Kinetics of hydrogen adsorption during catalytic reactions on transition metal surfaces

Yujung Dong and Francisco Zaera *
Department of Chemistry and UCR Center for Catalysis, University of California, Riverside, CA 92521, USA. E-mail: zaera@ucr.edu

Received 6th February 2017 , Accepted 18th June 2017

First published on 6th July 2017


A study of the kinetics of the hydrogenation of ethylene promoted by hydrogen was performed by using a high-flux molecular beam. The experiments were designed to probe the intermediate pressure regime (around the mTorr range) associated with the transition between the surface-science research carried out under ultrahigh vacuum (UHV) environments and the catalytic studies performed under atmospheric conditions, the so-called “pressure gap” that is often mentioned but seldom addressed in catalytic work. In addition, the experiments were also focused on the characterization of the kinetics of both hydrogen isotope scrambling and ethylene hydrogenation simultaneously, by using H2 + D2 + C2H4 reaction mixtures and by following the time evolution of all of the products, HD and the different C2X6 isotopologues (X = H or D), in order to estimate the influence of the kinetics of the hydrogen uptake on the overall olefin conversion rates under reaction conditions. It was found that the addition of ethylene to the H2 + D2 beam leads to a significant decrease in the probability for HD production, but that the olefin hydrogenation reaction can still be sustained catalytically under the conditions of our experiments. Three kinetic regimes were identified with increasing partial pressure (or flux) of ethylene in the reaction mixture. The first, seen for mixtures with less than 10 parts per million of ethylene, shows a steady-state production of ethane with kinetics similar to those reported from UHV studies, with a rate law dependent linearly on both ethylene partial pressure and hydrogen atom coverage. The surface is mainly covered with hydrogen, and ethane formation occurs primarily via a previously unidentified “reverse” Eley–Rideal mechanism where olefin molecules from the gas phase pick up two hydrogen (deuterium) atoms upon impingement on the surface. A second, intermediate regime is seen for mixtures with up to about 1% of ethylene. In that case, HD production is still relatively fast, albeit the rate decreases slowly with increasing ethylene pressure, and the catalytic activity is mainly controlled by the coverage of the reversibly adsorbed ethylene, which partially blocks the hydrogen uptake (some di-σ ethylene and ethylidyne, Pt3[triple bond, length as m-dash]C–CH3, a species that forms via ethylene dehydrogenation, are also strongly adsorbed on the surface). The probability for ethane formation decreases noticeably in this regime, and the reaction mechanism switches to the stepwise hydrogen incorporation sequence proposed by Horiuti and Polanyi many years ago. Finally, for reaction mixtures with more than 1% of ethylene, the ethylidyne surface layer reaches coverages close to saturation and controls the HD and ethane formation kinetics via site blocking; this latter regime is the one operational under most typical catalytic runs. It was also shown that the relative importance of the reversibly adsorbed ethylene and irreversibly adsorbed ethylidyne species in the reaction kinetics depends on surface temperature, and that the ethylidyne layer can be removed at temperatures around 500 K to restore the full catalytic activity of the clean Pt surface.


1. Introduction

Hydrogen is one of the most widely used molecules in chemical applications. For instance, there has been much interest recently in using H2 as a clean and sustainable chemical fuel.1–3 At present, hydrogen is primarily produced by4–6 and consumed in catalytic processes, the latter mainly for the production of ammonia7 and the processing of fossil fuels.8 Much of this catalysis, from fuel cells9 to oil reforming,10 is based on transition metals, mostly in the form of nanoparticles dispersed on a high-surface-area oxide support such as silica or alumina. Consequently, extensive research has been carried out on the activated adsorption and surface chemistry of hydrogen on transition metal surfaces.11–13 However, most of that work has been carried out under controlled environments, specifically under ultrahigh vacuum (UHV) conditions, and the behavior observed there may not be representative of the regime operational during catalysis.

This issue is particularly critical when referring to hydrocarbon conversions, because in those cases the catalytic surfaces are believed to be not clean but rather covered by strongly adsorbed carbonaceous deposits made via the decomposition of the reactants.14–19 Moreover, hydrogen adsorption on the remaining bare metal patches competes with that of the other gases in the mixture, and, in the case of the hydrogenation of unsaturated organics at least, that competition is quite unfavorable for H2. Therefore, the kinetics of the activation and uptake of hydrogen in catalytic processes is likely to be significantly slower than what has been measured in surface-science studies.20 Unfortunately, in spite of its central role in defining the overall catalytic kinetics of many industrial processes, there has been only very limited research on this subject.

Here, we report results from our investigation of the kinetics of hydrogen adsorption on platinum surfaces during ethylene hydrogenation catalysis. Our study relied on the use of high-flux molecular beams made out of mixtures of H2 + D2 and ethylene in order to sustain steady-state catalytic conversions under controlled UHV conditions, and on isotope labeling, using the detection of HD production from H2 + D2 mixtures as a proxy for the characterization of the kinetics of hydrogen dissociation, surface diffusion, and recombination under catalytic conditions. It was found that, perhaps unsurprisingly, the uptake of hydrogen on the platinum surface is severely hindered by the addition of olefins to the gas mixture. H–D isotope scrambling still takes place at a measurable rate, however, sufficient to promote olefin hydrogenation at relatively high turnover frequencies, on the order of 1–10 ML s−1 (ML = monolayers, reported in terms of molecules per surface Pt atom).17,21,22 The steady-state rate of HD production is inversely related to the relative pressure of ethylene, and its poisoning by ethylene is partially reversible: some, albeit not all, of the initial activity is regained once the olefin is removed from the gas mixture. By varying the beam total flux and composition, the background pressure, and the surface temperature, we were able to assess the relative contributions of the two effects mentioned above, namely, the poisoning by strongly bonded hydrocarbon fragments (alkylidynes) and the competition between hydrogen and ethylene adsorption, to the overall decrease in HD production activity. It was also found that the reduced hydrogen uptake rate clearly affects the rate of formation as well as the isotopologue distribution of the ethane produced via ethylene hydrogenation.

2. Experimental details

The experiments described here were carried out in an ultrahigh vacuum (UHV) apparatus described in previous publications.23–25 The catalytic reactions were promoted on the front surface of a polished 1 cm-in-diameter, 1 mm-thick, polycrystalline Pt disk, placed at the center of the vacuum chamber by using an on-axis vertical manipulator capable of XYZθ motion and of cooling and/or resistive heating to any temperature between 100 and 1100 K (measured by a chromel–alumel thermocouple spot-welded to the side of the Pt disk and controlled by home-made feedback electronics). The platinum surface was cleaned before each experiment by a combination of argon ion bombardment, annealing, and thermal treatments with O2 to burn any remaining surface carbon contaminants until known temperature programmed desorption (TPD) spectra for CO and H2 were reproduced.

The central element of our instrument is the high-flux gas doser, which is made out of a single capillary 1.2 cm in length and 150 μm in diameter. The reaction mixtures were premixed in a calibrated volume and fed to the doser via a leak valve to control the total flux, which was quantified by following the drop in pressure at the backing volume versus time using an MKS Baratron capacitance manometer. Typical total fluxes for the H2 + D2 beams were set at approximately 2 × 1017 molecules s−1, which corresponds to a flux density of ∼1 × 1025 molecules m−2 s−1 right at the exit of the capillary of the doser. It was also independently determined that the beam that hits the Pt surface, which is placed approximately 1 mm away from the source, is about 2 mm in diameter, which means that the average flux at the surface in these experiments is ∼6.5 × 1022 molecules m−2 s−1, or approximately 4500 ML s−1 (assuming a surface Pt atom density of 1.5 × 1019 atoms m−2); for reference, such flux leads to an impinging frequency on the surface equivalent to a gas pressure of approximately 5 mTorr at room temperature. The H2 + D2 beam composition in all the experiments reported here was equimolar (that is, one to one), and ethylene was added to the reaction mixture in one of two modes: (1) by combining the olefin with the H2 + D2 premixed gas behind the doser or (2) by introducing it directly into the UHV chamber via a side leak valve. The results from both approaches were in general comparable, as discussed in more detail in the Results section. All of the gases were purchased from commercial sources (H2 from Liquid Carbonic, >99.995% purity; D2 from Matheson, 99.5% atom purity; C2H4 from Matheson, 99.5%), and used as supplied.

The gas composition in the vacuum chamber during the course of the catalytic reactions was monitored continuously by using a UTI 100C quadrupole mass spectrometer placed out of the line of sight of the doser and the solid sample to avoid angular distortions due to direct scattering. This mass spectrometer was interfaced to a personal computer to afford the recording of the time evolution of the signal intensities of up to 15 ion masses simultaneously. Typically, raw mass spectrometry data were acquired in each kinetic run in the 2–4 and 26–37 amu ranges, and deconvoluted to obtain values for the partial pressures of all ethylene and ethane isotopologues by using a matrix-based analysis reported in previous publications.26 The mass spectrometer signals were calibrated against pressure readings from a nude ion gauge placed in the UHV chamber, and the cracking patterns of the individual ethylene and ethane isotopologues, which are needed for the data deconvolution, were either measured with our instrument or extracted from the literature.27–29

3. Results

For reference, data were first obtained for the HD production rate (expressed in terms of the HD partial pressure measured with the mass spectrometer) during the conversion of H2 + D2 mixtures on the clean Pt surface as a function of both beam flux and surface temperature. The results are summarized in Fig. 1. The left panel shows how the conversion rate is linear with beam flux for all the temperatures tested, a result easy to explain in terms of a kinetic model that assumes Langmuir-type adsorption in the limit of low coverages. It is interesting to note that Bernasek and Somorjai reported, in experiments combining a D2 beam with H2 from the background, first-order kinetics in D2 and half-order in H2.30 It is likely that in their experiments the coverage of normal hydrogen was much higher than that attained with their low-flux D2 molecular beam; the latter result with deuterium is the one that matches ours in terms of the first-order dependence of the rate of hydrogen/deuterium flux.
image file: c7cy00216e-f1.tif
Fig. 1 HD yields for the conversion of H2 + D2 mixed beams, expressed in terms of the HD partial pressure measured by the mass spectrometer in the UHV chamber, as a function of: total beam flux (left panel) and temperature (right). The HD yields are directly proportional to the beam flux, meaning that the reaction probability does not change with flux. The temperature dependence, reported here in the Arrhenius form, shows two regimes: (1) below 500 K, where the HD production displays an apparent activation barrier of 11.1 ± 0.7 kJ mol−1, and (2) above 500 K, where the temperature dependence is much weaker. The lines are the best linear fits to the data.

More interesting is the temperature dependence displayed on the right panel of Fig. 1, which shows two regimes: (1) between 300 and 500 K, where a typical Arrhenius behavior is seen with an apparent activation barrier of approximately 11.1 ± 0.7 kJ mol−1, possibly associated with the heat of hydrogen adsorption, and (2) above 500 K, when the HD production rate plateaus (a linear fit to the data yields a small, ∼1.1 ± 0.3 kJ mol−1, activation energy). The value for the activation barrier in the low-temperature regime is much lower than that reported from surface-science studies under vacuum, but Johansson and coworkers have shown that this is due to the higher steady-state surface coverages attained at higher H2 pressures/fluxes, as is the case here; the activation barrier at 1 Torr H2 pressure, extracted from their reported data, is approximately 9 kJ mol−1.20 Regarding the high-temperature results, it could be argued that the yields are insensitive to changes in temperature, which may be because the system may have reached equilibrium. However, the total HD production estimated from our measurements amounts to only about 30% of the total H2 + D2 mixture, not the expected 50%, which means that other factors may be at play. Clearly, the high fluxes used here modify the kinetics of hydrogen isotope scrambling even on the clean Pt surface. It should also be pointed out that the high probability for HD production reported here is likely the result of conversion on surface defects, since the activity of the Pt(111) plane has been shown to be quite low.12

The rate of HD exchange is reduced significantly upon introduction of ethylene to the gas-phase mixture. Fig. 2 shows the data obtained from a typical kinetic run, performed at a surface temperature of 300 K, to illustrate this point. Shown in that figure is the evolution of the mass spectrometer signals at 3 and 27 amu, corresponding to HD and C2H4, respectively, as a function of time; the signals, reported here in arbitrary units, are proportional to the partial pressures of the corresponding gas-phase species, and, because of the high pumping speed used to attain UHV conditions in our chamber, these are in turn proportional to rates of reaction (to the rate of HD production and of C2H4 dosing, respectively). In this example, a H2 + D2 beam (with a flux of 1 × 1025 molecules s−1 or ∼4500 ML s−1 at the Pt surface) was first turned on and allowed to reach a steady state, at a value of RClean. Ethylene was then introduced into the chamber via a side leak valve, starting at approximately t = 50 s, and its pressure stabilized at ∼3.5 × 10−8 Torr. At that stage, the HD partial pressure decreased until it leveled off at a fraction of the initial value (RSS ∼ 0.44 × RClean), signifying the partial poisoning exerted by the olefin. The dosing of C2H4 was stopped at t = 580 s, at which point the partial pressure of HD increased again but only to a value RAfter ∼ 0.51 × RClean. This indicates that two different effects contribute to the changes seen in the rate of hydrogen H–D scrambling, as mentioned in the introduction: (1) competitive adsorption between H2/D2 and reversibly π-bonded C2H4 under steady-state conditions, as manifested by the difference between RSS and RAfter, and (2) the irreversible poisoning (at low reaction temperatures, around room temperature) of the surface by strongly bonded hydrocarbon fragments, ethylidyne31 plus some di-σ bonded ethylene, seen in the difference between RAfter and RClean.


image file: c7cy00216e-f2.tif
Fig. 2 Data from a typical kinetic run in the form of mass spectrometer signals for HD (3 amu blue line) and ethylene (27 amu, red trace) as a function of time. In this experiment, the H2 + D2 beam was first directed at the clean Pt surface, and the HD steady-state production rate (RClean, which is proportional to the 3 amu signal) allowed to stabilize. Ethylene was added from a side leak valve at t = 50 s, after which the new (reduced) HD production rate was let to stabilize again (at RSS). The ethylene gas was removed at t = 580 s, allowing for the HD production to partially increase to a new RAfter value. These data show how the addition of ethylene poisons some of the HD production activity and how that drop in activity is partially, but not totally, reversible.

The effect of the reversibly-adsorbed (π-bonded) ethylene on the rate of HD production is highlighted in Fig. 3, which shows a different kinetic run where the partial pressure of ethylene was varied over time. As in the case of Fig. 2, the HD partial pressure drops considerably upon the initial introduction of the olefin to the vacuum chamber, suggesting that the surface becomes almost saturated (and poisoned) with hydrocarbon adsorbates immediately upon exposure to ethylene. The HD partial pressure drop is larger than that in Fig. 2 because the reaction was carried out at 400 K. Pointedly, that pressure is also seen to vary somewhat with changes in P(C2H4), as better shown in the amplified (green) trace of P(HD) − P(HD)Baseline (P(HD)Baseline being the value obtained for the baseline P(C2H4) used over the 300–1500 s period of time, which is approximately 6 × 10−8 Torr). Fig. 3 shows that the HD pressure, which is proportional to RSS, decreases with increases in ethylene pressure (which is varied to a maximum of ∼4 × 10−7 Torr in this experiment, reached at 1150 s), that is, with increases in the steady-state coverage of the π-bonded ethylene. The effect is small, representing a minor fraction of the changes seen upon the initial introduction of ethylene to the reaction mixture (and even of the total rate increase seen upon full removal of the olefin from the gas phase), but this can be explained by the fact that the adsorption of π-bonded ethylene exhibits Langmuir-type behavior,32 meaning that a high surface coverage is attained upon exposure of the surface to relatively low C2H4 pressures and changes relatively little upon further increase of the pressure. As indicated above, the full effect of the reversibly adsorbed ethylene on the rate of H–D exchange is indicated by the difference between RSS and RAfter.


image file: c7cy00216e-f3.tif
Fig. 3 Another kinetic run to illustrate the effect of the addition of ethylene to the production of HD from H2 + D2 mixed molecular beams, this time highlighting the variations induced by changes in the partial pressure of ethylene (amplified in green for the 300–1500 s time range).

Results from a more systematic study of the dependence of the poisoning of the HD formation rate on the partial pressure of ethylene are reported in Fig. 4. The left panel shows the HD partial pressures measured under steady-state catalysis versus the pressure of the ethylene added to the background, which was varied in the range from approximately 10−10 to over 10−7 Torr. The right panel, by contrast, displays similar data for experiments where the olefin was added directly to the beam gas mixture, in C2H4/(H2+D2) ratios going from 10−5 to 10−1; both sets of data are provided for reactions carried out at 300 K and at 400 K. Although there are a few minor differences between these two sets of data, the general trends are the same in both types of experiments, with a C2H4/(H2 + D2) ratio in the beam of 10−4 corresponding roughly to the independent leaking of 1 × 10−9 Torr of C2H4via the side leak valve. A large initial drop in HD scrambling activity is observed at 400 K upon the introduction of ethylene in both cases, followed by a slower additional decrease with further increases in ethylene pressure. Note, however, that the latter changes occur over a range of several orders of magnitude in P(C2H4). Also important to highlight is the difference in the extent of initial poisoning of the HD exchange seen between 300 and 400 K; the effect is much more pronounced at 400 K because of the more extensive decomposition of ethylene to ethylidyne in that case. At 300 K, more of the poisoning is exerted, in relative terms, by the π-bonded, reversibly-adsorbed ethylene instead. This is discussed in more detail next.


image file: c7cy00216e-f4.tif
Fig. 4 HD partial pressure (a signal proportional to the rate of HD production) versus the partial pressure of the ethylene added via the side leak valve (left panel) and the fraction of ethylene added to the H2 + D2 beam (right). Data are provided in both panels for two surface temperatures, 300 K (filled blue circles) and 400 K (open red squares). The initial steep drop in HD production activity seen upon addition of even small amounts of ethylene (especially at 400 K) is followed by a more nuanced decrease with ethylene pressures above 10−9 Torr or C2H4/(H2 + D2) ratios of 10−5.

The effects of reaction temperature on both the reversible and the irreversible components of the poisoning by C2H4 addition are summarized in Fig. 5. These experiments were carried out by adding ethylene (approximately 5 × 10−8 Torr) to the reaction mixture via the side leak valve. It can be seen that the main result of increasing the surface temperature is an increase in the relative extent of poisoning of the HD production rate, as mentioned above: the rate under steady state (RSS) drops from approximately 45% of the initial value (RClean) at 300 K to ∼10% at 400 K. It should be indicated that these are relative changes, and that the effect is not as dramatic in absolute terms, because the rate of HD production on clean Pt also goes up with temperature (Fig. 1). In fact, the absolute rate of HD formation does go up with temperature in the presence of gas-phase ethylene (Fig. 4). On the other hand, the changes seen upon turning off the C2H4 dosing are also larger at higher temperatures. Consequently, the changes in the irreversible loss of HD activity due to the deposition of carbonaceous deposits are not as dramatic as it would appear at first sight (although those too increase with increasing reaction temperature, as indicated by the drop in the RAfter/RClean ratio). It would seem that, although more poisoning is seen at higher temperatures in absolute terms, the reversible component of the reaction poisoning increases in relative importance with increasing reaction temperature. There are two possible reasons for this: (1) changes in the steady-state coverage of the π-bonded ethylene on the surface, as we have been discussing up to now, and (2) the removal of a larger fraction of the irreversibly adsorbed species. As discussed next, we think that the effect seen in Fig. 5 is mainly due to the latter.


image file: c7cy00216e-f5.tif
Fig. 5 Ratios of irreversible (RAfter/RClean, red hatched bars) and reversible + irreversible (RSS/RClean, solid blue) HD production rates relative to the activity on the clean Pt as a function of surface temperature. No value is available for RSS/RClean at 450 K, as indicated by the asterisk. Higher temperatures lead to a higher drop in activity but also to an increase in the relative importance of the reversible component.

The di-σ ethylene and ethylidyne species that form upon exposure of the surface to ethylene and poison the hydrogen uptake and HD scrambling are stable around room temperature but can be hydrogenated and removed from the surface at higher temperatures.19,23,33 This is in fact, we believe, the reason why a larger fraction of the lost activity of the catalyst is restored at higher temperatures once the source of ethylene gas is removed (see previous paragraph). We therefore hypothesized that exposing the Pt catalysts to the H2 + D2 beam at high enough temperatures should afford the full restoration of the catalytic activity of the clean surface. The data in Fig. 6 provides supporting evidence for this. In the experiment reported in that figure, the HD exchange was first probed at 400 K, starting with the clean surface. Ethylene was then introduced via the side leak valve, at which point the HD production rate dropped to a value of approximately RSS = 0.08 × RClean. Some of that activity was restored upon removal of the ethylene from the gas mixture, which was carried out at t = 130 s, and eventually the HD steady-state production rate settled at a new value of RAfter = 0.30 × RClean (at t = 600 s). At that point, the temperature of the surface was raised to 500 K while keeping the H2 + D2 beam on in order to facilitate the hydrogenation of the ethylidyne layer. The HD production rate in this time period exceeded the initial value, but this is because of the higher reaction temperature used (Fig. 1). The temperature was finally dropped back to 400 K after about 700 s of hydrogen exposure at 500 K (between ∼700 and 1400 s), and a new steady-state HD production rate was established after about 1600 s in this kinetic run. Interestingly, the new value amounted to approximately 92% of the initial rate on the clean Pt at 400 K. It appears that, indeed, most of the initial activity can be restored upon hydrogenation of the carbonaceous layer at higher temperatures. Perhaps longer hydrogenation times may lead to full recovery of the initial activity, but care should be taken not to use excessive temperatures, because the alkylidyne layer eventually dehydrogenates on the surface and leads to the formation of more strongly bonded species that cannot be as easily removed via a hydrogenation process.19,33


image file: c7cy00216e-f6.tif
Fig. 6 Kinetic run designed to illustrate the possibility of removing the strongly bonded carbonaceous deposits (ethylidyne, Pt3[triple bond, length as m-dash]C–CH3) believed to contribute to the irreversible component of the drop in HD production rate seen in Fig. 2, 3 and 5. First, the H2 + D2 beam was made to impinge on the clean Pt, kept at 400 K, and the HD steady-state production rate (proportional to the 3 amu signal) let to stabilize. Second, ethylene was added via the side leak valve from t = 50 s to t = 130 s, at which point the HD production drops to its new value, RSS. The ethylene leak was stopped at t = 130 s, and the HD partial pressure was allowed to reach a new value of RAfter. After 600 s, the sample was briefly cooled and then heated to 500 K (in the presence of the H2 + D2 beam); a new, high HD partial pressure was reached after ∼1200 s. Finally, after 1400 s, the surface was cooled down back to 400 K. By 1600 s the HD production rate settled at approximately 92% of the initial value, signifying the removal of most of the ethylidyne layer formed earlier during the ethylene exposure.

The gas composition of the reaction mixture affects not only the yield of the HD made from isotope scrambling of the H2 + D2 mixture but also the production of ethane that takes place concurrently. The details of the kinetics of this reaction under the conditions afforded by the high-flux molecular beam have been reported already,23–25,34 but we here provide some new data obtained with the H2 + D2 + C2H4 mixtures to add to our understanding of the reaction mechanism. Fig. 7 displays the data obtained from a typical kinetic run; the left panel shows the time evolution of the traces for 27 to 33 amu, whereas the right panel reports the results from deconvolution of the data in terms of partial pressures for the different C2H6−nDn isotopologues obtained (as well as for the unreacted C2H4). The main products in this case are ethane molecules produced via the direct addition of H or D atoms to the C[double bond, length as m-dash]C double bond, namely, C2H6, C2H5D and C2H4D2, but small amounts of C2H3D3 and C2H2D4 were detected as well, being the result of H–D isotope exchange steps on the adsorbed intermediates. Much of the observed kinetics, specifically the characteristic exponentially decaying distribution of ethane isotopologues with increasing deuterium substitution seen in Fig. 7, can be accounted for by a stepwise and reversible sequence of atomic hydrogen additions, as initially proposed by Horiuti and Polanyi many years ago,17,35 although we have recently shown that a second Eley–Rideal direct addition pathway may also open up under the conditions of the molecular beam experiments.34,36


image file: c7cy00216e-f7.tif
Fig. 7 Typical kinetic data from a catalytic run with a H2 + D2 + C2H4 mixed beam. In this case the total flux was about 4500 ML s−1, the beam composition H2[thin space (1/6-em)]:[thin space (1/6-em)]D2[thin space (1/6-em)]:[thin space (1/6-em)]C2H4 = 500[thin space (1/6-em)]:[thin space (1/6-em)]500[thin space (1/6-em)]:[thin space (1/6-em)]1, and the surface temperature 400 K. Shown are the raw signals from the mass spectrometer in the 27–32 amu range (left panel) and the corresponding partial pressures of ethylene and the ethane isotopologue products, calculated via deconvolution of the raw data (right), all versus reaction time. Ethanes with various multiple deuterium substitutions were detected in these experiments.

The product distributions obtained in these hydrogenation reactions using H2 + D2 mixtures are reported as a function of the amount of ethylene added in Fig. 8. The data correspond to experiments where the ethylene was leaked independently (left panel) as well as from runs using H2 + D2 + C2H4 mixtures (right). Both types of experiments yielded similar results, highlighting two opposing trends: (1) an initial increase of relative C2H4D2 production at the expense of C2H6, up to C2H4 pressures of about 1 × 10−9 Torr or C2H4/(H2 + D2) ratios in the beam of 1 × 10−4, and (2) a reversal of that trend and the dominance of C2H6 production (with a drop in C2H5D yield as well) in the ethylene-richer regime. It is worth indicating that the product distribution for P(C2H4) = 5 × 10−9 or C2H4/(H2 + D2) = 5 × 10−4 is close to the statistical expectation (C2H6[thin space (1/6-em)]:[thin space (1/6-em)]C2H5D[thin space (1/6-em)]:[thin space (1/6-em)]C2H4D2 = 1[thin space (1/6-em)]:[thin space (1/6-em)]2[thin space (1/6-em)]:[thin space (1/6-em)]1) from assuming direct H(ads) or D(ads) addition to the C[double bond, length as m-dash]C double bond in ethylene via the Eley–Rideal mechanism mentioned above.25,34 With higher ethylene pressures, the Horiuti–Polanyi pathway may become more important, in which case multiple H–D exchange steps can take place before ethane desorption; a normal kinetic isotope effect may then explain the favorable incorporation of regular H, leading to the predominance of C2H6 production, as seen in Fig. 8. The results in the low ethylene-pressure range are more difficult to explain but may have been influenced by the fact that the absolute ethylene exposures in this regime are quite low, leading to ethane yields on the order of a few monolayers at most; the hydrogenation steps may be affected by the addition of normal H atoms to the surface during the decomposition of the initial adsorbed ethylene to ethylidyne (H2C[double bond, length as m-dash]CH2 + Pt(s) → Pt3[triple bond, length as m-dash]C–CH3 + H(ads)).


image file: c7cy00216e-f8.tif
Fig. 8 Isotopologue ethane product distributions obtained from kinetic runs such as that shown in Fig. 7 as a function of ethylene partial pressure from the side leak valve (left panel), and of the C2H4/(H2 + D2) ratio in the molecular beam (right). Both sets of experiments show similar trends, with a maximum in C2H4D2 production at intermediate ethylene pressures and dominance of C2H6 formation with ethylene-rich mixtures.

Finally, the overall hydrogenation yield, obtained by adding the yields of all the ethane isotopologues obtained, is reported as a function of the amount of ethylene in the beam in Fig. 9. The general trend is a decrease in reaction probability with increasing olefin pressure, from values close to 100% with hydrogen-rich beams (C2H4/(H2 + D2) < 0.01%), to less than 0.1% for a beam with ∼5% of ethylene. This general behavior attests again to the poisoning effect of reversibly adsorbed ethylene toward the uptake of hydrogen. In fact, the trend observed here can be explained quantitatively by assuming that the rate of ethylene hydrogenation is proportional to the relative coverage of hydrogen (and/or deuterium) atoms on the surface, θX. The solid lines in Fig. 9 correspond to a fit of the data to such a model, assuming a competitive Langmuir model for the adsorption of hydrogen and the olefin, as indicated by the formula provided in the inset. The numerical fit yields a ratio of adsorption constants highly favorable toward ethylene image file: c7cy00216e-t1.tif, as has been discussed several times already in this report.


image file: c7cy00216e-f9.tif
Fig. 9 Ethane (C2X6) total yield as a fraction of the partial pressure of ethylene in the H2 + D2 + C2H4 mixed beams versus ethylene fraction in the beam (top scale, filled blue circles), and atomic hydrogen surface coverage θX (bottom scale, open red squares). The surface coverage of hydrogen (deuterium) was calculated by using a Langmuir competitive adsorption model, given by the formula provided in the inset (where KH2 and KC2H4 are the adsorption equilibrium constants for hydrogen and ethylene, respectively, and FH2 and FC2H4 the corresponding gas fluxes). The ethane production probability shows a linear dependence on hydrogen coverage (solid lines) until mixtures with approximately 1% ethylene are reached, at which point the hydrogen coverage reaches values below 1% and the reaction rate for ethane formation drops abruptly. We ascribe the latter observation to the formation of a nearly saturated layer of ethylidyne on the surface.

It should be mentioned that the model does break down at high ethylene pressures, failing to describe the large drop in reaction probabilities seen experimentally for beam compositions with more than 1% of ethylene (leading to hydrogen surface coverages below 1%). We believe that the reason for such abrupt reduction in catalytic activity is the qualitative change that takes place on the surface due to an increase in poisoning via the formation of ethylidyne, which may reach surface coverages close to saturation and dominate the kinetics of ethylene hydrogenation and HD exchange;37 this justifies the extremely low conversion probabilities (p ≤ 1 × 10−6) associated with typical catalytic conditions.18,23,24 The model used here, where the ethane formation rates depend linearly on hydrogen coverages, applies only to the hydrogen-rich mixtures, where reversible ethylene adsorption is the main process suppressing the uptake and isotope scrambling of hydrogen, and where a significant fraction of the di-σ ethylene and ethylidyne species that form on the surface can be removed by the hydrogen beam (note that the reactions in this case were carried out at 400 K, a temperature where a significant fraction of the catalytic activity can be restored upon removal of the ethylene reactant from the gas phase; see Fig. 5). Our argument is somewhat subtler, however, because some poisoning by strongly bonded carbonaceous deposits also occurs in the low-pressure regime, as indicated by the data in Fig. 2–5. It is important to point out that the kinetics of H–D formation has been shown to exhibit strongly non-linear kinetic effects as a function of ethylidyne coverage, and to display a sharp transition at a given threshold value for the coverage of the ethylidyne layer.37 It may very well be that the transition in activity seen here with ethylene-rich beams, as well as the breakdown of our model, is associated with that same threshold in ethylidyne coverage reported in the previous study.37

4. Discussion

Much research has been directed at developing a better molecular-level understanding of catalytic reactions, yet many issues still remain unresolved. In the case of the conversion of organic molecules, in reactions such as the hydrogenation and isomerization of unsaturated bonds with hydrogen, one recurring theme is that the surfaces of the transition-metal nanoparticles usually used for their promotion do not remain clean but rather become covered with a layer of strongly bonded hydrocarbons, the result of the initial decomposition of the reactants.18,19,38–48 This fact makes the chemistry associated with catalysis potentially quite different to that characterized extensively under controlled UHV conditions. The expected disparity between these two regimes has been long recognized and dubbed the “pressure gap”15,49–52 but has not been properly addressed to date; many studies have been performed either under UHV or under the atmospheric pressures typical in catalytic processes, but almost no work has been performed in the intermediate pressure regime, between approximately 10−6 and 1 Torr, where the transition is expected to occur.

We have over the last few years focused on addressing this specific issue. We have used the hydrogenation of ethylene promoted by platinum surfaces as a prototypical reaction to represent most olefin hydrogenations and isomerizations, and more generically, the hydrogenation of unsaturated organics. We have developed a high-flux molecular beam in order to be able to emulate steady-state catalytic conditions in the mTorr pressure range inside a UHV chamber, which affords great control over the cleaning, preparation, and characterization of the catalytic surface.24–26 The use of this system has been combined with more conventional studies using a so-called high pressure cell,49,53,54 where catalytic reactions are carried out in a batch reactor with a platinum crystal that can be transferred to a UHV chamber for sample preparation and for post-mortem analysis without exposure to the outside environment; mass spectrometry has been used in our set-up to monitor the kinetics of the reactions continuously during the catalytic runs while infrared absorption spectroscopy has been employed to characterize in situ the adsorbed species present on the surface during reaction.22,23,26 Our combined work using both the high-flux molecular beam and the high-pressure cell has allowed us to obtain new insights into the role of strongly bonded carbonaceous deposits in catalytic hydrogenations.19,22–25,32,34,37 The data reported here add to that knowledge by providing key information on how the kinetics of adsorption and surface reactivity of hydrogen are affected by both the carbonaceous deposits and the pressure and composition of the gas mixture.

The key results from the present work relate to the kinetics of isotope scrambling in H2 + D2 mixtures on platinum surfaces as they become affected by the addition of ethylene to the reaction mixture. This reaction was used here as an indirect way to probe the adsorption, surface diffusion, and recombination of hydrogen, as has been done multiple times in the past both under vacuum12,55,56 and for catalytic studies.17,38,57 We have shown that the incorporation of even small amounts of ethylene into the reaction mixture reduces the rate of HD formation by an order of magnitude or more (Fig. 2 to 5). This is so even though the experiments were designed to emulate relatively low total pressures (compared to the conditions typically used in catalysis), on the order of a few mTorr, and relied on high hydrogen-to-ethylene ratios; such conditions are needed to attain high reaction probabilities and to sustain steady-state conversion in hydrogenation catalysis under vacuum.24 Three regimes were identified as a function of the relative partial pressure of ethylene compared to the flux of the H2 + D2 beam (Fig. 4, 8, and 9): (1) P(C2H4) < 10−9 Torr or C2H4/(H2 + D2) < 10−5, where most of the drop in HD production probability occurs; (2) 10−6 > P(C2H4) > 10−9 Torr or 10−2 > C2H4/(H2 + D2) > 10−5, after which the rate decreases further, but at a much slower pace, over a change in reaction mixture composition of several orders of magnitude; and (3) P(C2H4) > 10−6 Torr or C2H4/(H2 + D2) > 10−2, at which point the coverage of the irreversibly adsorbed ethylidyne layer reaches values close to saturation, and its poisoning of surface sites dominate the catalytic behavior.

Regarding the first of the pressure regimes listed above, it was shown that some, but not all, of the initial HD conversion seen on the clean Pt surface can be reinstated upon the removal of ethylene from the reaction mixture (Fig. 2, 3 and 5). In fact, the steady-state rate of HD production depends to some extent on the steady-state pressure of ethylene (Fig. 3). Because of the reversibility of this effect, we associate such behavior to a competition for adsorption sites between hydrogen (or deuterium) and ethylene. It would seem that reversibly adsorbed ethylene, most likely a π-bonded species,16,32,58 blocks adsorption sites for hydrogen during reactions, and that this happens even with small fractions of ethylene in the reaction mixture. Some strongly bonded carbonaceous deposits (di-σ ethylene and ethylidyne adsorbed species) are expected to be present on the Pt surface under these conditions but to not be the main contributing factor that defines the catalytic activity. It is interesting to note that the reaction probability for ethylene hydrogenation is still near unity under such conditions, that is, for C2/(H2 + D2) ratios below 10−4, where the hydrogen coverages are still above 60% of a monolayer (Fig. 9), which means that the rate of ethane formation is roughly proportional to the pressure of ethylene. It is clear that in this low-ethylene-pressure regime ethylene poisons the adsorption of hydrogen to some extent but olefin hydrogenation proceeds with high probabilities. The catalytic active sites in this case are covered mainly with a mixture of hydrogen (deuterium) and π-bonded ethylene, and the olefin hydrogenation rate law is approximately first order in ethylene pressure and in atomic hydrogen surface coverage, as seen in UHV experiments.32 Also, the hydrogenation of ethylene appears to be dominated by a direct reverse Eley–Rideal mechanism by which ethylene molecules from the gas phase may react directly with hydrogen (deuterium) atoms adsorbed on the surface to produce ethane;34 the isotopologue distribution is close to what would be expected from such a mechanism (C2H6[thin space (1/6-em)]:[thin space (1/6-em)]C2H5D[thin space (1/6-em)]:[thin space (1/6-em)]C2H4D2 = 1[thin space (1/6-em)]:[thin space (1/6-em)]2[thin space (1/6-em)]:[thin space (1/6-em)]1, Fig. 8). The Eley–Rideal mechanism reported here is unique to this pressure regime, and has not, to the best of our knowledge, been reported before.

A transition is then seen at ethylene molar fractions above approximately 10−4 (P(C2H4) > 10−9 Torr). At this stage, the rate of HD production continues to decrease with ethylene addition but at a much slower pace (Fig. 4). The rate of ethane formation, on the other hand, decreases more abruptly (Fig. 9). In this regime, the calculated coverage of hydrogen (deuterium) atoms on the Pt surface is relatively low, going from θX ∼ 0.6 ML for C2/(H2 + D2) = 10−4 to θX ∼ 0.01 ML for C2/(H2 + D2) = 10−2. Whereas the rate of HD production is reduced by approximately a factor of 2 in that range (Fig. 4), the probability for ethane formation drops by an order of magnitude (Fig. 9). Of course, the absolute rate of ethane production goes up (because it is given by the product of the reaction probability times the flux or pressure of ethylene, and the latter increases by two orders of magnitude), but it is clear that the kinetics of both hydrogen uptake and olefin hydrogenation change in this transition. This is also evidenced by the change in ethane product distribution (Fig. 8); the main product becomes C2H6, presumably because of a kinetic isotope effect favoring H (vs. D) incorporation sequentially into the olefin first and into the resulting ethyl surface intermediate afterwards, according to the Horiuti–Polanyi stepwise mechanism.17,35 We believe that in this regime the kinetics of the surface reactions may still be controlled by the coverages of π-ethylene and atomic hydrogen, but that the latter is now much lower than the former and that more ethylidyne is present on the surface. This is still not the regime operational during typical catalysis but rather a transitional state only seen in the so-called “pressure gap”; as far as we know, no kinetics for this regime have been reported before.

The fraction of the surface irreversibly poisoned by the strongly bonded ethylidyne fragments under steady-state reaction conditions can be estimated by comparing the rates of HD production before versus after the addition of C2H4 to the H2 + D2 mixture (Fig. 2). This fraction is not negligible but also not enough to dominate the catalytic kinetics until the ethylene pressure reaches approximately 1% of the total reaction mixture (at least at 300 K). In fact, even at that point the rate of HD production on the ethylene-treated surface amounts to a significant fraction of the activity on the clean Pt. This behavior is temperature-dependent though, as indicated in Fig. 5. It is seen there that, as the reaction temperature is increased, the drop in catalytic activity with ethylene addition is higher, but also that a large contribution to this change comes from reversibly adsorbed species; the poisoning of the surface by ethylidyne reduces the HD scrambling activity by 70% at the most at a reaction temperature of 450 K. Two more things can be said about the role of ethylidyne in modifying the kinetics of the catalytic processes: (1) it is removable at higher temperatures (around 500 K),19,23,33,59 after which the Pt regains its clean-surface activity (Fig. 6), and (2) with reaction mixtures containing more than 1% ethylene, poisoning occurs much more extensively, so the probability for ethane production can no longer be predicted by the hydrogen coverage estimated from the competition between H2/D2 and reversibly adsorbed ethylene adsorption (Fig. 9). In this last regime, the kinetics of both hydrogen adsorption/isotope scrambling and ethylene hydrogenation appear to be controlled by the coverage of ethylidyne on the surface, which is expected to be near saturation.37 This is the regime most commonly applicable to practical catalytic processes.

5. Conclusions

The kinetics of HD production from H2 + D2 mixtures promoted by a platinum polycrystalline foil was studied using molecular beams with fluxes equivalent to pressures in the mTorr range. The conversion on the clean surface exhibits high probabilities, increasing with increasing temperature between 300 and 500 K but then leveling off at approximately 60% of the maximum expected from statistical considerations. The HD production rates are linearly proportional to the beam flux within the range (600–4000 ML s−1) studied.

Addition of even small amounts of ethylene to the reaction mixture, either by mixing the olefin in the beam or by introducing it independently to the UHV chamber via a side leak valve, results in a drop in HD production activity of an order of magnitude or more. This change is critical, because it leads to a condition where the kinetics of hydrogen dissociative adsorption becomes comparable to that of the overall olefin hydrogenation. Beyond that initial decrease, additional HD formation rate reductions are seen with further increases in ethylene pressure or C2H4/(H2 + D2) ratio but at a much slower pace; the HD rate is only halved by increasing the ethylene pressure by 2–3 orders of magnitude. Some of the HD conversion rate reduction is reversible, and in fact is weakly dependent on the flux/pressure of ethylene used, but only 50% or less of the initial activity can be restored (at 300–400 K) by removing the ethylene from the mixture. The relative ratios of the reversible versus irreversible rate reductions are dependent on the temperature of the surface; higher temperatures reduce the overall HD activity but also the fraction that can be reversibly restored upon the removal of the ethylene from the gas phase. These two components of the catalytic activity reduction are ascribed to a competition of hydrogen and ethylene adsorption and to the formation of strongly adsorbed carbonaceous deposits on the surface, specifically ethylidyne (Pt3[triple bond, length as m-dash]C–CH3). The latter can in fact be removed with the same hydrogen beam at higher surface temperatures (∼500 K), after which the initial activity of the clean Pt is restored.

A mixture of ethane isotopologues is produced concurrently with HD during the conversion of ethylene with H2 + D2 beams promoted by the Pt surface. The total probability for the hydrogenation of ethylene to ethane is near unity until the ethylene fraction in the beam reaches values of about 10 parts per million, after which it drops almost linearly with increasing P(C2H4). Nevertheless, the reaction rate is still proportional to the coverage of atomic hydrogen on the surface, as estimated by a competitive Langmuir adsorption model with a strong preference for the uptake of the olefin. Only with reaction mixtures with 1% ethylene or more does the ethane reaction probability drop sharply, presumably because of the blockage of surface reaction sites by the near-monolayer coverage of the ethylidyne that forms on the surface. In terms of the ethane product composition, a statistical deuterium distribution typical of direct addition prevails for beams with 10 to 100 parts per million of ethylene, but the formation of C2H6 becomes dominant afterwards.

The three kinetic regimes identified in our studies with increasing concentration of olefin in the reaction mixture are associated with the evolution from the non-catalytic surface chemistry seen for the hydrogenation of double bonds under UHV conditions and the steady-state alkane production that can be sustained catalytically at atmospheric pressures. This transition, which has not been identified previously, is explained in terms of the predominant species adsorbed on the surface; first atomic hydrogen, second reversibly adsorbed π-bonded ethylene, and third ethylidyne, and in relation to the limiting factors controlling the kinetics – first a competitive adsorption between hydrogen and ethylene and later the poisoning of catalytic sites by the formation of ethylidyne surface species. These changes explain the different behavior seen under UHV versus in catalytic processes, and truly bridge the pressure gap. The intermediate pressure regime also affords the operation of a new Eley–Rideal mechanism not observed in previous studies.

Acknowledgements

Funding for this project was provided by a grant from the U.S. National Science Foundation (CHE-1359668).

References

  1. J. A. Turner, Science, 2004, 305, 972–974 CrossRef CAS PubMed .
  2. J. R. Rostrup-Nielsen, Catal. Rev.: Sci. Eng., 2004, 46, 247–270 CAS .
  3. P. E. Dodds, I. Staffell, A. D. Hawkes, F. Li, P. Grünewald, W. McDowall and P. Ekins, Int. J. Hydrogen Energy, 2015, 40, 2065–2083 CrossRef CAS .
  4. J. D. Holladay, J. Hu, D. L. King and Y. Wang, Catal. Today, 2009, 139, 244–260 CrossRef CAS .
  5. R. Chaubey, S. Sahu, O. O. James and S. Maity, Renewable Sustainable Energy Rev., 2013, 23, 443–462 CrossRef CAS .
  6. B. Dou, Y. Song, C. Wang, H. Chen and Y. Xu, Renewable Sustainable Energy Rev., 2014, 30, 950–960 CrossRef CAS .
  7. M. Appl, in Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH, 2000,  DOI:10.1002/14356007.a02_143.pub2 .
  8. J. G. Speight, The Chemistry and Technology of Petroleum, CRC Press, Boca Raton, FL, 5th edn, 2014 Search PubMed .
  9. M. K. Debe, Nature, 2012, 486, 43–51 CrossRef CAS PubMed .
  10. B. C. Gates, J. R. Katzer and G. C. A. Schuit, Chemistry of Catalytic Processes, McGraw-Hill, New York, 1979 Search PubMed .
  11. K. Christmann, Surf. Sci. Rep., 1988, 9, 1–163 CrossRef .
  12. R. A. Olsen and L. B. F. Juurlink, in Dynamics of Gas-Surface Interactions: Atomic-level Understanding of Scattering Processes at Surfaces, ed. R. Díez Muiño and H. F. Busnengo, Springer, Berlin, 2013, ch. 5, vol. 50, pp. 101–129 Search PubMed .
  13. M. Wilde and K. Fukutani, Surf. Sci. Rep., 2014, 69, 196–295 CrossRef CAS .
  14. S. J. Thomson and G. Webb, J. Chem. Soc., Chem. Commun., 1976, 526–527 RSC .
  15. P. S. Cremer, X. Su, G. A. Somorjai and Y. R. Shen, J. Mol. Catal. A: Chem., 1998, 131, 225–241 CrossRef CAS .
  16. T. Ohtani, J. Kubota, J. N. Kondo, C. Hirose and K. Domen, J. Phys. Chem. B, 1999, 103, 4562–4565 CrossRef CAS .
  17. G. C. Bond, Metal-Catalysed Reactions of Hydrocarbons, Springer, New York, 2005 Search PubMed .
  18. F. Zaera, Phys. Chem. Chem. Phys., 2013, 15, 11988–12003 RSC .
  19. J. Simonovis, A. Tillekaratne and F. Zaera, J. Phys. Chem. C, 2017, 121, 2285–2293 CAS .
  20. M. Johansson, O. Lytken and I. Chorkendorff, J. Chem. Phys., 2008, 128, 034706 CrossRef CAS PubMed .
  21. J. Horiuti and K. Miyahara, Hydrogenation of Ethylene on Metallic Catalysts, C.-C. Kinetics Report NSRDS-NBC No. 13, National Bureau of Standards, U. S. Government Printing Office, Washington DC, 1968 Search PubMed.
  22. A. Tillekaratne, J. P. Simonovis and F. Zaera, Surf. Sci., 2016, 652, 134–141 CrossRef CAS .
  23. A. Tilekaratne, J. P. Simonovis, M. F. López Fagúndez, M. Ebrahimi and F. Zaera, ACS Catal., 2012, 2, 2259–2268 CrossRef CAS .
  24. M. Ebrahimi, J. P. Simonovis and F. Zaera, J. Phys. Chem. Lett., 2014, 5, 2121–2125 CrossRef CAS PubMed .
  25. Y. Dong, M. Ebrahimi, A. Tillekaratne, J. P. Simonovis and F. Zaera, Phys. Chem. Chem. Phys., 2016, 18, 19248–19258 RSC .
  26. J. Wilson, H. Guo, R. Morales, E. Podgornov, I. Lee and F. Zaera, Phys. Chem. Chem. Phys., 2007, 9, 3830–3852 RSC .
  27. Y. Amenomiya and R. F. Pottie, Can. J. Chem., 1968, 46, 1741–1746 CrossRef CAS .
  28. Registry of Mass Spectral Data, ed. E. Stenhagen, S. Abrahamson and F. McLafferty, Wiley-Interscience, New York, 1974 Search PubMed .
  29. TRC Spectral Data-Mass (formerly Selected Mass Spectral Data), Thermodynamics Research Center Hydrocarbon Project, Texas Engineering Experimental Station, The Texas A&M University System, College Station, 1984 Search PubMed.
  30. S. L. Bernasek and G. A. Somorjai, J. Chem. Phys., 1974, 62, 3149–3161 CrossRef .
  31. F. Zaera, Langmuir, 1996, 12, 88–94 CrossRef CAS .
  32. H. Öfner and F. Zaera, J. Phys. Chem. B, 1997, 101, 396–408 CrossRef .
  33. S. M. Davis, F. Zaera, B. Gordon and G. A. Somorjai, J. Catal., 1985, 92, 240–246 CrossRef CAS .
  34. Y. Dong, M. Ebrahimi, A. Tillekaratne and F. Zaera, J. Phys. Chem. Lett., 2016, 7, 2439–2443 CrossRef CAS PubMed .
  35. I. Horiuti and M. Polanyi, Trans. Faraday Soc., 1934, 30, 1164–1172 RSC .
  36. G. C. Bond, Trans. Faraday Soc., 1956, 52, 1235–1244 RSC .
  37. J. Simonovis and F. Zaera, Chem. Sci., 2016, 7, 4660–4666 RSC .
  38. G. H. Twigg, Discuss. Faraday Soc., 1950, 8, 152–159 RSC .
  39. S. M. Davis, F. Zaera and G. A. Somorjai, J. Catal., 1982, 77, 439 CrossRef CAS .
  40. T. P. Beebe, Jr., M. R. Albert and J. T. Yates, Jr., J. Catal., 1985, 96, 1–11 CrossRef .
  41. J. Houzvicka, R. Pestman and V. Ponec, Catal. Lett., 1995, 30, 289–296 CrossRef .
  42. M. J. Hostetler, R. G. Nuzzo and G. S. Girolami, J. Am. Chem. Soc., 1995, 117, 1814–1827 CrossRef CAS .
  43. A. Fási, J. T. Kiss, B. Török and I. Pálinkó, Appl. Catal., A, 2000, 200, 189–200 CrossRef .
  44. A. L. Marsh and G. A. Somorjai, Top. Catal., 2005, 34, 121–128 CrossRef CAS .
  45. D. Stacchiola, F. Calaza, T. Zheng and W. T. Tysoe, J. Mol. Catal. A: Chem., 2005, 228, 35 CrossRef CAS .
  46. D. Teschner, A. Pestryakov, E. Kleimenov, M. Hävecker, H. Bluhm, H. Sauer, A. Knop-Gericke and R. Schlögl, J. Catal., 2005, 230, 186–194 CrossRef CAS .
  47. B. Brandt, J.-H. Fischer, W. Ludwig, J. Libuda, F. Zaera, S. Schauermann and H.-J. Freund, J. Phys. Chem. C, 2008, 112, 11408–11420 CAS .
  48. U. Jung, A. Elsen, Y. Li, J. G. Smith, M. W. Small, E. A. Stach, A. I. Frenkel and R. G. Nuzzo, ACS Catal., 2015, 5, 1539–1551 CrossRef CAS .
  49. C. T. Campbell, Adv. Catal., 1989, 36, 1–54 CAS .
  50. H. J. Freund, N. Ernst, T. Risse, H. Hamann and G. Rupprechter, Phys. Status Solidi A, 2001, 187, 257–274 CrossRef CAS .
  51. G. A. Somorjai and J. Y. Park, Surf. Sci., 2009, 603, 1293–1300 CrossRef CAS .
  52. F. Gao and D. W. Goodman, Annu. Rev. Phys. Chem., 2012, 63, 265–286 CrossRef CAS PubMed .
  53. D. W. Blakely, E. I. Kozak, B. A. Sexton and G. A. Somorjai, J. Vac. Sci. Technol., 1976, 13, 1091–1096 CrossRef CAS .
  54. D. W. Goodman, Acc. Chem. Res., 1984, 17, 194–200 CrossRef CAS .
  55. M. Salmeron, R. J. Gale and G. A. Somorjai, J. Chem. Phys., 1977, 67, 5324–5334 CrossRef CAS .
  56. A. T. Gee, B. E. Hayden, C. Mormiche and T. S. Nunney, J. Chem. Phys., 2000, 112, 7660–7668 CrossRef CAS .
  57. A. Farkas, L. Farkas and E. K. Rideal, Proc. R. Soc. London, Ser. A, 1934, 146, 630–639 CrossRef CAS .
  58. P. S. Cremer, X. Su, Y. R. Shen and G. A. Somorjai, J. Am. Chem. Soc., 1996, 118, 2942–2949 CrossRef CAS .
  59. A. Wieckowski, S. D. Rosasco, G. N. Salaita, A. Hubbard, B. E. Bent, F. Zaera, D. Godbey and G. A. Somorjai, J. Am. Chem. Soc., 1985, 107, 5910–5920 CrossRef CAS .

This journal is © The Royal Society of Chemistry 2017