Precursor kinetics and nanoparticle synthesis studied in a shock wave reactor

Transcription

1 Precursor kinetics and nanoparticle synthesis studied in a shock wave reactor P. Roth and A. Giesen Institut für Verbrennung und Gasdynamik, Universität Duisburg-Essen, Duisburg, Germany Abstract. Kineticists all over the world use the shock tube as a high temperature wave reactor for obtaining rate coefficient data under diffusion free conditions, because it provides a nearly one dimensional flow with practically instantaneous heating of the reactants. The temperature range under which the reactions could be studied can be extended far beyond that of conventional flow reactors. Compared to homogeneous chemical reactions, the study of heterogeneous kinetics is in an early stage. The reasons are that both, the degree of reaction complexity and the difficulties in the diagnostics, are significantly higher. For reactions in dispersed systems, the surface area of the reacting particles needs to be considered, which is experimentally not easy to access. Also, optical absorption diagnostics for measuring concentrations of gaseous species is significantly disturbed by the particles, because of light scattering and light extinction. In the present paper, some examples of different types of shock wave induced chemical reactions during the synthesis of nanoparticles will be illustrated. Examples are taken from the homogeneous decomposition of iron pentacarbonyl (IPC, FeCO 5), the nucleation of iron clusters, and the formation of iron particles. 1 Introduction Shock tube based research has, over the last five decades, uncovered several potential areas for scientific investigation. Though the main thrust was focused on applying the shock tube for aerodynamic and high temperature kinetic studies, several interdisciplinary areas have also been greatly benefited. A few examples of such interdisciplinary research, involving shock waves, can be seen in many intriguing medical applications of shock wave focusing; shock wave phenomena in geoscience and astrophysics; shock waves in condensed matter and shock wave initiated material synthesis. Paul Vielle [1], who operated the first shock tube in 1899 to understand gas explosions in mines, could not have foreseen the great potential of this experimental tool. The gas phase homogeneous kinetic experiments carried out in the recent past using shock wave reactors are characterized by two factors, namely very high dilution of the reactants by an inert gas (usually argon) and high sensitivity of the diagnostic techniques employed to monitor species. The main advantage of diluting the reactants with an inert gas is that the exothermicity or endothermicity of the reactions involved will not greatly alter the constant temperature conditions during the investigation. Secondly, by using very low initial reactant concentrations (up to 10 ppm), the influence of subsequent reactions can be totally avoided or reduced. This facilitates the study of just one or two elementary reactions with high accuracy and without being strongly disturbed by fast secondary reactions.

2 2 P. Roth and A. Giesen The shock tube as a wave reactor provides an excellent environment for the study of nucleation and growth of particles from the vapor phase at high temperatures. Apart from providing nearly instantaneous and uniform heating of reactants, it allows rapid quenching of products leading to particle condensation and growth. The effect of varying initial temperature, pressure, and mixture composition on the size and yield of the particles produced, can be conveniently studied in a shock tube. Beside nanoparticle synthesis in conventional flame and high temperature flow reactors, the shock tube can also be used for carrying out systematic high temperature investigations into the nanoparticle formation, their shape, size distribution, and yield. Several years ago, we have started a new series of investigations into particle formation behind shock waves, including carbonaceous particles, Si particles, TiN particles, and Fe based particles. The present paper summarizes the results obtained during the synthesis of iron nanoparticles. It starts with the precursor kinetics, illustrates the formation of iron clusters and demonstrates the formation of nano ironparticles. 2 Experimental The experiments were carried out in a conventional diaphragm type shock tube. The tube has an internal diameter of 80 mm, a driver section of 3.5 m, and a driven section of 5.7 m in length. Downstream to the measurement section a tube extension of 0.5 m in length can be installed to provide sufficiently long measurement times behind the incident shock wave. The gas mixtures used were prepared manometrically in a stainless steel UHV-vessel using IPC as iron precursor and Ar as diluentant. During the experiment the incident shock velocity was measured by means of four piezo-electric pressure transducers placed along the shock tube. The temperature and pressure behind incident and reflected shock wave were computed from the measured incident shock speed and the speed attenuation using a one-dimensional shock model. A more detailed description of the shock tube, the gas mixing system, and the experimental procedure is given elsewhere, see [2,3]. Microwave power supply Fe- Hollow cathode lamp HeNe-Laser Vacuum monochromator CO: 151 nm PMT Scope Lenses and filters Pressure transducers Microwave discharge lamp Diode Monochromator Fe: and nm PMT PMT Quartz window Driven section Fig. 1. Schematic of the measurement section of the shock tube equipped with various optical diagnostics for gas species and particles. Different optical in-situ measurement techniques were applied to the measurement section. Figure 1 shows their schematic setup. For concentration measurements of gaseous species a resonance absorption spectroscopic setup was used. The Fe-atom resonance absorption diagnostic (λ = nm and nm) consists of a pulsed Fe-hollow cathode

3 Precursor kinetics and nanoparticle synthesis studied in a shock reactor 3 lamp, the optical path length, a monochromator, and a photomultiplier. Perpendicular to the Fe-ARAS system, the CO-MRAS diagnostic (λ = 151 nm) consisting of a microwave excited discharge lamp, a vacuum ultraviolet monochromator, and a solar blind photomultiplier is arranged. Both diagnostics require calibration due to the unknown spectral profiles of these line emission - line absorption techniques. Therefore series of shock-wave experiments with known concentrations of Fe and CO have been performed to relate the measured absorptions to the corresponding concentrations. For more details see [4,5]. Cw-laser extinction and time-resolved laser-induced-incandescence (TR-LII) were used to monitor the particle formation and growth. The TR-LII-setup consists of a pulsed Nd:YAG-laser for particle heating. The subsequent thermal emission of the particles is observed by a fast photomultiplier with a narrow interference filter (λ = 633 nm). The laser extinction setup consists of an HeNe-Laser, a narrow interference Filter and a photo diode. See also [6]. 3 Results and Discussion The formation of iron nanoparticles from the gas phase can be divided into three subprocesses. First a condensable iron vapor has to be formed by the thermal decomposition of the precursor, here IPC. Secondly the iron vapor condenses in a nucleation process to form iron clusters. The third process is the growth of the formed particles. In the following sections the investigation of these processes using a shock tube is presented. 3.1 Formation of Fe-atoms The thermal decomposition of IPC was studied behind incident shock waves at temperatures between 540 K and 730 K at pressures between 0.3 and 0.45 bar using gas mixtures of 5 and 10 ppm IPC/Ar. In all experiments, the absorption of CO-molecules and Fe-atoms were measured simultaneously and were transferred into concentration profiles by applying the calibration functions. A typical example of the measured concentration profiles during an experiment at 705 K is shown in Fig. 2, see noisy lines. Both species concentrations increase immediately after the shock wave heats the mixture. At a reaction time of about 300µs, the CO signal reaches the level of total available CO (5 [Fe(CO) 5 ]). At about the same reaction time, the Fe-atom concentration reaches a maximum which is close to the total available Fe-concentration. After this peak, the Fe concentration decreases during the remaining reaction time. The high dilution of the test gas mixtures facilitates the data interpretation as secondary reactions have only minor influence to the thermal decomposition of IPC. The experimental data proves, that the IPC decomposition proceeds via CO-abstraction. As all CO-signals show uniform increases up to the level of the total CO-concentrations, they provide no direct information on one of the five individual CO-abstraction steps. We therefore treat the complete IPC decomposition as a single reaction step: Fe(CO) 5 k 1 Fe + 5 CO (R1)

4 4 P. Roth and A. Giesen [CO] / cm x [Fe(CO) 5 ] Smirnov Fe - data [7] Smirnov Fe(CO) 5 - data [7] Lewis et al. [8] Didenkulova et al. [9] [Fe] / cm [Fe(CO) 5 ] 0 5 ppm Fe(CO) 5 T = 705 K p = 0.36 bar k 1 / s this study: 5 ppm Fe(CO) 5 10 ppm Fe(CO) K / T Fig. 2. Left: Measured CO- and Fe concentration profiles during the thermal decomposition of 5 ppm IPC at 705 K. Right: Arrhenius diagram for the rate coefficient of the IPC decomposition. and derived the corresponding rate coefficient by fitting calculated profiles to the measured ones, see solid lines in the concentration plots of Fig. 2. All individual values of the rate coefficient were determined in this way and are given in Fig. 2 with literature values. They can be summarized by an Arrhenius expression of k 1 = exp( kj/(rt )) [1/s]. For more details see [4]. The experimentally observed decrease of the iron concentration after the formation process has an inverse temperature dependency and is due to the formation of iron clusters, which is discussed in the following section. 3.2 Nucleation of Fe-atoms The condensation of Fe-atoms was studied behind incident shock waves in the temperature range of 750 K T 1150 K and pressures of 0.3 to 0.45 bar in gas mixtures of 30 to 100 ppm Fe(CO) 5 diluted in Argon. Typical examples illustrating the temperature dependency of the Fe-atom and CO-molecule resonance absorption in a 100 ppm Fe(CO) 5 gon mixture are given in Fig. 3. All Fe and CO absorption profiles show a fast increase due to the thermal decomposition of IPC to form Fe-atoms and COmolecules. In case of the highest temperature of 1110 K, the Fe-absorption shows after a few microseconds a constant absorption level, indicating no further reaction within 1 ms. A decrease of the experimental temperature, starting with T 1080 K, leads to a decreasing Fe-atom absorption with an inverse temperature dependency. At temperatures below 950 K, a significant Fe-consumption within 1 ms was measured. The simultaneously measured CO-absorption profiles show a fast increase due to the decomposition of the precursor and subsequent constant absorption levels during the whole observation time, indicating a complete decomposition of IPC and no further reaction of CO. Light extinction by particles was not observed. The Fe-concentration seems to be too low for a sufficient particle formation to interfere the spectral absorption measurements. In a second series of experiments with lower initial IPC concentration of 30 ppm the general

5 Precursor kinetics and nanoparticle synthesis studied in a shock reactor 5 1,0 1, K 0,8 0,8 Absorption by Fe-Atoms 0,6 0,4 0,2 830 K 970 K 900 K Absorption by CO-Molecules 0,6 0,4 0,2 0,0 0, Fig. 3. Measured Fe- and CO-absorption profiles for a 100 ppm Fe(CO) 5 different post schock temperatures. mixture at behavior of the measured absorption profiles is similar. The temperature at which the first Fe-consumption appears, is shifted to a lower value of T 940 K. The observed consumption of Fe atoms showing an inverse temperature dependency must mainly be caused by the stability and growth of Fe-clusters. For a kinetic interpretation of the experimental results a simplified reaction mechanism has been proposed, which contains the following subsystems: decomposition of the precursor Fe(CO) 5, formation and dissociation of small clusters, as well as growth and coagulation of clusters [5]. The bottleneck in the mechanism is the recombination of Fe-atoms and its reverse reaction which trigger all further Fe-condensation steps: Fe + Fe + M k 2 Fe 2 + M Fe 2 + M k 2 Fe + Fe + M. (R2) (R-2) The growth mechanism of Fe-clusters is considered to be a process of Fe-addition and abstraction. The rate coefficient of the exothermic reaction (R2) was determined from an experiment, where the reverse reaction is negligible. The value obtained is k 2 = cm 6 mol 2 s 1, which is in quite good agreement with the theoretically determined value of [10]. The rate coefficients for reactions of single clusters with Fe-atoms were estimated to be of the same magnitude like (R2). The rate coefficients for cluster decomposition were calculated from the rate coefficients of the forward reactions and the temperature dependent equilibrium constants of [11], except for reaction (R-2). Reaction (R-2) is regarded to be the most crucial one for interpreting the measured Fe-consumption behavior. Based on the reaction mechanism computer simulations were performed to determine the rate coefficient k 2 by fitting calculated to measured Fe-profiles. Figure 4 shows the result of a fitting procedure for an experiment performed at 815 K, see noisy line experiment, full line computer simulation. The dashed lines in Fig. 4 indicate the sensitivity of the Fe-concentration to variations

6 6 P. Roth and A. Giesen ppm Fe(CO) ppm Fe(CO) 5 Bauer and Frurip [10] Krestinin et al. [12] [Fe] / cm k -2 / 2 k -2 * k -2 / cm 6 mol -2 s K / T Fig. 4. Left Fe-Concentration profile of an experiment at T = 815 K using a 30 ppm Fe(CO) 5/Ar mixture. The dashed lines represent the sensitivity of the reaction mechanism to variations of (k 2). Right: Arrhenius plot of the rate coefficient k 2. of the rate coefficient k 2 by factors of 2. Best fit values of k 2 obtained from all experiments are summarized in the Arrhenius diagram of Fig. 4. The data points scatter around a straight line, which can be approximated by the following Arrhenius expression: k 2 = ± 0.40 exp( ±700 K / T) cm 3 mol 1 s 1. A comparison of the present values of k 2 with data from literature is also presented in Fig. 4. The kinetic model used to interpret the experiments showed that the thermal stability of Fe 2 -clusters represented by the recombination reaction of Fe-atoms and its reverse reaction are mostly responsible for nucleation. More details can be found in [5]. The application of a numerical nucleation model considering homogeneous nucleation and surface growth on the experimental data showed a very good agreement, when applying size dependent vapor pressure and surface energy terms [13]. 3.3 Formation and Growth of Fe-particles For efficient particle formation and growth, experiments were performed with a higher initial precursor concentration of 0.5% IPC/Ar. To measure the evolution of the particle diameter the TR-LII technique and Cw-laser extinction were applied. The principle of laser induced incandescence (LII) for particle sizing is based on a fast particle heat up due to the absorption of a short laser pulse and the observation of the subsequent thermal radiation. Time-resolved LII (TR-LII) makes use of the time behavior of the LII signal during particle cooling, as a measure of the particle size. Larger particles with a larger volume to surface area ratio need longer to cool down than smaller ones. Figure 5 shows a TR-LII signal of an experiment performed at 945 K. The laser was triggered after µs. Induced by the laser pulse the particles are heated up and the signal increases within 15 ns to a maximum before decreasing back to the background level within 600 ns. As a measure of particle cooling the 50%-decay time τ 50% is plotted.

7 Precursor kinetics and nanoparticle synthesis studied in a shock reactor 7 0,3 0.5 % Fe(CO) 5 in Ar T = 945 K p = 1.0 bar 20 Experiment Numerical model PM signal / V 0,2 0,1 0,0 τ 50% = 79 ns Mean particle diameter / nm ,5 752,0 752, Fig. 5. Left: Tr-LII signal measured µs after reaction start. Right: Resulting iron particle diameter at different reaction times for an average post shock temperature of 1100 K. For the interpretation of the TR-LII signals, the physics of particle heating and cooling have to be considered. This has been done by Roth and Filippov [14]. For moderate laser pulse energies, which were applied in the quoted experiments, evaporation and ionization of the particles could be excluded. The particle temperatures at the end of the laser puls can be estimated with the laser energy density and the complex refraction index of the material. For the particle cooling (the decreasing part of the TR-LII signal), three physical processes have to be considered: heat transfer to the carrier gas through convection, heat transfer through radiation, and particle evaporation. Particle evaporation can be excluded because the calculated particle peak temperature is below the vaporization limit. The heat transfer by radiation is very small compared to convection and can therefore be neglected. The time-dependent particle temperature during the cooling period can thus be assumed to be controlled by convective heat transfer alone. Based on the energy conservation, the time dependence of the particle temperature can be calculated and is a function of the particle size. The measured property in the TR-LII experiments is the particle thermal emission and not the particle temperature. Therefore the thermal emission of the particles is calculated from the particle temperature and can directly be compared to the measured signal. The transformation of all TR-LII signals into particle diameters is shown in Fig. 5 (left part). It represents the growth of the particles at an average temperature of 1100 K, depending on the reaction time. The TR-LII results are consistent with particle probes collected from the shock tube which were analyzed with a transmission electron microscope. A calculation of the iron particle growth based on Smoluchowsky coagulation and rapid sintering, see solid line, seems to fit the data points in Fig. 5 quite nicely. More details about this study can be found in [6].

8 8 P. Roth and A. Giesen 4 Conclusion The versatility of a shock tube as a high temperature wave reactor to study processes during the formation of nanoparticles has been illustrated. Especially the nearly instantaneous heat up of the reactants and the feasibility of isothermal reaction conditions provide an ideal basis to study such process. The combination of the shock tube with the resonance absorption spectroscopy gives a deep insight into the elementary reactions of the particle formation. This technique allowed the study of the decomposition of the precursor and the cluster formation. The application of the laser-induced incandescence and laser extinction provide macroscopic information about the particle growth. The combination of all techniques give a good view of the complete particle formation and growth process, which is presented here for the example iron. The achieved information are a valuable foundation for the understanding and the development of models of particle formation and growth. References 1. P. Vieille: Compte Rendus 129, 1228 (1899) 2. D. Woiki, P. Roth: Shock Wave 4, 95 (1994) 3. K. Thielen, P. Roth: Combust. Flame 69, 41 (1987) 4. D. Woiki, A. Giesen, P. Roth: Proc. Int. Symp. Shock Waves 23, 447 (2001) 5. A. Giesen, J. Herzler and P. Roth: J. Phys. Chem (2003) 6. R. Starke, B. Kock and P. Roth: Shock Waves 12(5), 351 (2003) 7. V. N. Smirnov: Kinetics and Catalysis 34, 523 (1993) 8. K. E. Lewis, D. M. Golden and G. P. Smith: J. Am. Chem. Soc (1984) 9. I. I. Didenkulova, M. L. Perepletchikov and Y. A. Aleksandrov: Russ. J. Chem. Phys. 64, 1836 (1993) 10. S. H. Bauer and D. J. Frurip: J. Phys. Chem. 81, 1015 (1977) 11. D. E. Jensen: J. C. S. Faraday II 76, 1494 (1980) 12. A. V. Krestinin, V. N. Smirnov and I. S. Zaslonko: Sov. J. Chem. Phys. 8(3), 689 (1991) 13. A. Giesen, A. Kowalik and P. Roth: Phase Trans. 77, 115 (2004) 14. P. Roth and A. V. Filippov: J. Aerosol Sci. 27, 95 (1996)