Aerosol processing for nanomanufacturing
- First Online:
- Cite this article as:
- Girshick, S.L. J Nanopart Res (2008) 10: 935. doi:10.1007/s11051-007-9331-6
- 201 Views
Advances in nanoparticle synthesis are opening new opportunities for a broad variety of technologies that exploit the special properties of matter at the nanoscale. To realize this potential will require the development of new technologies for processing nanoparticles, so as to utilize them in a manufacturing context. Two important classes of such processing technologies include the controlled deposition of nanoparticles onto surfaces, and the application of chemically specific coatings onto individual nanoparticles, so as to either passivate or functionalize their surfaces. This paper provides an overview of three technologies related to these objectives, with an emphasis on aerosol-based methods: first, the deposition of nanoparticles by hypersonic impaction, so as so spray-coat large areas with nanoparticles; second, the use of aerodynamic lenses to produce focused beams of nanoparticles, with beam widths of a few tens of microns, so as to integrate nanoparticle-based structures into microelectromechanical systems; and third, the coating of individual nanoparticles by means of photoinduced chemical vapor deposition (photo-CVD), driven by excimer lamps. We also discuss the combination of these technologies, so that nanoparticle synthesis, together with multiple processing steps, can be accomplished in a single flow stream.
KeywordsHypersonic plasma particle depositionFocused particle beamsNanoparticle coatingNanomanufacturingSynthesisProcessing
A variety of gas-phase methods are used to synthesize nanoparticles. Synthesizing nanoparticles in the gas phase, as an aerosol, rather than in the liquid phase, as a colloid, has potential advantages that include greater purity, the ability to generate nanoparticles in continuous rather than batch mode, higher throughputs, and avoiding the need to manage environmentally hazardous solvents. Gas-phase methods also present a number of challenges, including the question of how to manipulate and control nanoparticles after they are synthesized, so that they can be utilized in a manufacturing context. While some studies have reported sequential operations involving gas-phase synthesis of nanoparticles followed by their processing in liquid solution (Mangolini et al. 2006), there would be clear advantages to accomplishing both synthesis and processing in a continuous gas flow stream.
In this paper we give an overview of technologies that we have developed for gas-phase processing of nanoparticles. Three specific technologies are discussed: spray-coating surfaces with nanoparticles by means of hypersonic impaction; focusing nanoparticles into narrow beams, using aerodynamic lenses, for use in microfabrication; and coating individual nanoparticles by means of photoinduced chemical vapor deposition (photo-CVD). Finally we consider the potential integration of nanoparticle synthesis, coating and focusing into a single process flow stream.
Spray-coating nanoparticles onto surfaces
The primary motivation for HPPD is that inertial impaction produces higher deposition efficiencies and higher impact velocities than alternative deposition mechanisms such as diffusion and thermophoresis. Both of the latter mechanisms can be expected to result in “soft landings” of nanoparticles onto surfaces, producing relatively porous and non-adherent films, whereas in HPPD, the combination of high impact velocity and high deposition efficiency produces relatively dense, adherent films, with high deposition rates. Recent molecular dynamics simulations of Si nanoparticles in the 5- to 10-nm-diameter range, impacting onto Si (001) substrates, indicate that impact velocities around 1 km s−1 are sufficient to generate significant deformation, including phase transformation, within the nanoparticle, and to promote good adhesion to the substrate (Valentini and Dumitrica 2007).
The materials deposited to date by HPPD include various combinations of the system of elements (Si, Ti, C, N). In experiments on stationary substrates, typical deposition rates lie in the range 1–10 μm min−1, depending mainly on reactant flow rates. These deposition rates are high enough to make it practical to coat larger areas by translating the substrate across the nanoparticle spray.
By design HPPD synthesizes and deposits nanoparticles, but the thin reactive boundary layer above the substrate may also be conducive to film growth by chemical vapor deposition (CVD) (Beaber et al. 2007; Girshick and Hafiz 2007), provided that not all of the condensible vapor has condensed into particles, which is supported by simple calculations. In HPPD a stationary bow shock is located, for typical conditions, ∼2 mm in front of the substrate. The gas temperature, which drops to quite low values in the hypersonic flow, recovers immediately past the shock to close to the stagnation temperature of ∼4000 K, and then drops steeply across the boundary layer to the substrate temperature. Boundary layers with such steep temperature gradients can provide high fluxes of radicals to a substrate, promoting high-rate film growth by CVD, as has been demonstrated for materials including diamond (Girshick 1998) and SiC (Liao et al. 2003).
Simultaneous film growth by nanoparticle impact and CVD may in fact be desirable, as the properties of a nanostuctured material depend both on the nanoscale grains and the interfacial material between the grains. Without interfacial material, even the closest packing of nanoparticles contains some degree of porosity, which can be deleterious to mechanical properties. If such a co-deposition process could be controlled, then one could envision growing films with a microstructure in which nanoparticles deposited by hypersonic impaction are distributed in a vapor-deposited matrix. Such an approach could potentially incorporate ceramic–metal combinations of vapor and particle phases to form nanocomposites with improved strength (Zhang and Chen 2007) and ductility (Pal 2005).
Results from recent HPPD experiments provide evidence that film growth does indeed occur by both nanoparticle impact and CVD. In studies involving SiC deposition by HPPD, deposits were found to consist of small (∼5-nm-diameter) single-crystal β-SiC grains, embedded in a matrix that included both a crystalline Si phase and an amorphous Si-C phase (Hafiz et al. 2006a). In separate experiments with the same run conditions, in situ measurements of particle size distributions, obtained using a scanning mobility particle sizer with an aerosol sampling probe located at the same location as the film deposition substrate (Wang et al. 2005b), demonstrated that the size distribution of the impacting nanoparticles was virtually the same as that of SiC particles observed in high-resolution transmission electron microscopy images. Taken together, these results indicate that the β-SiC grains were deposited by nanoparticle impact, while the matrix was deposited by CVD (Girshick and Hafiz 2007).
There are challenges to achieving uniform coatings with co-deposition by nanoparticle impact and CVD. For a stationary substrate, film growth by ballistic impact of nanoparticles follows a characteristic spatial distribution, with a concentration peak in the center of the deposit and a decaying population density of particles radially away from the center. In contrast, for a film grown by CVD with a stagnation flow boundary layer, the radial uniformity of the boundary layer (Schlichting 2000) implies that the film should be uniform over the substrate, assuming uniform substrate temperature and neglecting edge effects. Thus the film thickness is greater in the center than in the outer regions, suggesting that substrate rotation, with the substrate located acentrically with respect to the flow axis, is needed to achieve a uniform co-deposited film. Furthermore, CVD would in general be expected to be much more sensitive to substrate temperature than nanoparticle impact, as CVD typically involves activated heterogeneous chemical reactions. Thus, substrate temperature is expected to be an important parameter in determining the relative contributions of CVD and nanoparticle impact.
One indicator of the extent of CVD growth in an HPPD film is the texture coefficient, which quantifies the preferred growth orientation with respect to the principal crystal planes of the material. If a film were grown purely by nanoparticle impact, then one would expect the film to be randomly oriented with respect to each of the crystal planes of the material, whereas for CVD growth, a preferred orientation is often found. The texture coefficient TC with respect to a crystal plane (hkl) is determined by measuring the intensities I(hkl) of peaks associated with different crystal planes in an X-ray diffraction (XRD) measurement, and comparing them to a standard powder (hence, randomly oriented) of the material. The texture coefficient is defined by
If the film were randomly oriented, the texture coefficient for each growth direction would equal unity. As seen in Fig. 2, the film shows a predominantly (111) texture, for which the magnitude of the associated texture coefficient increases radially outward from the center. The (111) texture is consistent with previous studies of SiC film growth by CVD, by both thermal plasma (Liao et al. 2003, 2005) and other CVD methods (Kim and Choi 1999), particularly for substrate temperatures below ∼1,200 °C. The fact that texturing becomes stronger as one moves radially outward from the center is consistent with the fact that nanoparticle deposition, which results in a randomly orientated film, is concentrated in the center, whereas CVD growth, which favors texturing, is spread uniformly.
Spatially resolved measurements were also made of hardness, modulus and fracture toughness for HPPD SiC films. Details of these measurements are in Beaber et al. (2007). For films deposited with substrate temperatures ranging from 950 to 1,350 °C, the average hardness measured near the outside of the deposit (at a radial location r = 8 mm) was approximately 15 GPa higher than near the center (at r = 2 mm). For the same film as in Fig. 2, the hardness measured near the outer edge of the film averaged 39.4 ± 1.0 GPa, and the modulus averaged 370.9 ± 12.6 GPa. This hardness, which is substantially greater than the ∼28 GPa value for standard commercial SiC, lies on the border of the “superhard” regime, conventionally defined as hardness greater than 40 GPa. The fracture toughness of the same film, based on an average of 32 measurements made over an area extending from near the outer edge to a radial location 3 mm from the center of the deposit, equalled 6.03 ± 1.95 MPa m1/2, much higher than the nominal value of 3.3 MPa m1/2 for bulk SiC (McColm 1990).
It is of interest to compare these films to the superhard SiC films we previously deposited by thermal plasma chemical vapor deposition (TPCVD) (Liao et al. 2005). The TPCVD films were nanocrystalline but were grown by pure CVD, without nanoparticle impact. While the measured hardness of the TPCVD films reached above 50 GPa, well into the superhard regime, those results were for substrate temperatures around 1,200 °C. For substrate temperatures in the 1,000–1,100 °C range, the hardness measured 40–45 GPa, slightly higher than the 39.4 GPa value for the outer regions of the HPPD film deposited at 1,050 °C, and the hardness–modulus relationship for films deposited by the two methods (modulus equals approximately 10 times the hardness) was similar.
On the other hand the fracture toughness of the HPPD films is notably higher than the value of 3.9–4.8 MPa m1/2 measured for the TPCVD films (Liao et al. 2005). While not yet proven, a reasonable hypothesis is that the exceptional fracture toughness of the HPPD films is due to the effect of nanoparticles in the film arresting crack propagation in the vapor-deposited matrix.
In summary, HPPD is a one-step process that sprays nanoparticles onto a surface downstream of their synthesis in a plasma. Recent investigations demonstrate that film growth occurs by both nanoparticle impact and CVD. This co-deposition may potentially produce films with mechanical properties and wear behavior that are superior to either method alone. The vapor-deposited matrix provides interfacial material that reduces the films’ porosity, while the presence of nanoparticles suppresses crack propagation. Experiments to date have involved stationary substrates, for which the films are spatially nonuniform with respect to the relative proportions of the material grown by nanoparticle impact versus vapor deposition. A substrate rotation system has recently been developed, and experiments now in progress will facilitate the growth of more uniform films. Also of interest will be to explore whether films can be deposited for which the nanoparticles and the vapor-deposited material are of different chemical composition, for example nanoparticles of a hard ceramic material in a vapor-deposited matrix of a softer, more ductile material.
Microfabrication with nanoparticle beams
Aerodynamic lenses were developed by McMurry and coworkers in the early-to-mid-1990s (Liu et al. 1995a, b). In an aerodynamic lens, an aerosol flows through a thin-plate orifice, contracting as it approaches the orifice and then re-expanding downstream of the orifice. The behavior of particles entrained in the gas depends mainly on their Stokes number, or ratio of a particle’s inertia to its viscous drag. Particles with Stokes numbers much smaller than unity are able to follow the rapid contraction and re-expansion, and thus follow the gas streamlines. Large-Stokes-number particles are unable to follow the flow, and follow trajectories that carry them toward bounding walls. Particles with intermediate Stokes numbers, roughly around unity, have enough inertia to deviate from gas streamlines as the flow re-expands, but not enough to deviate so far as to reach the opposite side of the streamline bundle on which they originated. Hence they are pushed toward the centerline of the flow. After passing through several lenses in series, such particles are focused to a tightly collimated beam, which exits through a critical orifice which terminates the lens assembly.
While the original motivation (and still the major use) for producing focused particle beams was for application in aerosol characterization instruments, we demonstrated in 2000 that such focused beams had an interesting potential as a tool for microfabrication (Di Fonzo et al. 2000; Rao et al. 2005). In particular, the focused beam can be used to “write” two-dimensional lines and patterns, and to construct three-dimensional structures, composed of nanoparticles. For our typical conditions, beam widths equal a few tens of μm, and thus this technology does not compete with methods such as nanoxerography (Barry et al. 2003) that produce sub-micron line widths. Rather, the beam dimensions are suitable for integration of focused nanoparticle beams into the fabrication of microelectromechanical systems (MEMS), in particular where an advantage could be conferred by incorporating structures composed of nanoparticles, that would potentially exhibit the unique properties possessed by nanoparticles—mechanical, optical, magnetic, etc. In practical implementation, aerodynamic lenses could be interfaced between a gas phase nanoparticle synthesis reactor and a microfabrication chamber. Alternatively, the lenses could be decoupled from the synthesis process, and instead be fed by aerosolized nanoparticles produced by any method.
Some properties of nanoparticles, e.g. photoluminescence, depend on quantum confinement effects that, depending on the material, are important only for particle diameters in the sub-10-nm regime. Focusing such small particles with aerodynamic lenses is difficult because Brownian diffusion broadens the beam. As particle diffusion coefficients scale inversely on the square of particle diameter, this effect becomes increasingly severe as particle size is reduced. Recent efforts have focused on developing design guidelines for aerodynamic lens systems that consider diffusion. This work has resulted in guidelines for focusing unit-density spherical particles as small as 3 nm in diameter (Wang et al. 2005a, c).
Coating individual nanoparticles by photoinduced chemical vapor deposition
For many applications it is desirable to coat, or otherwise chemically modify, the surfaces of nanoparticles. In some cases a coating is necessary to passivate the particle surfaces, for example to prevent oxidation, while in other cases the coating may impart a specific functionality, for example by attaching substances that bind to specific sites in the human body for medical applications.
While much work on coating nanoparticles has involved the use of colloids, there are obvious advantages, for the case of nanoparticles synthesized in the gas phase, to utilize a gas-phase coating process, in which the core nanoparticles exiting the synthesis reactor flow directly into a nanoparticle coating reactor. A variety of gas-phase methods have been employed for coating nanoparticles, including heated flow tubes, flames and plasmas.
Recently we developed a new approach to coating nanoparticles that utilizes photo-CVD (Zhang et al. 2008). Potential advantages of photo-CVD include that it can be conducted at room temperature and atmospheric pressure (or at other temperature and pressure, if desired), and that it utilizes excimer lamps, which are simple to use, available in both cylindrical and planar configurations, and already used in industry for growing coatings by photo-CVD onto macroscopic substrates (Kogelschatz et al. 2000).
Integrating multiple nanoparticle processing steps
Utilization of nanoparticles in manufacturing will require the development of innovative processes that go beyond nanoparticle synthesis to manipulate and control the nanoparticles, for example by their controlled deposition onto surfaces or by chemically modifying the surfaces of individual nanoparticles. For cases where nanoparticle synthesis occurs in the gas phase, downstream processing in the gas phase offers obvious advantages compared to collecting the particles and subsequently processing them in liquid solution or other method. This paper discusses three specific nanoparticle processing technologies: hypersonic plasma particle deposition, which spray-coats nanoparticles onto surfaces; focused nanoparticle beam deposition, which allows construction of MEMS-scale lines, patterns and objects composed of nanoparticles; and photo-CVD nanoparticle coating, which provides passivation or functionalization of nanoparticle surfaces in a convenient, room-temperature environment. Current research on HPPD is focused on growing films simultaneously by nanoparticle impact and CVD, tailoring the coating nanostructure so as to achieve optimum mechanical properties. One can also envision the integration of focused nanoparticle beam deposition and photo-CVD nanoparticle coating, so that nanoparticle synthesis, coating and focused deposition can be accomplished in a single flow stream.
The author acknowledges the assistance of A. Beaber, who contributed recent unpublished results on the HPPD process, shown in Figures 2 and 3 and discussed in the accompanying text.
This research was partially supported by the National Science Foundation under grants CTS-0506748 and CBET-0730184, by the Army Research Office under grant DAAD-190110503, and by the Minnesota Supercomputing Institute. Use of facilities of the Minnesota Nanotechnology Cluster (MINTEC) is gratefully acknowledged.