In Situ Studies of Chemistry and Structure of Materials in Reactive Environments

See allHide authors and affiliations

Science  14 Jan 2011:
Vol. 331, Issue 6014, pp. 171-174
DOI: 10.1126/science.1197461


Most materials and devices typically operate under specific environmental conditions, many of them highly reactive. Heterogeneous catalysts, for example, work under high pressure of reactants or in acidic solutions. The relationship between surface structure and composition of materials during operation and their chemical properties needs to be established in order to understand the mechanisms at work and to enable the design of new and better materials. Although studies of the structure, composition, chemical state, and phase transformation under working conditions are challenging, progress has been made in recent years in the development of new techniques that operate under a variety of realistic environments. With them, new chemistry and new structures of materials that are only present under reaction conditions have been uncovered.

Author Presentation

Embedded Image

Authors Franklin (Feng) Tao and Miquel Salmeron offer a video discussion of aspects of this topic on the Notre Dame Science Web site.

In reactive and corrosive environments (humid air, reactive gases or liquids, acidic solutions, etc.), the surfaces of most materials are likely to restructure, adapting their geometrical and electronic structure to the environments. Such processes have profound effects on the properties of materials and their applications (1, 2). In many cases, the structure and composition of surfaces in vacuum differ markedly from those in the reactive or corrosive environment where they operate. This difference has been illustrated dramatically by using recently developed in situ microscopy and spectroscopy techniques (315).

Until recently the study of surfaces of materials in such environments could only be performed with techniques based on photons, such as Raman and infrared spectroscopy (11, 16), sum frequency generation (17), and extended x-ray absorption fine structure (18). Resonance techniques, such as Mossbauer spectroscopy (19), nuclear magnetic resonance (20), and a few others (2), have also contributed substantially. However, some of them are sensitive to bulk materials as well as surfaces. Their use in surface studies is limited to special cases, such as grazing incidence geometries. The most notable techniques recently developed for in situ studies include high-pressure scanning tunneling microscopy, environmental scanning/transmission electron microscopy (SEM/TEM), ambient-pressure photoelectron spectroscopy, and in situ scanning transmission x-ray microscopy. The principle of operation of these techniques, along with detailed descriptions of the instruments, can be found in recent literature (3, 7, 2125). Here, we focus on recently uncovered new phenomena that illustrate in a dramatic way how the chemistry and structure of materials change in reactive environments.

Growth of Nanomaterials

The development of improved methods to synthesize semiconductor, metallic, and dielectric nanoparticles depends on a thorough understanding of the growth mechanisms that occur at solid-liquid or solid-gas interfaces. A kinetic model for the growth of particles through colloid chemistry was developed by LaMer, Dinegar, and Reiss in the 1950s (26, 27). In this model, an initial period of rapid self-growth from monomeric species leads to a distribution of particle sizes. Size homogenization is reached progressively through a mechanism called size distribution focusing, where small particles grow at a higher rate than large ones, so that their size catches up with that of the larger ones.

A recent study of the microkinetics of Pt nanoparticle growth in solution was performed in a TEM equipped with a thin liquid cell for in situ growth of nanoparticles (8, 28). This liquid cell consists of a pair of 100-μm-thick silicon wafers, which is coated with 25 nm of silicon nitride on each wafer and sealed with a cover. The center area of the two silicon wafers is then etched to create two electron-transparent windows in the space between a pair of reservoirs containing the Pt reagents and surfactants (22). The spatial resolution achievable using this cell is in the range of nanometers, limited by the thickness of view windows and the liquid in between.

In addition to self-growth, which creates a single nanocrystal, a punctuated growth mode takes place through particle attachment. The sequence of TEM images shows how two or more small particles coalesce into a large one (Fig. 1, A to C). In this event, the coalesced particle is polycrystalline. A relaxation period follows, during which no size increase is observed and the shape changes only slightly (8), while a transition from polycrystalline to single crystalline takes place. The study revealed that the relaxation time increases with particle size and follows a power law. This relaxation contributes to the size focusing because growth of the smaller particles through addition of monomers can continue during the relaxation time and catch up with the large size of the coalesced particles.

Fig. 1

Video frames captured at 12.1 s (A), 24.2 s (B), and 77.0 s (C) in a liquid cell inside the pole pieces of a TEM. They show a particle growing by coalescence of two smaller particles at 12.1 s (A) and the subsequent relaxation. Video frames (D) to (F) show a particle growing through steady accretion of atoms [from (28)].

Another example of in situ studies using TEM is the growth of carbon nanofibers. These materials are grown by dissociation of methane in hydrogen at pressures in the torr range on a nickel catalyst supported on MgAl2O4 at a temperature of 430° and 500° to 540°C (29). Monoatomic steps form on the Ni catalyst surface, with the growth of graphene sheets terminating at each of these steps (Fig. 2). The growth of carbon nanofibers is driven by formation and restructuring of the monoatomic step edges of Ni (Fig. 2, B to D), because migration of Ni atoms from bulk to the Ni-graphene interface continuously creates new active sites for the growth of graphene. Density functional theory (DFT) calculations explained these observations by showing that carbon binds more strongly to undercoordinated sites of the step edge compared with sites on the close-packed facets of the Ni catalyst (29). Thus, the diffusion of Ni atoms from deep layers to surface and the strong binding of carbon atoms to the Ni interface results in the formation of graphite and the reshaping of the catalyst. The visualization of the growth processes at solid-gas interface at the atomic scale revealed in a clear way the continuous generation of catalytic sites during the reaction in a process that is different from the traditional assumption of a fixed number of stationary active sites in heterogeneous catalysis.

Fig. 2

(A) In situ TEM image showing a whisker-type carbon nanofiber. The lattice fringes in the nanofiber correspond to the (002) planes of graphite. Scale bar indicates 5 nm. (B to D) Image sequence of a growing carbon nanofiber extracted from a video sequence. Scale bar in (B) is 5 nm; (C) and (D) have the same size as (B). (B) to (D) show the elongation/contraction process. Schematic drawings below (B) to (D) are included to help locate the positions of monoatomic Ni step edges at the Ni-graphene interface. These images were acquired in situ with CH4:H2 1:1 at a total pressure of 1.6 torr with the sample heated to 500° to 540°C (A) or 536°C [(B) to (D)] [from (29)].

Metal Catalysts in Oxidizing and Reducing Environments

Many heterogeneous catalysts perform oxidation reactions, and the interaction of oxygen with most transition metal surfaces can vary with temperature and pressure. At lower temperatures and pressures, oxygen chemisorbs, forming a bond to a surface metal atom. Further reaction at higher temperatures or pressures can incorporate oxygen into the subsurface region and create a surface layer of metal oxide. The build-up of oxygen through the formation of a subsurface oxide on several noble metal catalysts was suggested to be crucial for the catalytic oxidation of hydrocarbons and carbon monoxide (1, 2, 29, 30).

Oxygen incorporation into the subsurface region of metal catalysts was investigated by quenching single crystal surfaces such as Ru(0001) and Rh(110) after high exposures at low pressure and long dosing time, followed by ex situ examination in ultrahigh vacuum (UHV) (31, 32). However, it is very likely that the catalyst may have reverted partially or totally to its low-pressure phase after removal from the reactor. Thus, the ex situ studies might not provide access to the phases formed at high pressure.

A recent study using ambient pressure–x-ray photoelectron spectroscopy (AP-XPS) illustrates this point by showing how the phase diagram of the palladium-oxygen system exhibits new phases under pressures in the torr range that are not found in ex situ studies (10). In situ x-ray photoelectron spectra of the Pd3d and O1s core levels, which are sensitive to chemical oxidation state, show that three different phases were observed during exposure of Pd(111) to oxygen at a pressure larger than 10−6 torr. These results show that certain phases, such as subsurface oxide and bulk oxide, cannot be formed at a pressure and temperature lower than certain threshold conditions. A phase diagram containing four different oxide structures that were observed as a function of pressure and temperature is shown in Fig. 3A. First, a chemisorbed O (2 × 2) structure is formed at low pressure that is kinetically stable at temperatures below a few hundred degrees Celsius under UHV conditions. Then, the topmost surface layer is oxidized, and a two-dimensional oxide with a 6 × 6 reconstruction is generated when the pressure is larger than 10−6 torr and temperature is below 500° to 630°C. A subsurface oxide is formed at oxygen pressures between 10−4 and 0.3 torr and temperatures below 380° to 480°C. Upon further increasing the pressure of oxygen to the Torr range, PdO bulk oxide is formed. The phases formed under oxidizing conditions can be complicated with numerous kinetically stabilized structures that are thermodynamically metastable but still can play a crucial role during catalytic oxidation and reduction reactions.

Fig. 3

(A) Phase diagram showing the experimentally measured stability regions of the different palladium oxide structures as a function of oxygen partial pressure, p, and temperature. These include a O(two-by-two) chemisorbed phase, a Embedded Image × Embedded Image-monolayer-thick surface oxide, a subsurface oxide, and a bulk PdO. The bottom image is a diagram illustrating the structure of the observed phases. (B) Restructuring of the shell region (three to four atomic layers) of bimetallic nanoparticles of Rh0.5Pd0.5 (~15 nm in diameter) at 300°C in the presence of gases in the torr range [from (5, 10)].

One method for improving metal catalysts with regard to activity and selectivity is to add other metals and form alloys. The ratio of constituent elements in the surface region of an alloy can vary from its bulk composition substantially, even under high vacuum conditions. In situ studies using AP-XPS have revealed that the surfaces of RhxPd1–x alloy catalysts supported on a silicon wafer exhibit different compositions after changes in the composition of the reactive environment (Fig. 3B) (5). In oxidizing environments (pure NO), the surface region is enriched in Rh, whereas the atomic fraction of Pd is increased under reducing conditions. At 300°C, the changes occur rapidly, on the time scale of the experiment, and affect the entire nanoparticle, which becomes equilibrated with the environment by virtue of its small size (~15 nm). The restructuring is driven by the preferential segregation of the constituent metal with lowest surface free energy in reducing environments. Under oxidizing conditions, the formation of Rh-rich shell is driven by the preferential formation of the metallic oxide with higher formation energy.

In situ visualization of Cu nanoparticle catalysts with TEM has also shown that the adsorption energy of reactant molecules on different surfaces can give rise to substantial restructuring (9). Cu nanocrystals supported on oxides are widely used as catalysts in several important industrial reactions, including water-gas shift and the formation of alcohols from CO and H2 (1, 2). At 220°C and in 1.1 torr of pure H2, Cu nanoparticles supported on ZnO have a morphology rich in (111) and (100) surface orientations (Fig. 4A). When the pure H2 environment (Fig. 4A) was switched to a mixture of H2O (0.27 torr) and H2 (0.83 torr) (Fig. 4B), the particles changed shape, exhibiting a higher fraction of (110) facets (9). This process was driven by the high adsorption energy of H2O on the (110) compared with (111) surface, which results in the preferred formation of nanocrystals with a morphology that minimizes the surface free energy. After purging the H2O and restoration of a pure H2 environment, the nanocrystals reverted to their original shape (Fig. 4C) observed at 220°C in 1.1 rorr of H2, with a lower fraction of (110) facets. Clearly, surface faceting is reaction condition–dependent and reversible. It shows that the active surface structure formed under reaction condition could not be observed after a reaction, further suggesting the necessity of studying surface structure and chemistry of materials under reaction conditions.

Fig. 4

In situ TEM images showing the reversible shape change of a Cu nanocrystal catalyst collected under different reactants at 220°C. (A) Pure H2 at 1.1 torr, (B) H2:H2O (3:1) at a total pressure of 1.1 torr, and (C) pure H2 at 1.1 torr. The corresponding Wulff diagrams of the Cu nanocrystals that illustrate the surface planes were inserted [from (9)].

Surface Structure of Catalysts Under High Pressures and Temperatures and in Liquids

From the standpoint of thermodynamics, it is necessary to consider the entropic component in the Gibbs free energy of the system, which consists of the catalyst surface and the gas phase of reactants in a reaction system. The entropic contribution is important at high pressure as shown by the roughly 0.4-eV (~40 kJ/mol) difference between the free energy in UHV (<10−10 torr) and 760 torr at room temperature. This entropic fraction of the free energy is responsible for the formation of very high coverages at temperatures higher than would be needed at lower pressure. The high coverage structures of reactant molecules obtained by cooling the material to cryogenic temperatures may be metastable, and the thermodynamically stable structure may not be accessed because of kinetic barriers that are insurmountable at low temperature.

Scanning tunneling microscopy (STM) studies under high pressure have been recently performed on Pt(111), Ni(111), and Pt(110) single crystal catalysts (13, 33, 34). There is no large-scale change in the surface structure of flat and compact Pt(111) and Ni(111) surfaces from low to high pressure (33, 34). On Pt(111), an increase in the CO pressure from 10−9 to 1 torr caused the coverage to increase from 0.5 to 0.65 and to finally saturate to a value of 0.68 at 100 torr (Fig. 5A).

Fig. 5

(A) Represented coverage of CO on Pt(111) in the presence of CO at various pressures [from (33)]. (B) Coverage of CO on Pt(557) in the presence of CO at various pressures [from (4)]. (C) STM image (40 nm by 50 nm) of Pt(557) with a homogeneous distribution of terraces with a width of 1.2 nm separated by monoatomic steps in UHV; the inset is the structural schematic [from (4)]. (D) STM image (40 nm by 50 nm) of the restructured surface consisting of triangular nanoclusters with a size of ~2 nm; the inset is the structural schematic [from (4)]. (E and F) STM images (100 nm by 100 nm) of the surface of a Au-Ni alloy catalyst in 10 torr of CO at 0 min and 75 min [from (12)].

Movie S1

Animation simulating pressure-dependent restructuring of stepped Pt catalyst , Pt(557). The left-top panel shows how surface morphology changes from low pressure CO (1 x 10-10 Torr) to high pressure (1 Torr). The simultaneous evolution of Pt4f photoemission features is listed on the right-top panel. The corresponding CO coverages on the stepped Pt catalyst are plotted in the right-bottom panel. The structural models calculated with DFT are shown in the left-bottom panel. Note: this movie was made on the basis of data points collected at a few pressures of CO. [Download larger version in Supporting Onine Material.]

Industrial catalysts can contain particles with sizes ranging from 1 nm to a few hundred nanometers (1, 35, 36). Such small particles have a large fraction of atoms at step edges and corners, which are thought to be the active sites. To model such systems, surfaces with steps and kinks of controlled structure and density can be prepared by cutting a crystal at small angles relative to a low Miller index direction (Fig. 5C) (37). Recent studies have shown that the coverage of CO on Pt(557), which consists of 1.2-nm-wide (111)-oriented terraces separated by monoatomic steps, can be as high as 1.0 (Fig. 5B) (4). Most importantly, the high coverage of molecules produced when the pressure is higher than 0.1 torr at room temperature results in a substantial restructuring of the catalyst surface, where the terraces break up into triangular-shaped clusters with lateral dimensions of ~2 nm (Fig. 5D). This restructuring significantly increases the density of Pt atoms with low coordination by the increase of density of step edges (4). The formation of this new equilibrium structure is favored by the large molecular repulsion resulting from the high density of CO and the facile displacement of step atoms compared with atoms in the flat surface with no steps. The energy cost of this rearrangement is compensated by the decrease in surface free energy through the tilting of CO molecules bound to the low-coordinated edge sites of the nanoclusters, which decreases their repulsion (4).

There is an increased interest in the search of bimetallic catalysts for energy conversion and environmental remediation technologies such as control of exhaust emission of vehicles and fuel cell technologies. In these applications, the catalysts operate either under harsh conditions of high temperature and pressure or in acidic solutions. It has been shown that under these conditions some bimetallic catalyst nanoparticles experience evaporation or dissolution and that in many cases one of the constitutive metals is leached away preferentially. An example of this behavior is the evaporation of Ni of Au-Ni catalysts through the formation of nickel carbonyl under reaction condition (12). In situ studies using STM show that adsorption of carbon monoxide could induce changes in surface morphology. Figure 5E shows a STM image of the surface of a Au-Ni alloy catalyst before exposure to CO. Under 10 torr of CO, significant changes in the surface structure occurred in minutes (Fig. 5F), where one-atom-high nanoclusters of Au are formed because of the loss of nickel by formation of carbonyls (12). Fast imaging of the surface morphology revealed that the rate of Ni removal is linearly dependent on the pressure of CO. Theoretical calculations suggest that this phase separation is a thermodynamic process, in which the increased energy spent to compress CO to a high coverage and to form nickel carbonyl outweighs the decreased energy resulting from the formation of Au-Ni alloy.

The leaching of atoms of one constituent element in bimetallic catalysts occurs at solid-liquid interfaces as well. For example, Cu atoms at the surface of Au-rich Cu-Au alloys are leached away under suitable electrochemical conditions (38). Similarly, atoms of 3d elements in Pt-M (where M is a 3d metal) bimetallic catalysts dissolve in acidic solutions under the working conditions of low-temperature fuel cells (39).

Many reactions occur on metal catalysts dispersed on oxide surfaces. The presence of reactant gases can not only change the structure and composition of the metal particles but also induce the formation of an active surface phase of a catalyst. One example is the formation of an active phase in Fischer-Tropsch reactions on Fe2O3 catalysts dispersed on support (14, 36). Upon activation of the catalysts in hydrogen at 750 torr and 350°C, α-Fe, Fe2SiO4, and Fe3O4 are formed. After CO is added to the H2 gas at 250°C, FexCy species are formed. These species are suggested to be the active phase. This active phase is identified by using scanning transmission x-ray microscopy (STXM).

Future Challenges and Opportunities

The most active components in various catalysts and sensors are particles with sizes ranging from 1 nm to 1 μm. Many of their unique chemical and physical properties arise from the interactions of their surfaces with the reactive environments. In heterogeneous catalysis, the complexity of the solid-gas and solid-liquid interfaces and their structural and chemical evolution under reaction conditions make understanding catalytic mechanisms extremely challenging. Such understanding requires obtaining in situ information of catalyst systems, particularly their surface structure and chemistry under reaction conditions. In the past, the repertoire of techniques available for this purpose was limited to photon-based techniques and a few others. This has changed recently because new surface-sensitive techniques involving electrons capable of operating under ambient or near-ambient conditions, like those presented in this article, have been developed. However, there are many important issues in the area of heterogeneous catalysis that are still open.

Most processes in energy conversion, such as photo-driven water splitting, involve reactions occurring at solid-liquid or solid-gas interfaces. In situ studies of such processes are far behind compared with the spectacular advances in materials synthesis of the last decades. To bridge this gap, development of additional techniques and methods with the ability to provide information with high spatial and temporal resolution under reaction conditions is highly desirable. Because many chemical processes occur on nanoscale particles, there should be plenty of new chemistry for nanomaterials under reactive environments. Because catalysis on nanoscale materials at solid-gas or solid-liquid interface is important in most of the processes of energy conversion, in situ studies relevant to energy conversion technologies would be crucial for understanding catalytic mechanisms and designing new and efficient catalytic materials for a wide range of energy conversion processes.

Supporting Online Material

Movie S1

References and Notes

  1. This work was supported by the director, Office of Basic Energy Sciences, Materials Sciences Division of the U.S. Department of Energy under contract no. DE-AC02-05CH11231, and Department of Chemistry and Biochemistry, College of Science, Sustainable Energy Initiative, Radiation Lab, and Office of Research at University of Notre Dame. F.T. would like to acknowledge K. Davis of the University of Notre Dame for assistance with the animation movie.
View Abstract

Navigate This Article