7208
Langmuir 2006, 22, 7208-7216
Atomic Oxygen Recombination on Quartz at High Temperature: Experiments and Molecular Dynamics Simulation L. Bedra,† M. Rutigliano,‡ M. Balat-Pichelin,† and M. Cacciatore*,‡ Laboratoire Proce´ de´ s, Mate´ riaux et Energie Solaire, PROMES-CNRS, Font-Romeu, France, and Istituto Metodologie Inorganiche e Plasmi, IMIP-CNR c/o Dipartimento di Chimica, Via Orabona N.4, 70126 Bari, Italy ReceiVed January 4, 2006. In Final Form: May 18, 2006 A joint experimental and theoretical approach has been developed to study oxygen atom recombination on a β-quartz surface. The experimental MESOX setup has been applied for the direct measurement of the atomic oxygen recombination coefficient γ at TS ) 1000 K. The time evolution of the relative atomic oxygen concentration in the cell is described by the diffusion equation because the mean free path of the atoms is less than the characteristic dimension of the reactor. The recombination coefficient γ is then calculated from the concentration profile obtained by visible spectroscopy. We get an experimental value of γ ) 0.008, which is a factor of about 3 less than the γ value reported for O recombination over β-cristobalite. The experimental results are discussed and compared with the semiclassical collision dynamics calculations performed on the same catalytic system aimed at determining the basic features of the surface catalytic activity. Agreement, both qualitative and quantitative, between the experimental and the theoretical recombination coefficients has been found that supports the Eley-Rideal recombination mechanism and gives more evidence of the impact that surface crystallographic variation has on catalytic activity. Also, several interesting aspects concerning the energetics and the mechanism of the surface processes involving the oxygen atoms are pointed out and discussed.
Introduction Investigations of the heterogeneous recombination of atomic oxygen over ceramic materials is of importance in many laboratories and natural plasma systems, and it is also crucial for understanding the heat overload on space vehicles during the earth re-entry phase. Under the latter conditions, the concentration of gaseous oxygen atoms in the gas-wall interlayer is related to the high temperatures reached around the space shuttle wall when a shock wave arises as the shuttle is making contact with the dense layers of the atmosphere. It is a well-established fact that oxygen atoms recombine at the wall of the spacecraft, thus being responsible for a large part of the chemical heat overload and thermal damage to the spacecraft’s thermal protection system.1 Depending on different scales of study,2 several parameters are used to evaluate the heat released on a material surface by gassolid heterogeneous reactions. Among them is the recombination coefficient γ, which represents the fraction of total oxygen atoms in the incoming flux that ends in molecular form after colliding with the catalytic surface. The knowledge of this collisional coefficient is essential to determining the heat released on the catalytic surface3 as well as for the kinetic modeling of reactive oxygen plasmas under diffusive conditions. This article deals with the heterogeneous oxygen atoms’ recombination over a silica polymorph, β-quartz, under extreme thermal conditions. The first part of this study is concerned with the experimental results obtained within the MESOX setup.4 Hence, the oxygen recombination coefficient is determined with respect to the quartz surface temperature ranging from 850 to 1450 K at an air pressure of 200 Pa. In the second part, the * To whom correspondence should be sent. E-mail: mario.cacciatore@ ba.imip.cnr.it. Tel: +39 080 544 2101. Fax: +39 080 544 2024. † PROMES-CNRS. ‡ IMIP-CNR. (1) Jumper, E. J. J. Phys. D: Appl. Phys. 1996, 29, 1021-1031. (2) Balat-Pichelin, M.; Bedra, L. High Temp. Mater. Proc. 2004, 8, 283-292. (3) Goulard, R. Jet Propul. 1958, 28, 737-745. (4) Balat-Pichelin, M.; Badie, J. M.; Berjoan, R.; Boubert, P. Chem. Phys. 2003, 291, 181-194.
molecular dynamics simulation of the interaction between the oxygen atoms and β-quartz is described, according to which a β-quartz “model” surface is built up and the lattice dynamics is calculated using an interatomic potential taken from the literature. In the second step, a tentative semiempirical interaction potential derived from basic physical considerations is proposed for the O/silica system. Then, the complete collision dynamics of the oxygen atoms propagating in the gas phase or at the silica surface and the phonon dynamics is obtained by solving self-consistently the relevant equations of motion within the framework of the semiclassical collisional method.5 The theoretical results are discussed and compared, whenever possible, with the experimental results obtained for β-quartz and with those already published for β-cristobalite.6,7 Several aspects concerning the energetic and dynamics of the collisional processes involving the oxygen atoms are presented and discussed. The ultimate goal of this study is to demonstrate, using a combined experimental and theoretical approach, the hypothesis according to which spatial arrangement of the atoms in the solid surface generates, even in case of a close lattice structure such as the silica polymorphs used in this work, a remarkable impact on the heterogeneous recombination of oxygen atoms. Experimental Results for Two Silica Polymorphs The MESOX setup is an appropriate experimental tool for measuring the heterogeneous oxygen atom recombination over materials (ceramics, metal alloys, etc.) under extreme thermal conditions. By associating a solar radiative energy concentrator and a microwave plasma generator, the MESOX device is able to reproduce similar thermodynamic conditions undergone by space vehicles during earth re-entry (i.e., low pressures and high tem(5) Billing, G. D. Dynamics of Molecule Surface Interactions; WileyInterscience: New York, 2000. (6) Balat-Pichelin, M.; Bedra, L.; Issoupov, V. Proceeding of 9th International Symposium on Materials in a Space EnVironment; Noorddwijk: The Netherlands, 2003. (7) Cacciatore, M.; Rutigliano, M.; Billing, G. D. J. Therm. Heat Transfer 1999, 13, 195-203.
10.1021/la060032l CCC: $33.50 © 2006 American Chemical Society Published on Web 07/12/2006
Atomic Oxygen Recombination on Quartz
Langmuir, Vol. 22, No. 17, 2006 7209
Figure 1. Scheme (a) and photograph (b) of the MESOX experimental setup for the optical emission spectroscopy measurement: (1) waveguide, (2) mirror, (3) viewport, (4) sample, (5) pyrometer, (6) aiming slit, (7) lens, (8) spectrometer, (9) CCD-3000, (10) computer, and (11) mass flowmeters. peratures). Atmospheric re-entry conditions can be partially simulated on this setup with the great advantage that pressure and temperature can be reproduced independently with high accuracy. This experimental device (Figure 1) is placed at the focus of the 6 kW solar furnace equipped with a variable opening shutter. It can be moved away from the focus to be replaced by a calorimeter to measure the incident solar flux. The available incident concentrated solar flux can reach 5 MW‚m-2.
The temperature measurements on front and back faces of the sample are realized using a single monochromatic optical pyrometer (λ ) 5 µm) with a system of one rotating mirror and two stationary mirrors. The experimental reactor consists of a quartz tube (50 cm length, 5 cm diameter) with CaF2 viewports for pyrometric measurement. This reactor crosses the refrigerated waveguide that contains the sample (25 mm diameter, 3 mm height) placed in the stagnation point position at the center of the discharge. A leakage
7210 Langmuir, Vol. 22, No. 17, 2006
Bedra et al.
gate, a gauge, and a vacuum pump are used to control the total pressure precisely during the experiment. The relative oxygen concentration profile in the microwave discharge, above the sample, is determined by optical emission spectroscopy and actinometry methods. The spectroscopic bench is composed of an optical sampling system including a lens, a mirror, and a monochromator (Triax 550 spectrometer, Jobin-Yvon) equipped with an optical multichannel analyzer. The microwave discharge is imaged by a silica lens (magnification 0.1) on the slit entrance of the monochromator. The 55 cm focal length monochromator working with a 1200 grooves/ mm grating and a 100 µm width slit allows a spectral resolution of 0.2 nm. The dispersed light is analyzed by means of the CCD matrix (1024 × 128) of the OMA detector. Each of the 128 lines of the matrix gives information on the relative atomic oxygen concentration at different distances from the surface of the sample with a spatial resolution of around 270 µm. A spectral analysis over the 128 lines is performed very quickly after solar radiation breaking. The total duration of a scan is 200 ms. Therefore, all of the necessary spectral and spatial information is taken simultaneously, allowing good accuracy. The actinometry technique is used to follow the relative atomic oxygen concentration profile along the discharge. A low known quantity of argon is introduced into the flow, and the evolution of the intensity ratio IO/IAr of the oxygen line (at 844.6 nm) to the argon line (at 842.4 nm) is measured along the discharge zone with the following assumptions: -the actinometer is introduced in low quantity as not to disturb the plasma; -the excited species have to be produced by electronic impact from the ground state; -the de-excitation of the species has to be essentially radiative; and -the energy dependencies of the cross sections of the electronic excitation of O and Ar have to be identical in theory, and at the least, the energy thresholds of the transition must be similar. The experimental conditions chosen are constant microwave power of 300 W, total air pressure of 200 Pa, and a total flow rate of 1.11 × 10-6 m3 s-1 with 5% argon. A cylindrical volume corresponding to the discharge zone is considered. Because the mean free path of the atoms (0.043 cm at 200 Pa) is less than the diameter of the reactor (5 cm), atom diffusion is given by the diffusion equation. We suppose that the convective transfer and the radial gradient in the reactor compared to the axial ones are negligible. Moreover, the stability of the ratio IO/IAr in the reactor allows us to neglect the recombination in volume and on the reactor wall. Finally, under steady-state conditions, the intensity ratio obtained by actinometry leads to the determination of the recombination coefficient γ from the following equation γ)
(
)
Io/IAr|x)L TS 4Do,air -1 Io/IAr|x)0 TL VL
(1)
where the spectral line intensity ratio of atomic oxygen and argon IO/IAr stands for the relative atomic oxygen concentration infinitely close to the wall surface (x ) 0) and for the boundary layer external limit (x ) L), which are respectively at temperatures TS and TL. Do,air is the binary diffusion coefficient of the gaseous oxygen atom in air, and V is the mean square velocity of the incident gaseous atoms. Whereas the sample surface temperature TS is directly given by optical pyrometer measurements, the gas temperature TL results from the N2 rotational temperature 1 mm above the surface. The total relative incertitude on the recombination coefficient is estimated to be (30%. However, better accuracy in the critical measurements in the gas phase, especially the temperature, will be reached by recording a complete profile in the reactive layer. The experimental results concerning the recombination coefficient for silica polymorphs used in this study, β-quartz and β-cristobalite, are plotted in Figure 2 with respect to the reciprocal surface temperature.
Figure 2. Experimental recombination coefficient γ for O2 formation on quartz and β-cristobalite as a function of the reciprocal surface temperature. Consequently, one could deduce the total recombination activation energy by finding the Arrhenius curve that gives the best fit of the experimental points. According to these curves, it is evident that the two materials have different behaviors with regard to the recombination reaction, and β-cristobalite is much more catalytic than β-quartz. The activation energies obtained from the Arrhenius recombination coefficient curves are 17.9 ( 0.6 kJ mol-1 for β-quartz, which stands in comparison to the activation energy of 17.3 kJ mol-1 found by Kim,8 and 27.5 ( 0.6 kJ mol-1 for β-cristobalite.4 Two hypotheses that could explain the catalytic difference between these two polymorph materials are the surface roughness and the top-layer lattice structure because the plasma thermodynamic parameters and the sample geometry are the same for all experiments. Thus, the surface atoms’ arrangement may lead to a strong impact on the adsorption process because of steric effects and the number of adsorption sites affected. Environmental scanning electron microscopy (ESEM) micrographs of our samples are shown in Figure 3, and AFM images are shown in Figure 4. It appears in Figure 3 that β-quartz has a higher specific area because of the numerous dislocations appearing on the surface. One could then imagine that this material would be more catalytic than β-cristobalite, which has a smoother surface, but experimental results contradict this assumption (Figure 2). Figure 4 presents AFM images obtained for both samples on the scale of 5 × 5 µm2 performed in collaboration with Gerasimova from Ural State University. The higher roughness of the quartz sample is confirmed by the results obtained from AFM: the respective Ra and RMS values are 134 and 164 nm for quartz and 85 and 104 nm for β-cristobalite. These analyses confirm that the recombination reaction has to be studied on the atomic scale and cannot be discussed from micrographs, even at large magnification such as that obtained by AFM for polycrystalline and amorphous materials. However, perfect unreconstructed crystalline structures of silica materials are well known and are characterized from a morphological point of view, although real surfaces cannot be totally defined because of all of the defects that could appear during the experiments. Therefore, computer simulation via MD calculations can be an appropriate tool for predicting the impact of different crystallographic arrangements on the atom recombination reaction efficiency.
Molecular Dynamics Computational Simulation The MD approach is multistage work, starting from the building up of the crystal surface to the dynamics of the physicochemical processes at the surface. To model a sample of β-quartz, a 3D lattice containing 198 atoms (132 O atoms coordinated to 66 Si atoms and disposed over 8 layers: K19/38/14/32/14/38/19/24) was constructed by (8) Kim, Y. C.; Boudart, M. Langmuir 1991, 7, 2999-3005.
Atomic Oxygen Recombination on Quartz
Langmuir, Vol. 22, No. 17, 2006 7211
Figure 3. ESEM surface micrographs with a magnification of 2000× for quartz and β-cristobalite.
Figure 4. AFM images on the scale of 5 × 5 µm2 for quartz and β-cristobalite.
Figure 5. Top (left) and side (right) views of a β-quartz cluster consisting of 198 atoms. The Si atoms are shown by black spheres.
extending the elementary cell9 of the β-quartz and checking both its stoichiometry and electroneutrality. The surface top layer is modeled with silicon atoms with dangling bonds. In Figure 5, the top and side views of the lattice considered in the simulations are reported. With heterogeneous interactions being clearly influenced by lattice motions, the lattice dynamics description is an essential step in MD simulations of chemical reactions taking place at a (9) Wyckoff, R. W. G. Crystal Structures; Wiley & Sons: New York, 1963; pp 13-25.
surface. To describe the phonon dynamics of the silica surface, a reliable interatomic potential is necessary. Among the interaction potentials proposed in the literature10-12 that are able to reproduce on an appropriate level different structural properties of real materials, the semiempirical van Beest-Kramer-van Santen (10) Vashista, P.; Kalia Rajiv, K.; Rino, J. P.; Ebbsjo¨, I. Phys. ReV. B 1990, 41, 12197-12209. (11) Tsuneyuki, S.; Tsukada, M.; Aoki, H.; Matsui, Y. Phys. ReV. Lett. 1988, 61, 869-872. (12) Van Beest, B. W. H.; Kramer, G. J.; Van Santen, R. A. Phys. ReV. Lett. 1990, 64, 1955-1958.
7212 Langmuir, Vol. 22, No. 17, 2006
Bedra et al.
Table 1. BKS Potential Parameters Used in Equation 2 BKS potential Vij ) Aije-bijrij - Cij/rij6 + 14.4(qiqj/rij) Aij(eV)
bij(Å-1)
Cij(eV Å6)
18003.7572 1388.7730
4.87318 2.76000
133.5381 175.0000
qi Si O Si-Si Si-O O-O
2.40 -1.20
(BKS) potential12 is often cited as a prime potential to use with silica polymorphs. The BKS potential is a pairwise potential given by
VBKS(rij) ) Aije-bijrij -
Cij rij
+ 14.4 6
qiqj rij
(2)
where the first term describes the purely repulsive interaction between the two lattice atoms, i and j, with rij being the distance between the two atoms. The second term is the C6 attractive dispersion term, and the last term is the Coulomb interaction between the charged Si and O atoms. The potential parameters of eq 2 are reported in Table 1. The phonon density of states for β-quartz is deduced from local perturbation theory (LPT). Hence, the components of the dynamical force constant matrix are obtained by calculating the second derivatives of the pairwise BKS potential applied to the nearest atoms of the lattice
[ ] [ KiR,jβ µij
)
i
