CLAY MINERALOGY OF SEDIMENTS OF THE WESTERN NILE DELTA

Transcription

1 Clay Minerals (1975), 10, 369. CLAY MINERALOGY OF SEDIMENTS OF THE WESTERN NILE DELTA A. H. WEIR, E. C. ORMEROD AND I. M. I. EL MANSEY* Rothamsted Experimental Station, Harpenden, Herts., England and *Academy of Science and Technology, Cairo, Egypt (Received 10 August 1974; revised 10 December 1974) A B S T R A C T: Investigation of the clay mineralogy of forty-seven samples of sediments from boreholes in the western N lie Delta, an area little studied hitherto, and from surface sites on the mouth of the Nile and adjacent coast shows that the clay fractions consist of dominant iron-rich, dioctahedral, randomly interstratified smectite-illites together with kaolinite, illite and chlorite. Amounts of the constituent minerals of the clay fractions are estimated from their X-ray diffraction intensities, supported by selective dissolution chemical data, and a new method is used to estimate the proportion of expanding layers in randomly interstratified smectite-illite. The results, which confrm and extend the work of previous investigators, also show that there is little correlation between the clay mineral composition and texture of the sediments, only kaolinite being weakly linearly correlated with clay content. Transformation of 2: 1 layer silicate minerals occurs within the buried sediments; chlorite is transformed and smectite and illite interlayers redistributed within randomly interstratified smectite-illites. INTRODUCTION Studies of the clay mineralogy of the suspended load of the River Nile, its alluvial plain, delta and basin of accumulation in the eastern Mediterranean Sea were made by Elgabaly & Khadr (1962), Hamdi (1959, 1967), Hamdi & Iberg (1952), Hashad & Mady (1963), Gaith & Tanious (1964), Rateev, Emel'yanov & Kheirov (1966), Fayed (1970), Nir & Nathan (1972) and El-Attar & Jackson (1973). These studies showed that smectite is generally dominant in the sediments, with lesser amounts of kaolinite, mica and quartz. Hamdi (1967) showed that the coarser part of the clay contains abundant mica, and Fayed reported interstratified illite-montmorillonite from well samples near Cairo. El-Attar & Jackson reported chlorite and regularly interstratified layer silicates in addition to those already mentioned; they attempted to relate the clay minerals in the Delta and alluvial terrace sediments to those found in the soils of the drainage basins of the Blue and White Nile. The building of the Aswan Dam stopped the seasonal accumulation of silts in the Nile Delta, causing the lower Nile to flow silt-free and the Delta to be actively eroded along its Mediterranean coast. In the work reported here, which is part

2 370 A. H. Weir, E. C. Ormerod and L M. L E1 Mansey of a large beach protection project, we have studied the clay mineralogy of fortyseven samples of sediments taken from the river bed near the mouth of the West Rosetta branch of the Nile, off-shore samples from the Mediterranean coast near Abu Qir Bay, and samples from the Delta itself from seven boreholes sited between Abu Qir Bay and Lake Edkou. Most of the work was based on X-ray diffractometry of <2~m fractions, but four of the samples were analysed by selective chemical procedures similar to those used by El-Attar & Jackson, and four other samples were analysed in detail by subdividing their <2t~m clay fractions into coarse (2-0.2t~m), medium (0-2-0"04ttm) and fine (<0.04t~m) fractions. In addition to clay mineralogy, particle size determinations and measurements of ph, calcium carbonate content and organic matter, as organic carbon, were made on all samples for which sufficient material was available. Methods of analysis in this investigation were chosen to facilitate comparison between this and previously published work, and the detailed work on separated sub-fractions of clays was carried out to distinguish between smectites and smectite-rich randomly interstratified smectite-illites, minerals that frequently have all been referred to as smectites in the literature. SAMPLES Figure 1 shows the approximate location of the sampling sites. The Nile samples N1 to N4 were taken approximately 400, 600, 900 and 1400 metres from the mouth of the river, measured along the west bank. The samples were taken from the surface of the river bed; numbers following the colon, 20, 200, etc. represent the distance in metres from the west bank to the sampling points. Off-shore samples were also taken from the surface of the deposited sediment. Those from the east of the river are marked EB (El Bous); EB4 was taken from the river mouth and EB3, 2 and 1 from 900, 1300 and 1600 metres further to the east. The WR (West Rosetta) samples are also numbered from east to west and come from sites that are 1900 and 2900 metres from the river mouth. The EM (El Maidia) samples came from the centre of Abu Qir Bay at the mouth of E1 Maidia Inlet and the AQB (Abu Qir Bay) sample from off-shore at the western extremity of the bay. The borehole samples, numbered BH1 to BH7 come from the locations shown. Details of the boreholes, including particle size distributions for the many beds of different texture within each borehole, are to be published elsewhere. The samples brought to England for clay mineral analysis were selected to give a range of textures and calcium carbonate and organic matter contents. The depth of the sampling points in metres below the surface are shown as numbers following a colon, BHI:I, BH4:23, etc. METHODS Particle size determinations were done by a combination of sieving and pipette analysis. Calcium carbonate was determined by calcimeter (Bascomb, 1961) and organic carbon by the method of Kalembasa & Jenkinson (1973). Clay fractions, <2~m, were separated by the sedimentation of samples treated with sodium acetate,

3 Nile Delta clays 371 I I0 km i EB WF Abu Qir / Bay /.4 AQB ~,,7 c" \" EM~_r~E~ko~ ~'~ "--~-~ / ' FIG. 1. Sketch map of part of the Nile Delta showing approximate locations of the sampling sites. ph5, to dissolve calcium carbonate, and with H20~ to destroy organic matter, before dispersion in 0.15 % sodium hexametaphosphate solution. The separated clay fractions were saturated with magnesium or calcium ions from 0.1 N chloride solutions, washed salt-free and dried from acetone and petroleum ether. To subdivide the clay fractions, samples were redispersed in dilute sodium hexametaphosphate solution with ultrasonic agitation, and sedimen{ed by centrifugation for appropriate times and speeds. Oriented aggregates for X-ray diffractometry were made by drying dispersed gels of the clays in water on to glass slips. Treatments given to these speciments before diffractometry are described in the appropriate sections below. Selective dissolution analysis was carried out as described by Alexiades & Jackson (1966), except that kaolinite was determined by the method of Hashimoto & Jackson (1960). Total chemical analysis of four <0.04/~m fractions was done by X-ray fluorescence spectrometry using the method of Norrish & Hutton (1969). Sodium determinations on these samples were made by emission flame spectrometry of HF:H2SO4 digests of the samples. RESULTS Table 1 shows particle size distributions, calcium carbonate and organic carbon contents of all the samples for which clay minerals were determined. Particle size

4 372 A. H. Weir, E. C. Ormerod and L M. L E1 Mansey data was obtained at phi intervals, but has here been summarized in the three divisions of sand, silt and clay. Because the borehole samples represent incomplete sequences, the variations of textures with depth in the boreholes cannot be discussed in detail, but it is obvious that the range of textures is large and that the sediments vary from sands to clays to silty clays within a few metres of depth. The surface samples from the off-shore and Nile bed sites also have variable textures, the offshore samples tending to be sandier, and those from the Nile bed less sandy, than the borehole samples. All except EB4 contain calcium carbonate, but with the exception of BH4:13, 4:17 and 1:17 the amounts are small. Organic matter contents TABLE 1. Particle size distribution on a carbonate-free weight percent basis. Calcium carbonate and organic carbon on a whole soil weight percent basis Borehole samples Clay Silt Sand Calcium Organic ( < 2/~m) (2-63 p.m) ( p.m) carbonate carbon Depth (m) BHI 1 73"0 25"6 1"4 1 0"1 6 58" ' " "1 2 0" "0 23"6 51 "4 10 0'5 BH2 7 67" "8 2 13" "5 26"1 57"4 2 0" " 1 36" BH '9 44"3 3"8 2 1"2 12 9"6 7'4 83"0 1 0" '7 20"5 5"8 3 5" " ' BH4 2 35"5 35"7 28" "0 1 '6 2 1 ' "1 0"7 1 2" "3 22"0 55" '5 18"7 41" ' 8 4' BH5 2 59"3 34"2 6" "0 35"7 1 " "1 35' " " BH6 6 32'1 17"1 50"9 2 1"6 8 57'9 21 "2 22"7 1 4" '4 36"7 12" '8 36"5 2"7 4 3" '0 33'6 12" ' '4 BI-I7 4 54' "6 3 4"5 7 62'9 19' "4 14 2"1 2"1 95'8 1 3"1

5 Nile Delta clays 373 TABLE 1--continued Clay Silt Sand Calcium Organic ( < 2/~m) (2-63 t~m) ( t~m) carbonate carbon Off-shore samples AQB 46" "9 2 3"5 EB1 51 "9 44"3 3"8 6 1 "2 EB "7 91 "5 2 0"1 EB4 17'5 28"3 54" WR2 3" 4 12" 5 84"2 2 0' 1 WR3 5" "0 1 0"2 Nile bed samples N "3 41"1 15" "6 53"0 6"7 3 1" "6 N "9 43"8 35"3 2 0" "3 2" " "5 N "3 1 " ' "4 3 1 '4 N ' " ' "5 3 1"1 vary widely, with peat beds being The most peat-rich samples were analysis. cut by the borings at various levels in the delta. excluded because they yielded too little clay for Clay mineralogy Fine clay fractions (<0-04/~m). Samples of the fine clays of BHI:I, 3:18, 4:23 and N3:200, chosen because they are representative of the range of clay mineralogy in the Delta sediments, were examined by X-ray diffractometry of Na- and Casaturated forms, and chemically analysed in Ca and Mg forms. X-ray diffractometry showed that these samples contain interstratified expanding minerals, slightly different between samples, but substantially no other mineral phases such as illite, kaolinite, chlorite or hydroxy interlayers. The first three may be seen to be below the limits of detection in the traces illustrated in Fig. 2. Hydroxy interlayers were judged to be absent because Na-saturated specimens collapsed to 10A on heating to 60~ (Weir & Rayner, 1974). The chemical compositions of ignited Ca-saturated samples are given in Table 2. Analysis of Mg-saturated samples showed that they contain negligible amounts of non-exchangeable calcium ions; as other Ca-bearing minerals are absent it follows that the total CaO contents of the samples given in Table 2 may be taken as the exchange capacities of the minerals. Table 2 also shows the amounts of silica, alumina and iron oxide removed from the specimens by the extraction procedure of Hashimoto & Jackson (1960). These, mainly amorphous, oxides and the phosphate

6 (o) (b) , J, (c) I (d) I I0 o 28 Cu Ka FIG. 2. X-ray diffractograms of Ca-saturated,.ethylene glycol solvated oriented aggregate specimens of <0"04/~m clay fractions: (a) BH 1:1, (b) BH 3:18, (c) BH 4:23, (d) N 3:200.

7 Nile Delta clays TABL~ 2. Composition of Ca-saturated fine clay (< 0-04 tzm) fractions, weight percent of ignited samples 375 BH1 : 1 BH3 : 18 BH4: 23 N3 : Si TiOz A " "70 1" '59 20"49 Fe2Oa 17"56 4"10 15"31 2" " MnO ' "05 MgO 3"57 2"88 3" CaO 2"63 3' K20 1 " Na " P2Os 0"50 0"36 0' Total 99" Elemental composition of samples, as oxides. 2 Free oxides dissolved from samples by the alkali extraction procedure of Hashimoto & Jackson. associated with them, taken up from the dispersing agent, are excluded from the chemical composition used to calculate the unit formulae because they are considered to be separate from the 2:1 minerals. The manganese and titanium oxides are also excluded. Ferrous oxide amounts to between 0.4 and 0.5% of the samples, but as all these fine clay fractions contain organic matter some and probably all of this ferrous iron was produced by reduction of ferric iron during analysis. Unit formulae for the four minerals based on 22 negative charges are as follows: BH 1 : 1 C%.20(Ko.15Nao.02)(Alo.95Feo.72Mgo.38)(Si3-65A10-35) ~ atoms BH3:18 Cao.23(Ko.04Nao.o1)(Al i.08feo.65mgo.29)(si3.71alo.29)o a atoms BH4:23 Cao.23(Ko.04Nao.o1)(All.02Feo.69Mgo.32XSi3.68Alo.32)O atoms ' N3:200 Cao.24(Ko.o6Nao.ol)(All.o7Feo.66Mgo.32)(Si3.59Alo.41)01~ 2.05 atoms The minerals are all iron-rich, dioctahedral 2:1 layer silicates with slightly more tetrahedral than octahedral substitution producing the permanent negative charge. The interlayer charge varies between 0.57 and 0.51/Si4Oll, of which non-exchangeable K and Na ions account for between 0.17 and 0-05/Si401~. If an interlayer charge of 0"7/Si401~ is assumed for the non-expanding interlayers (Weir & Rayner, 1974), the following compositions can be deduced from the unit formulae: BH1 : 1, 76 : 24 smectite : illite, N3 : 200, 90 : 10 smectite: illite, BH3 : 18 and 4 : 23, 93 : 7 smectite : illite.

8 376 A. H. Weir, E. C. Ormerod and I. M. L El Mansey X-ray diffractometer traces of Ca-saturated ethylene glycol solvated oriented aggregates are shown in Fig. 2. They represent randomly interstratified smectite-illites varying from nearly pure smectite in BH3:18 and 4:23 through N3:200 to BH1 : 1, which contains sufficient illite layers to give it a sequence of basal reflections that is clearly irrational. The traces were matched against a set of computer simulated traces from models composed of varying proportions of 16.9A smectite layers and 10A illite layers in particles containing from three to seven interlayers each (Weir & Rayner, 1974). The best fit, based on peak positions, was obtained for BH3:18 with 95% expanding layers, 4:23 with 90~ N3:200 with 85% and BHI:I with 65%. The fit of the traces was not perfect, in particular the reflections from the samples were broader than those from the models, even for such small diffracting units. The proportions of smectite layers given by their diffraction peak positions average 4~176 less than those deduced from the unit formulae, which, considering the assumptions involved, may be regarded as showing good agreement. An alternative method of estimating the proportion of expanding layers in the minerals is to measure a function of the shape of the first reflection. As the proportion of expanding layers increases, this reflection increases in intensity while central scatter due to interstratification decreases. Thus, the ratio a/b (Fig. 4) changes and a plot of a/b or log a/b versus ~ expanding layers gives a sensitive measure of the proportion of expanding layers in the mineral. The values of a/b from the computer simulated traces used for calibration in this method are very sensitive to changes in the models used, both for the minerals and for the instrumental conditions. The results must therefore be regarded as relative within a group of similar samples made under identical experimental conditions. Measurements of log a/b ratios for the fine clays give the following proportions of expanding layers: BH3:18, 87 %; 4:23, 86 ~ N3:200; 79 O/o and BHI:I 64 %. These are within 10 %> of those obtained from matching peak positions, but are lower, probably because the fine clay fractions have smaller diffracting units than the model and, consequently, less well resolved first reflections. The application of this method to the medium clays should be satisfactory as the diffracting units in model and sample should be of comparable size, whereas the larger units in the coarse clays may cause the proportions of expanding layers in this fraction to be slightly over-estimated. Satisfactory estimates of proportions of expanding layers should also be obtained from the first reflections of the whole clays as they represent composites of the first reflections of the separated clay fractions. Medium ( ttm) and coarse (0-2-2/~m) clay [ractions. Diffractometer traces of medium and coarse clay fractions of BHI:I, BH3:18, BH4:23 and N3:200 together with those of their <2/~m fractions are illustrated in Fig. 3. The medium fractions contain expanding minerals, illite and kaolinite. Estimation of kaolinite by selective dissolution analysis gave values of 14, 20, 16 and 18% respectively. From the relative intensities of the 10 and 7-1 A reflections (this method is discussed in the next section) the illite contents of these fractions are estimated as 7, 3, 3 and 5% respectively. By difference, expanding minerals form ~ of the fractions. In the coarse clays kaolinite and, particularly, illite are more abundant. The 14 A

9 _o o V v / 2~ E I..... ~ v t~ re) o ~~ -3 V~ 6 t v 0d v -3 o o ~ v~ 6.~..~.."22.."22. v M ~

10 378 A. H. Weir, E. C. Ormerod and I. M. I. El Mansey JO D 7O w 6O 5O I III b I I I I I [ I I i I I I'0 Log a/13 FIG. 4. First reflection of a computer simulated trace from a randomly interstratified mixture of 16-9 and 10 A layers containing 3-7 interlayers per diffraction packet and showing a and b, the parameters used to measure the resolution of the reflection. The graph is a plot of logloa/b against per cent expanding layers (the proportion of 16'9 A layers used to compute the trace). reflection of chlorite is visible in the trace of N3:200 and the 3-2A reflection of feldspar in those of BH3:18 and 4:23. Measurements of a~ b ratios of first reflections from all the size fractions (Table 3) show that the fine clay fractions contain smectite-illites with the greatest proportions of expanding layers. Those in the <2 t~m fractions are similar to those in the coarse and medium fractions, which in these samples form the bulk by weight of the <2 txm clays. Figure 3 shows the advantages of using the a/b ratio of the first reflection rather than the peak positions of the sequence of basal reflections for estimating proportions of expanding layers. In the illustrated traces it is difficult to identify the second and subsequent basal reflections of the interstratified minerals even for the well-oriented medium clay fractions. For the coarse and whole clay fractions it is impossible, partly because reflections from kaolinite and illite interfere, but mainly because of the low intensity of the reflections themselves. Thus, the best hope of identifying randomly interstratified expanding minerals in diffractograms

11 Nile Delta clays TABLE 3. Percentages of expanding layers of randomly interstratified smectite-illites, obtained from Fig. 4, in fine, medium, coarse and whole clay fractions 379 Fine Medium Coarse Whole clay <0-04/~m 0"044)'2t~m 0.2-2tLm <2t~m BHI:I BH3: BH4: N3: of mixtures of clay minerals lies in utilizing the information given by the first reflection. In addition to the Ca-saturated, ethylene glycol solvated samples described above, Mg-saturated glycerol solvated samples were examined by X-ray diffractometry. These gave first reflections of randomly interstratified smectite-illites very similar to those described above, but, in addition, a small 14A reflection in all coarse and medium fractions except those of BH3:18. This indicates a vermiculitic phase, but as only the single reflection, distorted by the proximity of the intense 18A reflection, was observable, it is unclear whether the samples contain separate vermiculite or a second interstratified phase containing vermiculite layers. These results are referred to again in the discussion section below. Whole clay fractions (<2 tkm). Clay minerals were identified by X-ray diffractometry and their relative proportions in the samples estimated from their diffraction intensities by the method outlined below. Four samples were further analysed by selective dissolution analysis and the results compared with those from X-ray diffractometry. Selective dissolution analysis. Results obtained on BHI:I, BH4:23, BH6:21 and BH7:7 are given in Table 4. The methods used were those of Alexiades & Jackson (1966), except for the determination of kaolinite for which the original TABLE 4o Mineralogical composition of < 2/~m fractions from selective dissolution analysis (SDA) BHI:I BH4:23 BH6:21 BH7:7 Amorphous hydrated SiO2 q- A120~ Free iron as FeO(OH) 2"1 2-0 l "7 1 "4 Kaolinite Illite Smectite Vermiculite 8 12 t I 9 Chlorite Quartz 1-6 2" 1 2"0 2"8 Feldspar 2"2 1'9 2"2 2"

12 380 A. H. Weir, E. C. Ormerod and L M. L El Mansey method of Hashimoto & Jackson (1960) was preferred. It was found that the sequential extraction procedure of Alexiades & Jackson (1966) dissolved iron from the 2:1 silicate minerals. Attempts to correct for this either by the method advocated by Briner & Jackson (1970), or by substituting factors obtained from the actual composition of the 2:1 minerals given in Table 2, were unsuccessful because the boiling alkali used did not extract stoichiometric amounts of silicon, aluminium and iron from these minerals. We found, however, that the parallel extraction procedure of Hashimoto & Jackson (1960) dissolved much less iron than sequential extraction, and estimates of kaolinite obtained by this method were therefore preferred. Table 4 shows that the samples contain from 47 to 64 % of expanding minerals that are divided in the ratio smectite:vermiculite equal to 5:1; in addition the samples contain 7-17 % illite, % kaolinite and 0-2 ~ chlorite. Free oxides, 6-7 ~ and 4-5 % of quartz + feldspar are also present. X-ray difjractometry. The relative proportions of the layer silicate minerals in BHI:I, BH4:23, BH6:21 and BH7:7 were also estimated from their X-ray diffraction intensities. The method employed, based on an unpublished method for oriented aggregate specimens by G. Brown, uses the 7.1 A kaolinite reflection as an internal standard. Diffractometer traces are recorded from replicate Ca or Mg-saturated specimens, dried in air at 20~ solvated with ethylene glycol, and heated to 300~ to collapse expanding minerals to 10A. From the resulting three traces the reflection intensities of illite, I~ (10 A reflection, air-dry and glycol specimens), kaolinite, IK (7"1 A reflection from all three traces), illite + expanding minerals I~+Ex (10 A reflection, heated specimen) and chlorite, Ich (14 A reflection, heated specimen) are measured. The intensities are adjusted by simple proportion to equalise that of kaolinite from the three traces. The proportions by weight of the mineral phases are then calculated using the following simplifying assumptions: on a weight for weight basis (1) expanding minerals collapsed to 10 A give the same reflection intensity as illite, (2) chlorite gives twice the intensity of illite, and (3) kaolinite gives 2-5 times the intensity of illite. From normalized intensities I~ gives the proportion of illite, I~+zx--I~ that of the expanding minerals, IK/2"5 that of kaolinite and Ich/2 that of chlorite (there is not sufficient chlorite in the Nile Delta samples for its 7 A reflection to interfere with the 7.1 A reflection of kaolinite). The calculations are rapidly made and results are reproducible to within 20% of the amount of the phase considered. Table 5 compares the results obtained by the two methods of analysis described above. The data in Table 4 has been reduced to figures for kaolinite, illite, expanding minerals and chlorite only, recalculated to 100 %. Comparison of the results from selective dissolution analysis (SDA) and X-ray diffractometry (XRD) shows that kaolinite by SDA averages 21.8 % for the four samples, compared with 17-5 ~ by XRD, a difference of 25 % of the amount estimated by XRD. This difference is almost balanced by the difference in chlorite content as estimated by the two methods. In the SDA method an over-estimation of kaolinite leads to an under-estimate of chlorite, which suggests that despite the substitution of the parallel extraction for the sequential extraction procedure, kaolinite may still be

13 Nile Delta clays TABLE 5. Comparison of estimates of the mineralogical composition of < 2 t~m fractions by selective dissolution analysis (SDA) and X-ray diffractometry (XRD) 381 BH1 : 1 BH4: 23 BH6: 21 BH7: 7 Kaolinite SDA XRD Illite SDA XRD Expanding minerals SDA XRD Chlorite SDA XRD slightly over-estimated by SDA as a result of alkali attack on the iron-rich 2:1 minerals in these samples. The estimations of illite and expanding minerals may properly be considered together when comparing the two methods because K-containing illitic interlayers are grouped with illite by SDA, but with expanding minerals by XRD--they contribute to the 10 A reflection of the heated specimen. Table 5 shows that illite + expanding minerals average 77 ~ by SDA and 79 ~ by XRD, a result that suggests that there is no systematic difference between the methods. In summary, SDA gives more information than XRD, e.g. about amorphous oxides, and different information about expanding minerals--the subdivision into vermiculite and smectite is in effect information about the K-fixing exchange properties of expanding interlayers of 2:1 minerals that may or may not be interstratified--but overall the amounts found for the main phases are comparable. We found that the full SDA analysis was very time-consuming, man-hours per sample, compared with 2 hr per sample for the XRD method, so the clay mineral compositions of the-complete series of Nile Delta samples given in Table 6 were obtained by XRD. The average composition of the forty-seven Nile Delta samples listed in Table 6 is: kaolinite 15 ~ illite 8 ~/o, randomly interstratified smectite-illite (containing 69 ~/o expanding layers) 72 ~ and chlorite 5 yo. Correlation of clay mineral contents of the clay fractions (Table 6) with sand, silt and clay contents of the whole sediments (Table 1) shows that kaolinite is weakly positively linearly correlated (r = 0.55) with clay content and negatively correlated with sand content (r =- 0.49), whereas the other clay minerals show no correlation with texture (r = 0.16 or less). However, Table 6 also shows that kaolinite is uniformly distributed throughout all samples, and it can be deduced that the linear variation with clay content only spans approximately one quarter of the mean kaolinite content of the clay fractions. A possible explanation of what is, therefore, a very small effect, is that kaolinite in clay-rich sediments tends to be more effectively buffered against dissolution by circulating water than that in sandy sediments. If the borehole, off-shore and Nile bed samples are considered as separate groups, the means and standard deviations of the amounts of their constituent minerals may be used to show similarities or differences between these types of deposit. For

14 382 A. H. Weir, E. C. Ormerod and I. M. I. El Mansey TA~LE 6. Clay mineralogy of < 2 /zm fractions Interstratified Kaolinite Illite Chlorite smectite-illite expanding layers in smectite-illite Borehole samples BH BH BH BH BH BH BH Mean and s.d "6 3" "3 6"3 Off-shore samples AQB EM EBI EB EB WR WR Mean and s.d " " " "

15 Nile Delta clays 383 TABLE 6--continued expanding Interstratified layers in Kaolinite Illite Chlorite smectite-illite smectite-illite Nile bed samples NI N N N Mean and s.d instance, as stated above, kaolinite is uniformly distributed throughout all the samples, whereas the subsurface borehole samples contain more smectite-illite, but less illite or chlorite than the surface off-shore or Nile bed samples. The last column in Table 6 shows that the percentage of expanding layers in the smectiteillites have similar means for the three groups of samples, but their standard deviations indicate that there is markedly more variation in composition amongst the borehole samples. In detail, the smectite illites from borehole 5 are very uniform in composition whereas those from 1 or 4 vary greatly. There does not seem to be any direct relationship, however, between the composition of the smectite-illites and their depth of burial. CONCLUSIONS Our results, like those of Elgabaly & Khadr (1962) and others, show that smectite-like expanding minerals are dominant in the clay fractions of Nile Delta sediments. The fractionation of samples confirmed Hamdi's findings that illite is more abundant in the coarse clay than in other fractions (Hamdi, 1967), although in general we find it less abundant in the Delta sediments than in the surface Nile bed and off-shore samples. This work also confirms and extends Fayed's findings (Fayed, 1970) that the Delta sediments contain interstratified expanding minerals. Fayed distinguished smectites with rational sequences of basal reflections from interstratified minerals with irrational sequences. As sequences are only clearly irrational from minerals with, say, less than 70 ~ expanding layers, Fayed's interstratified minerals would correspond to approximately half those described here. This work shows that interstratified minerals with less than 70 % expanding layers are common in the upper

16 384 A. H. Weir, E. C. Ormerod and I. M. L El Mansey 80._o g- E c_ v 70 7 / 9 / / 9 //o O// / f // 9 / J J X / / J 10//'0 9 / 9 J // 9 9 o o 0 0 J / J / 9 9 / /./ / /" / X X 60 I 1 I I l I % Exponding layers in smectite-illite FIG. 5. Plot of percentage expanding layers in interstratified smectite-illite against percentage smectiteillite in < 2 t~m clay fractions: dots~borehole samples, crosses--off-shore samples, circles--nile bed samples. Details of the plot of the linear regression equation (dashed line) fitted by least squares method to the borehole data are as follows: r = 0'85, m = 0.72, b = 24.76, cr x = 6"26,.~ = 68"23, % = 5"30, ~r = 74"19, variance accounted for = 72 ~. 25 metres of the Delta sediments, as well as at depths greater than 150 metres near Cairo. Results on the clay mineralogy of soils from the Delta obtained by EI-Attar & Jackson (1973) appear at first sight to differ markedly from those reported here. They reported some vermiculite, more chlorite, but less kaolinite in all samples. Vermiculite was reported by us in Table 4 (using the same techniques as El-Attar & Jackson) as constituting 1/6 of the total expanding minerals of the four samples analysed--approximately the same fraction as shown in the soils by El-Attar & Jackson in their Fig. 5. For the main analysis of clay fractions in our Table 6 we interpreted the diffractometer traces of Ca-saturated glycol solvated samples in terms of randomly interstratified smectite-illites. Had we used Mg-saturated glycerol solvated specimens instead, many of the specimens would have shown 14 A reflections from one-layer glycerol complexes formed in interlayers that coincide, at least in part, with those measured as fixing K ions in the exchange capacity measurements. The differences observed in the reporting of vermiculite may thus be attributed to the different techniques used. The relationship between traces given by the same

17 Nile Delta clays 385 clay when either Ca-saturated and ethylene glycol solvated or Mg-saturated and glycerol solvated is not well understood. The first frequently seems to be consistent with a simple two-component interstratification, the second is more complicated if it has a separate 14.& phase. If further work shows it can be interpreted in terms of simple models it may give more information about expanding interlayers in interstratified minerals than glycol solvation. The explanation of the discrepancy in the measurements of kaolinite and chlorite in the two pieces of work is more complicated. When we attempted to correct our kaolinite figures by the method of Briner & Jackson (1970) we grossly over-corrected, as we did when using instead the measured composition of the fine clay of BHI:I. It appeared that iron, silicon and aluminium were not dissolved stoichiometrically from the 2:1 minerals by the alkali extraction, so that attempts to correct on the basis of the amount of iron dissolved gave low values for kaolinite. We therefore substituted a less drastic extraction procedure for estimating kaolinite. It seems possible, therefore, that EI-Attar & Jackson (1973) under-estimated kaolinite in their samples and, because the two methods are interrelated, over-estimated chlorite. However, this is probably not the whole explanation of the discrepancy between the two sets of results, because chlorite (Table 6) was estimated from the intensity of the 14 A reflection given by a specimen heated to 300~ and confirmed from specimens heated to 550~ whereas El-Attar & Jackson (1973) stated that little of their chlorite was in this form. Additionally, the low-angle reflections from ordered and disordered phases with hindered collapse shown by the K-saturated soil clays were not observed in any of our traces, neither those from the Ca-saturated <2/,m clays nor the Na-saturated fine clay fractions. It therefore appears that pedogenic hydroxy interlayers may be a feature of the near-surface soil clays, but absent at depth in the Delta and from the subaqueous off-shore and Nile bed samples. A striking feature of the results in Table 6 is the increase in the mean value of smectite-illite and decrease of illite and chlorite in passing from the surface offshore and Nile bed samples to the buried Delta deposits. The near uniformity of the kaolinite contents of the three types of deposit suggest that this is a real effect. By contrast, the percentage of expanding layers in the smectite-illites show no similar variation. When individual values of percentage expanding layers in the smectite-illites arc plotted against percentage smectite-illite in the <2 tlm clay fractions (Fig. 4), there is a strong positive linear correlation (correlation coefficient r = 0.85) between the two parameters for the borehole samples, but no correlation for the surface samples. Such a result should be treated with caution because the two parameters involved were obtaiv.ed by measuring related diffraction reflections produced by different preliminary treatments. However, the fact that the surface samples show no correlation using the same method of measurement suggests that for the borehole samples there is a real tendency for the expanding minerals to be more smectite-like where they are more abundant. A possible genetic explanation for this is that transformation (Millot, 1970) is active within the depth of the zone sampled. If it is, it is not simply related to time or the depth of burial

18 386 A. H. Weir, E. C. Ormerod and L M. L El Mansey or permeability of the sediments as indicated by the simple physical parameters given in Table 1, because there are no or only very weak linear correlations between the clay mineral parameters and depth of sample, or clay, calcium carbonate or organic carbon content of the sediments. If the surface samples from the Nile bed and off-shore sites represent the range of original clay composition in the samples now buried, the changes required to produce the assemblages observed would include the redistribution of K ions, presumably in percolating pore solutions within the sidements, to produce interstratified minerals with more collapsed layers as well as a loss of interlayer cations from illite and/or chlorite to produce a greater proportion of more smectite-rich expanding minerals. Although such an assumption about the nature of the original clay composition is highly speculative, the existence of the linear correlation between the proportions of expanding layers and totals of expanding minerals does suggest that some process of transformation, such as that outlined above, is active in the clay fractions of the Delta sediments. ACKNOWLEDGMENTS We thank the Director of the Nile Beach Protection Project for permission to publish this work, and F. Cowland and E. M. Thomson for preparing the figures. El Mansey was supported by a UNESCO Fellowship during his stay at Rothamsted. REFERENCES ALEXIADF~ C.A. & JACKSON M.L. (1966) Clays Clay Miner. 14, 35. BASCOMB C.L. (1961) Chem. Ind BRINER G.P & JACKSON M.L. (1970) IsraelJ. Chem. 8, 487. EL-ATTAR H.A. & JACKSON M.L. (1973) Soil Sci. il6, 191. ELGABALY M.M. & KHADR M. (1962) J. Soil Sci. 13, 333. FAYED L.A. (1970) Int. J. Rock Mech. Mill, Sci. 7, 249. GAITH A.M. & TANIOUS M. (1964) Trans. 8th Int. Soc. Soil ScL (Bucharest), $, 535. HAMDI H. (1959) Soils and Fertilisers, 22, HAMDI H. (1967) Egypt J. Soil ScL of the U.A.R. 7, 15. HAMDI H. & IBF.RG R. (1952) Z. PflErnahr. Dung. Bodenk. 67, 112. HASHAD M.N. & MADY F. (1963) Soils and Fertilisers, 28, 100. HASHIMOTO I. & JACKSON M.L. (1960) Clays and Clay Minerals, 7th Conj'., p Pergamon Press, Oxford. KALEMBASA S.J. & JFr~'KINSON D.S. (1973)J. Sci. FdAgric. 24, MmLOT G. (1970) Geology o/clays, Chap. II, p. 33. Chapman and Hall, London. NIR Y. & NATHAN, Y. (1972) Bull. Groupefranc. Argiles, 24, 187. NORRISH K. & IquTroN J.T. (1969) Geochim. Cosmochim. Acta, 33, 431. RATEEV M.A., EMEL'VANOV E.M. & KHEmOV M.B. (1966)Litologie i Paleoznye Iskopamie, 4, 6. WEIR A.H. & RAVNER J.H. (1974) ('lay Miner. 10, 173.