesis rates, Science 345 (2014) 197–200, http://dx.doi.org/ evolution, J. Am. Chem. Soc. 127 (2005) 5308–5309, http://dx.doi.org/
10.1126/science.1253486. 10.1021/ja0504690.
[42] A. Vojvodic, J.K. Nørskov, F. Abild-Pedersen, Electronic structure effects in [54] J. Greeley, I.E.L. Stephens, A.S. Bondarenko, T.P. Johansson, H.A. Hansen, T.F.
transition metal surface chemistry, Top. Catal. 57 (2013) 25–32, http:// Jaramillo, et al., Alloys of platinum and early transition metals as oxygen
dx.doi.org/10.1007/s11244-013-0159-2. reduction electrocatalysts, Nat. Chem. 1 (2009) 552–556. <http://www.ncbi.
[43] L.G.M. Pettersson, A. Nilsson, A molecular perspective on the d-band model: nlm.nih.gov/pubmed/21378936>.
synergy between experiment and theory, Top. Catal. 57 (2013) 2–13, http:// [55] J.S. Hummelshøj, F. Abild-Pedersen, F. Studt, T. Bligaard, J.K. Nørskov, CatApp: a
dx.doi.org/10.1007/s11244-013-0157-4. web application for surface chemistry and heterogeneous catalysis, Angew.
[44] B. Hammer, J.K. Norskov, Why gold is the noblest of all the metals, Nature 376 Chem. Int. Ed. Engl. 51 (2012) 272–274, http://dx.doi.org/10.1002/
(1995) 238–240, http://dx.doi.org/10.1038/376238a0. anie.201107947.
[45] A. Nilsson, L.G.M. Pettersson, B. Hammer, T. Bligaard, C.H. Christensen, J.K. [56] A. Jain, S.P. Ong, G. Hautier, W. Chen, W.D. Richards, S. Dacek, et al., The
Nørskov, The electronic structure effect in heterogeneous catalysis, Catal. Lett. Materials Project: a materials genome approach to accelerating materials
100 (2005) 111–114, http://dx.doi.org/10.1007/s10562-004-3434-9. innovation, APL Mater. 1 (2013) 011002, http://dx.doi.org/10.1063/1.4812323.

You might also like