Porous cobalt spheres for high temperature gradient magnetically assisted fluidized beds

Article abstract of DOI:10.1016/S0025-5408(02)01072-3

Porous metallic cobalt spheres have been prepared as high temperature capable media for employment in gradient magnetically assisted fluidization and filtration technologies. Cobalt impregnated alginate beads are first formed by extrusion of an aqueous suspension of Co₃O₄ into a Co(II) chloride solution. The organic polymer is thermally decomposed yielding cobalt oxide spheres, followed by reduction to the metallic state, and densification. Cobalt beads have been produced with porosities ranging between 10 and 50%, depending upon sintering conditions. The product media have been characterized by scanning electron microscopy (SEM), nitrogen adsorption porosimetry, and vibrating sample magnetometry.

Full text of DOI:10.1016/S0025-5408(02)01072-3

Materials Research Bulletin 38 (2003) 395±407
Porous cobalt spheres for high temperature gradient
magnetically assisted ¯uidized beds
James E. Atwatera,*, James R. Aksea, Goran N. Jovanovicb,
Richard R. Wheeler Jr.a, Thana Sornchamnib
aUMPQUA Research Company, P.O. Box 609, Myrtle Creek, OR 97457, USA
bDepartment of Chemical Engineering, Oregon State University, Corvallis, OR 97331, USA
Received 21 October 2002; accepted 11 December 2002
Abstract
Porous metallic cobalt spheres have been prepared as high temperature capable media for employment in
gradient magnetically assisted ¯uidization and ®ltration technologies. Cobalt impregnated alginate beads are ®rst
formed by extrusion of an aqueous suspension of Co3O4 into a Co(II) chloride solution. The organic polymer is
thermally decomposed yielding cobalt oxide spheres, followed by reduction to the metallic state, and
densi®cation. Cobalt beads have been produced with porosities ranging between 10 and 50%, depending upon
sintering conditions. The product media have been characterized by scanning electron microscopy (SEM),
nitrogen adsorption porosimetry, and vibrating sample magnetometry.
# 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Magnetic materials; Metals; Chemical synthesis
1. Introduction
Fluidized bed contacting is a very effective method for conducting gas±solid and liquid±solid reactions,
particularly those which are limited by mass transfer rates [1,2]. In ¯uidization, a dynamic balance is
produced between drag, buoyancy, and gravitational forces. Maximum gas or liquid ¯ow rates within
¯uidized bed reactors are limited by the ¯uid properties and also by the size and density distributions of
the granular ¯uidization media. For a given gas±solid or liquid±solid system, higher rates of mass transfer
can be achieved using magnetically stabilized ¯uidized bed (MSFB) methods which were pioneered by
Rosensweig et al. [3±8]. In MSFB technology, magnetically susceptible solids are ¯uidized within a
uniform magnetic ®eld oriented parallel to the ¯ow. Here, the magnetic forces between the particles
*
Corresponding author. Tel.: ‡1-541-863-7770; fax: ‡1-541-863-7775.
E-mail address: jatwater@urcmail.net (J.E. Atwater).
0025-5408/03/$ ± see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0025-5408(02)01072-3

396
J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
Fig. 1. Forces acting upon a ferromagnetic particle in zero gravity, (a) under a magnetic ®eld gradient and (b) without an
applied magnetic ®eld. FdÐdrag force, FmÐmagnetic force.
augment the apparent size of the ¯uidization particles and thereby allow higher ¯uid ¯ow rates under
equivalent ¯uidization conditions. In recent years, ¯uidization phenomena [9±15], mass transfer rates
[16,17], and ¯uid dynamics [18±20] of MSFB processes have been extensively studied.
Our research interest is in the application of magnetic forces acting upon suf®ciently susceptible
media to achieve both ¯uidization and ®ltration [21] in microgravity and hypogravity environments to
facilitate solid waste gasi®cation processes supporting NASA's long-term regenerative life support
requirements for future extended duration manned space mission scenarios such as a Lunar base, Mars
transit, and Mars base [22]. In the absence of a gravitational restoring force, conventional ¯uidization is
not possible. However, using modi®ed MSFB based methods in which magnetic ®eld gradients are
employed to produce additional force, liquid±solid ¯uidization can be achieved, as shown in Fig. 1.
Using gradient magnetically assisted ¯uidized bed (GMAFB) methods this has been demonstrated in
microgravity during two series of ¯ight experiments aboard NASA's KC-135 aircraft [23].
In traditional MSFB systems, ferromagnetic particles are magnetized within a uniform magnetic ®eld
as in (1)
M ˆ wmH (1)
where the vectors M and H represent magnetization and magnetic ®eld intensity, respectively, and
wm is the magnetic susceptibility of the medium. The resulting magnetic ¯ux density (B) is given
by,
B ˆ m0ꢀH ‡ M† ˆ m0ꢀH ‡ wmH† ˆ mH
(2)
where m0 is the magnetic permeability of free space and m ˆ m0ꢀ1 ‡ wm†. The pole strength, p, of a
magnetized ferromagnetic particle is related to the ¯ux density by
m p
4p r2
^
i
B ˆ
(3)
^
where i is a unit vector in the direction of the ¯ux density. In the uniform magnetic ®eld, forces are
produced between the particles as in (4).
m p1p2
4p r2
^
i
F ˆ
(4)

J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
397
In (3) r is the distance from the pole at which B is measured, while in (4) r is the distance separating the
two poles. In the gradient magnetically assisted ¯uidized bed (GMAFB) additional forces are produced
through the application of a non-uniform magnetic ®eld.
F ˆ rꢀM Á B† (5)
Using these additional forces, granular ferromagnetic media can be either ¯uidized or consolidated into
a packed bed in the absence of gravity.
Gasi®cation reactors necessarily require high temperature operating conditions, typically between
400 and 700 8C [24±26]. Because magnetic susceptibility is a strong function of temperature, it is
important to utilize ferromagnetic media with a suitably high Curie temperature (Tc). This requirement
eliminates nickel (Tc ˆ 358 8C) and the ceramic ferrites from contention. While the suitability of iron
(Tc ˆ 770 8C) is marginal, the high temperature capability of cobalt (Tc ˆ 1120 8C) renders it clearly
superior to most other candidate materials.
In the current research, cobalt oxide impregnated cobalt alginate spheres were ®rst prepared via an
alginate gelation methodology [27±30]. The organic polymer was then thermally decomposed in an
inert environment, resulting in cobalt oxide beads which were reduced to the metallic state under
hydrogen. Densi®cation of the spheres was achieved through further heat treatment. The relationships
between sintering temperature, surface area, porosity, pore size distribution, and magnetic properties of
the resulting materials were characterized by optical microscopy, SEM, nitrogen adsorption
porosimetry, and vibrating sample magnetometry. The results of a preliminary investigation into the
production of both metallic cobalt beads, and cobalt impregnated barium titanate beads have been
reported previously [31].
2. Experimental
Spherical cobalt beads were prepared utilizing a four step process, including: (a) production of cobalt
oxide impregnated polymeric spheres by alginate gelation, (b) oxidative and pyrolytic decomposition of
the organic polymer, yielding cobalt oxide spheres, (c) reduction of cobalt to the zero-valent state, and
(d) sintering of the metallic cobalt. An aqueous slurry was prepared containing 1.1% medium viscosity
sodium alginate, 0.14% citric acid, 0.04% tannic acid, 2.45% ammonium hydroxide, and 13.75% cobalt
oxide (Co3O4). The slurry was well mixed so that Co3O4 particles did not settle signi®cantly during the
gelation step. Droplets of the slurry were extruded through a 1 mm i.d. ori®ce, and allowed to fall into
an aqueous 11.9% cobalt(II) chloride solution, where polymerization occurred via divalent cation
promoted cross-linking reactions. The cobalt oxide impregnated cobalt alginate beads were aged in the
CoCl2 solution for 12 h, then thoroughly rinsed with deionized water and dried. This produced
spherically symmetrical beads varying between 2 and 3 mm in diameter.
To decompose the alginate binder, the beads were heated under ambient air to 550 8C at a rate of
2 8C/min, and held at this temperature for 4 h in a tube furnace. The binder-free cobalt oxide beads
were then reduced to cobalt metal using a 5% mixture of H2 in N2 at 550 8C. Following the complete
reduction of cobalt oxide to cobalt metal, sintering runs to endpoint temperatures of 700, 800, 900, and
1000 8C were carried out to determine the thermal requirements for densi®cation. In each case, the
beads were heated from 550 8C to the ®nal temperature under 5% hydrogen in nitrogen at 10 8C minÀ1.
The beads were held at the ®nal temperature for 4 h, and then cooled to ambient temperature.

398
J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
Grain sizes and porosities of the cobalt spheres were estimated by optical microscopy of polished
cross-sections, using a linear point counting methodology. BET surface areas and pore size distributions
were determined by nitrogen adsorption porosimetry. High resolution images of polished cross-sections
and surfaces were acquired using a JEOL model JSM6300 XV scanning electron microscope. Magnetic
properties of the material were characterized using a Lake Shore Model 7307 vibrating sample
magnetometer.
3. Results and discussion
The time required for completion of the reduction step is strongly dependent on the gas ¯ow rate. The
cobalt beads are then densi®ed in a sintering step. Sintering temperatures between 700 and 1000 8C were
investigated. Most typically, sintering was accomplished in 240 min at a lower reducing gas ¯ow rate to
minimize temperature gradients within the tube furnace. The combined process produces highly porous
and mechanically strong cobalt media. Numerous technical hurdles were overcome in the development
of this process. To produce more than a few grams of cobalt beads at a time, improvements in the control
of the alginate slurry rheology, strength of dried gelled beads, removal of alginate binder during thermal
processing, and thermal process scale-up were required. The alginate-cobalt oxide formula was varied
until stable and physically strong beads of ꢁ1 mm diameter resulted. The reduction of cobalt oxide to
cobalt metal required signi®cant modi®cation of the reaction conditions from those of the initial small
scale experiments. With larger batches (i.e. 50±100 g), the reduction of cobalt oxide was limited by the
availability of hydrogen in the reducing gas, since the maximum safe hydrogen content of the nitrogen
process gas is 5%. For large batches, hydrogen became rapidly depleted by reduction reactions during
the ramp up to sintering temperature, so that sintering temperatures were reached before complete
Fig. 2. Sintering temperature versus porosity of cobalt beads.

J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
399
reduction of the cobalt oxide was achieved. This situation resulted in inconsistent sintering and
densi®cation within each batch. At this point, an attempt to substitute cobalt metal for cobalt oxide was
made since cobalt metal does not require a reduction step. This proved unsuccessful due to the reaction
of cobalt with water to produce cobalt oxide and a mixture of hydrogen and oxygen. Consequently, the
Fig. 3. Pore size distribution for cobalt beads sintered at 700 8C (top) and 900 8C (bottom).

400
J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
original cobalt oxide reduction process was altered by providing longer reaction times and higher
hydrogen/nitrogen gas ¯ows at 550 8C to completely reduce cobalt oxide prior to the sintering step.
Following the complete reduction of cobalt oxide to cobalt metal, sintering runs to endpoint
temperatures of 700, 800, 900, and 1000 8C were carried out to determine the thermal requirements for
densi®cation. In each case, the beads were heated from 550 8C to the ®nal temperature under 5%
hydrogen in nitrogen at 10 8C minÀ1. The beads were held at the ®nal temperature for 4 h, and then
Fig. 4. Surface (top) and cross-section (bottom) SEM images of cobalt bead sintered at 700 8C.

J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
401
cooled to ambient temperature at the maximum furnace cooling rate. Densities and porosities of the
resulting cobalt beads were estimated utilizing a linear point count methodology. Bulk densities were
49.6, 58.4, 68.8, and 87.0% of theoretical for samples processed at 700, 800, 900, and 1000 8C,
corresponding to porosities of 50.4, 41.6, 31.2, and 13.0%, respectively. The relationship between
sintering temperature and bead porosity is shown in Fig. 2. Brunauer±Emmett±Teller (BET) surface
Fig. 5. Surface (top) and cross-section (bottom) SEM images of cobalt bead sintered at 800 8C.

402
J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
areas and pore size distributions for samples sintered at 700 and 900 8C were characterized by nitrogen
adsorption porosimetry. The resulting pore size distributions are shown in Fig. 3. A comparison of the
results for cobalt spheres prepared at the two temperatures indicates that the relative distributions of
pore diameters are similar but the absolute magnitudes are much greater for beads sintered at the lower
temperature. This is in agreement with the observed general decrease in porosity with increasing
Fig. 6. Surface (top) and cross-section (bottom) SEM images of cobalt bead sintered at 900 8C.

J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
403
sintering temperature. BET surface areas were 1.24 and 0.639 m2/g, for cobalt beads densi®ed at 700
and 900 8C, respectively.
Scanning electron photomicrographs taken at 2500Â magni®cation for both the cobalt bead surfaces
and polished cross-sections of samples ®red at these temperatures are shown in Figs. 4±7, respectively.
At the lowest sintering temperature, the reduction of the sub-micron cobalt oxide powder resulted in a
Fig. 7. Surface (top) and cross-section (bottom) SEM images of cobalt bead sintered at 1000 8C.

404
J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
Table 1
Summary of magnetic properties of beads sintered at temperatures between 700 and 1000 8C
T (8C)
Ms
Mr
Hc
A/m
emu/cm3
A/m
emu/cm3
A/m
Oe
700
800
6.4 Â 105
8.2 Â 105
7.1 Â 105
1.26 Â 106
6.4 Â 102
8.2 Â 102
7.1 Â 102
1.26 Â 103
4.46 Â 103
3.81 Â 103
2.29 Â 103
1.55 Â 102
4.46
3.81
2.29
0.155
5.47 Â 103
4.55 Â 103
4.45 Â 103
5.2 Â 102
68.7
57.2
55.9
6.53
900
1000
loosely connected metal compact with an average grain size of ꢁ0.5 mm. The higher thermal treatment
temperature of 800 8C increased the grain sizes to ꢁ1 mm. Both of these samples were mechanically
weak, and therefore, unsuitable for use in a gasi®cation reactor. At 900 8C, the grain size increased to
between 1.5 and 2 mm. Beads produced at this temperature were mechanically robust and relatively
dense (6.12 g/cm3). The grain size in the beads processed at 1000 8C increased to between 2 and 3 mm.
The magnetic properties of these beads were studied using a vibrating sample magnetometer. Test
results are summarized in Table 1. Room temperature magnetization±demagnetization curves for beads
sintered at 700, 800, 900, and 1000 8C are shown in Figs. 8±11, respectively. All cobalt beads exhibited
strong magnetic susceptibility and very little hysteresis, indicating low remnant magnetization (Mr).
Coercive ®eld strengths ranged between 6.53 and 68.7 Oe. The saturation volumetric magnetization
(Ms) increased from 647 emu/cm3 at the sintering temperature of 700 8C to 1261 emu/cm3 at 1000 8C,
Fig. 8. Magnetization±demagnetization curves for cobalt beads sintered at 700 8C.

J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
405
Fig. 9. Magnetization±demagnetization curves for cobalt beads sintered at 800 8C.
Fig. 10. Magnetization±demagnetization curves for cobalt beads sintered at 900 8C.

406
J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
Fig. 11. Magnetization±demagnetization curves for cobalt beads sintered at 1000 8C.
correlating with increased densi®cation. The beads prepared at 900 8C were anomalous in this respect,
showing somewhat lower saturation magnetization than would be expected based upon the metallic
cobalt content of the beads. However, these beads still showed very strong magnetic susceptibility.
Magnetization±demagnetization experiments using the beads prepared at 900 8C were repeated,
yielding equivalent results. Given the volumetric saturation magnetization of 1422 emu/cm3 for pure
cobalt metal with no porosity at 20 8C, it is evident that the ferromagnetic properties of the porous
cobalt beads produced in this study generally correlate with the bulk densities of the materials and their
cobalt contents on a volumetric basis.
Based upon these results, a sintering temperature of 900 8C was selected for the production of media
in quantities to support both laboratory and ¯ight experiments. Media prepared under these conditions
exhibit good mechanical strength and suf®cient porosity. In subsequent runs, after 4 h at 900 8C, a
reproducible density of 5.61 g/cm3 was achieved, with bead diameters varying between 1 and 1.2 mm.
These beads were mechanically strong and retained interconnected porosity. This combination provides
the physical robustness necessary for employment in a ¯uidized bed reactor at elevated temperature
with the surface area and porosity characteristics required for catalytic activity.
Acknowledgements
This work was funded by the Of®ce of Physical and Biological Research of the National Aeronautics
and Space Administration under grant NAG9-1181. The authors thank John W. Fisher at NASAÐAmes

J.E. Atwater et al. / Materials Research Bulletin 38 (2003) 395±407
407
Research Center for his helpful suggestions and support and Dr. David C. Johnson of the Materials
Science Institute at the University of Oregon for technical assistance.
References
[1] O. Levenspiel, Chemical Reaction Engineering, Wiley, New York, 1972.
[2] D. Kunii, O. Levenspiel, Fluidization Engineering, Butterworth-Heinemann, Boston, 1991.
[3] R.E. Rosensweig, Ferrohydrodynamics, Cambridge University Press, New York, 1985.
[4] R.E. Rosensweig, J.H. Siegell, W.K. Lee, T. Mikus, AIChE Symp. Ser. 77 (1981) 8.
[5] R.E. Rosensweig, Science 204 (1979) 57.
[6] R.E. Rosensweig, Ind. Eng. Chem. Fundam. 18 (1979) 260.
[7] R.E. Rosensweig, G.R. Jerauld, M. Zahn, in: O. Brulin, R.K.T. Hsieh (Eds.), Continuum Models of Discrete Systems 4,
North-Holland, New York, 1981, pp. 137±144.
[8] R.E. Rosensweig, M. Zahn, W.K. Lee, P.S. Hagan, in: R.E. Meyer (Ed.), Theory of Dispersed Multiphase Flow,
Academic Press, New York, 1983, pp. 359±384.
[9] M.A. Burns, D.J. Graves, Chem. Eng. Commun. 67 (1988) 315.
[10] L.M. Sajc, Z.R. Jovanovic, G.V. Novakovic, G.N. Jovanovic, R.D. Pesic, D.V. Vucovic, Trans. IChemE 72 (1994) 236.
[11] G.N. Jovanovic, L.M. Sajc, Z.R. Jovanovic, G.V. Novakovic, B. Kundacovic, B. Obradovic et al., J. Serb. Chem. Soc. 57
(1992) 345.
[12] G.N. Jovanovic, L. Sajc, Z.R. Jovanovic, B. Bugarski, G.V. Novakovic, J. Serb. Chem. Soc. 61 (1996) 319.
[13] T.T. Ames, R.M. Worden, Biotechnol. Prog. 13 (1997) 336.
[14] M.J. Rhodes, X.S. Wang, K.S. Gan, S. Phadtajaphan, Chem. Eng. Sci. 56 (2001) 5429.
[15] Y.-C. Lin, L.-P. Leu, Powder Technol. 120 (2001) 199.
[16] M. Franzreb, R. Hausmann, C. Hoffmann, W.H. Holl, React. Polym. 46 (2001) 247.
[17] M. Al-Mulhim, Enhancement of Mass Transfer Coefficient in a Magnetically Stabilized Liquid±Solid Fluidized Bed,
M.S. Thesis, Oregon State University, Corvallis, Oregon, 1995.
[18] C.-M. Chen, L.-P. Leu, Powder Technol. 117 (2001) 198.
[19] D. Bohm, B. Pittermann, Chem. Eng. Technol. 23 (2000) 309.
[20] L. Honerez, Fluid Dynamic Characteristics of a Magnetically Stabilized Liquid±Solid Fluidized Bed, M.S. Thesis,
Oregon State University, Corvallis, Oregon, 1994.
[21] T. Abbasov, Powder Technol. 115 (2001) 215.
[22] J.E. Atwater, J.R. Akse, J.L. DeHart, G.N. Jovanovic, T. Sornchamni, S. Yoo, J.W. Fisher, SAE Tech. Paper Ser. 1999-01-
2183 (1999) 1.
[23] T. Sornchamni, The Prediction of Voidage Distribution in a Non-uniform Magnetically Assisted Fluidized Bed: Theory
and Experiment, M.S. Thesis, Oregon State University, Corvallis, Oregon, 2000.
[24] G. Schuster, G. Loffler, K. Weigl, H. Hofbauer, Bioresource Technol. 77 (2001) 71.
[25] J.M. Encinar, F.J. Beltran, A. Ramiro, J.F. Gonzalez, Fuel Process. Technol. 55 (1998) 219.
[26] L. Garcia, M.L. Salvador, J. Aruzo, R. Bilbao, Fuel Process. Technol. 69 (2001) 157.
[27] C.D. Gilson, A. Thomas, J. Chem. Technol. Biotechnol. 62 (62) (1995) 227.
[28] G.T. Grant, E.R. Morris, D.A. Rees, P.J.C. Smith, D. Thom, FEBS Lett. 32 (1973) 195.
[29] A. Haug, Composition and Properties of Alginates, Report No. 30, Norwegian Institute of Seaweed Research, Trondheim,
Norway, 1964.
[30] H.L. Huguet, A. Groboillot, R.J. Neufeld, D. Poncelet, E. Dellacherie, J. Appl. Polym. Sci. 51 (1994) 1427.
[31] J.E. Atwater, J.R. Akse, G.N. Jovanovic, T. Sornchamni, J. Mater. Sci. Lett. 20 (2001) 487.