Structural Evolution of Atomically Dispersed PT Catalysts - P Christopher Nat Mat

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Articles

https://doi.org/10.1038/s41563-019-0349-9

Structural evolution of atomically dispersed Pt


catalysts dictates reactivity
Leo DeRita1,8, Joaquin Resasco1,8, Sheng Dai2,8, Alexey Boubnov3, Ho Viet Thang4,
Adam S. Hoffman   3, Insoo Ro   1, George W. Graham2,5, Simon R. Bare3, Gianfranco Pacchioni4,
Xiaoqing Pan2,6,7 and Phillip Christopher   1*

The use of oxide-supported isolated Pt-group metal atoms as catalytic active sites is of interest due to their unique reactivity
and efficient metal utilization. However, relationships between the structure of these active sites, their dynamic response
to environments and catalytic functionality have proved difficult to experimentally establish. Here, sinter-resistant cata-
lysts where Pt was deposited uniformly as isolated atoms in well-defined locations on anatase TiO2 nanoparticle supports
were used to develop such relationships. Through a combination of in situ atomic-resolution microscopy- and spectroscopy-
based characterization supported by first-principles calculations it was demonstrated that isolated Pt species can adopt a
range of local coordination environments and oxidation states, which evolve in response to varied environmental conditions.
The variation in local coordination showed a strong influence on the chemical reactivity and could be exploited to control
the catalytic performance.

T
he synthesis of oxide-supported Pt-group catalysts typically were provided by in situ spectroscopy and microscopy correlated to
produces metal particles with dimensions of a few nanome- first-principles calculations.
tres1. Recent work has shown that Pt-group species can coex- Catalysts were synthesized via strong electrostatic adsorption21
ist as nanoparticles and single atoms2–4, and that careful synthetic (see Supplementary Fig. 1) to deposit [Pt(NH3)4]2+ complexes onto
approaches can produce exclusively single atoms4–8. Interest in the 5-nm-diameter anatase TiO2 nanocrystals at a weight loading of
reactivity of supported isolated Pt atoms (Ptiso) and other rare met- ~0.025%, which correlates to nominally ~0.5 Pt atoms per TiO2 par-
als stems from the maximized metal utilization efficiency, unique ticle22. Through controlled deposition of on average <1 Pt atom per
reactivity or selectivity, connection to organometallic catalysis and TiO2 particle, Pt cluster formation was mitigated by the energeti-
the potential for making well-defined active sites8–10. It has proved cally difficult transfer of Pt atoms between TiO2 particles, enabling
challenging to characterize the intrinsic catalytic activity of these analysis of Ptiso species under conditions where they are mobile but
dispersed active sites on oxide supports at a level that relates local unable to form clusters. Before analysis, catalysts were calcined in
electronic and geometric structure to function. The difficulty arises air at 450 °C to remove remaining ligands from synthesis.
from their atomic dispersion, heterogeneity in the local coordina- To understand the influence of environmental conditions on
tion of active sites on most catalysts (that is, isolated species sit at the local environment of Ptiso species, CO probe molecule Fourier
different sites on the support)11, dynamic changes in local coor- transform infrared (FTIR) spectroscopy was executed at −140 °C
dination under reactive environments12 and often the low load- following pretreatment of the catalyst in oxidative and reduc-
ing of metal that is required to achieve site isolation13. Exploring tive conditions. Figure 1a shows normalized FTIR spectra of CO
the dynamic restructuring of supported metal catalysts has long adsorbed at −140 °C on Ptiso/TiO2 following pretreatment via in situ
been of interest14–16, with recent experimental work analysing the oxidation (300 °C in O2), mild reduction (250 °C in H2) or harsh
dynamic formation and fragmentation of supported ~1 nm metal reduction (450 °C in H2). Different characteristics of adsorbed CO
clusters under reactive environments17–19, but similar insights into were observed: (1) oxidation resulted in an adsorbed CO stretch-
the dynamic behaviour of dispersed Pt-group species have been dif- ing frequency of 2,135 cm−1 with a full-width at half-maximum
ficult to develop20. (FWHM) of 13.5 cm−1, (2) mild reduction resulted in an adsorbed
In this work, using a sinter-resistant catalyst consisting of CO stretching frequency of 2,112 cm−1 and a FWHM of 10.5 cm−1
Ptiso species uniformly deposited at a well-defined location on a and (3) harsh reduction resulted in two CO stretching frequencies
nanoparticle anatase TiO2 support, we demonstrate that environ- centred at 2,077 and 2,050 cm−1 with a FWHM of 28 and 36 cm−1,
mental conditions dictate the local coordination and charge state respectively. These vibrational frequencies are generally consis-
of Ptiso species, which are related to chemical and catalytic reactiv- tent with CO bound to (1, 2) cationic Pt22, and (3) near-neutral,
ity. Insights into the evolution of local coordination and proper- Ptδ+, sites23–26. The local coordination and oxidation state of the
ties of Ptiso species in response to varying environmental conditions Ptiso species can influence the vibrational frequency of CO. Thus,

1
Department of Chemical Engineering, University of California Santa Barbara, Santa Barbara, CA, USA. 2Department of Materials Science and Engineering,
University of California Irvine, Irvine, CA, USA. 3Stanford Synchrotron Radiation Light Source, SLAC National Accelerator Laboratory, Menlo Park, CA,
USA. 4Dipartimento di Scienza dei Materiali, Università di Milano-Bicocca, Milano, Italy. 5Department of Materials Science and Engineering, University of
Michigan, Ann Arbor, MI, USA. 6Department of Physics and Astronomy, University of California Irvine, Irvine, CA, USA. 7Irvine Materials Research Institute
(IMRI), University of California Irvine, Irvine, CA, USA. 8These authors contributed equally: Leo DeRita, Joaquin Resasco, Sheng Dai.
*e-mail: pchristopher@ucsb.edu

746 Nature Materials | VOL 18 | JULY 2019 | 746–751 | www.nature.com/naturematerials


NaTure MaTerials Articles
a b c
1.2 450 °C reduction 1.2 450 °C reduction 450 °C reduction
250 °C reduction 250 °C reduction 250 °C reduction
300 °C oxidation 300 °C oxidation 10–4 300 °C oxidation
2,135 cm–1 2,112 cm–1 2,077 cm–1
1.0 1.0

Normalized absorbance
Normalized absorbance

Rate (mol s–1 gPt–1)


Ea = 48 kJ mol–1
0.8 0.8

0.6 0.6
2,050 cm–1 Ea = 78 kJ mol–1

0.4 10–5
0.4
13.5 10.5
28 cm–1
cm–1 cm–1 Ea = 72 kJ mol–1
0.2 0.2
36 cm–1

0.0 0.0
2,150 2,125 2,100 2,075 2,050 2,025 –100 –50 0 50 100 150 0.24 0.26 0.28 0.30
Wavenumber (cm–1) Temperature (°C) 1/RT (mol kJ–1)

Fig. 1 | Controlling chemical and catalytic reactivity of Ptiso via pretreatment. a, FTIR spectra of CO adsorbed at saturation coverage and −140 °C to
Ptiso/TiO2 following pretreatment via in situ oxidation (300 °C in 760 torr O2), mild reduction (250 °C in 760 torr H2) and harsh reduction (450 °C in
760 torr H2). The spectra are normalized to the maximum intensity of the CO stretch for each pretreatment condition. The band centre is listed above each
spectrum and the FWHM within. b, Normalized intensity of the CO stretch absorbance as a function of temperature during TPD experiments associated
with the three different catalyst pretreatments. c, Arrhenius plots showing the temperature dependence of the per-gram Pt rate (mol s−1 gPt−1) for CO
oxidation on each catalyst from 150 to 200 °C in a 200 sccm flow of 1% CO, 1% O2 and balance He at 760 torr. The apparent activation energy (Ea) is shown
for each pretreatment condition with error bars indicating standard error.

unambiguous determination of the Ptiso oxidation state on the been reported for small Pt clusters, substantiating the assertion that
basis of adsorbed CO stretching frequency alone is not possible27. Pt remained as isolated atoms throughout all pretreatments30. In the
Re-oxidation of the catalyst following harsh reduction and exposure case of oxidized Ptiso/TiO2, CO2 evolution was observed to occur
to CO showed the same sharp stretches and desorption below room simultaneously with CO desorption initiating at −110 °C, and thus
temperature observed for Ptiso exposed to oxidation conditions (see a CO adsorption energy was not estimated (Supplementary Figs. 4
Supplementary Fig. 2 for details), which is in contrast to PtxOy clus- and 5).
ters produced by oxidation of 1 nm Pt nanoparticles that bind CO To relate variations in Ptiso properties to catalytic reactiv-
strongly (desorption above 200 °C)22. This suggested retained exis- ity, CO oxidation was performed under strict kinetic control
tence of Ptiso species following harsh reduction. (Supplementary Fig. 6). This reaction has been extensively studied
The FWHM of adsorbed CO stretches after oxidation (13.5 cm−1) with atomically dispersed Pt catalysts due to the interest in decreas-
and mild reduction (10.5 cm−1) were narrower than reports of zeo- ing Pt loading in automotive exhaust applications31. Figure 1c shows
lite-supported Ptiso species, where well-defined anchoring sites are that the reactivities of Ptiso sites following oxidation and mild reduc-
present (15–25 cm−1)28, and Ptiso species on other oxide supports tion were similar, with apparent activation energies, Ea, in the range
(>25 cm−1 for Ptiso/CeO2, FeOx and so on)4–6. A comparable ~10 cm−1 of 70–80 kJ mol−1. By contrast, following harsh reduction, Ptiso sites
FWHM for the adsorbed CO stretch was reported for Ptiso species showed an increase in activity of two- to fivefold, depending on
at a well-defined location on a Cu2O model system in a surface sci- temperature, and a decrease in Ea to 48 kJ mol−1, close to the Ea for
ence study25, which provides evidence that following oxidation and small Pt clusters on TiO2 of 53 kJ mol−1 (ref. 22). CO and O2 partial-
mild reduction pretreatments the Ptiso species observed here were pressure-dependent measurements and microkinetic modelling
predominately located at a single adsorption site on the support11. suggested that CO oxidation proceeded by a Mars–van Krevelen
The similar FWHM of the CO stretch following both pretreatments mechanism with kinetically relevant O2 dissociation at the interface
suggests that Ptiso species did not move significantly between the of an oxygen vacancy in TiO2 and Ptiso for all pretreatment condi-
two pretreatments, and that a modification in local coordination tions (see Supplementary Figs. 7 and 8 and associated discussion).
or Pt oxidation state induced the change in adsorbed CO stretch The rate law is consistent with reports of metal clusters on reducible
frequency. By contrast, harsh reduction resulted in broadened CO oxide supports, where metal atoms at the interface with the support
stretches with FWHMs of 28 cm−1 and 36 cm−1, accompanied by a serve as the active site32. Ptiso/TiO2 catalysts exhibited comparable
decrease in vibrational frequency. This suggests that Ptiso species apparent activation barriers and turnover frequencies to the small-
were reduced to a near-neutral charge state, became mobile on the est Pt clusters on reducible oxides22, which have enhanced CO oxi-
TiO2 surface and adopted a range of coordination environments. dation activity per surface metal atom compared to small Pt clusters
Following CO adsorption at −140 °C, temperature-programmed supported on irreducible oxides33 and unsupported Pt34.
desorption (TPD) experiments were executed (Supplementary Fig. The results in Fig. 1 suggest that during oxidation and mild
3). The intensities of the CO stretches were normalized to saturation reduction pretreatments Ptiso species did not significantly move on
coverage (Fig. 1a), and are shown as a function of temperature in the TiO2 surface, but existed in distinct local coordination environ-
Fig. 1b. During the TPD measurement, the CO stretch frequency, ments when probed by CO at −140 °C. On the basis of the simi-
symmetry and FWHM were essentially constant for the oxidized larities in site location and steady-state catalytic reactivity, it is
and mildly reduced cases, providing evidence that Ptiso species were hypothesized Ptiso species pretreated by oxidation and mild reduc-
spatially separated and existed in a predominant local environment tion dynamically evolved to similar local coordination environ-
(Supplementary Fig. 3). CO adsorption energies estimated using ments under CO oxidation conditions. By contrast, harsh reduction
Redhead analysis29 on mildly and harshly reduced Ptiso/TiO2 were induced mobility and left Ptiso species in close-to-neutral states. It
84 kJ mol−1 and 115–145 kJ mol−1, respectively (see Supplementary is hypothesized that the near-neutral Pt charge state facilitated the
Fig. 3 and associated discussion30). The inferred CO binding energy rate-limiting O2 dissociation step in CO oxidation, as compared
for the harshly reduced case is significantly weaker than what has to the more cationic state of Ptiso species that existed following

Nature Materials | VOL 18 | JULY 2019 | 746–751 | www.nature.com/naturematerials 747


Articles NaTure MaTerials

a b c d
300 °C, oxidation 250 °C reduction 450 °C, reduction Pt
1.0
0.9 300 °C
0.8

Normalized intensity (a.u.)


oxidation
0.7
0.6
0.5
0.4
1.0 Pt
0.9
0.8 450 °C
0.7 reduction
0.6
0.5
1 nm 0.4
0.3
0 0.5 1.0 1.5 2.0 2.5
Distance (nm)

Fig. 2 | In situ AC-STEM characterization of Ptiso/TiO2. a–c, In situ AC-STEM images of Ptiso/TiO2 after 30 min at different annealing conditions: 300 °C,
760 torr of O2 for 30 min (a); 250 °C, 760 torr of 5% H2 (balanced with Ar) for 30 min (b); 450 °C, 760 torr of 5% H2 (balanced with Ar) for 30 min (c).
The yellow circles identify the same Pt single atom. A false-colouring scale was used in a–c to enhance contrast on the Pt atom. d, Intensity profile of a line
scan (along the yellow arrows) shown in a and c normalized to the identified Pt atom.

pre-oxidation and mild reduction (see discussion in Supplementary measurements, although again we refrain from explicitly assign-
Information, step 4 in mechanism). ing oxidation states based on the XANES alone (see Supplementary
In situ aberration-corrected scanning transmission electron Information for discussion). Extended X-ray absorption fine-
microscopy (AC-STEM) at atmospheric pressure was executed to structure (EXAFS) spectra associated with these states of the cata-
examine the mobility of Pt atoms on the TiO2 support and corrobo- lyst (shown in Fig. 3c and Supplementary Table 1) demonstrate:
rate inferences from infrared measurements. An in situ AC-STEM lack of measurable Pt–Pt scattering for all pretreatments, consis-
experiment was performed on Ptiso/TiO2 during sequential exposure tent with the assertion that Pt remained atomically dispersed in
to 760 torr of O2 at 300 °C, 760 torr of 5% H2 at 250 °C and 760 torr all explored conditions; a decrease in the extent of Pt–O coordina-
of 5% H2 at 450 °C. Small probe currents and minimal acquisition tion comparing oxidizing to reducing conditions; and well-defined
times were used to minimize beam effects (see Supplementary Pt–O and Pt–Ti scattering paths for the pre-oxidized and mildly
Information for discussion). Previous studies have demonstrated reduced catalysts, indicating uniformity in the local coordination
the ability to conduct these measurements without artefacts from of the Pt single atoms.
electron beam effects or sample drift24,35,36. XANES and EXAFS measurements obtained under CO oxida-
Figure 2a shows a high-angle annular dark-field (HAADF)- tion reaction conditions showed that the Ptiso species evolved, with
STEM image of one Ptiso species on TiO2 following 30 min of in the Pt in the oxidized and mildly reduced states approaching a sim-
situ oxidation treatment. After 30 min of exposure to H2 at 250 °C, ilar oxidation state and local coordination environment under CO
this Pt atom remained at the same location on the TiO2 support oxidation conditions (Fig. 3b,c). This rationalizes the similarity in
(Fig. 2b). Following 30 min of exposure to H2 at 450 °C, the Pt atom observed reactivity and is consistent with the similar location of the
moved 1.6 nm, demonstrating mobility imparted to the single Pt Pt atom in both cases. Measurements under CO oxidation condi-
atom without the formation of a Pt cluster. Line-scan intensity tions following harsh reduction showed that the nearly neutral Pt
analysis and in situ STEM experiments as a function of time were state was maintained. This is consistent with the enhanced reac-
used to substantiate the identification of Ptiso species and dem- tivity in CO oxidation observed for this sample, as a more neutral
onstrate that the electron beam did not induce Pt motion (Fig. Pt atom would be expected to more effectively facilitate the rate-
2d and Supplementary Fig. 9). While exposure to H2 at 450 °C limiting O2 dissociation step.
induced mobility of the Pt atom, a relatively stable adsorption site To develop structural models of the local coordination of Ptiso
must have been reached to allow the atom to be imaged. Exposure species on TiO2, ex situ STEM images were collected along various
to H2 at 450 °C also induced minor structural changes in TiO2, zone axes of the TiO2 support for Ptiso/TiO2 following 250 ˚C reduc-
demonstrated by the loss of visible lattice fringes in Fig. 2c and tion in H2 (see Supplementary Information). Figure 4a presents a
Supplementary Fig. 9. The direct imaging of induced Ptiso mobility HAADF-STEM image taken along the [021] zone axis of anatase
only during the harsh reduction pretreatment and the structural TiO2, where (100) and (012) planes are indexed in the atomic-reso-
modification of TiO2 are consistent with inferences made on the lution image. Ti and O atomic columns are not overlapping in this
basis of sample-averaged infrared measurements, substantiating projection and from the Z-contrast the visible spots in Fig. 4a must
the hypothesis that Ptiso species adopted a range of local coordina- correspond to Ti atoms38. The brighter spots (highlighted by the
tion environments following harsh reduction. yellow circles), exhibited a higher Z-contrast than surrounding Ti
In situ X-ray absorption spectroscopy (XAS) experiments were columns, corresponding to the existence of Pt single atoms22. The
executed under pretreatment and reaction conditions to provide intensity profile (Fig. 4b) taken along the lines indicated by the yel-
complementary sample-averaged information regarding the evo- low arrows in Fig. 4a, and along the perpendicular direction across
lution of the Ptiso oxidation state and local coordination37. Figure the same Pt atom (Supplementary Fig. 11), demonstrates that this is
3a shows in situ Pt L3-edge X-ray absorption near-edge structure the case. Thus, following 250 °C reduction Pt atoms sat in place of,
(XANES) spectra for each state of the catalyst (see Supplementary or above the position of Ti columns (see also Supplementary Fig.
Information for experimental details). For the pretreated samples 12). On the basis of the in situ STEM in Fig. 2, it is assumed that
(before CO oxidation), the white-line intensity decreased in the Ptiso species must also sit in place of, or above, the position of Ti
order of oxidized  > mildly reduced >  harshly reduced, consis- columns following oxidation, and probably at a range of positions
tent with the trends in oxidation state inferred from the infrared on the TiO2 surface following harsh reduction.

748 Nature Materials | VOL 18 | JULY 2019 | 746–751 | www.nature.com/naturematerials


NaTure MaTerials Articles
a b c
2.0 2.0 12 450 °C reduction
450 °C reduction 450 °C reduction
250 °C reduction 250 °C reduction 250 °C reduction
300 °C oxidation 300 °C oxidation 300 °C oxidation

Fourier transform EXAFS


9
Normalized absorption

Normalized absorption
1.5 1.5

(k2-weighted)
9
1.0 1.0 Reaction conditions

3
0.5 0.5

0
Pretreated Reaction conditions Pretreated
0.0 0.0
11,550 11,560 11,570 11,580 11,590 11,550 11,560 11,570 11,580 11,590 0 1 2 3 4 5 6
Energy (eV) Energy (eV) R (Å)

Fig. 3 | XAS of Ptiso/TiO2 catalysts. a, XANES spectra collected at 200 °C immediately following various pretreatments (shown using solid lines). b, XANES
spectra collected under CO oxidation reaction conditions (200 °C, 1% CO, 1% O2, balance He). Corresponding spectra following pretreatment are shown using
dotted lines to emphasize changes induced by exposure to reaction conditions. The arrows show the change induced by exposure to CO oxidation conditions.
c, EXAFS of catalysts after each pretreatment and under CO oxidation reaction conditions. The magnitude and imaginary portion of the Fourier transforms of
the EXAFS spectra are shown using solid and dotted lines, respectively. Fits are shown in magenta. See also Supplementary Fig. 10 for EXAFS spectra.

Density functional theory (DFT) calculations were per- with a bond distance of 2.0 ± 0.1 Å, consistent with the calculated
formed to identify stable Ptiso structures on the (101) terrace and structure of Pt sitting in an octahedral Ti site at the (145) step edge
(145) stepped surfaces of anatase TiO2 (ref. 39), and to supplement (CN of 6, bond distance of 1.97–2.04 Å). Mild reduction decreased
recently reported calculations (see Supplementary Figs. 13–16 the total first shell Pt–O coordination, with two distinct Pt–O bond
and Supplementary Information for details30). Among the consid- distances of 1.90 ± 0.05 (CN 1.2 ± 0.6) and 2.10 ± .05 (CN 1.7 ± 0.7).
ered structures, (PtO2)ads species at the (101) anatase TiO2 terrace, This showed agreement with the calculated structure of (PtO2)ads
where the Pt atom is bound to two extra O atoms on the surface, at the (145) step edge, with 1 Pt–O bond distance of 1.83 (CN 2)
showed excellent agreement with observed characteristics for and 1 Pt–O bond distance of 2.13 (CN 2). Agreement between fits
mildly reduced Ptiso/TiO2 including the CO vibrational frequency to experimental spectra and theory was also found beyond first-
(2,112 cm−1 measured versus 2,121 cm−1 calculated) and CO adsorp- shell fits. EXAFS fitting was not performed on spectra taken under
tion energy (84 kJ mol−1 measured versus 103 kJ mol−1 calculated). reaction conditions, due to uncertainty in description of a physical
Good agreement was also found between characteristics of (PtO2)ads model for these structures.
at the (145) TiO2 step site and the experimental results, with calcu- On the basis of the comparison of the experimental and cal-
lated CO vibrational frequency and adsorption energy of 2,107 cm−1 culated characteristics of the different states of the Ptiso species, it
and 50 kJ mol−1, respectively. The (PtO2)ads structures at either the is proposed that oxidation of Ptiso/TiO2 resulted in Pt substituting
step or terrace site of TiO2 are consistent with STEM measurements, into the Ti6c position; mild reduction pulled Ptiso out of the lattice to
showing Pt in line with Ti cation columns when observed along the form (PtO2)ads species; and harsh reduction induced the formation
[021] axis (see Fig. 4c). of (PtOH)ads species that were mobile and adsorbed to both step and
Calculations of CO adsorbing to Ptiso species replacing a six- terrace sites on the support (Fig. 4d). Based on the better agreement
fold coordinated Ti (Ti6c) atom at the bottom of the (145) step of the calculated and measured XANES and EXAFS characteristics
edge, or on the (101) terrace, showed barrierless CO2 formation, for the oxidized and mildly reduced Ptiso species at the TiO2 (145)
which agreed with experimental observations for oxidized Ptiso/ step edge, as compared to the (101) terrace, it is hypothesized that
TiO2 (Supplementary Figs. 4 and 5). This Ti6c-substituted Pt posi- Ptiso species were predominantly located at TiO2 step edges. The
tion is vertically in line with (PtO2)ads along the <021> zone axis, proposed Ptiso structures have formal oxidation states of +4 for Pt
consistent with the lack of lateral motion observed for Ptiso spe- substituted at the Ti6c site, +2 for (PtO2)ads and +1 for (PtOH)ads.
cies between oxidation and mild reduction conditions via in situ These charge assignments are consistent with trends from XANES,
STEM. Finally, CO adsorbed to (PtOH)ads species on the TiO2 step infrared and Bader charge analysis (see Supplementary Tables 2 and
and terrace sites showed reasonable agreement with the measured 3 and Supplementary Information for discussion). While Pt charge-
vibrational frequencies (2,068 cm−1, step, and 2,087 cm−1, terrace, state estimations showed consistent trends, each method has unique
calculated, versus 2,050 and 2,077 cm−1, measured) and adsorp- limitations in the ability to quantitatively describe charge states and
tion energies (184 kJ mol−1, step, and 210 kJ mol−1, terrace, versus exact values are dependent on the method used41.
115–145 kJ mol−1, measured), given the known over-prediction of Future analysis aims to interrogate the mechanism of transfor-
CO binding energy in DFT calculations40. Observable changes to mations between these various states of the Ptiso species and how
the TiO2 support after harsh reduction indicate that this pretreat- these states are modified by exposure to CO oxidation conditions.
ment introduced surface defects, resulting in a variety of sites for We expect that these changes were driven not only by energetics
(PtOH)ads species to occupy. of the Pt cation, but also by structural transformations of reduc-
The DFT-calculated structures described above (Pt at the Ti6c ible supports. For example, TiO2 surfaces are known to become
site, (PtO2)ads and (PtOH)ads) were used to calculate Pt L3-edge mobile following high-temperature reduction14. Future work will
XANES spectra, which showed agreement in shape and white-line investigate how these structural transformations depend on choice
intensity with the experimental measurements (Supplementary Fig. of support. It is hypothesized that here the transition from (PtO2)ads
17). Fitting of EXAFS data after pretreatment showed close agree- to (PtOH)ads occurs by release of the first O atom, forming H2O,
ment in coordination numbers (CNs) and bond distances with the and hydroxylation of the second O atom during high-tempera-
DFT-predicted structures (Supplementary Tables 1 and 2, respec- ture reduction, consistent with observations of temperature-pro-
tively). In particular, oxidation resulted in Pt–O CNs of 6.0 ± 0.7 grammed reduction experiments (Supplementary Fig. 18).

Nature Materials | VOL 18 | JULY 2019 | 746–751 | www.nature.com/naturematerials 749


Articles NaTure MaTerials

a b Pt + Ti

Normalized intensity (a.u)


1.0
3.77 Å
(100) 0.8

0.6 Ti Ti Ti Ti

0.4

0.2
2.95 Å
0.0
(012)
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Distance (nm)

1 nm
<021> zone axis <021> ZA

d (PtO2)ads Harsh reduction

Oxidation
(PtOH)ads,terrace

Mild
(PtOH)ads,step
reduction

PtTi6c

Fig. 4 | Structural models of Ptiso/TiO2. a, An ex situ AC-STEM image of Ptiso/TiO2 following reduction at 250 °C along the <021> zone axis of TiO2.
The yellow circles highlight the Pt single atoms. b, Normalized intensity profile taken along the line indicated by the yellow arrows in a (see Supplementary
Fig. 11 for additional line scans). c, DFT-calculated model of the Ptiso/TiO2 along the anatase TiO2 <021> zone axis (ZA) for the (PtO2)ads structure. It
is seen that the Pt is in line with Ti columns, consistent with the STEM image in a. d, Schematic showing the proposed dynamic evolution of Ptiso/TiO2
catalysts following oxidation, mild reduction and harsh reduction (shown left to right, respectively). Structures were derived from DFT calculations and are
consistent with experimental data. Ti, O, Pt and H atoms are blue, red, white and light-pink spheres, respectively.

Ptiso species can adopt a range of local coordination environ- Received: 7 January 2019; Accepted: 15 March 2019;
ments, which dynamically form in response to varied environmen- Published online: 22 April 2019
tal conditions, analogous to the dynamic behaviour of metal ions
in zeolites42. This variation in local coordination showed a strong References
influence on the chemical reactivity and catalytic performance for 1. Campbell, C. T., Parker, S. C. & Starr, D. E. The effect of size-dependent
systems where the Ptiso active site was adsorbed at well-defined loca- nanoparticle energetics on catalyst sintering. Science 298, 811–814 (2002).
tions on the support. This work demonstrates that to understand the 2. Fu, Q., Saltsburg, H. & Flytzani-Stephanopoulos, M. Active nonmetallic
behaviour of atomically dispersed catalysts at the level of defining Au and Pt species on ceria-based water-gas shift catalysts. Science 301,
935–938 (2003).
structure–function relationships, it is not sufficient to merely show 3. Van’t Blik, H. F. J. et al. Structure of rhodium in an ultradispersed Rh/Al2O3
that all metal species are atomically dispersed; a detailed character- catalyst as studied by EXAFS and other techniques. J. Am. Chem. Soc. 107,
ization of the uniformity of the local coordination environment and 3139–3147 (1985).
how this local coordination responds to environmental conditions 4. Qiao, B. et al. Single-atom catalysis of CO oxidation using Pt1/FeOx.
must be considered. Nat. Chem. 3, 634–641 (2011).
5. Jones, J. et al. Thermally stable single-atom platinum-on-ceria catalysts via
atom trapping. Science 353, 150–154 (2016).
Online content 6. Ding, K. et al. Identification of active sites in CO oxidation and water-gas
Any methods, additional references, Nature Research reporting shift over supported Pt catalysts. Science 350, 189–192 (2015).
summaries, source data, extended data, supplementary informa- 7. Kistler, J. D. et al. A single-site platinum CO oxidation catalyst in zeolite
tion, acknowledgements, peer review information; details of author KLTL: microscopic and spectroscopic determination of the locations of the
platinum atoms. Angew. Chem. 126, 9050–9053 (2014).
contributions and competing interests; and statements of data and 8. Matsubu, J. C., Yang, V. N. & Christopher, P. Isolated metal active site
code availability are available at https://doi.org/10.1038/s41563- concentration and stability controls catalytic CO2 reduction selectivity.
019-0349-9. J. Am. Chem. Soc. 137, 3076–3084 (2015).

750 Nature Materials | VOL 18 | JULY 2019 | 746–751 | www.nature.com/naturematerials


NaTure MaTerials Articles
9. Flytzani-Stephanopoulos, M. & Gates, B. C. Atomically dispersed supported 34. Berlowitz, P. J., Peden, C. H. F. & Goodman, D. W. Kinetics of carbon
metal catalysts. Annu. Rev. Chem. Biomol. Eng. 3, 545–574 (2012). monoxide oxidation on single-crystal palladium, platinum, and iridium.
10. Bruix, A. et al. Maximum noble-metal efficiency in catalytic materials: atomically J. Phys. Chem. 92, 5213–5221 (1988).
dispersed surface platinum. Angew. Chem. Int. Ed. 53, 10525–10530 (2014). 35. Dai, S. et al. In situ atomic-scale observation of oxygen-driven core–shell
11. Hoffman, A. S., Fang, C.-Y. & Gates, B. C. Homogeneity of surface sites in formation in Pt3Co nanoparticles. Nat. Commun. 8, 204 (2017).
supported single-site metal catalysts: assessment with band widths of metal 36. Matsubu, J. C. et al. Adsorbate-mediated strong metal–support interactions in
carbonyl infrared spectra. J. Phys. Chem. Lett. 7, 3854–3860 (2016). oxide-supported Rh catalysts. Nat. Chem. 9, 120–127 (2017).
12. Parkinson, G. S. et al. Carbon monoxide-induced adatom sintering in a 37. Resasco, J., Dai, S., Graham, G., Pan, X. & Christopher, P. Combining in-situ
Pd–Fe3O4 model catalyst. Nat. Mater. 12, 724–728 (2013). transmission electron microscopy and infrared spectroscopy for
13. Hoffman, A. S. et al. Beating heterogeneity of single-site catalysts: MgO- understanding dynamic and atomic-scale features of supported metal
supported iridium complexes. ACS Catal. 8, 3489–3498 (2018). catalysts. J. Phys. Chem. C 122, 25143–25157 (2018).
14. Tauster, S. J., Fung, S. C. & Garten, R. L. Strong metal-support interactions. 38. Crewe, A. V., Wall, J. & Langmore, J. Visibility of single atoms. Science 168,
Group 8 noble metals supported on titanium dioxide. J. Am. Chem. Soc. 100, 1338–1340 (1970).
170–175 (1978). 39. Diebold, U., Ruzycki, N., Herman, G. S. & Selloni, A. One step towards
15. Gänzler, A. M. et al. Tuning the structure of platinum particles on ceria bridging the materials gap: surface studies of TiO2 anatase. Catal. Today 85,
in situ for enhancing the catalytic performance of exhaust gas catalysts. 93–100 (2003).
Angew. Chemie Int. Ed. 56, 13078–13082 (2017). 40. Schimka, L. et al. Accurate surface and adsorption energies from many-body
16. Zugic, B. et al. Dynamic restructuring drives catalytic activity on nanoporous perturbation theory. Nat. Mater. 9, 741–744 (2010).
gold–silver alloy catalysts. Nat. Mater. 16, 558–564 (2016). 41. Walsh, A., Sokol, A. A., Buckeridge, J., Scanlon, D. O. & Catlow, C. R. A.
17. Liu, L. et al. Generation of subnanometric platinum with high stability during Oxidation states and ionicity. Nat. Mater. 17, 958–964 (2018).
transformation of a 2D zeolite into 3D. Nat. Mater. 16, 132–138 (2017). 42. Paolucci, C. et al. Dynamic multinuclear sites formed by mobilized copper
18. Moliner, M. et al. Reversible transformation of Pt nanoparticles into single ions in NOx selective catalytic reduction. Science 357, 898–903 (2017).
atoms inside high-silica chabazite zeolite. J. Am. Chem. Soc. 138,
15743–15750 (2016). Acknowledgements
19. Wei, S. et al. Direct observation of noble metal nanoparticles transforming to We acknowledge S. Hanukovich for his assistance with the microkinetic modelling
thermally stable single atoms. Nat. Nanotechnol. 13, 856–861 (2018). in Supplementary Fig. 8 and the TPR experiment in Supplementary Fig. 18. P.C.
20. Liu, L. & Corma, A. Metal catalysts for heterogeneous catalysis: from single acknowledges funding from National Science Foundation (NSF) CAREER grant number
atoms to nanoclusters and nanoparticles. Chem. Rev. 118, 4981–5079 (2018). CBET-1823189. The UCSB MRL Shared Experimental Facilities are acknowledged for
21. Schreier, M. & Regalbuto, J. R. A fundamental study of Pt tetraammine use of the inductively coupled plasma–optical emission spectrometry equipment and
impregnation of silica: 1. The electrostatic nature of platinum adsorption. are supported by the MRSEC Program of the NSF under award no. DMR 1720256;
J. Catal. 225, 190–202 (2004). a member of the NSF-funded Materials Research Facilities Network (www.mrfn.
22. DeRita, L. et al. Catalyst architecture for stable single atom dispersion enables org). The work of H.V.T. and G.P. was supported by the Italian MIUR through the
site-specific spectroscopic and reactivity measurements of CO adsorbed to Pt PRIN Project 2015K7FZLH SMARTNESS. Use of the Stanford Synchrotron Radiation
atoms, oxidized Pt clusters, and metallic Pt clusters on TiO2. J. Am. Chem. Lightsource, SLAC National Accelerator Laboratory, is supported by the US Department
Soc. 139, 14150–14165 (2017). of Energy, Office of Science, Office of Basic Energy Sciences under contract no. DE-
23. Kale, M. J. & Christopher, P. Utilizing quantitative in situ FTIR spectroscopy AC02-76SF00515. Co-ACCESS is supported by the US Department of Energy, Office
to identify well-coordinated Pt atoms as the active site for CO oxidation on of Science, Office of Basic Energy Sciences. TEM experiments was conducted using the
Al2O3-supported Pt catalysts. ACS Catal. 6, 5599–5609 (2016). facilities in the Irvine Materials Research Institute at the University of California-Irvine.
24. Avanesian, T. et al. Quantitative and atomic-scale view of CO-induced Pt The work at UC Irvine was supported by the National Science Foundation through the
nanoparticle surface reconstruction at saturation coverage via DFT grant number DMR-1506535.
calculations coupled with in situ TEM and IR. J. Am. Chem. Soc. 139,
4551–4558 (2017).
25. Therrien, A. J. et al. An atomic-scale view of single-site Pt catalysis for Author contributions
low-temperature CO oxidation. Nat. Catal. 1, 192–198 (2018). L.D. developed the catalyst synthesis; L.D. and J.R. performed infrared characterization
26. Ivanova, E., Mihaylov, M., Thibault-Starzyk, F., Daturi, M. & Hadjiivanov, K. and catalytic measurements; S.D. performed and analysed TEM measurements
FTIR spectroscopy study of CO and NO adsorption and co-adsorption on Pt/ with supervision by X.P. and G.W.G.; L.D., A.S.H., A.B. and I.R. performed the XAS
TiO2. J. Mol. Catal. A 274, 179–184 (2007). supervised by S.R.B.; A.B. performed the XAFS analysis; H.V.T. performed the DFT
27. Aleksandrov, H. A., Neyman, K. M., Hadjiivanov, K. I. & Vayssilov, G. N. Can calculations supervised by G.P.; I.R. performed the inductively coupled plasma–optical
the state of platinum species be unambiguously determined by the stretching emission spectrometry measurements. All authors analysed and interpreted the results
frequency of an adsorbed CO probe molecule? Phys. Chem. Chem. Phys. 18, and contributed to the preparation of the manuscript. P.C. conceived the project and
22108–22121 (2016). oversaw all portions.
28. Zholobenko, V. L., Lei, G.-D., Carvill, B. T., Lerner, B. A. & Sachtler, W. M.
H. Identification of isolated Pt atoms in H-mordenite. J. Chem. Soc. Faraday Competing interests
Trans. 90, 233–238 (1994). The authors declare no competing interests.
29. Redhead, P. A. Thermal desorption of gases. Vacuum 12, 203–211 (1962).
30. Thang, H. V., Pacchioni, G., DeRita, L. & Christopher, P. Nature of stable
single atom Pt catalysts dispersed on anatase TiO2. J. Catal. 367, Additional information
104–114 (2018). Supplementary information is available for this paper at https://doi.org/10.1038/
31. Freund, H.-J., Meijer, G., Scheffler, M., Schlögl, R. & Wolf, M. CO oxidation s41563-019-0349-9.
as a prototypical reaction for heterogeneous processes. Angew. Chem. Int. Ed. Reprints and permissions information is available at www.nature.com/reprints.
50, 10064–10094 (2011). Correspondence and requests for materials should be addressed to P.C.
32. Bunluesin, T., Cordatos, H. & Gorte, R. J. Study of CO oxidation kinetics on
Rh/ceria. J. Catal. 157, 222–226 (1995). Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
33. Allian, A. D. et al. Chemisorption of CO and mechanism of CO oxidation on published maps and institutional affiliations.
supported platinum nanoclusters. J. Am. Chem. Soc. 133, 4498–4517 (2011). © The Author(s), under exclusive licence to Springer Nature Limited 2019

Nature Materials | VOL 18 | JULY 2019 | 746–751 | www.nature.com/naturematerials 751


Articles NaTure MaTerials

Methods error inherent in DFT calculations was partly corrected using the DFT + U
Catalyst preparation. The synthesis of Ptiso catalysts on TiO2 has been approach where the Hubbard’s U parameter was empirically set to 3 eV47. The
reported previously22. Briefly, a pH-adjusted solution (12.2) containing projector augmented wave48,49 method was used and the valence electrons explicitly
tetraammineplatinum(II) nitrate was slowly injected into a solution containing treated were: Ti(3s2, 3p6, 3d2, 4s2), Pt (5p6, 5d9, 6s1), O (2s2, 2p4), C (2s2, 2p2) and
TiO2 (anatase, 5 nm diameter, 290 m2 g−1, US Research Nanomaterials). After H (1s1). We used a plane-wave basis set with a cutoff energy of 400 eV and the
drying, the catalyst was calcined in a tube furnace at 450 °C for 4 h in flowing air calculation was performed at the gamma point. The optimized structures were
after ramping up the temperature at a rate of 10 °C min−1. considered converged when the ionic forces were <0.01 eV Å−1.
We adopted the models of (3 × 1) and (2 × 1) supercell slabs including five
Infrared spectroscopy. FTIR experiments were carried out in a diffuse reflectance atomic layers for the anatase TiO2 (101) or the stepped TiO2 (145) surfaces,
reaction chamber (Harrick Scientific) equipped with ZnSe windows, mounted respectively. A vacuum of more than 15 Å was used to avoid the interaction
inside a Praying Mantis diffuse reflectance adapter (Harrick Scientific), and between slabs50. The TiO2 (101) surface terminates with two types of Ti cation,
coupled to a Thermo Scientific Nicolet iS10 FTIR spectrometer with a liquid- five-fold coordinated Ti (Ti5c) and six-fold coordinated Ti (Ti6c), and two
nitrogen-cooled HgCdTe (MCT-A) detector. During measurements, the FTIR and types of O anion, two-fold coordinated O sites (O2c) and three-fold coordinated
Praying Mantis diffuse reflection accessory were purged with dry N2. O sites (O3c).
CO adsorption energies were calculated as the difference between isolated
Electron microscopy. AC-STEM was performed on a JEOL Grand ARM300CF species and the adsorption complexes. The CO frequencies were computed within
microscope equipped with two spherical aberration correctors, operated at 200 kV. the harmonic approximation, in which CO and its first neighbouring atoms were
HAADF-STEM images were recorded using a convergence semi-angle of 22 mrad, considered in the calculation. In the gas phase, the CO bond length (1.144 Å) and
and inner- and outer- collection angles of 83 and 165 mrad, respectively. To stretching frequency (2,125 cm−1) were obtained at the Perdew–Burke–Ernzerhof
minimize beam irradiation, a small beam current (<10 pA) was used for imaging, level. A scaling factor α = 2,143/2,125 = 1.0085 was applied to all frequencies to
while the acquisition time was always controlled to be less than 12 s for each frame. take into account the difference with respect to the experimental CO frequency in
In situ STEM experiments were performed in a Protochips Atmosphere gas cell the gas phase (2,143 cm−1).
system, which allows for dynamic observation of nanomaterials heated under
atmospheric pressure inside a transmission electron microscope. Data availability
All the data reported in this paper are available from the corresponding author
XAS measurements. XAS spectra were collected at beamline 9-3 at the Stanford upon request.
Synchrotron Radiation Lightsource at the SLAC National Laboratory. In the
experiments, 16 mg of 0.025 wt% Pt/TiO2 was diluted in mesoporous silica (MCM-41)
and loaded into a flow-through quartz tube, with an outer diameter of 3.0 mm and an References
inner diameter of 2.8 mm, held at a 45° angle relative to the X-ray beam43. Fluorescence 43. Hoffman, A. S., Singh, J. A., Bent, S. F. & Bare, S. R. In situ observation of
XAS measurements were collected using a Canberra 100-element Ge detector phase changes of a silica-supported cobalt catalyst for the Fischer–Tropsch
orthogonal (90°) to the beam path with 4 layers of Al foil to filter Ti fluorescence lines, process by the development of a synchrotron-compatible in situ/operando
as well as lead shielding to minimize contributions from scattering of beamline optics. powder X-ray diffraction cell. J. Synchrotron Radiat. 25, 1673–1682 (2018).
A Pt foil reference was scanned simultaneously for energy calibration. 44. Ravel, B. & Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: data analysis
The raw data were processed using the Athena interface of the Demeter for X-ray absorption spectroscopy using IFEFFIT. J. Synchrotron Radiat. 12,
software package44. The spectra were energy-calibrated, merged and normalized. 537–541 (2005).
The EXAFS was extracted in k-space and Fourier-transformation was conducted 45. Kresse, G. & Furthmüller, J. Efficiency of ab-initio total energy calculations
on the k2-weighted EXAFS function (k = 3.0–12.0 Å−1) to amplify the oscillations at for metals and semiconductors using a plane-wave basis set. Comput. Mater.
high k-values. Phase shifts and amplitudes for relevant back-scattering paths were Sci. 6, 15–50 (1996).
calculated using FEFF6. 46. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation
made simple. Phys. Rev. Lett. 78, 1396–1399 (1997).
Kinetic control and reactivity experimental details. After pretreatment and 12 h 47. Chen, H.-Y. T. Y. T., Tosoni, S. & Pacchioni, G. Adsorption of ruthenium
exposure at 200 °C in a feed of 1% CO, 1% O2 and 98% He, temperature-dependent atoms and clusters on anatase TiO2 and tetragonal ZrO2 (101) surfaces: a
measurements were taken between 150 °C and 200 °C, for 2 h at each temperature. comparative DFT study. J. Phys. Chem. C 119, 10856–10868 (2015).
Catalyst dilution studies showed that differences in activity between the catalysts 48. Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50,
are of kinetic origins, rather than a result of mass and heat transport limitations. 17953–17979 (1994).
Partial-pressure-dependent measurements were executed at 200 °C. 49. Kresse, G. & Joubert, D. From ultrasoft pseudopotentials to the projector
augmented-wave method. Phys. Rev. B 59, 1758 (1999).
DFT computational methods. Spin-polarized periodic DFT calculations were 50. Chen, H.-Y. T., Tosoni, S. & Pacchioni, G. A. DFT study of the acid–base
performed using the Vienna Ab initio Simulation Package45. The Perdew–Burke– properties of anatase TiO2 and tetragonal ZrO2 by adsorption of CO and CO2
Ernzerhof generalized gradient approximation was used46. The self-interaction probe molecules. Surf. Sci. 652, 163–171 (2016).

Nature Materials | www.nature.com/naturematerials

You might also like