Separation of coal fly ash using continuous steric ... - ACS Publications

Separation of Coal Fly Ash Using Continuous Steric Field-Flow Fractionation. Mark R. Schure/ Marcus N. Myers, Karin D. Caldwell, Colleen Byron, Kin Pe...
1 downloads 0 Views 3MB Size
Environ. Scl. Technol. 1905, 19, 686-689

Separation of Coal Fly Ash Using Continuous Steric Field-Flow Fractionation Mark R. Schure,+ Marcus N. Myers, Karin D. Caldwell, Colleen Byron, Kin Peng Chan, and J. Calvin Glddlngs"

Department of Chemistry, University of Utah, Salt Lake City, Utah 841 12

rn The principles and advantages of continuous steric field-flow fractionation (CStFFF) for the size separation of complex environmental samples is briefly reviewed. Following prefractionation according to magnetic and density properties, CStFFF was successfully applied to four well-characterized fly ash samples. The mode size of ash particles in different fractions followed the theoretical prediction. Fractionation was observed to be best for samples that were prefractionated. Introduction The fractionation of naturally occurring particulate materials (e.g., soils) and anthropogenically produced particles (e.g., coal fly ash), which dominate important parts of our environment, is a difficult task for which most present methodology is inadequate. Such devices as particle sieves, although time tested and well-defined (1, 2), are only crude separation tools. One may, of course, rely on sample characterization by microscopic methods. Although this mode of analysis does give information as to particle shape and size, information on density and chemical composition is lacking. A prestep of density separation in a centrifuge is usually required before one can imply composition. For the most thorough analysis, however, it is desirable to isolate particle fractions for further examination. This requires some type of separation method. Furthermore, the separation device should preserve the sample integrity for accurate morphological characterization. This requirement tends to diminish the value of sieves where, for the specific case of coal fly ash, it has been suggested (3) that particles may be abraded by the mesh or destroyed by impaction from other particles. Although frequently used, separations involving the cascade impactor ( 4 ) suffer from a number of problems such as low resolution, material losses, and particle destruction upon impaction. These drawbacks have been discussed by Natusch and Wallace (5) and by Smith et al. (6). The method of air elutriation has been used successfully in size-separating coal fly ash (6);however, even in elutriation the particle collection involves impaction on the wall of the chamber. In this paper, we will examine the separation of coal fly ash using a scheme incorporating simple density separation followed by fractionation and preconcentration by continuous steric field-flow fractionation (CStFFF). This relatively new technique is expected to be a gentle and thorough technique for environmental particles. The technique has been used for the fractionation of simple substances such as glass beads; the separation of coal fly ash offers a chance to evaluate the technique under more challenging conditions. The primary goal of this paper is to test and expand the utility of CStFFF under these more demanding circumstances. Principles of CStFFF The operational and theoretical principles of CStFFF have been described in our initial publication on this t Present address: Digital Equipment Corp., Marlborough, MA 01752.

686

Envlron. Sci. Technol., Vol. 19, No. 8, 1985

technique (7); these principles will be reviewed here briefly for the purpose of clarity. Steric FFF is a limiting form of general FFF methodology (8-10) in which particles are rolled or tumbled along the smooth surface of a narrow ribbonlike channel by a flow stream (11). In the case of continuous steric FFF, sedimentation is coupled into the system by tilting the channel some angle $ from the vertical as shown in Figure 1. Clearly, the smaller the angle $, the less the involvement of steric effects; in the limit of $ = 0, pure sedimentation coupled with flow will enact the separation. The steric element is anticipated to be of high importance in separating particles according to size and to some extent shape (7). As illustrated in Figure 1,the angle of deflection 0 will differ for particles of different size. This angle is a function of the two particle velocities: U, along flow axis z and Uy along sedimentation axis y. We have tan 0 = U y / U z (1) where Uy and U, have been identified as ( 7 ) U, = 3y d(u)/w

(2)

and

Uy = d2Apg COS $/(lSqP) (3) where d is the particle diameter, y is a hydrodynamic factor of order unity, ( u ) is the average flow velocity, w is the channel thickness, and Ap is the difference between particle density and fluid density. In addition, g, 7, and are the gravitational acceleration constant, the fluid viscosity, and a correction factor (>1)allowing for asphericity and the increased drag of a particle being sedimented along a solid surface. " Combining eq 2 and 3 with the relationship Z = b/tan 8 (b is the channel breadth and Z is the distance down the channel at which a particle enters a collection trough), we get 5 4 d u )MrP) Z= (4) dwApg cos $ This equation predicts that position Z will be inversely proportional to particle diameter d. Also, Z will be expected to depend on particle shape through factor yP and particle density through Ap. Experimental Section Apparatus. The apparatus used in these experiments is similar to that described in ref 7 with the following exceptions and specific details. The tilt angle 4 was 20 deg. The spacer was constructed of 0.254-mm mylar with eight sampling ports. The arrangement of the ports is indicated by the block diagram of the unit shown schematically in Figure 2. The flow rates were 0.053 mL/min through pump 1 and 3.97 mL/min through pump 2. The average linear flow velocity ( u ) through the channel was 0.426 cm/s. Three in-house built ultrasonic generators were fitted to the lower glass wall so that particles that appeared to stick to the wall could be pulsed loose. The pulse time was approximately 0.1 s, and the repetition rate was approximately 2 s for each generator.

0013-936X/85/09 19-0686$01.50/0

0 1985 American Chemical Society

70

Flgure 1. Shematic diagram of ths CStFFF system. Actual unit has triangular end pieces as shown in Figure 2.

I

01 0

I

I

I

I

2

3

I

I

4

5

( l / t ) x 1 0 0 (ern-') Flgure 3. Plot of particle mode size d, vs. l l Z for the high density Corrette ash sample, 2.3 Q p Q 3.0. CR

= =

Flgure 2. Schematic of the CStFFF apparatus, including sample beaker (SB). magnetic stirrer (MS).pumps (PI-P3). Carrier resewoir (CR). collection tubes (Cl-8). Rrst overflow (C9). and second overflow (C10). Arrows show direction of flow.

C a r r i e r Liquid. The choice of carrier used here is critical. In previous attempts to fractionate coal fly ash, we have found that water causes irreversible aggregation and leaching. The addition of surfactant in large quantities proved only marginally helpful in avoiding aggregation. The surfactant addition also proved to be a nuisance since it tended to gel the particles when attempts were made to remove the fluid. The leaching is to be expected on the basis of the chemical composition (mainly calcium oxides and sulfates) of the fly ash. Methanol, like water, tends t o leach particles. Heavy hydrocarbons do not cause leaching but tend to cause aggregation and furthermore hinder the separation due to high viscosity. It was found through trial and error that the short exposures to butanol which the fractionation entails did not leach the particles and caused a minimum of aggregation. A 24-h Soxhlet extraction of the Corrette ash with butanol showed no soluble components, as judged by thermal vaporization of the extract up to 300 OC in conjunction with time-resolved mass spectroscopy. In addition, particles can be completely freed of butanol by evaporation. Fly Ash Samples. A number of different fly ash samples were employed in this study; these will be referred to as the Corrette, Huntington, and “Western” ash samples. The Corrette ash was collected from the haghouse of a large coal-fired power plant in Montana known as the Corrette power plant. This power plant used low sulfur, subbituminous, Montana Rosebud coal. The ash has been characterized in a number of studies (12,13). The Western ash was collected from the stack (after the electrostatic precipitator) of a 750-MW coal-fired power plant. This power plant burned low sulfur, high ash, high moisture coal. This ash has also been the subject of a number of previously published studies; details of the collection and characterization are given in ref 3 and 14. The ash had been prefractionated by aerodynamic size in

a particle impaction device (3) from which the cut 2 fraction (14) was generously donated by Dr. Gerald L. Fisher. The Huntington ash was collected from the stage 14 bin of a 14stage electrostatic precipitator. A wet scrubber was used prior to the electrostatic precipitator for removal of SO2. The bituminous coal burned in this plant is low in sulfur (0.48%). The plant is located near Huntington, UT, and the sample was supplied by Utah Power and Light Co. Sample Prefractionation. Initial efforts at characterizing bulk coal fly ash revealed the complexity of the samples. In an effort to make the analysis more specific, the magnetic particles were first removed from most samples by the use of a permanent magnet. For the Corrette ash, the nonmagnetic ash material was then separated into density fractions by using a small laboratory centrifuge. The fluids used were mixtures of carbon tetrachloride and diiodomethane mixed to give densities of 1.8, 2.3, and 3.0 g/mL. The most dense fraction invariably contained magnetic particles which had been missed in the first stage of prefractionation. Microscopy. Upon termination of the CStFFF experiments (aU of which were run for 24 h), a small amount of ash was removed from the sample tube and placed on a microscope slide. All subsequent light microscopy was done by using an Olympus Model 7-C121 microscope with phase optics. Photographs were taken with an Olympus Om-1 SLR 35-mm camera.

Results and Discussion Table I summarizes the overall results of the CStFFF fractionation for the four samples studied in terms of the mode particle sized,. In the case of nonspherical particles, the size used was the average of the longest and shortest dimensions as measured by light microscopy. The table shows clearly that fractionation by size has been realized. Fraction 8, which is the last fraction collected prior to the overflow, is frequently seen to contain anomalously large particles in addition to spheres of the predicted, small diameter. These large particles are highly irregular and have a “lacy” appearance. In order to test the fractionation process quantitatively, we have plotted d, as a function of 1/Z. According to eq 4, a plot of this type should form a straight line. Figure 3 shows this plot for the Corrette ash prefractionated into the density range 2.3-3.0. The fit is seen to be satisfactory, yielding reasonable 70 values in the range 0.94-1.38 and a correlation coefficient of 0.939. The other Corrette ash sample yielded a correlation coefficient of 0.979. The correlation coefficient for the Western ash sample was 0.935, and for the Huntington ash it was 0.851; these somewhat lower values are perhaps due to the lack of a Environ. Scl. Technol., Vol. 19. No. 8. 1985 687

D

C

Flgure 4. Photomicrographsof fractions of tha high density (2.3 d p d 3.0) Corrette ash at port 2 (A). port 3 (E). port 4 (C). and port 8 (D).

Table 1. Particle Mode Size d. (am)

Corrette, Corrette, 2.3 5 1.8 5 cm p 5 3.0 p 5 2.3

Z,

port no. 1 2 3 4 5 6 7 8 first overflow second overflow

12 22 32 42 52 62 72 82

b 53 45 30 20 30 20 15.3810 5

b 60 40 30 25 25 15 22 15 6.3

Huntingtan

140 70 115 70 50 40, 100 30 17 19 9,30

Western

loo, 4ff

60 52 50 30 10 5 20 5 3

'Two sizes indicate two discrete particle types. 'No particles observed.

prior density fractionation. We also note that the slope of the line for the Corrette samples was greatest for the ash of least density as predicted by eq 4;the low density Corrette ash displayed a slope of 1240 r m c m , and the higher density fraction had a slope of 1116 pm.cm. The slopes of the other ash samples tended t o be higher (Western, 1220; Huntington, 1510), probably because of the lack of density fractionation and the likely content of low density particles. Figure 4 shows the fractionation of the high density Corrette ash as viewed by optical microscopy. Fairly well-defined samples of ash were obtained with respect to particle size. However, we note that ash collected a t port 8 contains a number of large anomalous particles. The shapes of the darker (probably carbonaceous) particles appear to be more irregular than the typical glassy spheres shown in abundance in Figure 4D. It is not clear why the large dark particles appear in this fraction in view of the prior density separation unless the latter separation was incomplete. 808

Environ. Sci. Technol.. Voi. 19, No. 8, 1985

A

B

Flgure 5. Fractions of Huntington ash from port 2 (A) and port 4 (6).

The Huntington ash contained almost none of the magnetic material typically found in coal fly ash. In addition, very few glassy spheres were identified. The density separation technique described previously was attempted on this ash and revealed a virtual absence of particles with density greater than 2.3. Presumably these particles were taken out of the flue stream by the water scrubber and electrostatic precipitator systems. The density separation did reveal a large amount of carbonaceous material which was found t o have a density less than 1.8. The Huntington ash was run with only the magnetic fraction removed. Two CStFFF fractions of the Huntington ash are shown in Figure 5. The black irregularly shaped particles are of carbonaceous origin. A small amount of attraction is exhibited by the carbonaceous particles in solution; this explains the appearance of aggregation shown in Figure 5. The particles do separate by mechanical action, and the appearance of aggregation is presumed to occur in the microscopic evaluation only. Figure 5 shows the fractions to he composed of two distinct types of particles-black irregular particles and opaque glassy particles. If the Huntington sample had

H40fm

C

B Flgure 6.

Fractions of

Western

ash at port 1 (A), port 2 (E). and the first overflow (C).

been density prefractionated, the resolution of these particle types would undoubtedly he more complete. Since the Western ash fractionation was done with a sample prefractionated by aerodynamic size, some degree of size discrimination with respect to bulk ash is already inherent in the sample. This sample had previously been found to have a log-normal particle size distribution with a count median diameter of 2.58 r m (3, 14). The fines would not be resolved by CStFFF in ita current implementation and were expected to end u p in the overflow, whereas the coarse particles were expected to fractionate according to size. Photomicrographs of fractions of Western ash obtained by the CStFFF technique are shown in Figure 6 for ports 1, 2, and the first overflow. The magnetic and carbonaceous fractions are almost exclusively confined to port 1 (this sample was not magnetically prefractionated), although some smaller magnetic material was found in port 6. The second port (Figure 6B) shows a well-defined sample of predominately glassy spheres mixed with irregular lacy ( I 4 ) particles. Of particular interest is the rather clear fraction of the smaller particles obtained from the first overflow and shown in Figure 6A. The aggregates shown in Figure 6C can be dispersed by mechanical force and are therefore assumed to be artifacts of microscopy. Conclusions From the results presented above, it can be seen that the technique of continuous steric field-flow fractionation has considerable promise as a preparation tool for particle study. The technique can undoubtedly be improved with additional work and possibly extended into the small particle (