Multi-modal surface analysis of porous films under operando conditions

The surface science approach to study catalysis begins with a simplified model of a practical catalyst and then builds on this scheme with increasing complexity to bridge the “materials gap.”¹ For example, initial investigations can involve well-defined single crystalline metal surfaces followed by oxide thin films and finally supported nanostructures, which mimic the material complexity of a practical catalyst. In addition to materials considerations, working catalysts operate at elevated temperatures and pressures and often readily undergo surface alterations in response to such harsh environmental conditions. The dynamic surface modifications incurred under working conditions strongly affect the catalyst activity and selectivity due to active site changes on the molecular level. Therefore, a fundamental science approach to uncover structure–property relationships in heterogeneous catalysis reactions at complex surfaces and interfaces is critical to understand and, ultimately, control catalyst performance through the rational design of catalysts. This goal requires advanced multi-modal operando characterization methods to capture changes in surface structure and chemistry of the catalyst under conditions relevant to applications.

The number of active sites for catalytic reactions is directly proportional to the available surface area. Metal catalysts rely upon coordinatively unsaturated sites for reactivity, i.e., open surface, nanoparticle, or single-atom configurations; consequently, the bulk of the material is inactive. Advanced catalytic materials, such as microporous materials, avoid this issue by utilizing a porous scaffold with channels and pores distributed throughout the bulk of the material.² The dimensions of the pore channel system can act as a filter by only allowing gas molecules with a particular size and shape to enter and interact with active sites. Therefore, a major contribution to reaction selectivity depends on shape, confinement, and diffusion effects within the pore channel system. Additionally, the active sites are often embedded in the framework caused by differences in polarity between adjacent atoms. For instance, the aluminosilicate backbone of zeolites features hydroxyl groups in the presence of Al³⁺ sites due to charge imbalance with lattice Si⁴⁺. The charge-balancing hydroxyl group is a Brønsted acid site, which is responsible for many acid-catalyzed reactions, such as cracking, isomerization, and alkylation reactions in fossil fuel production.^3,4 The acid strength and density of active sites in zeolites and other similar materials, i.e., metal–organic frameworks (MOFs),^5–8 prompt changes in reactivity and catalytic activity. Furthermore, reaction mechanisms potentially undergo significant changes as a result of reactant activation inside the pore channel system. Characterization at atomic and molecular level active sites on these microporous materials is very challenging.

As briefly mentioned above, catalysts operating under working conditions can undergo surface alterations and, thus, require operando tools for in situ tracking of physical, structural, and chemical changes.^9,10 Microporous materials can experience topological restructuring,¹¹ compositional changes,¹² variations in acidity,¹³ and deactivation primarily from coking and blockages.¹⁴ Microscopy techniques visually monitor the surface structure and changes in morphology, while spectroscopy tools detect the evolution of chemical species, i.e., catalyst, reactants, intermediates, and products, during a chemical reaction.¹⁵ Combining the two techniques offers a powerful characterization that can yield insight into a catalytic system that cannot be obtained from one technique alone. These methodologies can be applied to microporous films, where the pore structure relates to catalytic mechanisms.

Similarly, a multi-modal approach coupling vibrational and electronic structure investigations is useful to assess reaction intermediates^16,17 in an effort to deconstruct reaction mechanisms and understand the influence of the catalyst, promoters, and poisons.⁹ Vibrational spectroscopy is sensitive to the chemical bonding environment of adsorbed species with a clear spectroscopic detection of similarly bonded species and their adsorption geometries.^18,19 Photoemission spectroscopy is a complementary technique with additional element- and orbital-specificity to identify active site oxidation states and reaction intermediates in addition to quantifying reaction components.^17,20 Such vibrational and electronic measurements are particularly relevant to porous materials. Chemically complex materials, such as MOFs, can have overlapping binding energies for some elements and functional groups; vibrational spectroscopy offers a complementary approach to further identify chemical bonds.

Consequently, multi-modal investigations can shed light on complex systems by probing different aspects of the whole system, i.e., physical, structural, vibrational, and electronic environments, which cannot be accessed with one technique alone. Research investigations implementing a multi-modal approach using microscopy and spectroscopy techniques under UHV and operando conditions are routinely performed on individual instruments. A suite of complementary spectroscopy and microscopy tools at the Proximal Probes Facility in the Center for Functional Nanomaterials at the Brookhaven National Laboratory is uniquely poised to perform multi-modal characterization of systems under in situ and operando conditions. With an integrated sample transfer system and a vacuum suitcase, measurements of the same sample in several stand-alone UHV instruments with in situ measurement capabilities are possible.

The environmentally controlled parameters for each instrument in the Proximal Probes Facility are presented in Fig. 1. Many of the analysis chambers are built to sustain a wide range of gas pressures (10⁻¹⁰ Torr–10 Torr) and sample temperatures (5 K–1500 K) to simulate reaction conditions while simultaneously gathering data. Briefly, spectroscopy techniques include probes to investigate the vibrational landscape {(polarization-modulation) infrared reflection-absorption spectroscopy [(PM-IRRAS)], photothermal infrared spectroscopy (PTIR), scattering-type scanning near-field optical microscopy (s-SNOM), and Raman spectroscopy} and the electronic environment [ambient pressure x-ray photoemission spectroscopy (APXPS)]. Microscopy techniques provide visual information about atomic arrangement [low temperature-scanning tunneling microscopy (LT-STM), room temperature-STM (RT-STM), and reactor-STM (R-STM)] and film morphology [low-energy electron microscopy (LEEM) and scanning electron microscopy (SEM)]. A detailed description of each instrument is provided in Note 1 of the supplementary material. Note that the research investigations presented in this paper were exclusively performed using this suite of instrumentation.

The vacuum suitcase can be used to transport up to five samples from one instrument to another while maintaining pristine sample and film quality under UHV conditions or inert atmosphere. The vacuum suitcase is primarily comprised of commercially available components; however, on some instruments, in-house adaptations or modifications have been utilized (see Fig. 1 of the supplementary material). Briefly, the optimized vacuum suitcase design is comprised of a series of differentially pumped gate valves, a six-way cube, full range pressure gauge, sample receiver stack, linear transfer arms, and a getter pump. The getter pump can maintain pressures better than 1 × 10⁻⁸ Torr for more than two weeks with no power or moving parts, easily accommodating the shipment of samples between institutions. The suitcase can be used to load samples in an inert glove box atmosphere and then mounted into a UHV system without contaminating the sample and thus allowing for the possibility of studying reactive samples (e.g., those containing lithium). The entire apparatus is sealed with a differentially pumped gate valve and can be readily removed, transported, and installed on the vast majority of vacuum systems in the facility. The vacuum suitcase prevents exposure to air and other contaminants during transfer, which decreases the time needed to re-clean or remake air-sensitive samples. Furthermore, multi-modal measurements on the same sample can be achieved with several independent instruments, as opposed to building a complex instrument with many measurement capabilities.

In this paper, we present our integrated approach to characterize several microporous films and their interactions with gases using a combination of microscopy and spectroscopy techniques. Section II describes the thin film synthesis and preparation of the microporous films studied in this paper. Section III offers examples that tackle research questions using a multi-modal approach under in situ and operando conditions for film growth, characterization, and adsorption experiments. We focus on the thin film analogs of industrially relevant catalysts in three main examples: (1) bilayer silica, (2) MFI nanosheets, and (3) UiO-66 thin films. Section IV concludes with a summary and outlook.

A Pd(111) crystal (Princeton Scientific Corporation) was mounted on an Elmitec sample holder with a built-in tungsten filament on the backside of the sample for electron-beam heating. A C-type thermocouple spot-welded to a molybdenum ring makes indirect contact with the sample via a tungsten plate. Temperature readings were calibrated with a Si(111) crystal beforehand, noting the 7 × 7 to 1 × 1 structural phase transition in low-energy electron diffraction (LEED) at 1103 K.²¹ 

The Pd(111) crystal was cleaned by several cycles of argon-ion sputtering and repeated electron-beam heating cycles in an oxygen environment (1020 K, 1 × 10⁻⁷ Torr O₂) for 5 min, followed by vacuum annealing to 1200 K for a few seconds. After imaging using the low-energy electron microscope to verify little to no contaminants populating the Pd surface, the bilayer (BL) silica film was prepared in situ while monitoring the low-energy electron diffraction (LEED) pattern at 17.5 V corresponding to the structure of bilayer silica formation. First, a chemisorbed oxide layer was created after annealing to 1200 K in an oxygen atmosphere of 1 × 10⁻⁶ Torr. Second, using a Si rod, a pre-determined amount of silicon necessary to create a silica bilayer (2 monolayer equivalents, 10 min) was evaporated onto the oxygen-covered surface at room temperature in an oxygen atmosphere of 1 × 10⁻⁶ Torr. Third, the sample was slowly annealed in oxygen ambient to the crystallization temperature (1100 K) and held there for 15 min before slowly cooling down to room temperature. A highly ordered bilayer silica film structure appeared in the LEED pattern and was further characterized with Auger electron spectroscopy and infrared reflection-absorption spectroscopy.

The bilayer silica film on Pd(111) was mounted on a stainless steel flag with an in situ K-type thermocouple for APXPS and IRRAS experiments. The sample was cleaned before gas adsorption studies by oxygen annealing (5 × 10⁻⁵ mbar O₂ for 5 min at 750 K) followed by hydrogen annealing (0.1 mbar H₂ for 1 min at 550 K). The effectiveness of this cleaning protocol was confirmed with APXPS analysis.

Similar surface cleaning and film growth procedures were used when growing a separate silica film on Ru(0001). The bare Ru(0001) crystal was cleaned by alternating cycles of Ar ion sputtering (1 kV) and annealing to 1300 K until the surface was found to be clean by XPS. The main difference in silica film growth between Ru(0001) and Pd(111) is the crystallization temperature, which is 1200 K for the former.²² This film, which consisted of a mix of monolayer and bilayer silica regions, was imaged using low-temperature scanning tunneling microscopy.

Zeolite-based MFI nanosheets were synthesized by seeded growth with bis-1,5(tripropyl ammonium) pentamethylene diiodide (dC5), as reported previously;²³ however, the synthetic conditions were slightly modified to improve thickness uniformity. Briefly, the MFI nanosheets were fractured by using a horn-sonicator. The resulting fractured nanosheets were separated from the large aggregates or MFI nanocrystals with cylindrical morphology by using centrifugation. The nanosheet fragments were further processed with dC5-silica sol at 155 °C for 3 days, yielding rectangular nanosheets with ∼2 µm lateral dimension and 7 nm predominant thickness. The crystallized MFI nanosheets were transferred to a clean Au(111) crystal. Note that the Au(111) single crystal was cleaned through repeated cycles of argon-ion sputtering and vacuum annealing at 900 K.

Using the floating particle coating method, MFI nanosheets were transferred onto the surface of clean Au(111), yielding high-density nanosheet monolayers.²⁴ Specifically, a conical-shape Teflon^TM trough housing the Au(111) crystal was filled with de-ionized (DI) water. A solution of ethanol (5 vol. %) was deposited on the surface of the water, stimulating aqueous dispersion of MFI nanosheets. Upon draining the water in the trough, the nanosheet monolayers at the surface of the water were lowered and subsequently transferred onto the surface of Au(111). The acquired MFI nanosheet monolayers were calcined at 673 K for 6 h to remove the occluded organic structure directing agent in the pores.

A thin film of the zirconium-based metal–organic framework UiO-66 was synthesized using the vapor-assisted conversion method.²⁵ UiO-66 was grown directly on a Au-coated silicon wafer by placing a drop of precursor solution (ZrOCl₂•8H₂0, terephthalic acid) on the substrate and heating 373 K for 3 h in a jar that contained a solvent reservoir. Further details can be found in Ref. 25 where the procedure was followed for film No. 58 presented in the supplementary material of that publication.

The resulting crystalline UiO-66 film on the Au-coated Si wafer was mounted with a Cu spacer on a stainless steel flag with an attached in situ K-type thermocouple for APXPS and IRRAS studies. The sample was heated to 450 K to remove the residual solvent, as verified by IRRAS. Hydroxyl groups located in the zirconium oxide node also underwent evaporation with the removal of solvent contaminants in the vacuum annealing process. To restore the depleted hydroxyl sites, the film was exposed to 0.05 mbar D₂O for 5 min at 300 K. D₂O or H₂O treatment is a common practice for UiO-66 powders and films to saturate the unoccupied zirconium sites induced by vacuum annealing. Both ν(OH) and ν(OD) modes were observed in IRRAS.

Zeolites are a class of microporous aluminosilicates and, in some instances, contain small amounts of metal (Sn, Zn, and Ti)^26,27 to perform reactions such as dehydrogenation, a common reaction in the petrochemical industry. The intrinsic crystalline network of pores and channels provides an effective means of adsorption necessary for heterogeneous catalysis,^27,28 ion-exchange,²⁹ filtration,^30,31 and storage.³² Zeolites are naturally occurring; however, their synthetic counterparts achieve important physical properties not found in nature. For example, a bilayer silicate film, a 5 Å thin film analog of chabazite zeolite,³³ was synthetically produced as an avenue to understand the properties of bulk zeolites.^34–39 Open-facing pore sites not found in bulk zeolites provide a platform to analyze these materials using high-resolution microscopy^40,41 and advanced spectroscopy methods.⁴² In Subsections III A 1–III A 3, we investigate the native physical, vibrational, and electronic properties of a bilayer silica film and probe molecular interactions with CO adsorption studies using a multi-modal approach.

Bilayer silica thin films were grown in situ using a low-energy electron microscope (LEEM) with full-field imaging capability.⁴³ This type of microscopy allows for real-time monitoring of the crystal surface using the microscope image or diffraction pattern to follow the evolution of the film growth through changes in morphology and structure. Initially, the Pd(111) support was cleaned, as verified by LEEM imaging [Fig. 2(a)] and μ-spot diffraction that shows a hexagonal low-energy electron diffraction (LEED) pattern at 35 V indicative of the (111)-facet. Atomic terraces and steps of Pd appear as lines stretching diagonally across the image with a small amount of carbon on the surface, i.e., dark spots.

Structural changes were monitored with LEED during the crystallization of the silica film. An inner hexagon emerged in the diffraction pattern at 17.5 V upon crystallization, indicating a highly ordered film grown in the same direction as the underlying Pd lattice [Fig. 2(b)].⁴⁴ Higher start voltages (35 V) reveal Pd diffraction spots with additional spots from the superstructure of the adsorbed CO molecules from background pressure at temperatures below 573 K. Visualization of the silica film in the LEEM imaging mode reveals a spotted contrast due to the silica film with Pd terraces showing through. Further film characterization is presented in Subsection III A 2 to confirm the growth of bilayer silica spectroscopically and probe film construction on the nanoscale.

To achieve an atomic-level understanding of the silica film structure and morphology, we turn to low-temperature scanning tunneling microscopy (LT-STM). Figure 2(c) shows a silica film grown on Ru(0001), where two different contrasts can be seen (Fig. 2 of the supplementary material shows a reactor-STM image of the same film). In one of them (darker contrast), an expected honeycomb pore structure can be observed. This arrangement has been seen before for both monolayer and bilayer silica films. In the second (brighter) contrast, larger ring sizes can be observed as well. This is consistent with a surface having both bilayer and monolayer regions, as they can both coexist on Ru(0001) when the Si coverage is lower than that needed for bilayer synthesis.⁴⁶ This unique duality is suited for silica films grown on Ru(0001) due to lattice matching; however, Pd(111) is thought to only host bilayer silica films. While nanoscale surface imaging sheds light on atomic arrangement, film morphology, and coverage, imaging alone does not unequivocally confirm film construction, the analysis of the chemical bonding environment using spectroscopic methods is needed.

IR spectroscopy can detect subtle vibrational changes between the normally aligned Si–O–Si linkages of free-standing bilayer silica films compared to the M–O–Si linkages of monolayer silica films that are anchored to a metal substrate (M). IR reflection-absorption spectroscopy (IRRAS) is uniquely poised to detect and amplify the signal from such lateral vibrational stretches in thin films. Figure 2(d) (gray) shows a IRRA spectrum of a bilayer silica film on Pd(111) confirming the signature Si–O–Si vibrational stretch at 1288 cm⁻¹ with an amorphous silica tail at lower wavenumbers.⁴⁴ IRRAS is a global IR technique that is blind to local heterogeneities in the film structure, such as the origin of the amorphous silica signal.

To distinguish local heterogeneities, a multi-modal technique that combines atomic force microscopy (AFM) and photothermal infrared spectroscopy (PTIR) was employed to investigate the bilayer silica film on Pd(111).⁴⁵ PTIR modes include point spectroscopy, IR imaging, and hyperspectral imaging that combines point spectroscopy and imaging to create a grid of spectra on the selected points of the topographic landscape. We utilized both PTIR point spectroscopy and IR imaging in this experimental study. To begin, an AFM topographic map of the bilayer silica film revealed the patches of crystalline film growth running on and between Pd terraces with clusters primarily decorating the step edges [Fig. 2(e)]. PTIR point spectroscopy was utilized to spectroscopically analyze the distinguishing features of the crystalline film and clusters. A PTIR spectrum of the crystalline silica film on a Pd terrace revealed a sharp feature centered at 1256 cm⁻¹ with a broad shoulder extending toward lower wavenumbers, analogous to the spectral features observed in IRRAS [Fig. 2(d), red].⁴⁴ For an explanation of the energy discrepancy of the Si–O–Si vibrational stretch measured with the two different IR techniques, refer to Note 2 of the supplementary material. Another PTIR spectrum probed a cluster decorating the Pd step edges and revealed an attenuated bilayer phonon stretch consistent with amorphous silica present on top of the bilayer film [Fig. 2(d), blue]. Silica clusters likely formed from excess silicon deposited in the film growth process. The PTIR point spectra probed the distinct features present in the AFM topographic map of the bilayer silica film and revealed intensity changes in the Si–O–Si vibrational stretch. Consequently, IR imaging of the bilayer phonon stretch should capture film uniformity.

Figure 2(f) shows a chemical map that follows the spatial evolution of the bilayer phonon stretch at 1256 cm⁻¹ in conjunction with the AFM topographic map in Fig. 2(e). The extended regions of the bilayer silica film show subtle height changes due to Pd steps and within the Pd terraces. These changes are not mirrored in the IR imaging of similar regions but instead show a continuous distribution across multiple Pd terraces, therefore deviating from the AFM height profile and revealing chemical uniformity. The predominant change in IR signal occurs on clusters that show attenuation of the bilayer phonon stretch, as featured in the PTIR point spectra. A multi-modal approach using AFM-based IR spectroscopy to characterize a bilayer silica film ties together both structure and chemical information to demonstrate film uniformity, despite uneven terrain. Next, we investigate the reactivity of the 2D interfacial space under the silica cover, taking advantage of the strong interaction of CO molecules with Pd.

Previous experimental work performed with both synchrotron- and lab-based APXPS^46,47 and IRRAS^48,49 details CO adsorption effects on bare Pd(111) surfaces, and an operando study of CO on a Pd(100) surface with multi-modal APXPS and IRRAS shows the value of combining the two techniques.⁵⁰ We investigate CO adsorption on Pd(111) with an additional interface of the bilayer silica film that acts as a membrane for the permeation of small molecules and potential confinement below the film where the imposed confinement effects can play a significant role in reaction chemistry. CO adsorption is a model reaction for metal surfaces as a first step to understand the mechanism of more complex reactions such as CO oxidation in catalytic converters and CO hydrogenation to produce hydrocarbon fuels. CO has a high binding affinity to the Pd surface and shows similar bonding profiles with and without the presence of the porous silica film, suggesting the infiltration of CO.⁴⁴ Literature suggests favorability of low CO coordination, i.e., CO bound on-top of a Pd atom, designated as onefold atop sites, at room temperature and below.⁵¹ Alternatively, CO binds more strongly between three Pd atoms, commonly referred to as threefold hollow sites, at room temperature and above.⁴⁸ Some literature sources also report a twofold bridged adsorption geometry for CO adsorbates; however, structural considerations using photoelectron diffraction rule out the twofold bridged configuration.⁵² Operando IRRAS of CO adsorption on silica-covered Pd(111) monitors the temperature-dependent CO adsorption geometry in a high-pressure CO environment.

IRRAS is acutely sensitive to surface adsorbates with out-of-plane vibrational stretches and bends. Surface- or interface-confined species are also distinguishable from the gaseous ambient environment. Interestingly, CO adsorption on Pd and other transition metal surfaces has a range of vibrational signatures not only due to the coordination environment but also due to the surface coverage, confirmed with multi-modal diffraction experiments.⁵² Increased CO surface saturation [up to 0.55 ML on Pd(111) at 300 K⁵³] illustrates the strong dipolar coupling and polarization effects that result in spectral shifts of the CO adsorbate signatures, in this case threefold hollow sites.

To easily visualize CO desorption behavior, the difference between IRRA spectra at different temperatures is presented in Fig. 3(a). The temperature was systemically increased from 300 K to 750 K in 50 K steps, and each IRRA spectrum was divided by the previous temperature spectrum. Initial CO adsorption is shown in the top IRRA spectrum (black) at 300 K under 0.01 mbar CO as compared to 300 K in UHV conditions. The bilayer phonon stretch at 1288 cm⁻¹ progressively shifts to lower wavenumbers at elevated temperatures. The different CO coordination environments are immediately observed including onefold atop sites at 2072 cm⁻¹ and threefold hollow sites ranging from 1942 cm⁻¹ to 1832 cm⁻¹. The onefold atop sites overlap with a rotational branch of gas phase CO centered at 2144 cm⁻¹; a clear onefold atop feature can be observed in Fig. 3 of the supplementary material where the p-polarized spectrum was divided by the s-polarized spectrum, therefore canceling gas phase contributions and clarifying adsorbate features. The threefold hollow sites have two configurations including CO molecules adsorbed on hexagonal close packed (hcp) or face-centered cubic (fcc) packing arrangements of Pd sites, resulting in a spread of energies. CO molecules preferentially bind to threefold hcp sites at 1942 cm⁻¹ with a small amount of threefold fcc sites at 1842 cm⁻¹.^52,54 CO desorbs by adsorption site geometry, initially onefold atop, followed by threefold hcp, and finally threefold fcc sites. A similar experiment was repeated on bare Pd(111). On this sample, the IRRA spectra showed nearly equivalent amounts of each threefold site on bare Pd(111) (see Fig. 4 of the supplementary material). Consequently, the silica cover influenced CO binding configurations on the active Pd surface by selectively adsorbing CO on threefold hollow hcp sites as opposed to no preference on the open surface.

Distinguishing between the coordination environments is less evident in APXPS but instead provides new insight into how the metal and bilayer silica film respond to CO infiltration. The CO molecules adsorb immediately upon interaction at 300 K and persist to 400 K, as seen in the continued appearance of the C 1s signal from CO adsorbates [Fig. 3(b)]. The bilayer silica film responds to a different electronic environment with CO adsorbates under the silica overlayer, as evidenced by a rigid shift in Si 2p of 0.3 eV [Fig. 3(c)].^42,55 This electronic change corroborates the permeation of CO under silica to interact with the support. Furthermore, APXPS of Pd 3d_5/2 shows a change in bonding environment characteristic of CO bound species populating the Pd surface, represented by a decrease in the intensity of bulk Pd (green) and an introduction of CO adsorbates classified by coordination site type, including onefold atop (magenta) and threefold hollow (purple) sites. Thus, multi-modal IRRAS and APXPS of CO adsorption on silica-covered Pd(111) illustrates the selectivity of CO adsorption geometry under the silica overlayer and changes in the electric field environment in the 2D confined region induced by interfacial CO adsorbates.

To conclude, following the same material system through growth, characterization, and gas adsorption using a multi-modal approach provided a means to control film growth, understand fundamental materials properties, and unravel reaction mechanisms by monitoring structural, vibrational, and electronic changes. The key advantage of studying one film using multiple measurement techniques is obtaining a complete picture of all aspects of the film. Section III B spotlights a more complex zeolite-based microporous film compared to bilayer silica, namely, MFI nanosheets.⁵⁶ 

MFI is a particularly relevant zeolite for industrial adsorbents due to large-sized channels (6 Å) in the framework.⁵⁷ Unlike bilayer silica films, the porosity of MFI supports selective permeation of relatively-large molecules (up to 0.55 nm) into the channel system and down to the surface region of the support. Shape selectivity is useful for gas separation processes including filtration and molecular sieve technology.⁵⁸ Subsections III B 1 and III B 2 take advantage of coupling imaging and spectroscopy to unpack insights between structural and vibrational information and separately test the adsorption properties of MFI nanosheets using operando IRRAS.

The two-dimensional MFI nanosheets stack as separate layers on top of the metal support, as revealed in a topographic AFM map [Fig. 4(a)] accompanied by a chemical map following the PTIR signature of the MFI nanosheets at 1208 cm⁻¹ [Fig. 4(b)]. Each nanosheet grows as a rectangle with lateral dimensions of ∼2 µm and vertically stack to almost completely cover the surface. Stacked MFI nanosheets exhibit a fairly continuous PTIR signal demonstrating the decoupling effects of the two-dimensional nature of MFI nanosheets. Local PTIR spectroscopy identifies the metal surface from the multi-layered nanosheets with different vibrational information [Fig. 4(c)]. Specifically, the vibrational stretch at ∼1260 cm⁻¹ is present at the surface region but is highly diminished in multi-layered MFI regions. The unexpected vibrational structure on the Au surface may result from a thin silica film formed during calcination of the sample post deposition of the MFI nanosheets on Au(111), a cleaning protocol to remove contaminants after transferring the nanosheets to the metal. This surprising result is the subject of future research work. The ability to distinguish the origin of vibrational frequencies is critical to understanding the system as a whole. Therefore, the multi-modal approach of AFM-based PTIR provides an ideal experimental platform to investigate heterogeneity and identify spatially distributed material properties that would be otherwise lost in a spatially averaged technique. In Subsection III B 2, we incrementally expose MFI nanosheets to p-xylene and observe the vibrational response from MFI nanosheets and adsorbed p-xylene molecules.

Xylene isomers are valuable precursors and widely used as industrial solvents, making them a commodity in many industries such as medicines and plastics. The petrochemical industry is a large producer of xylenes; however, separation of the xylene isomers has proven to be difficult. Here, we provide an example of p-xylene adsorption in nano-sized pores of zeolite-based MFI nanosheets using polarization-modulated IRRAS (PM-IRRAS) using a stand-alone instrument described in more detail in Ref. 59. Note that polarization-modulation experiments using a photoelastic modulator in this instrument allow us to eliminate the contribution of gas phase species to the spectrum.

PM-IRRAS selectively detects p-xylene molecules adsorbed within the channels of MFI nanosheets separately from gas phase species, as illustrated in Fig. 4(d). PM-IRRA spectra were recorded for MFI nanosheets exposed to deuterated p-xylene(dimethyl, D6) at different vapor pressures. Three phonon modes corresponding to the MFI framework vibrations were observed at 1095 cm⁻¹, 1176 cm⁻¹, and 1248 cm⁻¹, respectively [Fig. 4(e)], and a characteristic vibration corresponding to p-xylene[D6] appeared at 1438 cm⁻¹ [Fig. 4(f)]. The MFI framework vibrations are observed at higher wavenumbers in PM-IRRAS compared to PTIR measurements, as explained in Note 2 of the supplementary material. The p-xylene peak at 1438 cm⁻¹ first appears at 0.53 mbar and steadily increased in intensity with higher partial pressures of p-xylene. We carried out the same experiment on the bare Au(111) surface in which no gas phase species are seen in the pressure range used in this work. This allows us to conclude that the p-xylene features observed are related to the molecules within the MFI framework or at the MFI/Au(111) interface. In response to p-xylene adsorption, the intensities and frequencies of the framework phonon modes changed. The intensities of all three features corresponding to the frame vibrations were slightly diminished as the p-xylene partial pressure increased, implying that this change is likely attributed to the interaction between the zeolite framework and adsorbed p-xylene molecules. In addition, the feature at 1176 cm⁻¹ shifted to a lower frequency with increasing p-xylene partial pressure, similar to previous experiments in which a similar red-shift was observed when methanol interacted with the framework.⁶⁰ This frequency shift of the framework vibrational mode is highly sensitive to a low concentration of p-xylene molecules as the frequency change started at 0.12 mbar in the absence of the p-xylene peak at 1438 cm⁻¹. The same phenomenon was previously observed for the case of methanol absorption, where the shift in the framework phonon vibration is seen at coverages below those at which methanol modes are observed. These experiments confirmed p-xylene adsorption into the channels of MFI nanosheets. The next step would be to repeat these studies under a gas mixture of xylene isomers to study the isomeric separation of p-xylene compared to o- and m-xylene. Section III C turns from zeolite-based materials and instead focuses on the budding class of porous materials known as metal–organic frameworks.

Metal–organic frameworks (MOFs) are a class of micro- to meso-porous materials with increased chemical complexity, diversity, and pore size tunability compared to the previous examples of zeolite-based microporous materials.^7,61–63 MOFs are solution processed, the precursor materials of which contain metal oxide complexes and organic linkers that self-assemble into the nodes and connected struts of the open framework structure, respectively. Synthetically tunable pore sizes and variability of chemical species make MOFs extremely useful materials across many industries that rely upon adsorption and separation processes and heterogeneous catalysis reactions.^5,66

A zirconium-oxide based MOF, abbreviated UiO-66, is robust under relatively high temperatures and humidity conditions. UiO-66 is composed of dense Zr₆O₄(OH)₄ clusters linked with benzene carboxylate groups.⁶⁴ The hydroxyl groups are known to desorb upon heating to generate coordinatively unsaturated Zr sites (Zr_cus). These sites can be hydroxylated again or functionalized with a carboxy group. The chemical functionalization and electronic structure of the zirconium oxide nodes play a significant role in the adsorption of chemical warfare agents⁶⁵ and catalytic reactions.⁵ Here, we show how in situ spectroscopy studies yield electronic and chemical structure information relevant to such applications.

Similar to most MOFs, UiO-66 is an electrically insulating material that can be difficult to investigate by electronic characterization methods due to potential charging effects. However, the scanning electron microscopy (SEM) on the ScientaOmicron Nanoprobe system, which is equipped with a UHV Zeiss Gemini Column, is uniquely suited to handle insulating materials due to a low emission current (1 nA) while sustaining a high impact field (10 kV). The SEM image [Fig. 5(a)] shows film domains as opposed to unreacted bulk crystallite formations. Complementary spectroscopic methods were used to confirm the structure by probing bonding environments using Raman spectroscopy, IRRAS, and APXPS.

Figure 5(b) shows a rich vibrational structure in both Raman (dark blue) and IRRA (green) spectra that match UiO-66 reported in literature sources.^64,66 The region between 1800 cm⁻¹ and 800 cm⁻¹ corresponds to vibrational modes of the benzene carboxylate linker groups in the framework, the most intense of which originate from OCO asymmetric and symmetric stretches.⁶⁴ 

Probing the electronic structure of the UiO-66 framework with photoemission techniques requires special treatment to counteract spatial charging effects that persist in insulating materials as a result of the accumulation of photon-generated holes. A consequence of positive charge accumulation near the surface manifests as a second set of core-level features at higher binding energies than the original. For instance, UHV scans of O 1s show a shoulder at a higher binding energy, as shown in Fig. 5(c) (the same effect was observed in Zr 3d and C1s core-levels, see Fig. 5 of the supplementary material). However, charge compensation was successfully demonstrated using two stand-alone treatment methods. First, a low energy stream of electrons (15 V) produced by a commercial electron source neutralized the charge accumulation at the surface of UiO-66. Charge depletion was complete at low emission currents of 1.8 μA and 3.0 µA, as illustrated by the removal of shoulder features in all core-levels. Second, a special feature of a high-pressure analysis chamber is the possibility of counteracting charging effects of insulating materials with a gas environment. The gaseous species donate electrons to the surface, which dissipates the accumulated positive charge. While nearly all gaseous species have this effect to varying degrees, inert gases are used when the goal is electronic characterization of the sample. For the UiO-66 film, XPS data were collected under Ar pressure. The charge dissipation is observed by monitoring intensity changes in the shoulder of O 1s. A diminished shoulder is observed at 1 mbar Ar but is not completely removed until 2 mbar–3 mbar Ar. By maintaining either a low-energy electron beam or high pressure Ar atmosphere, it was possible to measure the complex electronic structure of UiO-66 using APXPS. In Sec. III C 2, we indirectly characterize active sites in UiO-66 using operando IRRAS.

Porous materials including zeolites and MOFs contain acidic sites often contributing to their adsorption and catalytic properties. The ideal structure of UiO-66 has four bridging hydroxyl groups (μ₃-OH) per Zr node, which act as Brønsted acid sites. By dehydroxylating UiO-66, Lewis acid sites are generated in the form of coordinately unsaturated Zr (Zr_cus). IR probe molecules have been previously employed to identify the nature of these active sites and understand the adsorption properties of bulk zeolites and MOFs. Common probe molecules include CO, H₂, N₂, and NO with the selection based on the target adsorption site.⁶⁷ Here, we probe the hydroxylated and Zr_cus sites in the UiO-66 film by using CO as a probe molecule and monitoring key vibrations with IRRAS.

Due to the weak interactions of CO with μ₃-OH and Zr_cus sites, the sample was cooled to 122 K to facilitate CO adsorption. OH groups were observed in the UHV surface characterization of thin film UiO-66 at 3676 cm⁻¹. OH–CO adducts were observed under a pressure of 0.12 mbar CO. The ν(CO) modes identify three CO species including OH–CO adducts at 2155 cm⁻¹, physisorbed CO at 2135 cm⁻¹, and Zr_cus–CO sites at 2179 cm⁻¹ [Fig. 5(d)].^68,69 These results suggest extensive hydroxylation of UiO-66 after initial heat treatment due to a low intensity Zr_cus–CO and almost complete OH saturation of Zr sites due to high intensity OH–CO adducts. In addition, the ν(OH) mode red-shifts by 81 cm⁻¹ upon CO adsorption [Fig. 5(e)]. The surface acidity of UiO-66 is directly linked to hydroxyl acidity and the spectral shift correlates with an average hydroxyl acidity for the MOF family of porous materials (23 cm⁻¹–170 cm⁻¹)⁶⁹ but is considerably lower than most zeolites by a couple hundred wavenumbers.⁷⁰ Tracking IRRAS signatures of CO adducts revealed Brønsted and Lewis acid sites in UiO-66 that are key locations for adsorption and catalysis reactions.

We characterized and investigated gas adsorption on three microporous films (model systems for zeolite-analogs and MOFs) using a multi-modal approach in each case. Each thin film was inspected using microscopy-based probes and further characterized using spectroscopic tools to confirm good film quality before performing gas adsorption measurements. The examples highlight the possibility of comprehensive atomic and molecular characterization of porous catalytic systems. In the case of CO permeating a bilayer silica film to bind with a Pd(111) support, IRRAS data show the adsorption geometry, while XPS data quantify the electronic interaction of the bilayer film after CO adsorption. In the case of MFI films, p-xylene was effectively captured in the channels of MFI nanosheets supported on Au(111) only at near ambient pressures, as characterized by the framework vibrations. Finally, the catalytic performance of many chemical reactions in zeolites is based on the acid strength and density of active sites in the microporous material; analogously, MOFs are thought to have the same physical properties. The surface acidity of a UiO-66 film was indirectly detected by using CO as an IR probe molecule to interact with hydroxyl and Zr_cus sites, which are active sites in catalytic reactions. Overall, the molecular-level analysis of microporous films interacting with gas molecules elucidated adsorption geometries, oxidation state changes, and active site information.

The state-of-the-art in situ and operando instrumentation and integrated transfer system at the Proximal Probes Facility provided a means to study one sample using advanced characterization tools from initial film growth and characterization to gas treatments for adsorption and catalysis reactions. The vacuum suitcase enables these experimental studies by sustaining sample integrity and efficiently transferring samples between multiple stand-alone instruments in the facility. Catalysts are stripped down to model systems that resemble parts of a practical catalyst, and the material complexity is built upon by incorporating complex surfaces and interfaces that mimic an industrial grade catalyst. Furthermore, catalyst performance is tested under working conditions by studying molecular level changes using a multi-modal approach to unravel reaction mechanisms that define catalytic activity, reactivity, and selectivity properties.

See the supplementary material for instrument descriptions, vacuum suitcase designs, and additional data and explanations that support the claims made in the main text of this article.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

This research used resources of the Center for Functional Nanomaterials, which is a U.S. DOE of Science Facility, at the Brookhaven National Laboratory under Contract No. DE-SC0012704. J.-Q.Z. was supported by BNL LDRD Project No. 15-010. For the work related to MFI, we acknowledge support from the Catalysis Center for Energy Innovation, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award No. DE-SC0001004.