Jan Svoboda. Magnetic Techniques for the Treatment of Materials

Подождите немного. Документ загружается.

430 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

Figure 5.100: Viscosity of a kerosene-based ferroﬂuid, as a function of the ap-

plied magnetic ﬁeld, for three densities of the ferroﬂuid. Viscos-

ity of kerosene 

=1.75×10

3

Pa×s (adapted from Gubarevich

[G17]).

ﬁeld. As a result, an additional frictional coupling between ﬂuid layers is intro-

duced thereby increasing the viscosity [C22]. The fractional increase in viscosity

is larger when the ﬁeld is applied parallel to the stream, e.g. in a capillary, than

when it is applied perpendicular to the stream [C23].

By including the rotational component in the Einstein formula, several au-

thors have shown that the saturation values of the viscosity parallel (

)and

perpendicular (

) to the ﬁeld direction are given by:



= (1 +

i), 

= (1 +

i) (5.27)

where  is given by eq. (5.25).

The dependence of the viscosity of a kerosene-based ferroﬂuid on the applied

magnetic ﬁeld, for three values of the ferroﬂuid density, is shown in Fig. 5.100.

The measurements were made at the constant viscosity of kerosene and shear

stress. It can be seen that at a magnetic induction greater than 0.4 T, the

viscosity of ferroﬂuid is constant.

5.6. SEPARATION IN MAGNETIC FLUIDS 431

Mix

FeCl3.6H2O

FeCl2.4H2O

H2O

Co-precipitate

Concentrated

NH4OH

Peptize

Heat

Mix

Dispersant

(oleic acid)

Organic solvent

(kerosene)

Phase

separation

Aqueous salt

residue

Filtration

Dilution

Organic solvent

Final product

Figure 5.101: Preparation of a ferroﬂuid by chemical precipitation (adapted

from [R16]).

Preparation of ferroﬂuids

There are two basic methods of preparing a ferroﬂuid - size reduction and pre-

cipitation. Size reduction by wet grinding of ferrite in a ball mill, for a period of

1000 hours or longer, in the presence of a surfactant, was originally investigated

by Papell [P5], while Rosensweig et al. [R16] developed the technique further

and produced kerosene-based ferroﬂuids. The technique has been extended even

further by Kaiser and Rosensweig who succeeded in dispersing magnetic parti-

cles in other groups of solvents, including hydrocarbons, water, aromatics and

esters [R16, C22, R24].

Magnetite particles can also be prepared by precipitating the magnetite from

a solution of ferric and ferrous ions using an excess of an alkali hydroxide so-

lution. The process is schematically shown in Fig. 5.101. In the peptization

step, the particles are transferred from the aqueous phase to an organic phase

432 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

containing a dispersion agent, such as, for instance, oleic acid. Subsequently,

the particle dispersion in an organic phase is separated from the aqueous salt

residue, ﬁltered, and solvent adjusted to give the desired product concentration

[R16, B32].

Stability of ferroﬂuids

The stability of ferroﬂuids in gravitational and magnetic ﬁelds is one of the

fundamental parameters that ensures high selectivity of separation of materials

in ferroﬂuids. Basic criteria for such stability can be obtained by investigating

the balance of forces acting on the colloidal magnetic particles suspended in

the carrying ﬂuid. The relationship between the randomizing action of the

thermal energy and the destabilizing eect of the gravitational ﬁeld, surface

forces, magnetic interaction between the particles and the non-homogeneous

magnetic ﬁeld, and between the particles themselves, determines the conditions

under which a ferroﬂuid is stable.

It is fairly easy to show [R16] that while the force of gravity is of limited

threat to the segregation of ferroﬂuids, the eect of magnetic agglomeration

and of the ﬁeld gradient can be eliminated only when the particle diameter is

smaller than approximately 8 nm. However, molecular van der Waals forces that

arise as a result of ﬂuctuating electric dipole-dipole forces are always present in a

colloidal system. It transpires from eq. (3.121) that inﬁnite energy is required to

separate a particle pair. Therefore, contact between individual particles must

be prevented in order to obtain a stable colloidal suspension. Such a steric

stabilization can be achieved by adding a chemical, for instance oleic acid, that

can both adsorb on the surface of the particle and be solvated by the carrier

liquid. Such long-chain molecules are chosen in such a way that their tails have

dielectric properties similar to these of the surrounding carrier ﬂuid. This results

in the formation of a bound liquid sheath around each particle [K24]. The most

commonly used surfactants give a surface layer thickness of 2 nm to 3 nm.

The stability of kerosene-based ferroﬂuids used in ferrohydrostatic separators

extends over periods as long as one year or longer [G17, S8]. Rosensweig [R16]

reported that the properties of a ferroﬂuid prepared in kerosene did not deteri-

orate after 18 years. However, water-based ferroﬂuids usually contain particles

somewhat larger than the particles in organic carriers. The destabilizing eect

of magnetic agglomeration and the inﬂuence of the gravitational ﬁeld might

reduce the long-term stability of such ferroﬂuids. Moreover, the highly polar

nature of the carrier means that dispersants are not as eective as in organic

carrier ﬂuids.

A simple instrument that can be used to observe the long-term stability

of ferroﬂuids was developed by Bissell et al. [B33]. In a scanning column

magnetometer ferroﬂuid samples are introduced into glass tubes that are placed

in a gradient magnetic ﬁeld. The behaviour of particles as a function of time

can be observed and the stability of the ﬂuid determined.

5.6. SEPARATION IN MAGNETIC FLUIDS 433

Figure 5.102: The dependence of the apparent density of ferroﬂuid on tempera-

ture, for various physical densities of the ferroﬂuid (adapted from

[G17]).

The eect of temperature on the physical properties of a ferroﬂuid

A ferroﬂuid placed in a separation chamber of a ferrohydrostatic separator is,

over a period of time, exposed to an increasing temperature as a result of the

temperature increase in the windings of the electromagnet. The temperature

increase results in an increase in the volume, and a decrease in the physical and

apparent densities of the ferroﬂuid, at a constant concentration of magnetite.

The change of the apparent density of the kerosene-based ferroﬂuid with tem-

perature variation is illustrated in Fig. 5.102. It can be seen that when the

temperature increased from 20

Cto50

C, the apparent density of 3800 kg/m

of a ferroﬂuid with a physical density of 960 kg/m

decreased by 14%.

In addition to density, magnetic susceptibility also undergoes changes when

the temperature of the ﬂuid changes. It was observed [G17] that a temperature

increase by 15

C results in the reduction of the magnetic susceptibility of the

kerosene-based ferroﬂuid by 7% to 15%, depending upon the magnetic ﬁeld

strength to which the ferroﬂuid is exposed.

It is clear that temperature control of the ferroﬂuid and of the working

environment in which a ferrohydrostatic separator operates, are essential for

accurate performance of the equipment. Continuous detection of density varia-

tions and automated adjustment of the operating parameters of the separator

are essential components of modern ferrohydrostatic separators.

434 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

Table 5.33: Typical physical properties of ferroﬂuids, at 300 K.

Property Hydrocarbon Water Ester

Saturation polarization [T] 0.01-0.1 0.01-0.04 0.01-0.08

Density [kg/m

] 850-1250 1100-1400 1150-1400

Viscosity [Pa×s] 0.003-0.01 0.007-0.01 0.014-0.04

Boiling point [K] 350 299 422

Surface tension [10

3

N/m] 25-28 26 26

Thermal conductivity [W/m/K] 0.15 0.59

Electrical resistivity [m] 1.5×10

50-1000

Review of physical properties of ferroﬂuids

The chemical, mechanical and physical properties of ferroﬂuids correspond closely

to those of the carrier liquid. All commercially available ferroﬂuids utilizing or-

ganic carrier liquids are essentially electrically non-conductive. Their magnetic

saturation can be proportionately varied by dilution with the carrier ﬂuid. Most

ferroﬂuids can be freeze-thawed without damage. Table 5.33 lists various phys-

ical properties of commercially available ferroﬂuids.

Selection of a magnetic ﬂuid for material separation

When exposed to a non-homogeneous external magnetic ﬁeld, a paramagnetic

liquid will exhibit an apparent density given by eq. (3.144): 

= 





EuE.

In order to achieve the apparent density equal, for instance, to the density of

diamond, that is 3500 kg/m

, in a saturated solution of manganese chloride (

= 1450 kg/m

, " =5.65×10

7

/kg and  =8.2×10

4

(SI)), the product of

the magnetic induction and its gradient EuE must be equal to 30.8 T

/m. In a

magnet that can generate 2 T, the ﬁeld gradient must be 15.4 T/m or 1.54×10

G/cm. This is a very high value indeed. In a superconducting magnet, at the

ﬁeld induction of 20 T, the required ﬁeld gradient is 1.54 T/m, or 154 G/cm.

The application of paramagnetic liquids is more feasible on a small laboratory

scale [P14] using, for example, a Frantz laboratory isodynamic separator. In

this separator, the product EuE can reach maximum value of 100 T

/m and

an apparent density of 10 000 kg/m

was achieved with maximum throughput

of the separator at 40 g/h and a maximum particle size of 1 mm [P14].

It is clear, therefore, that for separation in paramagnetic liquids very high

magnetic ﬁelds are required. This requirement results in high capital and run-

ning costs, and in the need to remove even very feebly magnetic material from

the feed. In addition, the working volume of such a separator would necessar-

ily be very small, which would limit the upper particle size, and its throughput

would be very low. Even in a massive and expensive installation, the throughput

would not exceed a few kilograms per hour.

These practical di!culties make separation in paramagnetic liquids (usually

termed magnetohydrostatic separation - MHS) an intriguing curiosity rather

5.6. SEPARATION IN MAGNETIC FLUIDS 435

Figure 5.103: Saturation polarization of several types of ferroﬂuids, as a function

of their physical density [S80].

than a viable technique. It was realized a long time ago [L17] and not without

disappointment, that separation in paramagnetic liquids would never become

a widely used processing technique. Only under special circumstances would

production of a high-value concentrate from a small volume of coarse feed be

economically justiﬁed.

In contrast to paramagnetic liquids, ferroﬂuids are moderately strongly mag-

netic and modest magnetic ﬁelds and gradients are required to generate apparent

densities over a wide range of values. Figure 5.103 reviews the saturation mag-

netization of various ferroﬂuids used for material separation, as a function of

their physical density.

When exposed to a non-homogeneous magnetic ﬁeld, ferroﬂuids exhibit an

apparent density that is given by eq. (3.145). For example, the apparent density

of 3500 kg/m

can be obtained with a kerosene-based ferroﬂuid of density 1000

kg/m

, having magnetic polarization of 0.02 T (200 G), at a ﬁeld gradient

of 1.54 T/m (154 G/cm). Such a gradient can be fairly easily generated by

laboratory or pilot-scale FHS units at a magnetic ﬁeld as low as 0.2 T. Table

5.34 compares conditions required to achieve an apparent density of 3500 kg/m

in paramagnetic liquids and ferroﬂuids

Table 5.34 illustrates the considerable advantage of ferroﬂuids over para-

magnetic liquids, while Fig. 5.104 compares the operating density ranges of

various sink-and-ﬂoat techniques. The potential of ferrohydrostatic separation

436 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

Table 5.34: Typical parameters required to achieve an apparent density of 3500

kg/m3 in laboratory or pilot-scale separators with a paramagnetic

liquid and a ferroﬂuid. The costs of liquids are for the same value

of magnetization, i.e. 0.0016 T (16 G).

Parameter MnCl

·4H

O Kerosene FF

Polarization @ 2 T 16×10

4

Saturation polarization 200×10

4

Density [kg/m

] 1400 1000

Apparent density [kg/m

] 3500 3500

Viscosity @ B = 0 1.5×10

3

Pa×s 2.5×10

3

Pa×s

Magnetic induction 2 T or 20 T 0.2 T

Field gradient 15.4 T/m @ 2 T 1.54 T/m

1.54 T/m @ 20 T

Cost (US$/c) 10 0.7 (commercial)

0.07 (in-house)

Figure 5.104: The operating density ranges of various sink-and-ﬂoat separation

methods.

5.6. SEPARATION IN MAGNETIC FLUIDS 437

Table 5.35: Comparison of MHS and FHS separation techniques.

MHS FHS

High magnetic ﬁeld (2 T or higher) Modest magnetic ﬁeld (0.2 - 0.5 T)

High ﬁeld gradient (10 - 20 T/m) Modest ﬁeld gradient (1 - 2 T/m)

Maximum density: 10 000 kg/m

Maximum density A 22 000 kg/m

High-intensity magnetic scalping Low-intensity magnetic scalping

Low throughput (max. 1 kg/h) Easy scale-up to 1 t/h

Narrow size fractions Limited sensitivity to particle size

Max. particle size approx. 1 mm Max. particle size A 20 mm

High cost of liquid Moderate cost of ferroﬂuid

High capital and running costs Moderate capital and running costs

Large mass and ﬂoor space Modest mass and ﬂoor space

Di!cult handling of liquids High stability of ferroﬂuids

Di!cult re-concentration of liquids Recovery and recycling of FF simple

in ferroﬂuids is obvious. The main advantages of ferrohydrostatic separation

compared to magnetohydrostatic separation in paramagnetic liquids are sum-

marized in Table 5.35.

Water-based or kerosene-based ferroﬂuid? For separation purposes, water-

based and kerosene-based ferroﬂuids are commercially available. Generally

speaking, ferroﬂuids with hydrocarbon carrier ﬂuids tend to have a high de-

gree of monodispersity and a minimum level of aggregation. A wide range of

dispersants is available for such ferroﬂuids, the cost of which can be very modest

as a result of a range of iron-containing low-cost waste materials available from

the petrochemical industry. Given this wide range of available materials, it is

possible to make relatively low-cost kerosene-based ferroﬂuids having a wide

range of saturation magnetization and viscosity. These ferroﬂuids can be fairly

easily removed from the products of separation, recycled and re-concentrated,

and returned to the separation circuit.

As can be seen in Fig. 5.103, kerosene-based ferroﬂuids with saturation

polarization as high as 0.05 T (500 G) can be produced. In a typical range of

apparent densities from 3000 kg/m

to 9000 kg/m

, a ﬂuid with a saturation

polarization of the order of 0.0150 T to 0.02 T (150 G to 200 G) is required. Such

a ﬂuid can be obtained by diluting the concentrated ferroﬂuid, which further

reduces the viscosity and the cost of the process. Such a dilution must, however,

be carried out with a certain level of caution. Excessive dilution could result in

desorption of the protective surfactant from the surface of the particles in order

to maintain the equilibrium between adsorbed and free surfactant. This could

considerably reduce the colloidal stability of the ﬂuid and change its magnetic

and physical properties [C24].

Waste water resulting from the production of kerosene-based ferroﬂuid con-

sists mainly of organic waste such as kerosene and oleic acid. Disposal of such

438 CHAPTER 5. PRACTICAL ASPECTS OF MAGNETIC METHODS

Table 5.36: Selected physical and chemical properties of kerosene.

Property Value

Boiling point 200

Cto260

Freezing point - 45.6

Density 800 kg/m

(at 15

Reactivity with water no reaction

Reactivity with common materials no reaction

Flammable limits in air 0.7% to 5%

Ignition temperature 229

Fire combustible

Exposure irritating to skin and eyes

Protective equipment gloves, goggles

a waste either by incineration or by applying an aerobic reactor system, using

microorganisms obtained from activated sludge, can be easily achieved [V7].

Some physical and chemical properties of kerosene are shown in Table 5.36.

The alternative separation medium is a water-based ferroﬂuid. In water-

based colloids, the highly polar nature of the carrier renders dispersants less

eective than in hydrocarbon ﬂuids. Long-term stability and stability in high-

gradient ﬁelds is, therefore, limited Furthermore, the inclusion of a higher

concentration of dispersant, possibly of a more complex nature, tends to increase

the viscosity of the water-based ferroﬂuid compared with a hydrocarbon-based

material of the same concentration.

As illustrated in Fig. 5.103, because of the complexity of water-based ﬂuids,

it is somewhat di!cult to obtain a wide range of magnetization, and the satura-

tion polarization of 0.02 T (200 G) is typical for water-based ferroﬂuid used in

material separation. Such a concentrated ﬂuid must be used undiluted, in con-

trast to kerosene-based ﬂuid. This results in an increased viscosity and cost of

water-based ferroﬂuid. In addition, because of greater volatility of water, com-

pared to hydrocarbons, the values of magnetization and other parameters are

susceptible to greater variations with time and temperature. Recycling and re-

concentration of water-based ferroﬂuids, for instance by evaporation, although

seemingly simple, are rather costly, and time and space consuming.

The image of kerosene-based ferroﬂuids as hazardous and environmentally

unfriendly, resulted, in some instances, in exclusion of the FHS technique from

production-scale application. Nevertheless, certain obstacles that such ﬂuids

present can be fairly easily contained within a beneﬁciation circuit. Moreover,

water-based ferroﬂuid, as a result both of the increased level of dispersants and

their complexity, may not be as non-polluting as imagined from a simplistic

point of view. However, the entire material separation circuit can be made

simpler when water-based ferroﬂuids are used [F25].

5.6. SEPARATION IN MAGNETIC FLUIDS 439

Figure 5.105: The 1983 cost of kerosene-based ferroﬂuid as a function of the pro-

duction rate. Spent hydrochloric acid pickle liquor was considered

as the source of iron ions (adapted from Farkas [F26]).

The cost of ferroﬂuid

If the ferrohydrostatic separation technology is to be successful on a produc-

tion scale, ferroﬂuids must be available in su!ciently large quantities and at

reasonable cost.

The current situation in the market is not conducive to the introduction of

FHS on large scale. There are several small suppliers who provide speciality

ferroﬂuids in small quantities and at high cost. Technical-grade kerosene-based

ferroﬂuids for ferrohydrostatic separation are manufactured in larger quantities

by one or two suppliers in Russia. The 2000 price for a 0.04 T (400 G) ferroﬂuid

was US$40 or more per litre. Ferrotec (Japan), formerly Ferroﬂuidics Corp.,

supplies, on a larger scale, water-based ferroﬂuids for separation applications.

The 2002 price for 0.017 T (170 G), 1230 kg/m

ferroﬂuid was US$45 per litre.

Although these prices are acceptable for laboratory and possibly for pilot-

scale applications, a large industrial application of FHS requires reliable supply

of low-cost ferroﬂuid of consistent quality. Union Carbide Corp., in the early

1970s, developed a pilot plant for kerosene-based ferroﬂuid for their separation

application [F26]. The capacity of the plant was 4.3 c/h of 0.035 T ferroﬂuid.

Based on manufacturing experience, su!cient information was obtained

which permitted calculation of the cost of ferroﬂuid in an automated plant

of expanded capacity. The 1983 cost data shown in Fig. 5.105 were based on

the cost of materials, labour and supervision, and utilities [F26]. The diagram

illustrates that, as the production rate increases, the unit labour cost diminishes

and the production cost per litre, for large volumes, approaches the costs of the