Mineralogical, Geochemical, and Sedimentological Characteristics of Clay Deposits from Central Uganda and Their Applications George W.A

Journal of African Earth Sciences 35 (2002) 123–134 www.elsevier.com/locate/jafrearsci

Mineralogical, geochemical, and sedimentological characteristics of clay deposits from central Uganda and their applications George W.A. Nyakairu a,1, Hans Kurzweil b, Christian Koeberl a,* a Institute of Geochemistry, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria b Institute of Petrology, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria Accepted 10 October 2001

Abstract In Uganda, Precambrian rocks have undergone extensive weathering and erosion, and are locally altered to form considerable clay deposits. We have studied the geochemical, mineralogical, and sedimentological characteristics of clay deposits from central Uganda to determine their composition, source rocks, deposition, and possible use in local industry. Samples were collected from the Kajjansi, Kitiko, Masooli, and Ntawo deposits (near Kampala), all of which are currently used for both industrial and traditional brick, tile, and pottery manufacture. The deposits are widely scattered individual basins, with clays deposited under lacustrine and alluvial environmental conditions, and were all found to belong to the sedimentary group. The clays are composed of silt–sand fractions and predominantly consist of kaolinite and have a relatively high Fe2O3 content. The studied deposits are chemically homogeneous, except for the samples richer in sand fraction, which have higher SiO2 and K2O values. The chemistry of the studied samples, compared to European clays, shows that they need elaborate treatment to render them suitable for ceramics production. An analysis of the chemical and mineralogical composition of the clays has demonstrated that, taken as a whole, they possess characteristics satisfactory for brick production. Ó 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Clay deposits; Precambrian rocks; Uganda

1. Introduction the ever-increasing market demand for the construction materials needed. Traditional methods of production, Clays occur widely in many parts of Uganda. Besides which do not take account of the chemical and miner- their geological interest, they are of importance for local alogical characteristics, are still practiced. In the tradi- industry. They have been used to produce rather poor- tional method of brick production, rawclay material is quality bricks, tiles and pottery by primitive methods for mixed with water and covered for about a week. The several years. Scattered clay pits and brick kilns along paste is placed in a wooden mould as shown in Fig. 2(a) the roadsides document the uncontrolled and low-tech- and (b). The bricks are spread and covered with cut nology exploitation of the Uganda clay occurrences. grass until they are dry. However, during the rainy Apart from artisan brick producers, there are organized season, plastic sheets are used to cover the bricks (Fig. clay works, such as Uganda Clays and Pan African Clay 2(c)). The bricks are fired in field kilns, which consist of Products at Kajjansi, and Allied Clays at Masooli along a large pile of unfired bricks with tunnels in the bottom the Gayaza road, which supply the construction indus- of the pile (Fig. 2(d)). The pile is cemented with clay and try in Kampala and surroundings. Starting in 1986, contains 10,000–15,000 bricks. A wood fire is built in the there has been an increase in construction activity in tunnel and kept burning for 4–6 days and the tunnels are Kampala. The above-mentioned industries cannot meet then closed with unfired bricks and also cemented with clay. The hot exhaust from the wood fire flows through the pile, and heats the center of the pile enough to fire * Corresponding author. Tel.: +43-1-4277-53110; fax: +43-1-4277- the bricks in the core of the pile. The pile is then allowed 9531. E-mail address: [email protected] (C. Koeberl). to cool and dismantled. 1 Current address: Department of Chemistry, Makerere University, Fewstudies have been made of the clays used in the P.O. Box 7062, Kampala, Uganda. brickworks or of raw materials used for pottery in

0899-5362/02/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved. PII: S0899-5362(01)00077-X 124 ...Naar ta./Junlo fia at cecs3 20)123–134 (2002) 35 Sciences Earth African of Journal / al. et Nyakairu G.W.A.

Fig. 1. Generalized geological map of the study and surrounding areas extracted from the geology map of Kampala sheet NA 36-14 (Geological Survey of Uganda, 1962). The inset is a map of Uganda. G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134 125

Fig. 2. Photographs showing the traditional brick production methods used in Uganda. (a) After a heap of raw clay material is covered for a week, the paste is placed into a mould. (b) Mould with clay paste ready for drying. (c) Two moulds used to make bricks, spreading of bricks for drying, cut grass for covering the bricks and also the plastic sheets used in the rainy season in the background, and finished bricks for kiln construction. (d) Partly finished typical brick kiln under construction, with tunnels in which wood fires are built to fire the bricks.

Uganda. Harris (1946) and Kagobya (1950) studied the 2. Geology clay deposit at Ntawo, 25 km from Kampala on the Jinja road (Fig. 1). It was reported that Ntawo clay The study areas are indicated on Fig. 1. The areas are exhibited marked shrinkage and cracked on firing, and mostly underlain by Precambrian rocks that include the quality of product was inferior when it was evalu- sedimentary and metasedimentary lithologies, which ated for pottery production. McGill (1965) studied the comprise fine-grained sandstones, slates, phyllites, and nature and distribution of clays from several occur- schists. The more highly metamorphosed rocks include rences around Kampala, and determined their plasticity quartzites, muscovite–biotite gneisses, and subordinate with a view to establish a fine ceramics industry. Tu- schist, which may locally contain cordierite. The above humwire et al. (1995) measured the physical properties rocks, together with amphibolites and epidosites, are and discussed the geology of the Kajjansi and Kitiko part of the Buganda series, which make up a wider deposits located 13 km from Kampala on the Kampala– Buganda–Toro system with the Toro series of Western Entebbe road. A study of some clay samples from Uganda (Schlueter, 1997). In deeply weathered areas, various deposits in Uganda indicated that they are me- parts of the basement are exposed in the form of un- dium-quality kaolinitic–illitic clays (Nyakairu and Ka- differentiated gneisses and late granites, as well as mig- ahwa, 1998, and references therein). There is ample matized and remobilized parts of the Buganda series. demand for quality bricks and other clay products, and, These rocks are overlain in places by swamp deposits, thus, the present study evaluates the mineralogical and alluvium, and lacustrine deposits. The underlying gneis- chemical characteristics of the rawmaterial used, not sic and granitoid rocks of the Precambrian basement only in individual brick and tile works, but also by the have been extensively weathered and transported to traditional potters. This will help to give a better un- produce clays. Some of these clays are weathering derstanding of the clay materials, as well as of their products of schists and amphibolites, and of basic rocks geochemistry and source rocks. of the younger Buganda series. According to McGill 126 G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134 (1965), the clays can generally be classified as sedimen- Aliquots of 20–30 g of each dried sample were pow- tary and transported alluvial clays. It is from these clays dered in an agate mill. The chemical analyses (major that the samples used in this study were obtained. elements) were performed on powdered samples using Clays derived from gneissic and granitoid rocks of the X-ray fluorescence (XRF) spectrometry at the Univer- basement are leached and enriched in quartz. These sity of the Witwatersrand, Johannesburg, South Africa. clays are thought to have formed by leaching of the For details on procedures, precision, and accuracy, see decomposed bedrock, and are normally separated from Reimold et al. (1994). Mineralogical analysis was per- the bedrock by a layer of large quartzite pebbles (Harris, formed at the Institute of Petrology, University of Vi- 1946). The clays occur as surficial layers with a general enna, Austria, on bulk rock powders using a powder thickness varying between 2 and 5 m (cf. Kaliisa, 1983). X-ray diffractometer (Philips PW 3710) operated at The main features of the studied clays are their lowwet- 45 kV/35 mA using Ni-filtered CuKa radiation, with to-dry shrinkage, refractory nature, and an extremely automatic slit and on-line computer control. The sam- high plasticity, which is attributed to high kaolinite ples were scanned from 2° to 40° 2h. Mineral identifi- content. Due to their high plasticity, these clays can be cation on the diffractograms was processed using Philips classified as ball clays, which makes them suitable for PC-APD software, version 3.5B. The quantitative min- pottery, earthenware, and binders in refractory material eralogical composition was evaluated using the norma- production. In some areas, other clay occurs together tive calculation method of Fabbri et al. (1986). Total with the clays derived from gneissic and granitoid rocks carbon was determined with a LECO elemental analyzer of the basement in the same deposits. The bulk of these (Multiphase carbon determinator, RC-412) at the In- other clays are essentially micaceous schist derivatives, stitute of Petrology, University of Vienna, Austria. which gives them a yellowish-brown color, as opposed to the dull gray appearance of the clays derived from the 3.2. Grain size analysis results gneissic and granitoid rocks of the basement. However, most of the clay is medium gray to brownish gray. In Table 1 lists the results of the grain size analysis of the some areas this type of clay is more thoroughly weath- clay rawmaterials used in brick production for the ered and leached than in some locations found in the construction industry in Kampala, Uganda. The ana- Kajjansi clay field (McGill, 1965). lyzed samples showa large variation in grain sizes. For example, the sand contents range from 8 to 65 wt%, silt content ranges from 19 to 56 wt%, and the clay content 3. Methods and results ranges from 5 to 42 wt%. In Shepard’s diagram, the samples generally plot in the clayey silt, sand–silt–clay, 3.1. Methods and silty sand fields (Fig. 3) (cf. Shepard, 1954). The McManus (1988) diagram (Fig. 4) indicates that the In this study we analyzed 24 clay samples taken from samples are moderately well sorted, with moderately deposits that supply the construction industry in Kam- high porosity and permeability. Thus, according to pala and surrounding trading centers. These samples McManus (1988) the samples are suitable for the con- belong to the Kajjansi (n ¼ 6), Kitiko (n ¼ 8), Masooli struction industry. The Kajjansi clay samples are not (n ¼ 4) and Ntawo (n ¼ 6) locations, as indicated in easy to classify, as they have varying sizes, ranging from Fig. 1. Well-mixed large samples, which were reduced to sand to clay-sized particles (Table 1). The samples plot about 250 g each by coning and quartering, were taken within the silt clay, clayey silt, and sand–silt–clay fields from pits of at least 5 m depth, dug at each of the de- (Fig. 3). The Kitiko samples posses modest percentages posits. Samples were dried at 60 °C, and divided for of sand and variable contents of silt and clay (Table 3), grain size, mineralogical, and chemical analyses. For and belong to the silty clay, clayey silt, sand–silt–clay, grain size analysis, sample aliquots of 100 g were oxi- and silty sand categories (Fig. 3). The Masooli clays are dized and disaggregated with 15% H2O2 and were left to rather uniform in terms of grain size (Table 1). The sand stand overnight (cf. Dalsgaard et al., 1991). Water (200 fraction (>50 wt%) is more abundant than the silt frac- ml) was added to each of the samples, and was disag- tion, with low clay content. The samples can be classified gregated with an ultrasonic probe for 3 min. The sam- as silty sand, clayey sand, and sand–silt–clay (Fig. 3). ples were wet sieved into fractions of 2, 1, 0.5, 0.25, The Ntawo clay samples contain variable sand, silt, and 0.125, 0.063, and 0.032 mm. These fractions were dried clay fractions. Therefore, these rawmaterials are part of at 80 °C, and weighed to 0.1 g. The <32 lm fractions (10 the field of sand–silt–clay, silty sand, and sandy silt clays g) were mixed with sodium hexametaphosphate (20 ml) (Fig. 3). and ultrasonically disaggregated, and grain size analysis The parameters median, mean, sorting, and skewness was performed by X-ray monitoring of gravity sedi- were calculated from the results of the grain size anal- mentation (Micromeritics SediGraph 5100) up to 0:2 lm ysis. The positive skewness of the samples is attributed (cf. Coakley and Syvitski, 1991). to the presence of silt- and clay-size fractions. This result G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134 127

Table 1 Grain size parameters of clays from deposits supplying the construction industry in Kampala, Uganda, obtained by Sedivision 2.0 Sample Wt% Fraction (wt%) First–third moment mm

Clay Silt Sand >63 lm 4–63 lm<4lm >20 lm 2–20 lm<2lm x ra3 CM Kajjansi deposit Kaj-1 41 47 12 12.18 43.58 44.24 49.9 9.24 40.86 7.10 4.05 0.50 0.93 0.02 Kaj-2 28 48 24 24.04 46.12 29.84 65.37 6.57 28.06 6.26 4.19 0.70 1.39 0.04 Kaj-3 27 35 38 37.91 33.79 28.3 68.34 4.83 26.83 5.71 4.41 0.73 1.57 0.05 Kaj-4 38 54 8 7.65 51.33 41.02 51.02 10.9 38.08 7.45 3.96 0.40 0.67 0.03 Kaj-5 22 23 55 46.61 29.19 24.2 66.14 11.91 21.95 4.85 4.61 0.68 3.20 0.08 Kaj-A 26 36 38 37.73 33.47 28.8 62.5 11.08 26.42 5.86 4.31 0.66 1.67 0.04 Kitiko deposit Ks1-1 25 19 56 36.79 36.34 26.87 68.36 6.66 24.98 5.10 4.77 0.56 3.54 0.09 Ks1-2 39 48 13 12.91 45.41 41.68 48.66 12.07 39.27 7.56 4.37 0.17 2.13 0.02 Ks1-3 25 39 36 35.76 36.34 27.9 66.13 8.34 25.53 6.04 3.99 0.84 1.56 0.04 Ks1-4 31 53 16 15.68 47.47 36.85 40.44 28.5 31.06 7.09 3.90 0.23 1.87 0.01 Ks2-1 14 25 61 50.02 34.61 15.37 82.37 3.2 14.43 4.03 3.94 1.21 2.67 0.10 Ks2-2 42 43 15 14.47 41.22 44.31 49.48 8.23 42.29 7.64 4.64 0.07 1.90 0.02 Ks2-3 28 55 17 17.3 47.97 34.73 38.66 33.1 28.24 6.89 4.01 0.21 1.33 0.01 Ks-A 32 36 34 32.47 33.21 34.32 58.37 9.68 31.95 6.34 4.61 0.37 3.01 0.04 Masooli deposit Mas-L 21 38 41 41.56 36.64 21.8 76.76 2.14 21.1 5.27 4.09 1.12 1.23 0.05 Mas-M 26 20 54 54.04 18.39 27.57 66.46 7.93 25.61 5.37 4.47 0.81 1.33 0.08 Mas-U 13 30 57 56.67 28.44 14.89 79.52 7.2 13.28 4.29 3.55 1.41 1.34 0.09 Mas-A 14 35 51 51.36 33.85 14.79 82.67 3.44 13.89 4.43 3.59 1.49 1.49 0.07 Ntawo deposit Ntw-1 24 40 36 36.54 35.06 28.4 58.89 17.38 23.73 5.86 3.92 0.68 1.19 0.04 Ntw-2 6 29 65 64.95 28.12 6.93 91.37 2.22 6.41 2.34 3.47 1.42 3.74 0.33 Ntw-3 5 39 56 55.94 37.84 6.22 89.51 5.07 5.42 3.26 2.85 1.35 2.45 0.10 Ntw-B 13 56 31 30.47 53.13 16.4 65.37 21.18 13.45 5.28 3.16 1.06 0.93 0.04 Ntw-G 27 49 24 23.76 45.53 30.71 56.71 16.21 27.08 6.21 4.12 0.47 1.50 0.04 Ntw-A 34 37 29 28.81 34.66 36.53 57.89 8.07 34.04 6.76 4.39 0.48 0.90 0.04 CM values after Passega (1957, 1964) and Passega and Pyramjeeqffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (1969); grain size fractions are classified after Shepard (1954) and Winkler (1954). 2 2 3 3 Mean, x ¼ðq1x1 þÁÁÁþqnxnÞ=100; standard deviation, r ¼ ðq1ðx1 À xÞ þÁÁÁþqnðxn À xÞ Þ=100; skewness, a3 ¼fq1ðx1 À xÞ þÁÁÁþqnðxn À xÞ g= 100r3; where x is the midpoint of the grain size fraction (measured in phi units); q ¼ percentual frequency of the fraction; C ¼ one percentile; and M ¼ median.

Fig. 4. Ternary diagram of studied clay sediments from Uganda fol- lowing the relation between sand, silt, and clay components and their Fig. 3. Classification of studied clay rawmaterials from Uganda based controls over porosity and permeability, after McManus (1988). WS, on the sand–silt–clay ratios (Shepard, 1954). well sorted; PS, poorly sorted; MWS, moderately well sorted. 128 G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134 can be ascribed either to different sources or the envi- most samples is of lacustrine origin, whereas the other ronmental deposition conditions of the clays. The binary group of samples is formed by fluvial deposition. The diagrams, Fig. 5(a)–(c), showthat the samples fall into sorting indicates that the sediments are moderately to two distinct groups, which indicate different conditions poorly sorted and the cumulative curve (Fig. 6) shows of environmental deposition. The group containing the that most samples plot on the silt–sand size boundary.

(a) (b)

(c)

Fig. 5. Binary plots of studied Uganda clay sediments after Friedman (1967): (a) mean against sorting, (b) skewness against sorting and (c) skewness against mean. Samples fall into two distinct groups of environmental or hydraulic conditions that controlled their deposition.

Fig. 6. Smoothed cumulative frequency distribution of sample Kaj-1, a typical sample, obtained by combining sieve and SediGraph data using computer software Sedivision 2.0. The shape of the curve indicates that the sample is composed of silt to sand sizes. All clay samples give curves with similar shapes. G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134 129 posited by rolling as a transport mechanism. Climatic changes that took place in the Pleistocene and Holocene, with corresponding changes in water levels of Lake Victoria (Bishop, 1969), may have led to clay deposition.

3.3. Chemical results

Tables 2 and 3 showthe chemical and the normative mineralogical composition of the samples, respectively. The chemical data correlate with the mineralogical composition and the silica and alumina contents agree with the quartz and kaolinite contents (Fig. 8). The main oxides are SiO2,Al2O3,Fe2O3, and TiO2, whereas MnO, MgO, CaO, Na2O, K2O, and P2O5 are present only in small amounts. The SiO2 content in Kajjansi clay varies Fig. 7. CM diagram for clay sample from Uganda, after Passega and inversely with the Al2O3 content, and a relatively low Pyramjee (1969). C ¼ one percentile; M ¼ median; fields I, II, (<1 wt%) content of the alkali and alkali earth oxides is III ¼ rolled grains; VII ¼ suspension sediments and rolled ones which apparent. The concentrations of Fe2O3 vary between are graded and uniform (Passega and Pyramjee, 1969). 3.11 and 12.2 wt%. Compared to the Kajjansi clays, Kitiko clays have slightly higher SiO2 and Al2O3,and lower Fe2O3 contents. Considering the chemical com- According to the Passega and Pyramjee (1969) binary position (Table 2), Masooli clays have higher SiO2, plot of median, M, and one percentile, C, called a CM Al2O3,Fe2O3, and TiO2 contents compared to the other diagram (Fig. 7), the sample data plotted within class deposits. The alkali contents are in the same range as the III, which is defined by Passega and Pyramjee as de- other deposits. Chemically, there is a significant amount

Table 2 Chemical composition (wt%) of clays from deposits supplying the construction industry in Kampala, Uganda

Sample SiO2 TiO2 Al2O3 Fe2O3 MnO MgO CaO Na2OK2OP2O5 CO2 LOI Kajjansi deposit Kaj-1 51.59 1.29 22.91 9.68 0.04 0.57 0.38 0.10 0.83 0.08 0.05 12.39 Kaj-2 54.33 1.17 20.62 12.20 0.05 0.13 0.17 0.06 0.74 0.09 0.05 10.58 Kaj-3 70.39 1.67 16.41 3.11 0.04 0.05 0.15 0.08 0.97 0.05 0.06 7.51 Kaj-4 55.26 1.38 23.96 6.09 0.03 0.24 0.25 0.10 0.66 0.06 0.12 12.44 Kaj-5 76.08 1.02 12.29 3.27 0.03 b.d. 0.11 0.07 0.62 0.04 0.11 6.88 Kaj-A 69.07 1.32 14.88 5.19 0.05 0.06 0.17 0.10 0.71 0.09 0.04 8.66 Kitiko deposit Ks1-1 53.43 1.39 23.88 6.84 0.04 0.14 0.16 0.12 1.01 0.25 b.d. 11.88 Ks1-2 58.08 1.20 25.78 3.35 0.04 b.d. 0.09 0.05 0.80 0.10 0.10 10.97 Ks1-3 70.89 1.25 16.78 2.25 0.05 b.d. 0.09 0.05 1.00 0.03 0.09 7.38 Ks1-4 70.88 0.72 14.41 3.62 0.04 0.05 0.20 0.08 0.70 0.04 0.31 8.98 Ks2-1 71.50 1.07 14.42 4.90 0.03 b.d. 0.07 0.07 0.81 0.14 0.10 7.38 Ks2-2 55.25 1.06 27.08 3.38 0.03 0.04 0.16 0.14 0.63 0.04 0.14 12.78 Ks2-3 74.98 0.58 11.92 2.70 0.03 b.d. 0.18 0.08 0.56 0.03 0.38 9.26 Ks-A 64.11 1.11 20.00 4.61 0.03 b.d. 0.13 0.07 0.72 0.13 0.05 9.42 Masooli deposit Mas-L 64.00 0.91 19.28 4.54 0.03 0.10 0.14 0.13 0.99 0.04 0.02 9.72 Mas-M 80.51 0.87 9.96 2.80 0.04 b.d. 0.05 0.12 1.09 0.03 0.03 5.14 Mas-U 80.83 0.81 9.49 2.40 0.10 b.d. 0.06 0.10 0.98 0.04 b.d. 5.66 Mas-A 77.26 0.84 11.01 2.92 0.07 b.d. 0.08 0.11 0.98 0.04 b.d. 6.71 Ntawo deposit Ntw-1 64.65 1.00 16.14 7.64 0.04 0.05 0.13 0.08 0.69 0.10 b.d. 9.84 Ntw-2 84.70 0.82 8.59 1.46 0.04 b.d. 0.04 0.04 0.65 0.02 0.06 4.14 Ntw-3 82.80 0.96 7.61 2.36 0.10 b.d. 0.06 0.06 0.68 0.05 b.d. 5.42 Ntw-B 78.77 1.49 9.38 1.74 0.05 b.d. 0.10 0.12 0.99 0.06 0.24 7.77 Ntw-G 74.79 1.44 13.67 1.88 0.04 b.d. 0.12 0.11 0.75 0.04 0.16 7.45 Ntw-A 70.31 1.31 16.59 2.57 0.04 0.11 0.23 0.17 1.01 0.03 0.06 8.24

All Fe reported as Fe2O3; LOI ¼ loss on ignition; b.d. ¼ belowdetection limit. 130 G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134

Table 3 Calculated normative mineralogical composition (wt%) of clays from deposits supplying the construction industry in Kampala, Uganda, after method of Fabbri et al. (1986) Sample Kaolinite Illite Chlorite Quartz Albite Calcite Hematite Organic matter Accessories Kajjansi deposit Kaj-1 48.9 9.8 2.5 22.6 0.8 0.7 10.0 b.d. 4.7 Kaj-2 44.2 8.7 0.6 29.0 0.5 0.3 13.4 0.1 3.2 Kaj-3 31.1 11.4 0.2 49.8 0.7 0.3 3.4 0.1 3.0 Kaj-4 53.4 7.8 1.0 25.7 0.8 0.4 6.4 0.6 3.8 Kaj-5 24.4 7.3 60.8 0.6 0.2 3.6 0.5 2.7 Kaj-A 29.9 8.4 0.3 50.4 0.8 0.3 5.7 0.8 3.5 Kitiko deposit Ks1-1 49.4 11.9 0.6 23.7 1.0 0.3 7.4 0.7 4.9 Ks1-2 56.7 9.4 26.8 0.4 0.2 3.7 0.2 2.6 Ks1-3 31.8 11.8 50.0 0.4 0.2 2.5 0.2 3.1 Ks1-4 28.8 8.2 0.2 52.9 0.7 0.4 4.0 0.5 4.3 Ks2-1 27.8 9.5 53.5 0.6 0.1 5.5 0.3 2.8 Ks2-2 61.4 7.4 0.2 22.2 1.2 0.3 3.7 0.5 3.1 Ks2-3 24.0 6.6 60.1 0.7 0.3 3.0 1.0 4.3 Ks-A 42.8 8.5 39.6 0.6 0.2 5.1 0.4 2.7 Masooli deposit Mas-L 37.9 11.6 0.4 39.7 1.1 0.3 4.9 0.4 3.6 Mas-M 13.4 12.8 67.3 1.0 0.1 3.1 0.5 1.8 Mas-U 13.4 11.5 68.4 0.8 0.1 2.7 1.3 1.8 Mas-A 17.2 11.5 63.0 0.9 0.1 3.2 1.0 3.0 Ntawo deposit Ntw-1 33.3 8.1 0.2 44.6 0.7 0.2 8.4 0.3 4.1 Ntw-2 14.8 7.6 73.8 0.3 0.1 1.6 0.3 1.4 Ntw-3 11.9 8.0 73.0 0.5 0.1 2.6 1.5 2.4 Ntw-B 12.9 11.6 66.4 1.0 0.2 1.9 2.5 3.4 Ntw-G 26.3 8.8 57.6 0.9 0.2 2.1 0.6 3.4 Ntw-A 30.8 11.9 0.5 49.0 1.4 0.4 2.7 0.1 3.2 Note: Simple normative calculations were based on the assumptions of: uniform mineralogical composition (i.e., all samples contain the same minerals); MgO is used entirely to calculate a chlorite-like phase (Mg:Fe ¼ 1:1); the remaining iron oxide is converted to hematite; CaO is all supplied by calcite; Na2O is used to quantify albite; illite is calculated using all K2O with reference to a hydromica composition (8.5 wt% K2O); the remaining Al2O3 is converted to kaolinite; the remaining silica is quartz; organic matter is estimated by subtracting the CO2 contribution of calcite and converting the remainder to organic carbon by multiplication with a factor of 1.7.

of SiO2, but lower Al2O3 in Ntawo clay compared to the their refractory temperature and, therefore, they are other deposits (Table 2). The alkali contents are in the fluxes (Bain, 1987). The loss on ignition (LOI) is an same range as for other deposits and these samples are important characteristic of clays. In addition to being a depleted in MgO. vital determination in the routine chemical analysis of Because the deposits are located in swamps, plants ceramic materials, the total weight loss is also used as extract K, Na, Ca, and Mg, and add organic com- a means of identifying some minerals and as a rough pounds, such as tannic and humic acid. The relative means of estimating their percentage compositions in abundance of SiO2 indicates a rather high content various samples. In entirely kaolinitic clay the loss on of quartz, whereas Al2O3 can be correlated with clay ignition (13.9 wt%) gives a good measure of the clay minerals and feldspars. Varying amounts of quartz in- content (Searle and Grimshaw, 1959). fluence the plasticity and drying behavior of the clays. A X-ray diffraction (XRD) shows that the samples relatively high iron oxide ðFe2O3Þ content provides contain mainly kaolinite, chlorite, and quartz (Fig. 8). a characteristic reddish-brown color to the fired clay. No iron oxides were detected by XRD, which means However, Fe2O3 is not the only factor responsible for that iron could be present in the form of amorphous the coloring of ceramic wares (Kreimeyer, 1987). Other oxides or hydroxides, or is too lowto be detected by constituents such as CaO, MgO, MnO, and TiO2 can XRD. Kaolinite is more common than illite, whereas appreciably modify the color of the fired clay. The tem- chlorite and smectite are present only in traces. The perature of firing, relative amounts of Al2O3, and the Kajjansi clays are characterized by a kaolinite range furnace atmosphere all play an important role in the from 24.4 to 53.4 wt%, little variation in the illite con- development of color in the fired clay products (Fischer, tent (9 Æ 2 wt%), and quartz content between 22.6 and 1984). The main effect of alkalis in clays is to reduce 60.8 wt%, with organic matter being present at less than G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134 131

Fig. 8. X-ray diﬀraction patterns of bulk clay samples, Ks-A, Ntw-A, Kaj-A, and Mas-A. K ¼ kaolinite; Ch ¼ chlorite; Cl ¼ clay minerals; Q ¼ quartz; and Fp ¼ feldspar.

1 wt%. All but one of the Kajjansi samples have low chlorite content. The clay samples from the Kitiko deposit consist mainly of kaolinite (24–61.4 wt%), illite ð9:5 Æ 2%Þ, quartz (22.2–60.1 wt%), and little chlorite and organic matter. The Masooli clays have lower kaolinite content, but relatively higher illite and albite contents, and much higher quartz contents compared to the Kajjansi and Kitiko clays (Table 3). The clays in the Ntawo deposit contain kaolinite content comparable to that of the Masooli clays (Table 3). The illite content is comparable to that of the Kajjansi and Kitiko clays. These clays are rich in quartz, and have a lowFe 2O3 and an appreciable amount of organic matter. The absence of a significant amount of smectitic Fig. 9. Ternary diagram: quartz/carbonates + Fe-oxides + accesso- minerals will ensure a ceramic body against possible ries + feldspars/clay minerals for the studied clays from Uganda, after difficulties during drying. Sedimentary clays can contain Fiori et al. (1989). considerable amounts of organic matter, which influ- ences the firing properties of clays (Robbins, 1984). The abundance of organic matter is less than 1 wt% for all quartz contents and those that are rich in clay minerals. samples, except for sample Mas-U, which has 1.3 wt%. On the other hand, these samples were classified into Organic matter may change the clay color to light gray, two basic types according to their clay mineralogy blue, brown, or black. Colors, non-clay minerals, and (Table 3): those containing kaolinite and chlorite, and organic materials are similar in all four clay deposits; the those containing only kaolinite. On the basis of these clays differ, however, in the type of the clay minerals. results, and the criteria established by Fiori et al. (1989), samples with the lowest clay fractions are most suitable for manufacturing porous and white ceramic bodies. 4. Suitability for industrial applications The chemical data were plotted in a ternary diagram (silica–alumina–other oxides), as used by Fabbri and The ternary diagram of Fiori et al. (1989) distin- Fiori (1985) to classify rawmaterials and industrial guishes two groups of samples (Fig. 9): those with high ceramic bodies (Fig. 10). This diagram shows ceramic 132 G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134 products (Konta, 1995). According to the results of particle size analysis (Table 1), the rawmaterials cannot be distinguished by grain size as they were distributed all over the fields (Fig. 3). The data were grouped into three fractions (>20 lm, 20–2 lm, and <2 lm) for plotting on Winkler’s diagram to evaluate their suitability for different ceramic products (Winkler, 1954) (Fig. 11). The diagram shows that most of the samples are rich in the silt fraction, resulting in the need for special treatment and processing to render them suitable to produce structural ceramic products. From chemical and mineralogical analysis, the studied clay samples are ball clays (cf. Robbins, 1984).

Fig. 10. Triangular diagram of Uganda clays: SiO2=Al2O3=total ox- 5. Environmental aspects of brick production ides are plotted, where a ¼ red stoneware (Italy); b, b0,b00 ¼ white stoneware for German, English, and French industries, respectively Brick manufacture is associated with a number of (data from Fabbri and Fiori, 1985). environmental implications, some of which are benefi- cial, and others which are potentially detrimental. Tra- ditionally, bricks have been and are produced in compositional fields and reflects the overall chemical factories, which are sited adjacent to the source of the composition of the Ugandan rawmaterials. In this di- clay. The life of the production site may extend beyond agram, some samples plot into the white bodies field, but the life of an associated quarry, or beyond individual most of the samples do not. Taking into account the phases of quarrying, yielding large excavations in the ideal composition for an optimum white body product vicinity of the plant. These are ideal for consideration according to the mentioned authors (SiO ¼ 72 wt%, 2 as waste disposal sites, as the ‘bedrock’ is clay-rich, re- Al O and total oxides ¼ 8 wt%), samples outside the 2 3 sulting in lowpermeability, and little or no upgrading white bodies field need processing in order to reduce may meet the requirements of the waste disposal au- their iron oxide and dilute their quartz contents. thorities. The loss on ignition of the chemical analysis Due to their high iron oxide content, these Ugandan includes not only water and carbon dioxide but also clays cannot be used for the production of fine ceramics. harmful species such as sulfur dioxide, chlorine, fluorine, From their chemical composition, they could be con- and nitrogen oxides, which escape as acid gases during sidered as rawmaterial for use in structural ceramic firing unless mineralogical reactions retain them in the brick (Heller-Kallai et al., 1988). This is caused by the decomposition of clay minerals, micas, organic matter, and also iron sulfides that occur in subordinate amounts. These volatiles have a harmful influence on the natural environment (Eckhardt et al., 1990). In an assessment of the potential to generate acid gases, it is important to assess the composition not only of the rawmaterials but also of the fired products, to ensure that their fate during firing is known.

6. Conclusions

Uganda offers considerable prospects for the exploitation of rawmaterials for the ceramic industry. The studied clay deposits are derived from gneissic and Fig. 11. Grain size classification of clay rawmaterials from Uganda in Winkler’s diagram (cf. Winkler, 1954). Fields indicate: (I) common granitoid rocks of the basement, and from metamor- bricks, (II) vertically perforated bricks, (III) roofing tiles and masonry phosed schists, as well as from amphibolites and basic bricks, and (IV) hollowproducts. rocks of the Buganda series. The boundaries between G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134 133 one clay type and another are commonly sharp, and it References can be concluded that the clay deposits are of lacustrine and fluvial origin, respectively. The differences in the Bain, A.J., 1987. Composition and properties of clays used in various nature of the clays are a function of progressive stages of fields of ceramics. Part II. Ceramic Forum International 63 (1–2), 44–48. leaching or transport activities, as indicated by a CM Bishop, W.W., 1969. Pleistocene stratigraphy in Uganda. Memo, diagram (Fig. 7). The clays mainly formed by weather- Geological Surveys and Mines Entebbe, Uganda, 10, pp. 1–128. ing of basement rocks during the tropical Pliocene and Coakley, J.P., Syvitski, J.P.M., 1991. SediGraph technique. In: accumulation of the detrital material in lakes and Syvitski, J.P.M. (Ed.), Principles, Methods and Application of swamps. Particle Size Analysis. Cambridge University Press, USA, pp. 129– 142. Overall, the clays studied here exhibit a wide range of Dalsgaard, K., Jensen, J.L., Sorensen, M., 1991. Methodology of chemical and mineralogical compositions, and grain size sieving small samples and calibration of sieve set. In: Syvitski, distributions. The clay rawmaterials fundamentally J.P.M. (Ed.), Principles, Methods and Application of Particle Size consist of kaolinite, illite, and quartz, with rare feldspars Analysis. Cambridge University Press, USA, pp. 64–75. and minor chlorite. The content of iron oxide is gener- Eckhardt, F.J., Roosch,€ H., Stein, V., 1990. Brick-clays from lower- Saksony (FRG) – technical and environmental problems. Sciences ally high. An important role seems to be played by the Geeologiques Meemoire (Strasbourg) 89, 15–24. minor mineralogical components, which influence the Fabbri, B., Fiori, C., 1985. Clays and complementary rawmaterials firing color. However, the characteristics of the samples for stoneware tiles. Mineralogica et Petrographica Acta 29A, 535– analyzed here are quite far from the requirements of 545. high-quality ceramics production. Even though all these Fabbri, B., Fiori, C., Krajewski, A., Valmori, R., Tenaglia, A., 1986. Comparison between traditional mineralogical and computerized materials are currently exploited in the production of rational analyses of ceramic rawmaterials. Journal de Physique various ceramic products, most of them are not com- Colloque Cl (Suppl.) 47 (2), 1–57. parable to commercially marketed European counter- Fiori, C., Fabbri, B., Donati, F., Venturi, I., 1989. Mineralogical parts (Fig. 10). The Ugandan rawmaterials need composition of the clay bodies used in the Italian tile industry. elaborate treatment to render them suitable for such use. Applied Clay Science 4, 461–473. Fischer, P., 1984. Some comments on the color of fired clays. This may be largely due to these clays being of sedi- Ziegelindustrie International 37 (9), 475–483. mentary origin, and due to their not having been Friedman, G.M., 1967. Dynamic processes and statistical parameters properly washed and sorted. A centralized clay mineral compared for size frequency distribution of beach and river sands. dressing and preparation plant might be the best way to Journal of Sedimentary Petrology 37, 327–354. overcome the rawmaterials problems. Geological Survey of Uganda, 1962. Geological map of Kampala. Sheet NA 36–14, scale 1:250,000. Nevertheless, the studied samples showsome inter- Harris, N., 1946. Report on the pottery clay deposits at Mukono esting features for application in the ceramic sector if (Ntawo). Unpublished Report, Geological Surveys and Mines well treated, especially considering the high iron oxide Entebbe, Uganda, 2 pp. content. However, due to the generally high quartz Heller-Kallai, L., Miloslavski, I., Aizenshtat, Z., Halicz, L., 1988. contents they may possess a refractory behavior. These Chemical and spectrometric analysis of volatiles derived from clays. American Mineralogist 73, 376–382. clays have chemical and mineralogical compositions Kagobya, A.L., 1950. A report on a visit to Mukono clay. Unpub- that indicate their usefulness for brick, ceramic, and lishied Report, Geological Survey and Mines Entebbe, Uganda, 6 earthenware production. Further systematic applied pp. testing of the clays has yet to be carried out to determine Kaliisa, K.F.A., 1983. Industrial mineral deposits of Uganda. their physical, mechanical, and technological properties. Unpublished Report, Geological Survey and Mines Entebbe, Uganda, pp. 35–50. Konta, J., 1995. Clay and man: clay rawmaterials in the service of Acknowledgements man. Applied Clay Science 10, 271–273. Kreimeyer, R., 1987. Some notes on the firing color of clay bricks. Applied Clay Science 2, 175–183. We thank W.U. Reimold (Univ. Witwatersrand, Jo- McGill, M., 1965. Clays in Uganda. Internal Report MG1, Geological hannesburg) for XRF analyses, S. Gier (Inst. of Pe- Surveys and Mines Entebbe, Uganda, 100 pp. trology, Vienna) for assistance with XRD and grain size McManus, J., 1988. Grain size distribution and interpretation. In: analysis, and M. Dondi (CNR-IRTEC, Faenza) for the Tucker, M.E. (Ed.), Techniques in Sedimentology. Blackwell, Oxford, pp. 63–85. normative mineralogical analysis software. The authors Nyakairu, A.G.W., Kaahwa, Y., 1998. Phase transitions in local clays. are grateful to the Austrian Academic Exchange Service American Ceramic Society Bulletin 77 (6), 76–78. (OOAD)€ for a Ph.D. stipend and partial financial assis- Passega, R., 1957. Texture as characteristic of clastic deposition. tance of fieldwork (to N.G.W.A.). Laboratory work was American Association Petrology and Geological Bulletin 41, 1952– supported by the Austrian FWF, project Y58-GEO (to 1984. Passega, R., 1964. Grain-size representation by CM patterns as a C.K.). We are grateful to D. Brandt and P. Eriksson for geologic tool. Journal of Sedimentary Petrology 34, 830–847. helpful and constructive comments, which led to im- Passega, R., Pyramjee, R., 1969. Grain-size image of clastic deposits. provements in the manuscript. Sedimentology 13, 233–252. 134 G.W.A. Nyakairu et al. / Journal of African Earth Sciences 35 (2002) 123–134

Reimold, W.U., Koeberl, C., Bishop, J., 1994. Roter Kamm impact Shepard, F.P., 1954. Nomenclature based on sand–silt–clay ratios. crater, Namibia: geochemistry of basement rocks and breccias. Journal of Sedimentary Petrology 24, 151–158. Geochimica et Cosmochimica Acta 58, 2689–2710. Tuhumwire, J., Gita, J., Biryabarema, M., 1995. Kajjansi and Kitiko Robbins, J., 1984. Ceramic whiteware an overview of raw material clay deposits. Journal of Uganda 1 (1), 34–44. supply. Industrial Minerals (204), 31–63. Winkler, H.G.F., 1954. Bedeutung der Korngroossenverteilung€ und des Schlueter, T., 1997. Geology of East Africa. Borntraager,€ Berlin, 484 pp. Mineralbestandes von Tonen fuur€ die Herstellung grobkeramischer Searle, A.B., Grimshaw, R.W., 1959. The Chemistry and Physics of Erzeugnisse. Berichte der Deutschen Keramischen Gesellschaft 31 Clays. Intersciene, NewYork, 942 pp. (10), 337–343.