Study of the kinetics of condensation and evaporation of water vapor over atmospherically relevant pure and doped ice films

This thesis presents the different steps in the development of a multiple diagnostic experimental setup and the results concerning the kinetics of condensation and evaporation of water over pure ice, as well as ice films doped with a known amount of atmospheric trace gases. The goal of this project is to obtain reliable data concerning the evaporative flux, Jev [molec cm-2 s-1], and the concomitant rate of condensation, kcond [s-1], of H2O molecules in the presence of an ice film. A further aim of this work is the detailed investigation of the impact on the aforementioned kinetic processes of the deposition of atmospherically relevant trace gases such as hydrogen chloride (HCl), hydrogen bromide and nitric acid (HNO3) under well defined conditions including those that are representative of the atmosphere. The new experimental apparatus consisting of a stainless steel reactor may be used under three different regimes, namely static, stirred flow and dynamic condition which correspond to equilibrium, low rate of pumping and molecular flow, respectively. The separation of the kinetics into individual processes is principally afforded by the application of these 3 regimes. First, using a combination of the first two regimes, we have measured Jev and γ of H2O in the case of pure as well as HCl- or HBr-doped ice in the range 190-240 K. The resulting Arrhenius expressions for pure ice are: in the presence of an average mole fraction of HCl, ΧHCl, in the range 3.2·10-5 to 6.4·10-3 we have found: and for ΧHBr at a HBr mole fraction smaller than 6.4·10-3 was found, where R=1.987 cal mol-1 K-1. Owing to the rate separation, γ applies to an individual rate process and should be called an accommodation rather than an uptake coefficient. Of utmost importance is the fact that during most of the ice film evaporation the vapor pressure remained that of pure ice even in the presence of small amounts of HCl or HBr on the order of a fraction to several formal monolayers. This is consistent with the enthalpies of sublimation that may be derived following the relation ΔHsubl0 = Econd - Eev. It leads to different thermochemical parameters A (pre-exponential factor) and E values (energy of activation) for the three cases that are identical within experimental uncertainty. In addition, FTIR spectroscopy of the ice films involved in these experiments in transmission have revealed that different species such as amorphous HCl·H2O mixtures or the crystalline HCl·6H2O hexahydrate are formed depending on the doping protocol. These two species significantly affect the temporal change of Jev each in a different manner. For the sake of precision and accuracy of the kinetic results, a quartz crystal microbalance (QCM) was subsequently added to the experimental apparatus and used as a substrate for ice film growth. The development of a QCM-based device for the investigation of the kinetic processes of ice films at atmospherically relevant temperatures was a challenge we have successfully surmounted whereas previously published attempt had given barely convincing results. The Arrhenius representation of the zero order evaporative flux Jev (molec cm-2 s-1) of H2O from pure ice displays a discontinuity at 193±2 K that could be observed owing to the accuracy and precision of the QCM measurements. It leads to: In addition, γ of H2O onto ice was found to be lower than 0.5 for temperatures below 240 K in contrast to the value of 1.0 which is often encountered in the literature. Subsequently, the deposited ice films have been doped with HNO3 using different protocols depending on parameters such as temperature, rate of deposition of HNO3 onto ice as well as number ofHNO3 molecules deposited. The reproducible and quantitative dosing of the doping gas was key to the quality of the results. This study has shed light on a complex interplay between several of these parameters that affect the continuous decrease of Jev during ice evaporation. Using atmospherically relevant conditions in the doping of ice by HNO3 we have observed crystalline α-NAT being most likely the first species that forms in/on ice films. During most of the evaporation of the ice film doped with known amounts of HNO3 we have checked that the vapor pressure over the ice film remains that of pure ice. Finally, HCl-doped ice films have been investigated in experiments that are complementary to the first set in order to verify that HCl doping has a similar impact on the change of Jev of H2O in the presence of small albeit known amounts of HCl. The doping protocol is a fundamental parameter determining the decrease of Jev with H2O evaporation time of doped ice films. In summary, we have developed from scratch a new experimental apparatus that allows the separation of the kinetic processes of condensation and evaporation of pure as well as doped ice films using multiple diagnostics under the same experimental conditions which was heretofore not attainable. The main result for pure ice films is that the uptake coefficient, γ of H2O on ice is always lower than approximately 0.5 for atmospherically relevant temperatures. Its negative temperature dependence has been confirmed. An accurate control of the doping protocol has allowed us to observe that for a fraction to a few monolayers of the doping gases, HCl, HBr and HNO3, the nature of the aforementioned protocol is the predominant factor influencing the continuous temporal decrease of Jev and γ during doped ice film evaporation. Of utmost importance is the fact that the vapor pressure over the doped ice films remains that of pure water ice during almost all of the evaporation process. These results may be of importance in atmospheric modeling for determining the lifetime of ice particles involved in clouds in regions that are undersaturated in the vapor pressure of H2O over ice. In this way the physical process of evaporation affects the importance of heterogeneous chemistry of atmospheric trace gases on ice substrates representative of Cirrus clouds, aviation induced Cirrus clouds and contrails as well as type II PSC's.

Heterogeneous kinetics of some atmospherically relevant reactions

A low pressure reactor (Knudsen cell) coupled to molecular beam modulated mass spectrometry was used to study the heterogeneous kinetics of the following three systems: (1) NO2 on amorphous carbon, (2) H2O vapor on ice and (3) gaseous HCl on ice. The first system was investigated at ambient temperature while the two last ones were investigated at temperatures typical of the stratosphere, that is roughly between 160 and 220K. The solid phase samples correspond to model surfaces whose physicochemical properties come very close to the ones of atmospheric particulates. Firstly, we studied the interaction of gaseous NO2 with three different commercial samples of amorphous carbon having widely differing physicochemical properties. Using in situ laser-induced fluorescence, the only major product was unambiguously determined to be NO with the oxidation product of carbon apparently remaining on the surface. Probing the gas phase with mass spectrometry (MS), both pulsed valve dosing and steady state experiments were performed and revealed a complex reaction mechanism for both the uptake as well as product formation. The initial uptake coefficient γo for NO2 was (6.4±2.0)·10-2 and proved to be identical within experimental uncertainty for all three types of amorphous carbon. The initial NO2 uptake rate was independent of the mass of the carbon sample and scaled with its external geometrical surface. Applying a simple chemical kinetic model to both pulsed dosing and steady state experiments resulted in an effective surface for uptake on amorphous carbon ranging between a factor of 2.8 to 8.4 larger than the geometrical surface area, but inversely proportional to the measured BET surface area of the three types of amorphous carbon. This means that the amorphous carbon with the highest BET surface (FW2) had the lowest external surface for NO2 adsorption. The chemical kinetic model included competing processes between Langmuir-type adsorption and inhibition controlling the adsorption kinetics of NO2 and revealed that the NO generation rate differed greatly between the three carbon samples examined. All samples showed saturation effects of differing degree that were partially reversible through prolonged pumping at 10-4 Torr and/or heating. Virgin amorphous carbon samples did not take up H2O vapor at 20 mTorr, and no HONO and/or HNO3 was detected in simultaneous NO2/H2O exposure experiments. CO and CO2 were detected when a sample previously exposed to NO2 was heated by an incandescent lamp. Moreover, upon heating such a sample, a MS signal m/e 62 originating from NO3 and/or N2O5 was detected. In addition to the experiments conducted on the three commercial carbon samples, we carried out uptake experiments of NO2 on acetylene soot. Originating from the flame of an acetylene burner, this soot was freshly deposited on glass dishes before each experiment. The initial uptake coefficient γo for NO2 was measured to be (3.0±1.1)·10-2 is smaller by a factor of two compared to γo observed in the uptake on commercial samples whose mass was a factor of 10 higher. In addition to NO, a net formation of HONO (m/e 47) was observed. This HONO formation is strongly dependent on the instantaneous water desorption from the sample as well as on the NO2 partial pressure in the reactor. On the other hand, it is not enhanced by an external water flow which did not show any interaction with the carbon sample. Upon heating the sample using an incandescent lamp, significant amounts of water desorbed, suggesting that the water vapor had condensed into the micropores of the soot during combustion. No quantitative results are presented yet. However, as no correlation between HONO formation and NO partial pressure has been observed so far, we suggest a reaction mechanism that only involves NO2 and H2O but no NO: 2 NO2 (g) + H2O HONO + HNO3 In a second study, the kinetics of condensation and evaporation of water on ice was determined in the temperature range of 170 to 220K. The rate of evaporation Fev corresponds to the evaporation of 70±10 monolayers of H2O per second at 200K. The condensation rate constant kc has a negative activation energy of -3.l±1.5kcal/mol which is significantly larger than the one recently measured by Haynes et al. We report the first real time kinetic measurements of water condensation on ice at stratospherically relevant temperatures. At temperatures above 160K, pulsed dosing experiments show a dose dependence of the condensation rate coefficient. For example, the condensation rate coefficient increases by a factor of five with a 1000 fold increase of the H2O dose from 1·1015 to 1·1018 molecules per pulse at 180K. This pressure dependence of the condensation rate constant may be explained by an autocatalytic mechanism involving the competition of two condensation channels. These two channels feed two different surface species, one of which leads to autocatalytic condensation. This mechanism gives insight into the manner in which a gas phase molecule is stepwise incorporated into tetrahedrally coordinated bulk ice. In a third and ongoing study, the uptake kinetics of HCl on ice has been investigated by pulsed valve experiments in the temperature range of 180 to 210K and are discussed as a function of the state of the surface. In fact, some of the pulsed valve experiments present a relaxation to a steady state level. It is shown that this level originates from the evaporation of a liquid solution corresponding to a temperature specific HCl concentration. The threshold of the injected dose leading to this solution layer is estimated to be 8·1014 molecules. This threshold corresponding to a fraction of a monolayer may be explained by the formation of small domains along grain boundaries or other lattice imperfections. Within detection limit, no difference in keff has been observed for pulses on the solution or on a previously exposed sample without formation of the solution. In the case of pulses forming a solution layer, a fraction of 20±20% of the injected molecules remains permanently adsorbed on the surface. In the temperature range from 210 to 180K, the pulsed valve experiments yield values of keff roughly between 5 and 25s-1 and an activation energy of -1.6±0.7kcal/mol which are consistent with the values observed by Flueckiger in steady state experiments using the 15mm escape aperture (kuni varying from 12 to 23s-1 and an activation energy of -1.9±0.25 kcal/mol).

EPFL1996

Production d'hydrogène dans un réacteur microstructuré

Pierre Reuse

The aim of this work was to operate the steam reforming of methanol and the total oxidation of methanol in a microstructured two-passage reactor. The steam reforming enables the production of hydrogen at relatively low temperatures (250 °C) with only a small amount of carbon monoxide (~ 1%). After a cleaning step (CO preferential oxidation, for example), this hydrogen would be a good feed for a fuel cell to produce electricity for mobile applications. Steam reforming is an endothermic reaction. The energy needed to run this reaction is produced by the total oxidation, a strongly exothermic reaction. To allow an efficient heat transfer between these two reactions, a microstructured reactor was used. These structures, with dimensions in the micron range, are widely known to have far better heat transfer capacity than traditional heat exchangers. Furthermore, these structures have some unique features with respect to classical reactors. A narrow residence time distribution, close to that of a plug flow reactor, can be obtained, keeping selectivity high for complex reactions. Microstructured reactors also have small residence time, allowing very quick variations in the feed and experimental conditions. In chapter 4, the concepts relating to the channel coatings will be presented. A Cu/Zn/Al catalyst provided by Süd-Chemie was transformed into a suspension of nanoparticles in isopropanol by using an attrition grinding apparatus. A thin layer (~5 µm) of catalyst was formed on the microchannel walls. Several tests were performed to characterize the adhesion of this layer. The influence of the layer on the residence time distribution was also investigated; it was possible to maintain Bodenstein numbers as high as 50. Chapter 5 is focused on the steam reforming of methanol, the reaction that generates hydrogen. A kinetic analysis in differential and integral modes established that the rate of the reaction is inhibited by hydrogen, the main product of the reaction; the partial order is - 0.2. The influence of water on the rate of reaction is negligible (partial order of 0.1), and methanol has a positive partial order of 0.7. A kinetic constant of 5.57 · 10-6 mol0.4 · (m3)0.6 · gcat-1 · s-1 at 200 °C and an activation energy of 73.9 kJ · mol-1 complete the description of the kinetics of the steam reforming. Throughout the comparisons, we showed that the catalyst coated in the microchannels has an initial rate for the steam reforming 30% higher than the original catalyst in fixed bed. We can therefore conclude that the catalyst modification procedure does not diminish the catalyst activity. This was also shown by comparing our results with results available in the scientific literature for similar catalysts. CO2 selectivity remains high and constant then decreases as 90% conversion is approached. Working with an excess of water with respect to methanol contributes to keeping the selectivity high. A molar ratio of 1.2 is a good compromise between selectivity and increased energy costs due to the vaporization of the additional water. It's also beneficial to work at conversions below 90% to be in the high selectivity domain and to diminish the inhibiting influence of hydrogen. Finally, we showed that this catalyst deactivates over 100 hours at a temperature of 300 °C. A kinetic model for the deactivation at temperatures between 300 °C and 330 °C was established. In chapter 6 the total oxidation of methanol was studied. A cobalt based catalyst was developed to allow the synchronization of the two reactions temperatures. The minimum temperature for complete conversion for the total oxidation reaction must clearly not exceed an acceptable working temperature for the steam reforming catalyst (i.e. a temperature that results in an acceptable catalyst lifetime). A kinetic analysis determined partial orders for methanol and oxygen as 0.67, the kinetic constant at 230 °C as 5.45 · 10-3 mmol-0.33 · m4 · (gcat · s)-1 and the activation energy as 130 kJ · mol-1. The copper based catalyst used for the steam reforming of methanol was also active for the total oxidation of methanol. Unfortunately the minimal temperature to ensure a total conversion was not compatible with the working temperature of the steam reforming. In chapter 7, the results concerning the coupling of the two previously studied reactions are presented. The numerous fittings and cables necessary for the operation of the reactor didn't allow the reactor to be run in an autothermal mode. Therefore constant, not regulated, heating was used to compensate the thermal losses. By coupling the two reactions and by studying the dynamic response time of the reactor (temperature and conversion of the steam reforming reaction) at the time of stopping and starting the oxidation reaction, we've seen that the temperature increase was around +10 °C within 60 seconds and that the conversion for the steam reforming of the methanol fitted well with the temperature increase. This reactor also has good dynamics and responds quickly to feed variations. When the coupling mode was changed from co-current to counter-current we noticed a significant increase in conversion. Furthermore, selectivity in counter-current mode remained high, even for conversion approaching 100%.

EPFL2003

Study of N2O decomposition over Fe-ZSM-5 with transient methods

Petra Prechtl

Nitrous oxide is the third most important gas contributing to global warming and its concentration in the atmosphere is still on the rise. Nitric acid production represents the largest source of N2O in the chemical industry, with a total annual emission of about 400 kt N2O. An efficient way to reduce the N2O is the catalytic decomposition into N2 and O2. ZSM-5 zeolites containing iron are known to be effective in this reaction involving Fe(II)-extraframework species as active sites. But the nature of these sites and the reaction network is still under debate. Often the N2O is in the exhaust together with other gases like NOx, COx, oxygen, and water vapor, which may influence the catalyst efficiency towards N2O decomposition. The effect of these gases on the reaction mechanism is still not understood yet. The aim of this work is to study the decomposition mechanism of N2O over isomorphously substituted Fe-ZSM-5 (0.02-0.55% Fe) using transient response methods. Kinetic studies under steady-state conditions give overall information about the reaction, whereas transient experiments lead to information on the individual steps and the possibility to suggest a reaction model. Transient response methods are applied in vacuum using a Temporal Analysis of Products (TAP) setup and at ambient pressure using a Micromeritics AutoChem 2910 analyzer. The influence of different gases such as NO, H2O and NO2 on the decomposition of N2O over Fe-ZSM-5 is investigated to gain insight into their effect on this reaction. The TAP multifunctional reactor system is used to perform transient pulse experiments for micro-kinetic analysis of complex heterogeneous catalytic reactions. This setup allows to dose precisely reactants and to monitor product formation with submillisecond time resolution implying the absence of external mass or heat transfer. A reproducible low hydroxylation level of the catalyst can be attained in vacuum. The results from the transient responses with the Micromeritics analyzer are obtained under conditions more relevant to "operando". The mechanism of N2O decomposition over Fe containing ZSM-5 was studied by transient response methods at ambient pressure with the Micromeritics analyzer and under vacuum with the TAP setup. Both approaches showed that the mechanism strongly depends on the reaction temperature, but in all cases involves atomic oxygen loading from N2O with N2 evolution as the first reaction step. The surface oxygen, (O)Fe, formed from N2O possesses very high reactivity and oxidizes CO to CO2 already at 373 K. A complete saturation of the active sites can be reached by the catalyst exposure to N2O at low temperatures (T < 600 K in vacuum, T < 523 K at ambient pressure). Under these conditions oxygen is stored and does not desorb. On that basis, the concentration of sites active for the surface atomic oxygen loading from N2O is determined by pulse experiments using the TAP setup. The concentration of active sites is found to be the same as in a flow experiment with the Micromeritics analyzer. The mechanism of surface atomic oxygen recombination and desorption as O2 depends on the temperature. At temperatures higher than 773 K, the mechanism probably involves direct (O)Fe recombination. The investigation of the response profiles upon N2O pulses at 803 K evidences the O2 formation is the rate-determining step since the oxygen desorption is found to be slow compared to N2O and N2. Furthermore, a fit of the oxygen desorption curve corresponded to a first order dependence on the oxygen surface coverage. In the range of 523 to 773 K, the O2 formation is facilitated by NO, which is formed on the catalyst surface from N2O. The addition of NO to the gas phase confirmed the accelerating effect of NO on the oxygen desorption. NO and (O)Fe are adsorbed on adjacent, but different sites and are not competing for the same ones. Adsorbed NO accommodates oxygen from N2O forming higher oxidized nitrogen oxide species NO2/NO3 which decompose back to NOads and molecular oxygen. Thus, the recombination of surface oxygen loaded from N2O is accelerated by a NO/NOx redox cycle whereas adsorbed NOx acts as co-catalyst. Adsorbed NO is oxidized by N2O easily as shown also by temperature-programmed desorption (TPD) experiments. Therefore, the interaction of gaseous NO2 with Fe-ZSM-5 at 523 K in vacuum and at ambient pressure is investigated and demonstrates the formation of two different NOx surface species accompanied by the evolution of gaseous NO. These surface species block the sites for oxygen loading from N2O whereas the amount of reversible adsorbed N2O stays constant compared to the standard pre-treated catalyst. The formation of reactive surface oxygen from NO2 could neither be evidenced in TPD nor with CO oxidation experiments. Adsorbed water strongly influences the N2O decomposition inhibiting it completely, at least at temperatures ≤ 673 K. Water pulses at 523 and 593 K result in partial desorption of the loaded atomic surface oxygen as O2, indicating the competition of water for the same adsorption sites. The residual adsorbed oxygen desorbs during TPD experiments at 849 K instead of 715 K for the dry catalyst and it can not oxidize CO to CO2. It means that the active sites are transformed into an inactive form in the presence of water. The oxygen which stays on the catalyst seems to participate in the Fe(II) oxidation to the hydroxylated Fe(III) species. These Fe(III) species are inactive in N2O decomposition and could be activated again by high temperature treatment in He or in vacuum. During TAP experiments only small amounts of water are dosed and the water desorption is facilitated due to vacuum. Therefore, the effect of water could be shown to be reversible at temperatures > 700 K. At 803 K oxygen desorbs significantly faster with increasing water amount suggesting a water assisted O2 desorption. Simultaneously, the catalyst activity decreases to about a half of the initial activity of the dry zeolite. To conclude, from the TAP reactor data obtained under vacuum conditions and the results from the transient flow experiments at ambient pressure it follows that the mechanism of the N2O decomposition over Fe-containing ZSM-5 depends strongly on the reaction temperature and on the presence of NO, H2O and NO2. In general, below 523 K only surface atomic oxygen loading from N2O takes place with simultaneous evolution of N2 in the gas phase. The O2 formation proceeds via direct (O)Fe recombination at temperatures above 773 K. In the range of 523 to 773 K, the recombination of the atomic oxygen involves NOx adsorbed on the catalyst surface.

EPFL2007