Observation of Superaggregates from a Reversed Gravity Low-Sooting Flame

Abstract

Copyright 2012 American Association for Aerosol Research

Superaggregates (SAs) are formed due to late stage aggregation occurring near the gelling point of an aggregating system (Dhaubhadel et al. 2006). Percolation theory best describes their formation mechanism (Sorensen and Chakrabarti 2011). SAs have a long-range overall structure described by a fractal dimension D_f ≈ 2.6 and a short-range local structure described by D_f ≈ 1.8 (Sorensen et al. 1998; Sorensen and Chakrabarti 2011). Recently, based on the mechanism of SA formation in an aggregating system, Dhaubhadel and co-workers (Dhaubhadel et al. 2007) carried out enclosed-chamber aerosol aggregation to create “aerosol gels”—a novel material similar in properties to aerogels (Akimov 2003; Lerner 2004). Flame synthesis—which has been shown to be a cost-effective, scalable, and versatile route for large-scale production of nanoscale materials (Ulrich 1984)—is yet to be established as a viable method for production of aerosol gels. An upward-pointing flame system cannot facilitate steady and consistent formation of SAs in its product outflow upstream because of the aggregate fragmentation effects associated with buoyancy-driven flickering and shear stress of the flame's “tipping” front (Shaddix and Smyth 1996).

Here, we report our observations of steady outflow of SAs from a nontipping, reversed gravity (-g, up-side-down configuration) laminar methane–air diffusion flame. Figure 1 shows a schematic drawing of our up-side-down diffusion burner that is fed with methane through the inner tube and coflow air through the outer tubes, as in a Burke–Schumann arrangement (Stipe et al. 2005; Kirchstetter and Novakov 2007). The flame is cylindrical in shape, and the tip of the flame remains open. No flame flickering was observed past a few seconds after ignition. A portion of the exhaust was extracted using an ejector pump through the downstream port. Particle flow exiting the ejector pump was split isokinetically for characterization of particle morphology through two parallel setups—a large-angle static light scattering (SLS) setup (Kim et al. 2006), and a scanning electron microscopy (SEM) sampling unit (Chakrabarty et al. 2009). Additional information regarding the experimental setup and analysis methods used in this study can be found in the accompanying supplemental files.*

FIG. 1 Left, schematic diagram of the experimental setup used for aggregate generation and characterization. Right, photograph of the up-side-down diffusion flame. (Color figure available online).

SEM image of a typical SA observed during this study is shown in Figure 2a. The partial fragmentation of the fragile network SA seen in Figure 2 could be due to the effects of impaction during the sampling process. To confirm the morphology of SAs to be consistent with previously reported observations (Chakrabarty et al. 2006; Kim et al. 2006), we quantitatively analyzed the SA micrographs using the perimeter method image analysis technique (Dhaubhadel et al. 2006; Chakrabarty et al. 2011). Analyses were performed on approximately 25 micrographs for investigating SA morphology. Past experimental and simulation studies (Jullien et al. 1994; Dhaubhadel et al. 2006) have confirmed the perimeter fractal dimension D_p of an SA to be well defined and related to its mass fractal dimension D_f using the following empirical relationship:

FIG. 2 (a) SEM image of a typical superaggregate observed in the outflow of the up-side-down diffusion flame used in our study. (b) Perimeter fractal analysis of part (a). The negative perimeter analysis slope equals the perimeter fractal dimension, D_p = 1.4. The mass fractal dimension, D_f is found via Equation (1a) to be 2.46.

Figure 2b shows the value of D_p measured using the perimeter method for the SA in Figure 2a. The SAs observed in this study ranged in length scales between 2 and 10 μm, and yielded D_f = 2.5 ± 0.12 from Equation (1a). Our measured values of D_f imply universality in SA morphologies as predicted by the percolation theory. They are consistent with previous results for gelation in simulations and experiments (Fry et al. 2004; Dhaubhadel et al. 2006).

Using the large-angle SLS setup, we quantitatively characterized the morphology of aggregates, which constituted the SAs. Our SLS apparatus facilitated measurements of aggregate morphologies with length scales between 150 nm and 1 μm. Aggregates constituting the repeating units of the SAs were calculated to exhibit D_f = 1.82 ± .05 (Figure 3 a,b). These values confirm the growth mechanism of the aggregates in the dilute regime of the flame system to be consistent with the diffusion limited cluster agglomeration (DLCA) mechanism (Jullien and Botet 1987), and its subsequent transition into the percolation regime.

Our observation of superaggregation of DLCA-grown aggregates in an up-side-down flame configuration is explained by the following hypothesis. A “stagnation plane” is formed at the nontipping, lower end of the diffusion flame which confines the DLCA-grown D_f ≈ 1.8 aggregates. By enhancing the aggregate residence time within the stagnation plane, aggregates evolve from dilute to percolation regime to form SAs. Figure 4b depicts a schematic of this hypothesis. When a flame is inverted, the effects of buoyancy on the flow-field are quite drastic. Buoyancy provides an upstream feedback mechanism that balances the opposing flow forces to create a zero-acceleration plane of stagnation downstream from the flame zone (Bédat and Cheng 1996). Creation of this plane pushes the flame front wide open and makes it curl outward. This mechanism is contrary to what takes place in a normal gravity upright flame, where buoyant forces accelerate the flow forces, causing a narrowing of the flame front to tip (Figure 4a). For a low-sooting fuel such as methane, a narrowing and accelerating flame front cannot provide the combination of high number density and aggregating time for aggregates to cross over into the percolation regime, while we have shown here that this is possible for the reversed gravity flame. Additionally, buoyant forces in upright flames cause flickering, which results in aggregate structure fragmentation.

FIG. 3 (a) SEM image of a single aggregate. These aggregates constitute the repeating units of the superaggregates observed in this study. (b) Large-angle structure analysis measured an ensemble of the aggregates in part (a) to have a negative slope (D_f ) ≈ 1.8, consistent with the DLCA growth mechanism.

FIG. 4 Schematic depiction of our hypothesis explaining superaggregate formation in an up-side-down diffusion flame system. (a) In an upward-rising diffusion flame system, a narrowing and accelerating flame front cannot provide the threshold aggregating time for DLCA-grown D_f ≈ 1.8 aggregates to cross over into the percolation regime. (b) In an up-side-down flame system, a “stagnation plane” is formed at the nontipping end of the flame, which confines the aggregates and their residence time within the plane to facilitate their evolution from dilute to percolation regime to form superaggregates.

Further investigation of our hypothesis and/or any other mechanism that might be involved in the production of SAs from an inverted diffusion flame is needed. Future research needs to be directed toward characterization and optimization of our inverted flame system using different fuels and operating conditions (e.g., temperature, flow parameters, equivalent ratios), and toward identifying its limitations. In situ aerosol growth analysis, using a small-angle SLS apparatus at different heights in the flame (Beaucage et al. 2004), might provide further insights into the growth kinetics of gelation, which is currently an unresolved question (Sorensen and Chakrabarti 2011).

Acknowledgments

*Supplementary materials are available for this article. Go to the publisher's online edition of Aerosol Science and Technology to view the free supplementary files.

This material is based upon work supported by NASA EPSCoR under Cooperative Agreement No. NNX10AR89A, by NASA ROSES under grant NNX11AB79G and by the Desert Research Institute.