CHAPTER 1

Introduction

1.1 Background

Fluorosis caused by high fluoride intake predominantly through drinking water

containing F- concentrations > 1 mg.L -1, is a chronic disease manifested by
mottling of teeth (dental fluorosis) in mild cases and changes in bone structure
(skeletal fluorosis), ossification of tendons and ligaments, and neurological
damage in severe cases (Wang and Reardon, 2001; Ghorai and Pant, 2002).
Today increasing concern is being expressed that these adverse effects of
fluorosis are irreversible.

With the rural population of South Africa in many areas currently using drinking
water with high F- concentrations supplied from wells and boreholes, the
development of a long term solution for the defluoridation of F- contaminated
groundwaters is of critical importance. This would require appropriate water
treatment procedures. Appropriate technology must be technically simple, cost-
effective, easily transferable, use local resources and must be accessible to the
rural community.

The removal of fluoride from water using defluoridation techniques is a common

practice world-wide, both in industry and domestically. Current methods of
fluoride removal from water include adsorption onto activated alumina, bone char
and clay, precipitation with lime, dolomite and aluminium sulfate, the Nalgonda
technique (Srimurali et al., 1998), ion exchange (Mohan Rao and Bhaskaran,
1988), and membrane processes such as reverse osmosis, electrodialysis and
nanofiltration. These processes are discussed in more detail in Section 2.1.

1
1.2 The origin and distribution of fluoride in groundwaters

1.2.1 International

High F- concentrations in groundwater are found in many countries around the

world, notably the United States of America, Africa, and Asia (Czarnowski et al.,
1996; Azbar and Turkman, 2000). The most severe problems associated with
high F - waters occur in China (Wang et al., 2002), India (Argawal et al., 2003), Sri
Lanka (Disanayake, 1996) and Rift Valley countries in Africa. High fluoride
groundwaters have been studied in detail in Africa, in particular Kenya and
Tanzania (Moges et al., 1996; Gaciri and Davies, 1999; Chernet et. al., 2002;
Mjengera and Mkongo, 2002; Moturi et al., 2002). The abundance of F- in Rift
Valley groundwaters is due to the weathering of alkaline volcanic rocks rich in F-.
Typical fluoride concentrations of towns in the Rift Valley are between 1 and 33
mg.L -1. High fluoride groundwater is also found in the East Upper Region of
Ghana (Apambire et al., 1997). The concentration of fluoride was found to be
between 0.11 and 4.60 mg.L -1.

1.2.2 South Africa

High fluoride groundwaters in the Republic of South Africa (see Figure 1.1) have
recently been studied. Few of the studies, however, cover the whole of South
Africa. Detailed studies were done in Bushveld and Pilanesberg regions (Bond,
1947; Fayazi, 1995; McCaffrey and Willis, 2001). These two areas were seen as
areas deserving special investigation because of endemic dental fluorosis.

Fluoride concentrations in rivers flowing through the Kruger National Park were
determined by Raubenheimer et al. (1990). All five rivers (Olifants, Luvuhu,
Letaba, Sabie and Crocodile rivers) had higher concentrations of F- during the
latter part of the dry winter season.

2
Fayazi (1995) measured the fluoride level in groundwater of the Northern
Springbok Flats and attributed high fluoride groundwater to fluoride-bearing
minerals, such as fluorite (CaF2). Fluorite is known to occur in the granites of the
Bushveld Complex which surround the basins of the Springbok Flats.

Figure 1.1
South African groundwater with F- concentration > 1.5 mg.L-1 (McCaffrey
and Willis, 2001)

There is a wide variation in fluoride levels in the natural waters of South Africa.
This was evident from an analysis of ca. 3000 boreholes in the Pilanesberg and
Western Bushveld area (see Figure 1.2). More than 30% of the boreholes have
F- concentrations > 1 mg.L -1. In alkaline waters (pH > 9) F- concentrations up to
30 mg.L -1 were recorded (McCaffrey and Willis, 2001).

3
 500
450
400
350

Frequency
300
250
200
150
100
50
0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
-1
[F]/mg.L

Figure 1.2
Distribution of F- concentrations in ca 3000 boreholes in the North
West Province.

1.3 Fluorosis

1.3.1 Background

Fluoride has certain physiological properties (Muller et al., 1998; Notcutt and
Davies, 1999) of great importance in human health. The role of fluoride in the
process of mineralisation of certain tissues is important. At low concentrations
fluoride stabilises the skeletal system by increasing the size of apatite crystals
and reducing their solubility (Moges et al., 1996). Although beneficial effects can
be demonstrated at low concentrations, it has detrimental effects when
concentrations exceed the threshold (Li et al., 2001).

The relationship between fluoride and dental caries was first noted in the early
part of the 20th century when it was observed that residents of certain areas of
U.S.A. developed brown stains on their teeth. In the 1930s it was observed that
the prevalence and severity of this type of mottled enamel was directly related to
the amount of fluoride ingested (Dean, 1934).

4
Endemic fluorosis is known to be global in scope, occurring in all continents and
affecting many millions of people. Cases of skeletal fluorosis have been reported
all over the world (Hillier et al., 2000). According to a report (Susheela, 2001)
from UNICEF, fluorosis is endemic in at least 25 countries across the globe.

The fluorosis problem is most severe in the two most populous countries of the
world, China and India (Cao et al., 1997; Fung et al., 1999, Mekonen et al.,
2001). Thus for example, in China some 38 million people are reported to suffer
from dental fluorosis and 1.7 million from the more severe skeletal fluorosis. In
India, 15 states are endemic for fluorosis, with over six million people seriously
afflicted (Kailash et al. 1999).

The drinking water standards for fluoride ion stipulated by various authorities are
tabulated in Table 1.1. There is uncertainty with regard to the precise
concentrations of fluoride in drinking water which are optimal for human health.

Table 1.1: Drinking water standards for fluoride ion prescribed by various
authorities (Smet, 1992)

Authority Maximum Fluoride

concentration (mgL-1)
WHO (International 0.5
Standard)
US Public Health 0.7-1.2
South African Bureau 1.0
of Standards

5
1.3.2 Symptoms of fluorosis

1.3.2.1 Skeletal fluorosis

Fluorapatite is an order of magnitute less soluble than hydroxyapatite, the

principal mineral constituent of bone. The F- ion aggressively substitutes for the
OH- ion, leading to a buildup of F- in bone tissue which eventually lead to skeletal
fluorosis.

Early in the development of fluorotic changes in the skeleton, the patient often
complains of a vague discomfort in the limbs and the trunk. Pain and stiffness in
the back appear next, especially in the lumbar region.

In severe fluorosis, in addition to joint problems, some victims can experience

deformation of their bones. The stage at which skeletal fluorosis becomes
crippling usually occurs between 30 to 50 years of age in endemic regions. The
factors which govern the development of skeletal fluorosis are (a) the prevalence
of high levels of fluoride intake, (b) continual exposure to fluoride, (c) strenuous
manual labour, (d) poor nutrition and (e) impaired renal function due to disease
(Nicolay et al., 1999).

1.3.2.2 Dental fluorosis

Human beings throughout history have suffered from dental fluorosis, but until
the 20th century the cause of the condition was unknown. Given the common
incidence of high F- groundwaters in the East African Rift Valley (Walvekar and
Qureshi, 1982), it is evident that ancient people could have suffered from dental
fluorosis.

Mottling of teeth is one of the earliest and most easily recognized symptoms
(Choi and Chen, 1979). It is the permanent teeth that are affected the most and

6
they lose their normal creamy white translucent colour and become rough,
opaque and chalky white. Dental fluorosis is a developmental disturbance which
increases with time. Therefore primary teeth are less severely affected than the
permanent teeth, and those teeth which erupted first (the incisors and first
permanent molars) are less affected than those erupting later, the premolars and
other permanent molars (Pereira and Moreira, 1999).

The first signs of dental fluorosis (moderate dental fluorosis) are thin white lines
running across the entire enamel surface and which can only be seen after
drying of the tooth surface (Dean, 1942). With more fluorosis these thin lines
become broader, merge and may be clear without the need for drying. At slightly
greater severity the tooth surface shows distinct, irregular, opaque or cloudy
white areas, caused by increasing porosity of the tooth enamel.

Dental fluorosis has been studied in several parts of South Africa (Carstens et
al., 1995; Mothusi, 1995; Louw and Chikte, 1997). The findings by Mothusi
(1995) indicated that in North West province, dental fluorosis has resulted in
cases of psycological trauma, particularly amongst adolescents. In areas with
fluoride levels of drinking water exceeding 3 mg F- L-1, local inhabitants demand
that their teeth be extracted and replaced with dentures. Dental fluorosis was
also investigated in areas surrounding the Pilanesberg complex (Rudolph et al.
1995). They reported that two of the villages had high F- groundwater. They
found that severe dental fluorosis occurred in 28% of the population while 41%
had moderate dental fluorosis.

The same results were reported in the study by Du Plessis et al (1995). They
found that 39% of school children in the Bloemfontein area had evidence of
dental fluorosis. The spatial variations of dental fluorosis in seven villages was
studied by Zietsman (1989).

7
1.4 Ways of solving the problem: defluoridation techniques

The prevention of fluorosis through treatment of drinking water in rural areas is a

difficult task because of economical and technological restrictions. Defluoridation
of water is the only measure to prevent fluorosis and many different
defluoridation methods have been developed (Chaturvedi et al., 1990). However,
not many are applicable in the areas where the problems occur. This section
gives a brief overview of defluoridation methods (See Chapter 2 for a more
detailed discussion).

Defluoridation processes are categorised into four main groups:

Adsorption methods: In these methods sorbents such as bone charcoal,

activated alumina, and clay are used in column or batch systems.

Ion exchange methods: These methods require expensive commercial ion

exchange resins.

Co-precipitation and contact precipitation methods: These methods coprecipitate

F- with for example aluminium sulfate and lime (Nalgonda technique.) or
precipitate F - for example with calcium and phosphate compounds.

Membrane processes: These include reverse osmosis, nanofiltration and

electrodialysis methods.

Taking into account the realities of the problem as outlined in this Chapter the
provision of an affordable and technologically simple solution must obviously lie
in empowering the local communities to construct viable defluoridation systems
from local and readily available materials.

8
1.5 Motivation for this study

High fluoride groundwaters have been reported in various parts of South Africa
(McCaffrey, 2001) and since groundwater forms a major source of drinking water
in rural areas, rural populations are facing a major health problem in these areas.

Fluoride removal using activated alumina and reverse osmosis was proposed as
a solution to the problem in South Africa (Schoeman and Steyn, 2000). However,
the capital cost for household defluoridation was estimated to be about R 5000
for a 50 L per day unit, using either of the methods. This turned out to be too
expensive. There is thus a need to develop low cost methods to remove fluoride
from water. The removal of fluoride using locally available clays was studied in
many countries where the problem occurs. Adsorption of fluoride onto clay
minerals was, however, not studied in South Africa before.

The main objective of this WRC-sponsored research was therefore to study the
adsorption of fluoride onto clay sorbents from a mechanistic point of view, and to
evaluate South African clays as possible sorbents in defluoridation systems.

In order to achieve this outcome the specific objectives were:

(i) The fundamental study of F- adsorption mechanisms and the derivation of a

proper modelling algorithm for fluoride adsorption onto mineral surfaces. This
includes the study of the effect of mineralogical and structural characteristics of
clays, F- solution chemistry, surface properties of the sorbent, and how this is
related to F- adsorption capacity. Chapter 4 discusses adsorption mechanisms
and adsorption modelling using the model developed in this study.

(ii) The characterisation of South African clays for their F- adsorption potential
and usefulness in defluoridation syste ms. This is the first study of this nature on
South African clays and is reported in Chapter 5.

9
(iii) The effect of physical and chemical pre-treatment procedures on enhancing
adsorption efficiencies of clays and the mechanistic explanations for these
procedures (Chapter 5).

(iv) The development of laboratory scale defluoridation columns to study the

efficiency of fluoride removal using different sorbents (Chapter 6).

10
 CHAPTER 2
Literature study

2.1 Defluoridation techniques

The process of removal of fluoride is generally termed defluoridation. A

comprehensive search of the literature reveals that fluoride removal techniques
fall into three major categories:

Chemical precipitation and coprecipitation

Adsorption/Ion exchange
Membrane processes.

The materials studied under each category are tabulated in Table 2.1

Table 2.1: Materials and methods of defluoridation

Method Process/Material Reference

Coprecipitation Aluminium salts Dahi, 1996
(Nalgonda Technique)
Precipitation Calcium and phosphate Larsen et al., 1993
compounds
Adsorption/Ion Activated alumina Hao et al., 1986; Schoeman
exchange and Steyn, 2000
Fly ash Chaturvedi et al., 1990
Clays Srimurali et al., 1998
Soils Omueti and Jones, 1977
sulphonated carbonaceous Mohan Rao and Bhaskaran,
materials 1988

11
Membrane processes Reverse Osmosis Schoeman and Steyn, 2000
Nanofiltration Lhassani et al., 2001

2.1.1 Precipitation methods

Precipitation methods can be divided into two categories, those based on

coprecipitation of adsorbed F- and those based on the precipitation of insoluble
fluoride compounds.

2.1.1.1 Methods based on coprecipitation:

Coprecipitation (eg. the Nalgonda Technique) is the process by which aluminium

salts (aluminium chloride and aluminium sulphate) is added to F- contaminated
drinking waters for treatment (Yang et al., 1999; Yang and Dluhy, 2002). This
process is used in three ways.

A bucket system designed to be used on household scale

Fill and draw plants to be used on community scale
A waterworks flow system developed for larger communities.

(i) Bucket system

The bucket defluoridation system was first practised for domestic use in
Tanzania (Mjengera and Mkongo, 2002). The two chemicals (aluminium chloride
and aluminium sulphate) are added simultaneously to the raw water bucket and
stirred with a wooden paddle. Lime is added to adjust the pH of water to about
6.7. After addition of the chemicals it is left to settle for about 1 hour.

12
This process is suitable for a daily routine, where one bucket of water is treated
for one days water supply of about 20 L. The process produces water with
residual F - between 1 and 1.5 mg.L -1 (Dahi et al.; 1996).

(ii) Fill and draw system

This system is also used in Tanzania for the defluoridation of drinking water
(Mjengera and Mkongo, 2002). It consists of a cylindrical vessel equipped with a
hand operated stirring mechanism. The vessel is filled with raw water and a
similar procedure for defluoridation using bucket system is performed. Raw water
is pumped onto the tank and the required amounts of alum, lime and bleaching
powder are added. The contents are stirred slowly for ten minutes and allowed to
settle for two hours. The defluoridated supernatant water is withdrawn and
supplied through stand-posts. The settled sludge is discarded.

(iii) Waterworks flow system

A bigger defluoridation system is used for larger communities. This system

involves the combined use of alum and lime for the defluoridation process (Dahi,
1996). It consists of several components, namely, reactors, sump well, sludge
drying beds, elevated service reservoir, electric room and chemical storehouse.
The raw water from the source is pumped to the reaction-cum-sedimentation-
tank which is referred to as reactor. A sludge pipe with sluice valve is provided to
withdraw the settled sludge once a day.

The Nalgonda technique has been introduced in many countries, e.g. India,
Kenya, Senegal and Tanzania. However, the method has a number of
disadvantages. These include:

13
 The treatment efficiency is about 70%, which means the process cannot
be used in cases of high fluoride contamination.
A large dosage of aluminium sulphate, up to 700-1200 mg.L -1 may be
needed.
The adverse health effects of dissolved aluminium species in the treated
water.

2.1.1.2 Methods based on F- precipitation with calcium and phosphate

compounds:

Many methods of precipitation of fluorides with salts of calcium, aluminium and

iron are reported in the literature (Lawler and Williams, 1984; Larsen et al., 1993;
Qafas et al., 2002). Precipitation processes are governed by the solubility of a
forming salt (Parthasarathy et al., 1986). The most common method of treatment
is the precipitation of calcium fluoride using calcium from either lime or calcium
chloride.

The fundamental problem that exists using lime arises from the low solubility of
the calcium hydroxide. It therefore requires excess of reagent to complete
precipitation. The relatively high solubility of the calcium fluoride does not allow a
complete removal of F-. An additional difficulty with lime precipitation is the poor
settling characteristics of the precipitate.

The lime-based fluoride removal can be improved by using CaCl2-lime mixture.

The highly soluble CaCl2 provides more calcium than lime without increasing pH.
Fluoride removal by lime and CaCl2-lime costs about the same.

2.1.2 Adsorption methods

Fluoride can be removed by adsorption onto many adsorbent materials. The

criteria for selection of suitable sorbents are: cost of the medium and running

14
costs, ease of operation, adsorption capacity, potential for reuse, number of
useful cycles and the possibility of regeneration. Some of the most frequently
encountered sorbents a re reviewed in this section.

2.1.2.1 Activated alumina

Activated alumina is a granular form of aluminium oxide (Al2O3) with very high
internal surface area, typically in the range of 200-300 m2/g. This high surface
area allows the material a very large number of sites where adsorption can
occur. It has been widely used for removal of F- from drinking water (Hao et al.,
1986; Schoeman and MacLeod, 1987).

The mechanism of F- removal from water is similar to those of a weak base ion
exchange resin. Fluoride removal efficiency is excellent (typically > 95%), and is
dependent on pH. Fluoride removal capacity is best in the narrow range of pH
5.5 to 6.

Fine (28-48 mesh) particles of activated alumina are typically used for F-
removal. The adsorption sites on the activated alumina are also attractive to a
number of anions other than F-. The selectivity sequence (Johnston and Heijnen,
2002) of activated alumina in the pH range of 5.5 to 8.5 is:

OH->H2AsO4->Si (OH)3O->HSeO3->F->SO42->CrO42->>HCO3->Cl->NO3->Br->I-

Activated alumina can be regenerated by flushing with a solution of 4% sodium

hydroxide which displaces F- from the alumina surface (Schoeman and
MacLeod, 1987). This procedure is followed by flushing with acid to re-establish
a positive charge on the surface of the alumina.

15
2.1.2.2 Clays and soils

The first comprehensive study of fluoride adsorption onto minerals and soils was
published in 1967 (Bower and Hatcher, 1967). Since the above -mentioned paper
was published, several workers studied the adsorption of fluoride. These studies
include the use of Ando soils of Kenya ( Zevenbergen et al., 1996), Illinois soils
of USA (Omueti and Jones, 1977), Alberta soil (Luther et al., 2996), illite- goethite
soils in China (Wang and Reardon, 2001), clay pottery (Chaturvedi et al., 1988;
Hauge et al., 1994), fired clay (Bardsen and Bjorvatn, 1995), fired clay chips in
Ethiopia (Moges et al., 1996), kaolinite (1997), bentonite and kaolinite (Kau et al.,
1998; Srimurali et al., 1998), and fly ash (Chaturvedi et al., 1990).

2.1.2.3 Other sorbents

In addition to activated alumina, clays and soils other materials such as spent
bleaching earth, spent catalyst, rare earth oxides, bone charcoal and activated
carbon were studied as sorbents for F-. Mahramanlioglu et al (2002) investigated
the adsorption of F- using spent bleaching earth. They found that the removal of
F- depends on the contact time, pH and adsorbent concentration. Lai and Liu
(1996) studied the F- removal from water with spent catalyst. Their findings
showed that spent catalyst could be utilized as adsorbent for F- removal. Its
adsorption capacity was comparable to that of activated alumina. Raichur and
Basu (2001) studied the adsorption of F- onto rare earth oxides. Rare earth
oxides showed great potential for F- removal from water. Lu et al (2002)
investigated the removal of F- using red mud. The removal of F- using red mud
was found to be 82%.

16
2.1.3 Ion exchange resins

Ion exchange resins are effective in removing F- from water. Mohan Rao and
Bhaskaran (1988) studied the removal of F- using ion exchange materials such
as sulphonated material from coconut shell, Carbion, Tulsion and Zeocarb 225.
From the results, it was evident that Zeocarb 225 has the highest F- removal
capacity and sulphonated material of coconut shell has the lowest. It was also
indicated that the ion exchange material can be regenerated by aluminium
sulphate solution (2-4%). Castel et al (2000) studied the removal of F- by a two
way ion exchange cyclic process. This system used two anion exchange
columns. The results show that this process can effectively remove fluoride from
water.

The use of anion exchange resins for F- removal is not common because of their
relatively high costs. The presence of other anions such as chloride and sulphate
also presents a major problem when using ion exchange resins for F- removal.
Since F- removal is accompanied by sorption of other anions, the sorption
capacity is normally lower than 0.5 mg F -.L-1 (Veressinina et al., 2001).

2.1.4 Membrane processes

Membrane processes such as reverse osmosis, nanofiltration and electrodialysis

are recently developed methods for F- removal from water (Schoeman and
Steyn, 2000; Lhassani et al., 2001; Garmes et al., 2002). Not much research has
gone underway using these membrane processes. However, the study by
Lhassani et al (2001) indicates that F - can be removed using nanofiltration.

17
2.2 Clay adsorption studies

Bower and Hatcher (1967) indicated that the adsorption of F- onto minerals and
soils is accompanied by the release of OH- ions. It was also found that the F-
adsorption is concentration dependent and it is described by the Langmuir
adsorption isotherm. After this study, many studies on the adsorption of F- using
clay minerals and soils were undertaken. Some of the studies involving F-
adsorption are discussed below.

2.2.1 Fired clays

Several researchers have studied the removal of F- using fired clays (Hauge et
al., 1994; Bardsen and Bjorvatn, 1995; Moges et al., 1996). Hauge et al (1994)
studied the defluoridation of drinking water using pottery. The study investigated
the effect of temperature on F- adsorption. The results show that clays fired at
temperature up to 600C gave higher F- adsorption. Moges et al (1996) studied
the defluoridation of water using fired clay chips in Ethiopia. Their findings
indicated that F- adsorption is affected by factors such as initial concentration,
mass of adsorbent and the pH of the solution.

2.2.2 Low cost materials

Srimurali et al. (1998) investigated the removal of F- using low cost materials
such as kaolinite, bentonite, charfines, lignite and nirmali seeds. Their results
show that F- adsorption using nirmali seeds and lignite is low (6 to 8%). The
removal of F- by kaolinite is slightly better (18.2%) while charfines and bentonite
give higher F- removal capacity of 38 and 46% respectively. Chemical pre-
treatment was used to investigate its effect on the removal capacity of these
materials. Kau et al. (1998) investigated the adsorption of F- by kaolinite and
bentonite. The results show that bentonite was found to have higher F-
adsorption than kaolinite. F- removal using China clay was studied Chaturvedi

18
(Chaturvedi et al., 1988). The results show that low F- concentration, high
temperature and acidic pH are factors favouring the adsorption of F-. It was
concluded that the alumina constituent of the China clay is responsible for F-
adsorption. Chaturvedi et al. (1990) also studied the defluoridation of water by
adsorption on fly ash. This study confirms their previous findings that low F-
concentration, high temperature and acidic pH favour the adsorption of F -.

2.2.3 Soils

Several studies investigated the adsorption of F- using soils (Omueti and Jones,
1977; Chhabra et al., 1980; Zevenbergen et al., 1996; Bjorvatn et al., 1997;
Wang and Reardon, 2001). Chhabra et al. (1980) investigated the effect of
varying levels of exchangeable sodium on the adsorption of F- onto sodic soils. It
was noted that at equilibrium F- concentration, a decrease in adsorption of F- with
increase in the soil exchangeable sodium percentage was observed. Omueti and
Jones (1977) studied the adsorption of F- by Illinois soils. They reported that at
low concentrations F- adsorption onto soils was described by both Langmuir and
Freundlich isotherms. It was also suggested that F - adsorption onto soils was due
to the presence of the amorphous aluminium hydroxides. Bjorvatn et al. (1997)
studied the defluoridation of water using soil samples from Ethiopia. It was
reported that five soil samples from highland areas around Addis Ababa reduced
the fluoride content of the water from about 15 to 1 mg.L -1. From this study, it was
concluded that the highland soil may be useful for removal of excessive F- from
drinking water. Zevenbergen et al. (1996) studied the defluoridation of water
using the Ando soil of Kenya. It was concluded that the use of Ando soils
appears to be an economical and efficient method for defluoridation of drinking
water.

The structure of the clay plays a very important role in determining the charge on
the clay surface and type of exchange that can occur with ions in solution. In
general the more positive the surface the better the sorption will be for negatively

19
charged ions, such as F-. As discussed in detail in Chapter 4, pH plays a
dominant role in determining the adsorption capacity as pH modifies the charges
on edge positions in phyllosilicates and also those of variably charged minerals
such as gibbsite, hematite and goethite. Charges are generally positive under
acid conditions and negative in an alkaline environment. The specific pH range
for positive and negative surface charge will of course be a function of the pKa
values of the metal hydroxides present.

Chemical pre-treatment which includes the use 1% Na2CO3 and 1% HCl was
reported to improve the adsorption capacity of clays and soils (Srimurali et al.,
1998). In general, it was found that firing and chemical pre-treatment both
improve the adsorption capacity of some clays and soils.

Studies on the surface coating of clays and soils were reported (Coleman and
Thomas, 1964; Agarwal et al., 2002; Zhuang and Yu, 2002). The coating of
clays and soils with Al and Fe hydroxides improves their adsorption capacity.

2.3 South African studies on defluoridation and F - adsorption

South African studies on defluoridation of drinking water was first started in mid
1980 (Schoeman, 1985). Most of these studies have been concerned with the
use of activated alumina for fluoride removal (Schoeman, 1985; Schoeman and
Leach, 1986; Schoeman and Steyn, 2000). Schoeman (1985) developed two full
scale activated alumina defluoridation plants. These plants reduced the F-
content of the underground mine water from 8 to less than 1.5 mg.L -1. This study
was followed by a critical evaluation on the performance of the two defluoridation
plants (Schoeman and Leach, 1986). In recently published work on this topic,
Schoeman and Steyn (2000) evaluated the use of activated alumina and reverse
osmosis for the defluoridation of water.

20
 CHAPTER 3
Experimental procedure and methodologies

The procedures described here were developed and used in the study of F-
adsorption mechanisms onto mineral substrates (Chapter 4), the characterisation
of South African clays for their F- adsorption potential (Chapter 5), and laboratory
scale column defluoridator systems discussed in Chapter 6. These methods are
discussed in detail in this chapter and will only be referred to in the follo wing
chapters.

3.1 Mineralogical characterisation of the clays

3.1.1 Sample preparation

A representative portion of each clay sample was dried overnight at 30C, after
visible pieces of plant material and pebbles were taken out by hand. Each
sample was lightly crushed in a swing-disk mill to an arbitrary chosen grain size
of <180 m. Each sample was then dispersed in distilled water using an
ultrasonic bath. The clay fraction was separated by using standard centrifugal
techniques (USGS, open file report 01-041). Decantation was not an acceptable
alternative to centrifugation in this study because normal gravitational methods of
particle sedimentation take an inordinate amount of time, and for particles finer
than < 0.5 m Brownian motion interferes with settling (Folk, 1974; Syvitski,
1991). For each sample, three sedimented samples were prepared on glass
slides and allowed to dry slowly to produce orientated deposits. One was heat-
treated for 30 min at 550C, one was left in a desiccator with ethe lene glycol at
40C for 24 h, and one was kept in a desiccator for X-ray diffraction.

21
3.1.2 Identification of clay minerals from X-ray diffraction

Step scans of 4 s per step of 0.05 2? from 2.50 2? to 16.00 2? were done on
air-dried, glycolated and heat-treated samples using a Phillips PW 1710 X-ray
diffractometer with the following settings:

Tube anode material Cobalt

Generator potential 40 kV

Wavelength Ka1 1.78896

Wavelength Ka2 1.79258

Intensity ratio I Ka1/I Ka2 2

Divergence diaphragm 1
Detector diaphragm 0.1 mm

The changes in peak position on swelling and heating were used to confirm
mineral identification (Velde, 1995). X-ray powder diffraction analyses were
conducted on each sample using side loaded aluminium sample holders to
determine the overall mineralogical composition.

3.2 Analytical methods for the determination of F -

3.2.1 Background

Various methods have been reported for the determination of F- in aqueous

solutions, such as spectrophotometric eg. the SPADNS-method (Crosby et al.,
1968), conductometric (Buffle et al., 1985), complexometric (Pickering, 1986;
Saha, 1993), ion chromatographic (IC), and potentiometric or fluoride ion
selective electrode (FISE) methods (Van den Hoop et al., 1996; Yuchi et al.,
1999; McCaffrey, 2001). From the methods listed above, FISE has been widely
used due to its simplicity and short analysis time. Recent developments in ion

22
chromatographic separation techniques have led to an improvement in the
determination of F- in terms of selectivity and sensitivity. Ion chromatography is
therefore increasingly used for the determination of various ions and in particular
the F- ion. In this study both FISE and IC have been used for the determination of
F- .

3.2.2 Reagent and standard solutions

3.2.2.1 Standards

A 1000 mg F-.L-1 sodium fluoride (NaF) stock solution was prepared using
deionised water. Standards at a required concentration range were prepared by
appropriate dilution of the stock solution.

3.2.2.2 Total ionic strength buffers (TISAB) for FISE determinations

TISAB III (Orion cat No 940911, Thermo Orion) was obtained from Thermo
Orion (Beverly, USA). This commercial product consists of a mixture of
ammonium chloride (NH4Cl), ammonium acetate (CH3COONH4), CDTA
(C 14H22N2O8 .H2O), and cresol red (C 21H18 O5S) in water. 2 mL of this buffer was
added to the test and standard solutions of 20 mL according to the
manufacturers recommended procedure for the determination of F -.

TISAB IV was prepared as follows: 84 mL concentrated HCl (36-38%), 242g Tris

(hydroxymethyl) aminomethane, and 230g sodium tartrate (Na 2C4H4 O62H2O)
were added to 500 mL deionised water, allowed to dissolve, and then made up to
1L in a volumetric flask with deionised water. An equal volume of TISAB IV was
added to standards or samples before analysis.

23
3.2.3 HPIC instrumentation

A Dionex DX-120 ion chromatographic system equipped with a Dionex anion

exchange column system (AG 14A + AS14A) was used for the determination of
F- concentrations < 0.1 mg.L -1 and also other anions leached from the clay
sorbents during adsorption tests. The standard HCO3-/CO32- (1.0 mM HCO3- +
3.5 mM CO32-) eluent recommended for anion determinations with this column
was used in all determinations.

3.2.4 FISE instrumentation

Orion F- ion selective electrodes were used for routine determination of F- for
concentrations > 0.1 mg.L -1. The fluoride concentration in each of the prepared
solutions containing one of the two buffers were measured using a fluoride ion
selective electrode (Orion 9609, Orion Research Inc., Beverly, MA) connected to
a single junction calomel reference electrode (Orion Model 9609) and pH/ISE
meter (Orion Model 520A, Orion Research Inc., Beverly, MA).

3.2.5 The role of total ionic strength buffers in F - determinations

One of the major problems encountered in the analysis of total F- concentrations

is the impact of metal ions such as aluminium, iron, magnesium and calcium on
the analytical response. Due to binding of F- with these metal ions, only a fraction
of the total F- concentration is determined. Elimination of these interferences may
be achieved by addition of metal complexing agent. A Total Ionic Strength
Adjustment Buffer (TISAB III or TISAB IV) containing complexing agents was
therefore added to standards and samples. This buffer ensures:

that a constant ionic strength is maintained in test solutions and

standards. This is important because the F- electrode actually

24
 measures activity and not concentration. The measured result is
therefore strongly dependent on the ionic strength of the medium
because the activity a, given by the equation below,

a = F[F-]

will depend on the activity coefficient of the F- ion, F at the particular

ionic strength of the solution.

that the pH is kept constant at ca. 5.5. This is achieved by an acetic

acid buffer in the TISAB III solution and a tartrate buffer in the case of
TISAB IV. It is important to maintain the pH of samples and standards
above 4 because increasing protonation of F- at pH < 4 could lead to a
negative error. The F- sensitive membrane does not respond to F-
when it is in its protonated form, HF.

that F- is released from its metal ion complexes if present. Of particular

importance are AlF 2+ and AlF 2+. TISAB III contains CDTA (cyclohexane
diamine tetra acetic acid) that forms a very strong complex with metal
ions such as Al3+ replacing the F- in the process. For very high levels of
Al3+ and other metals forming complexes with F-, TISAB IV was used.
TISAB IV contains hydroxymethyl aminomethane as complexing agent.
In the case of low concentration F- determinations a special low level
TISAB solution is recommended because TISAB III and IV do not
produce accurate results when applied under these conditions. The
low level TISAB, however, does not contain a complexing agent such
as CDTA and can therefore not compensate for interferences caused
by complexation of F - with metals such as Al3+ .

25
FISE offers linear calibration graphs for concentrations larger than 0.1 mg.L -1.
Below 0.1 mg.L -1, curvature in the calibration curve increases and it becomes
more difficult to compensate for electrode drift and special procedures need to be
followed to produce accurate results. These procedures include a 20 min
equilibration of the membrane in 0.1 M F- solution prior to calibration and the
analysis. Even then the electrode requires more than 10 min to stabilise after
each transfer to a new sample or standard. Groundwaters in the study area in
most cases have pH values > 7. In alkaline waters the presence of metal cations
such as Al3+ that can form complexes with F- is unlikely and the inclusion of
CDTA in the TISAB solution is not mandatory. In this study F- determinations
were also required in acidic solutions where Al3+ leached from the clay substrates
could interfere with the measurement. This would lead to an underestimation in
the F- concentration. In this study TISAB III and IV were therefore used as ionic
strength adjustment buffers.

3.3 Comparison of the FISE and IC methods

3.3.1 Background

Consistent differences in the determination of F- in natural waters using FISE and

HPIC have been reported (McCaffrey, 1994) using the older generation of anion
exchange columns and F- membrane electrodes. One of the problems with these
columns using the standard HCO3-/CO32- eluent was that the F- peak appeared
immediately after the water dip which made the accurate determination of the
baseline difficult. The Dionex AG14A + AS14A colum n systems used in this study
produced a F- peak well separated from the water dip. To ensure accurate
analytical results it was nevertheless decided to perform a limited study to
compare the two techniques for F- determination in the test solutions
encountered in this study. The aim of this study was to compare the results of F-
determination in standard solutions, deionised and tap water, and clay extracts
samples as obtained by the two methods

26
3.3.2 Experimental procedure

The procedure used in this comparative study included the following steps. The
two techniques were calibrated using exactly the same set of standards. In case
of FISE, however, TISAB IV was added to both samples and standards.

Method:
Spiked samples of F- in deionised water, tap water and solutions obtained from
clay extracts were prepared as follows: 25 mL of water was added into a plastic
beaker and spiked with 0, 0.1, 0.2, 0.5, 1, 2, 3, 5, 7, 10 mg.L -1 F-. Volumes of
water equal to the volume of spiking solution needed for a specific concentration
were removed prior to spiking to keep the volume constant. For clay extract
preparation, 300 mL of deionised and tap water was added to 10g of clay. The
mixture was stirred for 1h, centrifuged for 5 minutes and decanted. 25 mL of
extract was added into plastic beakers and spiked in the same way as with
deionised and tap water.

3.3.3 Results

3.3.3.1 Comparison of calibration curves

The calibration curves obtained for 0.1, 0.2, 0.5, 1, 2, 5 and 10 mg.L -1 F-
standards for the two methods are shown in the Fig. 3.1 and 3.2 below.

27
 Figure 3.1
Fluoride calibration curve using NaF dissolved in MiliQ water and IC.

The calibration curve for ion chromatography shows excellent linearity as

indicated by correlation coefficient of 0.9995.

80
y = -54.317x + 62.173
60 2
Pot (mV)

R = 0.9998
40
20

0
0 0.5 1 1.5
-
Log [F ]

Figure 3.2
Fluoride calibration curve using NaF dissolved in MilliQ water and FISE.

The calibration curve above shows a sensitivity (response factor) of 54.317 mV

per decade and excellent linearity, as indicated by the correlation coefficient of
0.9998. This calibration curve was combined with calibration values for lower F-

28
concentrations (0.2 and 0.5 mg.L -1 F-) on a graph paper. Determinations of lower
F- concentrations were extrapolated from the combined graph.

Table 3.1 shows the results obtained by these two methods and in Fig. 3.3 ISE
results using TISAB IV buffer are plotted against F- concentrations obtained by
IC. It can be seen that the results are in good agreement. The plot of HPIC
versus FISE, which gives a straight line obtained at a slope of 1, indicates that
the two methods are comparable. The detection limit for ISE using the linear
calibration line section was 0.1 mg.L -1 F- and for IC 0.03 mg.L -1 F-. Because of
the higher detection limit for ISE, IC was used for samples where low residual F-
concentrations where expected. In general, however, ISE was used because the
method is a simple, accurate, quick and economical way to determine F-
concentrations.

TABLE 3.1
F concentrations (mg.L-1) determined by IC and FISE
-

(TISAB IV)
ISE IC
- -
Known F conc Measured F Measured F-
(mg.L-1) (mg.L-1) (mg.L-1)
0.2 0.22 0.26
0.5 0.54 0.53
1 0.99 1.24
2 2.10 2.13
5 4.90 5.08
10 10.0 9.93

29
 10

F /mgL determined by IC
y = 1.0107x - 0.1041
8
R 2 = 0.9991
6

2
-1

0
-

0 2 4 6 8 10
-1
F-/mg.L determined by ISE

Figure 3.3
F- concentrations determined by ISE using TISAB IV vs determination
by IC

3.3.3.2 Comparison of F- determination in spiked samples of deionised water, tap

water and solutions obtained from clay extracts.

(i) Deionised and tap water

Test solutions obtained from spiking deionised water, tap water and aqueous
extracts from selected clay adsorbents were analysed in triplicate by both
methods. The results of spiking deionised and tap water obtained by the two
methods are summarised in Table 3.2. Tap water contained 0.09 mg.L -1 F- and
the measured samples were therefore corrected by subtracting this amount.

30
 TABLE 3.2
F- determined in spiked water samples by IC and ISE using TISAB III and TISAB IV.
Concentrations in mg.L-1
ISE
TISAB III
Matrix Deionised water Tap water
Spiked F- F- in SD %recovery % error F- in SD %recovery % error
-1 -1 -1
in mg.L mg.L mg.L
0.1 0.10 0.001 100.0 0.0 0.09 0.002 90.0 -10.0
0.2 0.20 0.003 100.0 0.0 0.18 0.001 90.0 -10.0
0.5 0.49 0.001 98.0 -2.0 0.49 0.002 98.0 -2.0
1 1.01 0.003 101.0 1.0 0.96 0.003 96.0 -4.0
2 2.01 0.017 100.5 0.5 1.95 0.029 97.5 -2.5
3 2.93 0.016 97.7 -2.3 2.89 0.008 96.3 -3.7
5 5.00 0.014 100.0 0.0 4.92 0.014 98.4 -1.6
7 6.94 0.019 99.1 -0.9 6.81 0.098 97.3 -2.7
10 10.07 0.112 100.7 0.7 9.77 0.028 97.7 -2.3
ISE
TISAB IV
Matrix Deionised water Tap water

Spiked F- F- in SD %recovery % error F- in SD %recovery % error

in mg.L-1 mg.L-1 mg.L-1
0.1 0.10 0.001 100 0.0 0.1 0.002 100.0 0.0
0.2 0.20 0.002 100 0.0 0.2 0.002 100.0 0.0
0.5 0.50 0.002 100 0.0 0.5 0.004 100.0 0.0
1 1.01 0.001 101 1.0 0.98 0.001 98.0 -2.0
2 1.98 0.003 99 -1.0 1.99 0.015 99.5 -0.5
3 3.01 0.01 102 0.3 2.96 0.011 98.7 -1.3
5 4.96 0.02 99.2 -0.8 4.95 0.02 99.0 -1.0
7 6.99 0.01 99.9 -0.1 6.96 0.032 99.4 -0.6
10 9.98 0.03 99.9 -0.2 9.98 0.025 99.8 -0.2

31
 ION CHROMATOGRAPHY

Matrix Deionised water Tap water

Spiked F- F- in SD %recovery % error F- in SD %recovery % error
in mg.L-1 mg.L-1 mg.L-1
0.1 0.10 0.002 100.0 0.0 0.09 0.002 90.0 -10.0
0.2 0.21 0.001 105.0 5.0 0.18 0.004 90.0 -10.0
0.5 0.49 0.001 98.0 -2.0 0.53 0.001 106.0 6.0
1 0.99 0.003 99.0 -1.0 0.99 0.002 99.0 -1.0
2 2.03 0.005 101.5 1.5 1.97 0.003 98.5 -1.5
3 3.01 0.002 100.3 0.3 2.96 0.002 98.7 -1.3
5 4.95 0.01 99.0 -1.0 4.97 0.011 99.4 -0.6
7 7.00 0.009 100.0 0.0 7.01 0.02 100.1 0.1
10 10.02 0.021 100.2 0.2 9.95 0.024 99.5 -0.5

It can be seen that the spike recovery values obtained are in the range of 90-
100%. Both methods give good and stable fluoride recoveries when using water
samples. These results suggest that either TISAB III and TISAB IV were suitable
for the accurate determination of fluoride in water samples. Typical
chromatograms of water samples spiked with 0.2 mg.F -.L-1 are shown in Figure
3.4 and 3.5 below.

32
 Figure 3.4
A sample chromatogram of spiked deionised water (0.2 mg F-.L-1)
determined by ion chromatography. Concentrations (mg.L-1): 1. F- =
0.21, 2. Cl- = 0.02 and 3. SO42- = 0.01.

Figure 3.5
A sample chromatogram of spiked tap water (0.2 mg F-.L-1) determined
by ion chromatography. Concentrations (mg.L-1): 1. F- = 0.27, 2. Cl- =
42.56, 3. Br- = 0.02, 4. NO3- = 2.17 and 5. SO42- = 13.54.

In Figures 3.6 and 3.7 below ISE results for spiked deionised and tap water
samples using TISAB III and TISAB IV buffers are plotted vs F- concentrations
obtained by IC. Both figures show a straight line with correlation coefficient
between 0.9999 and 1. The results show a good correlation at F- concentrations

33
between 0.1 and 10 mg.L -1. Both techniques are therefore capable of analysing
correctly in relatively undemanding matrices such as deionised and tap water.

F-/mg.L-1 determined by
12
y = 0.9996x - 0.0021
10 2
R = 0.9999
8
T III
ISE 6
T IV
4
2
0
0 2 4 6 8 10 12
- -1
F /mg.L determined by IC

Figure 3.6
-
F concentrations in spiked tap water samples determined by ISE using
TISAB III and IV vs determinations by IC.

The general observation is that application of the ISE using TISAB III and
IC procedures leads to lower F- values at low F- concentrations in tap
water.
F /mg.L determined by

12
y = 0.9971x + 0.0014
10
R2 = 1
8
T III
ISE

6
T IV
4
-1

2
0
-

0 2 4 6 8 10 12
- -1
F /mg.L determined by IC

Figure 3.7
-
F concentrations in spiked deionised water samples determined by
ISE using TISAB III and IV vs determination by IC

34
(ii) Clay extracts

Extracts were prepared by extracting 10g of clay with 300 mL deionised water
and tap water for 1h. Samples were centrifuged for 5 minutes and pH of each
sample was measured. Blank values were determined before spiking with F-.
Table 3.3 gives a summary of pH of four clay samples used. The clay samples
are described in Chapter 5.

TABLE 3.3
pH of clay extracts in deionised and tap water
determined after stirring for 1 hour .
Clay samples pH determined for each extract.

Deionised water Tap water

ALU 4.80 5.45
RBM 9.64 8.25
MD2 5.95 7.42
CAL 9.27 8.71

Table 3.3 shows pH values for each sample extract. In case of clay samples
(ALU and MD2) the pH of deionised water is lower than that of tap water. For
clay samples (RBM and CAL) the pH of deionised water is higher than that of tap
water. It is a well known fact that pH plays an important role in the determination
of F - in sample matrices.

Accurate determination of F-, however, becomes rather challenging in more

complex matrices such as the extracts obtained from adsorption experiments on
clay samples. Table 3.4 compares the results obtained by analysing spiked
extracts obtained from the sorbents ALU (activated alumina), RBM, MD2 and

35
 CAL. The RBM extracts contained 2.58 mg.L -1 F-, MD2 extract contained 0.09
mg.L -1 F-, CAL extract contained 0.55 mg.L -1 F-, ALU extract contained 0.04
mg.L -1 F- and the measured samples were therefore corrected by subtracting this
amount.

TABLE 3.4
-
F determined in spiked clay extract samples by IC and ISE using TISAB III and TISAB IV.
Concentrations in mg.L-1
RBM TISAB III TISAB IV IC
- - - -
Spiked F F in %recovery % error F in %recovery % error F in %recovery % error
-1 -1
in mg.L mg.L mg.L-1 mg.L-1
0.1 0.12 120.0 20.0 0.12 120.0 20.0 ND ND ND
0.2 0.3 150.0 50.0 0.22 110.0 10.0 0.74 370 270
0.5 0.49 98.0 -2.0 0.54 108.0 8.0 1.13 226 126
1 0.97 97.0 -3.0 1.10 110.0 10.0 1.47 147 47
2 1.90 95.0 -5.0 2.13 106.5 6.5 2.45 122.5 22.5
3 3.00 100.0 0.0 3.26 108.7 8.7 2.96 98.7 -1.3
5 5.23 104.6 4.6 5.33 106.6 6.6 4.84 96.8 -3.2
7 7.69 109.9 9.9 7.55 107.9 7.9 6.60 94.3 -5.7
10 11.03 110.3 10.3 10.25 102.5 2.5 8.61 86.1 -13.9

ALU TISAB III TISAB IV IC

- - - -
Spiked F F in %recovery %error F in %recovery %error F in %recovery %error
-1 -1 -1
in mg.L mg.L mg.L mg.L-1
0.1 0.06 55.0 -45.0 0.09 90.0 -10.0 0.06 60.0 -40.0
0.2 0.08 40.5 -59.5 0.16 80.0 -20.0 0.13 65.0 -35.0
0.5 0.29 58.0 -42.0 0.41 82.0 -18.0 0.35 70.0 -30.0
1 0.60 60.0 -40.0 0.83 83.0 -17.0 0.73 73.0 -27.0
2 1.10 55.0 -45.0 1.70 85.0 -15.0 1.41 70.5 -29.5
3 1.06 35.3 -64.7 2.71 90.3 -9.7 1.93 64.3 -35.7
5 3.84 76.8 -23.2 4.36 87.2 -12.8 4.06 81.2 -18.8

36
 7 5.86 83.7 -16.3 6.30 90.0 -10.0 5.57 79.6 -20.4
10 8.51 85.1 -14.9 8.60 86.0 -14.0 85.6 85.6 -14.4

MD2 TISAB III TISAB IV IC

Spike F- F- in %recovery %error F- in %recovery %error F- in %recovery %error
in mg.L-1 mg.L-1 mg.L-1 mg.L-1
0.1 0.06 60.0 -40.0 0.06 60.0 -40.0 0.05 50.0 -50.0
0.2 0.11 55.0 -45.0 0.11 55.0 -45.0 0.11 55.0 -45.0
0.5 0.23 46.0 -54.0 0.33 66.0 -34.0 0.34 68.0 -32.0
1 0.62 62.0 -38.0 0.72 72.0 -28.0 0.7 70.0 -30.0
2 1.47 73.5 -26.5 1.41 70.5 -29.5 1.59 79.5 -20.5
3 2.40 80.0 -20.0 2.41 80.3 -19.7 2.46 82.0 -18.0
5 4.60 92.0 -8.0 4.41 88.2 -11.8 4.52 90.4 -9.6
7 6.60 94.3 -5.7 6.18 88.3 -11.7 6.35 90.7 -9.3
10 9.90 99.0 -1.0 8.69 86.9 -13.1 8.85 88.5 -11.5

CAL TISAB III TISAB IV IC

Spike F- F- in %recovery %error F- in %recovery %error F- in %recovery %error
in mg.L-1 mg.L-1 mg.L-1 mg.L-1
0.1 0.10 100.0 0.0 0.08 80.0 -20.0 0.03 30.0 -70.0
0.2 0.13 65.0 -35.0 0.16 80.0 -20.0 0.16 80.0 -20.0
0.5 0.49 98.0 -2.0 0.43 86.0 -14.0 0.40 80.0 -20.0
1 1.00 100.0 0.0 0.62 62.0 -38.0 0.78 78.0 -22.0
2 1.91 95.5 -4.5 1.17 58.5 -41.5 1.30 65.0 -35.0
3 2.88 96.0 -4.0 2.26 75.3 -24.7 2.29 76.3 -23.7
5 5.76 115.2 15.2 4.50 90.0 -10.0 4.76 95.2 -4.8
7 8.02 114.6 14.6 6.00 85.7 -14.3 6.21 88.7 -11.3
10 11.17 111.7 11.7 8.92 89.2 -10.8 8.95 89.5 -10.5

Chromatograms of the extracts were recorded before analysis of the spiked

extracts. A sample chromatograph in Figure 3.8 shows the anions present in the

37
CAL extract. using deionised water. These include anions such as: F-, Ac-, Cl-,
NO2-, Br-, NO3-, PO43- and SO42-.

Figure 3.8
Chromatogram for the CAL extract using sodium
carbonate/bicarbonate as eluent and an AS4A anion exchange
column. Concentrations (mg.L-1): 2. F- = 0.55, 3. Ac - = 0.04, 4. Cl- =
1.0, 5. NO2- = 0.05, 6. Br- = 0.01, 7. NO3- = 0.03, 8. PO43- = 0.03 and 9.
SO42- = 9.63.

A chromatograph for the RBM extract spiked with 0.2 mg F -.L-1 is shown in Figure
3.9. The chromatograph shows excellent resolution where all anions elute away
from the water dip and at retention times > 4 min. Unresolved peaks, however,
occur between fluoride and acetate. It can be seen that acetate peak elutes next
to fluoride, hence the fluoride concentration is not easily determined. Chloride,
nitrate and sulphate were both resolved with sharp peaks. Nitrite, bromide and
phosphate occurred in lower concentrations.

The accurate determination of F- in these matrices proved to be problematic

when using IC. Lower F- concentrations from 0.1 to 2 mg.L -1 were over estimated
and higher concentrations underestimated. It should be noted that the presence
of anions such as acetate, chloride, nitrite, bromide, nitrate, phosphate and
sulphate can interfere with the F- determinations. The presence of polyvalent

38
cations (Al3+, Fe3+, Mg2+, and Ca2+) may also cause interferences under
operating conditions of the IC, as it does with the ISE method. Typical
concentrations of these polyvalent cations in the clay extracts were found to be in
the range of 1-4 mg.L -1 aluminium, 0-5 mg.L -1 calcium, 0-11 mg.L -1 magnesium
and 0-0.5 mg.L -1 iron.

Figure 3.9
A chromatogram for the RBM extract using sodium
carbonate/bicarbonate eluent and an AS4A anion exchange column
after spiking with (2 mg.F-.L-1). Concentrations (mg.L-1): 1. F- = 3.41, 2.
Ac - = 0.16, 4. Cl- = 0.81, 5. NO2- = 0.09, 6. Br- = 0.01, 7. NO3- = 5.17, 8.
PO43- = 0.03 and 9. SO42- = 3.33.

The ion selective electrode method of fluoride determination has found wide
spread use. However, when applied to samples in which aluminium
concentrations may be high (clay samples) the buffer systems (TISAB buffers)
were not effective in combating aluminium interference. Some clay samples
contained sufficient Al3+, Fe3+ , Ca2+ and Mg2+ to interfere seriously in the
determination of fluoride ions when using the fluoride ion selective electrode. In
the presence of these metals the spiking method therefore gave erroneous
results.

39
In comparison with IC, the results seem to improve when using ISE. This is
because of the availability of TISAB being used. TISAB III contains CDTA
(cyclohexane diamine tetra acetic acid) and TISAB IV contains Tris
(hydroxymethyl) aminomethane (THA) as complexing agents. The success of the
ISE method depends on the effectiveness of the system used to buffer pH and
ionic strength and to complex possible interferences such as Al3+, Mg2+ , Ca2+ and
Fe3+. The F- recovery using these reagents depends on the system under
investigation. For instance, both CDTA and THA yield quantitative results of F-
analysis in deionised and tap water. In more complex matrices, such as clay
extracts, the efficiency of these complexing agents seems to decrease. In case of
TISAB III, quantitative F- recoveries were not obtained. The results obtained
using ISE with TISAB IV buffer produced the most accurate results and were
used as standard analytical procedure in this study.

Chemical interferences are difficult to eliminate completely. Fluoride readily forms

complexes with many ions in solution. In order to measure total fluoride present,
such complexes need to be destroyed. Polyvalent cations (Al3+, Fe3+, Mg2+ , and
Ca2+) complexes F- and it becomes necessary to destroy these complexes.
Aluminium as was already mentioned presents a significant interference even at
levels below 0.2 mg Al / L. Considering stability constants of aluminium fluoride
complexes, it is expected that part of the F- in the original sample will be bound
by aluminium. Obviously, the added decomplexing agents hydroxymethyl
aminomethane for TISAB IV and CDTA (cyclohexane diamine tetra acetic acid)
for TISAB III were not able to fully decomplex F- bound by metals. This means
the decomplexing agents applied are not able to decomplex F- completely from
its aluminium complex form. Thus, for clay extracts which contain aluminium,
measured F- concentrations obtained by using ISE method were a fraction of the
total F - concentration.

The chemical states and dissociation behaviour of fluoride have been thoroughly
investigated. In an aqueous solution, fluoride has three dissolved forms,

40
depending on the pH: which are HF, H2F2 and F-. Among these dissolved forms,
the fluoride ion (F -) is the most dominant form in water with pH values ranging
from 3 to 9. However, when tri or tetravalent cations of certain metallic elements
such as aluminium or iron exist along with fluoride ions, complexes with fluoride
ions are readily formed. Aluminium fluoride is one of the stable complexes which
can be formed with fluoride anions. Several different forms of aluminium fluoride
can be found with various F-/Al3+ atomic ratios. One aluminium ion (Al3+) in the
complex structure is surrounded by a maximum of six octahedrally arranged
fluoride ions. This means that the presence of one aluminium ion could form a
complex with six free fluoride ions. This would, however, not occur at the
concentrations levels encountered in this study.

Therefore, the significant interference by aluminium ions during fluoride

determinations using both IC and ISE was caused by the strong affinity of
fluoride ions for aluminium ions. Although both decomplexing agents used
obviously dissociated aluminium ions from the aluminium fluoride complexes, this
process was not complete.

The interference of aluminium in fluoride ion determinations can also be

attributed to the formation of insoluble aluminium fluoro-species. Aluminium
forms colloidal hydrous oxide particles in the pH range 4-9. This solid can
strongly adsorb fluoride, resulting in low fluoride measurements with both IC and
ISE.

Anion interferences are particularly problematic when using ion chromatography.

These interferences are normally caused by anions with retention times that are
close to each other. This could lead to an overlap of one anion to the anion of
interest.

41
For determinations based on anion chromatography, much remains to be learned
about aluminium interference effects. With excess of aluminium present, low
values seem inevitable unless the aluminium is removed.

3.3.3.3 Effect of acetate on F - determination with IC

The effect of acetate or other organic species extracted from the clays and that
elute close to F- was investigated by spiking activated alumina (ALU) extracts
with 1 mg.L -1 fluoride and 5 mg.L -1 acetate. Table 3.6 shows results obtained with
ISE using TISAB III and IV and IC.

TABLE 3.6
- -1 - -1 -
F determined in ALU extract spiked with 1 mg.L F and 5 mg.L Ac by IC and ISE using
TISAB III and IV.
DEIONISED WATER
ALU TISAB III TISAB IV IC
- - - -
Spiked F F Recovery Error F Recovery Error F Recovery Error
-1 -1 -1 -1
mg.L mg.L (%) (%) mg.L (%) (%) mg.L (%) (%)
1 0.59 60 -40 0.62 62 -38 0.46 46 -54

TAP WATER
TISAB III TISAB IV IC
- - - -
Spiked F F Recovery Error F Recovery Error F Recovery Error
-1 -1 -1 -1
mg.L mg.L (%) (%) mg.L (%) (%) mg.L (%) (%)
1 0.76 76 -24 0.67 67 -33 0.52 52 -48

The results show lower fluoride concentrations with both IC and ISE. Consistent
lower F - concentrations with both methods might also be attributed to the effect of
aluminium present in the samples. Ion chromatography studies show that the
acetate interferes with fluoride as the two anions elute close to each other. The
chromatogram in Figure 3.10, shows the effect of acetate on the fluoride peak.

42
All anions elute well away from the water dip. From the chromatogram, the
interference of acetate with fluoride peak is clear as the two peaks are next to
each other. The presence of acetate in the sample results in the underestimation
of the fluoride. Therefore, it is difficult to obtain accurate and quantitative fluoride
measurements in the presence of acetate. The retention times of other anions
also seem to differ when large amounts of acetate were present.

Figure 3.10
Chromatogram showing interference of acetate with the fluoride peak
in an activated alumina (ALU) extract spiked with 1 mg.L-1 F- and 5
mg.L-1 acetate. Concentrations (mg.L-1): 1. F- = 0.48, 2. Ac - = 5.63, 3.
Cl- = 23.5, 4. NO3- = 0.14, 5. PO43- = 0.03 and 6. SO42- = 0.65.

3.3.4 Conclusions

A good agreement for F- determinations was found for the two methods. The
results emphasise the point that the determination of fluoride depends on the
matrix characteristics. Consistently lower F- concentrations were found in spiked
clay extracts. The presence of Al in the samples resulted in lower F- values using
both techniques. Acetate has also proved to be a major interference on fluoride
determinations. In the analysis of clay extracts in this study care was taken to
develop efficient precleaning procedures to reduce the presence of interfering
ions on F - determinations. See pre-treatment procedures in Section 3.4.2.

43
3.4 Experimental methods

3.4.1 Determination of elemental composition

3.4.1.1 Solid samples

The major element composition of selected clay samples was determined by

XRF by Set Point Technologies.

3.4.1.2 Solutions

Inductively coupled plasma optical emission spectrometry (ICP-OES) was used

to determine the elemental composition in solution samples. A method was
developed for the determination of the elements: Al, Fe, Mn, Cu, Ca, Mg, Cr, Mn,
Ni, Zn, and As using a Varian Liberty 100 ICP-OES spectrometer. A 1000 mg.L-1
Multi IV (Merck SA) multi-element ICP-OES standard was used to prepare
calibration standards by appropriate dilution with 2% HNO3 .

The analytical wavelengths and experimental conditions used in the ICP-OES

measurements are summarised in Table 3.7 and Table 3.8, respectively.

44
 TABLE 3.7
Analytical wavelengths for the determination of metals
in solution by ICP-OES
Elements Analytical Instrumental
Wavelength detection limit
(nm)
Al 396.153 0.10
Fe 259.940 0.05
Cu 324.754 0.02
Ca 393.366 0.05
Mg 279.553 0.05
Cr 267.716 0.05
Mn 257.610 0.02
Ni 231.604 0.05
Zn 213.856 0.05
As 188.979 0.05

TABLE 3.8
ICP-OES operational conditions
Parameter Setting
Generator power 1.3 kW
Plasma gas flow 15.0 L.min-1
Nebuliser pressure 150 kPa
Sample uptake rate 1 mL.min-1
Integration time 3s
Viewing height 2-8 mm
PMT voltage 450-650 V
Order 2

45
3.4.2 Pretreatment procedures

3.4.2.1 Wash procedure to remove water extractables

All clay samples were extracted by shaking with deionised water in a 1:10
mass(g) to volume(mL) ratio for 10 min, centrifuged at 3900 rpm for 5 min and
decanted. Extraction was repeated until clays were free from water extractable
ions indicated by the absence of Cl- in ion chromatograms of the extracts. The
washed clay was dried overnight at 105 C, pulverised in an agate mortar and
stored in a desiccator.

3.4.2.2 Heat treatment

Clay samples, placed in porcelain crucibles, were heat-treated at the desired

temperature in a preheated muffle furnace for 2 h. To study the effect of heating
clays at different temperatures, the same procedure was followed through the
temperature range 200 to 900oC in steps of 100oC. A temperature of 600 C was
selected for heat treatment. After heat treatment clays were cooled, pulverised
and stored in a desiccator.

3.4.2.3 Chemical pretreatment

Clay samples containing F- were chemically treated to remove F- through an ion

exchange process with hydroxide. Chemical treatment was also used to activate
clay surfaces before F- adsorption. In the chemical pretreatment procedure clays
were stirred with 0.1 M NaCO3 for 30 min in a mass(g) to volume (mL) ratio of 1
to 5, centrifuged for 5 min and decanted. The residue was washed with deionised
water to remove excess sodium carbonate and again centrifuged and decanted.
The residue was then treated with 1% HCl for 30 min in a mass(g) to volume(mL)
ratio of 1 to 5 and rinsed until Cl--free. The rinsed solution was tested with

46
AgNO3 solution initially and finally with HPIC. The residue was dried overnight at
105 C, pulverised in an agate mortar and stored in a desiccator.

3.4.2.4 Chemical treatment for alkaline clays

Alkaline clays such as clays containing substantial amounts of dolomite or Ca

and Mg carbonates, required modification of the chemical pretreatment
procedure. The acid treatment was excluded from the procedure described in
Section 3.4.2.3, because adding acid to dolomitic clays would cause dissolution
of the Ca and Mg carbonates according to the reaction:

CaCO3 + HCl ? Ca2+ + 2Cl- + H2 O + CO2

3.4.2.5 Sequential pretreatment procedure

In the sequential treatment, samples were first heated for 1 h at 600oC and then
treated chemically with Na 2CO3 and HCl as above.

Table 3.9 summarises the different pretreatment procedures and the notation
used in this study.

TABLE 3.9
Notation of pretreatment procedures
Procedure Notation
Removal of water extractables CR
Heat treatment C
Chemical treatment CT
Heat followed by chemical CTC
treatment

47
3.4.2.6 Substrate regeneration procedure

Regeneration of clay substrates was done using the same solutions, 0.1 M
Na2CO3 and 1% HCl as for chemical pretreatment.

3.4.3 Ammonium oxalate extraction

The ammonium oxalate extraction (Parfitt, 1989) removes materials without a

well-developed crystalline structure and is specifically designed to dissolve Al, Fe
and Mn from amorphous soil phases containing these elements. A sample is
shaken for 4 h in the dark with ammonium oxalate at pH 3.

0.5 g of clay sample was weighed into a 100 mL polyethylene flask which was
wrapped in aluminium foil to protect it from light induced reactions. 50 mL 0.2 M
ammonium oxalate at pH 3 was added. The flask was then placed in a shaker for
4 h. After filtration and appropriate dilution the Al, Fe and Mn concentrations were
determined in the filtrate by ICP-OES.

3.4.4 Batch adsorption procedures

3.4.4.1 Adsorption at pH 6

Adsorption capacities were determined at pH 6 by shaking 1 g of washed and

dried (2 h at 105oC) clay with 50 mL of 10 mg.F -.L-1 NaF solution (C 0) for 2 h in
polyethylene bottles at 22 C. The initial pH was adjusted to approximately 6
using NaOH or HCl depending on the acid or base properties of the sample. After
equilibration the solutions were centrifuged, the pH measured to ensure that the
pH was within 0.2 pH unit from the target pH, and the residual F- concentration
(C e) determined using ion chromatography or a F- ion selective electrode. The %
adsorption was calculated from the residual F - concentration using the equation:

48
 (CO Ce )
% Adsorption = 100 x
Co

To ensure that F- does not adsorb on the inner walls of the adsorption vessels,
blank runs were performed. In this procedure a 10 mg.F -.L-1 solution was added
to a polyethylene vessel and the F- concentration measured after 2 h and again
after 12 h. No reduction in F - concentration was found.

3.4.4.2 Adsorption curves (% adsorption vs pH)

The same method as described in Section 3.4.4.1 was followed for equilibrium
solution pH values of about 3, 4, 5, 6, 7, 8, and 9.

3.4.4.3 Competitative adsorption

The effect of other anions on the adsorption of F- was studied by determining the
residual F- concentration in the presence of Cl- and SO42-. The same procedure
was used as in Section 3.4.4.1.

3.4.4.4 Determination of adsorption kinetics

Adsorption kinetics were determined using the batch adsorption procedure

described in Section 3.4.4.1. Residual F- concentrations were, however,
measured at the time intervals 10, 20, 30, 45, 60, 120, and 300 min.

3.4.5 Zeta potentials

Zeta potentials were determined by shaking 1g of clay with 50 mL of deionised

water. The pH was adjusted to desired values (3-9) using 0.1 M NaOH or HNO3

49
depending on the acid or base properties of the sample. Zeta potentials were
measured using a Malvern ZetaMaster.

3.4.6 Determination of adsorption isotherms

Adsorption isotherms were determined using chemically pretreated clays. 1 g of

clay was equilibrated in 45 mL 0.1 M NaClO4 at pH 6 for 14 h. Appropriate
amounts of a 1000 mg.L -1 F- stock solution and 0.1 M NaClO4 were then added
to the equilibrated samples to prepare 50 mL test solutions with initial
concentrations: 2, 5, 10, 20, 50, 100, and 200 mg.F -.L-1. These solutions were
equilibrated for 2 h to complete the adsorption process and the residual fluoride
concentration and final pH determined.

3.4.7 Determination of pH curves

Clay samples of 1 g each were equilibrated for 14 h in 50 mL 0.1 M NaClO4

solution at pH 3 to 9. After equilibration 0.5 mL of a 1000 mg.L -1 F- stock solution
was added so that the initial F - concentration was 10 mg.L -1.

The pH was adjusted with 1M HClO4 or NaOH. The pH was measured at 1h after
pH adjustment (pH1h), again after 14h equilibration (pHi), and then at different
times after addition of 10 mg.L -1 fluoride (For example pH15 denotes 15 m after
addition of F -).

3.4.8 Kinetics of adsorption and pH equilibration

Kinetics of equilibration were determined by weighing 1g of clay and add 47 mL

of deionised water and 2.5 mL of 1 M NaClO4 . The pH was adjusted with 1 M
HClO4 or NaOH. The pH was measured before adding fluoride (pHi). Fluoride
solution of 10 mg.L -1 (0.5 mL of 1000 mg.L -1 F-) was added. The pH was then

50
measured at different times after addition of fluoride using the same method as
described in Section 3.4.7

3.4.9 Leaching procedures

Heavy metals leached from clays during the F- adsorption procedure were
determined by ICP-OES. 1g clay sample was extracted with 50 mL of F- solution
(1 and 10 mg.L -1) at pH 6 for 2 h. The suspension was centrifuged and the final
pH was measured. The extracted metals determined in the solution after
decanting by ICP-OES.

3.4.10 Particle sizes

1g of clay was suspended in 50 mL of deionised water. The samples were

subjected to ultrasonic treatment for 2 min. Samples were then left for 5 min for
particles to settle and the particle size distribution was measured using a Malvern
ZetaMaster. Larger particles were determined using Coulter particle size counter
(PAMAS 3108 FM) in differential mode.

In differential presentation the measured results for a given channel include the
diameter of that channel as well as the total number of particles of all larger sized
particles, but smaller than the size of the higher channel.

3.4.11 Coated particles

3.4.11.1 Preparation of clays for coating experiments

The same method as in Section 3.4.12.1 was followed for preparing clays for
coating experiments.

51
3.4.11.2 Coating method

Coated samples were prepared by mixing a required amount of clay with 1 M

AlCl3 in a 1:25 mass(g) to volume(mL) ratio. Then the same volume of 0.5 M
NaOH was added dropwise while stirring vigorously. The mixture was stirred for
2 hours, and then centrifuged and decanted. The same washing method as in
Section 3.4.2.1 was used to remove Cl-.

3.4.11.3 Adsorption method

The same method as described in Section 3.4.4.1 was followed to determine the
adsorption capacities of the coated samples.

3.4.12 Experiments with mini columns

3.4.12.1 Preparation of clays for column experiments

Clay samples were washed with deionised water until F- free, dried at 105 C,
compacted in a Perkin Elmer pellet press at 5 t and pellet diameter 35 mm, heat-
treated at 600 C for 2 h, cooled in a desiccator, and crushed in a Siebtechnik
mill to particle size < 2 . This procedure was followed to stabilise the clay
samples to prevent loss of small particles during column operation.

3.4.12.2 Description of columns

Mini columns consisted of 180 x 14 mm glass tubes fitted with no 1 porosity frits
on both ends. The columns were connected to a Gilson Minipuls 3 peristaltic
pump using Tygon tubing. Columns were typically packed with 4 g sorbent. The
flow direction was from bottom to top. The pH of F- feed solutions were adjusted
to pH 6 to ensure proper conditions for adsorption to occur.

52
3.4.12.3 Determination of breakthrough curves and adsorption capacities

Columns were packed with 4 g dry sorbent (or other mass as indicated). The
columns were then equilibrated by pumping deionised water at pH 6 through the
column using a Gilson Minipuls 3 peristaltic pump. The flow rate was 4 mL.min-1.
F- feed solutions were adjusted to pH 6 before being pumped through the
columns. Samples were taken at regular intervals and the F- concentration
determined by FISE. Residual F- concentration was plotted against time.
Breakthrough times were determined graphically and used to calculate
breakthrough volumes and adsorption capacities of the columns.

3.5.13 Preparation of hematite

Hematite was prepared by dissolving 16g of Fe (NO3)3 . 9H2 O in 1 L of deionised

water. Sodium hydroxide (1 M NaOH) was added under stirring to raise the pH to
10.5 10.8. The resulting brown precipitate (Fe(OH)3) was washed with
deionised water 10 15 times until the pH reached ~ 9.3. After final rinsing cycle,
into about 200 mL of the dispersion 20 mL of 1 M HCl and 2.5 mL of 0.1 M
NaH2PO4 solutions were added. The solution was filled up with deionised water
to 2 L. The suspension was aged at 60 C for 2 days in an oven. The precipitate
was repeatedly washed on a Millipore filter (filter paper 0.45 m). Finally, the
precipitate was dried in an oven at 105 C and kept in a desiccator.

53
 CHAPTER 4
Adsorption mechanisms and adsorption modelling
for fluoride on to mineral substrates

4.1 Background

Adsorption of F- onto different sorbents including clays and soils has been
studied with regard to rate of adsorption (Wu et al, 1979), preheating of clays at
high temperatures (Hauge et al, 1994), effect of pH (Choi and Chen, 1979), and
the effect of concentration through the measurement of adsorption isotherms
(Chhabra et al, 1980; Chaturvedi et al 1988). Very few studies, however, focused
on gaining a proper understanding of the specific adsorption mechanisms
involved in F- adsorption onto materials, other than activated alumina (Hao et al,
1986).
The lack of a proper understanding and detailed description of the mechanism of
F- adsorption onto different sorbents of mineral origin prompted this systematic
study of the theoretical aspects of F- adsorption. Adsorption was studied in
relation to the chemistry and structural properties of the sorbent and the aqueous
solution chemistry of F-. The results include experimental evidence based on
adsorption isotherms, adsorption kinetics, adsorption curves, and changes in pH
during sorbent equilibration and F- adsorption. The evidence compiled in this way
provided the basis for the development of an adsorption modelling procedure.
This model was used to predict optimum solution conditions and best structural
properties of sorbents to achieve maximum adsorption capacity.

The specific objectives of the research reported on in this chapter are listed
below.

To investigate the mechanisms of F - adsorption in relation to the

mineralogical structure and surface chemistry of selected clay types.

54
 To model the F- adsorption process with the aim of predicting the
optimal surface properties of the sorbent.

4.2 Adsorption mechanisms

An ion exchange mechanism (Hao et al, 1986) whereby F - is exchanged with OH-
groups in the mineral structure, is generally assumed to be the rate determining
step in the adsorption process. This ion exchange process can be presented by
the following equilibria, here given for an aluminium oxide substrate in hydrated
form. Similar equilibria can be written for other metal oxides such as Fe oxides.
Underlined species denote the solid state:

AlOH2+ + F- ? AlF + H2O 4.1

AlOH + F- ? AlF + OH- 4.2

Easily exchangeable OH- groups are those found in metal hydroxides or hydrated
metal oxides such as aluminium or iron oxides, although it is known that not all
structural forms of metal oxides have exchangeable OH- groups with F-, notably
magnetite (Fe4 O6) and corundum (-Al2O3). Lattice OH- groups such as found in
aluminiumsilicates, for example kaolinite, have a low tendency to be replaced by
F- .

The pH of the system is a very important parameter and determines the degree
of protonation of the OH- exchange sites and the degree of protonation of F-. The
pH will therefore determine the specific charge of an exchange site and therefore
ultimately also the adsorption tendency of the substrate. This is because the
surfaces of clays generally contain pH dependable ionisable functional groups
which constitute the active exchange sites. If the clay contains metal oxides,
commonly Fe oxides and less frequently Al oxides, exchange sites consist of OH-

55
groups depending on the degree of hydration and the specific structural
characteristics of the mineral. For a metal oxide such as gibbsite, the OH- group
can undergo proton transfer in the following way.

AlOH2+ ? AlOH + H+ 4.3

AlOH ? AlO- + H+ 4.4

The dissociation constants for these reactions are defined in the same way as
acid dissociation constants for soluble acids:

[ AlOH ][ H + ]
K a1 = 4.5
[ AlOH 2+ ]

[ AlO ][ H + ]
Ka2 = 4.6
[ AlOH ]

The process of protonation (by adding acid) changes the substrate surface from
negative to positive charge and therefore also forms a neutral surface at a
specific pH, the point of zero charge given by the symbol pHZPC . Either a positive
surface when the pH < pHZPC or a negative surface when the pH > pHZPC can
therefore exist for each substrate depending on the pH of the medium. This pH at
the point of zero charge will differ from substrate to substrate depending on the
acid dissociation constants (pKa) of the functional groups. The pH at the point of
zero charge for metal oxide surfaces is given by the equation:

pHZPC = (pKa1+pKa2) 4.7

56
In Table 4.1 pKa values for selected metal oxide surfaces are listed. The strong
dependence of the numerical values of pKas on mineral structure even for
minerals with the same chemical composition is evident.

TABLE 4.1
Surface acidity in terms acid dissociation constants for mineral
oxides
Mineral pKa1 pKa2 pHzpc Reference
Activated alumina 4.2 10.5 7.4 Hao et al, 1986
Type F1
Activated alumina 6.0 9.4 7.7 Gosh and Yang,
Type F1 1984
-Al2O3 7.7 9.3 8.5 Huang and
Stumm, 1973
a-Al2O3 8.5 9.7 9.1 Yopps and
Fuerstenau, 1964
a-FeOOH 7.0 8.4 7.7 Atkinson et al,
1967
a-Fe2O3 8.9 9.8 9.3 Atkinson et al,
1967
a-SiO 2 2.0 7.2 3.1 Huertas et al,
1998

The distribution of the protonated and deprotonated surface species as a function

of pH can be calculated using fractions. Equations for fractions can easily be
derived from the dissociation constant expressions and are given below for
MOH2 +, MOH, and MO- species, typical for metal oxide surfaces.

57
 [H + ]2
MOH+2 = 4.8
[ H + ] 2 + | K a1[ H + ] + K a1 K a 2

[ H + ] K a1
MOH = 4.9
[ H + ] 2 + | K a1[ H + ] + K a1K a2

K a1 K a 2
MO = 4.10
[ H + ] 2 + | K a1[ H + ] + K a1 K a 2

The distribution curves for the surface species as a function of pH for Al, Fe, and
Si oxide surfaces are given in Figure 4.1. They were calculated in Excel using the
constants given in Table 4.1. The derivation of the equations and the numerical
spreadsheet results are given in Appendix A. It is clear from these distribution
diagrams that the number of protonated groups, and therefore the surface
positive charge, will steadily increase as the pH falls below the point of zero
charge. This means that the adsorption of F- onto an aluminium oxide surface
should steadily increase as the pH decreases from 8 to lower values. It is
important to note that the region of positive charge shifts to lower pH values in
the following sequence in metal oxides: Al>Fe>Si. In the case of Si, a positive
surface only occurs at low pH values, pH<2. Because silicate is a major
component in most clays the occurrence of negatively charged clay particles is
very common in the pH range (5 to 8) normally observed in natural waters. It is
understandable that clays preferentially adsorb positively charged metal ions and
are known to be good sinks for metal pollutants (Benjami n, 1981). F- on the
other hand is negatively charged in these pH ranges and will not be readily
adsorbed onto negatively charged clay surfaces. For F- to be adsorbed from
solution it is therefore important to maximise the number of positive sites in the
pH range 5 to 8. Figure 4.1 indicates that Al oxides give the highest
concentration of positive charge in the required pH range. The range in which

58
iron oxides have a positive surface extend to slightly lower pH values and would
therefore be more suitable to extract F - from more acidic waters.

1.0

Alpha fraction
0.8 + -
AlOH2 AlOH AlO
0.6
0.4
0.2
0.0
1 2 3 4 5 6 7 8 9 10 11 12
pH

1
Alpha fraction

0.8 + -
FeOH2 FeOH FeO
0.6
0.4
0.2
0
0 1 2 3 4 5 6 7 8 9 10 11 12
pH

1.0
Alpha fraction

0.8 -
SiOH SiO
0.6
0.4
SiOH2+
0.2
0.0
1 2 3 4 5 6 7 8 9 10 11 12
pH

Figure 4.1
Distribution of surface species MOH2+, MOH, and MO - as a function of
pH for Al, Fe and Si oxide surfaces.

59
The pH of the solution also effects the degree of protonation of F -.

HF ? H+ + F- Ka = 6.8 x 10-4 4.11

The distribution of F- species as a function of pH is given in Figure 4.2. It is clear

that F- is completely protonated below about pH 3. The neutral HF species
existing in acidic media will not be adsorbed. At pH 4 about 50% of F- is
protonated and at pH > 5, F- is completely deprotonated and will exist as free F-
ions in the absence of complexing agents. It is therefore expected that maximum
F- adsorption will occur at pH > 5.

1.00
Alpha fraction

0.80
0.60 HF
0.40 F
0.20
0.00
0 1 2 3 4 5 6 7 8 9 10
pH

Figure 4.2
Distribution curves for F - species as a function of pH

4.3 Verification of fluoride-hydroxide exchange mechanism

through pH measurement.

From inspection of the adsorption equilibria (Eq. 4.1 and 4.2) for aluminium oxide
given above one would expect that the adsorption of F- would be accompanied
by a concomitant rise in the pH of the solution as a result of the neutralisation of

60
H+ in the first equilibrium and the release of OH- in the second. This was,
however, not consistently observed in experiments reported in the literature (Hao
et al, 1986; Wang and Reardon, 2001) casting some doubt on the proposed
mechanism. The fact that linear Langmuir isotherms (see Section 3.4.6) were
observed in many studies of F- adsorption onto activated alumina is also not
reconcilable with an ion exchange mechanism (see Section 4.5.2 and Eq. 4.27).
In view of these uncertainties it was necessary in this work to investigate the
mechanism of F- adsorption and to provide reliable experimental evidence to
enable the correct description of the adsorption process.

The chosen method was to study the change in pH after the addition of F- to an
adsorbent at different initial pH values. The procedure is described in Section
3.4.7.

4.3.1 Equilibration of sorbent surfaces

Experiments were done to ensure that proper pH equilibration of the activated

alumina and other sorbents, which were used as substrates in these tests, was
achieved.

Results obtained for activated alumina in the acid form (ALU), given in Table 4.2
show the pH value 1 h after pH adjustment and again 14 to 18 h later. The pH
remained constant after overnight equilibration, that is after a period of > 14h,
indicating that equilibrium was established between solid surfaces and the
solution. Since the pH of a suspension of ALU in 0.1 M NaClO4 was about 4.6,
acid (HClO4) was added to prepare solutions with pH < 5 and NaOH added for
solutions with pH > 5. NaClO4 was added to keep the ionic strength of solutions
constant.

61
 TABLE 4.2
pH at different equilibration times after pH adjustment of activated
aluminium (ALU).
1h 14 h 15 h 16 h 17 h 18 h
3.26 4.21 4.18 4.23 4.23 4.23
4.04 4.82 4.80 4.81 4.81 4.81
5.25 5.27 5.22 5.24 5.25 5.25
6.20 5.50 5.45 5.47 5.50 5.50
7.23 5.88 5.74 5.74 5.78 5.78
8.11 7.16 7.05 7.00 6.98 6.98
9.13 8.46 8.33 8.29 8.28 8.28

The change in pH, pH is calculated by subtracting the non equilibrium pH at 1 h

from the equilibrium pH. pH data for ALU are given in Table 4.3. In Figure 4.3
pH is plotted against the pH value after 1h for equilibration of activated alumina
at different pH values. A positive pH reflects an increase in pH and a negative
pH a decrease in pH during equilibration.

TABLE 4.3
pH data for ALU
1st set of results 2 nd set of results
1h 16 h pH 1h 16 h pH
1.70 2.61 0.91 1.86 2.81 0.95
2.30 4.03 1.73 2.65 3.85 1.20
4.01 4.61 0.60 3.95 4.58 0.63
6.67 5.70 -0.97 6.80 5.69 -1.11
8.50 7.05 -1.45 8.25 6.70 -1.55
9.32 8.25 -1.07 9.50 8.50 -1.00
10.05 9.02 -1.03 10.10 9.11 -0.99

62
 2
1.5
1
Change in pH 0.5
0
-0.5 0 2 4 6 8 10
-1
-1.5
-2
pH(1h)

Figure 4.3
pH vs pH1h for the equilibration of activated aluminium (ALU) at different pH
values for duplicate measurements.

A steady decrease in pH is observed for pH values up to a pH of about 5. At

this point pH equals zero. At pH > 5 the pH values become increasingly
negative up to pH of about 8. Above pH 8 the curves turn upward indicating
smaller pH values. The values are still negative which means that the pH
decreased during equilibration. The data give an important insight into the
protonation and deprotonation behaviour of the aluminium oxide surface of
activated alumina in the acid form. The results can be explained in terms of the
rather slow equilibration rate and the effect of pH on the distribution of protonated
and deprotonated species as shown in Figure 4.1. The curve, however, is partly
defined by the experimental procedure that was followed. After the addition of
acid to adjust the pH to values less than 5 at equilibrium, the protonation reaction
takes place in which AlOH2+ species are formed from sites still not protonated.
The pH value therefore rises as protons are removed from solution until
equilibrium is attained. The pH in solution of the activated alumina that was used
for these studies was 4.6. The amount of acid added to attain an equilibrium pH
of 3, for example, would of course be more than that needed to attain pH 4.
Because of the slow kinetics of the protonation reaction, the excess added will

63
also be more and the difference between initial pH measured at 1h after addition
of acid and equilibrium pH will be more in the case of the lower pH value. This
explains the decreasing pH for the pH range 3 to 5. In this range the reaction is
governed by the reciprocal of the first dissociation constant, 1/K a1 for the
protonation reaction:

AlOH + H+ ? AlOH2+ 4.12

At pH above 5, OH- had to be added to produce the desired equilibrium pH. In

the pH range 5 to 7 pH again increases for the same reasons as above but the
sign is now reversed because the excess is OH-. In this pH range the relevant
reaction is governed by the constant K a1/Kw :

AlOH2+ + OH- ? AlOH + H2O 4.13

Above pH 7 the curve suddenly reverses trend and the pH values become
smaller. This turning point coincides with the beginning of the formation of AlO-
species according to the deprotonation reaction governed by the constant K a2/Kw :

AlOH + OH- ? AlO- + H2O 4.14

It is important to note that the equilibration characteristics will differ substantially

between surface types. In the case of RBM (an amorphous Al2O3 ), the surface is
basic and equivalent to AlO-. The pH of a RBM suspension in 0.1 M NaClO4 was
about 9.2. The ?pH curves shown in Figure 4.4 and the ?pH data for RBM given
in Table 4.4 therefore, are quite different from that for ALU. In this case pH
decreased but never reversed its positive sign as the equilibrium pH progressed
towards the substrate pH. Due to the high initial pH, acid was added to effect pH

64
adjustments in all cases . These observations can be explained in terms of the
protonation reactions

AlO- + H+ ? AlOH 4.15

AlOH + H+ ? AlOH2+ 4.16

TABLE 4.4
?pH data for RBM
1st set of results 2nd set of results
1h 16 h ?pH 1h 16 h ?pH
2.95 7.01 4.04 2.95 7.01 4.0
3.88 6.72 2.84 3.88 6.63 2.80
4.86 6.63 1.77 4.86 6.72 1.78
6.09 8.34 2.25 6.09 8.44 2.30
6.74 8.53 2.44 6.74 8.51 2.39
8.14 8.83 0.64 8.14 8.78 0.61
8.98 8.98 0 8.98 8.95 -0.02

4.5
4
3.5
3
2.5
?pH

2 2h
1.5 6h
1
0.5
0
pH
-0.5 0 2 4 6 8 10

Figure 4.4
?pH vs pH 1 h for the equilibration of RBM at different pH values.

65
To support this interpretation and to compare the kinetics of pH equilibration for
ALU and RBM the pH change over time was measured over shorter time
intervals after adjustment of the pH with acid or base. The results are shown in
Table 4.5 and 4.6 and Figure 4.5 and 4.6 for ALU and RBM, respectively. The
curves show that equilibrium is reached within 2 h after pH adjustment. As
expected the change in pH (?pH) decreases as the initial pH is closer to the
substrate pH of about 5 for ALU and 9 for RBM. The curves for ALU show
symmetrical but reversed behaviour above and below the substrate pH.

In the next section the observed changes in pH during adsorption onto these
substrates will be explained in terms of the acid-base properties of the surfaces
as discussed above.

TABLE 4.5
pH at different time intervals after pH adjustment at
varying starting pH. Activated aluminium (ALU)
pH at different time intervals after adjustment of initial pH
Time (h) pH (2.92) pH (5.00) pH (7.41) pH (9.34)
1/2 4.17 4.90 6.59 8.69
1 4.41 4.96 6.37 8.58
1.5 4.46 4.97 6.35 8.55
2 4.48 5.00 6.35 8.56
3 4.56 5.01 6.42 8.58
4 4.58 5.16 6.47 8.60
5 4.73 5.23 6.50 8.72
6 4.77 5.10 6.43 8.71
7 4.78 5.28 6.52 8.63
8 4.89 5.22 6.48 8.66

66
 10

8
pH 2.92
6 pH 5.00

?pH
4 pH 7.41
pH 9.34
2

0
0 2 4 6 8 10
Time/h

Figure 4.5
pH vs time for the equilibration of ALU for different starting pH
values.

TABLE 4.6
pH at different time intervals after pH adjustment at
varying starting pH. Amorphous aluminium oxide (RBM)
pH at different time intervals after adjustment of initial pH
Time (h) pH (2.79) pH (5.16) pH (7.32) pH (8.90)
4.64 6.23 8.37 8.88
1 5.66 6.74 8.61 8.89
1.5 6.50 7.23 8.73 8.97
2 5.86 7.59 8.77 8.99
3 5.86 7.76 8.84 8.99
4 5.95 7.84 8.75 8.94
5 6.14 7.95 8.72 8.95
6 6.30 7.97 8.77 8.98
7 6.14 7.92 8.73 8.91
8 6.31 7.94 8.72 8.86

67
 10

8
pH 2.79
6 pH 5.16

?pH
4 pH 7.32
pH 8.90
2

0
0 2 4 6 8 10
Time/h

Figure 4.6
pH vs time for the equilibration of RBM at different starting pH
Values.

4.3.2 Change in pH during adsorption

The change in pH during the adsorption process was studied after the addition of
F- solutions to properly equilibrated aluminium oxide suspensions. Data for ALU
and RBM are given in Table 4.7 and 4.8, respectively. pH was calculated by
subtracting the pH at different times after addition of F- from the initial pH. The
initial pH, pHi, is the pH after 14 h equilibration of the substrate at a particular pH
level.

A plot of pH vs the initial pH value for ALU shown in Figure 4.7 reveals
important characteristics of the adsorption process. In the pH range 2.5 to 6 a
slight rise in pH was observed after addition of F- to the test sample of activated
alumina in the acid form. The rise in pH is particularly strong at short time
intervals, eg. 1.3 pH units after 15 min, after the addition of F- for solutions with
pH between 4 and 6. The pH in this pH range drops to an equilibrium value only
slightly above the initial value after about 45 min. The decreasing pH values
over time indicated that OH- released in the exchange reaction with F- was
removed from the solution by some process. This process is the neutralisation of

68
remaining acid sites (AlOH2+ ) on the activated alumina. The rise in pH therefore
suggests that the rate of the F- exchange reaction is faster than the rate of the
neutralisation reaction. Above pH 6 the pH curve at equilibrium rapidly drops to
0 and then move into negative range, signifying a decrease in pH after addition of
F-. Note, that 15 min after addition of F- all pH values are positive as was
expected in accordance with the ion exchange model.

TABLE 4.7
The pH at different times intervals after the addition of 10 mg.L-1
F- solution to activated alumina (ALU) equilibrated at the initial
pH, pHi at 14h.
1st set of results 2nd set of results
pHi pHi + pHi + pHi + pHi pHi + pHi + pHi +
15min 30min 45min 15min 30min 45min
2.61 2.77 2.83 2.81 2.81 2.93 2.96 2.97
4.03 4.35 4.37 4.35 3.85 3.99 4.07 4.06
4.61 5.21 5.09 5.02 4.58 5.28 5.06 4.98
5.70 7.01 6.62 6.25 5.69 7.10 6.50 6.16
7.05 7.38 7.29 7.26 6.70 7.11 7.16 7.05
8.25 8.31 8.06 7.90 8.50 8.75 8.75 8.69
9.02 9.26 9.21 9.11 9.11 9.51 9.61 9.60

69
 1.5
15min-1st

Change in pH
1 30min-1st
45min-1st
0.5
15min-2nd
0 30min-2nd
0 2 4 6 8 10 45min-2nd
-0.5
pH(initial)

Figure 4.7
?pH vs the initial pH for ALU at different times after the addition
of 10 mg.L-1 F - solution

The pH changes vs initial pH during adsorption of F- onto RBM are given in Table
4.8 Figure 4.8 shows the same trend as that for ALU with a maximum ?pH
occurring at pH 6. The direction of the change in pH with time allowed for
equilibration is, however, reversed. The ?pH increases with equilibration time,
that is the pH rises with time. This phenomenon can be understood in terms of
the fact that the RBM has a basic surface while the surface of the activated
alumina is acidic. In the case of ALU the OH- released on adsorption, is partially
neutralised by the acidic surface whereas with RBM no such mechanism for the
removal of OH- is available. The pH thus increases with time as OH- is released
through F- exchange. After 60 min a maximum is reached when the exchange
process is complete. The study of the kinetics of exchange as discussed in
Section 4.7, confirms that the adsorption process is practically completed in less
than 60 min for most substrates.

Adsorption curves, % ads vs pH, for activated alumina and RBM given in Figure
4.9 show adsorption maxima at pH 6. This is in agreement with the maximum

70
positive pH value also observed at pH 6. Recall that a positive pH represent
an increase in pH as a result of OH- released in the exchange reaction with F-
during adsorption.

TABLE 4.8
The pH at different time intervals after the addition of 10 mg.L-1 F-
solution to amorphous aluminium oxide (RBM) in the basic form
equilibrated at the initial pH, pHi at 14h.
pHi pHi + pHi + pHi + pHi + pHi + pHi + pHi +
15min 30min 45min 60min 2h 4h 6h
2.98 3.02 3.04 3.08 3.08 3.07 3.06 3.07
3.83 3.90 3.92 3.98 4.00 4.04 4.01 4.05
4.95 5.70 6.00 6.19 6.21 6.21 6.23 6.24
6.03 6.99 7.22 7.39 7.59 7.58 7.60 7.59
7.03 7.27 7.37 7.43 7.49 7.49 7.53 7.55
7.90 8.18 8.23 8.28 8.30 8.30 8.35 8.34
8.86 8.90 8.93 8.94 8.98 9.07 9.03 9.03

1.5 15min
30min
?pH

1 45min
60min
0.5
2h
0 4h
0 2 4 6 8 10 6h
pH

Figure 4.8
?pH vs the initial pH for RBM at different times after the addition
of 10 mg.L-1 F - solution

71
 100

% F Adsorption
80
60 ALU
40 RBM

20
0
0 2 4 6 8 10
pH

Figure 4.9
Adsorption curves for activated alumina (ALU) and amorphous
aluminium oxide (RBM)

4.4 Zeta potentials

4.4.1 Background

An understanding of the zeta potential at a solid-liquid interface is an important

factor in explaining the adsorption mechanisms and the wa y in which it is
affected by factors such as pH, concentration or ionic strength (McFadyen,
2001). The availability of instrumentation to facilitate the determination of zeta
potential has undoubtedly aided the rapid growth in the number of reported
measurements over recent years. The zeta potential is used to characterize the
outer, diffuse part of the double layer and it is useful for evaluating the interaction
between a surface and solution components (Hasselbrink et al., 2001).

In this study, zeta potentials of different clay samples were investigated in order
to compile information useful in explaining F - adsorption mechanisms.

The procedure is described in Section 3.4.5.

72
4.4.2 Results and Discussion

The results obtained in this study are shown in Table 4.9

Table 4.9
Zeta potentials (mV) of clays at different pH values.
pH ALU RBM SLV MD2 CAL PTA-LAT
3 37.4 42.0 28.5 18.3 0.8 1.0
4 35.1 24.9 -0.5 -3.0 -3.4 -2.4
5 32.4 21.4 -7.3 -2.4 -3.7 -3.0
6 35.1 -2.3 -5.1 -5.6 -1.5 -2.6
7 34.7 -5.8 -10.8 -17.6 -3.4 -6.4
8 29.1 -5.6 -12.9 -24.3 -4.7 -4.7
9 -2.1 -20.4 -35.3 -30.9 -11.1 -37.3

The zeta potentials of the clay samples as a function of solution pH values are
shown in Figure 4.10. The zeta potentials for ALU are positive up to pH 8 but
negative at pH > 9. The point of zero zeta potential was obtained at
approximately pH 8.9. This value is much higher than the calculated point of zero
charge. For RBM, the zeta potentials are positive under acidic conditions but
negative in basic condition, with the point of zero charge occurring at about pH 6,
much lower than that of ALU. Below pH 7, the zeta potentials of RBM decreased
almost linearly with increasing solution pH. This can be attributed to the
protonation of the alumina surface. From pH 7 to 8, the negative zeta pontentials
did not appear to decrease significantly with the increase of pH. This may
indicate that the alumina surface in the RBM was not fully deprotonated under
these pH conditions.

73
The zeta potentials for SLV and MD2, which contained both Al2O3 and Fe2O3
were substantially different from ALU and RBM. Both showed a point of zero zeta
potential around pH 4. The negative zeta potentials of both SLV and MD2 may
be due to the presence of high quatities of SiO2 sites in the samples. The zeta
potentials for CAL and PTA-LAT, show points of zero zeta potential around pH 3
to 3.2 reflecting the effect of quartzitic components in these clays.
Zeta potentials (mV)

40
ALU
20 RBM
MD2
0 SLV
0 2 4 6 8 10 CAL
-20
PTA-LAT
-40
pH

Fig 4.10
Zeta potentials as a function of pH for selected clay samples.

4.4.3 Conclusions

The results indicated that chemical composition of the samples as expected have
a dominant effect on the zeta potential values. The results also confirm the
surface charge distribution as a function of pH as was predicted by the a
distribution curve for Al, Fe and Si oxides in Figure 4.1.

74
4.5 Adsorption isotherms

4.5.1 Background

Adsorption isotherms can be used to obtain information about the mechanism of

adsorption. It can also be useful in determining adsorption capacities and
Langmuir constants. Langmuir constants are useful in comparing adsorption
characteristics of different substrates. In many studies (Chaturvedi et al, 1988)
linear Langmuir adsorption curves or F- adsorption onto metal oxide surfaces
have been found. In this study adsorption isotherms for a selection of clays and
activated alumina were determined according to the procedure described in
Section 3.5.6. The results summarized in Table 4.10 confirm the phenomenon of
a linear Langmuir adsorption relationship (inverse of the amount of F- adsorbed
vs equilibrium F- concentration) and are shown in Figure 4.11 for activated
alumina (ALU) and amorphous Al2O3 (RBM) at pH 6. Only the linear portions of
the graphs are shown. Similar curves were obtained for clay sorbents. At low
equilibrium concentrations slight curvature is observed. Linear regression
analysis of the linear portion of the graph gives the slope and intercept values
which can be used to calculate the Langmuir constants Q, the adsorption
capacity, and b (see Eq. 4.20). Table 4.11 compares the Langmuir constants
obtained in this study with values reported in the literature.

75
4.5.2 Results and discussion

TABLE 4.10
Adsorption isotherms for ALU and RBM at pH 6
ALU RBM
Initial F- Ce qe Ce/qe Ce Ce qe Ce/qe Ce
-1 -1 -1 -1 -1 -1 -1 -1
(mg.L ) mg.L mg.g g.L mmol.L mg.L mg.g g.L mmol.L-1
0 0 0 0 0 0 0 0 0
2 0 0.10 0 0 0 0.10 0 0
5 0 0.25 0 0 0.06 0.25 0.24 0.03
10 0.42 0.48 0.90 0.02 1.0 0.45 2.20 0.05
20 4.5 0.80 3.33 0.24 4.50 0.78 5.80 0.24
50 24 1.30 18.5 1.30 33 0.83 40 1.60
100 62 1.90 32.6 3.30 78 1.10 71 3.80
200 149 2.60 57.3 7.80 169 1.50 109 8.90

The values of Q and b in Table 4.11 were calculated using Eq. 4.20. The inverse
of the adsorption capacity (Q-1) is equal to the slope of the curve. The Langmuir
constant , b is equal to the intercept.

TABLE 4.11
Langmuir constants for different sorbents
Sorbent Q b Reference
-1 -1
mg.g L.mg
gibbsite 2.22 0.25 Bower 1967
activated alumina 2.90 0.76 This work
Aluminium oxide (RBM) 0.45 0.24 This work

76
 2.5 y = 0.3518x + 0.7658

qe (mg.g-1 )
2
1.5 R2 = 0.9849
1
0.5
0
0 1 2 3 4
Ce(mmol.L-1)

(a)

1
y = 2.244x + 0.2408
0.8
qe (mg.g-1)

2
R = 0.9999
0.6
0.4
0.2
0
0 0.1 0.2 0.3
-1
C e (mmol.L )

(b)
Figure 4.11
Langmuir isotherm for (a) activated alumina (ALU) and (b) aluminium oxide (RBM)

Langmuir adsorption assumes monolayer adsorption. At equilibrium the rate of

adsorption, k1(1-)[F] and the rate of desorption, k2 of a monolayer of F- will be
the same. is the fraction of the surface covered with adsorbed F -. Therefore

k1(1-)[F] = k2 4.17

77
 and
K [F ]
= 4.18
1 + K[ F ]
or written as a linear equation

F 1
= + [F ] 4.19
K

If is written in terms of measured experimental quantities, where C0 and Ce are

the initial and equilibrium F- concentrations respectively, w the mass of sorbent,
then the Langmuir equation can be written in the form:

Ce 1 Ce
= + 4.20
qe bQ Q
where
C 0 Ce
qe = 4.21
w

and is proportional to q e: Q-1 is the slope and b the intercept.

The fact that F- adsorption seems to comply with conditions for monolayer
Langmuir adsorption resulting in almost linear isotherms is somewhat surprising
if one considers the hydroxide exchange mechanism commonly assumed to be
valid for F- adsorption onto metal oxides such as activated alumina.
Rearrangement of the equilibrium expression for the exchange mechanism
reveals that a linear relationship between [F -] and adsorbed fraction can only
occur if the [OH-] remains constant. This is not the case as was shown in Section
4.3.2.

78
Consider the ion exchange reaction

AlOH + F- ? AlF + OH- 4.22

with equilibrium constant Kex and let f be the fraction of OH- exchanged with F-, f
is therefore the fraction of surface sites containing F-. The equilibrium constant
can then be written as:

[ AlF][OH ]
K ex =
[ AlOH][F ]
4.24
f [OH ]
=
(1 f )[ F]
4.25

Rearrangement gives:

K ex [ F ]
f = 4.26
[OH ] + K ex [ F ]

or written in a form equivalent to the linear Langmuir equation:

1 [OH ]
= 1+ 4.27
f K ex [ F ]

It is clear that 1/f plotted against 1/[F] will only be a straight line if [OH] is
constant. The fact that Langmuir isotherms are rectilinear either means that the
exchange mechanism is wrong or that the [OH] remains constant during
adsorption. In many reported (Chhabra et al, 1980; Chaturvedi et al 1988)

79
 measurements of pH changes during adsorption onto Fe and Al oxide surfaces
no pH changes were observed. A detailed study of the pH changes during
adsorption as discussed in Section 4.3.2 shows that pH changes do occur
during adsorption if the proper account is taken of the buffer action of the
substrate and provision is made for the equilibration of the substrate at a
particular pH. It therefore seems that many researchers did not manage to
control and take into account the buffer action of the substrate. The explanation
for the anomalous linear isotherms lies in the fact that the [OH] is indeed kept
relatively constant because of the buffer action of the substrate. This reaction
is, however, slow and if insufficient time for equilibration is allowed substantial
changes in [OH] may be measured.

4.6 Effect of other ions

4.6.1 Background

Hao et al. (1986) stated that increase of the concentration of sulphate and other
anions decreases the adsorption capacity of fluoride onto hydrous alumina. This
competition effect may also reduce the efficiency of defluoridation in other
processes. In summary, the type and concentration of other anions play an
important role in the adsorption of fluoride. This work aims to elucidate the effects
of other anions on adsorption.

Borehole waters normally contain high concentrations of many anions such as

chloride, nitrate, sulphate, etc (Schoeman and Steyn, 2000). In the present study
all tests were done with single ion i.e. F- solutions. However, in real systems
several other ions are present which can compete with F- (Nishimoto et a., 2001;
Raichur and Basu, 2001). The effect of presence of two ions i.e. chloride and
sulphate on F- adsorption was studied. Tests were conducted in the presence of
10 mg.L -1 of F - and 0-5000 mg.L -1 each of chloride and sulphate ions.

80
4.6.2 Methods

The effect of competing anions on the adsorption capacity of activated alumina

(ALU) and RBM was studied at pH 6. The procedure is described in Section
3.4.4.3.

4.6.3 Results and Discussion

The results obtained for the effect of Cl- and SO42- on F- adsorption on clay
samples using different anion concentrations at pH 6 are shown in Table 4.12.

TABLE 4.12
Effect of anions on F- adsorption at pH 6 on % adsorption
Anions Concentrations ALU RBM
-1
(mg.L )
-
Cl 0 100 98
100 98 97
200 97 96
500 93 97
1000 90 85
5000 70 80

SO42- 0 100 99
100 96 98
200 92 94
500 84 79
1000 77 50
5000 50 35

81
The defluoridation efficiency was 100% or close to 100% in the absence of Cl-
and SO42-, dropping to 70% and 80% in the solutions containing Cl- and dropping
to 50% and 35% in the solutions containing SO42-. It was found that the
defluoridation efficiency did not change very much with Cl- concentrations, but
decreased as the concentration of SO42- ions increased.

4.6.4 Conclusions

The presence of other anions such as Cl- and SO42- decreased the adsorption of
F- .

4.7 Development of a mathematical procedure for the modelling

of F - adsorption onto mineral substrates.

In this section the derivation of a mathematical modelling procedure for the

simulation of F- adsorption onto mineral surfaces as a function of pH is
presented. The model is based on the acid-base characteristics of the sorbent as
determined by the dissociation constants, the protonation of F-, and assuming a
surface complexation mechanism for the adsorption of F- (Hao etal., 1986). The
assumption of a surface complexation mechanism is based on the fact that
almost linear Langmuir adsorption isotherms were obtained for many sorbents
containing metal oxide components. Numerous studies reported in the literature
confirm that F- adsorption in this conte xt corresponds to a Langmuir isotherm
(Bower and Hatcher, 1967; Chaturvedi et al., 1988). This is despite the fact that
the mechanism proposed for F- adsorption is ion exchange and not monolayer
adsorption which is the primary condition for simple Langmuir adsorption. The
significance of this assumption is that the data support a monolayer adsorption
mechanism for which the Langmuir adsorption equation is valid. The Langmuir

82
equation was therefore used to describe the adsorption process. This
assumption, however, contradicts the ion exchange model. Equations 4.27 show
that the ion exchange model should not lead to linear adsorption curves. This
observation was discussed in Section 4.5.2.

A simplified model for F - adsorption is developed below, based on:

the known thermodynamic constants for protonation of metal oxide

sorbents which occur in most clays,
the protonation constant for F- derived from the dissociation constant for
HF, and
the Langmuir equation for monolayer adsorption can be derived as shown
below.

If it is assumed that the total number of active adsorption or exchange sites, Nt

for a metal oxide surface is given by:

Nt = [MOH2+] + [MOH] 4.28

and is the fraction of active sites occupied by F-, the number of sites substituted
with F - is given by

NF = kNt 4.29

where k is a constant.

Derivation of an expression for

For monolayer chemisorption from solution with equilibrium F- concentration [F],

the Langmuir equation is given by

83
 K [F ]
= 4.30
1 + K[ F ]

where K is the Langmuir adsorption constant.

The free F - concentration is pH dependent and is proportional to F,

Ka
[F ] ~ F = 4.31
K a + [H + ]

where Ka is the dissociation constant for HF and F is the fraction of total F-, CF in
the F - form.

[F ]
F = 4.32
CF

and

C F = [ F ] + [ HF ] 4.33

Substitution of Eq. 4.31 into 4.30 gives in terms of the Langmuir constant,
dissociation constant for HF, and [H+].

=
(
K K a K a + [H + ] )
( )
4.34
1 + K Ka Ka + [H + ]

Derivation of an expression for N t

If the total concentration of active sites, Nt is assumed to be proportional to the

sum of the fractions for the two types of active site, then

84
 N t = k1 (1 + 2 ) 4.35

where

1 =
[MOH ] = +
2 [H + ]2
4.36
CM [ H + ]2 + K a1[ H + ] + K a1 K a 2

2 =
[MOH ] = K a1[ H + ]
4.37
CM [ H + ] 2 + K a1 [ H + ] + K a1 K a2

[ ]
C M = MOH 2+ + [MOH ] 4.38

By expressing the fractions and surface species concentrations in terms of the

acid dissociation constants, Ka1 and Ka2 and H+ and equation for Nt can be
derived.

[ H + ] 2 + K a1[ H + ]

N t = k1 + 2 +
4.39
[ H ] + K a1 [ H ] + K a1 a 2
K

Substitution of Eq 4.34 and 4.39 into Eq 4.29 gives the adsorption modelling
equation for fraction of sites, NF substituted with F- in terms of the equilibrium [H+]
concentration and equilibrium constants.

K [ H + ] 2 + K a1[ H + ]
N F = k +

+ 2 +
4.40
1 + K + [ H ] K a [ H ] + K a1[ H ] + K a1K a2

The modelling equation can be used to calculate the form of the adsorption curve
as a function of pH. The theoretical form of the adsorption curve is shown in
Figure 4.12. The curve was calculated for an arbitrary selection of dissociation

85
constants, Ka1 = 10-6 and Ka2 = 10-11, approximately similar to that of an Al oxide
surface. The dotted lines indicate the effect of F- protonation and the formation of
HF which causes a sharp drop in adsorption at pH<4 and the effect of a rapidly
decreasing number of exchange sites at pH>8 where a similar drop in adsorption
is noted. In this pH region deprotonation of the metal hydroxide, leads to the
formation of a negative surface according to the reaction:

MOH ? MO- + H+ 4.41

Relative adsorption

1.00
0.80
0.60
0.40
0.20
0.00
0 1 2 3 4 5 6 7 8 9 10 11 12
pH

Figure 4.12
Theorectical form of adsorption vs pH curve. The dotted lines show the
effect of protonation of F- at pH < 3 and the deprotonation of MOH at
pH > 8

4.8 Experimental verification of modelling equation

The model was verified by fitting experimental adsorption data, both obtained in
this study and taken from the literature to the predicted curve. Fig 4.13 shows the
very good fit of normalised experimental data for F- adsorption onto activated
alumina to the theoretical curve predicted by the model. Absorption data were

86
obtained by determining the % adsorption as a function of pH according to the
procedure described in Section 3.4.4.2.

Fraction adsorbed
1.00
0.80
0.60
0.40
0.20
0.00
0 1 2 3 4 5 6 7 8 9 10 1 1 1 2
pH

Figure 4.13
Fit of modelling equation to experimental data for activated alumina

4.10 Conclusions

Results obtained in this chapter support the exchange mechanism for F-

adsorption. The accurate determination of pH changes during adsorption and a
thorough investigation of the acid-base properties of the substrate and its effect
on the ion exchange properties, provided enough evidence to resolve
uncertainties in the literature with regard to the nature of the F- adsorption
mechanism. This work confirms that a rise in pH should accompany the
adsorption of F - as predicted by the ion exchange model.

Adsorption modelling based on thermodynamic considerations, i.e. the thermo-

dynamic constants for the different equilibria in the adsorption process, proved to
be successful in prediction adsorption curves for metal oxide substrates. Since
metal oxides of Fe and Al are the predominant mineral structures found in clays
showing F- adsorption potential, the model provide a useful tool for the

87
interpretation and prediction of adsorption behaviour of substrates under different
pH conditions.

88
 CHAPTER 5
Adsorption characteristics of South African clays

5.1 Background

The literature review (see Chapter 2), revealed that studies on F- adsorption onto
to clays and soils, and many other materials, have been conducted in almost all
countries where high F- natural waters present a health problem to the
community South Africa being a notable exception. No such study has been
reported despite the fact that large parts of rural South Africa do not have access
to good quality municipal water and are in many cases dependent on borehole
waters containing dangerously high levels of F-. The extent of the high F-
concentrations in groundwater problem in South Africa was brought to the
attention by a comprehensive WRC report by McCaffrey and Willis (McCaffrey
and Willis, 2001). In this report the geochemistry of F- in the northern parts of
South Africa is addressed and the fluorosis problem associated with prolonged
consumption of water containing F- concentrations > than 1.5 mg.L -1 is
discussed. A case is made in this report for the development of cost-effective
house-based methods for defluoridation.

Results of F- adsorption studies on clays and soils indicate that large variations
in adsorption capacities occur among the different types. Adsorption capacities
are generally low and kinetics slow. These studies were generally conducted in
a non-systematic way without taking into account the relation between the
structural characteristics of the clay and its tendency to adsorb F-. Nevertheless,
useful defluoridation applications based on clays baked at low temperatures were
developed by trial and error methods. The clay brick system used in Sri Lanka
(Padmasiri, 1998) is a good example of a relative successful house-based

89
approach to defluoridation and for that reason a clay sample of the clay type
used for clay brick defluoridation in that country was included in this study.

In this chapter the characterisation of South African clays with regard to their F-
adsorption capacities is discussed. The selection of clay types with a high
probability of being good sorbents for F- was based on the predictions of the
model discussed in Chapter 4. Clay bodies with these characteristics were
sought in close proximity to the high F- areas in the Northern Province, an area
targeted for this study. Adsorption capacities for the selected clays, 25 samples
representing different mineral types, were determined at pH 6, a pH close to the
pH range predicted for maximum adsorption. A sub sample of four clays showing
potential to be used as sorbents for F- was selected for further characterisation.
Characterisation included the determination of adsorption curves (adsorption
capacity vs pH) and the effect of physical (baking at different temperatures) and
chemical pretreatment (wash with an acid and base) to determine whether
enhancement in adsorption capacity can be induced by these methods.

The specific objectives of the research project reported on in this chapter, are
listed below.

(a) To investigate the F- adsorption characteristics of South African clay types

and assess their potential as sorbents in a house-based F- removal
system.

(b) To develop chemical and physical pretreatment procedures to enhance

adsorption properties of clays.

90
5.2 Sampling and selection of clay samples

5.2.1 South African clay deposits

The term clay is often used in a non-specific way and could refer to: (a) soil
consisting of a range of small particle sizes (eg. particle size < 1/256 mm on the
Udden-Wentworth scale), (b) very fine-grained earthy substances comprising of a
combination of minerals, inorganic amorphous material and organic matter or (c)
a specific clay mineral. The clay minerals are minerals in the phylosilicate mineral
group and as the name implies (Greek: pyllon, leaf), most of these minerals have
a platy habit.

In this study the term clay refers to natural occurring, very fine-grained earthy
material composed primarily of clay minerals. Clays are potentially good
adsorbers of anions since they contain crystalline minerals such as kaolinite,
smectite and amorphous minerals such as allophane and other metal oxides and
hydroxides which could adsorb anions such as F-. The South African clay
deposits shown in Figure 5.1, can be broadly classified according to the
dominant clay mineral present into (1) Kaolin Fields (dominant clay mineral being
kaolinite) (2) Bentonite Fields (dominant clay mineral being montmorillonite,
which is part of the smectite group) and (3) Palygorskite Fields (dominant clay
mineral being palygorskite). Bauxite, which is a rock type composed of one or
more of the aluminium hydroxides gibbsite, boehmite or diaspore is potentially
also a good adsorber of anions and therefore the Bauxite Fields of South Africa
were also included in this study. Finally, the rock type laterite which represents a
group of deposits consisting of residual insoluble ferric and aluminium oxides,
clay minerals and quartz, formed by weathering of rocks, were also included in
this study as potential adsorbers of anions.

91
 Figure 5.1
Major clay deposits (Horn and Strydom, 1998) and bauxite deposits
(Barnardo, 1998) in the northern parts of South Africa

5.2.2 The selection of clays and sampling

In the selection of clay types for this study, two criteria were considered:

The clay structure most likely to have good F- adsorption

properties, i.e. exchangeable hydroxide-containing minerals and
clays high in aluminium and iron oxides or hydroxides such as
gibbsite and goethite/hematite minerals.

The best possible clay deposits with potential F- adsorption

properties in the areas where high F- groundwaters are used for
drinking water and where fluorosis was demonstrated to be
endemic. Although large parts of South Africa are under risk, this
study focused on the western Bushveld area because it was
extensively studied (McCaffrey and Willis, 2001) with regard to the

92
 F- geochemistry and the distribution of F- in groundwater. Clay
deposits situated within 500 km from the endemic area were
identified and sampled for this study.

Known major clay deposits in South Africa were identified from geological maps
and are shown in Figure 5.1. Table 5.1 lists the areas and the types of clay
deposit sampled for this study. Samples were collected using a manual auger to
a maximum depth of 2.5 m. In many cases existing open pits allowed sampling at
greater depths. In addition to samples collected in South Africa, clay samples
obtained from Sri Lanka and Australia were included in the study for comparison.

TABLE 5.1
Descriptions and localities of samples examined during this study

KAOLIN CLAYS
No Name Description Location

1 ZEB1 Pure white clay from open pit.

Zebediela Kaolin Field
Farm: Rooiboschbaak 107KS
White clay with greenish tint, from open
2 ZEB2
pit

3 ZEB6 Blood red clay, from open pit

Finely laminated orange and white clay in
Koster Area
4 RNF pan, 1 m to 2.5 m deep. Derived from
Farm: Renosterfontein 494JP
weathering of Timeball Hill Shale

PALYGORSKITE CLAYS
Light brown colour, under 800mm of black Modimolle Area, Springbok Flats
5 LW1
topsoil, 800 mm 950 mm deep Farm: Leeuwaarden 633KR

93
 TABLE 5.1
Descriptions and localities of samples examined during this study
Light brown clay from open pit. Below
Immerpan Palygorskite Field
6 CAL calcrete bed, 12 m deep. Weathering
Farm: Calais 563KS
product of basalt
Blue-white clay sampled at base of open Dwaalboom Palygorskite Field
7 DB
pit approximately 5m deep Artherstone Nature Reserve,
White clay bed, 1,2 m thick, below obtained from G&W Base and
8 DB2
approximately 500mm black topsoil Industrial Minerals

BENTONITE CLAYS
Koppies Bentonite Field
Bentonite sample - dry and finely milled
9 BEN Obtained from G&W Base and
product, greenish colour.
Industrial Minerals.
Pienaarsrivier Area, Springbok Flats
Black clay, 50 mm to 300 mm deep, in
10 BBK2 Farm: Blaawboschkuil 20JR
shallow depression or pan
Bela-Bela
Main white clay bed, approximately 2 m
11 CF3
thick. From open pit Naboomspruit Area, Springbok Flats
Main red clay bed, approximately 2 m Farm: Cyferfonteine
12 CF4
thick. From open pit

13 NHM Black clay, 2.00 m - 2.5 m deep

Northam Area

LATERITES AND BAUXITES

Laterite, orange red in colour, below
14 LAT Pilanesberg Area
600 mm brown topsoil
PTA-
15 Laterite. Orange-red colour Pretoria Area
LAT
Orange colour and surface texture that Howic Area
16 BAUX1
resembles a human brain Brainstone Farm: Lyndhurst
17 BAUX2 Orange colour, with dolerite core-stone Kwazulu-Natal

94
 TABLE 5.1
Descriptions and localities of samples examined during this study
18 BAUX3 Orange colour Australian bauxite
Orange colour, weathering profile, below
19 MD1
50cm black soil. Mooirivier Area
20 MD2 Orange colour, with dolerite core-stone Farm: Middeldraai
21 MD3 Orange colour, with dolerite core-stone
Ventersdorp Area
22 RYE1 Dark red nodular laterite Ryedale ferromanganese open pit
mine.

OTHER SAMPLES
Ochre to pale-brown nodules, dark-red to Sri Lanka clay used in clay brick
23 SLV
purple on fresh fractures defluoridation systems
Australian bauxite
Processed bauxite, 100% amorphous
24 RBM Obtained from Richards Bay
Al2O3
Minerals
25 ALU Activated aluminium Fluka, Type 504 C

5.3 F- adsorption characteristics of selected clay types

5.3.1 Methods

Preliminary screening tests were conducted in order to investigate the F-

adsorption capacities of 25 clay samples out of the 50 samples collected. From
this study, clay samples with good adsorption capacity were selected for further
studies. Detailed descriptions of the methods, procedures and analytical
techniques used in this study are given in Section 3.4.4.

95
5.3.2 Adsorption capacities at pH 6

Adsorption capacities for the different samples were determined by shaking 1 g

of washed and dried (2 h at 105oC) clay with 50 mL of 10 mg F-.L-1 NaF solution
for 2 h in polyethylene bottles at 22C. The initial pH was adjusted to 6 using
NaOH or HCl depending on whether acidic or basic clays were tested. After
equilibration the solutions were centrifuged, the pH measured to ensure that the
pH was within 0.2 pH unit from the target pH, and the residual F- concentration
determined using ion chromatography or a F - ion selective electrode.

The mineral composition and percentage F- adsorbed from 10 mg F-.L-1 solutions

at pH 6 are given in Table 5.2 for the selected 25 clay samples. The samples are
grouped according to the major clay types: bauxite, laterite, palygorskite,
bentonite and kaolinite. Adsorption capacities ranged from almost zero for
certain kaolinites to up to 80% for a bauxitic clay of Australian origin and 99% for
a processed bauxite obtained from Richard Bay Minerals and consisting of
armorphous Al2O3 . Anion chromatography and a F- ion selective electrode were
used to determine residual F- after completion of the adsorption process.
Reproducibility was determined by three repeat measurements. The standard
deviation ranged from less than 10% for the bauxite samples to about 15% for
the other sample types with lower adsorption capacities than the bauxites. For
the processed bauxite (RBM) and activated alumina (ALU) where adsorption was
close to 100% the reproducibility was better than 1%. The results show that high
adsorption can be correlated with bauxitic and lateritic samples where the
gibbsite and goethite/hematite contents are high. The lowest adsorption was
found for kaolin samples.

96
 TABLE 5.2
Mineralogical composition of clay samples and % F- adsorption at pH 6 from
10 mgF.L 1 solutions. xxxx: dominant(>50%); xxx: major (20-50%); xx: minor
(5-20%); x: trace (<5%)

Goethite/hema
palygorskite

mixed-layer
dolomite

smectite

Gibbsite
% F ads

kaolinite
Sample

quartz

clay

illite

talk

tite
BAUXITE (gibbsite / goethite)

ALU 100 xxxx

RBM 99 xxxx
BAUX3 80 xxx xxxx x
BAUX1 65 xx xxxx xx
BAUX2 50 xx xxxx xx
MD1 68 xxx xxx x xx xx xx
MD2 71 xxx x xxxx xx
MD3 73 xxx x xxxx xx

LATERITE (goethite / hematite / kaolinite)

SLV 72 xx xxxx xxx
RYE1 50 xxx xx x xxx
PTA/LAT 40 xxx xxx xxx
LAT 23 xxx xx xxx xx

PALYGORSKITE (dolomite / palygorskite)

CAL 58 x xxx xxx xxx
DB1 43 xxx xx xxxx
DB2 21 xx xxxx xxx
LW11 23 xxxx xxx xx xx

97
 TABLE 5.2
Mineralogical composition of clay samples and % F- adsorption at pH 6 from
10 mgF.L 1 solutions. xxxx: dominant(>50%); xxx: major (20-50%); xx: minor
(5-20%); x: trace (<5%)

Goethite/hema
palygorskite

mixed-layer
dolomite

smectite

Gibbsite
% F ads

kaolinite
Sample

quartz

clay

illite

talk

tite
BENTONITE (smectite)
NHM 45 xx xx xxxx
BBK2 30 xxx xxx
BEN 20 xx xxxx
CF3 15 xxx xxxx
CF4 0 xx x xxx xxx

KAOLINITE
RNF 25 xxx xxx xxx x xx x
ZEB1 5 xx xxxx x
ZEB2 4 xx xxxx xx
ZEB6 6 xx xxxx xx xx x

Table 5.3 gives the idealised chemical composition of the major mineral types
found in the clays investigated in the order in which they were found to adsorb F-
from aqueous solutions. Adsorption capacities were estimated from adsorption
data at pH 6 and initial F- concentrations of 10 mg F-.L-1. The results show that
clay samples with chemical composition of Al(OH)3 and FeO.OH give good
adsorption capacity. This confirms the predictions of the adsorption model
described in Chapter 4 that aluminium and iron oxide would play a major role in
the adsorption of F - using clays.

98
 TABLE 5.3
Major clay types and the ability to adsorb F- from aqueous solution at pH
6 from a solution containing 10 mg F-.L-1

Clay type Idealised chemical F- F-

composition adsorption adsorption
capacity
mg.g-1
gibbsite Al(OH) 3 0.25 - 0.4 high
goethite FeO.OH 0.2 0.3 high
palygorskite (Mg,Al)Si 8O20(OH) 2.8H2O 0.1 0.3 intermediate
smectite Ca(Mg,Fe)(Si,Al)4O10(OH) 2.nH2O 0.1 0.2 low
intermediate
kaolinite Al2Si2O5(OH) 4 < 0.05 low

In Table 5.4, published adsorption capacity data for different soil and clay types
are compared with the values obtained in this study. Data estimated for an initial
F- concentration of about 10 mg.L-1 and pH between 5 and 7 are included. The
comparison gives an approximate indication of F- adsorption trends for a wide
variety of soils and clays from all over the world although the numerical
adsorption values are not exactly comparable because experimental conditions
differed. These results support the conclusion that hydrated alumimium oxide
and iron oxide surfaces occurring in bauxites and goethites/hematites are useful
substrates for F- adsorption.

99
 TABLE 5.4
Comparison of F adsorption capacity data in mg.g-1 for different
-

soils and clays. Conditions: [F-]initial = 10 mg.L-1, 1 g clay, pH = 5 7,

equilibration time > 2h
Sorbent Sample name and Adsorption Reference
type description capacity
mg.g-1

BAUXITE
activated alumina Type 504C, Fluka 0.5 This work
gibbsite BAUX3 0.4 This work
Australia
BAUX1, BAUX2 0.25 - 0.4 This work
South Africa

GOETHITE
goethite / PTA-LAT 0.2 This work
kaolinite South Africa
goethite / Tertiary soil 0.23 Wang and
illite Shanxi China Reardon,
2001
goethite / SLV 0.35 This work
kaolinite Sri Lanka

PALYGORSKITE
palygorskite / DB1, CAL 0.21 0.29 This work
dolomite South Africa

BENTONITE
smectite BEN 0.1 This work
South Africa

100
 Bentonite trace Bower and
Wyoming, USA Hatcher, 1967
Alkaline soils 0.040.08 Bower and
USA Hatcher, 1967

KAOLINITE
kaolinite ZEB6 0.03 This work
South Africa
Acid soils 0.170.25 Bower and
USA Hatcher, 1967
Acid soils 0.13 Omueti and
Illinois, USA Jones, 1977
Potters clay, 30% 0.12 Brdsen and
Al2O3 Bjorvatn,
1106 White St 1995
Thomas
Clay pots 0.07 Moges et al.,
Ethiopia 1996

5.3.3 Conclusion

Results obtained in this section have proved that Al2 O3 and Fe2O3 are principal
components responsible for the adsorption of F- onto clays. The results agree
with previous findings (Chaturvedi et al., 1988) as clays consisting of Al2 O3 and
Fe2O3 give a higher F - adsorption.

101
5.4 Adsorption capacities vs pH

5.4.1 Background

Previous studies (Hao et al., 1986; Lu et al., 2002) have indicated that the pH of
the aqueous solution is an important variable which controls F- adsorption. The
effect of pH on the adsorption of F- was investigated using 5 selected clay
samples, namely RBM, CAL, SLV, MD2, PTA-LAT and laboratory prepared
hematite (Ozaki et al., 1984). The adsorption capacities vs pH results for these
clay samples were compared with results on similar clay types published by other
researchers.

5.4.2 Results and Discussion

The pH dependence of the adsorption process is illustrated in Figure 5.2 for a

few mineral types. Data for H-151 in (a) and GOETH in (b) were taken from the
literature (Hao et al. 1986). H-151 is a commercial activated alumina type and
GOETH is a goethite sample. The major component present in mineral sorbents
in Figure 5.2 (a) is aluminium oxide. Maximum F- adsorption is found at pH
between 5 and 6 for this group. The optimum pH value changes with the type of
sorbent used. Maximum F- adsorption is achieved at pH 3-4 for iron oxide types
such as goethite. The sorbents included in Figure 5.2 (b) contain varying
amounts of Fe2O3.

Clearly the adsorption capacity is determined by the structure of the clay and in
particular the surface charge distribution as a function of pH. The adsorption
capacity seemed to increase with increasing Al and Fe oxide content. Thus, the
more basic the surface and the more easily it can be protonated at pH values
above 3 the better were the adsorption characteristics of the clay. Typically the
adsorption is low at pH values < 3 and > 8. At pH values below 3, F- is

102
completely protonated to form neutral HF and therefore rapidly loses its ability to
adsorb onto positive surfaces.

100

% Adsorption
ALU
80
H-151
60
RBM
40
MD2
20
SLV
0
0 2 4 6 8 10
pH

(a)

100
% Adsorption

80 CAL
60 PTA/LAT
40 GOETH
20 HEM

0
0 2 4 6 8 10
pH

(b)
Figure 5.2
-
The pH dependence of F adsorption for different clays sorbents. (a) ALU =
activated alumina, H-151 = activated alumina type H-151; RBM = amorphous
Al2O3; MD2 = gibbsite/goethite and SLV = goethite/hematite/kaolinite.
(b) CAL = dolomite / palygorskite; PTA/LAT = geothite / hematite / kaolinite;
GOETH = goethite and HEM = laboratory prepared hematite.

103
5.5 Effect of chemical and physical pretreatment on adsorption
capacities

5.5.1 Background

Previous studies have indicated that adsorption of F- onto clay samples can be
enhanced by activation processes (Hauge et al., 1994; Srimurali et al., 1998).
The term activity denotes surface and physico-chemical reactivity, which is
usually traced to an increase in the surface area of the clay sample. The
following explanations illustrate the effect of pretreatment on the clay samples:

(a) Physical alteration of chemical composition and/or crystalline

structure by the effect of temperature (Balci and Gokcay, 2002).
(b) Chemical this is usually limited to ion exchange: It does not include
massive chemical destruction of the clay mineral structure.

5.5.2 Results and discussion

A few of the clay types that showed above average F- adsorption tendencies
were selected to determine the effect of chemical and physical treatment on the
possible enhancement of adsorption capacity. The selected clay types were:
gibbsite-goethite (MD2), palygorskite -dolomite (CAL), kaolinite-goethite (SLV),
and amorphous Al2 O3 (RBM). Activated alumina (ALU) was included for
comparison.

For clay sample MD2, consisting mainly of gibbsite and a minor amount of
goethite, a slight increase in adsorption capacity was observed with increasing
heat treatment temperature up to 600C followed by a decrease at temperatures
above 600 C as shown in Figure 5.3. Similar results can be found in the
literature (Hauge et al, 1994) for clays with high concentrations of Al oxides. The

104
decline in adsorption capacity above 600C can be attributed to the gradual
conversion of gibbsite into -Al2O3 that has a smaller tendency to adsorb F -.

Figure 5.3 below shows F- adsorption on a bauxitic clay (MD2) as a function of

heat treatment temperatures.

95
% Asdsorption

90
85
80
75
70
200 300 400 500 600 700 800
Temperature

Figure 5.3
Percentage adsorption for the bauxitic clay (MD2) as a function of
heat-treatment temperature

In the case of goethite a similar reduction in adsorption capacities after heat

treatment at temperatures above 600oC were observed (Wang and Reardon,
2001). This may be attributed to the formation of magnetites, a form of iron oxide
which show little tendency to adsorb F-. The temperature of 600oC therefore
represents an upper limit above which F- adsorption decreased for the clay types
containing Al and Fe oxides. Should clays be used as substrates for F- removal,
lower temperatures could be used without significantly changing adsorption
capacities.

The dolomite in samples such as CAL can also play a role in F - adsorption. It was
found that adsorption increased after these clays were heated to 600oC. This
could be attributed to the conversion of calcium and magnesium carbonates into

105
oxides of these metals. Subsequent exposure of the freshly formed CaO to F-,
could result in the formation of CaF2. This process could contribute to F- removal
from the solution through a precipitation mechanism rather than an adsorption or
ion exchange mechanism.

Figure 5.4 summarises the effect of heat treatment and chemical treatment on
the adsorption capacities of the selected clay types. Chemical treatment had a
significant effect on the adsorption capacities of certain clays such as
palygorskite-dolomite (CAL), kaolinite-goethite (SLV), and amorphous Al2O3
(RBM), but little change was observed for gibbsite-goethite clay (MD2).
Activated alumina was included as a reference. The pre-treatment procedures
were not expected to affect the adsorption characteristics of this sorbent.

100
90
80
70
% Adsorption

CR
60
C
50
CT
40
CTC
30
20
10
0
MD2 CAL SLV RBM ALU

Figure 5.4
The effect of heating and chemical treatment on adsorption capacities
of selected samples. CR = clean raw, C = heated for 1 h at 600C,
CT = chemically treated with 1% Na2CO3 and 1% HCl, CTC = heated
and followed by chemical treatment

X-ray diffraction data for heated and non heated clays provide an insight into the
crystallographical changes that take place during heating and can be used to

106
explain why heating at 600C increase adsorption capacities for certain clays.
The X-ray diffractograms given in Figures 5.5 to 5.8 show the structural changes
that occur in:

CAL, (Figure 5.5) where the dolomite structure is decomposed and

CaCO3 and MgCO3 are converted into Mg and Ca oxides. The
dolomite peaks disappear on heating while MgO peaks appear. A
dramatic increase in adsorption capacity was observed after treatment
probably reflecting the formation of CaF2 during the adsorption
process.

RBM, (Figure 5.6) where the amorphous or microcrystalline structure is

too small for proper XRD structural identification. RBM consist partly of
amorphous Al oxide, a mixture of - and -Al2 O3 which do not undergo
any change on heating at this temperature. Its adsorption
characteristics are therefore unchanged.

MD2, (Figure 5.7) where gibbsite is converted into aluminium oxide

and goethite into hematite, these changes do not influence the
adsorption characteristics.

SLV, (Figure 5.8) where kaolinite is decomposed and probably

converted into Al and Si oxides. The released Al oxides, probably -
Al2O3, then contributes to improved adsorption characteristics.

107
 Figure 5.5
X-ray diffractogram for CAL showing structural changes after heating at
600 oC for 2 h

Figure 5.6
X-ray diffractogram for RBM showing structural changes after heating at
600oC for 2 h

108
 Figure 5.7
X-ray diffractogram for MD2 showing structural changes after heating at
600oC for 2h

Figure 5.8
X-ray diffractogram for SLV showing structural changes after heating at
600oC for 2 h

109
5.6 Interpretation of adsorption capacities in terms of structure
and elemental composition

5.6.1 Background

To assess the contribution of amorphous components to the overall F- adsorption

capacity of clays, an ammonium oxalate extraction was performed. The F-
adsorption capacity was determined before and after the extraction. The clays
MD2 and SLV contain gibbsite and goethite and are therefore likely to also have
substantial amounts of amorphous aluminium and iron oxides as a result of
weathering.

5.6.2 Methods

Detailed description of methods used in this study was described in Section

3.4.3.

5.6.3 Results and discussion

Table 5.5 below shows results obtained before and after the removal of the
amorphous aluminium and iron oxides. The results show a drop of ca. 70% in the
adsorption capacity for MD2 at pH6 and a drop of ca. 80% for SLV after
extraction. This would indicate that amorphous Al and Fe oxides play a very
important role in the adsorption capacities of the clays. Ammonium oxalate is
used to dissolve by complexation Al and Fe from amorphous phases in clays and
soils (Parfitt, 1989).

110
 Table 5.5
-
F adsorption before and after ammonium oxalate extraction
using MD2 and SLV at pH 6.
Samples Clean raw (CR) Ammonium oxalate extraction
% F- Adsorption % F- Adsorption
MD2 69 25
SLV 72 15

The filtrates after extraction for both MD2 and SLV were analysed by ICP-OES to
measure the contents of Al, Fe and Mn. Table 5.6 shows the results of Al, Fe and
Mn extracted from the clays.

Table 5.6
The contents of Al, Fe and Mn contents after ammonium oxalate
extraction of MD2 and SLV determined by ICP-OES.
Elements Al Fe Mn
-1 -1 -1 -1 -1
Concentrations mg.L mg.g mg.L mg.g mg.L mg.g-1
MD2 41 3.3 81 6.5 5.2 0.42
SLV 38 3.1 60 4.8 BDL BDL

The major element composition of selected clay samples before extraction was
determined by XRF. The results give n as % metal oxide are given in Table 5.7
and provide support for the positive correlation between adsorption capacity and
the amount of Al and Fe oxides present in the clays.

111
 TABLE 5.7
Major element composition of selected clay samples and
adsorption capacity (% metal oxide)
Oxide ALU RBM MD2 CAL SLV
% Ads 100 80 80 60 67
SiO 2 0 0 9.7 25.3 25.5
Al2O3 100 92.9 32.0 2.2 17.5
Fe2O3 0 0.04 31.1 1.24 41.5
TiO 2 0 0 3.04 0.11 2.61
CaO 0 0.03 0 14.2 0.02
MgO 0 0 0 21.6 0
LOI 0 6.34 23.0 34.6 12.1

The data clearly show that the adsorption capacity decreases as the Al and Fe
oxide content decrease. In the case of CAL, the Al and Fe oxide contents are low
but the adsorption capacity is still relatively high. CAL is a dolomitic clay and the
high Ca and Mg content provides an alternative mechanism for F- removal from
the solution as discussed above. The high loss on ignition (LOI) of 35% for this
clay reflects the decomposition of Ca and Mg carbonates in their respective
oxides and the concomitant release of CO2.

5.7 Rate of F - adsorption

5.7.1 Methods

Detailed descriptions of the methods used in this study are given in Section
3.4.4.4.

112
5.7.2 Results and discussion

The residual F- concentrations as a function of time are shown in Figure 5.9.

Initially the rate of adsorption of F- onto clay samples was very fast. In the first 15
minutes most of the F- was adsorbed. With a further increase in time, a marginal
increase in adsorption was observed up to 60 min. There was no increase in F-
removal between 1 and 6 hours. Adsorption was completed in about 60 minutes
regardless of the type of the adsorbent used. The changes in the rate of removal
may be due to the fact that, initially, all adsorbents sites were vacant and the F-
concentrations gradient was high. Afterwards, the F- uptake rate by the
adsorbent decreased significantly, due to decrease in adsorption sites.
F- concentration (mg.L-1)

5.5
5
4.5 ALU (A)
4 RBM (CT)
3.5
3 MD2 (CR)
2.5
2 SLV (CT)
1.5 CAL (CT)
1
0.5 MD4 (CR)
0 RYE (CR)
0 100 200 300 400 500 6 0 0 ALU (N)
Time (min)

Figure 5.9
Residual F vs time curves for different substrates (1g) and 10 mg.L-1
-

solutions

113
5.8 Leaching of impurities from substrates during adsorption

5.8.1 Background

Clays contain significant quantities of various trace elements. Leaching of

impurities from clays during adsorption could have adverse health effects if these
impurities are of a toxic nature, Al being potentially the most problematic. At
concentrations of 0.08 mg.L -1 the presence of aluminium was found to result in
Alzheimer disease (Srinivasan et al., 1999). The presence of higher
concentrations of aluminium in clays will render the clays unfit to be used as
defluoridation substrates. This study was done to investigate the presence of
leaching impurities from clays.

5.8.2 Methods

The metals Al, Fe, Cu, Ca, Mg, Cr, Mn, Ni, Zn, and As were determined in
leachates by ICP-OES (see Section 3.4.9) after 5 g of clay was extracted with
water and 10 mg F -.L-1 solutions for 2 h.

5.8.3 Results and discussion

The results are summarised in Table 5.8 for clays washed with water only and
clays that were chemically pretreated.

114
 TABLE 5.8
Metal impurities (mg.L ) in clay leachates extracted with 10 mg.L-1 fluoride solutions
-1

before and after chemical pre-treatment with 0.1 M Na2CO3 and 1% HCl
Metal DL* ALU ALU RBM RBM MD2 MD2 SLV SLV CAL CAL
Before After Before After Before After Before After Before After
Al 0.10 1.15 < DL 3.67 < DL < DL < DL < DL < DL 0.71 1.68
As 0.05 < DL < DL < DL < DL < DL < DL < DL < DL < DL < DL
Ca 0.05 2.68 0.23 < DL < DL 0.18 < DL 1.4 < DL 4.64 115
Cr 0.05 < DL < DL < DL < DL < DL < DL < DL < DL < DL < DL
Cu 0.02 < DL < DL < DL < DL < DL < DL < DL < DL 0.02 < DL
Fe 0.05 < DL < DL < DL < DL < DL < DL < DL < DL 0.51 < DL
Mg 0.05 0.11 < DL < DL < DL 0.06 < DL 0.58 < DL 10.9 61.5
Mn 0.02 < DL < DL < DL < DL 0.23 < DL < DL < DL < DL < DL
Ni 0.05 < DL < DL < DL < DL < DL < DL < DL < DL < DL < DL
Zn 0.05 < DL < DL < DL < DL < DL < DL < DL < DL < DL < DL

*DL is detection limit of analytical method

It is clear that in the case of ALU and RBM, both being aluminium oxide
containing substrates, chemical pre-treatment reduced the amount of Al leached
out to below the detection limits of the analytical method. The reverse result was
obtained for CAL, a dolomitic clay, where chemical pre-treatment increased the
release of Al. The use of dolomitic clays for fluoride removal is therefore not
advisable since Al levels in the leachate before and after chemical treatment are
too high. For the other substrates chemical treatment effectively cleans the
material to such an extent that metal impurities in the leachate are negligible.
Although metals are released by the clay samples, their concentrations remain
low. The concentrations of Cr, Ni, Cu and Zn remained below the detection limit
of the analytical method before and after chemical pretreatment. In summary, the
use of clay samples as adsorbents for F- removal could be an alternative for

115
commercially available activated alumina due to their low cost and good
efficiency.

5.9 Determination of particle size distribution

5.9.1 Background

Many investigations have been carried out on the use of clay minerals for F-
adsorption, but little attention has been given to the particle size of the
suspended clay minerals. This section will deal with the particle size distributions
on selected clay samples.

5.9.2 Methods

Detailed descriptions of the methods used in this study are described in Section
3.4.10.

5.9.3 Results and discussion

Cumulative curves of the individual clay samples particle size fraction are given
in Figure 5.10 to 5.18.

The particle size distribution data for clays provide the information on larger
particles present when clays are suspended into deionised water. Figures 5.10 to
5.18 below show the particle size distribution as a function of diameter.

116
 2500000

2000000

Size ( m)
1500000

1000000

500000

0
0 20 40 60 80 100
Diameter (m )

Fig 5.10
Size (m) vs diameter for 1g of ALU suspended in 50 mL deionised
water.

1500000

1000000
Size (m)

500000

0
0 20 40 60 80 100
Diameter ( m )

Fig 5.11
Size (m) vs diameter (m) for 1g RBM suspended in 50 mL of
deionised water.

2 5 0 0 0 0 0

2 0 0 0 0 0 0
Size ( m)

1 5 0 0 0 0 0

1 0 0 0 0 0 0

5 0 0 0 0 0

0
0 2 0 4 0 6 0 8 0 1 0 0
D i a m e t e r ( m )

Fig 5.12
Size (m) vs diameter (m) for 1g of SLV suspended in 50 mL of de
ionised water.

117
 3 0 0 0 0 0 0
2 5 0 0 0 0 0

2 0 0 0 0 0 0

Size ( m)
1 5 0 0 0 0 0
1 0 0 0 0 0 0

5 0 0 0 0 0

0
0 2 0 40 60 8 0 1 0 0
Diameter ( m )

Fig 5.13
Size (m) vs diameter (m) for 1g CAL suspended in 50 mL of de
ionised wa ter.

3 0 0 0 0 0 0

2 5 0 0 0 0 0

2 0 0 0 0 0 0
Size (m)

1 5 0 0 0 0 0

1 0 0 0 0 0 0

500000

0
0 2 0 4 0 60 80 1 0 0
Diameter ( m )

Fig 5.14
Size (m) vs diameter (m) for 1g MD2 suspended in 50 ml deionised
water

2000000

1500000
Size (m)

1000000

500000

0
0 20 40 60 80 100
Diameter ( m )

Fig 5.15
Size (m) vs diameter (m) for MD4 suspended in 50 mL of
deionised water

118
 2500000

2000000

Size (m)
1500000

1000000

500000

0
0 20 40 60 80 100
D i a m e t e r ( m )

Figure 5.16
Size (m) vs diameter (m) for 1g RYE suspended in 50 mL of de
ionised water.

2500000

2000000
Size (m)

1500000

1000000

500000

0
0 20 40 60 80 100
Diameter ( m )

Fig 5.17
Size (m) vs diameter (m) for 1g coated Kaolin (Al(OH)3)
suspended in 50 mL of deionised water.

2500000

2000000
Size (m)

1500000

1000000

500000

0
0 20 40 60 80 100
Diameter ( m )

Fig 5.18
Size (m) vs diameter (m) for 1g coated Bentonite (Al(OH)3)
suspended in 50 ml of deionised water.

119
The particle size distributions of the clay samples ranges from 4 to 8 m. The
clay samples ALU, CAL, MD2, MD4, RYE and coated kaolin (RNF) show a
bimodel distributions at 4 to 8 m. Only RBM, SLV and coated bentonite show a
mono-dispersion around 2 to 3 m. The reason for a bimodel distributions
appeared on several clay samples is difficult to explain. In general, particle sizes
of different clays are close to each other which allow comparison of adsorption
data.

5.10 Coated particles

5.10.1 Methods

Detailed description of the methods and analytical techniques used in this study
are described in Section 3.4.11.

5.10.2 Results and discussion

The Table 5.9 below summarises the results obtained after coating kaolin and
bentonite with Al and Fe hydroxides. The results were compared with the ones
without coating.

120
 Table 5.9
F- adsorption at pH 6 of kaolin and bentonite coated with
Al and Fe oxides and comparison with natural samples.
Samples Coating % F- Ads
Kaolin (RNF) No coating 45.1
Al oxide 72.3
Fe oxide 61.3

Bentonite No coating 49.6

Al oxide 95.3
Fe oxide 80.4

The results show that coati ng clay samples with Al and Fe oxides improved the
F- adsorption of the clays. Coating clays using Al oxide improved the F-
adsorption more than Fe oxide does. Kaolin (RNF) after coating with Al oxide
increased F- adsorption 45.1 to 72.3% and bentonite from 49.6 to 95.3%. The F-
adsorption on bentonite after coating increased more than for kaolin.

5.10.3 Conclusions

Clay samples coated with Al and Fe oxides increased F- adsorption. This result
confirms that Al and Fe oxides plays a very important role in the adsorption
capacities of the clays. Clay particles coated with Al and Fe oxides or hydroxides
also provide an alternative source of sorbent material for defluoridation systems.

121
 CHAPTER 6
Laboratory scale column defluoridators

6.1 Background

Laboratory scale column defluoridators were developed to provide a quick and

cheap way of assessing the performance of clays as sorbents for F-.
Breakthrough volumes were determined to assess the effect of different
operational conditions on the defluoridation efficiency. This included column
regeneration procedures. The performance of several clay types are compared.

Mini columns packed with sorbent were used to study the suitability of selected
clays to be used in defluoridation columns. The main objective was to compare
the results obtained using the batch type adsorption procedure discussed in
Chapter 5 with that from the dynamic type adsorption conditions found in column
adsorption. Important factors that were studied included: the rate of F- OH-
exchange relative to flow rate through the column, column capacity, regeneration
efficiency, and the effects of flow rate, pH and concentration on column
performance.

6.2 Column design and operational conditions

Mini columns consisted of 180 x 14 mm glass tubes fitted with no 1 porosity frits
on both ends. The columns were connected to a Gilson Minipuls 3 peristaltic
pump using Tygon tubing. Columns were typically packed with 4 g sorbent. The
flow direction was from bottom to top. The pH of F- feed solutions were adjusted
to pH 6 to ensure proper conditions for adsorption to occur. A detailed
description of the procedure is given in Section 3.4.12.

122
Clay samples were washed with deionised water until F- free, dried at 105C,
compacted in a Perkin Elmer pellet press at 5t (pellet diameter was 35 mm), heat
treated at 600C for 2 h, cooled in a desiccator, and crushed in a Siebtechnik mill
to particle size < 2 . This procedure was followed to stabilise the clay samples
to prevent loss of small particles during column operation.

6.3 Comparison of the use of acetic acid and hydrochloric acid

for pH adjustment to pH 6 using ALU.

6.3.1 Background

This study was undertaken to investigate the use of acetic acid for pH
adjustment. Acetic acid (vinegar) is a cheap acid which can be used in domestic
units for pH adjustment instead of hydrochloric acid. In this work the acids were
compared for chemical preconditioning and pH adjustment.

6.3.2 Method

Description of the method used in this study is described in Section 3.4.4.1.

6.3.3 Results and Discussion

Table 6.1 shows the results obtained using HCl and HAc for pH adjustment.

123
 TABLE 6.1
Comparison on the use of acetic acid and hydrochloric acid
for pH adjustment to pH 6 using activated alumina (ALU)
Acid used % F- Adsorption
HCl 94
CH3COOH 97

The results show that acetic acid can be used for pH adjustment instead of
hydrochloric acid. At pH 6 F- adsorption using acetic acid gives 97% adsorption
as compared to the 94% using hydrochloric acid.

6.4 Breakthrough volumes and column capacities

Breakthrough volumes and column capacities were determined for activated

aluminium (ALU) samples at different flow rates using a 4 g column packing and
a 10 mg.L -1 F- feed solution.

Column capacities given in Table 6.2 were calculated from the breakthrough
volumes determined from curves for residual F- vs time curves shown in Figure
6.1.

124
 TABLE 6.2
Breakthrough volumes, residual F- concentrations in non-saturated
columns, and column capacities for a column with 4 g activated
aluminium and a 10 mg.L- 1 F- test solution. Flow rate 4 mg.L-1
Flow rate Residual Breakthrough Breakthrough Adsorption
-1 -
(mL.min ) [F ] time (m) vol (mL) capacity
-1
(mg.L ) (mg.g-1)
2 0.6 520 1040 2.7
4 0.8 260 1040 2.7
8 1.0 100 800 2
10 1.2 75 750 1.9

10
-1

8
Residual F/mg.L

2ml/min
6
-

4ml/min

4 8ml/min

10ml/min
2

0
0 100 200 300 400 500 600
Time/min

Figure 6.1
Residual F - concentration vs time after passage of a 10 mg.L-1 F feed solution
through a 4 g activated alumina column at different flow rates.

The curves show that the residual concentrations for non saturated columns
increased with flow rate from about 0.5 to 1.2 mg.L -1 a clear indication that
exchange kinetics required a limit to the flow rate at about 2 mL.min-1 in order to
maintain a F- at the ideal level at 0.6 mg.L -1. The drop in adsorption capacity as

125
the flow rate increased also point to a limitation on flow rates because of kinetic
considerations.

The effect of flow rate on the column performance for F- adsorption on activated
alumina was investigated in a packed-bed column. Initially, most of the F- was
adsorbed, so the F- concentration in the effluent was low. As the adsorption
continued, the effluent concentration increased, slowly at first but then abruptly.
The general position of the breakthrough curve along the volume axis depends
on the capacity of the column with respect to the feed concentration, flow rate
and bed volume.

The effect of flow rate was studied at 2, 4, 8 and 10 mL.min-1 while the initial F-
concentration was held constant at 10 mg.L -1. At lower flow rates of 2 mL.min-1,
the adsorption was very efficient. Even after breakthrough the column was still
capable of accumulating F- although at a progressively lower efficiency. As the
flow rate increased, the breakthrough curve became steeper and the break point
time and adsorbed F- concentration decreased. This behaviour can be explained
in terms of residence time. If the residence time of the solute in the column was
not long enough for adsorption equilibrium to be reached at that flow rate the
contact time of F- with activated alumina was very short, causing a reduction in
removal efficiency.

6.5 Concentration limitations

The effect of F- concentration in the feed solution on column capacity and column
efficiency was studied using an activated alumina column. The results are given
in Figure 6.2. Summary of breakthrough volumes, residual F- concentrations in
non-saturated columns, and column capacities for a column packed with 4 g
activated alumina at different feed solution concentrations are listed in the Table
6.3.

126
 TABLE 6.3
Breakthrough volumes, residual F- concentrations in non-saturated columns,
and column capacities for a column packed wit 4 g activated aluminium at
different feed solution concentrations. Flow rate 4 mL.min-1.
Feed [F-] Residual [F-] Breakthrough Breakthrough Adsorption
(mg.L-1) (mg.L-1) time (m) vol (mL) capacity (mg.g-1)
5 0.8 560 1020 2.6
10 0.9 260 1040 2.7
20 1.6 120 480 2.8
50 2.5 70 280 3.5

30

25
-1
Residual F /mg.L

20 5 mg/L F
10 mg/L F
-

15
20 mg/L F
10 50 mg/L F

0
0 100 200 300 400 500 600
Time/min

Figure 6.2
Residual F- concentration vs time after passage of different feed solution
concentrations through a 4 g activated alumina column at a flow rate of 4 mL.min-
1

As was expected the concentration of the feed solution is a determining factor in

the efficiency of the column. Residual F- concentrations for non-saturated
columns increased from 0.8 to 2.5 mg.L -1 as the feed concentration increased
from 5 to 50 mg.L -1. At a flow rate of 4 mL.min-1 a feed volume concentration

127
limit of = 10 mg.L -1 is required to maintain a residual F- concentration below 1
mg.L -1.

Removal of F- using a column defluoridator at different initial concentrations of 5

50 mg.L -1 was investigated at a constant flow rate of 4 mL.min-1. Fluoride
adsorption by clay samples in the packed column also depends on the initial
concentration. At the initial concentration of 5 mg F-.L-1, the adsorption with time
was very efficient. As the initial concentration increased, F- adsorbed decreased
and the breakthrough volume also decreased.

6.6 Regeneration procedures

Exhausted clay packings were regenerated by running Na 2CO3 and HCl solutions
through the columns.The efficiency of column regeneration with 0.1 M Na 2CO3
and 1% HCl was studied using activated alumina (ALU) and aluminium oxide
(RBM) packings of 2 g, flow rates of 4 mL.min-1, and feed concentrations of 10
mg.L -1. The results obtained after successive regenerations are shown in Figure
6.3 for ALU and Figure 6.4 for RBM. It is evident that even after 10 regenerations
the breakthrough volumes and hence the adsorption capacities of the columns
did not deteriorate very much. In Table 6.4 the column capacities and residual F-
concentrations are given for successive regenerations. Residual F-
concentrations show only a small increase after each regeneration, about 20%
for ALU and 30% for RBM. Adsorption capacities show the same trend namely a
15% decrease for ALU and a 30% decrease for RBM, after successive
regeneration.

128
 TABLE 6.4
-
Column capacities and residual F concentrations for successive regenerations of 2 g
packings of ALU and RBM. Flow rate 4 mL.min-1. FEED [F-] = 10 mg.L- 1
Regeneration ALU RBM ALU RBM ALU RBM
no. Residual Residual Breaktrough Breaktrough Adsorption Adsorption
- -
[F ] [F ] vol (mL) vol (mL) capacity capacity
-1 -1 -1
(mg.L ) (mg.L ) (mg.g ) (mg.g-1)
1 0.95 1.0 480 400 2.4 2.0
2 0.9 1.2 480 400 2.4 2.0
3 1.01 1.1 440 360 2.2 1.8
4 0.95 1.2 520 360 2.6 1.8
5 1.11 1.2 440 400 2.2 2.0
Ave 1st 5 0.98 1.1 472 384 2.3 1.9
6 1.2 1.1 400 360 2.0 1.8
7 1.12 1.3 400 240 2.0 1.2
8 1.2 1.4 480 320 2.4 1.6
9 1.18 1.4 480 240 2.4 1.2
10 1.20 1.3 440 200 2.2 1.0
Ave 2nd 5 1.18 1.3 440 272 2.2 1.3

Regeneration removes the previously sorbed F- from the surface. The

addition of Na 2CO3 releases F- from the clay surface. The use of 1% HCl
reactivates the surface for F- adsorption. The HCl re-establishes a fully
protonated surface which would be conducive to renewed F - adsorption.

129
 10
9
8

-1
Residual F /mg.L
7
1st
6

-
3rd
5
5th
4
3 7th
2
1
0
0 50 100 150 200
Time/min

Figure 6.3
Residual F - concentration vs time after passage of a 10 mg.L-1 F - feed solution
through a 2 g activated alumina (ALU) column at a flow rate of 4 mL.min-1 after
successive regenerations.

12
-1
Residual F /mg.L

10
1st Reg
8 3rd
-

6 7th
4 8th
2 9th
0 10th
0 20 40 60 80 100 120 140
Time/min

Figure 6.4
Residual F - concentration vs time after passage of a 10 mg.L-1 F - feed solution
through a 2 g aluminium oxide (RBM) column at a flow rate of 4 mL.min-1 after
successive regenerations.

130
6.7 Column capacities for selected clay packings

The suitability of selected clay packings for use in defluoridation columns was
compared by determining adsorption capacities and residual F- concentrations
for non saturated columns. Table 6.5 compares the adsorption capacities and
residual F- concentrations for 4 g sorbent and a feed solution of 10 mg.L -1 F-
passed through a column at 4 mL.min-1. Breakthrough curves for ALU, RBM,
CAL, SLV, MD2, MD4, RYE, coated kaolin and coated bentonite are shown in
Figure 6.5. It is clear that the ability of natural sorbents to remove F- from
solution is not nearly comparable with the activated alumina here used as
benchmark. RBM, amorphous aluminium oxide, however has excellent F-
removal characteristics, not too much different from that of activated alumina.
This oxide is obtained from processed Australian bauxite ore, and is available in
large quantities at Richards Bay Minerals where it is used as feedstock for the
aluminium smelter. It is a relatively cheap and easily available substance. Of the
South African clays only MD2 which is a low grade bauxite, shows any potential
as sorbent for F- removal. Its performance is similar to that of SLV, the Sri
Lankan clay used in clay brick defluoridation in that country. The residual F-
concentrations are ranging from 0.8 for ALU through 1.2 for RBM, 1.4 for MD2
and 1.8 for SLV to totally unacceptable levels of 4.5 mg.L -1 for MD4 and RYE.
MD4 is a low grade South African bauxite from the same area as MD2. This
result emphasises the local variability of clays taken from the same deposit. RYE
is a laterite with a significant Fe oxide component.

Column studies were also performed on coated kaolin (RNF) and bentonite. Both
performed better than CAL, SLV, MD2, MD4 and RYE. This result indicates that
coated bentonite can be used in defluoridation processes.

131
 TABLE 6.5
Breakthrough volumes, residual F- concentrations in non-saturated columns, and
column capacities for a column packed with 4 g of clay different sorbent at
different. Feed solution concentrations 10 mg.L-1. Flow rate 4 mL.min-1.
Clay ID Residual [F-] Breakthrough Breakthrough Adsorption
-1
(mg.L ) time (m) vol (mL) capacity (mg.g-1)
ALU 0.8 250 1000 2.5
RBM 1.2 180 720 1.8
CAL 2.1 90 360 0.9
SLV 1.8 70 280 0.7
MD2 1.4 100 400 1.0
MD4 4.5 40 160 0.4
RYE 4.2 20 80 0.2
Kaolin 2.1 100 560 1.0
(coated)
Bentonite 1.1 140 400 1.4
(coated)

11
10 ALU
Residual F -/mg.L -1

9
8 RBM
7 CAL
6 SLV
5
4 MD2
3 MD4(ENG)
2 RYE(ENG)
1
0 BEN(Coated)
KAOL(Coated)
0 100 200 300 400
Time (min)

Figure 6.5
Residual F concentration vs time after passage of a 10 mg.L-1 F - feed solution
-

through 4 g of different clay sorbents at a flow rate of 4 mL.min-1

132
6.8 Conclusion and recommendations

This study shows that clay types can differ enormously in their adsorption
capacities for F-. Big differences also occur within a specific clay type, even if
taken from the same deposit. A trend in adsorption capacity related to clay
structure can, however, be observed. Adsorption capacity decreases as the
concentration of exchangeable OH- groups decreases. Clay types consisting of
the metal oxides of Fe and Al were found to have the best potential as sorbents
for F - from aqueous solutions.

Chemical and physical pre-treatment increased the adsorption capacity in some

clay types where the protonation of the surface during acid treatment could
create a surface with positive charge.

The bauxitic clays and a processed bauxite (amorphous aluminium oxide)

showed the best overall potential to be used as defluoridation sorbents

133
 CHAPTER 7
Conclusions and Recommendations

7.1 Conclusions

Defluoridation of water using clays as substrates has become popular in many

countries to solve problems related to high F- concentrations in drinking water in
rural areas.

In this work, F- adsorption using South African clays as substrates was studied
for the first time. Results obtained support the notion that F - adsorption onto clays
followed an ion exchange mechanism. The rise in pH observed after F- addition
confirms that OH- was released into the solution. The adsorption model
developed in this study was used successfully in predicting F- adsorption
characteristics of mineral substrates.

Various clays showed good capacity for F- removal. Clays containing aluminium
and iron oxide surfaces, the bauxites and goethite/hematites in particular,
exhibited high adsorption capacity. Consequently, a more detailed study focused
on the bauxites and goethites/hematites (RBM, MD2, CAL and SLV) as sorbents
for defluoridation.

The pH of the solution was found to be the most important factor affecting
adsorption potential. Results obtained show that maximum adsorption was
achieved at pH 5-6 for aluminium oxide types of adsorbents and pH 3.5 -4 for iron
oxide type of adsorbents.

134
Chemical and physical pre-treatment enhanced the adsorption capacities of
some clays. Heat treatment at temperatures of up to 600C was observed to
increase F- adsorption. A decrease in F- adsorption was observed at
temperatures above 700C. Chemical pretreatment had a significant effect on the
adsorption capacities of certain clays. Clays such as palygorskite-dolomite
(CAL), kaolinite-geothite (SLV) and amorphous Al2 O3 (RBM) show an increase in
adsorption capacity after chemical treatment. Little change was observed for
gibbsite-geothite clay (MD2).

Coating of clays such as kaolin and bentonite with aluminium and iron hydroxide
improved their F - adsorption.

Selected clays showed good adsorption capacities using laboratory scale

defluoridator column. The column regeneration procedure showed a slight
decrease in the breakthrough volumes and the adsorption capacities with
successive regenerations of about 10. ALU and RBM gave an excellent F-
removal capacity compared to other clays.

The overall results obtained showed that few of South African clays have good
capacities for F- removal. MD2 is one of South African clays that shows
acceptable performance.

7.2 Recommendations

Findings in relation to the amorphous Al2 O3 (RBM) yielded results of particular

interest. The adsorbent was shown to have a high adsorption capacity both with
and without chemical pre-treatment, and almost identical to that of activated
alumina but at a fraction of the cost.

It is therefore recommended that this material be studied further in full scale

domestic defluoridation columns. This study also showed that a simple process

135
of coating clay particles such as kaolin and bentonite with aluminium hydroxide
can produce useful substrates for defluoridation columns. An investigation of the
feasibility of coated particles in domestic columns defluoridators is
recommended.

136
 BIBLIOGRAPHY

1. AGARWAL M, RAI K, SHRIVASTAV R and DASS S (2003) Defluoridation of

water using amended clay. J. of Cleaner Produc. 11 439-444.
2. APAMBIRE WB, BOYLE DR and MICHEL FA (1997) Geochemistry, Genesis
and health implications of fluoriferous groundwaters in the upper regions of
Ghana. Environ. Geol. 33 13-24.
3. ATKINSON RJ, POSNER AM and QUIRK JP (1967) Adsorption of potential-
determining ions at the ferric oxide aqueous electrolyte interface. J. Phys. Chem.
71 550-558.
4. AZBAR N and TURKMAN A (2000) Defluoridation in drinking waters. Water
Sci and Tech. 42(1-2) 403-407.
5. BALCI S and GOKCAY E (2002) Effects of drying methods and calcinations
temperatures on the physical properties of iron intercalated clays. Materials
Chem. and Phys. 76 46-51.
6. BRDSEN A and BJORVATN K (1995) Fluoride sorption isotherm on fired
clay. Proc. 1st Int. Workshop on fluorosis and defluorication of water. Publ. Int.
Soc. Fluoride Res. 46-49.
7. BARNARDO DJ (1998) Aluminium. In: Wilson MGC and Anhaeusser CR
(eds.) The mineral resources of South Africa. Handbook, Council for Geoscience
16 46-52.
8. BENJAMIN M and LECKIE J (1981) Amorphous iron oxyhydroxide. J. Colloid
Int. Sci. 79 209-221.
9. BJORVATN K, BARDSEN A and TEKLE-HAIMANOT R (1997) Defluoridation
of drinking water by use of clay/soil, Bergen. 2nd Int. Workshop on fluorosis and
defluoridation of water. Publ. Int. Soc. Fluoride Res.
10. BOND GW (1947) A geochemical survey of the underground water supplies
of the Union of South Africa. Department of Mines, Geological Survey Memoir 41
203pp.
11. BOWER CA AND HATCHER JT (1967) Adsorption of fluoride by soils and
minerals. J. Soil Sci. 3(103) 151-154.

137
12. BUFFLE J., PARTHASARATHY N. and HAERDI W. (1985) Importance of
speciation methods in analytical control of water treatment processes with
application to fluoride removal from wastewaters. Water Res. 19(1) 7-23.
13. CAO J, ZHAO Y and LIU J (1997) Brick tea consumption as the cause of
dental fluorosis among children from Mongol, Kazal and Yugu populations in
China. Food and Chem. Toxicol. 35 827-833.
14. CARSTENS IL, LOUW AJ and KRUGER E (1995) Dental status of rural
school children in a sub -optimal fluoride area. J. of Dent. of S.Afr. 50(9) 405-411.
15. CASTEL C, SCHWEIZER M, SIMONNOT MO and SARDIN M (2000)
Selective removal of fluoride ions by a two-way ion-exchange cyclic process.
Chem. Eng. Sci. 55(17) 3341-3352.
16. CHATURVEDI AK, PATHAK KC and SINGH VN (1988) Fluoride removal
from water by adsorption on China clay. Appl. Clay Sci. 3 337-346.
17. CHATURVEDI AK, YADAVA KP, PATHAK KC and SINGH VN (1990)
Defluoridation of water by adsorption on fly ash. Water, Air, and Soil Pollut. 49
51-60.
18. CHERNE T T, TRAFI Y and VALLES V (2002) Mechanism of degradation of
the quality of natural water in the lakes region of the Ethiopian rift valley. Water
Res. 35(12) 2819-2832.
19. CHHABRA R, SINGH A and ABROL IP (1980) Fluorine in sodic soils. Soil
Sci. Soc. AM. J. 44 33-36.
20. CHOI W and CHEN Y (1979) The removal of fluoride from waters by
adsorption. J. AWWA, Water Tech. 562-570.
21. COLEMAN NT and THOMAS GW (1964) Buffer curves of acid clays as
affected by the presence of ferric iron and aluminium. Soil Sci. Soc. Proceedings
28 187-190.
22. CROSBY NT, DENNIS AL AND STEVENS JG (1968) An evaluation of some
methods for the determination of fluoride in potable waters and other aqueous
solutions. Analyst 93 643-652.

138
23. CZARNOWSKI W, WRZESNIOWSKA K and KRECHNIAK J (1996) Fluoride
in drinking water and human urine in Northern and Central Poland. Sci. of the
Total Environ. 191 177-184.
24. DAHI E (1996) Contact precipitation for defluoridation of water. 22nd WEDC
Conference.
25. DAHI E, MTALO F, NJALO B and BREGNHJ H (1996) Defluoridation using
the Nalgonda technique in Tanzania. 22nd WEDC Conference.
26. DEAN HT (1934) Classification of mottled enamel diagnosis. J. Am. Dent.
Assoc. 211421-1426.
27. DEAN HT, ARNOLD FH and ELVOVE E (1942) Domestic water and dental
caries. Publ. Health Report 57 1115-1179.
28. DU PLESSIS JB (1995) What would be the maximum concentration of
fluoride in water that would not cause dental fluorosis? Fluoride and fluorosis.
The status of S. Afr. Res., North West Province. 4.
29. FAYAZI M (1995) Distribution of fluoride in groundwater of the Northern
springbok Flats. The status of S. Afr. Res., Pilanesberg National Park, North
West Province.
30. FOLK RL (1974) The petrology of sedimentary rocks. Hemphill Pub., Austin.
182 pp.
31. FUNG KF, ZHANG ZQ, WONG JWC and WONG MH (1999) Fluoride
contents in tea and soil from tea plantation and the release of fluoride into tea
liquor during infusion. Environ. Pollut. 104 197-205.
32. GACIRI SJ and DAVIES TC (1992) The occurrence and geochemistry of
fluoride in some natural waters of Kenya. J. of Hydrol. 143 395-412.
33. GARMES H, PERSIN F, SADEAUR J, POURCELLY G and MOUNTADAR M
(2002) Defluoridation of groundwater by a hybrid process combining adsorption
and Donnan dialysis. Desalination 145 287-291.
34. GHORAI S and PANT KK (2002) Investigations on the column performance
of fluoride adsorption by activated alumina in a fixed bed. Report, Indian Institute
of Technology.

139
35. GOSH MM and YANG YP (1984) Adsorption of trace arsenic and selenium
on activated alumina. Final Report, US Environmental Protection Agency No.
R809425-01-0.
36. HAO OJ, ASCE AM, HUANG CP and ASCE M (1986) Adsorption
characteristics of fluoride onto hydrous alumina. J. of Environ. Eng. 112(6) 1054-
1069.
38. HASSELBRINK ER, HUNTER MC, EVEN WR and IRVIN JA (2001)
Microscale zeta potential evaluation using streaming current measurements.
SANDIA Report SAND2001-8193.
39. HAUGE S, STERBERG R, BJORVATN K and SELVIG KA (1994).
Defluoridation of drinking water with pottery: Effect of firing temperature. Scand.
J. Dent. Res. 102 329-333.
40. HILLIER S, COOPER C, KELLINGRAY S, RUSSELL G, HUGHES H and
COGGON D (2000) Fluoride in drinking water and risk of hip fracture in the UK: A
case-control study. The Lancet 335 265-269.
41. HORN GFJ and STRYDOM JH (1998) Clay. In: Wilson MGC and
Anhaeusser CR (eds.) The mineral resources of South Africa. Handbook, Council
for Geoscience 16 46-52.
42. HUANG CP and STUMM W (1973) Specific adsorption of cations on
hydrous ?-Al2O3. J. Colloid Int. Sci. 43 409-414.
43. HUERTAS FJ, CHOU L and WOLLAST R (1998) Mechanism of kaolinite
dissolution at room temperature and pressure: Part 1. Surface speciation.
Geochim. et Cosmochim. Acta 62(3) 417-431.
44. JOHNSTON R and HEIJNEN H (2002) Safe water technology for Arsenic
removal. Report. World Health Organization (WHO).
45. KAILASH CA, GUPTA SK and GUPTA AB (1999) Development of new low
cost defluoridation technology (KRASS). Wat. Sci. Tech. 40(2) 167-173.
46. KAU PMH, SMITH DW and BINNING P (1997) Fluoride retention by kaolin
clay. J. of Contam. Hydrol. 28 267-288.
47. KAU PMH, SMITH DW and BINNING P (1998) Experimental sorption of
fluoride by kaolinite and bentonite. Geoderma 84 89-108.

140
48. LAI YD and LUI JC (1996) Fluoride removal from water with spent catalyst.
Separation Sci. and Tech. 31(20) 2791-2803.
49. LARSEN MJ, PEARCE EI and JENSEN SJ (1993) Defluoridation of water at
high pH with use of brushite , calcium hydroxite, and bone char. J. of Dent. Res.
72(11) 1519-1525.
50. LAWLER DF and WILLIAMS DH (1984) Equalization/Neutralization
modelling: An application to fluoride removal. Water Res. 11 25-32.
53. LHASSANI A, RUMEAU M, BENJELLOUN D and PONTIE M (2001)
Selective demineralisation of water by nanofiltration application to the
defluoridation of brackish water. Water Res. 35 3260-3264.
54. LI YH, WANG S, CAO A, ZHAO D, ZHANG X, XU C, LUAN Z, RUAN D,
LIANG J, WU D and WEI B (2001) Adsorption of fluoride from water by
amorphous alumina supported on carbon nanotubes. Chem. Phys. Lett. 38(3)
469-476.
55. LOUW AJ and CHIKTE UME (1997) Fluoride and fluorosis. The status of
research in South Africa, Tygerberg, South Africa. 2nd Int. workshop on fluorosis
and defluoridation of water. Int. Soc. Fluoride Res.
56. LU YC, ESENGUL K and ERSOZ M (2002) Removal of fluoride form
aqueous solution by using red mud. Sep. and Purif. Tech. 28 (1) 81-86.
56. LUTHER SM, POULSEN L, DUDAS MJ and RUTHERFORD PM (1996)
Fluoride sorption and mineral stability in an Alberta soil interacting with
phosphogypsum leachate. Can. J. Soil Sci. 76 83-91.
57. MAHRAMANLIOGLU M, KIZILCIKLI I and BICCER IO (2002) Adsorption of
fluoride from aqueous solution by acid treated spent bleaching earth. J. Fluorine
Chem. 115(1) 41-47.
58. McCAFFREY LP (1994) Fluoride electrode vs ion chromatographic analysis
of groundwater. Abstract Volume, Analytika 94, Stellenbosch, 8-13 December
1994, 87pp

141
59. McCAFFREY LP and WILLIS JP (2001) Distribution of fluoride-rich
groundwater in the Eastern and Mogwase region of the Northern and North West
Provinces. WRC Report No 526/1/01.
60. McFADYEN (2001) Zeta potential of macroscopic surfaces from streaming
potential measurements. Brookhaven Instruments Limited, Worcestershire B96
6ST.
61. MEKONEN A, KUMAR P and KUMAR A (2002) Integrated biological and
physicochemical treatment process for nitrate and fluoride removal. Water Res.
35(13) 3127-3136.
61. MJENGERA H. and MKONGO G. (2002) Appropriate defluoridation
techno logy for use in fluorotic areas in Tanzania. 3rd WaterNet Symposium Water
Demand Management for Sustainable Development.
62. MOGES G, ZEWGE F and SOCHER M (1996) Preliminary investigations on
the defluoridation of water using fired clay chips. J. of Afr. Earth Sci. 21(4) 479-
482.
63. MOHAN RAO NVR and BHASKARAN CS (1988) Studies on defluoridation of
water. J. of Fluorine Chem. 4117-24.
63. MOTHUSI B (1995) Psychological effects of dental fluorosis. Fluoride and
fluorosis. The status of S. Afr. Res. Pilanesberg National Park, North West
Province. 7
64. MOTURI WKN, TOLE MP and DAVIES TC (2002) The contribution of
drinking water towards dental fluorosis: A case study of Njoro division, Nakuru
district, Kenya. Environ. Geochem. and Health 24 123-130.
65. MULLER WJ, HEATH RGM and VILLET MH (1998) Finding the optimum:
Fluoridation of potable water in South Africa. Water SA 24(1) 21-27.
66. NICOLAY A, BERTOCCHIO P, BARGAS E, COUDORE F, CHAHIN GL and
REYNIER JP (1999) Hyperkalemia risks in hemodialysed patients consuming
fluoride-rich water. Clin. Chim. Acta 281 29-36.
67. NISHIMOTO J, YAMADA T and TABATA MASAAKI (2001) Solvent
extraction and fluorometric determination of fluoride ion at ng.L -1 level in the

142
presence of large excess of aluminium (III) and iron (III) by using an expanded
porphyrin, sapphyrin. Anal. Chim. Acta 428 201-208.
68. NOTCUTT G and DAVIES F (1999) Biomonitoring of volcanogenic fluoride,
Furnas Caldera, Sao Miguel, Azores. J. of Volcan. and Geoth. Res. 92 209-214.
69. OMUETI JAI and JONES RL (1977) Fluoride adsorption by Illinois soils. J.
of Soil Sci. 28 546-572.
70. OZAKI M, KRATOHVIL S and MATIJEVIC E (1984) Formation of
monodispersed spindle-type hematite particle. J. of Colloid and Int. Sci. 102(1)
146-151.
71. PADMASIRI JP and ATTANAYA KE MASL (1991) Reduction of iron in
groundwater using a low cost filter unit. J. of the Geol. Soc. of Sri Lanka 3 68-
77.
72. PARFITT RL (1989) Optimum conditions for extraction of Al, Fe and Si from
soils with acid oxalate. Commun. Soil Sci. Plant Anal. 20 801-816.
73. PARTHASARATHY N, BUFFLE J and HAERDI W (1986) Combined use of
calcium salts and polymeric aluminium hydroxide for defluoridation of waste
waters. Water Res. 20(4) 443-448.
74. PEREIRA AC and MOREIRA BHW (1999) Analysis of three dental fluorosis
indexes used in epidemiologic trials. Braz. Dent. J. 10(1) 1-60.
75. PICKERING WF (1986) The effect of hydrolysed aluminium species in
fluoride ion determinations. Talanta 33(8) 661-664.
76. QAFAS Z, KACEMI KE, ENNAASSIA E and EDELAHI MC (2002). Study of
the removal of fluoride from phosphoric acid solutions by precipitation of Na 2SiF6
with Na 2CO3 . Sci. Lett. 3(3) 1-11.
77. RAICHUR AM and JYOTI BASU M (2001) Adsorption of fluoride onto mixed
rare earth oxides. Sep. and Purif. Tech. 24(1-2) 121-127.
78. RAUBENHEIMER EJ, OKONSKA ET, DAUTH J, DREYES MJ, LOURENS C,
BRUWER CA and DE VOS V (1990) Fluoride concentrations in the rivers of the
Kruger National Park: S. Afr. J. Wildlife Res. 20 127-129.

143
79. RUDOLPH MJ, MOLEFE M and CHIKTE UME (1995) Dental fluorosis with
varying levels of fluoride in drinking water. Fluoride and fluorosis. The status of S.
Afr. Res. North West Province 5.
80. SAHA S (1993) Treatment of aqueous effluent for fluoride removal. Water
Res. 27(8) 1347-1350.
81. SCHOEMAN JJ and BOTHA GR (1985) An evaluation of the activated
alumina process for fluoride removal from drinking water and some factors
influencing its performance. Water SA 11(1) 25-31.
82. SCHOEMAN JJ and LEACH GW (1986) An invesigation of the performance
of two newly installed defluoridation plants in South Africa and some factors
affecting their performance. Water Sci. Tech. 19 953-965.
83. SCHOEMAN JJ and MACLEOD H (1987) The effect of particle size and
interfering ions on fluoride removal by activated alumina. Water SA 13(4) 229-
234.
84. SCHOEMAN JJ and STEYN A (2000). Defluoridation, denitrification and
desalination of water using ion-exchange and reverse osmosis technology. WRC
Report TT 124/00.
85. SRIMURALI M, PRAGATHI A and KARTHIKEYAN J (1998) A study on
removal of fluorides from drinking water by adsorption onto low-cost materials.
Environ. Pollut. 99 285-289.
86. SRINIVASAN PT, VIRARAGHAVAN T and SUBRAMANIAN KS (1999)
Aluminium in drinking water: An overview. Water SA 25(1) 47-55.
87. SUSHEELA AK (2001) Sound planning and implementation of fluoride and
fluorosis mitigation programme in an endemic village. Int. workshop on fluoride in
drinking water.
88. SYVITSKI JPM (Ed.) (1991) Principles, Methods, and Applications of Particle
Size Analysis. Cambridge Uni v. Press, New York. 368 pp.
89. Van Den HOOP MAGT, CLEVEN RFMJ, Van STADEN JJ and NEELE J
(1996) Analysis of fluoride in rain water: comparison of capilary electrophoresis
with ion chromatography and ion selective electrode potentiometry. J. of Chrom.
A 739 241-248.

144
90. VELDE B (1995) Composition and mineralogy of clay minerals. In: Velde B
(ed.) Origin and mineralogy of clays. Springer-Verlag, New York. 8-42.
91. VERESSININA Y, TRAPIDO M AHELIK V and MUNTER R (2001) Fluoride in
drinking water: The problem and its possible solutions. Proc. Estonian Acad. Sci.
Chem. 50(2) 81-88.
92. WALVEKAR SV and QURESHI BA (1982) Endemic fluorosis and partial
defluoridation of water supplies A public health concern in Kenya. Comm. Dent.
and Oral Epidem. 10(3) 156-160.
93. WANG Y and REARDON EJ (2001) Activation and regeneration of a soil
sorbent for defluoridation of drinking water. App. Geochem. 16 531-539.
94. WANG W, LI R, TAN J, LUO K, YANG L, LI H and LI Y (2002). Adsorption
and leaching of fluoride in soils of C hina. Fluoride 35(2) 122-129.
95. WU YC, ASCE M and NITYA A (1979) Water defluoridation with activated
alumina. J. of the Environ. Eng. Div. 105 375-367.
96. YANG M, HASHIMOTO T, HOSHI N and MYOGA H (1999) Fluoride removal
in a fixed bed packed with granular calcite. Water Res. 33(16) 3395-3402.
97. YANG CL and DLUHY R (2002) Electrochemical generation of aluminium
sorbent for fluoride adsorption. J. of Hazardous Materials 94(3) 239-252.
98. YOPPS JA and FUERSTENAU DW (1964) The zero point of charge of
alpha-Al2O3 . J. Colloid Sci. 19 61-71.
99. YUCHI A, MATSUNAGA K, NIWA T, TERAO H and WADA H (1999)
Separation and preconcentration of fluoride at the ng mL-1 level with a polymer
complex of zirconium (IV) followed by potentiametric determination in a flow
system. Anal. Chim. Acta 388 201-208.
100. ZEVENBERGEN C, VAN REEUWIJK LP, FRAPPORTI G, LOUWS RJ and
SCHUILING RD (1996) A simple method for defluoridation of drinking water at
village level by adsorption on Ando soils in Kenya. Sci. of the Total Environ. 188
225-232.
101. ZIETSMAN S (1987) The relationship between the geological variation, the
F- of the water and the spatial variation in the occurrence of dental fluorosis in an
endemic area. J. S. Afr. Geol. 71 102-108.

145
102. ZHUANG JIE and YU GUI-RUI (2002) Effects of surface coatings on
electrochemical properties and contaminant sorption of clay minerals.
Chemosphere 49 619-628.

146
 APPENDIX A
Derivations and calculations of alpha values

A.1 Background

This section deals with the derivation of the equations and the calculation of
distribution curves for Al, Fe and Si oxides as a function of pH. Calculations done
on Microsoft Excel are included.

A.2 Derivation of alpha equations

The distribution curves are obtained by plotting the relative concentrations of

polyfunctional acids and bases as a function of pH. These relative concentrations
are denoted as alpha values. If c T is the sum of metal oxide species in the
substrate, the alpha values ca be defined as:

+
[ MOH 2 ]
0 =
cT
[ MOH ]
1 =
cT

[ MO ]
2 =
cT
Where mass balance equation is cT = [MOH2+ ] + [MOH] + [MO-]
For the dissociation reactions:

MOH2 + ? MOH + H+ (1)

MOH ? MO- + H+ (2)

147
The dissociation constant expressions are:

[ H + ][ MOH ]
K a1 = +
(3)
[ MOH 2 ]

[ MO ][ H + ]
Ka2 = (4)
[ MOH ]

Equations (3) and (4) can be rearranged to:

+
K a1 [ MOH 2 ]
[ MOH ] = (5)
[H +]

K a2 [ MOH ]
[ MO ] = (6)
[H + ]

Substituting Equation (5) and (6) from the mass balance equation yields

+ +
+ K [ MOH 2 ] K a1 K a 2 [ MOH 2 ]
cT = [ MOH 2 ] + a1 + (7)
[H + ] [H + ]2

Simplifying Equation (7) gives

+ K K K
cT = [ MOH 2 ] 1 + a+1 + a1 + a22 (8)
[H ] [H ]

Upon rearranging Equation (8) we obtain

+
[ MOH 2 ] [H + ]2
= (9)
cT [ H + ]2 + K a1[ H + ] + K a1 K a 2

148
By definition,

+
[ MOH 2 ]
0 = (10)
cT

Therefore, a0 is given as

[ H + ]2
0 = (11)
[ H + ] 2 + K a1[ H + ] + K a1 K a 2

Expressions for a 1 and a 2 are derived in the same way as above:

K a1[ H + ]
1 = (12)
[ H + ] 2 + K a1 [ H + ] + K a1 K a 2

K a1K a2
2 = (13)
[ H ] + K a1[ H + ] + K a1 K a 2
+ 2

A.3 Calculations of alpha values for the distribution curves

Alpha values were determined using different pH values. The concentration of H+

was calculated from the equation pH = -log [H +].

The pKa values used for Al surfaces are pKa1 (5 to 6) and pKa2 (9 to 10). This
assumption was made since the surface of Al might be either in the form of an
oxide or hydroxide. The same assumption was made for Fe and Si.

The values of a0, a1 and a2 were calculated in Excel. Values used for calculations
are pKa1 = 5 and pKa2 = 9.5. The results are shown in Table A-1.

149
 TABLE A-1
Alpha values for Al oxide calculated at different pH
values.
pH [H+] ?a 0 a1 a2
1 0.1 0.999900019.999E-05 3.16196E-13
2 0.01 0.999001 0.000999001 3.15912E-11
3 0.001 0.990099010.00990099 3.13097E-09
4 0.0001 0.909090650.090909065 2.8748E-07
5 0.00001 0.499992090.499992094 1.58111E-05
6 0.000001 0.090882960.908829639 0.000287397
7 0.0000001 0.009870090.987008717 0.003121196
8 1E-08 0.000968410.968407847 0.030623745
9 1E-09 7.5969E-05 0.759689209 0.240234822
10 1E-10 2.4025E-06 0.240252496 0.759745101
11 1E-11 3.0653E-08 0.030653429 0.96934654
12 1.E-12 3.1523E-10 0.003152309 0.996847691

The pKa values used for Fe surfaces are pKa1 (4 to 5) and pKa2 (9 to 10). The
values of a0, a1 and a2 were calculated in Excel. Values used for calculations are
pKa1 = 4 and pKa2 = 9.5. The results are shown in Table A-2.

TABLE A-2
Alpha values for Fe oxide calculated at different pH
values.
pH [H+] a0 a1 a2
1 0.1 0.999001 0.000999001 2.997E-12
2 0.01 0.99009901 0.00990099 2.9703E-10
3 0.001 0.909090880.090909088 2.72727E-08
4 0.0001 0.49999925 0.49999925 1.5E-06
5 0.00001 0.090906610.909066116 2.7272E-05
6 0.000001 0.009898050.989805008 0.000296942
7 0.0000001 0.000996020.996015936 0.002988048
8 1E-08 9.7078E-05 0.970779536 0.029123386
9 1E-09 7.6922E-06 0.769224852 0.230767456
10 1E-10 2.5E-07 0.249999938 0.749999813
11 1E-11 3.2258E-09 0.032258064 0.967741932
12 1E-12 3.3223E-11 0.003322259 0.996677741

The pKa values used for Si surfaces are pKa1 (2 to 3) and pKa2 (6 to 7). The
values of a0, a1 and a2 were calculated in excel. Values used for calculations are
pKa1 = 2 and pKa2 = 6.5. The results are shown in Table A-3.

150
 TABLE A-3
Alpha values for Si oxide calculated at different pH
values.
pH [H+] a0 a1 a2
1 0.1 0.909090660.090909066 2.72727E-07
2 0.01 0.4999925 0.4999925 1.49998E-05
3 0.001 0.0908843 0.908843043 0.000272653
4 0.0001 0.009871670.987166831 0.0029615
5 0.00001 0.000969930.969932105 0.029097963
6 0.000001 7.6917E-05 0.769171602 0.230751481
7 0.0000001 2.5E-06 0.249999375 0.749998125
8 1E-08 3.2258E-08 0.032258063 0.967741904
9 1E-09 3.3223E-10 0.003322259 0.996677741
10 1E-10 3.3322E-12 0.000333222 0.999666778
11 1E-11 3.3332E-14 3.33322E-05 0.999966668
12 1E-12 3.3333E-16 3.33332E-06 0.999996667

151