Thermal Plasma Deposition of Nanostructured Films

A thermal plasma process for the synthesis of nanoparticles and their immediate assembly into nanostructured films is discussed. In this process, known as hypersonic plasma particle deposition, a thermal plasma with injected precursors is expanded through a nozzle to nucleate nanoparticles, which are then inertially deposited onto a cooled substrate in vacuum. A lightly consolidated nanostructured film results. Particle and film diagnostics along with images of the plasma flow are used to explain the formation of nanostructured silicon carbide films by this process. WE have recently developed a plasma process, hypersonic plasma particle deposition (HPPD), for direct deposition of nanostructured films [1]. In HPPD vapor phase precursors are injected into a flowing thermal plasma generated by a dc arc torch. The plasma is then quenched by supersonic expansion in a ceramic lined nozzle, resulting in the nucleation and growth of ultrafine particles. The particles are further accelerated in the hypersonic free jet downstream of the nozzle and are deposited by inertial impaction onto a temperature-controlled substrate. This results in the formation of a consolidated nanostructured film. Inertial impaction of particles in impinging flows is generally associated with a sharp separation between larger particles which deposit and smaller ones which remain suspended. In order to maximize the deposition rate, it is necessary that the particles produced in the nozzle have diameters larger than the critical particle diameter needed for inertial impaction. In this paper, images of the plasma under typical deposition conditions are used in estimating the critical particle size, which is then compared with particle size distributions measured in flight using a sampling probe coupled to an aerosol size spectrometer. The experimental facility used for HPPD consists of a thermal plasma reactor exhausting into a low pressure deposition chamber, along with associated systems for pumping, gas handling, power supply, data acquisition, diagnostics, etc. The plasma reactor assembly consists of a dc arc torch coupled to a reactant injection tube and converging nozzle. A temperaturecontrolled molybdenum substrate is placed facing the flow exiting the nozzle. On-line diagnostics include calorimetry for obtaining cross-section average plasma temperatures at Manuscript received June 26, 1998; revised October 21, 1998. This work was supported by NSF (CTS-9520147, ECD-87-21545) and the Minnesota Supercomputer Institute. A. Neuman, J. Blum, Z. Wong, N. P. Rao, P. H. McMurry, J. V. R. Heberlein, and S. L. Girshick are with the Department of Mechanical Engineering, University of Minnesota, Minneapolis, MN 55455 USA (e-mail: [email protected]). N. Tymiak and W. Gerberich are with the Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, MN 55455 USA. Publisher Item Identifier S 0093-3813(99)02586-2. Fig. 1. A nanoparticle-laden hypersonic plasma free jet impinges against a substrate to deposit a nanostructured film. the various points in the water-cooled reactor, a sampling probe coupled to aerosol instruments for obtaining particle size distributions downstream of the nozzle, and a color video camera with a telephoto zoom lens for recording images of the plasma through a quartz window. Deposited films are analyzed by scanning electron microscopy (SEM). Data from these diagnostics systems are used to explain the deposition process. We have synthesized SiC films by injecting SiCl and CH into an Ar–H plasma (83% Ar). Fig. 1 is an image of the plasma deposition process recorded on videotape using the video camera and later digitized using a frame grabber. The thermal plasma in Fig. 1 has undergone a large expansion, from a source pressure of 500 torr at the nozzle entrance (not shown) to a background pressure of 2.5 torr in the deposition chamber. When the flow impinges against the substrate, a detached recompression shock is formed [2]. The stagnation flow downstream of the shock is rapidly heated to a temperature on the order of the stagnation temperature at the nozzle entrance, estimated from calorimetry to be 3700K. This feature of the flow is vividly captured by the greenish-blue emission from excited atoms/molecules in the stagnation region. However, the recompression stagnation pressure is much lower than that at the nozzle source due to the strong shock. This incomplete recompression enables particles accelerated in the free jet to penetrate the stagnation flow layer 0093–3813/99$10.00  1999 IEEE IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 27, NO. 1, FEBRUARY 1999 47 Fig. 2. An aerosol probe samples a particle-laden subsonic plasma jet emerging from the plasma reactor. and reach the substrate without being appreciably decelerated or reevaporated. The critical particle size for deposition depends on the recompression pressure and shock detachment distance, both of which can be controlled by varying the nozzle-to-substrate distance [2]. Previous studies of impinging hypersonic jets have used Schlieren techniques to highlight the jet structure. Here we take advantage of the distinct plasma emission in the stagnation region of Fig. 1 to determine the shock detachment distance for a hypersonic plasma flow. Using image analysis software (NIH Image), the shock detachment distance in Fig. 1 was measured to be 2.5 mm for a nozzle-to-substrate separation of 20 mm and a nozzle exit diameter of 5 mm. This is in good agreement with the value predicted for the case of a monatomic gas [3]. The pressure in the stagnation layer has been estimated from one-dimensional compressible flow theory to be about 18 torr. Based on these values and assuming a particle density of 2.3 g/cm (a reasonable lower limit) a cut-size of 5.7 nm has been estimated for an adiabatic expansion, using the inertial impaction scaling laws discussed in [2]. For the more realistic case of a nonadiabatic nozzle, the cut size may be as high as 8.4 nm. These estimates of critical particle diameter were compared with particle size distributions measured using a sampling probe coupled to an aerosol size spectrometer [1]. Fig. 2 is a color photograph of a sampling experiment in progress taken with a 35 mm camera. Particle size distributions measured using nozzles of different lengths and shapes are shown in Fig. 3 for conditions close to that for actual deposition. The size distributions are all lognormal with a geometric standard deviation of 1.6, and number median diameters ranging from 9–14 nm. It is seen that a majority of the particles produced in the reactor are larger than the critical size for Fig. 3. Probe measurements of particle size distributions produced by nozzles of different lengths and shapes. Fig. 4. SEM micrograph of nanostructured SiC film deposited by hypersonic plasma particle deposition. inertial impaction. The mass fraction of subcritical particles is negligible. Fig. 4 is a scanning electron micrograph of an actual SiC deposit deposited under conditions similar to that of Figs. 1 and 3. The mean grain size in the deposit is on the order of 20 nm, reasonably close to that of particles measured in flight. The deposit morphology is consistent with our understanding of the film formation process, i.e., a piling up of impacting particles.