Photodestruction of CCl4 on MgO Films with/without Water - The...
17 downloads
161 Views
1MB Size
J. Phys. Chem. 1996, 100, 1055-1065
1055
Photodestruction of CCl4 on MgO Films with/without Water X.-L. Zhou and J. P. Cowin* EnVironmental Molecular Sciences Laboratory, Pacific Northwest Laboratory,† P.O. Box 999, M/S K2-14, Richland, Washington 99352 ReceiVed: August 15, 1995; In Final Form: October 9, 1995X
Ultrahigh-vacuum (UHV) studies show that UV irradiation (193 nm) of carbon tetrachloride adsorbed on a MgO surface produces phosgene both with/without coadsorbed water. Isotope studies show the oxygen required for the phosgene formation comes from the MgO lattice in the absence of water and from the water when coadsorbed. The reaction kinetics differ strongly between the wet and dry conditions. The observations suggest that UHV studies can reproduce important features seen in bulk studies of oxides for destruction of environmental hydrocarbons and that surface reactions may play a more important role in air or aqueous photocatalytic destruction of hydrocarbons than typically assumed.
Introduction Motivation. Carbon tetrachloride (CCl4) is a major solvent waste problem in the environment.1 This is particularly true at the nearby Department of Energy Hanford site, where 106 kg of CCl4 was dumped into the soil over 40 years of plutonium production. Its interaction with oxides is central to most issues, since soil is mostly oxides, as are the majority of the catalysts or scrubbing agents used or contemplated for destroying CCl4 in aqueous or air streams. Existing technology for destroying CCl4 or stripping it from soils has many serious shortcomings, in terms of incompleteness, expense, tendency to produce undersirable products, etc. Many aspects of CCl4 are similar to that for other chlorocarbons in the environment. However, CCl4 often is the exception to the rules and tends to be harder to destroy.2 Appropriately, there are many studies aimed at improving the situation: high-temperature incineration techniques for CCl4,3 low-temperature catalytic destruction,4,5 reactive scrubbers,6 or photocatalytic processes.2,7 Most applications and most studies to date on CCl4 destruction are on very fine powders. This clouds reaction mechanism determination since in traversing a typical powder, a molecule will make a random walk (or very slightly “downwind” preferred when a flowing gas is used), adsorbing and desorbing 1010 or more times. This makes it experimentally difficult to know whether a reaction that does occur had a 50% chance per surface visit or 1 in a million. Was the surface site responsible a typical site or that 1 in a million site? It is also difficult to ascertain the role of homogeneous (gas or liquid phase) reactions versus that due to heterogeneous surface reactions. Ultrahigh-vacuum (UHV) studies provide analytical tools and can clarify the issues of heterogeneous versus homogeneous. A major purpose of this study was to discovery whether the reaction kinetic conditions that can be explored in a UHV system provide access to the reaction phenomena that are seen in bulk powder work at high pressures (the answer is yes). Few prior UHV chemistry studies exist for CCl4 on any surface.8-12 The oxide reactions with CCl4 that occur in the course of thermally induced scrubbing reactions on powders have most typically been interpreted almost entirely in terms of surface reactions. Interestingly, the photocatalyzed air scrubbing experiments have been typically interpreted almost entirely in terms of gaseous reactions and in contrast, the photocatayltic * To whom correspondence should be addressed. † The Pacific Northwest Laboratory is a multiprogram national laboratory operated by Battelle Memorial Institute for the Department of Energy under Contract DE-AC06-76L0-1830. X Abstract published in AdVance ACS Abstracts, December 15, 1995.
0022-3654/96/20100-1055$12.00/0
destruction of chlorocarbons in aqueous streams was interpreted almost entirely in terms of a completely different set of homogeneous liquid phase reaction mechanisms. This is so despite often finding the same products and intermediates. One of the results of this study is to demonstrate that the heterogenous photolysis reaction of CCl4 to form phosgene proceeds easily, in the absence of either a gas phase or liquid phase. Another result is that a monolayer of water is all that is required to alter dramatically the kinetics. A monolayer of water will be often found not only under truly aqueous conditions, but also for many “dry” catalysts, at elevated temperatures. This suggests a more active role of surface reactions should be assumed in interpreting air or aqueous photocatalytic destruction and suggests that they hold more reaction steps in common. A very similar point has been made recently in the case of chlorinated ethanes/ethenes on TiO2,13 as will be discussed later in this paper. Phosgene as a product has had a long history of observation in chlorocarbon-oxide reactions and is obviously not a desirable final product when trying to remediate environmental problems. CCl4 passed over powdered TiO2 produced CO2, HCl, and OCCl2 and depended on the hydration state.14 CCl4 passed over a Fe2O3 produced CO2, Cl2, C2Cl4, and also a small amount of OCCl2. The latter was believed to be a “short-lived intermediate.”6a OCCl2 was produced from UV-irradiated liquid phase oxidation of CH2ClCH2Cl.15 Gas phase photodestruction over TiO2 powders of various chlorinated ethylenes was found to sometimes produce phosgene and Cl2 and other species in addition to HCl, H2O, and CO2.16 Golovanova et al.12 studied the photocatalytic decomposition of CCl4 on a silicon surface in the presence of chemisorbed water and reported OCCl2 formation. MgO was chosen, as it is a cheap and attractive reagent to scrub CCl4, to produce CO2 and disposable MgCl2. It is also a well-studied, prototypic base-type catalyst, with simple structure and stoichiometry. At this site another group has studied oxides (TiO2 and soda glass) for catalyzing/scrubbing CCl4 from air streams, after excitation of the steam by a room temperature corona discharge.5 Their observations of Cl retention/nonretention depending on water vapor presence, the formation of phosgene, and other details have strong parallels in our UHV studies on MgO and provide motivation for more detailed studies. We look for insight into its production via these UHV studies. We used lasers to initiate the reactions, rather than high temperatures or plasmas.17 This is because we wanted to cleanly separate the high activation chemistry that first dissociates the CCl4 molecule from the much lower activation energy © 1996 American Chemical Society
1056 J. Phys. Chem., Vol. 100, No. 3, 1996 channels that follow and determine the various final products. These studies were carried out on thin-film, single crystal MgO, grown on a (conductive) Mo(100) single crystal, similar to the method developed by Goodman’s group.18 Related Work. Both reagents CCl4 and H2O have been studied on MgO in UHV. Results for CCl4 adsorption/ desorption on MgO as studied by us are presented here and in ref 19. The adsorption of water on thin film MgO has been studied by Goodman and co-workers,18 Barteau and coworkers,20 and this group (presented here and in ref 21a). Their observations are similar (though not all the conclusions). A number of other papers deal with related photochemistry on MgO or of adsorbed CCl4.8,10,11,22-28 MgO has a band gap of 7.8 eV29 and is transparent to 193 nm photons. A surface resonance exists near 6.6 eV due to a self-trapped surface exciton (Mg2+ + O2- f Mg+ + O-).30 CCl4 has a broad UV absorption with a cross section in the gas phase of 1 × 10-18 cm2 at 193 nm31 and a quantum efficiency for dissociation near unity. (A few percent lead to Cl* and CCl2.32) In condensed phases the photochemistry is similar.33 Water is transparent at 193 nm. It can participate in photochemistry indirectly. For example, 193 nm photoexcited OHions (whether free in solution or adsorbed at a surface) can lose an electron to the water, leaving a reactive •OH.2,22 Most all the chlorine- and containing reaction products of intermediates are also subject to direct photochemistry. For example, CCl3 in the gas phase34 or in H2O35 shows a strong gross absorption (∼2 × 10-18 cm2) at 193 nm in what appears to be an overlap of a discrete and a continuous absorption. At 308 nm at least some of the photoabsorption leads to photodissociation.36 Experimental Section The experiments were carried out in an ultrahigh-vacuum system, described earlier.37 The procedures of mounting, cooling, heating, and cleaning of the Mo(100) crystal are described in more detail elsewhere21a and are similar to the methods developed by Goodman’s group.18 MgO thin films (20 monolayers thick) were prepared by simultaneously dosing Mg and O2 onto Mo(100) at 400 K18 and then raised to 10001100 K for about 10 s. This annealed surface was mostly flat terraces (via UHV scanning tunneling microscopy38), Auger electron spectroscopy showed only Mg and O, and low-energy electron diffraction showed a (1 × 1) pattern with diffuse background. Water adsorption studies showed it to be in an asymptotic limit with regard to increasing MgO coverage.21a CCl4, D2O, H218O (MDS Isotopes, 95 atom % 18O), and deionized H2O were dosed through a molecular beam doser,37 with the surface temperatures near 100 K. All the species dosed had sticking probabilities near unity on the surface, so that coverage was nearly proportional to exposure time to the molecular beam. One monolayer of CCl4 was defined as the exposure to fill the highest binding state as judged by the thermal desorption spectra. For H2O, one monolayer is defined as that needed to fill both the high-temperature state and the small 210 K state.19,21a Line-of-sight thermal desorption data were taken using an UTI mass spectrometer at a temperature ramping rate of 5-6 K/s. The UTI mass spectrometer as originally configured allowed a nanoamp-scale electron flux to leak to the sample, causing noticeable damage to the electron-sensitive adsorbates. This was eliminated by adding an isolated 80% open metal mesh enclosing the ionizer, which charged up to suppress the electron flow.39 The mass spectrometer was computer multiplexed, and surface temperatures (W-Re thermocouple) were measured via a 12 bit analog-to-digital converter and intermittently with a high-resolution microvoltmeter. Some apparent temperature shifts occurred over time due to
Zhou and Cowin
Figure 1. Thermal desorption of pure CCl4: 0.4, 1, and 2 monolayers of CCl4 adsorbed on the MgO surface are then desorbed during a 6 K/s heating ramp. The insert shows the desorption Ea versus coverage, calculated from the 2 monolayer data.
changes in crystal mounting and 12 bit ADC offsets and the cryogenic properties of the W-Re thermocouple. The second layer water desorption peak temperature on MgO and a single T (77 K) calibration were used to correct for these offsets, as needed. The self-consistency was within several degrees for most features, though (as in most surface studies) systematic errors of 10 K are possible. The laser was a Lambda Physik 101 excimer, operated at 193 nm, at about 10 pulses/s. The laser illuminated an aperture which was imaged onto the crystal at 45° from the normal. The fluence at the surface was 5-9 mJ/(pulse cm2) (per area at the crystal surface) and spatially uniform. The irradiated area was larger than the region dosed by the molecular beam. Three masses (mass 117, 166, and 201 amu) were monitored via the mass spectrometer to determine the desorption rates of CCl4, C2Cl4, and C2Cl6. The signal at mass 201 was due entirely to molecular C2Cl6. The mass spectrometer cracking patterns of all three compounds contribute to the signals at mass 117, and two contribute to mass 166. However, these two mass signals are predominantly due to a single parent, and small corrections are easily made to eliminate cracking pattern interferences. The cracking patterns of CCl4 and C2Cl4 were measured in our apparatus by leaking the pure gases into the main chamber (and were similar to published patterns40). We found that the molecular CCl4 desorption rate was given by (signal at mass 117) - 1.2(signal at mass 201) - 0.01(signal at mass 166), and the molecular C2Cl4 desorption rate was (signal at mass 166) - 0.52 (signal at mass 201). No sensitivity corrections were made. The prevalence of C2Cl4 was not recognized for much of the experiment, so mass 201 data are often not available. In that case the 166 amu data (C2Cl4 + C2Cl6) are reported. Cl2 was monitored at mass 70 amu. CCl4 gave a contribution to this mass at 0.016 times the signal at mass 117 amu. This was substrated from the reported Cl2 data. Results CCl4 Desorption. Figure 1 shows the temperature-programmed desorption (heating rate near 6 K/s) of 0.4, 1, and 2 monolayers of CCl4 from the clean, dry MgO surface. Adsorption of 0.4 monolayers results in a first-order-like desorption near 162 K. Completion of a close-packed layer (1 monolayer) occurs with little change in peak desorption temperature. Exposure to 2 monolayers of CCl4 gives two desorption peaks; the lower, new one (144 K) corresponds to CCl4 binding on top of itself. Nothing other than CCl4 was observed to desorb,
Photodestruction of CCl4 on MgO Films
Figure 2. Thermal desorption of pure D2O: 0.2, 0.8, and 1.7 monolayers of D2O adsorbed on the MgO surface are then desorbed by a 6 K/s heating ramp. The insert shows the desorption Ea versus coverage, calculated from the 1.7 monolayer data.
indicating that the MgO surface is chemically inert to CCl4, at least below the 162 K desorption temperature. The sticking probability was found to be unity. The signal is proportional to the desorption rate, and if one assumes that the desorption is a simple first-order desorption process with an Ahrrenius rate and a fixed preexponential, then the desorption activation energy as a function of coverage can be calculated directly from the data. For an assumed preexponential of 1014/s and using the 2 monolayer data, this gives the data shown in the insert of Figure 1. The 162 K peak implies that the CCl4 in the first monolayer has a binding energy of about 42 kJ/mol, rising to 44 kJ/mol for the first few percent of a monolayer. The latter may be due to defects in the surface. Water Desorption. Figure 2 shows the temperatureprogrammed desorption of 0.2, 0.8, and 1.7 monolayers of water (D2O) from the MgO (no CCl4 present). It too shows a first and second layer desorption peak. The first layer of water desorbs in a broad peak, centered near 240 K. This layer shows some evidence of being dissociated water, though this is a matter of controversy.18,20,21a For an assumed preexponential of 1014/s and using the 2 monolayer data, the desorption activation energy was calculated and is shown in the insert of Figure 2. The first monolayer of water desorbs with a 60-80 kJ/mol activation energy, with a plateau near 66 kJ/mol. As the sticking probability is one, this should be nearly equal to the binding energy21a and is much more than CCl4 feels on MgO. Coadsorption. We would expect that water would, in competition with CCl4 for surface sites, be very effective in displacing CCl4 to adjacent sites when available or on top of preadsorbed water when there is no more room in the first layer. The competitive adsorption for water and CCl4 on MgO is shown in Figure 3. Here we have placed 1/2 monolayer of water down on the surface first and then dosed what would have just sufficed to complete the first monolayer peak for CCl4 in Figure 1. (Reversing the order of deposition makes almost no difference for the 0.5 monolayer water case and minor differences for a whole monolayer of water.) Upon temperatureprogrammed desorption, we see that about 1/2 monolayer of CCl4 desorbs in a high-temperature peak, as from clean MgO regions, and the other 1/2 monolayer of CCl4 desorbs at lower temperature. The water remains after the CCl4 is gone and desorbs as if the CCl4 was never there. Less than a half monolayer of CCl4 fits nicely in the bare areas left by the 1/2 monolayer of water. In data not shown here,21a adsorption of 1 monolayer of water is just sufficient to completely block access of postadsorbed CCl4 to any bare MgO, eliminating the high-
J. Phys. Chem., Vol. 100, No. 3, 1996 1057
Figure 3. Thermal desorption of coadsorbed CCl4 and H2O: 0.25 and 1 monolayer of CCl4 desorbed from 0.5 monolayer of predosed D2O on the MgO surface, at a 5 K/s heating rate.
Figure 4. Dry MgO photochemistry: 1 monolayer of CCl4 on 20 monolayers of MgO was irradiated with 1.4 × 1019 photons/cm2 at 193 nm, at about 100 K, and then heated at 6 K/s. The dashed curves are the desorption spectra with no laser irradiation. Both CCl4 and C2Cl4 have data shown before and after subtracting out a mass spectrometer “cracking pattern” contribution from other species. The H for the HCl is from condensation of the background pressure of H2O: that shown is larger than typical. The CO2 data is from a separate run, at 1.1 × 1019 photons/cm2.
temperature peak for CCl4 altogether. Water and CCl4 seem to coadsorb in a simple manner, with the water taking preference for occupying the first monolayer. Dry MgO Photochemistry. Figure 4 shows temperatureprogrammed desorption data for 1 monolayer of CCl4 on 20 monolayers of MgO, without and with prior irradiation by 1.4 × 1019 photons/cm2 at 193 nm. The temperature-programmed desorption taken before laser irradiation shows the single monolayer peak at 165 K. After irradiation, there are two CCl4 peaks at 155 and 180 K. The area of the CCl4 peaks (versus time, not temperature) gives the amount of CCl4 remaining on the surface. After the irradiation, this integral was 0.5 that obtained without irradiation, indicating half of the surface CCl4 was photolyzed. After UV irradiation, temperature-programmed desorption shows desorption of new species: OCCl2, C2Cl4, C2Cl6, CO2, Cl2, and HCl. There are two peaks for OCCl2: a narrow one at 240 K and a broad one at 320 K. Cl2 desorbs near 200 K. CO2 commences desorption near 220 K and shows a broad peak. C2Cl4 and C2Cl6 each shows a peak at 240 K, and HCl shows a broad peak centered at 520 K. Auger electron spectroscopy taken after temperature-programmed desorption
1058 J. Phys. Chem., Vol. 100, No. 3, 1996
Figure 5. CCl4 versus laser fluence. CCl4 yield (117 amu) versus increasing 193 nm laser fluence is shown. Curves from the top down are at fluences of 0, 0.028, 0.056, 0.14, 0.28, 0.42, 0.7, 1.05, 1.4, 2.1, 2.8, 4.5 × 1019 photons/cm2, at about 100 K followed by heating at 6 K/s. The 0.028 and 0.056 curves tuncate, as they were off-scale. No cracking pattern corrections have been made, so contributions from C2Cl4 and C2Cl6 are present near 230 K.
Figure 6. C2Cl4 + C2Cl6 versus laser fluence. C2Cl4 + C2Cl6 yield (166 amu) versus increasing 193 nm laser fluence is shown. Curves from the bottom up are at fluences of 0, 0.028, 0.056, 0.098, 0.14, 0.21, 0.28, 0.42, 0.7, 1.05, 1.4, 2.1, 2.8, 3.6, and 4.5 × 1019 photons/ cm2, at about 100 K followed by heating at 6 K/s. No cracking pattern corrections have been made.
to 800 K showed chlorine but no carbon left on the surface. Surface chloride desorbs above 1300 K. The formation of C2Cl4 and C2Cl6 is not surprising as they should result from simple radicals reactions following from CCl4 + hν f CCl3 + Cl (primary channel), CCl4 + hν f CCl2 + 2Cl (or Cl2) (minor channel), and CCl3 + hν f CCl2 + Cl. Other people have commonly seen these products under various photolysis conditions: for example, C2Cl6 was found in the photolysis of CCl4 both in the gas phase and in ethanol solution at 254 nm,41,42 and C2Cl6 was found in laser photolysis of CCl4 adsorbed on GaAs(110).8 The CO2 produced was considerable in this figure. CO2 desorption studies are complicated by evolution of background CO2 from the chamber during heating of the sample. Largely for this reason (and unfortunate in retrospect) CO2 production was only monitored in the case shown. It is likely that the CO2 peak shown is directly evolved from the surface and is not a chamber artifact. HCl formation occurs because of the varying background pressure of water vapor that would develop (up to about 10-10 Torr) during the day’s runs of water adsorption/desorptions. Water adsorbed on the surface during the 5 min required for sample cooling or the tens of minutes required for photolysis.
Zhou and Cowin
Figure 7. OCCl2 versus laser fluence. OCCl2 yield (65 amu) versus increasing 193 nm laser fluence is shown. Curves from the bottom up are at fluences of 0.028, 0.056, 0.098, 0.14, 0.21, 0.28, 0.42, 0.7, 1.05, 1.4, 2.1, 2.8, 3.6, and 4.5 × 1019 photons/cm2, at about 100 K followed by heating at 6 K/s.
Figure 8. Cl2 versus laser fluence. Cl2 yield (70 amu signal, minus 0.016 times the signal at 117 amu) versus increasing 193 nm laser fluence is shown. Curves from the bottom up are at fluences of 0, 0.15, 0.3, 0.6, 0.9, 1.5, 2.2, 3.0, and 4.5 × 1019 photons/cm2, at about 100 K followed by heating at 6 K/s.
The amount would vary from day to daysthat shown in Figure 4 was somewhat larger than that typically seen. The effect of the water in such small amounts is illustrated later in more detail. The small shoulder at about 240 K on the CCl4 trace after correcting the the cracking pattern on the other products could arguably be due to errors in that correction. However, CCl4 does show a small desorption feature near 270 K, which is not due to any cracking of other identified species. Fluence Effects on Dry Photochemistry. Insight into the photochemical mechanisms can occur by looking at the fluence dependence of the reaction probability. The data of Figure 4 were taken at 1.4 × 1019 photons/cm2. We also took data spanning (0.028-4.5) × 1019 photons/cm2. Figure 5 shows the CCl4 desorption after laser fluences spanning this range. The high-temperature edge typically shows a small shoulder, probably largely due to C2Cl6 cracking in the mass spectrometer (no correction made for this figure for this). Figure 6 shows similarly collected data for mass 166 amu (C2Cl4 + C2Cl6), Figure 7 for phosgene, and Figure 8 for Cl2. Not shown is mass 36 curves (HCl), all of which show a peak near 500 K. The desorption peaks in Figures 5-8 and the HCl data were timeintegrated to get the relative yields versus laser fluence. The results are shown in Figure 9. The CCl4, C2Cl4 + C2Cl6, and
Photodestruction of CCl4 on MgO Films
Figure 9. “Dry” photolysis yields versus laser fluence. CCl4, (C2Cl4 +.5 C2Cl6 ) mass 166 amu), HCl, Cl2, and OCCl2 yields are given versus laser fluence. Curves are double-exponential fits to data, except for Cl2, which is line between points. CCl4 and mass 166 amu data closely follow curves, so data points omitted for simplicity. HCl data (∆), OCCl2 data (+). Note scale change at 1 × 1019 cm2 fluence. The double-exponential curves are used to “guide the eye”, except for CCl4, where quantitative use is made (see text).
Figure 10. Wet MgO photochemistry. At or below 100 K, 1 monolayer of H2O was dosed on 20 monolayers of MgO, followed by 1 monolayer of CCl4. Then it was irradiated with 1.4 × 1019 photons/ cm2 at 193 nm. The dashed curves are the temperature-programmed desorption spectra without laser irradiation. The heating rate was 6 K/s. Cracking pattern removal of signal at 166 amu (mostly C2Cl4) due to C2Cl6 is indicated.
Cl2 data points had little noise (mostly because they are narrow peaks) and were well fit by the curves shown (a doubleexponential decay, an exponential rise and decay, and a cubic polynomial, respectively). To avoid confusion with the more scattered data for OCCl2 and HCl, the data points for the less noisy data were omitted. Clearly, the results are not simply linear, nor show simple single-exponential decay. The mechanistic implications will be discussed later. Wet MgO Photochemistry. Irradiation with (1.4-14) × 1019 photons/cm2 at 193 nm for 1 monolayer of H2O alone on MgO showed no photoeffects (reaction or desorption) on the water. Figure 10 shows temperature-programmed desorption results for the case of having predosed the surface with 1 monolayer of water before dosing with 1 monolayer of CCl4. Temperature-programmed desorption without UV irradition shows a peak at 163 K for CCl4 and two peaks at 210 and 255 K for H2O (dashed curves). After an irradiation by 1.4 × 1019 photons/cm2, CCl4 desorption rate integral is reduced to 0.5 of the “dark” integral. H2O shows now four peaks between 170
J. Phys. Chem., Vol. 100, No. 3, 1996 1059 and 320 K. OCCl2, C2Cl4, C2Cl6, and HCl are also found, but their desorption kinetics are very different than in the absence of preadsorbed H2O. OCCl2 shows an intense peak at 145 K, C2Cl4 a peak at 180 K, and C2Cl6 a peak at 195 K, all three coming off at much lower temperatures than from the dry MgO surface. HCl is primarily in a broad peak centered at 490 K, though a small amount appears to desorb throughout 160-300 K. Predosing with a second layer of water prior to the CCl4 gave results nearly the same as shown in Figure 10, except that the phosgene product is shifted a few degrees colder yet. Figure 10 shows a second bump for desorption of C2Cl4, which appears real, but may be due to errors in the cracking pattern correction. Auger spectroscopy shows a complete absence of surface chlorine after this wet photochemistry and desorption of the HCl peak, unlike the dry surface case. 18O Isotope Study. In order to determine the origin of oxygen for OCCl2 formation, oxygen isotope-labeled water and gaseous oxygen were used. For simplicity, when no isotope for oxygen is indicated, it means 16O. As usual, the phosgene is monitored via mass 63 (or 65 or 67), corresponding to the predominant ion fragment COCl+ and its chlorine or oxygen isotope variants. Other work done by our group has already shown that adsorbed water does not exchange oxygen with this MgO substrate,21a consistent with higher temperature studies on MgO powders.21b Referring to Figure 11, for the case of CCl4/H2O/MgO, only CO35Cl+ and CO37Cl+ are seen in temperature-programmed desorption. For CCl4/H218O/MgO, the products C18O35Cl+ and C18O37Cl+ are dominant in temperatureprogrammed desorption, and the CO35Cl+ intensity is much weaker than for CCl4/H2O/MgO. The weak CO35C+ signal is primarily due to H2O, an impurity in the oxygen isotope-labeled water sample (95 atom % 18O). Similar results were found for CCl4 on multilayer H2O-covered MgO. On the dry MgO, the isotope labeling was done by using 18O2 to make the Mg18O. Referring again to Figure 11, the photolysis products C18O35Cl+ and C18O37Cl+ overwhelmingly dominate over C16O35Cl+. In this case, the weak signal for CO35Cl+ shows only a peak at 240 K and, unlike the other two isomers, almost none at 325 K. This small peak is due to adsorbed H2O from background, as illustrated in the next section. We also conclude that the lattice oxygen does not participate in the phosgene formation for the monolayer water-predosed surface. The roughly 10 K shift between the two dry surface data sets is undoubtedly due to drift of the electronics, which occasionally occurred between runs that could be used for temperature calibration. Effect of a Little Added Water. Water at less than a monolayer dosing will help understand how the dramatically different kinetics of Figures 4 and 10 transform into each other and also help understand the role of water from the chamber background pressure. Figure 12 shows the phosgene product 16OC35Cl+ from a Mg18O-labeled surface versus predose of H216O. Note that the narrow low-temperature phosgene peak grows rapidly in size and shifts continuously downward in temperature, while the broad peak changes slowly, at least initially. Similar data were taken at mass 65 (18OC35Cl+ and 18OC37Cl+) and some at mass 67 (18OC37Cl+). Figure 13 shows the evolution of C2Cl4, [16O]phosgene, [18O]phosgene, and H216O for photolysis of a monolayer of CCl4 on a 0.17 monolayer water predose. The [18O]phosgene narrow peak also shifts downward with water predose, closely matching the [16O]phosgene feature. These data allow plotting the yield integrals of the [16O]phosgene and the [18O]phosgene versus water predose. This is shown in Figure 14. The [18O]phosgene decreases with H2O dose dramatically. The [18O]phosgene is initially dominated by the broad portion of the integral, whereas
1060 J. Phys. Chem., Vol. 100, No. 3, 1996
Zhou and Cowin
Figure 11. Oxygen-18 study. Shown is the thermal desorption of 1 monolayer of CCl4 on 20 monolayers of MgO irradiated with 1.4 × 1019 photons/cm2 at 193 nm, showing OCCl+ (the most abundant mass spectrum fragment of phosgene). (a) Dry MgO: solid curves are for Mg16O and dashed curves are for Mg18O. (b) Wet MgO (predosed with 1 monolayer of H2O): solid curves are for H216O/Mg16O and dashed curves are for H218O/Mg16O. The heating rate was 6 K/s.
Figure 12. Effect of added H2O on OCCl2 production. Shown is 16Olabeled phosgene evolution (63 amu ) 16OC35Cl+), for an Mg18O surface, where H216O is added prior to dosing and photolysis of CCl4, at 1.4 × 1019 photons/cm2.
the [16O]phosgene is dominated always by the narrow peak. The data here agree generally with the desorption curves shown in Figure 11. Figure 13 shows that the C2Cl4 and C2Cl6 peaks did not change much in intensity with water addition. However, their temperatures of desorption decreased strongly with water addition, closely tracking the narrow phosgene peak over the range 0-0.17 monolayer water predose. As shown earlier, a full monolayer of water predose causes the phosgene, C2Cl4, and C2Cl6 to desorb all at slightly different temperatures. From 0 to 0.17 monolayer water predose, the high-temperature edge of the CCl4 desorption peak also tracks the rising edge of the desorption peaks for phosgene, C2Cl4, and C2Cl6. Again, this no longer occurs at a full monolayer. C2Cl4 Desorption. The various photolysis products seen to desorb during heating could come off at a temperature characteristic of the activated process that created them. Or they could be produced at a lower temperature, and the products could subsequently desorb with an activation energy characterized by
Figure 13. Products for 0.17 monolayer of H2O Predose. Shown are some of the products evolved after photolysis of 1 monolayer of CCl4 on a Mg18O surface, with 0.17 monolayer of H216O predose. Laser fluence about 1.4 × 1019 photons/cm2.
the product binding energy to the surface. In this study, this possibility was checked only for the product C2Cl4. This was done on the dry MgO surface and with 1 monolayer of water preadsorbed. It was not done on a prior photolyzed surface, which of course is different from that for the nonphotolyzed surface. Figure 15 shows the desorption for about 2 monolayers of C2Cl4 dosed on to the dry MgO surface and also for about 1 monolayer dosed on the MgO with a predose of 1 monolayer of H2O. The C2Cl4 seemed to desorb intact with no of loss of parent molecule. The amount of deposited C2Cl4 can be estimated from its dose time for this effusive source (50 or 25 s at 22 Torr, its room temperature vapor pressure) compared to the dosing time for a close-packed monolayer of CCl4 (25 s at 35 Torr). Time-of-Flight Studies. We used our line-of-sight mass spectrometer along with a Transiac transient digitizer to look for species ejected from the surface during the pulsed laser irradiation (similar to other time-of-flight studies we have done37). We saw for 1 monolayer of CCl4/MgO that some Cl
Photodestruction of CCl4 on MgO Films
J. Phys. Chem., Vol. 100, No. 3, 1996 1061 light, interact only via physisorption with the MgO. Both show a very distinctly stronger binding to the bare MgO surface in 1 monolayer or less adsorption compared to the binding of the second layer on top of the first. Water binds to the MgO more strongly than does CCl4. When they are coadsorbed, the water tends to preempt access to the bare MgO by the CCl4. When the water coverage is partial, the CCl4 will, as its coverage increases, first bind to the remaining empty sites on the MgO, and if more CCl4 is adsorbed, then take the next best choice: adsorb on top of the water. The reactions produce a number of products. Unraveling what it means in terms of mechanism will draw heavily on the order at which the products evolve and the strong effect of added water on the temperature of product evolution and its weak effect on the product distribution. Thermodynamics of Reactions. The possible reactions are better understood if the following thermodynamics are kept in mind (∆H°298 values in kJ/mol):43
CCl4 + 2MgO(s) f CO2 + 2MgCl2(s) ∆H°298 ) -367 (1) Figure 14. 18OCCl2 yield versus added H2O. Shown is the thermal desorption integral (yield) of phosgene from an Mg18O surface, versus H216O added prior to dosing and photolysis of CCl4, at 1.4 × 1019 photons/cm2. 63 amu is 16OC35Cl+. The 18OC35Cl+ is calculated by subtracting 0.33 of the 63 amu desorption curve from the 65 amu curve, to remove the 16OC37Cl+ contribution, before integration. The narrow low-temperature portion of the integral [O’s] and the broader hightemperature portion [X’s] are estimated from the data.
CCl4 + 3MgO(s) f MgCO3(s) + 2MgCl2(s) ∆H°298 ) -470 (2) CCl4 + MgO(s) f OCCl2 + MgCl2(s) ∆H°298 ) -155 (3) CCl4 + 2H2O f CO2 + 4HCl
∆H°298 ) -176 (4)
CCl4 + H2O f OCCl2 + 2HCl
∆H°298 ) -59
(5)
Cl + H2O f HCl + OH
∆H°298 ) +67
(6)
CCl3 + H2O f CCl3H + OH
∆H°298 ) +98
(7)
CCl3 + H2O f OCCl2 + HCl + 1/2H2 ∆H°298 ) -140 (8) Cl2 + CCl3 f Cl + CCl4
∆H°298 ) -174 (9)
Cl2 + CCl2 + H2O f OCCl2 + 2HCl ∆H°298 ) -140 (10) CCl2 + H2O f 2HCl + CO Figure 15. C2Cl4 thermal desorption. Shown are thermal desorptions to measure the desorption kinetics of intact C2Cl4. C2Cl4 vapor is used to deposit 2 monolayers on clean MgO (solid curve) and 1 monolayer on a 1 monolayer predose of H2O.
atoms were observed directly ejected, with CCl3, CCl2, and CCl radicals either absent or much weaker. The ejected radicals, at several tenths of an electronvolt translational energy, are consistent with photoexcitation of the CCl4 to a directly repulsive state. Discussion Photolysis of CCl4 on dry or “wet” MgO in ultrahigh vacuum yields the same products as typically seen in bulk, high-pressure (or liquid phase) reactions: C2Cl2, C2Cl6, phosgene, CO2 (probably), Cl2, and HCl. This occurs without a bulk water or gas phase, which suggests that the photomineralization reaction steps can occur entirely at the oxide surface. Some implications of this on proposed mechanisms for such reactions will be discussed below. First we will examine some of the details of the data presented above. Figures 1-3 show that CCl4 and water, in the absence of
∆H°298 ) -108
(11)
Reactions 1-5 remind us that reactions of CCl4 with either MgO or water are in principle energetically downhill to a variety of fairly stable products (including phosgene). Equations 6-8 illustrate that Cl or CCl3 radicals created by photolysis of CCl4 do not have a simple way to immediately react with water: they can be long-lived in water. The reactive photolysis of CCl4 can be thwarted if the radicals simply recombine. Quantifying Yield and Cross Section. The photochemistry could be due to photon adsorption either of the molecule CCl4 or of the solid, as both have direct adsorptions at this wavelength (the solid via charge transfer exciton states of the surface30). If we interpret the photochemistry seen in Figures 4 and 10 in terms of a molecular photoabsorption (assuming a molecular quantum efficiency of 1), whereby after 1.4 × 1019 photons/ cm2, one-half of the initial monolayer of adsorbed CCl4 was photolyzed, the effective cross section is 7 × 10-20 cm2. This must be corrected for the interface’s effect on the average optical electric field at the molecule via Maxwell’s equations. Assuming we can neglect the effect of the thin MgO film on the light fields, and using for the optical properties of Mo at 193 nm of 0.789 and 2.36 for the real and imaginary part of the refractive index,44 the average electric field intensity (E2) near the surface
1062 J. Phys. Chem., Vol. 100, No. 3, 1996 is 0.89 times that in the unpolarized incident beam. This gives a cross section of 8 × 10-20 cm2. The gas phase cross section is 1 × 10-18 cm2.31 The smaller value we observed could be due to the surface inducing quenching, lowering the molecular quantum efficiency to 0.08. (It is surprising that it would be identical for both the wet and dry MgO case, however.) Surface chemistry due to substrate excitations is often analyzed in terms of gross quantum yield: reactions per incident photon. After 1.4 × 1019 photons/cm2, one-half of the initial monolayer of adsorbed CCl4 was photolyzed. If we assume a surface density of CCl4 at 1 monolayer of that for 6 Å spheres close-packed, this gives 3.2 × 1014 molecules/cm2 equals 1 monolayer. This gives 10-5 as the quantum yield. We are not aware of any evaluation of the quantum yield for photochemistry or photoluminescence for MgO in the cases where the chemistry has been ascribed to substrate excitations.22 However, for chlorinated hydrocarbon reactions on TiO2 it is not unusual to achieve gross quantum efficiencies well over 10-2.16 Our 10-5 value suggests that if it is a substrate excitation, it is very inefficient. We cannot rule out a substrate excitation completely, but our data seem more consistent with a molecular excitation. When we replaced the single monolayer of predosed water with 2 monolayers (data not shown), the photochemistry yield or kinetics changed little. This again would suggest that the main excitation was a molecular excitation, as a 2 monolayer spacer layer would be expected to strongly alter the access to bulk excitation effects. The significance of the products observed depends on whether they are high-probability products or account for only a tiny fraction of the CCl4 consumed. The amount of CCl4 lost for 1.4 × 1019 photons/cm2 exposure was 1.6 × 1014 molecules/ cm2. The integrals of the thermal desorption spectra of the products measure the amount desorbed, if the response is calibrated directly against desorption of a known amount. Such information we have for only C2Cl4 (and reactants CCl4 and water). Desorption of CCl4 gives a desorption integral at mass 117 amu of 670 (arbitrary units) per 1014 molecules/cm2. For C2Cl4 data in Figure 15 the time integral is 840 in the first layer peak. Assuming van der Waals radii packing of this molecule flat on the surface implies 1 monolayer ought to be about 2 × 1014 molecules/cm2. We can separately estimate the surface density from the dosing conditions. One monolayer of C2Cl4 requires about 28 s at 22 Torr (pressure upstream of a small aperture), compared to 25 s at 35 Torr for a monolayer of CCl4. We can estimate the monolayer C2Cl4 coverage to be (22 Torr/ 35 Torr)(28 s/25 s)(152 amu/164 amu) ) 0.68 that of the 1 monolayer CCl4 dose or 2.2 × 1014 molecules/cm2 for a C2Cl4 monolayer. The two estimates are very close. Using the dosing time estimate gives 380 integral per 1014 molecules/cm2 for C2Cl4. The photochemical production of C2Cl4 from Figures 4 and 10 have integrals of 340 and 360 for the dry and wet cases, implying about 0.89 × 1014 and 0.95 × 1014 molecules/cm2 of C2Cl4 were produced. Since each molecule of C2Cl4 consumes two CCl4 molecules, this slightly more than accounts for the 1.6 × 1014 molecules/cm2 lost CCl4. C2Cl6 desorption yield can be crudely estimated by using only the published relative mass spectrometer sensitivity. Compared to CCl4 at the masses recorded, it is 0.9.40 Thus, the desorption integrals for C2Cl6 in Figures 4 and 10 (140 and 170) imply about 0.23 × 1014 and 0.28 × 1014 molecules/cm2 desorbed C2Cl6 molecules for the dry and wet case, respectively. These estimates and the general strength of the phosgene and CO2 (when it was measured) make it clear that the observed
Zhou and Cowin products are indeed important primary products and that the amount of lost CCl4 is easily accounted for (nominally somewhat overaccounted for) by the amount of carbon-containing products seen desorbing. Background Water. The data obtained by adding small amounts of water to the CCl4 + MgO system can be used to estimate the typical dosage of water via the background chamber pressure, which, given the HCl produced for the nominally “dry” MgO case, is clearly present. Figure 14 data for light phosgene production in the narrow peak can be extrapolated back to zero yield. This occurs at about -0.15 monolayer. The fluence for this data involved 5000 laser shots. At about 10 Hz, this requires 500 s. Roughly 1 monolayer of water would accumulate in 1 s at 10-6 Torr, so the background pressure is about 0.15/500 × 10-6 Torr or 3 × 10-10 Torr. This is severalfold higher than our residual gas analyzer suggests, however. Thermal desorption was occasionally performed without the laser firing, but waiting for the equivalent time: Some water was sometimes seen, but quite less than 0.15 monolayer. Figure 14 shows that from an Mg18O surface; the majority of the narrow peak phosgene evolved is 18Olabeledsnot due to background water adsorption. Nonetheless, for the longest laser exposures, background water adsorption should have provided perhaps 0.1-0.2 monolayer of water, which will be kept in mind as the data are interpreted. Photolysis Reaction Mechanism. The photochemistry of the nominally dry MgO shows (Figure 4) three products all evolving at the same temperature: C2Cl4, C2Cl6, and the narrow peak for phosgene. The data for CO2 evolution are very limited. However, it is seen in Figure 4 to commence at the same temperature. As water is added, these first three features stay locked together through 0.17 monolayer of water added, despite that large shift in temperature evolution. At low laser fluences, the narrow phosgene peak is absent (see Figure 7). This is quite likely because less water is present for these shorter duration experiments from background water. Yet the leading edge of the broad phosgene peak is still coincident with the C2Cl4/C2Cl6 peaks shown in Figure 6. This strongly suggests that all are associated with a common rate-limiting step. The temperature of the evolution of C2Cl4 from the dry surface is around 240 K, which is very much higher than the temperature that intact C2Cl4 desorbs from dry or wet MgO (Figure 15). This argues that C2Cl4 must have been produced on the surface at 240 K, and not earlier. Also interesting is that CCl4 ceases desorbing just as these products begin forming, very much as if the mechanism that produces these other products competitively shut down the rate of CCl4 desorption. Also note that despite the large change in kinetic rates due to water adsorption, almost no change in the relative amounts of products takes place. These facts make some sense if, after the 1.4 × 1019 photons/ cm2 at 100 K, there was almost no molecular CCl4 remaining on the surface. Looking at the CCl4 evolution at lower laser fluence, in Figure 5, it is clear that the thermal desorption is profoundly changed by much less than 1.4 × 1019 photons/ cm2. Figure 9 shows the integrated CCl4 desorption, represented by a fitted double exponential. Most of the curve is fit very well with an exponential decay consistent with a molecular absorption cross section of 8.5 × 10-20 cm2, similar to the overall cross section calculated earlier. However, 20% of the desorption integral is lost between zero fluence and the first integrable case for 0.1 × 1019 photons/cm2. The second exponential term is a fast one that gives that 20% loss and implies a cross section of roughly (1-2) × 10-18 cm2 or higher. This is approximately the value of the gas phase photolysis cross section of 1 × 10-18 cm2. We suggest that in fact the CCl4 is
Photodestruction of CCl4 on MgO Films almost entirely photolyzed with a gas-phase-like cross section in the range of fluences explored and that the CCl4 we see desorb comes from recombination of photolysis fragments. So at 0.1 × 1019 photons/cm2 most of the CCl4 would have been photodissociated to CCl3 and Cl. However, 80% of the CCl3 and Cl recombine to form CCl4 during the thermal desorption. The implied photolysis intermediates on the surface have several choices: recombine to form CCl4, form C2Cl4 or C2Cl6, or form other products like phosgene, Cl2, or CO2. This competition would synchronize the ending of CCl4 evolution with the rise of other products, as observed. The branching ratios to products we propose are controlled by the relative amounts of CCl3 and CCl2 present on the surface, a function of the laser fluence only (not the water predose). The radical CCl3 we suggest largely recombines with Cl to form CCl4, with some fraction forming C2Cl6, as in bulk photolysis/radiolysis of CCl4, where the only products seen are Cl2 and C2Cl6.45 Figure 9 shows that the production of C2Cl4 + C2Cl6 (mass 166) is abundant even at low fluences. The relative amounts of C2Cl4 to C2Cl6 (in the absence of mass 201 data, as is true for much of the data in Figure 9) can still be estimated using their cracking patterns at mass 117. The 117 amu data (mostly from CCl4) shows a shoulder on its hightemperature side (poorly displayed in Figure 5) that matches the 166 peak. At low laser fluences it has a relative intensity of 2 to 1 compared to the 166 amu peak, dropping to 1 to 2 at the highest laser fluence. This is compatible with the published cracking patterns of C2Cl4 and C2Cl6 only if mass 166 is predominantly from C2Cl6 at low laser fluence and predominantly C2Cl4 at high fluences (as it clearly is for Figure 4, at high laser fluence where 201 amu data are available). We suggest that the cross section of 8.5 × 10-20 cm2 for loss of CCl4 corresponds to the irreversible formation of CCl2, leading to phosgene. This can occur either via a second photon adsorbed by an intermediate CCl3 to produce CCl2 or by the suspected direct CCl2 production with a single photon.32 This is the source of the phosgene, we propose. The gross photoadsorption cross section for CCl3 at 193 nm, as mentioned in the Introduction, is about 2 × 10-18 cm2.34-36 So our observed cross section of 8.5 × 10-20 cm2 implies a quantum eficienty of 4.2%, which is reasonable. Similarly, a 5% branching ratio of the primary photolysis to CCl2 surmised in the gas phase32 would predict a CCl2 production cross section of 5 × 10-18 cm2. We propose both channels can contribute. The fluence dependence of the phosgene produced generally mirrors the loss of the CCl4 peak in Figure 9. The Cl2 formed desorbs at the same temperature as the hightemperature peak of CCl4. The laser fluence dependence of the Cl2 yield is markedly different than that of the other species: it shows initially a quadratic-like behavior. Cl2 requires two photons to occur but will give quadratic-like yields only if the yield is not limited by the amount of Cl available (which would give linear behavior). This means that the Cl2 could not be the typical fate for Cl atoms. We suggest that the Cl are more typically consumed in oxidizing surface O2- or water. Referring to Figure 4, we suggest that, after photolysis at 100 K, the surface has a lot of free CCl3, CCl2, and Cl radicals present and some residual CCl4, which are immobilized by diffusional barriers. The van der Waals size of the fragments considerably exceeds that of the initial monolayer of reagents, so only some of the fragments will find room to adsorb in the first layer. At 160 K, any undissociated CCl4 desorbs, along recombination from most weakly bound second layer radicals. At 180-190 K comes CCl4 from more strongly bond second layer fragments. Some of the mobile Cl’s form Cl2 as well,
J. Phys. Chem., Vol. 100, No. 3, 1996 1063 which promptly desorb. The cessation of CCl4 and Cl2 evolution occurring at 210 K occurs because the second layer fragments are consumed. Recombination of CCln occurs on the dry surface likely at the same temperature they are seen to desorb: The thermal desorption of 1 monolayer of C2Cl4 from dry and wet MgO at 180 and 160 K and the low evolution temperature (180 K) seen for C2Cl4 in the photolyzed wet MgO case strongly suggests the surface lifetime of the physisorbed products would be very short at 250 K and thus not rate limiting. The key to the two-photon process is the long life of the CCl3 intermediate. In solution this radical is surprisingly stable but would rapidly recombine in our time frame at room temperature. The low temperature of our experiment clearly can, through the reduction of diffusion, give the CCl3 radical a long enough lifetime to be photolyzed by a second photon. Even more generally, the oxide surface would be expected to form bonds with the CCl3 ranging from weak to modest, which could easily extend the two-photon channel to a wide temperature range. Charge Exchange. Though often not explicitly referred to for these kind of reactions, charge exchange is crucial to the photolysis. The reaction to extract oxygen from the lattice to form phosgene involves neutralizing a double negative oxygen anion. The thermodynamic stability of several of the reactions (1-3) that are strongly exothermic requires MgCl2 to form. This requires Cl accepting an electron from some donor. Cl neutral atoms should not strongly bind to Mg sites, as the Mg2+ has no covalent bonding ability, and the Cl should only feel weak induced-dipole binding, sitting over the cation. Similarly, the reactive carbene CCl2 sitting above a lattice O2- would also likely interact fairly weakly, as long as the O remained double negative (no covalent bonding ability). Both of these problems are remedied by an electron transferring from the O2- to the Cl, allowing the O- formed to covalently bond to CCln, as shown in Figure 16 and eqs 12 and 13.
Cl• + CCl3• + O2- f Cl- + CCl3O-
(12)
2Cl• + CCl2• + 2O2- f 2Cl- + CCl2(O-)2
(13)
The two oxygen atoms bound to the CCl2 in Figure 16 are diagonal across the MgO unit cell. The CCl3O- can evolve phosgene by losing an additional Cl-, or CCl2(O-)2 can produce phosgene by simple desorption or CO2 by losing two Cl-. These ionicly bound intermediates are our best speculation as to the thermally persistent predecessor to phosgene evolution occurring in the broad peak from 200 to 400 K. The Cl reaction to form chloride would not only permit phosgene to evolve but would displace CCl3 and CCl2 bound in the first layer, freeing them to recombine to form C2Cl4 and C2Cl6. This is because a high fraction of a monolayer of Cl is available, and there is simply not room for all the product Cland CCln radicals in the first layer. This would explain how evolution phosgene and these two recombination products would occur at the same temperature. When water added has reached a monolayer, the chlorination reaction has finally dropped to such a low temperature that the desorption of the products becomes rate limiting. Then they desorb at three different temperatures, corresponding to their respective binding energies to the surface. Water Role in Shifting Kinetics. One of the most constraining observations mechanistically is that as water is added to the reaction, the phosgene produced from this water evolves at a temperature which continuously drops as more water is added. We see that as this water is added, the source of oxygen
1064 J. Phys. Chem., Vol. 100, No. 3, 1996
Figure 16. Possible phosgene production mechanism.
continuously changes from the lattice to the water. If both reactions had activation energies that were independent of water coverage, we might expect to see two fairly fixed temperatures for evolving phosgene, with the amount evolving at each continuously shifting with water dose. This is not seen: water progressively lowers the activation energy itself. If the phosgene production reaction were first order with 1014 s-1 preexponential, then the data of Figure 12 say the phosgene production activation energy shifts from 59 to 38 kJ/mol, a 21 kJ/mol shift. This continuous shift with added water suggests to us that hydration by water of a charged species is involved in the rate-limiting step for phosgene production. We suggest that water’s fundamental role in lowering the reaction temperature is its ability to accelerate the chlorination step. This chlorine reaction inherently involves ions, which will be stabilized by coadsorbed water. This is the reaction step whose activation energy can be plausibly continuously lowered as water is added. Water not only accelerates the reaction to form phosgene; it supplies the oxygen for it and releases HCl as a product instead of refractory MgCl2. A simple possible mechanism for this is that, in the presence of water, the CCl2 reaction is still initially with a surface oxygen. Then water would react to displace the lattice oxygen from the intermediates illustrated in Figure 16. Relation to Bulk Gas/Liquid Catalysis Mechanisms. The reactions and mechanisms we discuss here are similar to those seen and proposed by a number of other workers for thermally driven CCl4 reactions on a variety of oxide powders at Torr and higher pressures. Primet et al.14 looked at products and surface intermediates using IR spectroscopy for CCl4 reacting with TiO2. They observed phosgene, CO2, and HCl as products. They found a strongly bound surface carbonyl-like intermediate, which they ascribed to something resembling the intermediates we show in Figure 16. He also found a sharp OH stretch which he correlated to a strong H bond to a surface chloride. A very similar complex may occur on the MgO surface and could help understand how the product HCl is evolved at such a high temperature (∼500 K). Reactions of CCl4 with CaO powder from 200 to 500 °C were studied by Klabunde and co-workers, who found phosgene, CO2, HCl, and C2Cl4 products.6c Klabunde and co-workers also studied6a CCl4 decomposition on Fe2O3. They saw similar products. They also proposed a reaction path similar to what we show in Figure 16 and, in particular, emphasized the role of CCl2 as an intermediate. Interestingly, photostimulated reactions involving many of the same powdered oxides have usually been interpreted in terms of liquid or gaseous reactions and have usually not reported phosgene production. One of the most studied such systems is that of chlorinated ethanes and ethenes on TiO2. A typical mechanistic scenario involves homogeneous-phase free radicals, with oxygen addition usually assumed to occur via addition of an OH• to an alkene, or O2 addition to a radical,14 and carboncarbon bond breaking via elimination of CO2, CO, or (seldom reported) OCCl2. In most cases the hypothesized reactions
Zhou and Cowin directly involving the solid have been minimized. Very recent photocatalytic work by several groups has strongly challenged this bias, showing the surface is more intimately involved in TiO2 reactions.14,46 Several groups14,47a,b very recently also have more strongly linked the photochemical reactions with the thermal ones, by finding that phosgene was a major product (or intermediate) in photolytic destruction of chlorocarbons, just as in the thermally driven reactions. Apparently most other workers either failed to detect or collect the more volatile, reactive phosgene or used longer residence times, which allowed the intermediate phosgene to transform to CO2.14,47a Additional evidence of the link between the thermal and photocatalytic mechanisms, and the importance of the role of CCl2, comes in the form of a recent paper on aqueous CCl4 photodegradation over TiO2, where CCl2 was suggested as an intermediate.48 Our work shows that the photochemical process of transforming CCl4 into OCCl2 and HCl on MgO requires no gas phase or liquid phase reactions and that it occurs readily, with small activation barriers, and does not require unusually special sites. Most of the powder studies involve such high surface areas with low “turnover rates” that it is not possible to conclude whether some obscure, rare defect or impurity state might have played a role in the reaction. Water Effects at Catalytic Conditions. Sensitivity of the kinetics of chlorocarbon degradation over oxides to water vapor has been often seen, including those with CCl4.5,22,25,49 A recent paper50 for example analyzed water vapor effects on the photooxidation over TiO2 of unsaturated organics. They found strong effects, typically increasing the rate at low water partial pressures and decreasing it for high values. This was well fit by a bimolecular surface reaction kinetics with water as a reagent, which also competed for the same surface binding sites as the organic molecule. Jacoby et al.13 found that water vapor altered the photolysis rate of trichloroethylene reaction over TiO2 and seemed to increase the rate of conversion of the intermediate phosgene to CO2. Our own study found that water strongly lowered the activation energy to turn the photolysis intermediates of CCl4 at the MgO surface into products. The role of surface water (or hydroxyls) is often ignored in gaseous reactions of oxides, often because from room temperature to several hundred degrees centigrade, where many such reactions are studied, it is presumed that negligible surface water exists. Our desorption kinetics calculated for water (Figure 2) allow us to estimate the surface coverage expected to exist at catalytic conditions, by placing the desorption rate predicted by the Arhennius equation in equilibrium with the impingement rate of water molecules at a given condition. The results are shown in Figure 17 at fixed water pressures from 1000 to 10-9 Torr. At 1000 Torr, for example, MgO powder would not be “dry” even at 1000 K, due to the high-temperature states for water adsorption. Even more striking is that at 10 Torr the surface will have nearly a monolayer of adsorbed water molecules for temperatures at or below 500 K. At room temperature, even parts per million water will lead to high surface coverages. Thus, for a reaction like the chlorocarbon destruction presented in this study, we would expect that MgO catalysts used with moist (like ambient) gas feeds would yield results most closely resembling the “wet” kinetics we showed in Figure 10, with no surface chlorination. The corona destruction experiments of CCl4 on powdered oxides here5 show similar effects, and we hope to address these more directly in future studies. Summary Phosgene, C2Cl4, C2Cl6, HCl, and surface chloride are observed from the photolysis of CCl4 on a single crystal thin
Photodestruction of CCl4 on MgO Films
Figure 17. Water isotherms (calculated) on MgO. Shown is the water coverage expected to exist on MgO versus temperature, at ambient water partial pressures from 10-9 to 1000 Torr, calculated using water/MgO desorption activation energies (see Figure 2).
film MgO surface. The kinetics and laser fluence dependence of their evolution suggest that CCl2 is the critical surface intermediate leading to phosgene production. Water causes the products to evolve at much lower temperatures, but without changing the product amounts. Hydration enhancement of the chlorination of the surface is suspected to play an important role. The observations here have many parallels in powder studies of MgO, CaO, TiO2, Al2O3, etc., which suggest that ultrahigh-vacuum studies should be able to identify intermediates and measure rates which will help discern mechanisms for the powder systems. Acknowledgment. Funding for this research was provided by the Office of Chemical Sciences of the Basic Energy Sciences Division of the Department of Energy. We appreciate the help of Martin Iedema in facilitating the experiment. References and Notes (1) Haag, W. R., Yao, C. C. D. EnViron. Sci. Technol. 1992, 26, 1005. (2) Ollis, D. F. In Homogeneous and Heterogeneous Photocatalysis; Pelizzetti, E., Serpone, N., Eds.; Reidel: Dordrecht, 1986; p 651. Ollis, D. F.; Pelizzetti, E.; Serpone, N. EnViron. Sci. Technol. 1991, 25, 1523. (3) Xieqi, Q. B.; Cicek, B.; Senkan, S. M. Combust. Flame 1993, 94, 131. Santolri, J. J. J. Chem. Eng. Prog. 1982, 69, 68. (4) Chatterjee, S.; Greene, H. L.; Park, Y. J. J. Catal. 1992, 138, 179. Petrosius, S. C.; Drago, R. S.; Young, V.; Grunewald, G. C. J. Am. Chem. Soc. 1993, 115, 6131. (5) Virden, J. W.; Heath, W. O.; Goheen, S. C.; Miller, M. C.; Mong, G. M. Department of Energy Report PNLSA20741, 1992; available via NTIS or via Pacific Northwest Lab. (6) (a) Hooker, P. D.; Klabunde, K. EnViron. Sci. Technol. 1994, 28, 1243. (b) Li, Y.-X.; Hui, Li; Klabunde, K. J. EnViron. Sci. Technol. 1994, 28, 1248. (c) Koper, O.; Li, Y.-X.; Klabunde, K. J. Chem. Mater. 1993, 5, 500. (7) Kittrell, J. R.; Quinlan, C. W.; Eldridge, J. W. J. Air Waste Manage. Assoc. 1991, 41, 1129. Sabin, F.; Tuerk, T.; Vogler, A.; J. Photochem. Photobiol. A 1992, 63, 99. (8) Liberman, V.; Haase, G.; Osgood, R. M., Jr. Surf. Sci. 1992, 268, 307. (9) Smentkowski, V. S.; Ellison, M. D.; Yates, J. T., Jr. Surf. Sci. 1990, 235, 116. (10) Dixon-Warren, S. J.; Jensen, E. T.; Polanyi, J. C. J. Chem. Phys. 1993, 98, 5938. (11) Sack, N. J.; Nair, L.; Madey, T. E. Surf. Sci. 1994, 310, 63. (12) Golovanova, G. F.; Petrov, A. S.; Silaev, E. A. Kinet. Katal. 1982, 23, 1275. (13) Jacoby, W. A.; Nimlos, M. R.; Blake, D. M.; Noble, R. D.; Koval,
J. Phys. Chem., Vol. 100, No. 3, 1996 1065 C. A. EnViron. Sci. Technol. 1994, 28, 1661. (14) Primet, M.; Basset, J.; Mathieu, M. V.; Prettre, M. J. Phys. Chem. 1970, 74, 2868. (15) Griesbaum, K.; Hayes, M. P.; Werli, V. Can. J. Chem. 1988, 66, 1366. (16) Nimios, M. R.; Jacoby, W. A.; Blake, D. M.; Milne, T. A. EnViron. Sci. Technol. 1993, 27, 732. (17) Bai, H.; Ault, B. S. Chem. Phys. Lett. 1992, 188, 126. (18) Wu, M.-C.; Corneille, J. S.; Estrada, C. A.; He, J.-W.; Goodman, D. W. Chem. Phys. Lett. 1991, 182, 472; J. Vac. Sci. Technol. A 1992, 10, 1467. Wu, M.-C.; Estrada, C. A.; Corneille, J. S.; Goodman, G. W. J. Chem. Phys. 1992, 96, 3892. (19) Zhou, X.-L.; Kwini, M.; Iedema, M. J.; Cowin, J. P. Manuscript in preparation. (20) Peng, X. D.; Barteau, M. A. Surf. Sci. 1990, 233, 283. (21) (a) Zhou, X.-L.; Iedema, M. J.; Cowin, J. P. Manuscript in preparation. (b) Li, Y.-X.; Klabunde, K. J. Chem. Mater. 1992, 4, 611. (22) Anpo, M.; Moon, S.-C.; Chiba, K.; Martra’s, G.; Coluccia, S. Res. Chem. Intermed. 1993, 19, 495-519. (23) Volodin, A. M.; Cherkashin, A. E.; Prokop’ev, K. N. Kinet. Catal. 1992, 33, 956. (24) Yanagisawa, Y.; Huzimura, R.; Matsumura, K.; Yamabe, S. Surf. Sci. 1991, 242, 513. (25) Watanabe, M. Surf. Sci. Lett. 1992, 279, L236. (26) McCarthy, M. I.; Gerber, R. B.; Trentelman, K. A.; Strupp, P.; Fairbrother, D. H.; Stair, P. C.; Weitz, E. J. Chem. Phys. 1992, 97, 5168. (27) Hodgson, L.; Ziegler, G.; Ferkel, H.; Riesler, H.; Wittig, C. Can. J. Chem. 1994, 72, 737. (28) Liu, Z. M.; Zhou, X. L.; Kiss, J.; White, J. M. Surf. Sci. 1993, 286, 233. (29) Henrich, V. E.; Kurtz, R. L. J. Vac. Sci. Technol. 1981, 18, 416. (30) Shulger, A. L.; Grimes, R. W.; Catlow, C. R. A.; Itoh, N. J. Phys.: Condens. Matter 1991, 3, 8027 and references therein. (31) Okabe, H. Photochemistry of Small Molecules; Wiley-Interscience: New York, 1978. Burton, G. R.; Chan, W. F.; Cooper, G.; Brion, C. E. Chem. Phys. 1994, 181, 147. (32) Matsumi, Y.; Tonokura, K.; Kawasaki, M.; Inoue, G.; Satyapal, S.; Bersohn, R. J. Chem. Phys. 1991, 94, 2699. Tiee, J. J.; Wampler, W. B.; Rice, W. W. Chem. Phys. Lett. 1979, 65, 425. (33) Chateauneuf, J. F. J. Am. Chem. Soc. 1993, 115, 1915 and references therein. (34) Danis, F.; Carapl, F.; Veyret, B.; Loirat, H.; Lesclaux, R. Int. J. Chem. Kinet. 1989, 21, 715. (35) Lesigne, B.; Gilles, L.; Woods, R. J. Can. J. Chem. 1974, 52, 1135. (36) Hintsa, E. J.; Zhao, X.; Jackson, W. M.; Miller, W. B.; Wodtke, A. M.; Lee, Y. T. J. Phys. Chem. 1991, 95, 2799. (37) Marsh, E. P.; Tabares, F. L.; Schneider, M. R.; Gilton, T. L.; Meier, W.; Cowin, J. P. J. Chem. Phys. 1990, 92, 2004. (38) Gallagher, M. C.; Fyfield, M. S.; Cowin, J. P.; Joyce, S. A. Manuscript in preparation. (39) Zhou, X.-L.; Schwaner, A. L.; White, J. M. J. Am. Chem. Soc. 1993, 115, 4309. (40) Cornu, A.; Massot, R. Compilation of Mass Spectrometric Data, 2nd ed.; Heyden & Son: Philadelphia, 1979; Vol. 1. (41) Davis, D. D.; Schmidt, J. F.; Neeley, C. M.; Hanrahan, R. J. J. Phys. Chem. 1975, 79, 11. (42) Kuchmii, S. Y.; Kulik, S. V.; Korzhak, A. V.; Belous, A. I.; Kryukov, A. I. Teor. Eksp. Khim. 1989, 25, 682. (43) Handbook of Chemistry and Physics, 74th ed.; Lide, D. R., Ed.; Chemical Rubber Co. Press: Boca Raton, FL, 1994. Hudgens, J. W.; Johnston, R. D., III, Timonen, R. S.; Seetula, J. A.; Gutman, D. J. Phys. Chem. 1991, 95, 4400. (44) Lynch, D. W.; Hunter, W. R. In Handbook of Optical Constants of Solids; Palik, E. D., Ed.; Academic Press: Orlando, 1985; p 311. (45) Sherman, W. V. In AdVances in Free Radical Chemistry; Williams, G. W., Ed.; Academic Press: New York, 1969; Vol. 3, p 53. (46) Pelizzetti, E.; Minero, C. Electrochim. Acta 1993, 38, 47 and references therein. (47) (a) Holden, W.; Marcellino, A.; Valic, D.; Weedon, A. C. In Photocatalytic Purification and Treatment of Water and Air; Olis, D. F., Al-Ekabi, H., Eds.; Elsevier Science: Dordrecht, 1993; p 393. (b) Nimlos, M. R.; Jacoby, W. A.; Blake, D. M.; Milne, T. A. In ref 47a, p 387. (48) Choi, W.; Hoffmann, M. R. EnViron. Sci. Technol. 1995, 29, 1646. (49) Petrosius, S. C.; Drago, R. S.; Young, V.; Grunewald, G. C. J. Am. Chem. Soc. 1993, 115, 6121. (50) Obee, T. N.; Brown, R. T. EnViron. Sci. Technol. 1995, 29, 1223.
JP952375Q
