Author’s Accepted Manuscript

Design of ceramic filters using Clay/Sawdust

composites: Effect of pore network on the
hydraulic permeability

Martine Youmoue, R.T. Téné Fongang, J. C

Sofack, Elie Kamseu, U. Chinje Melo, Ignas K.
Tonle, C. Leonelli, Sylvie Rossignol
www.elsevier.com/locate/ceri

PII: S0272-8842(16)32363-X
DOI: http://dx.doi.org/10.1016/j.ceramint.2016.12.101
Reference: CERI14411
To appear in: Ceramics International
Received date: 17 August 2016
Revised date: 6 December 2016
Accepted date: 19 December 2016
Cite this article as: Martine Youmoue, R.T. Téné Fongang, J. C Sofack, Elie
Kamseu, U. Chinje Melo, Ignas K. Tonle, C. Leonelli and Sylvie Rossignol,
Design of ceramic filters using Clay/Sawdust composites: Effect of pore network
on the hydraulic permeability, Ceramics International,
http://dx.doi.org/10.1016/j.ceramint.2016.12.101
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
Design of ceramic filters using Clay/Sawdust composites: Effect of pore
network on the hydraulic permeability

Martine Youmoue,1,3 R. T. Téné Fongang,2,3 J. C Sofack,3 Elie Kamseu,3,4 U. Chinje

Melo,3 Ignas K. Tonle,1,2 C. Leonelli,4 Sylvie Rossignol5

1
Electrochemistry and Chemistry of Materials, Faculty of Science, University of Dschang, P.O
Box 67 Dschang, Cameroon
2
Laboratory of Analytical Chemistry (LCA), University of Yaoundé I, P.O Box 812, Yaounde,
Cameroon
3
Laboratory of Analysis of Materials (LMA), Local Material Promotion Authority
(MIPROMALO), Yaounde, P.O Box 2396, Yaounde, Cameroon
4
Department of Engineering ‘Enzo Ferrari’, University of Modena and Reggio Emilia, Via
Vignoles 905/A, 41125 Modena, Italy

Groupe d’Etude des Matériaux Hétérogènes (G. E. M. H), Centre Européen de la Céramique,
5

Université de Limoges, 12 Rue Atlantis-87068 Limoges Cedex, France

* Corresponding author: E-mail address: youmstina@yahoo.fr (M. Youmoue),

Tel: (00237) 679 929 341/ (00237) 695 736 761; Fax: (00237) 243 691 508.

Abstract

Clay based ceramic composite materials with hydraulic permeability were

elaborated using sawdust as porogent agent. Their mechanical, morphological,

microstructural and pore network properties were investigated. Mixtures in various

ratios of two kaolinite clay minerals, Ba (highly plastic) and Va (sand-rich) constitute

the five ceramic matrixes studied (CM1, CM2, CM3, CM4 and CM5). Due to their high

flexural strength, CM3 and CM4 received 0, 5, 10 and 15% sawdust before firing, to

improve the porosity of the final matrixes. Results revealed that 900°C - 1000°C is the

range of temperature necessary to get good sintering and flexural strength (≥ 2 MPa). A

typical clay-sawdust based materials (parallepipedic bricks) present porosity ≥ 40% vol

and 1.5 g/cm3 density. Characterizations such as FTIR, SEM, MIP and flow

1
permeability of ceramic candles were performed. A Hydraulic permeability of ~ 10

mDarcy was obtained and the mean pore diameter varies from 0.05 to 0.1 µm, in

agreement with the microstructure exhibited by the ceramic candles. In the presence of

sawdust, pores with size up to 10 µm were observed, justifying the increase of flowing

permeability. The elaborated matrixes are promising candidates for microfiltration.

Keywords: Filter, hydraulic permeability, ceramic, porosity.

1. Introduction

A recent United Nations (UN) report on enhancement of resources in water

reveals that 663 billion people lack access to sources of potable drinking water and 2.4

billion live without adequate sanitation services [1,2]. Africa possesses the weakest

infrastructure and equipment for cleaning up. Yet, 36% of the African population lack

good sources of water and 70% do not have adequate sanitation services. Even where

the provision of drinking water is shown to be satisfactory, the question of

treatment/purification of wastewater remains problematic. In fact, in Africa and in most

developing countries, about 90% wastewater is rejected in surrounding environment

without treatment [3,4]. This situation leads to many consequences: Cholera and water

bone diseases affect thousands of people worldwide. Developing countries do not have

enough water purification system and the few part made from activated sludge suffer

from maintenance [5] or from climate accommodation [6]. Besides, the efficiency of

ceramic devices in removing various pollutants, bacteria elimination notably has been

clearly demonstrated [7,8,9]. The purification of wastewater by using porous ceramic

based on waste resources has been documented [10]. As shown by many authors, the

porosity and hydraulic permeability of some ceramic filters are improved by waste

resources such as sawdust, rice and coffee husks [11,12].

2
Porosity is known as the most meaningful property to describe a porous material as it

allows to indicate the volume of emptiness (cavity). Porosity is expressed as the fraction

of total volume taken up by pore space. Its quantification is essential to obtain

parameters such as thermal conductivity, mass transfer, diffusion coefficient and

permeability [13]. For a material, the permeability (k) is the ability to let itself cross by

a fluid under the effect of a pressure gradient. This constant is used to evaluate flowing

of liquid through a porous medium and is expressed in Darcy. The ease at which water

flows through a porous ceramic corresponds to the level of permeability [12]. The types

of pores in porous materials that contribute to the processes of filtration are isolated and

interconnected pores [14]. However, the composition of the ceramic matrix, as well as

heating are factors that influence the porosity [15,16]. It has been reported that the

microstructure of a ceramic filter present small particles sizes ranging from 0.02 to 4

µm and that their porosity can vary from 34 to 46% [9,17]. However, an additional pore

from the action of porogent agent in ceramic filter can contribute to increase the flow

rate. There is also a vesicular structure presenting interconnected round pores of 3-4 µm

forming the ideal channels for the hydraulic conduction [9]. So at a given particle size

distribution, pore radii and volumetric porosity correspond a precise permeability which

can decrease as grain radii decrease. Permeability is therefore a function of porosity,

grain size, packing arrangement (size and pores), capillary structure, pore connectivity

and orientation [18, 19]. The hydraulic conductivity (cm/s) can be high (k > 10-1),

median (10-3 ˂ k ˂ 10-5) or low (10-3 ˂ k ˂ 10-7) [9]. This corresponds to the cases of

industrial wastewater, simple wastewater and drinking water respectively. The hydraulic

retention rate of ceramic filter depends on the shape of the filter and can range from 5-

10 L/day according to literature [9,12,7].

3
 In spite of the abundance of clay minerals in Africa, very few research has been

carried out on the production of filter materials based on these low-cost and available

materials. Recently, Belibi et al. [20] have exploited some cameroonian clay minerals to

elaborate support for microfiltration. According to the index of the clay deposits in

Cameroon, the Southen part is endowed with a lot of kaolinite. For instance clay from

Center region in Mvan deposit is used to produce fired refractory bricks [21]. Clay

minerals from Bamessing (Nord-West region, Cameroon) are used locally for traditional

pottery [22]. To obtain the desired properties such as color, plasticity or mechanical

strength, the Local Material Promotion Authority (MIPROMALO, Yaounde,

Cameroon) uses mixtures of clay minerals from Bamessing and Mvan (Center region,

Cameroon) to produce decorative ceramic objects. To the best of our knowledge,

hydraulic and microstructural properties for filtration of fired clay from these deposits

have not been studied. It is therefore necessary to explore filtering capacity of these

clayey systems reinforced with sawdust to produce water purification devices. This

work aims to produce ceramic devices from a mixture of clay minerals with sawdust

and to evaluate their physic-chemical, mechanical and morphological properties.

2. Materials and methodology

2.1 Materials and purification

Two well-known clay minerals samples were selected for this work: a grey

blackish kaolinitic plastic material (referred as Ba) collected from Bamessing area

(Nord-West Region, Cameroon), and a whitish sand-rich kaolinitic clay (referred as Va)

from Mvan (Center Region, Cameroon). Wouatong et al.[22] have found that the main

mineral phases present in Ba sample are kaolinite (~15-16%), quartz (~41-42%), illite

(~15-22%), chlorite (~2-3%), with associated minerals such as goethite (~ 3.5-4%),

4
gibbsite (~0,73-0,74%) and anatase (~1-1.3%). According to Djangang [21, 26],

kaolinite (~67%), quartz (~27%) and illite (~2.3%) constitute the main mineral phases

of Va clay mineral, with anatase (~0.5%), ilmenite (~2.1%) and hematite as associated

phases. The chemical composition of Va is ~67% of SiO2, ~20% of Al2O3, and ~1.6%

of Fe2O3. Before using raw Ba and Va clay minerals, they were soaked in water for two

days and washed at 200 µm, decanted, dried on a plaster vase, then in an oven at

105±2°C, grinded and finally sieved at 63 µm. A sawmill by-product of Ayous

(triplochitonscleroxylon) sawdust was collected in Yaounde (Cameroon) and used as

porogent agent. The sawdust collected was washed, dried in the laboratory for 72 h then

in oven at 150°C for 24 h before grinding and softening ≤ 100 µm.

2.2 Characterization of raw materials

Plasticity parameters of Ba and Va raw clays (400 µm particle size) were

determined using the Casagrande method, on a CNTROLS Model 22-TOO 30/G

apparatus: the liquid limits were 99% for Ba and 35% for Va while the plastic indexes

were 51 and 16 for the same materials respectively.

2.3 Preparation of ceramic candles

2.3.1 Investigation of the best formulation

To determine the optimum Ba/Va/S composite formulations with appropriate

mechanical properties, parallepipedic (8 x 4 x 1 cm3) shape bricks were elaborated.

Ceramic matrixes were made by mixing Ba and Va clays in various weight ratios. They

were referred as CM1 (100 % Va + 0 % Ba), CM2 (75% Va + 25% Ba), CM3 (50%

Va+ 50% Ba), CM4 (25 % Va + 75% Ba) and CM5 (100% Ba + 0% Va). CM3 and

CM4 that exhibited good mechanical properties were reinforced with 0, 5, 10 and 15%

of sawdust fiber. Each formulation was weighed and then mixed with around 10% of

water, pressed on stainless mould with hydraulic press. The briquettes were first dried in

5
open air at room temperature during three days, then in oven at 105 ± 2°C for 24 h.

Dried specimens were fired respectively at 800, 900 and 1000°C in an electrical furnace

(Insuni-Mic model) at a heating rate of 3°C/min; during 270, 300 and 330 minutes with

the holding time of 2 h for each temperature. The samples were free cooled in the

furnace up to 25°C.

2.3.2 Design of candles

Purified clay materials were weighed in appropriate ratio, mixed and then

homogenized by adding water to control the plasticity. Mixed plastic paste was shaped

on an electrical tower (Shimpo Whell 21 model) rotating at 210 tr/min. Considering the

linear shrinkage of 4.3% maximum (obtained from the previous briquettes specimens),

the dimensions of green cylindrical ceramic candle were as follows: height 10.5 cm,

thickness 0.53 cm and diameter 4.16 cm. Green candles were kept on a plastic bag

during 4 days to allow the progressive migration of water [23], dried on open air during

two weeks, then fired in a furnace.

2.4 Characterization of parallelipedic bricks and ceramic candles

The dimensions of parallepipedic dried bricks (Ld, ld, ed), fired bricks (Lf, lf, ef)

and cylindrical candle ceramic filters were recorded with vernier calipers (10-1

precision). For each formulation of bricks, many samples were manufactured. The linear

shrinkage after fired was obtained using the formula 100 x (Ld - Lf) x Ld-1. The bulk

density of elaborated ceramic (in g/cm3) was obtained by calculating the ratio of the

fired specimen weight (Mf) to its volume (Vf). Fired ceramics were weighed and soaked

in pure water fo 24 h (Mw), and the apparent porosity of samples was determined

according to Equation (A.1) [20,24]. The flexural strength of specimens was recorded

on a three points bend test IGM hydraulic press (minimum strength value 0.02 kN) and

6
computed according to Equation (A.2) [17]. All tests were carried out at a rise speed of

0.06 mm/min.

Ba and Va raw clay minerals samples and powders (80 µm particle size) of CM3

and CM4 fired at 900°C were characterized using Fourier Transformed Infra-Red

spectroscopy on a Bruker alpha-p spectrometer. Thus, about 2 mg of each sample were

mixed with 200 mg of KBr to form pellets that were analysed in absorbance mode.

Microstructures of ceramic filter were observed on a Scanning Electron

Microscope Philips XL30.

The pore size distribution and the average pore diameter of sintered candle

filters were determined using Mercury Intrusion Porosimeter (Autopore IV 9500 V1.09)

working from 0.10 to 33000.00 psia, corresponding to pore size from 353 to 0.01 µm.

Hydraulic permeability of candles were evaluated by simply collecting the flow

volume over time of non soaked filters, filters soaked during 1 day, and filters soaked

during 5 days. For this procedure, a known volume of water was poured on the filter

suspended on gallows, the filtered water flowing through the funnel was collected in a

burette and the volume was recorded after each 10 min. Additional permeability values

(herein expressed in mDarcy) of these samples were reported by MIP.

3. Results

3.1. Phases transformation

Fig. 1 presents the FTIR spectra of Ba and Va raw clay minerals, and of their

selected mixtures fired at 900 and 1000°C. Regarding Ba raw material, the band

appearing at 3620 cm-1 is attributed to the stretching vibration of -OH bond

characteristic of water and to the –OH groups from the octahedral sheet of kaolinite

[25,26]. The asymmetric vibration of Si-O-Al bonds is observed at 1164 cm-1 [27,28].

7
Also, the symmetric vibration of Si-O-Si appears at 750 cm-1. Additional small band of

Si-O from quartz are observed at 683 cm-1 and 848 cm-1. Similar bands are identified in

Va sample. For this later, the high content of quartz sand justified the intensity of the Si-

O-Si bending vibration with slight reduction of water and clay content. Other series of

bands have been obtained between 543 cm-1 and 693 cm-1 that were attributed to Fe-O

stretching from hematite (Fe2O3) and goethite FeO(OH)2 [25,29] contained in raw clays.

The thermal treatment at 900°C (CM4-0) and 1000°C (CM3-0) contribute to eliminate

the bands at 3620 cm-1 and those at 1600 cm-1 with more structuration of the band at

1100 and 450 cm-1. These two main bands are characteristic of structural reorganization

of clay minerals with the formation of new crystalline phases as mullite and

cristoballite. Many authors [30,31] have suggested that the transformations occurring

when kaolinite is heated could be explained by the following equations:

T ≥ 500°C: 2SiO2.Al2O3.2H2O 2SiO2.2Al2O3 + 2H2O (1)

Metakaolinite

T ≥ 980°C: 2SiO2.Al2O3 SiAl2O4 + SiO2 (2)

Metakaolinite Spinel phase Amorphous silica
Or 2SiO2.Al2O3 Al2O3 + 2SiO2 (3)
γ-Alumina

T ≥ 10000°C: SiAl2O4 + SiO2 1/3(3Al2O3.2SiO2) + 4/3(SiO2) (4)

Mullite
Or 3Al2O3 + 4SiO2 3Al2O3.2SiO2 + 2SiO2 (5)

The spinel structure is amorphous while mullite is made of AlO6 octahedral and

(Al,Si)O4 tetrahedral chains. As suggested by the literature [30] β-quartz and tridymite

are formed above 870°C; elsewhere, the mullitisation of Bamessing clay Ba could begin

near 900°C [22]. So, the stretching bands from 692 to 778 cm-1 in CM4 fired at 900°C

may be attributed to β-quartz and tridymite while stretching bands at 779 and 693 cm-1

in CM3 heated at 1000°C may correspond to Al-O bond on γ-Alumina and mullite. The

8
absence of -CH band on FTIR spectra of ceramic candles means that the sawdust

material has been completely destroyed by the firing process.

3.3. Shrinkage, bulk density, apparent porosity and flexural strength properties

The variation of fired linear shrinkage, bulk density, apparent porosity and

flexural strength with temperature and sawdust content are summarized into Fig. 2, 3, 4

and 5 respectively. Fig.2a revealed that all the matrixes without sawdust fired at 800°C.

Matrixes CM2 and CM5 fired at 900 and 1000°C present linear shrinkage of ~ 0.5%.

Then, the linear shrinkage of CM3 and CM4 at 900°C was found to be about 1.5%.

These results are similar to those obtained by Ker Das et al. [32] for the mixtures of

several clays. Va clay is a limous low plastic clay which allows to reduce the plasticity

of ceramic composites. The presence of sand mineral phase in Va could also explain the

sintering behavior.

At 900°C, a linear shrinkage of ~3% for CM3 and ~4% for CM4 is observed. Ceramic

matrixes CM1 and CM2 have high percentage of Va which is a sand-rich clay and

doesn’t present any significant sintering up to 1000°C. This is in agreement with the

refractory behavior as confirmed by Djangang [26]. The low value of shrinkage for

CM5 is linked to the predominance of Ba. The formulation CM3 and CM4 corresponds

to mixed proportions with better sintering properties.

The percentage of sawdust added in the formulations did not highly affect the overall

value of shrinkage (Fig. 2b and 2c). Hence, the decrease of linear shrinkage compared

to other traditional ceramics is due to the presence of associated minerals such as quartz,

abundant in Va clay mineral.

One can notice that the addition of Ba tends to decrease the bulk density of the

mixtures. There is no significant evolution of the bulk density with the variation of

9
temperature (Fig. 3a). The value of the bulk density which is ~1.66 g/cm3 for CM1

doesn’t change upon variation of the temperature between 800 and 1000°C. The density

of CM3 and CM4 decreases inversely relatively to their high linear shrinkage (Fig. 2a),

as already indicated, this behavior is due to the sand-poor nature of the Ba clay mineral

used in this work. As it can be observed on Fig. 3b and Fig.3c, the bulk density

generally decreases from 1.66 to ~ 1 g/cm3 when the percentage of sawdust increases

from 0 to 15%. This is an interesting finding for the quality of filters that should be

lightweight materials. Meanwhile, it was observed that increasing the percentage of

sawdust induces an increase in the apparent porosity (Fig.4a and 4b) while the flexural

strength decreases (Fig. 5a & 5b). In fact the apparent porosity of the mixture of Ba/Va

is between 30-40% within the range of temperatures of 900-1000°C (Fig. 6a). For

composite membranes from natural aluminosilicate, Almandoz et al. [18] have obtained

a porosity in the same order of magnitude. With the addition of 5% sawdust, the

apparent porosity increases to 42%, and to almost 50% when the sawdust was added at

15% (Fig. 4a & 4b). Thus, the mixture of clays is molded with particles of sawdust.

This sawdust contains more than 90% of organic material which will be carbonized,

giving rise to a new class of porosity in the clay ceramic product. This may justify the

increase of the total porosity of the filters [12,33,34].

When the concentration of sawdust is high, the number of pores due to the

carbonization of sawdust is important. These capillary pores, due to their size, decrease

the mechanical properties of the composites. Fig. 5a describes how the fraction of Ba

into the mixture significantly improves the mechanical behavior of the prepared ceramic

composites. While none of both clays can achieve optimum flexural strength,

formulations with 50 and 75% of Ba (CM3 and CM4 respectively) give flexural

strength above 3 Mpa (Fig. 5a). It is known that during the sintering of clay, the

10
reactivity and phase transformation of constituents like quartz and hematite are

significant for their densification. Adding the sawdust ((5%), the flexural strength

decreases from 3 to 2 Mpa and even below 1 Mpa for 10 and 15% of sawdust. In

summary, ceramic with high density exhibit good resistance and upon adding 5% of

sawdust, ceramic slightly lose mechanical properties, in accordance with the results

achieved by Djangang [26]. Overall, there is a strong correlation between addition of

wood fibers, the apparent porosity, the bulk density and the flexural strength as reported

in literature [33, 35]. The range of porosity obtained is acceptable for microfiltration, as

previously reported by many authors [15,17,20].

Regarding the statistical analysis, the dispersion of the flexural strength of clay

matrices without wood fibers fired at 900°C is shown on Fig. 6. The variation

coefficients of matrixes at 800, 900 and 1000° C (fig.12) are 0.37, 0.14 and 0.29

respectively (Table 1). These results reveal that the dispersion decreases when the

temperature rises from 800 to 900°C. Thus, on the basis of statistical laws, the lowest

the variation coefficient is, the better is the composition for optimal flexural strength.

3.4. Microstructure

Some pictures of ceramics are presented in appendix (B). The color of designed

filters is affected by the relative amount of hematite (Fe2O3) and rutile (TiO2) found in

raw clay minerals, as previously mentioned [21,22,26]. The final color of ceramic

candles is yellow-brown, more significant for samples containing 75% of Ba clay. The

texture of the final products as indicated by the pictures is typical of that of fired clay

ceramics. At first glance, no porosity is visible, indicating that the pore’s network of

ceramic filters is formed with nano and micropores as analyzed with mercury intrusion

parameters (see next paragraph).

11
 Fig.7 shows the SEM images of the typical ceramic filters with optimum

properties (CM4 and CM3) fired at 900°C and 1000°C respectively. At low

magnification (Fig.7a and 7d), the microstructure exhibits the influence of firing

temperature. At 900°C, the matrix remains with high concentration of capillary pores.

The observation of surface fracture of both matrices indicates relative high mechanical

strength of matrix at 1000°C compared to that at 900°C, results that can be confirmed

with Fig. 5a. At higher magnification (Fig. 7b & 7e), the difference in compactness due

to temperature is obvious. The bonds due to the sintering at 1000°C are significant with

evidence of the slight vitrification responsible of the significant reduction of large pores.

Fig. 7c shows the close contact between grains which is optimized at 1000°C with

vitrification. Thus, at 900°C, the microstructure of fired ceramic candles can be

described as multiparticles system closely disposed forming agglomerated aggregates

with pores dispersed into the matrix. At 1000°C, it can be described as a continuous

network with the liquid phase that reduces the extent of individuation of particules

together with porosity. In both interparticles agglomerates and continuous vitrified

systems, grains of quartz are dispersed as a function of silica present into raw materials,

an evidence of non-reached residues [36]. Due to the dense structure of candles, others

minerals are not easily identified. This result is similar to the microstructure described

by Almandoz et al. [18].

The micrographs of CM3 (900°C) and CM4 (1000°C) ceramic candles with 5%

of sawdust as porogent agent are shown on Fig. 8. At low magnification, it is observed

the impact of the densification, which has been reduced with significant increase of the

porosity. The matrix is now near to a sponge-like structure (Fig. 7a). However the

increase of temperature from 900 to 1000°C contributed to the compactness of the

matrix due to the action of vitrification (Fig. 8d). As in the case of formulation without

12
sawdust, the vitrification is responsible for the optimum densification with reduction of

porosity and pore size (Fig. 8b, c, d & e). In the matrixes with sawdust, the cracks

propagation is significant (Fig. 8) justified with the increase of the total porosity (Fig.

4b & 4c) and the important reduction of the strength (Fig. 5a & 5b). The microstructure

investigation confirms that suggested firing temperature of ceramic candles based on

mixtures of Ba and Va clay minerals range from 900 to 1000°C.

3.5. Relation between the pore network and the hydraulic permeability

Appendix (B) shows the pictures of the ceramic candles filters with no open

larger porosity identified. This confirms the classes of porosity collected with mercury

intrusion porosimetry (Fig. 9 and Fig. 10). In fact, specimen with or without sawdust

show largest pores with size ≤ 200 µm, the cumulative pore volume of the ceramic

candles is ~ 0.16-0.23 mL/g which is found on most fired clay-based materials [17]. In

the presence of sawdust in CM3, the cumulative pore volume increases above 0.2 mL/g

(Fig. 11c). The most important issue is the increase of the connectivity of pores with the

presence of sawdust. The cumulative pore curve of filter CM3-0 (1000°C) is typical for

a monomodal pores with the peak at 0.1 µm. The band of pores cover size from 0.01 to

~ 1 µm in CM3-0 (1000°C), with just a small band between 1 to 10 µm and the band

with peaks centered at 100 µm. Very fine particle size of sawdust increase the pore

connectivity in the range 0.01 to ~1 µm without increase of pores > 1 µm in CM4-0

(900°C) while particles of sawdust produces new bands of larger capillary pores with

peak centered at 10 µm (Fig. 9c & 9d).

Fig. 11 presents the variation of cumulative pore volume with pore size

distribution of CM3-0 (1000°C), CM3-5 (900°C), CM4-0 (900°C) and CM4-5 (1000°C)

filters. Their values are respectively around 0.17, 0.19, 0.23 and 0.16 mL/g (Table 2).

Filter CM4-0 (900°C) presents the highest value of total intrusion volume. The pore

13
band centered at 10 µm observed on Fig. 11c and Fig. 11d for CM3-5 (900°C) and

CM4-5 (1000°C) are capillary pores present in typical firing ceramic whose volume

increase with the presence of coarse grains of sawdust.

The level of permeability of a porous ceramics gives a measure of the relative

ease at which water (or any other fluid) will flow through it. Fig. 10 presents the

behavior of the filters during the out-flow of distilled water through the non soaked

materials, and then filter soaked during 1 and 5 days. When the filters are not soaked, it

is observed that the out-flow of the filters without sawdust is lower; and that the passage

of water through the filters reinforced by the sawdust is raised more, precisely for filter

CM3-5 (900°C) than for CM4-5 (1000°C) during the first 30 minutes. However, the

tendency reverses itself from the 40th minute. When the reinforced filters were soaked 1

day in water, it was observed that the out-flow through the CM3-5 (900°C) was always

faster than for CM4-5 (1000°C). Similar results were obtained when the filters were

soaked during 5 days. This may be explained by the fact that when the filters are

soaked, the isolated pores are saturated completely and the passage of water makes itself

through the interconnected pores. These pores are created by the burnt organic matter

with consequent increase of the permeability [17,20]. The results of the filtration test are

in agreement with those of the mercury intrusion porosimetry. The permeability values

of CM3-5 (900°C) and CM4-5 (1000°C) are of 9.8 mDarcy and 9.1 mDarcy

respectively, high compared to the permeabilities of CM3-0 (1000°C) and CM4-0

(900°C) that are 4 mDarcy and 0.14 mDarcy respectively. The highest value of

permeability of CM3-5 (900°C) and CM4-5 (1000°C) may be due to the big pores

present in these matrixes according to the microstructure described in the previous

section.

14
Table 2 shows that the values of average pore diameter of CM4-0 (900°C) and CM4-5

(1000°C) are respectively 0.13 µm and 0.15 µm; suggesting that they can be used to

remove viruses whose size ranges from 0.01 to 0.1 µm. This result is in accordance with

those reported by Yakub [17]. However, the average pore diameter of CM3-0 (1000°C)

and CM3-5 (900°C) is 0.05 µm, suggesting that the high content of sand-rich Va clay on

these filters may contribute to decrease the mean pore size of ceramic candles.

4. Discussion

We successfully used sawdust to dimension the porosity (particularly the fine

fraction of pores) of fired ceramic filters (Fig. 9b & 9c). Amongst the two formulations

with optimum sintering and mechanical properties, CM4 (75% of Ba and 25% of Va) is

easily molded due to the content of Bamessing clay (Ba) which is more plasic than

Mvan (Va). The brown colour of fired products increases in intensity with Ba content. It

appears clearly that the success of having desired filters with indicated porosity and

permeability as well as color is linked to appropriate mixtures of clay materials.

Although having a porous structure, their stability and durability are insured with silica,

cristobalite and mullite that result from phase’s transformation during sintering. Due to

its sedimentary nature, Ba clay mineral provides glassy phase forming, and induces the

enhancement of sintering of illite and of iron minerals. A review of literature reveals

that most ceramic filters are porous in structure [12,20,34], with pores occupying 30 to

48% of total volume. Yakub et al. [12] obtained for unfired ceramic pastes pores

ranging from 36 to 47 vol%. The results obtained in this work (30-42% vol) are in the

same order of magnitude. However, attention was paid on the pore size distribution that

allow ceramic candles filters to have particular class of permeability (Fig. 10).

15
 The permeabilities of the filters produced in this work are similar to those

reported by Yakub et al. [12], but low compared to others [14,23]. The permeability of

ceramic candles with sawdust (~50 mL) is between 9.1 and 9.8 mDarcy, corresponding

to a flow rate of ~ 8 mL/H after complete saturation in water. With the success of

appropriate control of the pores size distribution and the possibility to avoid larger

capillary pores, it seems possible with these results to design ceramic filters with

prediction of permeability acting on fine pores range though the control of the

concentration and granulometry of sawdust. These filters with pore size between 0.01

and 1 µm (Figure 9) are porous and promising candidates for microfiltration [12].

The effect of the turtuosity and the total pore area are essential on the removal

efficiency of pathogenic organisms. Hence total pore area of ceramic candle CM3-

5(900°C) and CM4-5(1000°C) are respectively 13.04, and 4.32 m2/g while their

turtuosity are respectively, 19.02 and 16.88. From these results, it is clear that the

ceramic filter based on CM3 matrix is more tortuous and exhibited high total pore area.

It is important to notice that the particles of sawdust are randomly distributed within the

clay mineral during the mixing of ceramic plastic paste (described on the first section of

this work). It was realized through mercury intrusion parameter that pore size

distribution curves show an intensive band of pore from 0.01 to ~ 1µm (Fig. 9). But the

band is larger with filter based on CM4. When considering the filters with sawdust

soaked in water for one day, all the empty spaces are filled and the flow through CM3-5

(900°C) is higher than that through CM4-5 (1000°C), thereby justifying their

permeability and flow rate behavior (Table 2 and Fig. 10). In other way, pore area

reveals that filters based CM3 may adsorb pathogenic organisms than filters based on

CM4. The total pore area of ceramic candle CM3-5 (900°C) is high than that of filters

reported by Yakub et al. [12] and Van Halen [34]. Finally, it is expected that the

16
production of permeable and low-cost ceramic candles based on mixtures of Ba and Va

clay minerals, precisely on CM3 and CM4 matrixes fired at 900°C could be interesting

from an economic point of view.

5. Conclusion

In this work ceramic candle filters were elaborated. Some were made by mixing two

clay minerals and sawdust, whereas others were fabricated without sawdust. Their

formulations were obtained by studying pressed ceramic bricks considering their

flexural strength and apparent porosity. Composites which have exhibited flexural

strength above 2MPa with around 42% of apparent porosity have been found to be

interesting. It was found that formulations CM3-5 and CM4-0 present good mechanical

properties at 900°C while 1000°C is suitable to get good strength from CM3-0 and

CM4-5. To summarize:

 The mechanical behaviors of ceramic candles was attributed in one hand to the

composition of matrix, the firing temperature, the mineral phase (as mullite, amorphous

quartz) contained in sample after firing; and on other hand to the presence of capillary

pore in fired ceramics.

 The pore network is formed by pores channels and the pore distribution depends

on the composition of clay minerals mixtures and on firing temperature. The mean pore

of 0.05 µm and 0.1 µm obtained for the prepared filters could be helpful to filtrate

viruses. Filters CM3-5 (900°C) and CM4-5 (1000°C) have exhibited almost the same

permeability. Around 8-10 mL of water was flow through elaborated filters (50 mL), a

value convenient for microfiltration.

 The turtuosity and pore size area of ceramic filter CM3-5 (900°C) are slightly

high than CM4-5 (1000°C), meaning that the first filter is suitable to adsorb and remove

17
pathogenic agents. Nevertheless, the presence of nanoscale and micropores is interesting

to filter virus. It is therefore necessary in a future work to evaluate the performance of

filters to eliminate viruses and the efficiency to leach metallic compounds or dye

usually found in waste water.

Acknowledgments

The authors acknowledge the support of the "Agence Universitaire de la Francophonie

(AUF)" through the programme “Soutien aux Equipes de Recherche (Research project

AUF/2016-2017/N°282). They also thank MIPROMALO laboratory staff for their

assistance in the preparation of specimens.

References

[1] Nations Unies, Rapport Mondial des Nations Unies sur la mise en valeur des ressources en

eau, 2016, http://unesdoc.unesco.org/images/0024/002440/244041f.pdf:// (Access date:

01.07.2016).

[2] Nations Unies, Objectifs du Millénaire pour le développement, Rapport, 2015,

http://www.un.org/fr/millenniumgoals/reports/2015/pdf/rapport_2015.pdf (Access date:

30.09.2015).

[3] Nations Unies. Faits et chiffres sur la mise en valeur des ressources en eau. L’eau dans un

monde durable. Rapport Mondial, 2015,

http://unesdoc.unesco.org/images/0023/002322/232273f.pdf (Access date: 30.09.2015).

[4] UNESCO, World Water Development report 4, Managing water under uncertainty and risk,

2012. http://unesdoc.unesco.org/images/0021/002156/215644e.pdf (Access date:

30.09.2015).

[5] J. Nya, F. Brissau, I. M. Kengne, C. Drakides, A. Amougou, E. Atangana, T. Fonkou, P. L.

Agendia, Traitement des eaux usées domestiques au Cameroun : Performance épuratoire

18
 comparées du lagunage macrophytes et du lagunage macrophytes, Proceeding of

International Symposium on Environnemental Control on Water Management, 7-10

January Tunis (EPCOWM) (2002) 726-736.

[6] D. Kone, Epuration des eaux par lagunage à microphytes et à macrophytes en Afrique de

l’Ouest et du Centre : Etat des lieux, performances épuratoire et critère de

redimensionnement. Thèse de Doctorat, Ecole Polytechnique Fédérale de Lausanne,

Lausanne, Suisse, 2002.

[7] G. Rytwo, S. Nir, N. Shuali, Clay and water treatment, Appl. Clay Sci. 67 (2012) 117-118.

[8] S. Gourari, A. Achkari-Begdouri, Use of baked clay media as biomass supports for

anaerobic filters, App. Clay Sci. 12 (1997) 365-375.

[9] J. J. Simonis, A. K. Basson, Manufacturing a low-cost ceramic and filter system for the

elimination of common pathogenic bacteria, Phy Chem Ear. 50-52 (2012) 269-276.

[10] T. Kato, K. Ohashi, M. Fuji, M. Takahashi , Water absorption and retention of porous

ceramics fabricated by waste resources, J. Ceram. Soc. Jpn. 19(2) (2008) 212-215.

[11] W. Han, Q. Yue, S. Wu, Y. Zhao, B. Gao, Q. Li, Y. Wang, Application and advantages of

novel clay ceramic particles (CCPs) in an up-flow anaerobic bio-filter (UAF) for waste

water treatment, Biotechnol. 137 (2013) 171-178.

[12] I. Yakub, A. Plappally, M. Leftwich, K. Malatesta, K. C. Friedman, S. Obwoya, F.

Nyongesa, A. H. Maiga, A. O. B. Soboyejo, S. Logothetis, W. Soboyejo, Porosity, Flow

and Filtration Characteristics of Frustum-Shaped Ceramic Water Filters, J. Environ. Eng.

139 (2013) 986-994.

[13] A. Khalili , M. R. Morad, Matyka , B. Liu, B. R. Malekmohammadi, J. Weise, M. M.

Kuypers, Porosity variation below fluid-porous interface, Chem. Eng. Sci. 10 (2014)

311-316.

[14] S. C. Bloem, Silver Impregnated Ceramic filter, Flow rate versus the removal of pathogens,

PhD Thesis, Delft University of Technology, Netherlands, 2008.

[15] L. Lianga, J. Li, H. Lin, X. Yang, G. Guo, M. He, Relation between Porosity and

Permeability of Ceramic Membrane Supports, Adv. Mater. Res. 11-12 (2006) 39-42.

19
[16] L. C. Molina Félix, L. A. Belanche Mu˜noz, Representing a relation between porosity and

permeability based on inductive rules, J. Pet. Sci. Eng. 47 (2005) 23-34.

[17] I. Yakub, Micro- and Nano-porous adsorptive materials for removal of contaminants from

water at point of use. PhD Thesis, 2012. http://pratica.org/wp-

content/uplood/2014/Bloem-Sophie-onderzoe rapport.pdf

[18] M. C. Almandoz, C. L. Paqliers, N. A. Ochoa, J. Marchise, Composite ceramic membranes

from natural aluminosilicate application, Ceram. Int. 41 (2015) 5621-5633.

[19] Y. Yang, A. Aplin, A pemeability-porosity relationship for mudstones, Mar. Pet. Geol. 27

(2010) 1692-1697.

[20] P. Belibi Belibi , M. M. G Nguemtchouin , M. Rivallinb, J. Ndi Nsami, J. Sieliechi, S.

Cerneaux, M. B. Ngassoum, M. Cretin, Microfiltration ceramic membranes from local

Cameroonian clay applicable to water treatment, Ceram. Int. 41 (2014) 2752-2759.

[21] C. N. Djangang, A. Elimbi, U. C. Melo, G. L. Lecomte, C. Nkoumbou, J. Soro, J. Yvon, P.

Blanchart, D. Njopwouo, Refractory ceramics from clays of Mayouom and Mvan in

Cameroon, Appl. Clay Sci. 39 (2008) 10-18.

[22] S. L. Wouatong, S. Tchounang Kouonang, Hawa Mohamed, K. P Yerima, P. Bernard, U. C.

Melo, D. Njopwouo, Chemical and mineralogical characterization of clayey deposits for

traditional pottery and brickworks in the North West Region of Cameroon (Central

Africa), Bri. J. Appl. Sci. Tec. 9(1) (2015) 19-36.

[23] S. Iaich, D. Lahmani, L. Messaoudi, A. Chahlaou, N. Tijani, Elaboration et caractérisation de

nouveaux supports membranaires tubulaires monocanaux à base d’argiles de la région

Meknes-Tafilet destinées au traitement des eaux domestiques par filtration tangentielle,

Sci. Lib. 5 (2013) 1-26.

[24] T. Kato, K. Ohashi, M. Fuji, M. Takahashi, Water absorption and retention of porous

ceramics fabricated by waste resources, J.Ceram. Soc. Jpn. 19(2) (2008) 212-215.

[25] J. Madejovà, FTIR techniques in clay minerals studies, Vib. Spectrosc. 31 (2003) 1-10.

20
[26] C. N. Djangang, Argiles kaolinitiques de Mayoum et de Mvan : Caractérisation et utilisation

dans l’élaboration des matériaux réfractaires. Thèse de Doctorat, Université de Yaoundé

I, Yaoundé, Cameroun, 2007.

[27] A. Elimbi, H. K. Tchakoute, D. Djoupwouo, Effects of calcinations temperature of kaolinite

on the properties of geopolymers cements, Const. Buil Mater. 25 (2011) 2805-2812.

[28] J. M. Moutou, R. Mbedi, A. Elimbi, D. Njopwouo, J. Yvon, O. Barres, H. R. Ntekela,

Mineralogy and Thermal Behaviour of the Kaolinitic Clay of Loutété (Congo-

Brazzaville), Res. J. Env. Ear. Sci. 4(3) (2012) 316-324.

[29] A. S. Fish, P.M. Fredericks, Fourier transformed infrared studies of aluminous geothites and

hematites, Clays Clay. Miner. 32(5) (1983) 377-382.

[30] C. A. Jouenne, Traité de céramiques et minéraux, Ed. Septima, Paris, 2001.

[31] J. Anggono, Mullite ceramics: Its properties structure and synthesis, J. Tek. Mesin, 7(1)

(2005)1-10.

[32] S. K. Das, K. Dana, N. Singh, R. Sarkar, Shrinkage and strength behaviour of quartzitic and

kaolinitic clays in wall tile compositions, Appl. Clay Sci. 29 (2005) 137-143.

[33] F. O. Aramide, Production and characterization of porous insulating bricks from clay with

sawdust, J. Miner. Charc. Eng. 11 (2012) 970-975.

[34] D. Van Halem, S. G. J. Heijman, A. I. A. Soppe, J. C. Van Dijk, G. L. Amy, Ceramic silver-

impregnated pot filters for household drinking water treatment in developing countries:

Material characterization and performance study, Sci. Technol. Water Suppl. 7(5-6)

(2007) 9-17.

[35] G. Viruthagiri, S. Nithya Nareshananda, N. Shanmugam, Analysis of insulating fire bricks

from mixtures of clays with sawdust addition. Indian J. Appl. Res. 3(6) (2013) 469-473.

[36] E. Kamseu, C. Leonelli, P. Veronesi, P. Miselli, P. Giancarto, U. Chinje. Melo,

Characterization of porcelain compositions using china clays from Cameroon, Ceram.

Int. 33 (2007) 851-857.

[37] A. K. Chakravorty, D. K. Ghosh, Synthesis at 980°C and Phase Development of Some

Mullite Gels, J. Am. Ceram. 71 (1988) 978-87.

21
 Appendice A

M w  Mf
AP(%)  ……………………………………..………………………(Eq. A.1)
ρ e Vf

Mf = Weight of fired sample (g),

Mw = Weight of soaked sample during 24 h in water (g),

Vf = Volume of each fired briquette (cm3),

ρe = Volumic mass of water (~1g/cm3).

3.F.d
FS(MPa)  ……………………………………..…………………...……(Eq. A.2)
2.e. 2

F= Recorded strength (kN),

d = Distance between the two rollers of hydraulic press (mm),

e = thickness of briquette (mm)

ℓ = Width between of briquettes used on three points flexural strength (mm)

22
Appendice B: Photographs of elaborated ceramic candle filters. F1: CM3-0 (1000°C),
F2: CM3-5(900°C), F4: CM4-0 (900°C) and F4: CM4-5 (1000°C)

23
Figure captions

Fig. 1: FTIR spectra of raw Ba and Va clays and matrixes CM3 and CM4 without
sawdust fired at 1000°C and 900°C respectively.

Fig. 2: Variation of linear shrinkage at 800°C, 900°C and 1000°C: (a) with the ceramic
matrixes; (b) and (c) with the adding of sawdust in CM3 and CM4 respectively.

Fig. 3: Variation of bulk density at 800°C, 900°C and 1000°C: (a) with the ceramic
matrixes; (b) and (c) with the adding of sawdust in CM3 and CM4 respectively.

Fig. 4 : Variation of apparent porosity at 800°C, 900°C and 1000°C: (a) with the
ceramic matrixes, (b) and (c) with the adding of sawdust in CM3 and CM4 respectively.

Fig. 5: Variation of flexural strength at 800°C, 900°C and 1000°C: (a) with the ceramic
matrixes, (b) and (c) with the adding of sawdust in CM3 and CM4 respectively.

Fig. 6: Statistical distribution of flexural strength of ceramic matrix without sawdust,

fired at 900°C.

Fig. 7: Micrographs of ceramic candles without sawdust : (a), (b) and (c) are SEM of
CM4-0 (900°C) respectively at magnifications x50, x200 and x1000 while (d), (e) and
(f) are MEB of CM3-0 (1000°C) at the same magnifications.

Fig. 8: Micrographs of ceramic candles reinforced with 5% of sawdust : (a), (b) and (c)
are SEM of CM3-5(900°C) respectively at magnification x50, x200 and x1000 while
(d), (e) and (f) are MEB of CM4-5 (1000°C) at the same magnifications.

24
Fig. 9: Pore size distribution in ceramic candle filters: (a) CM3-0 (1000°C), (b) CM4-0
(900°C), (c) CM3-5 (900°C) and (d) CM4-5 (1000°C).
Fig. 10: Test of hydraulic permeability of ceramic candle filters. Flowing volume versus
time of non-soaked filters CM3-0 (1000°C), CM4-0 (900°C), CM3-5 (900°C) , CM4-5
(1000°C); and for filters CM3-5 (900°C) and CM4-5 (1000°C) soaked during 1 day, 5
days.

Fig. 11: Variation of cumulative pore volume with pore size diameter for ceramic
candle filters : (a) CM3-0 (1000°C), (b) CM4-0 (900°C), (c) CM3-5 (900°C) and (d)
CM4-5 (1000°C).

Table 1: Statistical analysis of flexural strength of ceramic matrices without sawdust at

800, 900 and 1000°C

800°C 900°C 1000°C

Matrixes Flexural strength (MPa)
CM1 0.69 ± 0.02 1.75 ± 0.02 0.85 ± 0.02
CM2 2.02 ± 0.02 1.98 ± 0.02 1.93 ± 0.01
CM3 1.71 ± 0.02 2.13 ± 0.01 1.92 ± 0.01
CM4 2.76 ± 0.02 3.28 ± 0.02 2.40 ± 0.02
CM5 2.57 ± 0.02 2.66 ± 0.02 3.01 ± 0.02
Variation Coefficient 0.37 0.14 0.29

Table 2: Mercury intrusion parameters of ceramic candle filters

MIP parameters CM3-0(1000°C) CM3-5(900°C) CM4-0(900°C) CM4-5(1000°C)

25
Total pore intrusion (mL/g) 0.17 0 .19 0.23 0.16

Total pore area

12.57 13.04 7.14 4.33
(m2/g)
Average pore diameter (µm) 0.05 0.06 0.13 0.15

Bulk density at 0,51 psia (g/L) 1.67 0.06 1.50 0 .13

Porosity (%) 28.32 26.48 35.91 24.51

Permeability (Darcy) 4.54 9.83 0.15 9.11

Turtuosity 24.82 19.02 130.95 16.88

Ba
Va
CM3-0 (1000°C)
CM4-0 (900°C)

4000 3500 3000 2500 2000 1500 1000 500

-1
Wavenumber (cm )

Figure 1

26
Figure 2

27
Figure 3

28
Figure 4

29
Figure 5

30
Figure 6

31
a b c

500 µm 100 µm 10 µm

d e f

500 µm 100 µm 10 µm

Figure 7

32
a b c

500 µm 100 µm 10 µm

d e f

500 µm 100 µm 10 µm

Figure 8

33
Figure 9

34
Figure 10

35
Figure 11

36