Structure–property relationships describing the buried interface between silicon oxide overlayers and electrocatalytic platinum thin films

Marissa E. S. Beatty , Han Chen , Natalie Y. Labrador , Brice J. Lee and Daniel V. Esposito *
Columbia University in the City of New York, Department of Chemical Engineering, Columbia Electrochemical Energy Center, Lenfest Center for Sustainable Energy, 500 W. 120th St., New York, NY 10027, USA. E-mail: de2300@columbia.edu; Tel: +1 212-854-2648

Received 19th July 2018 , Accepted 10th October 2018

First published on 10th October 2018


Encapsulation of an active electrocatalyst with a permeable overlayer is an attractive approach to simultaneously enhance its stability, activity, and selectivity. However, the structure–property relationships that govern the performance of encapsulated electrocatalysts are poorly understood, especially those describing the electrocatalytic behavior of the buried interface between the overlayer and active electrocatalyst. Using planar silicon oxide (SiOx)-encapsulated platinum (Pt)/titanium (Ti) bilayer thin films as model electrodes, the present study investigates the physical and electrochemical properties of the SiOx|Pt buried interface. Through a combination of X-ray photoelectron spectroscopy and electroanalytical measurements, it is revealed that a platinum oxide (PtOx) interlayer can exist between the SiOx overlayer and Pt thin film. The thickness and properties of the PtOx interlayer can be altered by modifying (i) the thickness of the SiOx overlayer or (ii) the thickness of the Pt layer, which may expose the buried interface to oxophilic Ti. Importantly, SiOx|Pt electrodes based on ultrathin Pt/Ti bilayers possess thinner PtOx interlayers while exhibiting reduced permeabilities for Cu2+ and H+ and enhanced stability during cycling in 0.5 M H2SO4. These findings highlight the tunability of buried interfaces while providing new insights that are needed to guide the design of complex electrocatalysts that contain them.


I. Introduction

Electrocatalysts are essential components in fuel cell and electrolysis technologies, which are expected to play key roles in a sustainable energy future by enabling the efficient interconversion between renewable electricity and chemical fuels.1–3 For most commercial electrolyzers and fuel cells, electrocatalysts take the form of metallic nanoparticles that are attached to conductive electrode supports while simultaneously remaining in contact with an ion-conducting electrolyte phase. Conventionally, these electrocatalysts are also exposed to a gaseous or liquid phase that contains the electroactive species of interest, creating so-called triple-phase sites. This “exposed” configuration is generally desirable because it allows for facile transport of reactants and products to and from the electrocatalytically active sites. However, it also presents challenges for electrocatalyst stability because electrochemical and/or physical processes can promote nanoparticle agglomeration and/or dissolution of the electrocatalyst.4

In recent years, alternative electrocatalyst designs have been explored in which the active electrocatalyst is entirely encapsulated by ultrathin oxide overlayers that are permeable to the electroactive species of interest.5–10 Encapsulation is an attractive approach to improve both the stability and catalytic properties of the active electrocatalyst compared to its exposed form. From a stability standpoint, overlayers can serve as a nanoscale adhesive that mitigates particle migration and loss of electrochemically active surface area (ECSA).7,8 Additionally, oxide overlayers can serve as selective barriers, or membranes, that can further promote electrocatalyst stability by blocking impurity species from poisoning the catalyst surface.10,11 Due to this membrane functionality, encapsulated electrocatalysts can be referred to as membrane coated electrocatalysts (MCECs). By leveraging selective transport properties, the overlayers of MCECs can alter the selectivity of competing electrochemical reactions by altering the concentrations of reactants at the buried interface between the active catalyst and overlayer.9,12,13

Encapsulation of electrocatalysts by thin overlayers can also alter reaction energetics and pathways by mechanisms other than selective transport.12 These non-transport mechanisms often rely on the fact that the buried interface between the metallic phase and overlayer represents a confined environment where reactive intermediates are interacting simultaneously with the metal and overlayer. When active sites are located in confined environments, the steric, chemical, and/or electronic properties of reactive intermediates can be expected to be significantly altered relative to the same species located at the interface between a conventional electrocatalyst and the bulk electrolyte phase.14–16 These confinement effects, combined with selective transport properties of the overlayer, make MCECs a highly tunable electrocatalyst architecture that may be engineered to achieve better selectivity and activity than conventional electrocatalysts. However, the large number of control knobs in MCECs, combined with inherent difficulties of characterizing buried interfaces, makes it extremely difficult to uncover the structure–property relationships that govern the performance of MCECs.

In order to better understand reaction energetics at buried interfaces and distinguish between different reaction mechanisms occurring there, it is essential that more detailed knowledge about the physical, chemical, and electronic properties of the buried interface be developed. If structure–property relationships of the buried interface can be clearly established and linked to electrocatalytic behavior, they can serve as design rules for optimizing the performance of MCECs. Towards this end, the current paper explores model MCECs based on ultrathin (1–10 nm thick) silicon oxide (SiOx) overlayers deposited on smooth Pt thin films (Fig. 1). The thickness of SiOx overlayers is controlled with nanoscale precision using a low temperature photochemical deposition process,11 and the planar nature of these samples makes them well-suited for characterizing the physical and electrochemical properties of the SiOx|Pt buried interface.


image file: c8ta06969g-f1.tif
Fig. 1 Schematics of membrane coated electrocatalysts (MCECs) comprised of silicon oxide (SiOx) overlayers deposited onto (a) “thick” and (c) “thin” electrocatalytic Pt thin films supported by smooth p+Si substrates. Representative AFM images are provided for 10 nm thick SiOx overlayers deposited onto (b) a thick 50 nm/4.5 nm Pt/Ti bilayer electrode and (d) a thin 2/3 nm thick Pt/Ti bilayer electrode. The values in the scale-bar for the AFM image correspond to 2 the deviation of the local z-height from the average z-height.

The current study expands on two recent publications by our group that investigated the electrocatalytic properties of SiOx|Pt electrodes based on two different Pt thin film substrates. In the first study,11 it was shown that SiOx overlayers deposited on ultrathin bilayers of 3 nm Pt on 2 nm titanium (Ti) can behave as membranes that support the diffusive flux of protons (H+) and hydrogen (H2) molecules while blocking transport of Cu2+, a poison for the hydrogen evolution reaction (HER). More recently, we reported that SiOx|Pt electrodes based on 50 nm thick Pt substrates exhibit substantially enhanced activity for the oxidation of carbon monoxide (CO) and methanol.17 Herein, we present a side-by-side analysis of SiOx|Pt electrodes based thick (50 nm) and thin (3 nm) Pt substrates under identical test conditions, revealing significant differences in their physical and electrochemical properties. Of particular interest are differences in the SiOx|Pt buried interface, as probed by X-ray photoelectron spectroscopy (XPS) and cyclic voltammetry (CV). While XPS was used to monitor differences in the composition and structure of the Pt film at the buried interface, CV was employed to characterize the electrochemical properties of these electrodes in 0.5 M H2SO4 over narrow (non-oxidizing) and wide (oxidizing) ranges of applied potential. After reporting on differences in the structural and electrochemical properties of the buried interface that result from variations in the SiOx and Pt layer thicknesses, we explore how these differences affect (i) SiOx permeability to electroactive species and (ii) stability of the SiOx overlayer during one hour of CV cycling. Overall, this study highlights the tunability of the MCEC architecture and reveals structure–property relationships that will be key to establishing a molecular-level understanding of electrocatalysis at buried interfaces.

II. Experimental methods

2.1 Electrode fabrication

Monocrystalline p+Si(100) degenerately doped wafers (Prime-grade p+Si, resistivity < 0.005 Ω cm, 500–550 μm thick, WRS materials) were used as flat, conductive substrates that were unresponsive to light. Titanium and platinum at 99.99% purity were sequentially deposited by electron-beam deposition under high vacuum (<6 × 10−8 Torr) without substrate heating to varying thicknesses at a rate of 0.5 and 1 Å s−1, respectively. Thicknesses were monitored using a quartz crystal thickness monitor. Wafers were cleaved into 1.4 cm × 2.5 cm pieces and washed sequentially in acetone, methanol, isopropanol, and deionized water under sonication. A solution of trimethylsiloxy terminated polydimethyl-siloxane (PDMS) in toluene was spin coated onto the Pt|Ti|p+Si substrates for 2.5 minutes at 4000 rpm. The SiOx films having target thicknesses of 2.0, 5.0 and 10.0 nm were fabricated from PDMS/toluene solutions with concentrations of 1.2, 3.0 and 5.3 mg mL−1, respectively. The PDMS-coated electrodes were dried in a vacuum oven at 80 °C for 1 hour before undergoing treatment in a UV ozone chamber for 2 hours (UVOCS, T10X10/OES) to convert the PDMS to SiOx. An electrical contact was made by soldering a copper wire onto the back side of the silicon substrate using indium solder with a soldering iron temperature set to 218 °C. The geometric area of the electrode was defined using 3M electroplater's tape, resulting in a circular 0.246 cm2 opening on the front of the electrode through which the electrocatalytic surface was exposed to the electrolyte.

2.2 Materials characterization

SiOx overlayer thicknesses were measured using a J. A. Woollam alpha-SE ellipsometer. The thicknesses of SiOx overlayers in the SiOx|Pt|Ti|p+Si samples were determined by sequentially fitting the raw data with an optical model that used optical constants and substrate thicknesses determined from fitting bare control samples lacking SiOx. When analyzing SiOx|Pt electrodes based on 50 nm thick Pt films, the Ti layer was not included in the model. Surface topography and roughness were measured with a Bruker Dimension Icon Atomic Force Microscope (AFM) in air using a ScanAsyst silicon tip on a nitride lever silicon cantilever with a 25 nm tip radius. Measurements were performed in peak force nanomechanical mapping mode using a scan rate of 0.977 Hz, a spring force constant of 0.4 N m−1, and a resonant frequency of 70 kHz.

XPS measurements were made with a Phi XPS system at pressures <2 × 10−10 Torr using a monochromatic Al Kα source (15 kV, 20 mA) and a charge neutralizer. All samples are tilted to 57.4 degrees relative to the X-ray source. Multiplex spectra were fit using XPSPEAK software using Shirley's algorithm for background subtraction. All peaks were calibrated such that the Pt 4f7/2 peaks were centered at 71.2 eV.18 The thicknesses of SiOx overlayers based on XPS measurements were calculated using an overlayer model, which is described in Section V of the ESI. Atomic ratios were calculated using each element's atomic sensitivity factor (ASF).19 ASFs used in overlayer calculations for the metallic elements in oxide compounds were computed by multiplying the ASF of the metallic element by the ratio of the atomic density of that element in the oxide to its density in a pure elemental state.

2.3 Electrochemical measurements

Most electrochemical measurements were performed in deaerated 0.5 M sulfuric acid prepared from concentrated sulfuric acid (H2SO4, Certified ACS plus, Fischer Scientific) and 18 MΩ deionized water (Millipore, Milli-Q Direct 8). All electrochemical measurements were conducted with a SP-200 BioLogic potentiostat and carried out in a standard three-neck round bottom glass cell with a commercial Ag|AgCl (3 M KCl) reference electrode (E° = 0.210 V vs. NHE, Hach, E21M002) and a Pt-mesh (Alfa Aesar, 99.9%) counter electrode. The electrolyte was deaerated by purging with nitrogen gas (N2) for 20 minutes before experiments, while the headspace was continuously purged with N2 for the duration of all experiments. Current densities were normalized with respect to the geometric area of the exposed electrode.

Narrow window CV measurements were performed for 228 cycles (1 h) between 0.06 V and 0.82 V vs. RHE, and wide window CV measurements were performed for 151 cycles (1 h) between 0.06 V and 1.22 V vs. RHE, respectively. Upon completion of CV cycling experiments, electrodes to be analyzed by AFM and/or XPS were removed from the glass cell, gently rinsed with 18 MΩ deionized water to remove residual electrolyte for ≈5 s, and dried under a stream of compressed N2 gas. Copper (Cu) stripping voltammetry was carried out with sequential linear sweep voltammograms (LSVs) in solutions of 0.1 M H2SO4 (background) and 0.1 M H2SO4 with 2 mM CuSO4 (Sigma-Aldrich, ReagentPlus Grade). For both LSV measurements, the electrode potential was held at +0.358 V vs. RHE for 50 seconds before sweeping the potential from +0.358 V to +1.0 V RHE at a scan rate of 100 mV s−1. The difference in the integrated charge between the two curves was used to calculate the electrochemically active surface area (ECSA) of the electrode using a conversion factor of 420 μC cm−2 for polycrystalline Pt.20

III. Results and discussion

3.1 Overview of SiOx|Pt electrodes investigated in this study

A series of SiOx|Pt electrodes of different overlayer thicknesses were synthesized from a PDMS precursor using a low temperature photochemical conversion process that has been described in detail elsewhere.21,22 Briefly, a PDMS/toluene mixture was spin-coated onto the Pt thin film substrate, followed by evaporation of the solvent and subsequent conversion of PDMS to a continuous SiOx overlayer within a UV-ozone cleaning system. During this process, reactive oxygen species generated within the system convert methyl groups in the PDMS into gaseous CO2 and H2O byproducts, leaving behind amorphous SiOx that is primarily comprised of SiO2. By adjusting the concentration of PDMS within the solvent mixture, precise control of the SiOx thickness can be obtained. In this study, SiOx|Pt electrodes with SiOx thicknesses between ≈1 nm and 10 nm were made on two different Pt substrates: (i) a “thick”, or bulk-like Pt film having a thickness of 50 nm (Fig. 1a) and (ii) a “thin” Pt thin having a thickness of 3 nm that was sequentially deposited onto a 2 nm thick Ti adhesion layer (Fig. 1c). Within a set of samples, electrodes with three different SiOx thicknesses (tSiOx) were typically fabricated with target thicknesses of tSiOx ≈ 2 nm, 5 nm, and 10 nm. Throughout this paper, the reported values of tSiOx are those measured by ellipsometry, unless stated otherwise.

3.2 Characterization of SiOx-encapsulated “thick” Pt films

Fig. 1b shows a representative AFM image of an as-synthesized 4.8 nm thick SiOx overlayer (tSiOx = 4.8 nm) deposited onto a 50 nm thick Pt substrate. The surface of the SiOx electrode is very smooth, having a root mean squared (rms) roughness of 0.48 nm that represents a slight decrease from a value of 1.00 nm measured for the bare 50 nm Pt substrate lacking any SiOx overlayer. AFM images show that no large cracks or holes in the SiOx overlayers are detected by AFM, although smaller pinholes (<20 nm diameter) are observed for the SiOx|(thick Pt) electrode with the thinnest SiOx overlayer (tSiOx = 1.0 nm) (ESI Fig. S1).

The thicknesses of SiOx overlayers were determined by ellipsometry and are reported in Table 1. Ellipsometry measurements show that the SiOx overlayer thicknesses are extremely uniform across a given sample, with variations in tSiOx across a large 3.5 cm2 sample typically being less than 5% from the average thickness. As-made samples were also characterized by XPS to determine chemical and physical properties of buried interfaces located within a few nanometers of a sample surface. Multiplex scans of the Pt 4f, Si 2p, O 1s, C 1s, and Ti 2p regions were taken for as-made SiOx|Pt and bare Pt control samples. The Si 2p and C 1s spectra of SiOx|(thick Pt) electrodes are provided in the Fig. S2, where it can be seen that the Si 2p peak centers for SiOx|Pt samples correspond to Si in the +4 oxidation state (SiO2) and small amounts of carbon (10–16 atomic%, excluding Pt signal) are present in/on all samples. Both findings are consistent with our previous studies of SiOx produced by the UV ozone process.7,14 The Si 2p and Pt 4f signals were also used to estimate the thicknesses of SiOx overlayers from standard XPS overlayer models, which gave values of tSiOx for the thinnest SiOx overlayers within 0.2 nm of the values measured by ellipsometry.

Table 1 Thicknesses of SiOx and PtOx layers on as-made electrodes. SiOx thicknesses were determined by ellipsometry, while PtOx thicknesses were calculated based on the charge associated with electrochemical PtOx reduction, as described in the text. (*) Indicates bare Pt control samples that were treated in the UV ozone chamber without any PDMS
Substrate SiOx thickness/nm PtOx thickness/nm
Thin platinum (2 nm Ti|3 nm Pt) 0 0.00
0* 0.03
1.58 0.30
4.76 0.21
7.88 0.08
Thick platinum (1 nm Ti|50 nm Pt) 0 0.00
0* 0.66
1.4 0.36
4.6 0.33
10.3 0.23


Pt 4f spectra also provided useful information about the nature of Pt atoms at the SiOx|Pt buried interface. Fig. 2a shows the Pt 4f spectra for a 1.4 nm SiOx|(thick Pt) sample and two bare 50 nm thick Pt control samples. The first control was an as-made, unmodified 50 nm thick Pt film (“Bare Pt”) that gives a standard Pt 4f spectra consistent with metallic Pt0. The second control was a 50 nm thick bare Pt thin film that was treated for 2 hours in the UV ozone chamber (“UV ozone Pt”). Shoulders at higher binding energies are evident in the Pt 4f spectra for the UV ozone Pt control and 1.4 nm SiOx|Pt samples, indicating the presence of Pt bound to oxygen, which is generically referred to as PtOx throughout this paper. In order to isolate the Pt 4f signal associated with PtOx species from metallic Pt0, the Pt 4f7/2 peak intensities of the UV ozone Pt control and 1.4 nm SiOx|(thick Pt) electrode were normalized to that of the untreated bare Pt control, after which the spectra of the latter was subtracted from that of the former to obtain Pt 4f difference spectra (Fig. 2b). The PtOx peaks in the Pt 4f difference spectra for the UV ozone treated bare Pt sample are shifted to higher binding energy (BE) compared to the SiOx|(thick Pt) sample, indicating that the PtOx species on the surface of that sample possess, on average, a higher oxidation state than those at the buried interface of the SiOx|Pt sample. Both difference spectra contain multiple pairs of Pt 4f peaks, which were assigned to three different PtOx species based on previously reported Pt 4f peak centers for adsorbed oxygen on Pt (PtOad), Pt monoxide (PtO), and Pt dioxide (PtO2).23 Analysis of the fitted PtOx Pt 4f peak areas for these species indicates that the PtOx layer of the UV ozone-treated Pt control contains 14% PtOad, 42% PtO, and 44% PtO2, while the PtOx interlayer at the buried interface of the 1.4 nm SiOx|(thick Pt) sample contains 42% PtOad and 58% PtO.


image file: c8ta06969g-f2.tif
Fig. 2 (a) XPS Pt 4f spectra of two unencapsulated Pt electrodes (“Bare Pt” and “UV ozone Pt”) and a SiOx|(thick Pt) electrode with tSiOx = 1.4 nm. The spectra were normalized to the maximum Pt 4f7/2 peak intensities for all samples. (b) Pt 4f difference spectra (grey curves) obtained by subtracting the spectrum of the untreated bare Pt electrode from those of the PtOx-containing electrodes. The black dashed lines represent the fitted background and overall fitted spectra.

The presence of PtOx species at the buried interface of as-synthesized SiOx|(thick Pt) electrodes was also verified by electrochemically reducing them during the first scan segment of cyclic voltammetry (CV) measurements in 0.5 M H2SO4. Immediately after immersing an electrode in the electrolyte, CV scans were initiated at a potential of 0.82 V vs. RHE, which is just below the onset potential for underpotential deposition (upd) of oxygen species. As the potential is scanned in the negative direction, broad peaks associated reduction of PtOx species are observed. The first CV scan for the untreated Pt control sample doesn't exhibit significant reduction current in the range of 0.3–0.8 V vs. RHE, confirming that there are negligible amounts of PtOx on that sample before initiating CV cycling. In contrast, the first CV cycles for the UV ozone treated bare Pt electrode and all of the SiOx-encapsulated Pt electrodes possess distinct reduction peaks between ≈0.3 V and 0.7 V vs. RHE. Fig. 3b shows that the PtOx reduction peaks are completely absent in the subsequent CV cycles, confirming that the PtOx species being reduced in the first cycle were not formed electrochemically but were present on the as-made samples as a result of the UV-ozone process.


image file: c8ta06969g-f3.tif
Fig. 3 (a) First segment of the first CV scan for tSiOx = 0, 1.4, 4.6, and 10.3 nm SiOx|(thick Pt) electrodes. Steady state CV curves measured over (b) narrow and (c) wide potential windows for the same electrodes used in (a). All CVs were conducted at 100 mV s−1 in deaerated 0.5 M H2SO4.

The background-corrected charge associated with PtOx reduction during the first CV cycle, Qr, was used to estimate the thicknesses of PtOx interlayers from Faraday's law. Assuming an average PtOx density of 14.1 g cm−3 (corresponding to that of PtO)24 and an average Pt oxidation state of +2, PtOx thicknesses in the range of 2.2–6.6 Å were obtained (Table 1). Taking the lattice constant of the PtOx layer to be 3.1 Å (similar to that of α-PtO2 and PtO25), the PtOx thicknesses determined from CV measurements correspond to ≈0.7–2.1 atomic layers of PtOx. The slight decrease in PtOx thickness for the 10.3 nm SiOx overlayer may result from a shielding effect whereby the thicker SiOx overlayer suppresses oxidation of the underlying Pt during the UV ozone conversion process. Additionally, it should be noted that the CV scans in Fig. 3a possess significant differences in the PtOx reduction onset potentials and peak centers, which arise from differences in the kinetics and/or energetics of reducing various PtOx species back to the metallic Pt0 state.26 Comparing the PtOx reduction peak centers for the series of SiOx|(thick Pt) electrodes, a shift to more negative potentials is observed with increasing SiOx overlayer thickness. This shift in peak potential may reflect differences in the PtOx species that are present at the buried interface, but could also be convoluted with differences in mass transport constraints imposed by the presence of the SiOx overlayers.

Fig. 3b shows steady state CV curves obtained for the same SiOx|Pt electrodes upon extended CV cycling between 0.06 V to +0.82 V vs. RHE. Over this relatively narrow potential range, none of the samples exhibit any features associated with the oxidation of Pt or formation of upd oxygen. The major features in these CV curves are associated with the underpotential deposition and desorption of hydrogen atoms, which are located between ≈0.0–0.3 V RHE. The integrated hydrogen upd (Hupd) signal is a measure of the electrochemically active surface area (ECSA) of Pt electrocatalysts, and is seen to be greatly suppressed for the bare Pt control sample compared to the SiOx-encapsulated samples. The lower-than expected Hupd signal and lack of sharp Hupd peaks for the bare Pt control are observations that are common for Pt electrodes that are not flame annealed or scanned to potentials sufficient to oxidize Pt surface atoms and adventitious impurity species that can suppress H adsorption.26 In contrast, the 1.4 nm and 4.6 nm SiOx-encapsulated samples exhibit sharp Hupd features and integrated Hupd signals that are 100–130% larger than that of the bare Pt control sample. The Hupd peaks for the tSiOx = 10.3 nm SiOx|Pt sample are less pronounced and slightly skewed, which may be attributed to slow proton diffusion through the thicker SiOx overlayer.11

When changes in the Hupd features are tracked as a function of cycle number (Fig. S5–S8), it is seen that the CV signals for the bare Pt and tSiOx = 1.4 nm electrodes slowly shrink with increasing cycle number, while the Hupd signal actually increases for tSiOx = 4.6 nm and 10.3 nm SiOx|Pt electrodes. The differences in Hupd behavior may be explained by the membrane-like properties of the SiOx layers. If trace amounts of carbon-containing contaminants are present in the electrolyte and cannot be oxidized over the narrow potential CV window, they can poison exposed Pt surface atoms, thereby decreasing Hupd signal. We hypothesize that the SiOx overlayer functions as a diffusion barrier that impedes transport of contaminants to the buried interface, thereby enabling sharp Hupd features that do not decrease with cycling.

CV cycling in deaerated 0.5 M H2SO4 was also performed on an identical set of SiOx|(thick Pt) electrodes over a wider potential window that extended to +1.22 V vs. RHE (Fig. 3c). Unlike the narrow window CV measurements of Fig. 3b, all samples with tSiOx < 10 nm produce CV curves in the wider window that possess sharp Hupd features that are characteristic of polycrystalline Pt electrodes in sulfuric acid.26,27 The Hupd peak locations for the 1.4 nm SiOx|(thick Pt) electrode remain consistent with those observed in Fig. 3b, while the Hupd peaks for the 10.3 nm SiOx|(thick Pt) electrode remain skewed due to slow proton diffusion through the SiOx overlayer. Interestingly, some of the Hupd peak locations for the tSiOx = 4.6 nm SiOx|(thick Pt) electrode are very different from those observed for the tSiOx = 1.4 nm SiOx|(thick Pt) and bare Pt control electrodes (see ESI Fig. S9a for a zoomed in view of Hupd features). The most notable difference is the presence of prominent Hupd peaks in the CV curves of the tSiOx = 4.6 nm SiOx|(thick Pt) electrode that are centered around +0.26 V RHE. This Hupd peak location has been commonly associated with Hupd on Pt(100) terraces, while the Hupd peaks centered around +0.12 V RHE are predominantly associated with Hupd on Pt(110) and Pt(111) orientated surfaces.27,28 This observation implies that the average Pt crystal orientation and/or crystalline defect densities at the SiOx|Pt buried interface depend on the thickness of the SiOx overlayer. It should be noted that the electrodes used for the CV measurements in Fig. 3 were all made from a common Pt substrate, meaning that the differences in Hupd features are unlikely to be caused by differences in the initial crystal structures of as-made substrates.

At more positive potentials, the wide window CV scans in Fig. 3c possess additional features associated with the oxidative adsorption of upd oxygen and formation of Pt oxides during the positive scan segment, as well as the subsequent reduction of PtOx species during the negative scan segment. The PtOx features of the bare Pt and 1.4 nm SiOx|(thick Pt) electrodes are very similar, but the PtOx oxidation and reduction peaks become much more pronounced with increasing thickness of the SiOx overlayer. The increase in the total PtOx signal for the thicker SiOx overlayers is accompanied by a ≈100 mV shift in the onset of PtOx formation to more negative potentials. A similar phenomenon has been reported for polycrystalline Pt electrodes encapsulated by ultrathin layers of atomic layer deposited (ALD) SiO2, although no explanation was provided.29 This negative shift in the PtOx formation potential may be related to restructuring of the Pt surface, which could produce a higher percentage of crystal facets and/or defects that are more susceptible to oxidation. CV investigations of Pt single crystals in 0.5 M H2SO4 have shown that Pt(100) electrodes exhibit a ≈50–100 mV shift in the PtOx onset potential with respect to Pt(110) and Pt(111) electrodes.27 This is consistent with our observation that the thicker SiOx|Pt electrodes exhibiting more pronounced Pt(100) Hupd features simultaneously show a negative shift in the onset potential of PtOx formation. However, these observations may also be related to confinement effects at the buried interface, where disruption of the electrochemical double layer30 and/or interactions with the SiOx overlayer17 could cause the Pt at the buried interface to become more oxophilic (i.e. more easily oxidized). For example, it is well known that the formation of adsorbed OH and O species on Pt can be suppressed due to competitive adsorption of anions.31 In this study, it is possible that SiOx overlayers may affect Pt oxidation by making it more difficult for anions to reach the buried interface, thereby allowing the formation of PtOx species at more negative potentials.

3.3 Characterization of SiOx-encapsulated “thin” Pt electrodes

The results presented up until this point have focused on the effects of varying the thickness of the SiOx overlayer on 50 nm “thick” Pt substrates, but it is also possible to tune the properties of MCECs by altering the composition of the metal layer. To demonstrate this, additional SiOx|Pt electrodes were prepared by depositing SiOx overlayers onto “thin” Pt substrates consisting of sequentially deposited 3 nm of Pt on 2 nm of titanium (Ti).

Analysis of AFM images (Fig. 1d and ESI Fig. S1a) of the bare thin Pt substrates shows that they are even smoother than the thick Pt substrates, having a typical rms roughness of 0.26 nm. A direct relationship between film roughness and thickness, as seen here, is commonly observed for various thin film growth methods.32,33 The difference in rms roughness at the surface of thin and thick Pt samples decreases after coating the Pt substrates with equivalent thicknesses of SiOx, but the larger difference in roughness of the bare substrates is likely to persist at the buried interface of SiOx|Pt electrodes. Close inspection of the AFM image of the 1.6 nm SiOx|(thin Pt) sample (Fig. S1b) reveals that it possesses numerous nanoscopic holes, <20 nm, similar to those identified for the thinnest SiOx|(thick Pt) electrodes. However, the holes are no longer evident for tSiOx > 5 nm (Fig. S1c).

Besides differing in their surface morphology, thick and thin Pt substrates were also found to vary in their composition due to the absence and presence, respectively, of Ti in close proximity to the front surface of the substrate. The source of the Ti is the adhesion layer deposited between inert p+Si(100) wafer and Pt thin film. One way that Ti may reach the front surface of thin Pt substrates is by alloying between Ti and Pt, which can occur during the e-beam deposition process. Evidence of alloying in e-beam Pt/Ti bilayers has previously been provided by cross-sectional transmission electron microscopy (TEM).34 In the current study, significant Ti 2p signal in XPS measurements of as-deposited and UV ozone-treated bare Pt/Ti substrates (Fig. 4a) give strong evidence of the presence of Ti in the near surface region of these samples. While no Ti 2p signal is observed for the thick Pt substrate, minor peaks associated with metallic Ti (2p3/2 peak center at 454.1 eV) and larger peaks associated with a higher oxidation state (2p3/2 peak center of 457.6 eV) are evident for the thin Pt substrate. Ti 2p signal is also seen for the thinnest SiOx|(thin Pt) electrodes, although its intensity is greatly attenuated by screening from the SiOx overlayer (Fig. S3). The Ti 2p peaks at higher binding energy exhibit multiplex splitting of 5.6 eV, which is very similar to that of TiO2, although the peak center binding energies are ≈1.2 eV lower.18 These observations indicate that most of the Ti present near the surface has TiO2-like character, although its XPS signature is not perfectly consistent with stoichiometric TiO2. Interestingly, UV ozone treatment of the thin Pt substrate in the absence of any PDMS results in a ≈230% increase in the Ti 2p signal intensity and concomitant increase in the atomic ratio of Ti[thin space (1/6-em)]:[thin space (1/6-em)]Pt from 0.19 to 0.91. This result shows that Ti is drawn to the surface of the metal bilayer not only by alloying between the Pt and Ti layers, but also by the highly oxidative UV ozone treatment. Given that (i) the UV ozone treatment time (2 hours) is significantly longer than the e-beam deposition time and (ii) the formation of TiO2 from its elements has a very large thermodynamic driving force, it is likely that the UV ozone treatment contributes the most to enrichment of Ti at the metal surface and SiOx|Pt buried interfaces.


image file: c8ta06969g-f4.tif
Fig. 4 XPS analysis of bare Pt and SiOx|Pt electrodes based on thin and thick Pt films. (a) Ti 2p spectra for bare Pt films before and after UV ozone treatment. (b) Pt 4f spectra for bare thick (50 nm) and thin (3 nm) Pt layers. In (b), the solid black curve is the raw spectra and the solid grey curve is the fitted background signal.

XPS characterization was also used to monitor differences in the PtOx interlayer between thin and thick Pt substrates, revealing substantially less PtOx Pt 4f signal for samples based on the former than the latter (Fig. 4b). The absence of PtOx signal is seen in the raw Pt 4f spectra, as well as the Pt 4f difference spectra (Fig. S4b) for the 2.1 nm SiOx|(thin Pt) electrode. This observation is consistent with previous studies showing that composite electrocatalysts based on the combination of TiOx and Pt group metals (or their oxides) such as Ru35 or Pt36 exhibit enhanced resistance to oxidation and/or dissolution of the Pt group metal. In fact, Ti and TiO2 are crucial ingredients for dimensionally stable anodes (DSA®) that are frequently employed under harsh oxidation conditions in various industrial electrochemical processes.37 In this study, the difference in PtOx interlayer thickness for the thin and thick Pt substrates was confirmed by analyzing the PtOx reduction charge (Qr) in the negative sweep of the first CV cycle in 0.5 M H2SO4 (Fig. 3a and 5a). The amounts of PtOx in the SiOx|(thin Pt) electrodes determined from analysis of the CV curves were found to be 17–65% lower than those present in to SiOx|(thick Pt) electrodes (Table 1). These numbers indicate that sub-monolayer amounts of PtOx, ≈0.25 to 0.97 atomic layers, are present at the buried interface of SiOx|(thin Pt) electrodes. Besides having lower Qr, the PtOx reduction features of the SiOx|(thin Pt) electrodes also differed from those of the SiOx|(thick Pt) samples in that they were broader and showed little variation in the location of the peak center (≈0.5 V vs. RHE) on tSiOx. These differences in PtOx reduction features in Fig. 3a and 5a suggest that the thickness and redox properties of the PtOx interlayers formed on the as-made electrodes depend not only on SiOx thickness, but also the substrate.


image file: c8ta06969g-f5.tif
Fig. 5 (a) First segment of the first CV scan for tSiOx = 0, 1.6, 4.8, and 7.9 nm SiOx|(thin Pt) electrodes based on a 3 nm thick Pt layer. Steady state CV curves measured over (b) narrow and (c) wide potential windows for the same electrodes used in (a). All CVs were conducted at 100 mV s−1 in deaerated 0.5 M H2SO4.

SiOx|(thin Pt) electrodes were also characterized by CV in 0.5 M H2SO4 over narrow (Fig. 5b) and wide (Fig. 5c) potential windows to compare their Hupd and PtOx redox features to those of SiOx|(thick Pt) electrodes. As shown in Fig. 5b, the steady state narrow window CV curves of all SiOx|(thin Pt) electrodes are all very similar to each other, with slightly larger Hupd signal than the bare thin Pt electrode. As with the SiOx|(thick Pt) electrodes, we attribute the larger Hupd features for the SiOx|Pt electrodes to the ability of the SiOx overlayer to block impurities from reaching the buried interface. In comparing Fig. 5b to Fig. 3b, it is seen that Hupd features for the SiOx|(thin Pt) electrodes are much less sharp than those of SiOx|(thick Pt) electrodes, while their integrated Hupd signals are significantly smaller than those for SiOx|(thick Pt) electrodes. The smaller Hupd signal may be partially explained by the lower rms roughness for electrodes based on thin Pt substrates, but could also result from a lower availability of upd sites at the buried interface due to the presence of Ti atoms that have segregated there after the UV-ozone treatment.

CV measurements conducted over the wider CV range (Fig. 5c) show that the integrated Hupd signal of SiOx|(thin Pt) electrodes greatly increases compared to that measured over the narrow potential window, becoming comparable to those calculated for SiOx|(thick Pt) electrodes. The similar Hupd signal intensities in wide window CVs indicates that differences in rms roughness can't be solely responsible for the lower Hupd signal of SiOx|(thin Pt) electrodes over the narrow potential window, suggesting that differences in Hupd coverage and energetics at the buried interface might also be factors. In going from the narrow to wide scan window, the Hupd peaks for SiOx|(thin Pt) electrodes also become much sharper, with peak locations that don't deviate substantially from those of SiOx|(thick Pt) electrodes. Despite similarities in Hupd features over the wide scan range, notable differences in PtOx redox features persist at more positive potentials. In particular, the SiOx|(thin Pt) electrodes show less pronounced PtOx features than the bare thin Pt control (Fig. 5c), while SiOx|(thick Pt) electrodes showed larger PtOx features than the thick Pt control (Fig. 3c). Increasing the SiOx thickness on thin Pt substrates leads to a small negative shift in the onset potential of PtOx formation and concomitant increase in PtOx reduction charge, but these changes are much less pronounced than they were for the SiOx|(thick Pt) electrodes in Fig. 3c. Overall, a comparison of the CVs in Fig. 3c and 5c indicates that the SiOx|(thin Pt) electrodes are less prone to electrochemical formation of PtOx at the buried interface than those made from the thicker 50 nm Pt substrate.

3.4 Substrate effects on transport of electroactive species through SiOx overlayers

Understanding the transport properties of MCEC overlayers is of great importance for several reasons. First and foremost, the permeabilities of reactants and products through the overlayer must be sufficiently high to avoid large concentration gradients and associated concentration overpotentials. These permeabilities, combined with the desired operating current density and the bulk concentrations of the reactant, set the maximum overlayer thickness that can be used before significant concentration overpotential losses are incurred.12 Transport properties of the overlayer are also important because differences in permeabilities of electroactive species can be leveraged to control reaction selectivity or poison resistance.11 Understanding the structure–property relationships that control transport of electroactive species through overlayers is therefore a key aspect of developing design rules needed for tuning the properties of MCECs.

In this study, the transport of electroactive Cu+ and H+ through SiOx overlayers was investigated for SiOx|Pt electrodes based on the thin and thick Pt substrates. The ability of Cu2+ to transport through the SiOx overlayers and reach the SiOx|Pt buried interface was viewed by performing Cu stripping voltammetry in an electrolyte containing 2 mM CuSO4 and 0.1 M H2SO4. In these measurements, electrodes were first held at a potential of 0.358 V vs. RHE for 50 s, during which time Cu2+ ions deposit on accessible Pt atoms by under potential deposition. At this potential, only a single layer of Cu can adsorb, making this technique useful for measuring the ECSA of electrocatalysts. Next, the potential was swept from +0.358 V to +1.0 V vs. RHE, resulting in oxidative stripping of upd Cu. An identical linear sweep voltammetry (LSV) curve was also carried out in the 0.1 M H2SO4, shown in Fig. S10, and the difference between the two LSV curves integrated to determine the total amount of Cuupd stripped from the surface. Fig. 6a (for thin Pt substrates) and Fig. 6b (for thick Pt substrates) contain the Cu stripping difference curves of SiOx|Pt electrodes. Consistent with our recent study on SiOx|(thin Pt) electrodes,11 the Cu stripping curves in Fig. 6a exhibit negligible Cuupd oxidation signal, indicating that Cu2+ is not able to penetrate through SiOx overlayers deposited on the thin 3 nm Pt substrates. However, significant Cu stripping signal is observed for SiOx|(thick Pt) electrodes. A comparison of the integrated Cuupd signal (Fig. 6c) reveals that the SiOx|Pt electrodes based on the thick Pt substrate show a ≈1.5–6 fold increase in Cuupd signal relative to the SiOx|(thin Pt) electrodes. This finding suggests that a greater fraction of sites at the buried interface of SiOx|(thick Pt) electrodes are electrochemically accessible to Cu2+, and may be explained by a higher permeability of Cu2+ through the SiOx deposited on thick Pt. It is also interesting to note that the shape of the Cu stripping signal for the SiOx|(thick Pt) samples deviates greatly from the bare Pt substrates, with most signal shifted to higher potentials, indicating a higher binding energy and different energetics of the Cuupd at SiOx|Pt interfacial sites compared to the bare Pt surface.


image file: c8ta06969g-f6.tif
Fig. 6 Cu stripping voltammetry difference curves for electrodes based on SiOx overlayers deposited on (a) thin (3 nm) Pt substrates and (b) thick (50 nm) Pt substrates. The curves in (a) and (b) are given by the difference between LSV curves measured at 100 mV s−1 in 0.1 M H2SO4 (supporting electrolyte) and 2 mM CuSO4 + 0.1 M H2SO4. (c) Comparison of the ECSA of the SiOx|Pt electrodes based on Cuupd signal, shown as a function of SiOx thickness (tSiOx).

The greatly suppressed ECSAs of SiOx|Pt electrodes based on Cu stripping (Fig. 6) contrasts with the large Hupd features seen in CV curves in 0.5 M H2SO4 (Fig. 3 and 5) that indicated relatively high coverages of Hupd at the SiOx|Pt buried interface sites for both thick and thin Pt substrates. However, the CVs used to measure Hupd features do not require high fluxes of protons. To better compare the rates of proton transport, the hydrogen evolution reaction (HER) properties of SiOx|Pt electrodes were evaluated by LSV measurements from 0.82 V to −0.38 V vs. RHE in dearated 0.5 M H2SO4 after 1 hour of CV cycling between 0.06 V and 0.82 V vs. RHE (Fig. 7). It should be noted that the LSV curves for the thin and thick bare Pt control samples overlap almost perfectly, indicating that the presence of any Ti in the former has negligible effect on the intrinsic HER activity of these SiOx-free samples. Thus, the differences in the LSV curves in Fig. 7 for the SiOx|(thin Pt) and SiOx|(thick Pt) electrodes must arise from differences in the SiOx overlayers and/or their buried interface with the underlying Pt.


image file: c8ta06969g-f7.tif
Fig. 7 LSV curves in dearated 0.5 M H2SO4 for SiOx|Pt electrodes based on (a) thin (3 nm) Pt substrates and (b) thick (50 nm) Pt substrates. LSV curves were carried out at 20 mV s−1 immediately after 1 hour of CV cycling in the same electrolyte between 0.06 and 0.82 V vs. RHE. No iR correction has been performed on this data. (c) A comparison of the overpotentials for the hydrogen evolution reaction (ηHER) at a current density of −20 mA cm−2 for all samples as a function of SiOx thickness (tSiOx).

Consistent with our previous study,11 the LSV curves of the SiOx|(thin Pt) electrodes exhibit a strong dependence on tSiOx, with large overpotentials evident for tSiOx = 4.8 nm and tSiOx = 7.9 nm. However, the SiOx|Pt electrodes based on the thick Pt substrates exhibit almost no difference in the LSV curves for SiOx thicknesses of 5 nm or less (Fig. 7b). A side-by-side comparison of the overpotentials at −20 mA cm−2 is provided in Fig. 7c, showing that larger HER overpotentials are required for SiOx|Pt electrodes made from the thin bilayer Pt/Ti substrates. Although it is possible that changing SiOx overlayer thickness could lead to slight changes in the HER reaction energetics at the buried interface, the monotonic decrease in current density with increasing SiOx thickness is most likely a result of concentration overpotential losses that develop across the SiOx overlayers.11 By extension, we interpret the differences in the LSV curves for the SiOx|(thin Pt) electrodes (Fig. 7a) and SiOx|(thick Pt) electrodes (Fig. 7b) to be a result of differences in the permeabilities of protons diffusing through the SiOx overlayers. The weaker thickness-dependence of the LSV curves of the SiOx|(thick Pt) electrodes can be attributed to a higher permeability for protons, consistent with the higher permeability for Cu2+ ions that was seen in Fig. 6b.

In evaluating the combined results of the Cu stripping voltammetry (Fig. 6) and H2 evolution (Fig. 7) measurements, we find that transport of both H+ and Cu2+ through SiOx overlayers is much more suppressed for SiOx overlayers deposited on thin Pt substrates than those deposited on thick Pt substrates. Although the exact origins of these “substrate effects” on the electrochemical properties of SiOx|Pt electrodes are not obvious, the stark differences in behavior of the SiOx(thin Pt) and SiOx(thick Pt) electrodes highlight the wide range of properties that are accessible with the MCEC architecture.

3.5 Stability of SiOx|Pt electrodes during CV cycling in sulfuric acid

In order for the overlayer of an MCEC to protect the encapsulated electrocatalyst, it is imperative that the overlayer itself also be stable under the operating conditions of interest. Takenaka et al. reported that SiO2-encapsulated Pt nanoparticles used for the oxygen reduction reaction (ORR) showed minimal loss in ECSA and no noticeable changes in the silica layers after CV cycling for 20[thin space (1/6-em)]000 cycles in 0.1 M HClO4 between 0.0 V and 1.2 V vs. RHE.7 Recent studies from our lab showed that UV ozone SiOx-encapsulated Pt nanoparticle8 and 3 nm Pt thin film11 electrodes exhibit stable performance as HER electrocatalysts in 0.5 M H2SO4. However, our more recent study investigating SiOx|(thick Pt) planar electrodes for alcohol oxidation indicated partial delamination of the SiOx overlayer during extended CV cycling in 0.5 M methanol + 0.5 M H2SO4.17 None of the aforementioned studies provided detailed physical characterization of the SiOx overlayers after extended operation. To better investigate the stability of SiOx coatings under CV cycling as a part of the current study, AFM, ellipsometry, and XPS were used to monitor changes in the properties of tSiOx ≈ 1 nm and tSiOx ≈ 5 nm SiOx|Pt electrodes resulting from CV cycling in 0.5 M H2SO4 over both narrow (0.06–0.82 V RHE) and wide (0.06–1.22 V RHE) potential scan ranges.

Representative AFM images of SiOx|Pt electrodes in their initial state and after CV cycling are provided for two different SiOx thicknesses for both thin (Fig. 8a) and thick (Fig. 8b) Pt substrates. For reference, AFM images of as-made bare Pt substrates are also provided. Looking first at Fig. 8a, AFM is not able to detect any significant differences in the morphology of the SiOx|(thin Pt) electrodes after CV measurements, with the exception of subtle coarsening of the tSiOx = 4.8 nm SiOx|(thin Pt) electrode after CV cycling over the wide potential window. AFM images in Fig. 8b show that the tSiOx = 1.0 nm SiOx|(thick Pt) electrodes also maintain their original morphology after CV cycling over both short and wide windows. In contrast, the surfaces of the tSiOx = 4.8 nm SiOx|(thick Pt) electrodes clearly changed after CV cycling in both the narrow and wide potential ranges, exhibiting morphologies that closely resemble the morphology of the bare thick Pt substrate. Rms roughness values computed for each sample (Fig. 8c) support these qualitative descriptions, showing that negligible changes in surface roughness were observed for all samples except the tSiOx = 4.8 nm SiOx|(thick Pt) electrodes. After both narrow and wide window CV cycling, the rms roughness of those electrodes nearly doubled. The exact rms roughness values obtained from the post CV AFM images of the tSiOx = 4.8 nm SiOx|(thick Pt) electrodes is very similar to that of the bare Pt substrate, indicating that most, but not all of the SiOx has been removed. Ellipsometry also indicated a substantial reduction in tSiOx occurred for the 4.8 nm SiOx|(thick Pt) electrode after CV cycling (Fig. S16).


image file: c8ta06969g-f8.tif
Fig. 8 AFM images of SiOx|Pt electrodes before and after 1 hour of CV cycling in deaerated 0.5 M H2SO4 over narrow (0.06–0.82 V vs. RHE) and wide (0.06–1.22 V vs. RHE) potential windows. Representative AFM images are shown for (a) SiOx|(thin Pt) electrodes and (b) SiOx|(thick Pt) electrodes with two different SiOx thicknesses. (c) Rms roughness values, taken as the average value obtained from 3 different images for each sample.

To further probe changes in the chemical and/or electronic properties of the SiOx overlayer and SiOx|Pt buried interface caused by CV cycling, XPS was used to characterize the same tSiOx = 1.0 nm SiOx|(thick Pt) and tSiOx = 1.6 nm SiOx|(thin Pt) electrodes. Changes in the Si 2p, O 1s, and C 1s spectra relative to as-made electrodes are seen for both electrodes, but are most noticeable for the SiOx|(thick Pt) electrode. As seen in Fig. S11, notable positive shifts are seen in the Si 2p (0.40–0.82 eV), O 1s (0.67–0.89 eV), and C 1s spectra (0.42–0.76 eV), with the largest shifts seen for the SiOx|(thick Pt) electrode that underwent CV cycling over the narrow potential window. The positive shifts in the Si 2p spectra could be explained by conversion of SiOx to a highly hydroxylated layer containing many silanol groups,38 which would not be surprising given that the permeable SiOx overlayers were operated in an acidic electrolyte with pH below the pKa values of SiOH groups.39 Also of interest in Fig. S11 is the O 1s spectra of the SiOx|(thick Pt) electrode taken after narrow window CVs, for which the O 1s peak center shifts by +0.8 eV to a binding energy of 532.4 eV. O 1s binding energies greater than those of the parent oxide have commonly been associated with hydroxylated oxides.40–42 Thus, the O 1s spectra in Fig. S11b may provide evidence that the SiOx overlayers deposited on the thick Pt substrates are themselves transformed to a highly hydroxylated and/or hydrated form during CV cycling. Volume expansion compared to the as-made sample would be expected to accompany such a transformation to a hydrated phase. Supporting this possibility, ellipsometry measurements performed on the 1.0 nm SiOx|(thick Pt) electrode after CV cycling indicate that tSiOx increased to 4.5 nm. This increase in overlayer thickness may indicate swelling of the overlayer, perhaps due to hydration, but could also result from chemical and/or structural transformations that alter the optical properties of the SiOx, and therefore the apparent thickness of the overlayer. In contrast, much smaller shifts are observed for the Si 2p (0.01–0.30 eV) and O 1s (−0.10–0.23 eV) spectra of the tSiOx = 1.6 nm SiOx|(thin Pt) electrodes, with post-CV ellipsometry measurements, indicating that little-to-no change in SiOx thickness had occurred from its as-synthesized state. However, one notable change in the SiOx|(thin Pt) electrodes was seen for the electrode that underwent CV cycling over the wide potential window; after CV cycling, the Ti 2p signal almost completely disappears (Fig. S12e), meaning that most of the TiOx near the buried interface had leached from the buried interface.

Si 2p and Pt 4f spectra were also used to monitor changes in the Si[thin space (1/6-em)]:[thin space (1/6-em)]Pt atomic ratio resulting from CV cycling (Fig. S15), which can be a useful measure of changes in the SiOx overlayer thickness. While the Si[thin space (1/6-em)]:[thin space (1/6-em)]Pt atomic ratio of the 1.0 nm SiOx|(thick Pt) electrodes decreased from ≈5.4 to 3.3 after both CV experiments, a moderate increase from Si[thin space (1/6-em)]:[thin space (1/6-em)]Pt ≈ 3.3 to 4.6 was calculated for the 1.6 SiOx|(thin Pt) electrode after the narrow window CV scan. However, the atomic Si[thin space (1/6-em)]:[thin space (1/6-em)]Pt ratio was found to decrease from ≈3.3 to 2.2 for the SiOx|(thin Pt) electrode after the wide window CV scan. XPS measurements performed on the 4.8 nm SiOx|(thick Pt) (Fig. S13) and 4.8 nm SiOx|(thin Pt) (Fig. S14) electrodes before and after CV cycling similarly reveal much better stability of SiOx overlayers deposited on the thin Pt substrates. As evidenced by the nearly complete absence of Si 2p signal (Fig. S13c) and dramatic reduction in the Si[thin space (1/6-em)]:[thin space (1/6-em)]Pt atomic ratio (Fig. S15), XPS measurements confirmed that CV cycling over both potential ranges almost completely removed the thicker 4.8 nm SiOx overlayers from the thick Pt substrates. On the other hand, the thicker 4.8 nm SiOx|(thin Pt) electrodes still maintain similar Si[thin space (1/6-em)]:[thin space (1/6-em)]Pt atomic ratios (Fig. S15) after CV cycling, consistent with AFM and ellipsometry measurements that had indicated the SiOx overlayers remained well-adhered to the thin Pt substrates. It is difficult to quantitatively assess changes in the SiOx thickness based on the Si[thin space (1/6-em)]:[thin space (1/6-em)]Pt atomic ratios due to possible film swelling and/or differences in screening from adventitious species. Nonetheless, the absence of extremely large changes in these ratios or in the intensities of the raw Si 2p spectra confirm that the electrodes based on the thinnest SiOx overlayers remain primarily intact, albeit with varying degrees of change to the SiOx composition.

To see whether or not chemical dissolution contributed to the decrease in SiOx overlayer thickness for the ≈5 nm SiOx|(thick Pt) electrode, the thickness of a freshly made 5.3 nm SiOx|(thick Pt) sample was measured by ellipsometry before and after soaking in 0.5 M H2SO4. After 20 hours of soaking, the measured SiOx thickness decreased by less than 0.2 nm (≈3% of total film thickness). This result indicates that chemical dissolution should not influence SiOx stability over the timescale employed for the electrochemical CV measurements (≈1 hour). Given that considerable losses in overlayer thickness were observed for this same sample type after 1 hour of CV cycling, our findings suggest that the primary mechanism of SiOx overlayer loss on the thick Pt substrates is delamination rather than dissolution. AFM images, ellipsometry, and XPS measurements collectively show that the durability of SiOx|Pt electrodes depends strongly on both the SiOx overlayer thickness and the choice of substrate.

3.6 Effects of PtOx and Ti on the SiOx|Pt buried interface

Several differences between the “thick” and “thin” Pt substrates may alter the structure–property relationships of the SiOx overlayer and SiOx|Pt buried interface of SiOx|Pt electrocatalysts. These include differences in the substrate roughness, substrate crystallinity, thickness of the PtOx interlayer, and the presence of Ti-species (e.g. TiOx, PtxTiy) near the buried interface. The amount of PtOx and presence of Ti appear to be especially important for explaining many key differences in electrochemical properties of SiOx|(thick Pt) and SiOx|(thin Pt) electrodes. XPS and CV measurements showed that the thickness of the PtOx is significantly reduced for SiOx|(thin Pt) electrodes, for which Ti originating from the adhesion layer may diffuse to the near-surface region and serve as an oxygen scavenger that suppresses PtOx formation. Combining the influence of Ti with the shielding effect of the SiOx overlayer during the UV ozone process, the SiOx|Pt|Ti electrodes presented in this study offer two control knobs for adjusting the thickness of the PtOx interlayer. The relationships between SiOx thickness, substrate type (thin or thick Pt), and the amount of the PtOx interlayer (using Qr as a proxy) are shown in Fig. 9a. This figure shows that the amount of PtOx is most sensitive to the choice of substrate, or rather, the presence of Ti. This relationship between the presence of Ti and PtOx thickness is illustrated schematically in Fig. 9, which shows a molecular-level view of as-made SiOx|(thin Pt) (Fig. 9b) and SiOx|(thick Pt) (Fig. 9c) electrodes based on knowledge gained from XPS and CV measurements.
image file: c8ta06969g-f9.tif
Fig. 9 (a) Relationship between the charge associated with reduction of PtOx species during the first CV cycle, Qr, and the thickness of the SiOx overlayer, tSiOx, as measured by ellipsometry. Qr values for the SiOx|(thick Pt) and SiOx|(thin Pt) electrodes were determined from analysis of the PtOx reduction signal shown in Fig. 3a and 5a, respectively. Schematic side-views of as-made SiOx|Pt electrodes made from depositing SiOx on (b) a thin bilayer Pt substrate and (c) a thick (bulk) Pt substrate. In both schematics, the dark blue and light blue spheres represent Pt atoms in metallic (Pt0) and oxidized (PtOx) states, respectively.

Fig. 9 highlights structural differences at the SiOx|Pt buried interface of as-made electrodes, but it is not trivial to relate these structural characteristics to the electrochemical properties of SiOx|Pt electrodes. PtOx interlayers formed during the UV ozone process are reduced during the first scan segment of the first CV scan, meaning that the structure(s) of the buried interface during and after operation are very likely to deviate from that of the as-made electrode. Nonetheless, we hypothesize that the initial thickness and composition of PtOx interlayers can greatly influence the physical and chemical characteristics of the buried interface during operation. Reduction of a relatively thick PtOx layer to form a more compact Pt layer can be expected to break bonds between PtOx and the SiOx overlayer, leading to a loosely connected buried interface. Accordingly, reduction of thicker PtOx interlayers is more likely to alter Pt crystal structure (e.g. reconstruction, defects, roughness) and/or the steric environment associated with nano-confined spaces at the buried interface compared to the initial state. On the other hand, an as-made SiOx|Pt electrode that possesses negligible PtOx is more likely to maintain many of the original bonding interactions between the substrate and overlayer, minimizing structural changes at the buried interface.

In this study, differences in the electrochemical properties of SiOx|Pt electrodes were seen in the form of shifts in Hupd peaks, large variations in Cuupd stripping features, shifts in PtOx formation onset potentials, alterations in HER LSV curves, and variability in overlayer stability after CV cycling. Many of the finer details of the relationships between these observed properties and structural characteristics of SiOx|Pt electrodes remain ambiguous, but some details have emerged. In general, our measurements indicate that SiOx overlayers deposited on thin Pt substrates exhibit better stability than SiOx overlayers deposited on thick Pt substrates, and we primarily attribute this to differences in initial PtOx interlayers that may affect adhesion between the overlayer and substrate. The detrimental role of the PtOx interlayer was verified with CV scanning over wide windows for the SiOx|(thin Pt) electrodes, which can promote electrochemical formation of PtOx. The presence/absence of PtOx at the buried interface of as-made electrodes also correlates strongly with electrochemical characteristics such as the location and magnitude of Hupd and Cuupd features. These dependencies are attributed to differences in the Pt crystal structure and/or energetics of adsorbates located at the buried interface. Needless to say, a more detailed molecular-level understanding of the processing–structure–property relationships would be invaluable for tuning the catalytic properties of MCECs.

While this study has presented several new insights into the structure–property relationships of SiOx|Pt electrocatalysts, many open questions remain. The presence of Ti species appears to indirectly affect the buried interface by altering the thickness of PtOx interlayers, but it remains to be seen whether TiOx may also directly impact structural and electrocatalytic properties of the buried interface. For example, it is unclear whether Ti species at the buried interface bonds with SiOx, thereby creating robust “anchor points” between the overlayer and substrate that can promote adhesion and enhance confinement effects at the buried interface. A related question pertains to the influence of the metal oxide species (PtOx and TiOx) on the structure–property relationships of the SiOx overlayer. This study showed that the morphological, chemical/electronic, and electrochemical characteristics of the SiOx overlayer were strongly affected by use of the thin or thick Pt substrate. In particular, stark differences in the Cuupd stripping voltammetry and H2 evolution performance were observed, indicating that the permeabilities of the SiOx overlayers for Cu2+ and H+ were greatly altered by changing the substrate. An explanation for the differences in permeabilities may be found in XPS characterization of SiOx films after CV cycling, which suggested that the SiOx overlayer deposited on the thick Pt substrate becomes significantly hydroxylated, while that deposited on the thin Pt substrate is better able to retain its initial, less hydroxylated state. Previous studies have indicated that proton transport through silica is facilitated by a “hydrogen hopping”, or Grotthuss-type mechanism involving transfer of protons between silanol groups in silica.43,44 If true, then one would expect to see substantially higher proton transport rates for the more heavily hydroxylated SiOx|(thick Pt) electrodes; based on the HER measurements in Fig. 7, this is exactly what was seen. Additional supporting evidence of differences in the overlayer composition was obtained from contact angle (θc) measurements of as-made ≈5 nm SiOx|(thin Pt) and SiOx|(thick Pt) samples (Fig. S17). These measurements showed that the SiOx|(thick Pt) samples (θc = 64.8 ± 1.1°) are significantly more hydrophilic than the SiOx|(thin Pt) samples (θc = 79.9 ± 1.7°), suggesting that there are differences in the chemical and/or physical properties of as-made overlayers that result from being deposited on different substrates. The smaller contact angle for the SiOx|(thick Pt) sample is consistent with the SiOx coating being more hydrated and/or hydroxylated than its SiOx|(thin Pt) counterpart. However, a detailed mechanistic understanding of how the substrate impacts SiOx structural and/or compositional characteristics that dictate its propensity to become hydroxylated, and therefore its ability to support cation transport, is lacking. With the aim of answering this and other fundamental questions about the structure–property relationships of oxide-encapsulated electrocatalysts, we expect that the use of in situ spectroscopies, aided by ab initio simulation tools, will be invaluable.42

IV. Conclusions

The structure–property relationships governing the performance of MCECs based on oxide-encapsulated metals are significantly different, and in many ways, more complex than those that govern the performance of conventional “exposed” electrocatalysts. In order to uncover some of these relationships, this study has examined the physical, chemical, and electrochemical properties of model SiOx-encapsulated Pt thin film electrodes, with a particular emphasis placed on characterizing the SiOx|Pt buried interface. Through a combination of physical, spectroscopic, and electroanalytical characterization techniques, it was found that Pt oxide (PtOx) interlayers are located at the buried interface of as-made samples, and that the thickness and composition of these interlayers can be tuned by adjustments in the thicknesses of the SiOx overlayer and a Pt/Ti bilayer substrate. Importantly, the presence and amount of PtOx interlayers in SiOx|Pt electrodes, whether present initially or formed during electrochemical measurements, are found to correlate with electrochemical properties such as upd adsorption/desorption features, HER performance, and stability of SiOx overlayers. The large variations in electrochemical properties observed over the parameter space investigated in this study highlights the large degree of tunability afforded by the encapsulated electrocatalysts. Although many fundamental questions about the operation of encapsulated electrocatalysts remain, the structure–property relationships reported here represent a step forward in the development of design rules for this tunable class of emerging electrocatalysts. As the complex design rules governing the performance of oxide-encapsulated electrocatalysts are further refined through additional experimental and computational investigations, we expect that this emerging electrocatalyst architecture will be extended to many other materials and electrochemical reactions.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

We acknowledge Columbia University Columbia Nano Initiative clean room facility for the use of the physical vapor deposition and the Shared Materials Characterization Laboratory for the use of AFM, ellipsometer and XPS. We would also like to acknowledge Amar Bhardwaj and Dr Ngai Yin Yip for providing assistance and facilities for contact angle measurements. This material is based upon work supported by the National Science Foundation under Grant Number (CBET-1752340). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

References

  1. J. Greeley and N. M. Markovic, Energy Environ. Sci., 2012, 5, 9246–9256 RSC.
  2. S. Gu, B. Xu and Y. Yan, Annu. Rev. Chem. Biomol. Eng., 2014, 5, 429–454 CrossRef CAS PubMed.
  3. Z. W. Seh, J. Kibsgaard, C. F. Dickens, I. Chorkendorff, J. K. Nørskov and T. F. Jaramillo, Science, 2017, 355, 1–12 CrossRef PubMed.
  4. P. J. Ferreira, G. J. la O', Y. Shao-Horn, D. Morgan, R. Makharia, S. Kocha and H. A. Gasteiger, J. Electrochem. Soc., 2005, 152, A2256 CrossRef.
  5. K. Maeda, K. Teramura, D. Lu, N. Saito, Y. Inoue and K. Domen, Angew. Chem., Int. Ed., 2006, 45, 7806–7809 CrossRef CAS PubMed.
  6. H. Matsumori, S. Takenaka, H. Matsune and M. Kishida, Appl. Catal., A, 2010, 373, 176–185 CrossRef CAS.
  7. S. Takenaka, H. Miyamoto, Y. Utsunomiya, H. Matsune and M. Kishida, J. Phys. Chem. C, 2014, 118, 774–783 CrossRef CAS.
  8. N. Y. Labrador, X. Li, Y. Liu, H. Tan, R. Wang, J. T. Koberstein, T. P. Moffat and D. V. Esposito, Nano Lett., 2016, 16, 6452–6459 CrossRef CAS PubMed.
  9. B. Endrődi, N. Simic, M. Wildlock and A. Cornell, Electrochim. Acta, 2017, 234, 108–122 CrossRef.
  10. M. Qureshi, T. Shinagawa, N. Tsiapis and K. Takanabe, ACS Sustainable Chem. Eng., 2017, 5, 8079–8088 CrossRef CAS.
  11. N. Y. Labrador, E. L. Songcuan, C. De Silva, H. Chen, S. J. Kurdziel, R. K. Ramachandran, C. Detavernier and D. V. Esposito, ACS Catal., 2018, 8, 1767–1778 CrossRef CAS.
  12. D. V. Esposito, ACS Catal., 2018, 8, 457–465 CrossRef CAS.
  13. A. T. Garcia-Esparza, T. Shinagawa, S. Ould-Chikh, M. Qureshi, X. Peng, N. Wei, D. H. Anjum, A. Clo, T. C. Weng, D. Nordlund, D. Sokaras, J. Kubota, K. Domen and K. Takanabe, Angew. Chem., Int. Ed., 2017, 56, 5780–5784 CrossRef CAS PubMed.
  14. A. D. Doyle, J. H. Montoya and A. Vojvodic, ChemCatChem, 2015, 7, 738–742 CrossRef CAS.
  15. Y. Fu, A. V. Rudnev, G. K. H. Wiberg and M. Arenz, Angew. Chem., Int. Ed., 2017, 56, 12883–12887 CrossRef CAS PubMed.
  16. H. Li, J. Xiao, Q. Fu and X. Bao, Proc. Natl. Acad. Sci. U. S. A., 2017, 114, 5930–5934 CrossRef CAS PubMed.
  17. J. E. Robinson, N. Y. Labrador, H. Chen, B. E. Sartor and D. V. Esposito, under Rev.
  18. J. F. Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben, Handbook of X-ray Photoelectron Spectroscopy, Physical Electronics, Inc., Eden Prairie, MN, 1995 Search PubMed.
  19. C. D. Wagner, L. E. Davis, M. V. Zeller, J. A. Taylor, R. H. Raymond and L. H. Gale, Surf. Interface Anal., 1981, 3, 211–225 CrossRef CAS.
  20. K. Varga, P. Zelenay and A. Wieckowski, J. Electroanal. Chem., 1992, 330, 453–467 CrossRef CAS.
  21. C. Mirley and J. Koberstein, Langmuir, 1995, 11, 0–3 Search PubMed.
  22. M. Ouyang, C. Yuan, R. J. Muisener, A. Boulares and J. T. Koberstein, Chem. Mater., 2000, 12, 1591–1596 CrossRef CAS.
  23. K. S. Kim, N. Winograd and R. E. Davis, J. Am. Chem. Soc., 1971, 93, 6296–6297 CrossRef CAS.
  24. CRC Handbook of Chemistry and Physics, ed. D. R. Lide, CRC Press, Boca Raton, FL, internet version, 2005, pp. 4–75 Search PubMed.
  25. J. R. McBride, G. W. Graham, C. R. Peters and W. H. Weber, J. Appl. Phys., 1991, 69, 1596–1604 CrossRef CAS.
  26. V. Climent and J. M. Feliu, J. Solid State Electrochem., 2011, 15, 1297–1315 CrossRef CAS.
  27. K. Yamamoto, D. M. Kolb, R. Kotz and G. Lehmpfuhl, J. Electroanal. Chem., 1979, 96, 233–239 CrossRef.
  28. N. Furuya and S. Koide, Surf. Sci., 1989, 220, 18–28 CrossRef CAS.
  29. G. Yuan, A. Agiral, N. Pellet, W. Kim and H. Frei, Faraday Discuss., 2015, 176, 233–249 RSC.
  30. L. Bellarosa, R. García-Muelas, G. Revilla-López and N. López, ACS Cent. Sci., 2016, 2, 109–116 CrossRef CAS PubMed.
  31. B. E. Conway, Prog. Surf. Sci., 1995, 49, 331–452 CrossRef CAS.
  32. T. Aaltonen, M. Ritala, T. Sajavaara, J. Keinonen and M. Leskela, Chem. Mater., 2003, 15, 1924–1928 CrossRef CAS.
  33. J. Singh and D. E. Wolfe, J. Mater. Sci., 2005, 40, 1–26 CrossRef CAS.
  34. D. V. V Esposito, I. Levin, T. P. P. Moffat and A. A. A. Talin, Nat. Mater., 2013, 12, 562–568 CrossRef PubMed.
  35. L.-Å. Näslund, C. M. Sánchez-Sánchez, Á. S. Ingason, J. Bäckström, E. Herrero, J. Rosen and S. Holmin, J. Phys. Chem. C, 2013, 117, 6126–6135 CrossRef.
  36. E. Ding, K. L. More and T. He, J. Power Sources, 2008, 175, 794–799 CrossRef CAS.
  37. S. Trasatti, Electrochim. Acta, 2000, 45, 2377–2385 CrossRef CAS.
  38. T. L. Barr, Appl. Surf. Sci., 1983, 15, 1–35 CrossRef CAS.
  39. M. Sulpizi, M. P. Gaigeot and M. Sprik, J. Chem. Theory Comput., 2012, 8, 1037–1047 CrossRef CAS PubMed.
  40. S. Yamamoto, H. Bluhm, K. Andersson, G. Ketteler, H. Ogasawara, M. Salmeron and A. Nilsson, J. Phys.: Condens. Matter, 2008, 20, 184025 CrossRef.
  41. S. Benkoula, O. Sublemontier, M. Patanen, C. Nicolas, F. Sirotti, A. Naitabdi, F. Gaie-Levrel, E. Antonsson, D. Aureau, F. X. Ouf, S. I. Wada, A. Etcheberry, K. Ueda and C. Miron, Sci. Rep., 2015, 5, 1–11 Search PubMed.
  42. T. A. Pham, X. Zhang, B. C. Wood, D. Prendergast, S. Ptasinska and T. Ogitsu, J. Phys. Chem. Lett., 2018, 9, 194–203 CrossRef CAS PubMed.
  43. M. Nogami, R. Nagao and C. Wong, J. Phys. Chem. B, 1998, 102, 5772–5775 CrossRef CAS.
  44. J. C. Fogarty, H. M. Aktulga, A. Y. Grama, A. C. T. van Duin and S. A. Pandit, J. Chem. Phys., 2010, 132, 174704 CrossRef PubMed.

Footnotes

Electronic supplementary information (ESI) available. See DOI: 10.1039/c8ta06969g
Both authors contributed equally.

This journal is © The Royal Society of Chemistry 2018