Embed Size (px)
Citation preview
In the format provided by the authors and unedited.
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
1
Supplementary Information
PtCu Single-Atom Alloys as coke-resistant catalysts for efficient C-H activation
Matthew D. Marcinkowski1ǁ, Matthew T. Darby2ǁ, Jilei Liu3ǁ, Joshua M. Wimble3, Felicia R. Lucci1,
Sungsik Lee4, Angelos Michaelides5, Maria Flytzani-Stephanopoulos3*,
Michail Stamatakis2* and E. Charles H. Sykes1*
1. Department of Chemistry, Tufts University, 62 Talbot Avenue, Medford, Massachusetts 02155, United States.
2. Thomas Young Centre and Department of Chemical Engineering, University College London, Roberts Building, Torrington Place, London WC1E 7JE, United Kingdom
3. Department of Chemical and Biological Engineering, Tufts University, 4 Colby Street, Medford, Massachusetts 02155, United States
4. X-ray Science Division, Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, Illinois 60439, United States.
5. Thomas Young Centre, London Centre for Nanotechnology and Department of Physics and Astronomy, University College London, Gower Street, London WC1E6BT, United
Kingdom ǁ denotes authors contributed equally to the work
*Corresponding author
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2915
NATURE CHEMISTRY | www.nature.com/naturechemistry 1
2
Contents
A. Model Catalyst Studies ......................................................................................................................... 3
A.1 Effect of Co-Adsorbed Iodine .............................................................................................................. 3
A.2 Methyl Iodide Decomposition Products ............................................................................................. 5
A.3 TPR of Deuterium Pre-covered 0.01 ML Pt/Cu(111) SAA ................................................................... 6
A.4 1.0 ML Pt/Cu(111) Surface .................................................................................................................. 8
A.5 Hydrogen Desorption after CH3I exposure ........................................................................................ 9
A.6 TPR Results from Pt/Cu Alloys ............................................................................................................ 9
A.7 KMC Reaction Network and Lattice Model ...................................................................................... 11
A.8 Rate Constants from DFT .................................................................................................................. 13
A.9 CxHy Adsorption Geometries ............................................................................................................. 14
A.10 Energy Landscape for the Full Dehydrogenation of Methane........................................................ 20
A.11 Calculations for Subsurface Pt Atoms ............................................................................................. 21
A.12 Areas of and Segregation Energies on (110) and (100) Surfaces .................................................... 22
A.13 STM of Surfaces Annealed to 1000 K .............................................................................................. 24
B. Ambient Pressure C-H Activation Studies ............................................................................................... 25
B.1 Supplementary methods for Nanoparticle Studies .......................................................................... 25
B.2 Nanoparticle Synthesis ..................................................................................................................... 25
B.3 CO-FTIR for fresh and spent catalysts ............................................................................................... 26
B.4 Extended X-ray absorption fine structure (EXAFS) characterizations .............................................. 31
B.5 B-D scrambling .................................................................................................................................. 33
B.6 Nonoxidative dehydrogenation of butane and propane .................................................................. 35
C. References .............................................................................................................................................. 38
3
A. Model Catalyst Studies
A.1 Effect of Co-Adsorbed Iodine
Supplementary Figure 1. Methane desorption from various Langmuir (1 L = 1X10-6 torr.s) exposures of methyl iodide on Cu(111) and 0.01 ML Pt/Cu(111). Each experiment was performed on a clean surface.
Iodine is known to strongly adsorb to metal surfaces and act as a site blocker.1,2 As shown in
Supplementary Figure 1, on Cu(111) as additional methyl iodide is deposited on the surface, the
temperature at which methane desorbs does not change, but the area of the methane peak begins to
drop after a 4.5 L dose. This is due to iodine blocking Cu sites from accommodating methyl groups,
causing methyl iodide to desorb intact (or causing methyl rejection as will be shown in Supplementary
Figure 3).1,2 On the 0.01 ML Pt/Cu(111) surface, there is also an eventual decrease in the amount of
desorbing methane, and the temperature shifts from 350 K to around 400 K after the exposure of
methyl iodide increases past 6 L. This is because iodine atoms block the single Pt sites and prevent low
temperature activation of C-H bonds.
To further demonstrate the site blocking properties of iodine, we performed experiments with
CO on pure Cu(111) (Supplementary Figure 2). The behavior of CO on Cu(111) is well documented. CO
4
forms three distinct packing phases on Cu(111), a densely packed (7x7) unit cell, an intermediate (4x4)
unit cell, and a least densely packed (√3 × √3) unit cell.3–5 These three phases desorb at 110 K, ~130 K,
and 170 K respectively. In addition, CO bound to Cu(111) steps desorbs at 220 K.6 In these experiments,
various amounts of methyl iodide were exposed to the surface followed by a heating ramp to remove all
species with the exception of iodine. After each methyl iodide exposure, the crystal was exposed to a
saturation dose of CO. When no methyl iodide is exposed to the surface, the familiar desorption trace of
CO from Cu(111) is seen. However, as more iodine is added to the surface, the signal from each CO
adsorption structure decreases; ultimately, after a 10.5 L exposure of methyl iodide, very little CO is able
to adsorb to the surface. The temperature CO desorbs from the surface is unaffected by the iodine,
indicating that it acts as a site blocker rather than surface modifier.
Supplementary Figure 2. Desorption of CO from surfaces previously exposed to various amounts of methyl iodide. The surface was thermally annealed to remove everything besides iodine prior to exposure to CO in each case.
5
A.2 Methyl Iodide Decomposition Products
Supplementary Figure 3. TPR traces for all desorption products observed following 4.5 L and 15 L exposures of methyl iodide on both Cu(111) and 0.01 ML Pt./Cu(111).
As mentioned in the main text, at relatively low coverage (4.5 L exposure) the main desorption
product detected after exposing the surface to methyl iodide is methane, which exhibits large signals for
both m/z 15 and 16. The second most abundant desorption product at these exposures is ethene which
is tracked via m/z 27. In addition to ethene, trace amounts of ethane and propene desorb from the
surface which are tracked via m/z 30 and m/z 41, respectively. The ratio of methane to carbon coupling
products is larger on the 0.01 ML Pt/Cu(111) surface, although the overall carbon count desorbing from
each surface is the same. The increase in methane production on the alloy surface is most likely due to
6
low temperature hydrogenation of methyl groups from small amounts of hydrogen adsorbed from the
chamber background. This can be seen by the low temperature shoulder apparent for the two methane
masses (15 and 16) where no coupling products are observed. At larger exposures of methyl iodide,
intact methyl iodide desorbs at low temperature on both surfaces. In addition, as previously reported by
Bent and co-workers, methyl rejection is also observed as there is a large desorption feature from m/z
15 at low temperature.7 Bent also reports a high temperature methyl rejection pathway occurring at 450
K at high coverage, but we do not see evidence of this in our work.7 In spectra presented in the main
text, all exposures were low enough to avoid desorption of intact methyl iodide and methyl rejection.
A.3 TPR of Deuterium Pre-covered 0.01 ML Pt/Cu(111) SAA
Supplementary Figure 4. (a) TPR spectra of possible methane products resulting from the reaction of methyl iodide on a D-pre-covered 0.01 ML Pt/Cu(111) surface. (b) Simulated TPR of H-pre-covered Pt/Cu(111) SAA showing methane desorption. (c) Schematic showing the interaction of D atoms with methyl groups on Pt/Cu.
7
We have previously demonstrated that Pt/Cu(111) SAAs are capable of dissociating H2.8,9 To
examine the direct hydrogenation of methyl groups to methane, we performed experiments in which
methyl iodide was introduced to a 0.01 ML Pt/Cu SAA that had been pre-covered with hydrogen at 80 K.
On this hydrogen pre-covered surface, the product methane desorbs at 200 K, well below the
temperatures methane desorbs from either clean surface. In addition, at 200 K no carbon coupling
products are observed. On this surface, C-H activation is not the rate limiting step. Rather, it is the
hydrogenation of methyl that is rate limiting in the production of methane. It is known that pre-covering
Pt(111) with hydrogen allows for hydrogenation of methyls.10 To confirm this on our SAA surface, we
performed a similar experiment using a deuterium pre-covered surface (Supplementary Figure 4a). M/z
15 through 20 were tracked to detect CH4, CH3D, CH2D2, CHD3, and CD4. We observe predominately m/z
17 with a large contribution from m/z 16 and a small amount of m/z 15. In other words, if we assign the
parent ion as M, we observe M, M-1, and M-2. M/z 16 (M-1) is ~70 % of the signal from m/z 17 while
m/z 15 (M-2) is ~20 % of the signal from m/z 17. Normally the primary contributor to CH4’s cracking
pattern is the parent ion m/z 16. According to NIST’s cracking pattern for methane, m/z 15 (M-1) and
m/z 14 (M-2) both contribute with relative intensities 89 % and 20 % of m/z 16 respectively.11 Therefore,
our observed cracking pattern is in good agreement with the sole desorption product at 200 K being
CH3D. This indicates that the methane results from hydrogenation of the methyl groups, and no C-H
activation occurs at this temperature. The production of methane at low temperature is limited by the
saturation of hydrogen on the 0.01 ML Pt/Cu(111) surface, which is 0.1 ML.8,9 As our coverage of methyl
groups present on the surface is greater than this, we still observe some CH4 desorbing at 350 K
resulting from C-H activation in methyl groups. In addition, we performed KMC simulations where we
populated the surface with CH3 and H atoms (Supplementary Figure 4b). Our simulation agrees well with
our experimental results and methane desorbs at 216 K in the simulation due to direct hydrogenation of
8
methyl groups. The conditions for the simulation are discussed in detail below and in the Methods
section of the main paper.
A.4 1.0 ML Pt/Cu(111) Surface
Supplementary Figure 5. STM image of 1 ML Pt/Cu(111) obtained at 30 K. Scale bar = 25 nm.
Depositing 1 ML of Pt onto a Cu(111) surface does not result in a monolayer Pt(111)-like film, rather the
surface is composed of large islands of intermixed Pt and Cu, between 1 and 3 layers high. These islands
do contain regions that are Pt(111)-like; however, some areas of the surface are not covered by islands,
but Pt still exists as single isolated species substituted into the exposed Cu(111) lattice.12
9
A.5 Hydrogen Desorption after CH3I exposure
Supplementary Figure 6. Hydrogen desorption after adsorption of 4.5 L CH3I on Cu(111), 0.01 ML Pt/Cu(111) and 1.0 ML Pt/Cu(111).
High temperature desorption of hydrogen is a good indication of hydrocarbon decomposition on
the surface, leaving carbon fragments which cause coking. Bent’s work showed only iodine remained on
a Cu(111) surface after TPRs of methyl iodide, indicating no coking occurs on this surface.7,13,14 Our
results confirm this, as there is no desorption of hydrogen from Cu(111) as shown in Supplementary
Figure 6. The same is true for the SAA 0.01 ML Pt/Cu(111) surface. However, on the 1.0 ML Pt/Cu(111)
surface significant hydrogen desorption is observed at 500 K. On Pt(111) hydrogen desorbing at this
temperature has been attributed to the decomposition of methylene groups and the results on 1.0 ML
Pt/Cu(111) are consistent with this.10,15
A.6 TPR Results from Pt/Cu Alloys To further analyze the effect of Pt on C-H activation, we performed TPR experiments with alloys
containing various concentrations of Pt, up to 1 ML of Pt. We have previously characterized these alloys
using STM.16,17 Based on our STM results, it is known that up to 0.05 ML of Pt, SAAs are formed. Above
0.05 ML Pt, fingers containing linear chains of Pt grow from the step edges. The higher the coverage, the
10
more prominent these features become. Above 0.5 ML Pt, areas at the edges of the fingers become
Pt(111)-like. Supplementary Figure 7 shows methane and ethene TPR spectra for Cu(111) and six
different Pt/Cu alloys. In general, the methane desorbs at lower temperature with increasing Pt
coverage. The temperature change is most drastic between Cu(111) and 0.01 ML Pt/Cu(111), where the
C-H activation temperature changes from 450 K to 350 K as previously discussed. Beyond this point the
decrease in temperature is nearly linear. At 0.05 ML, where Pt chains begin to form, the temperature
drops further to 323 K. By 0.5 ML, where some Pt(111) sites are present, the C-H bond activates at 280
K. Finally by 1 ML of Pt the C-H bond activates at 250 K as previously discussed. The production of
coupling products also decreases with increasing Pt coverage. There is initially a slight decrease in
ethene production, which is accompanied by an increase in methane production between Cu(111) and
0.01 ML Pt/Cu(111). However, both ethene and methane production remain roughly constant up to 0.1
ML Pt, where the surface still mostly consist of single Pt sites, with some Pt chains. Beyond this point,
ethene production decreases while methane production increases. These trends demonstrate that as
more Pt is added, the surface chemistry of the alloy becomes more similar to Pt(111). The lack of
coupling products at high Pt coverages imply that some carbon species remain on the surface, similar to
Pt(111). High temperature hydrogen as shown in Supplementary Figure 6 is first seen when the surface
coverage of Pt reaches 0.30 ML, the coverage where regions of Pt(111) are expected to start forming.
11
Supplementary Figure 7. TPR spectra of methane and ethene produced via reaction of 4.5 L of methyl iodide on various Pt/Cu(111) alloy surfaces.
A.7 KMC Reaction Network and Lattice Model We model exclusively a single terrace of the (111) surfaces of Cu, Pt/Cu SAA and Pt. Each metal
surface is fcc stacked and we approximate the lattice constant of Pt/Cu as that of pure Cu, due to the
very low concentration of Pt in the alloy. We construct the lattice as in the Graph Theoretical approach
of Stamatakis et al18,19 Our lattice is built of vertices corresponding to high symmetry surface adsorption
sites (top, hollow) that are connected by edges describing neighbouring relations. On pure metal lattices
(M = Pt, Cu), we have only two site types namely top and hollow. On the SAA lattice, the surface
symmetry is reduced by random substitutions of Cu atoms with Pt atoms that are isolated to represent
single, non-clustered surface Pt atoms; the density of these Pt dopant atoms is approximately 3 %. This
reduction in symmetry necessitates the definition of four site types, two for facets of pure Cu as before
and a further two sites for top of Pt single atoms and for adjacent hollow sites. Representations of a
small portion of each lattice are given in Supplementary Figure 8.
12
Supplementary Figure 8. Schematics representing a small portion of the lattice models for the (111) surfaces of Cu (left), Pt/Cu SAA (middle) and Pt (right). Atop sites are represented by circles for Cu atoms (red), Pt atoms as isolated single atoms (light blue) and Pt atoms in the pure metal (dark blue). Threefold hollow sites are depicted as upper and lower triangles for fcc and hcp site types respectively. Neighbouring relations up to second nearest neighbours are represented by connecting lines between sites.
The elementary events are all reversible by microscopic reversibility and include the dissociative
adsorption of CH4 (g) and C-H bond scissions in chemisorbed CHx* fragments. We also account for surface
diffusion of CHx* and H* surface species. A pictorial representation of these non-diffusive events on
Pt/Cu(111) in Supplementary Figure 9.
0 0.5 1 1.5 20
0.5
1
1.5
2
0 0.5 1 1.5 20
0.5
1
1.5
2
0 0.5 1 1.5 20
0.5
1
1.5
2
13
Supplementary Figure 9. Graph-theoretical KMC representations of all non-diffusive elementary events involved in the full dehydrogenation pathway of methane on Pt/Cu(111) SAA; colour schemes are in line with the schematics of the lattice in Supplementary Figure 8.
A.8 Rate Constants from DFT We compute first principles rate constants using DFT and TST. To compute 𝑘𝑇𝑆𝑇 for an elementary
event, we require the activation energy relative to the initial state. From a DFT perspective, this means
we must perform total energy minimization calculations for the initial and final (for the backward
process) states, and also locate the transition state (we do this with the dimer method).20,21 Moreover,
in the computation of 𝑘𝑇𝑆𝑇 we must calculate a prefactor for the exponent in the Eyring equation (1).
H CH
3
CH2
CH
180°
C H 180°
CH2
CH
3
180° 180°
CH2 H
CH3
180° 180°
CH H
CH2
H 180° 180°
CH3
CH2
CH
H
CH3
H 120° 120°
120° 120°
CH2
CH
C
H 120° 120°
14
We achieve this by using harmonic TST which breaks the prefactor into contributions from the molecular
partition functions of the initial 𝑄𝐼𝑆 and transition states 𝑄‡.
𝑘𝑇𝑆𝑇 = 𝜅.
𝑘𝐵𝑇
ℎ.
𝑄‡
𝑄𝐼𝑆. 𝑒𝑥𝑝 (
−𝐸𝑎
𝑘𝐵𝑇),
(1)
where 𝜅 is the transmission coefficient accounting for barrier re-crossings (typically taken to be 1), 𝑄 is
the molecular partition function and 𝐸𝑎 is the activation barrier. Further DFT vibrational frequency
calculations were performed to compute the vibrational contribution in each molecular partition
function (Supplementary Tables 1 and 2) and the rotational contributions were computed using bond
length measurements from the fully relaxed geometry of CH4 (C-H distance = 1.098 Å).
A.9 CxHy Adsorption Geometries We find the most stable configurations for all CxHy species (x={0,1}, y={0,1,2,3,4}) on the (111) surface of
Cu, Pt and single atom alloy Pt/Cu. Three high symmetry adsorption sites are considered for each
adsorbate on each surface, namely atop, bridge and hollow. We give the formation energies of all stable
surface configurations in Supplementary Table 3 and corresponding geometries for the most stable
adsorptions for each fragment in Supplementary Figure 10.
We ensure that formation energy calculations are well converged with respect to both k-points and
plane wave energy cut off (Supplementary Figure 11). We find that a Monkhorst-Pack k-point mesh of
13 × 13 × 1 (21 irreducible points) and a plane wave energy cut off with this grid size of 400 eV are
sufficient to ensure less than 4 meV variation between calculations of M-CH3 formation energy with up
to 72 irreducible k-points and a cut off energy of 600 meV.
15
Supplementary Figure 10. The most stable configurations for each CxHy species (x={0,1}, y={0,1,2,3,4}) on Cu(111) (top), Pt/Cu(111) SAA (middle) and Pt(111) (bottom).
Cu(111)
Pt/Cu(111)
Pt(111)
CH4 CH3 CH2 CH C H
16
Supplementary Figure 11. k-point (using 600 eV cut off) and planewave energy cut off (using m = 13 k-points) convergence plots for CH3 chemisorption on Cu(111), Pt/Cu(111) SAA and Pt(111).
5 10 15-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
m x m x 1 k-point mesh (m)
Form
atio
n En
ergy
(eV)
Pt(111)
5 10 150.3
0.32
0.34
0.36
0.38
0.4
0.42
0.44
0.46
0.48
m x m x 1 k-point mesh (m)
Form
atio
n En
ergy
(eV)
PtCu(111) SAA
5 10 150.78
0.8
0.82
0.84
0.86
0.88
0.9
m x m x 1 k-point mesh (m)
Form
atio
n En
ergy
(eV)
Cu(111)
200 300 400 500 600
0
0.2
0.4
0.6
0.8
1
1.2
Kinetic Energy Cutoff (eV)
Form
atio
n En
ergy
(eV)
Cu(111)PtCu(111) SAAPt(111)
17
Vibrational Frequencies / cm-1
ν1 ν2 ν3 ν4 ν5 ν6 ν7 ν8 ν9 ν10 ν11 ν12 ν13 ν14 ν15
PtCu(111) SAA
TS4 1848 547 440 407 320 822i
TS3 3039 1878 923 555 536 383 359 336 760i
TS2 3032 2965 1943 1356 878 789 575 508 310 277 59 705i
TS1 3094 3043 2964 1965 1410 1394 1176 842 759 399 109 93 61 21 890i
Cu(111)
TS4 1687 475 458 449 150 817i
TS3 3020 1388 816 599 586 527 415 292 827i
TS2 2952 2908 1678 1313 781 575 457 384 265 166 138 843i
TS1 3072 3067 2957 1437 1397 1322 1083 764 619 447 329 144 81 41 891i
Pt(111)
TS4 1879 639 550 472 418 942i
TS3 2967 1881 1089 738 593 502 427 416 472i
TS2 3036 2950 1993 1334 1020 928 713 559 350 333 175 757i
TS1 3097 3031 2941 1709 1389 1361 1181 855 830 442 327 115 111 62 913i
Supplementary Table 1. Vibrational frequencies (computed from DFT using the finite differences method) for each transition state for C-H scissions on each surface. Imaginary frequencies correspond to unstable modes and are not incorporated into the partition function.
18
Vibrational Frequencies / cm-1 ν1 ν2 ν3 ν4 ν5 ν6 ν7 ν8 ν9 ν10 ν11 ν12
PtCu(111) SAA CH3 3033 3032 2944 1396 1393 1153 725 725 474 83 81 43 CH2 3012 2942 1335 716 666 546 441 310 36
CH 3001 685 602 571 397 357 C 651 415 392
H 1424 581 480 Cu(111)
CH3 2915 2914 2853 1355 1355 1164 533 532 320 279 142 140 CH2 2956 2825 1313 646 461 397 311 234 101
CH 3005 596 595 540 387 383 C 494 483 481
H 1041 830 822 Pt(111)
CH3 3039 3035 2934 1385 1384 1170 765 764 509 111 109 22 CH2 3014 2926 1327 874 701 653 571 391 167
CH 2996 767 766 598 467 466 C 621 621 460
H 1039 597 583 Supplementary Table 2. Vibrational frequencies (computed from DFT using the finite differences method) for chemisorbed species in the most stable site type on each surface.
Species Site Type Formation Energy (eV)
PtCu(111) SAA CH4 atop -0.21 CH4 bridge - CH4 fcc - CH3 atop 0.44 CH3 bridge - CH3 fcc - CH2 atop 1.94 CH2 bridge 1.57 CH2 fcc - CH atop 3.57 CH bridge - CH fcc 2.13 C atop 4.41 C bridge - C fcc 3.36 H atop -0.27 H bridge - H fcc -0.29
19
Cu(111) CH4 atop -0.19 CH4 bridge -0.18 CH4 fcc -0.19 CH3 atop 0.88 CH3 bridge - CH3 fcc 0.68 CH2 atop 2.56 CH2 bridge 1.88 CH2 fcc 1.72 CH atop 4.08 CH bridge 2.52 CH fcc 2.37 C atop 5.72 C bridge - C fcc 3.79 H atop 0.33 H bridge - H fcc -0.26
Pt(111) CH4 atop -0.24 CH4 bridge -0.22 CH4 fcc -0.22 CH3 atop 0.01 CH3 bridge 0.64 CH3 fcc 0.73 CH2 atop 1.47 CH2 bridge 0.58 CH2 fcc - CH atop 3.29 CH bridge 1.22 CH fcc 0.53 C atop 3.87 C bridge 2.43 C fcc 1.62 H atop -0.52 H bridge -0.44
H fcc -0.46
Supplementary Table 3. Formation energies in eV with respect to CH4(g) and H2(g) for CxHy fragments x={0,1}, y={0,1,2,3,4}. Adsorption sites whereby a stable configuration cannot be found have formation energies reported as a dash.
20
Surface Reaction EF(TS) (eV) Ea (eV) ΔERxn (eV) Cu(111) CH4* CH3* + H* (TS1) 1.29 1.48 0.61 CH3* CH2* + H* (TS2) 2.10 1.42 0.78 CH2* CH* + H* (TS3) 2.68 0.96 0.40 CH* C* + H* (TS4) 4.34 1.97 1.17 Pt/Cu(111) CH4* CH3* + H* (TS1) 0.71 0.92 0.36 SAA CH3* CH2* + H* (TS2) 1.56 1.13 0.85 CH2* CH* + H* (TS3) 2.41 0.84 0.27 CH* C* + H* (TS4) 3.78 1.64 0.94 Pt(111) CH4* CH3* + H* (TS1) 0.34 0.58 -0.27 CH3* CH2* + H* (TS2) 0.70 0.69 0.05 CH2* CH* + H* (TS3) 0.82 0.23 -0.57 CH* C* + H* (TS4) 1.86 1.32 0.56
Supplementary Table 4. Formation energies of the transition state (𝑬𝑭(𝑻𝑺)) with respect to gaseous CH4 (g) and H2 (g), activation energies (𝑬𝒂) and reaction energies (𝚫𝑬𝑹𝒙𝒏 = 𝑬𝑭(𝐅𝐢𝐧𝐚𝐥) − 𝑬𝑭(𝐈𝐧𝐢𝐭𝐢𝐚𝐥)) for C-H bond scissions in methane derivatives on Cu(111), Pt/Cu(111) SAA and Pt(111).
A.10 Energy Landscape for the Full Dehydrogenation of Methane Along the reaction coordinate in Figure 2c in the main manuscript, the potential energy plot for
methane dehydrogenation on Pt/Cu(111) SAA is similar to Pt(111) for the first dehydrogenation steps,
whereas more Cu(111)-like as the reaction progresses. To evaluate the “intermediacy” of the SAA
reaction barrier with respect to pure Cu and Pt, we introduce parameters 𝛼 and 𝛽 where
𝛼 =
𝐸𝑎𝐶𝑢 − 𝐸𝑎
𝑆𝐴𝐴
𝐸𝑎𝐶𝑢 − 𝐸𝑎
𝑃𝑡 , 𝛽 = 1 − 𝛼. (2)
A value of 𝛼 = 0 versus 𝛼 = 1 indicates that the SAA activation barrier is equal to the barrier on Cu(111)
versus Pt(111), respectively. The first C-H scission (CH4* to CH3*) on the SAA has an activation barrier
closest to that on Pt(111) (𝛼 = 0.64, 𝛽 = 0.36). The next scission (CH3* to CH2*) is closer to the barrier on
Cu(111) (𝛼 = 0.40, 𝛽 = 0.60) and the penultimate scission (CH2* to CH*) has an activation energy most
similar to that on Cu(111) (𝛼 = 0.16, 𝛽 = 0.84). The final barrier (CH* to C*) is almost exactly
intermediate (𝛼 = 0.51, 𝛽 = 0.49). Interestingly, the change in relative intermediacy of the barrier with
respect to pure Pt(111) and Cu(111) is reflected experimentally in the ability of Pt/Cu(111) to activate
methyl, but not further dehydrogenation steps that lead to coking.
21
A.11 Calculations for Subsurface Pt Atoms We have performed DFT calculations of CHx fragment adsorption on Cu(111) with an isolated subsurface
Pt atom. The formation energies of all CHx fragments (for x = 0,1,2,3) reveal that their adsorption to the
Cu(111) surface with a subsurface Pt atom is weaker than adsorption either on pure Cu(111) or
Pt/Cu(111) SAA.
Supplementary Figure 12. Slab structures used for the subsurface Pt atom calculations along with the calculated formation energies for four CHx fragments.
The Pt atom in the subsurface is causing expansive strain in the surface layer Cu atoms which reduces
the overlap of the d-band with the molecular orbitals of adsorbing fragments. Thus, the adsorption of
CHx fragments is weakened compared to pure Cu(111) as the bar graph with our calculated formation
energies above reveals. Consequently, BEP scaling relations suggest the activation energy for C-H
scissions for the isolated subsurface Pt system will be greater than that on pure Cu(111), indicating
subsurface Pt atoms would actually have a negative effect on the catalytic performance of the
nanoparticles. The clear improvement in the TOF for Pt/Cu nanoparticles over pure Cu nanoparticles
contradicts the previous point and so we conclude that the observed promotional effect is due to
surface Pt atoms rather than subsurface Pt atoms.
22
A.12 Areas of and Segregation Energies on (110) and (100) Surfaces A Wulff construction by Tran et al. using Cu surface energies from DFT indicates that the surface area
fractions of (hkl) facets on a model truncated octahedral Cu nanoparticle are as follows: (111) = 0.6,
(100) = 0.2, (110) = 0.0, (221) = 0.03, (331) = 0.06, (311) = 0.08, (310) = 0.01, (210) = 0.03.22 Excluding
corners and edges, the relevant facets with the highest surface areas are the (111) and (100).
Supplementary Figure 13. Wulff construction for Cu nanoparticle (information from Tran et al.).22
We have calculated the segregation energy ΔESeg which is the energy difference for a Pt atom in the bulk
of a Cu slab (EPt-bulk) compared to the energy of the Pt atom in the surface layer of a slab of the (111) and
(100) facets (EPt-surf) i.e. ΔESeg = EPt-surf – EPt-bulk. More negative values of ΔESeg correspond to a more
thermodynamically favourable segregation of the Pt atom from the bulk into the surface. The values of
ΔESeg for Pt/Cu(111) and Pt/Cu(100) are -0.02 eV and +0.25 eV respectively. It is evident that Pt is much
more likely to reside in the (111) facet than the (100) facet from a thermodynamic point of view. Given
this and the very low (1:100) atomic ratio of Pt:Cu, the likelihood of the very few Pt atoms being in the
(111) surface is much greater than the (100) for SAAs. Thus, we conclude that even in the case of the
nanoparticles, it is most relevant to consider the (111) facet for our calculations.
(100) (100)
(111)
(111)
23
Supplementary Figure 14. Segregation energies for the Pt atoms on the Pt/Cu(111) versus (100) surface.
24
A.13 STM of Surfaces Annealed to 1000 K
Supplementary Figure 15. STM images of Cu(111) and 0.01 ML Pt/Cu(111) annealed to 1000 K. Both surfaces had previously been populated with methyl iodide. Scale bars = 5 nm.
Annealing both Cu(111) and 0.01 ML Pt/Cu(111) to 1000 K after reaction with MeI removes all
iodine from the surfaces. After a 1000 K anneal, the Pt in the SAA surface has diffused into the bulk so
the SAA surface appears identical to bare Cu(111). Most importantly, on both surfaces there is no
evidence of residual carbon species once the iodine is removed. Previous research demonstrated that
no carbon remains on Cu(111) after the reaction of methyl iodide, but this experiment shows the same
result for the SAA surface. 7,13,14
25
B. Ambient Pressure C-H Activation Studies
B.1 Supplementary methods for Nanoparticle Studies Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was conducted using a
Thermo Nicolet iS50 FT-IR equipped with a Harrick DRIFTS cell.23 The samples were pre-reduced in H2 at
400 °C in situ. Samples were degassed with helium for 15 minutes before introducing CO. After 10
minutes CO exposure, helium was passed through the sample while the IR spectra were collected.
X-ray absorption spectroscopy (XAS) was conducted at beamline 12-BM at Applied Photon
Source (Argonne National Laboratory). The samples were pre-reduced in H2 at 400 °C in situ. Then XAS
spectra are collected in helium at ambient temperature. XAS data was processed and analysed using
Athena and Artemis.
Temperature- programmed- oxidation (TPO) experiments were performed in a packed-bed flow
microreactor (L = 22 inch, O.D. = 1/2 inch) following the 12-hour butane-D2 isotope scrambling (B-D
scrambling) experiments as described in the main text. The samples were treated in Argon (pure, 50
ml/min) at 300 °C for 2 h prior to the TPO. After cooling to ambient temperature in Argon, 20 % O2/He
(20 ml/min) was introduced to the reactor and allow the system to stabilize for one hour. To perform
TPO, the temperature was increased from ambient temperature to 600 °C at 3 °C/min. The gas effluent
from the reactor was analyzed by a mass-spectrometer.
TEM was conducted on a JEOL 2010 electron microscope with 200 kV and 107 μA beam
emission. The specimens were obtained by adding one drop of ethanol solution of nanoparticles onto a
carbon film with a nickel covered microgrid.
B.2 Nanoparticle Synthesis This NP synthesis approach used here follows the preparation of PtCu SAAs as reported by Lucci
et al., Liu et al., Shan et al., and Marcinkowski et al.9,17,23,24 This work demonstrated that the dilute metal
26
alloys in single crystals act as good surrogates for nanoparticles. Unlike monometallic nanoparticles in
which reactivity is often dominated by under-coordinated edge atoms, in single atom alloys the most
active site is the single substituted atom. In order to confirm the formation and stability of SAA catalysts
used in this work, we performed CO-IR, TEM, and XAS.
Supplementary Table 5. Sample Information
Sample Composition[a] Preparation Method Pt Species
Cu-NP 3.6 wt.% Cu on SiO2 Colloidal ---
Pt-NP 0.5 wt.% Pt on SiO2 IWI[b] NPs
Pt0.03Cu-SAA 0.25 wt.% Pt, 3 wt.% Cu on SiO2
Galvanic replacement Single atoms in Cu surface[c]
Pt0.01Cu-SAA 0.1 wt.% Pt, 3.4 wt.% Cu on SiO2
Galvanic replacement Single atoms in Cu surface[d]
[a] Determined by ICP-AES.
[b] Incipient wetness impregnation (IWI).
[c,d] Determined by STEM in ref. 9 and EXAFS and CO-IR in this work.
B.3 CO-FTIR for fresh and spent catalysts Supplementary Figures 16 and 17 show the FTIR spectra for Pt0.01Cu-SAA and Pt0.03Cu-SAA with CO
chemisorption, respectively. The peaks at 2130 to 2112 cm-1 are assigned to linearly adsorbed CO on Cu
and Pt respectively.17,25–28 Peaks corresponding to linearly adsorbed CO and bridged CO on extended Pt
surfaces are not observed in the SAA samples, which typically appear at 2018-2043 cm-1 and 1780-1820
cm-1, respectively.28–32 This suggests that Pt atoms are fully isolated in the copper surface. The CO peaks
observed are not shifted as the CO coverage is changed, which indicates small dipole-dipole interaction
because of the high dispersion of individual Pt atoms and low CO coverage on the Cu surface. Similarly,
Supplementary Figures 18 and 19 which are FTIR spectra for used Pt0.01Cu-SAA and Pt0.03Cu-SAA catalysts
do not show any linearly or bridged adsorbed CO on extended Pt surfaces even when gas phase CO is
27
present (indicated by the 2170 cm-1 peak). This confirms that Pt atoms remain isolated after long term B-
D scrambling experiments.
28
Supplementary Figure 16. FTIR spectra of chemisorbed CO molecule on Pt0.01Cu-SAA catalyst. The time values on the graph indicate the time of helium gas purge after CO exposure.
29
Supplementary Figure 17. FTIR spectra of chemisorbed CO molecule on the Pt0.03Cu-SAA catalyst. The time values on the graph indicate the time of helium gas purge after CO exposure.
30
Supplementary Figure 18. FTIR spectra of chemisorbed CO molecule on the used the Pt0.01Cu-SAA. The time values on the graph indicate the time of helium gas purge after CO exposure.
31
Supplementary Figure 19. FTIR spectra of chemisorbed CO molecule on the used Pt0.03Cu-SAA catalyst. The time values on the graph indicate the time of helium gas purge after CO exposure.
B.4 Extended X-ray absorption fine structure (EXAFS) characterizations
Supplementary Figures 20 and 21 show the EXAFS of Pt0.01Cu-SAA, Pt0.03Cu-SAA and Pt foil
plotted in k-space and Fourier transform space (R-space). Pt0.01Cu-SAA and Pt0.03Cu-SAA show Pt-Cu
interaction peak at lower R value in comparison to that of Pt-Pt interaction peak of Pt foil. EXAFS model
fitting suggests there is no Pt-Pt interaction in Pt0.01Cu-SAA or Pt0.03Cu-SAA in the first coordination shell
but only in Pt-Cu interactions (Supplementary Table 6). This further confirms that Pt atoms are fully
isolated in the Cu NPs. The first shell bond length (2.59 and 2.61 Å) of SAA NPs is 100% Pt-Cu without Pt-
Pt contribution. This is consistent with the formation of SAAs.
32
Supplementary Table 6. EXAFS model fitting.
Sample Shell CN[a] R[b] (Å) σ2 (Å2) R-factor
Pt foil Pt-Pt 12 2.764±0.002 0.004 0.001
Pt0.01Cu-SAA Pt-Pt Pt-Cu
0 10.51±1.62
2.591±0.004
0.005
0.013
Pt0.03Cu-SAA Pt-Pt Pt-Cu
0 10.80±1.75
2.609±0.004
0.006
0.016
[a] CN, coordination number. [b] R, distance between absorber and backscattered atoms. R-factor, closeness of the fit, if < 0.05, consistent with broadly correct models.
Supplementary Figure 20. In situ Pt LIII EXAFS of (a) Pt0.01Cu-SAA, (b) Pt0.03Cu-SAA and (c) Pt foil plotted in k space.
33
Supplementary Figure 21. Fourier transform of k3-weighted Pt LIII EXAFS of (a) Pt0.01Cu-SAA, (b) Pt0.03Cu-SAA and (c) Pt foil plotted in R space
B.5 B-D scrambling Supplementary Figure 22 shows that the Cu-NPs are not active for the B-D scrambling reaction
until above 500 °C.
A linear relationship between the B-D scrambling rate and Pt loading in the catalysts was
demonstrated for the PtCu SAA catalysts, as shown in Supplementary Figure 23. This corroborates the
fact that Pt atoms are isolated in Pt0.01Cu-SAA and Pt0.03Cu-SAA.
We found the (non SAA) Pt0.39Cu catalyst is not stable in B-D scrambling condition, as shown in
Supplementary Figure 24. Pt clusters are formed at this Pt loading level as demonstrated previously.12 In
the first cycle, the reaction rate starts to drop at 390 °C, even with increasing temperature. Moreover,
the reactivity of the catalyst decreases with cyclic operation.
34
Supplementary Figure 22. Temperature programmed surface reaction data for the B-D scrambling reaction over Cu-NP catalyst.
Supplementary Figure 23. B-D scrambling rate as a function of Pt loading in the catalysts. Gas composition: 5% butane, 2% deuterium and balance argon. 50 ml/min, 100 mg catalysts.
35
Supplementary Figure 24. Reactor studies of B-D scrambling. TPSR data for the B-D scrambling reaction over Pt0.39Cu followed by mass spectrometry. (5 °C/min) Gas composition: 5% butane, 2% deuterium and balance argon. 50 ml/min, 100 mg sample.
B.6 Nonoxidative dehydrogenation of butane and propane To investigate the coke resistance of PtCu SAAs under realistic reaction conditions, we
conducted the nonoxidative dehydrogenation of butane to butane over Pt0.01Cu-SAA. We found the
reactivity was stable for at least 52 hours at 400 °C (Supplementary Figure 25) and only minimal carbon
deposition Pt0.01Cu-SAA compared to Pt-NP (Supplementary Figure 26).
Ongoing work in our lab to explore the use of single atom alloys for propane dehydrogenation
shows that the use of SAAs at high temperature is limited by copper sintering, and not by coke
deposition. To understand the deactivation mechanism of these materials TEM studies were performed
on fresh and spent catalysts. It was found that under propane dehydrogenation conditions at 550 oC the
average particle size of Pt0.01Cu SAA catalysts increased from 13nm to 29nm, as shown in Supplementary
Figure 27. Particle sintering for monometallic Pt was minimal (particle size increase from 1.8nm to
2.1nm). TPO of these samples shows that when normalized to the amount of surface Pt, the PtCu-SAAs
show significantly less coke deposition than the monometallic Pt (Supplementary Figure 28). These
results are consistent with deactivation by particle sintering and not coke deposition. We are presently
studying the stabilization of copper NPs at high temperature by other additives.
36
Supplementary Figure 25. Long-term butane dehydrogenation reaction over Pt0.01Cu-SAA catalyst. Reaction condition: 2.5% butane, 5% H2, bal. He. 50 ml/min, 0.1 g catalysts, 400 °C.
Supplementary Figure 26. Comparing the degree of coking in Pt NP vs PtCu SAA catalysts. Temperature–programmed-oxidation (TPO) of (a) Pt-NP and (b) Pt0.01Cu-SAA. (a) and (b) are spent catalysts after butane dehydrogenation reaction. (a) and (b) were treated in a He flow at 300 °C for 1 h before TPO to desorb the hydrocarbon adsorbates. TPO conditions: 10% O2/N2, 50 mL/min, 3 °C/min, 100 mg sample.
37
Supplementary Figure 27. TEM images and particle size distribution (insets) of (a) fresh PtCu-SAA catalysts and (b) spent PtCu-SAA catalysts after PDH at 550 °C.
Supplementary Figure 28. TPO data for Pt-NPs (black) and PtCu-SAAs (red) used in propane dehydrogenation. 100mg spent catalyst. Sample purged in Argon at 300oC followed by heating from room temperature to 600oC at 3oC/min using 20% O2 in Ar, total flow = 50mL/min.
38
C. References
1. Chao-Ming, C. & Bent, B. E. Methyl radical adsorption on Cu(111): Bonding, reactivity, and the effect of coadsorbed iodine. Surf. Sci. 279, 79–88 (1992).
2. Pascal, M. et al. Methyl on Cu(111)–– structural determination including influence of co-adsorbed iodine. Surf. Sci. 512, 173–184 (2002).
3. Raval, R., Parker, S., Pemble, M. & Hollins, P. FT-RAIRS, EELS and LEED studies of the adsorption of carbon monoxide on Cu(111). Surf. Sci. 203, 353–377 (1988).
4. Kirstein, W., Krueger, B. & Thieme, F. CO adsorption studies on pure and Ni-covered Cu (111) surfaces. Surf. Sci. 176, 505–529 (1986).
5. Kneitz, S., Gemeinhardt, J. & Steinrück, H. A molecular beam study of the adsorption dynamics of CO on Ru(0001), Cu(111) and a pseudomorphic Cu monolayer on Ru(0001). Surf. Sci. 440, 307–320 (1999).
6. Bonicke, I., Kirstein, W., Spinzig, S. & Thieme, F. CO adsorption studies on a stepped Cu(111) surface. Surf. Sci. 313, 231–238 (1994).
7. Lin, J.-L. & Bent, B. E. Two mechanisms for formation of methyl radicals during the thermal decomposition of CH3I on a Cu(111) surface. J. Phys. Chem. 97, 9713–9718 (1993).
8. Lucci, F. R., Marcinkowski, M. D., Lawton, T. J. & Sykes, E. C. H. H2 activation and spillover on catalytically relevant Pt–Cu single atom alloys. J. Phys. Chem. C 119, 24351–24357 (2015).
9. Lucci, F. R. et al. Selective hydrogenation of 1,3-butadiene on platinum–copper alloys at the single-atom limit. Nat. Commun. 6, 8550 (2015).
10. Zaera, F. Study of the surface chemistry of methyl iodide coadsorbed with hydrogen on Pt(111). Surf. Sci. 262, 335–350 (1992).
11. NIST Chemistry Webbook. NIST Chemistry Webbook Available at: http://webbook.nist.gov/chemistry/. (Accessed: 19th May 2016)
12. Liu, J. et al. Tackling CO Poisoning with Single-Atom Alloy Catalysts. J. Am. Chem. Soc. 138, 6396–9 (2016).
13. Lin, J.-L. & Bent, B. E. Iodomethane dissociation on Cu(111): Bonding and chemistry of adsorbed methyl groups. J. Vac. Sci. Technol. A 10, 2202–2209 (1992).
14. Jenks, C. J., Bent, B. E. & Zaera, F. The chemistry of alkyl iodides on copper surfaces. 2. Influence of surface structure on reactivity. J. Phys. Chem. B 3017–3027 (2000).
15. Henderson, M. A., Mitchell, G. E. & White, J. M. The chemisorption of methyl halides (Cl, Br, and I) on Pt(111). Surf. Sci. 184, L325–L331 (1987).
16. Lucci, F. R., Lawton, T. J., Pronschinske, A. & Sykes, E. C. H. Atomic scale surface structure of Pt/Cu(111) surface alloys. J. Phys. Chem C. 118, 3015–3022 (2014).
17. Liu, J. et al. Tackling CO Poisoning with single atom alloy catalysts. J. Am. Chem. Soc. 138, 6396–6399 (2016).
18. Stamatakis, M. & Vlachos, D. G. A graph-theoretical kinetic Monte Carlo framework for on-lattice chemical kinetics. J. Chem. Phys. 134, 214115 (2011).
19. Nielsen, J., D’Avezac, M., Hetherington, J. & Stamatakis, M. Parallel kinetic Monte Carlo simulation framework incorporating accurate models of adsorbate lateral interactions. J. Chem.
39
Phys. 139, 224706 (2013).
20. Henkelman, G. & Jónsson, H. A dimer method for finding saddle points on high dimensional potential surfaces using only first derivatives. J. Chem. Phys. 111, 7010–7022 (1999).
21. Henkelman, G., Uberuaga, B. P. & Jónsson, H. Climbing image nudged elastic band method for finding saddle points and minimum energy paths. J. Chem. Phys. 113, 9901–9904 (2000).
22. Tran, R., Xu, Z., Radhakrishnan, B., Winston, D. & Sun, W. Data descriptor: surface energies of elemental crystals. Sci. Data 3, 160080 (2016).
23. Marcinkowski, M. D. et al. Selective formic acid dehydrogenation on Pt-Cu single atom alloys. ACS Catal. 7, 413–420 (2017).
24. Shan, J. et al. Water co-catalyzed selective dehydrogenation of methanol to formaldehyde and hydrogen. Surf. Sci. 650, 121–129 (2016).
25. Greeley, J. et al. CO vibrational frequencies on methanol synthesis catalysts: a DFT study. J. Catal. 213, 63–72 (2003).
26. Mondelli, C., Ferri, D., Grunwaldt, J.-D., Ravasio, N. & Baiker, A. Redox properties of supported copper catalysts studied in liquid and gas phase by in situ ATR-IR and XAS. Catal. Today 178, 124–131 (2011).
27. Nishiyama, H. & Inoue, Y. IRAS study of surface acoustic wave effects on CO adsorbed on Cu surfaces. Surf. Sci. 594, 156–162 (2005).
28. Oxford, S. M. et al. Study of supported PtCu and PdAu bimetallic nanoparticles using in-situ X-ray tools. J. Phys. Chem. C 114, 17085–17091 (2010).
29. Qiao, B. et al. Single-atom catalysis of CO oxidation using Pt1/FeOx. Nat. Chem 3, 634–641 (2011).
30. Chandler, B. DRIFTS studies of carbon monoxide coverage on highly dispersed bimetallic Pt-Cu and Pt-Au catalysts. Catal. Today 65, 39–50 (2001).
31. Ferri, D., Bürgi, T. & Baiker, A. Pt and Pt/Al2O3 thin films for investigation of catalytic solid−liquid interfaces by ATR-IR spectroscopy: CO Adsorption, H2 -induced reconstruction and surface-enhanced absorption. J. Phys. Chem. B 105, 3187–3195 (2001).
32. Bazin, P., Saur, O., Lavalley, J. C., Daturi, M. & Blanchard, G. FT-IR study of CO adsorption on Pt/CeO2: characterisation and structural rearrangement of small Pt particles. Phys. Chem. Chem. Phys. 7, 187 (2005).
