CHEMICAL REACTIVITY OF SOME RESERVOIR ILLITES: IMPLICATIONS FOR PETROLEUM PRODUCTION

Transcription

1 Clay Minerals (1989) 24, CHEMICAL REACTIVITY OF SOME RESERVOIR ILLITES: IMPLICATIONS FOR PETROLEUM PRODUCTION C. R. HUGHES, R. C. DAVEY AND C. D. CURTIS* Department of Geology, University of SheJ~eld, Beaumont Building, Brookhill, Sheffield $3 7HF, and *Department of Geology, University of Manchester, Oxford Road, Manchester M13 9PL (Received 18 May 1988; revised 3 February 1989) A B S T RA C T : Authigenic illite samples have been isolated from preserved reservoir core and characterized by XRD, XRF, SEM and ATEM. Reactivity towards aqueous solutions containing acids, alkalis, complexing agents and reducing agents has been evaluated using both static (flasks) and dynamic (flow-cell) experiments at 80~ and atmospheric pressure. In flask experiments with dilute reagents, reaction appears to be simple stoichiometric dissolution. The same pattern extends to higher concentrations (molar), except for alkalis where Si is preferentially leached leaving an Al-rich residual phase. Under the more vigorous conditions of flow (continuous leaching by fresh reactant solutions), stronger acid and alkaline solutions both cause substantial dissolution although by different mechanisms. Acids displace K and cause illite to swell. In many of these experiments, gels were seen to precipitate downstream from the reaction site. These reactions are rapid, taking place in hours at 80~ Similar reactions can be expected to take place in consequence of a range of drilling, stimulation and production procedures. Clay dissolution, modification and, possibly, migration would be anticipated close to the well-bore and damage from scaling (gels) somewhat further away. The pores of many sandstone reservoirs are lined with authigenic illite. The clay characteristically exhibits a delicate fibrous or platy morphology with a very large surface area in contact with resident pore-fluids. Reservoir productivity can be so seriously impaired by the presence of illite that economic viability may be negated. As no natural illite dissolution textures have been reported it would appear that, once formed, illites remain thermodynamically stable throughout the range of diagenetic environments of interest to the petroleum geologist. Artificial fluids introduced during drilling and production operations, however, are very different from natural porewaters and the possibility of reaction, with further reservoir deterioration, must be considered. As a first approach, the nature of chemical rections between authigenic illite and some of the more reactive components of aqueous drilling fluid filtrates, cement filtrates and alkaline or acid flushes has been investigated in experiments at atmospheric pressure and temperatures up to 80~ The principal objectives were to: (1) Isolate and characterize pure authigenic illite samples from preserved reservoir core. (2) React these illite samples with a range of chemical reagents under controlled conditions. (3) Identify conditions where significant reaction occurs. (4) Determine the nature of the reaction The Mineralogical Society

2 446 C. R. Hughes et al. Batch and flow experiments were designed to simulate differing rection conditions: flow experiments use high solution/solid ratios and maximize solid alteration (solution compositions are modified only slightly). Batch experiments, with high solid/solution ratios, maximize fluid compositional changes. Analysis of solids from the former and solutions from the latter should provide a better description of reaction. These ambient-pressure experiments are performed relatively readily, their reproducibility is easily tested and many different reaction configurations (minerals, solutes, temperatures) can be investigated. This contrasts with high-pressure experimentation where few experiments can be run. EXPERIMENTAL PROCEDURES Separation of clays from preserved core Commercially available illite standards were not used because most are based on mudrocks. They bear little similarity to the authigenic illites found in sandstones. Three sections of preserved illitic sandstone were chosen on the basis of an initial X-ray diffraction (XRD) survey from material kindly provided by British Petroleum Development Ltd. The cores were unsealed and immediately submerged in distilled water. Any drilling-mud cake was washed off and the core examined for drilling fluid infiltration. Invasion rims of drilling mud were chipped away. Each clean core sample was crushed gently by hand under distilled water and the clay fraction decanted off. It was found that continued agitation of the disaggregated framework grains in distilled water removed most grain-coating clays by abrasion, this procedure resulting in only minimal breakdown of framework grains. The suspended clays were washed and further diluted until there were no signs of flocculation. Each clay suspension was then separated into < 2/zm, 2-8 #m and > 8 #m size fractions (equivalent spherical diameters) by gravitational settling. The finer size fractions were then concentrated by centrifuge, freeze-dried and homogenized. Characterization Clay samples were analysed by XRD and X-ray fluorescence (XRF) to determine overall mineralogy and bulk chemistry. In addition, each core was examined by standard thinsection petrography and scanning electron microscopy (SEM). It was anticipated that the separated illite fractions would be contaminated by other authigenic phases and minor framework-grain debris. In order to obtain compositional data for the illitic component alone, analytical transmission electron microscopy (ATEM) was used. Previous work (Curtis et al., 1985) has demonstrated that this is perhaps the only reliable technique for quantitative data since single crystals are analysed and other phases avoided. Two different methods for preparing electron transparent samples for ATEM were used. Ultra-thin rock sections were prepared following the ion-beam thinning technique of Phakey et al. (1972). Dispersed mounts of the clay separates were prepared on formvar-coated TEM grids. Chemical data were couected using a Philips 400T TEM/STEM system with EDAX energy-dispersive X-ray analysis. X-ray emission spectra were processed according to the ratio method of Cliff & Lorimer (1975). This assumes that below a certain critical thickness,

3 Chemical reactivity of illites and implications for petroleum production 447 sample matrix effects (primarily X-ray absorption and fluorescence) are sufficiently close to zero that characteristic X-ray intensities are effectively directly proportional to element concentrations. The ATEM system at Sheffield has been calibrated for the analysis of phyllosilcates by determining proportionality constants relative to silicon from sections of independently analysed macroscopic micas. Following an investigation into potential sources of analytical error in the Sheffield system (Hughes et al., 1989; Hughes, 1987), great care was taken to work within optimum operating conditions, avoiding especially any specimen disruption and X-ray matrix effects. Element ratios were converted to standard 24 anion phyllosilicate formulae for subsequent comparison. 9 Clay~solute reaction batch-dissolution experiments The < 2 and 2-8/an fractions from each core were then reacted in conventional batchleaching experiments with the following aqueous reagents: Distilled water 0.01 M HC! (hydrochloric acid) 0"1 M HCI,, 1.0 M HC M NaOH (sodium hydroxide) 0.1 M NaOH,, 1-0 M NaOH 0.1 M C6HsOT.H20 (citric acid) 1:0 M C6HsOT.H20,, 0.01 M NH2OH.HC1 (hydroxylamine hydrochloride) 0.1 M NH2OH.HCI,, 1.0 M NH2OH. HCI,, These reagents were chosen to evaluate the effects of acids, alkalis, complexing agents and reducing agents. Hydrofluoric acid experiments were not undertaken owing to limitations of equipment design. That HF reacts strongly with clays, of course, is not in doubt. 100 mg of each clay sample (<2/an and 2-8 /an fractions) were weighed into flasks containing precisely 10 g of each solution. The stoppered flasks were agitated (120 strokes per min) for 48 h at 80~ The suspensions were then filtered hot into previously weighed flasks containing precisely 10 g of I M HNO3 to prevent solute precipitation (Whatman 542 paper-- fine precipitate retention). A set of clay-free blanks was also run. Once cooled, the HNO3- diluted product solutions were analysed by inductively-coupled plasma spectrometry (ICP) and the concentrations of Si, AI, Fe, Mg, Ca and K in the original leachates calculated. Clay/solute reaction flow-cell experiments In flow-cell experiments, a small sample of illite is washed continuously by fresh reactant solution. Since the products are continually removed, conditions of reaction are particularly aggressive. Approximately 5 mg samples were prepared as thin films sandwiched between two 30 mm diameter filter papers. Within the flow cell, the sample is supported on a silica sinter disc (Fig. 1). Five sets of flow-cell experiments were carried out, each involving reaction of the 2-8/an illitic separates II and I2 with aqueous reagents as follows:

4 448 C. R. Hughes et al. Reactant Solution reservoir Peristaltic Pump ~ Flow cell body \ O-ring Product Solution Clay sample - sandwiched between two filter papers Silica sinter disc FI6. 1. Diagram of a single flow cell. Distilled H20 for 48 h at a flow rate of 0.5 ml/min at 80~ 0.01 M NaOH for 48 h at a flow rate of 0.5 ml/min at 80~ 0.1 ~l NaOH for 48 h at a flow rate of 0.5 ml/min at 80~ 0-01 M HCI for 48 h at a flow rate of 0.5 ml/min at 80~ 0.1 M HCl for 48 h at a flow rate of 0.5 ml/min at 80~ After reaction, the clays were separated and examined by XRD as thin smears on glass slides. The flow-cell experiments were limited to testing the effects of prolonged reaction of the illites with 'fresh' water, acid and alkali solutions. The experiments were much more time- ~onsuming and could not be run in large batches. They did, however, allow the effects of vigorous reaction on the clays to be evaluated. EXPERIMENTAL RESULTS AND DISCUSSION XRD, XRF and A TEM characterization of samples Table 1 summarizes mineralogical information for samples I1, I2 and I3 as revealed by XRD. The <2 #m fractions of I1 and I3 were almost pure illite (traces of kaolinite and quartz), whilst I2 proved to be a mixture of an interstratified illite/smectite and kaolinite together with traces of feldspar. For each sample, the < 2 #m and 2-8 #m fractions were similar in mineralogical composition, although the latter contained more framework-grain

5 E E Chemical reactivity of illites and implications for petroleum production 449 TABLE 1. Mineralogyofseparatedclay fraction from selected illitic reservoir sandstones as determined by XRD. Specimen no. Major Minor Trace I1. illite (<2/~m) I1. illite (2-8/an) I2. illite/smectite ( < 2/Jm) kaolinite I2. iuite/smectite (2-8/~m) kaolinite I3. illite (<2/~m) I3. illite (2-8/am) -- quartz kaolinite quartz kaolinite K-feldspar K-feldspar kaolinite quartz kaolinite quartz TABLE 2. ATEM mean data for ion-thinned (thin) and dispersed (disp.) preparations of illitic samples. Recalculated to mica formulae on a water-free basis (assuming 20 oxygens and 4 hydroxyls, and all iron to be ferric). S.D. denotes standard deviation assuming the population parameter n -1. It (thin) S.D. I1 (disp.) S.D. I2 (thin) S.D. I2 (disp.) S.D. I3 (disp.) S.D. Si Ai Al Ti Fe(III) Mg Ca Na K Octahedral Interlayer OH n debris. In all three samples the illitic material was originally present as grain coatings with both fibrous and platy morphologies. The mean composition of each illitic phase analysed by ATEM is presented in Table 2. The data for I1 and I3 recalculate to near-ideal dioctahedral mica formulae which are similar to other authigenic illites on the Sheffield database. IUite I2, with its lower interlayer occupancy, appears to be a mixed-layer iuite/smectite, consistent with the XRD trace. In general, the chemical data from ion-thinned preparations are very similar to those obtained from dispersed mounts. The latter are more easily obtained because the clay is dispersed across the whole grid, whereas electron-transparent areas are limited to the perimeter of small holes in the ion-thinned rock sections. Textural information, however, is preserved.

6 450 C.R. Hughes et al. TAnLE 3. Bulk XRF data for illitic clay separates. Recalculated to mica formulae on a water-free basis (assuming 20 oxygens and 4 hydroxyls, and all iron to be ferric). I1 <2/an I1 2-8/~m I2 < 2/~m I2 2-8/an I3 < 2/an I3 2-8 pan Si A AI " Ti Fe(III) Mg Ca Na K Octahedral Intedayer OH The clay separates were also analysed as bulk powders by XRF. For monomineralic samples this is obviously a good check on the quality of the experimentally more difficult analytical electron microscopy. The presence of any contaminant phase, however, will lead to differences. The XRF data, as mica formulae, are listed in Table 3. For samples I1 and I3, XRF and ATEM analyses are very similar, confirming both that the analytical methods are sound and that the bulk samples are relatively pure illite (contaminants identified as traces of quartz, kaolinite, carbonate and iron oxides). Major kaolinite in sample I2 (avoided in ATEM analysis) clearly influences the bulk composition. Clay/solute reaction batch-dissolution experiments The batch ICP leachate data are plotted in Figs 2 to 11.* In Fig. 2, leachate Si and AI concentrations are plotted for all experiments. This overview suggests immediately that dissolution may be occurring by different mechanisms. The results of many experiments plot with similar Si/AI ratios suggesting, perhaps, stoichiometric dissolution. Others, however, fall distinctly to the right, suggesting either relative enrichment in Si or depletion in AI. In Figs 3 to 11, the leachate data for the full range of reagents used have been plotted for the < 2/~m fractions of I 1, I2 and I3. For ease of comparison, various solute species have all been plotted against the concentration of Si. In Figs 3 and 4, AI and Si concentrations are plotted together with independently determined (XRF) AI/Si atomic ratios for the two illite samples (I1 and I3). For all but the strongly alkaline solutions (f and g), leachates give similar AI/Si ratios. Quartz contamination probably accounts for the departures from anticipated ratios based on the XRF data. The leachate data (major species, expressed as /zg g-') can be expressed as mineral formulae. Those for dilute NaOH (0.01 M) are recalculated in this way in Table 4, together * The complete data set will be published in a final contract report (see Acknowledgments) and may also be obtained directly from the authors.

7 Chemical reactivity of illites and implications for petroleum production 451 6" 5" 4" 2" r.. 9 ~..~...'o... 0 i i / i i i i i Si (~JM.g-1 ) FIG. 2. AI concentration vs. Si concentration for all batch.experiment eacbates. + / 5" Si:AI ratio m. 4" 3" c I d. " h "g 1" b e O' k "1 "f ; ; 10 1'1 Si (pa.g -~ ) FIG. 3. AI concentration vs. Si concentration from I1 <2pan fraction batch-experiment leachates, with A1/Si atomic ratio for original I 1 as determined by XRF. General key for Figs 3 to 10. (a) distilled water, (b) 0-01 M l-t HC1, (c) 0-1 M 1-t HCI, (d) 1.0 M I- l HCI, (e) 0.01 M I-I NaOH, (f) 0-1 M 1-1 NaOH, (g) 1.0 M 1 -l NaOH, (h) 0-1 M 1 -I C6HsOT.H20, (i) 1-0 M 1-1 C6HsOT.H20, (j) 0.01 M 1-1 NH2OH.HCI, (k) 0-1 M 1-1 NHeOH.HCI, (1) 1-0 M I -l NHeOH.HCI. with XRF formulae for original clay (Ca assumed to be located in carbonate). The agreement is good in both cases, and remarkably so for I1. In these dilute solutions, dissolution does appear to be stoichiometric. The strongly alkaline leachates (f and g) are rich in Si but with much less than the expected equivalent AI concentration. Either Si is preferentially leached (leaving an Al-rich residual surface phase) or simple dissolution has been followed by precipitation of an Al-rich phase (possibly a hydroxide gel) prior to analysis.

8 452 C. R. Hughes et al. 5" Si:AI ratio 3",c 2" 1 *g ~,,"-~.f 0 '.-i = = / i ~ = S Si (pm.g-1 ) FxG. 4. AI concentration vs. Si concentration from I3 < 2 #m fraction batch-experiment leachates, with A1/Si atomic ratio for original I3 as determined by XRF. TABLE 4. Leachate ICP data for 100 mg I1 and I3 < 2/an fractions reacted with 10 g 0.01 M NaOH for 48 h at 80~ compared with bulksample XRF data. All data recalculated to mica formulae on a waterfree basis (assuming 20 oxygens and 4 hydroxyls, and all iron to be ferric). ICP data recalculated assuming the illite to be Ca-free. I1 (ICP) I1 (XRF) I3 (ICP) I3 (XRF) Si A AI 3" Ti Fe(III) Mg Ca Na K Octahedral Interlayer 1 " "58 OH From these experiments it is clear that illite reacts significantly with a range of solute species. The extent of reaction depends on the nature of the solute (mineral acids and alkalis being most reactive) and on concentration (especially for alkaline solutions). Figs 5 and 6 document the pattern of K dissolution. Again, broadly stoichiometric behaviour is indicated except for the very strong sodium hydroxide and hydroxylamine hydrochloride solutions. Displacement of K by Na + and NH3OH from interlayer sites in addition to simple disolution is indicated.

9 Chemical reactivity of illites and implications for petroleum production Si:K ratio (XRF data) I 1-9 d".c i ~ ~ ~ 0" a i i = i = = = i i i ' Si (~M.g-1 ) FIG. 5. K concentration vs. Si concentration from I1 < 2 #m fraction batch-experiment leachates, with K/Si atomic ratio for original I1 as determined by XRF ~2- Si:K ratio 1.1 d =C k b h Si (tam.g-1) FIG. 6. K concentration vs. Si concentration from 13 < 2/tin fraction batch-experiment leachates, with K/Si atomic ratio for original I3 as determined by XRF. Figs 7 and 8 exhibit quite different behaviour, with no correlation between Si and Ca, Mg and Fe. Broadly constant values in the more reactive solutions suggest dissolution of contaminant carbonate and oxide: both are present in the original core samples. Figs 9, 10 and 11 are the equivalent plots for the mixed sample 12 (illite/smectite plus kaolinite). Dissolution patterns are essentially similar except in that strong alkali leachates retain both AI and Si in solution. Preferential leaching of Si is not indicated. This pattern of behaviour rather suggests that the explanation for anomalous Si/AI behaviour in Figs 3 and 4 (f and g) is due to selective leaching rather than gel precipitation since it is mineral-specific.

10 454 C. R. Hughes et al. " E 8 E o$ O 0 [] Calcium E LU n i R Magnesium and iron 0.0 n, 0 1,,-= _=,,, ; si (~M.g-1 ) F]G. 7. Ca, Mg and Fe concentrations vs. Si concentration from I1 < 2 #m fraction batchexperiment leachates. m. 2.0".g_ 1.5. o calcium ~ 1.0' 8 =q= o o a ~ 0.5" o =~ ~ 8 II magnesium and iron 0.0 n, ~l,,-~ i i i Si (~M.g-1 ) FIG. 8. Ca, Mg and Fe concentrations vs. Si concentration from 13 < 2 #m fraction batchexperiment leachates. Clay~solute reaction flow-cell experiments Several of the flow-cell runs proved difficult because of blockages in the plumbing caused by amorphous hydroxides precipitating down-line from the cell, particularly in the stronger acids and alkalis. This means that precise comparisons between runs are not possible: some samples were traversed by much larger volumes of leaching solution than others. On the other hand, the flow experiments certainly demonstrate that unstable solutions are formed by reaction and that formation damage by scale precipitation is to be anticipated within reservoirs subjected to these solutions. Table 5 summarizes the information obtained by visual inspection and from XRD traces of samples following reaction. As expected, dissolution/reaction is most marked for the strong acid and alkali experiments. Although not a concern of this particular paper, kaolinite is clearly more reactive than the illite/smectite in the same sample, especially toward alkali. Comparable experiments with kaolinites and chlorites will be described elsewhere.

11 6- ~ Si:AI ratio (XRF data) ~a 2 h "i 1 e e J, " f 0 J J i a j Si (}JM.g 1 ) FIG. 9. AI concentration vs. Si concentration from 12 < 2/tm fraction batch-experiment leachates, with AI/Si atomic ratio for original 12 as determined by XRF. 4" 3" d, " c Si:K ratio (XRF data) 1 h "i '~ ~ " " " I k " b f J, Si (~M.g 1 ) FIO. 10. K concentration vs. Si concentration from I2 < 2 ~n fraction batch-experiment leachates, with K/Si atomic ratio for original I2 as determined by XRF iron w c A A 0.0 ~ R, 0 I A A A magnesium A 4 a calcium o a o i i i i J Si (JJM.g 1 ) FIG. 11. Ca, Mg and Fe concentrations vs. Si concentration from 12 < 2 ~a fraction batchexperiment leachates.

12 456 C. R. Hughes et al. TABLE 5. Qualitative summary of flow-cell experiments: visual examination of sample followed by XRD analysis. Distilled water 0.1 M NaOH 0.01 M NaOH 0.1 M HCI 0.01 M HCI BP I1 Little reaction Slight dissolution No obvious dissolution Obvious marked dissolution Obvious dissolution BP I2 Little reaction Marked dissolution of kaolinite fraction only Less marked dissolution of kaolinite fraction only Obvious dissolution Slight dissolution j\ J t/l, \; Control Distilled water 0.01 HCI 0.1M HCI O.01M NaOH O.1M NaOH FIG. 12. XRD profiles for 001 reflections of I1 residues following flow-cell experiments. Lower trace: air dried; upper trace: glycolated. Figs 12 and 13 illustrate XRD profiles (air-dry and glycolated) for the products of these runs for I1 and 12 respectively. Traces cover only the illite and illite/smectite (001) reflection region. SUMMARY AND CONCLUSIONS There have been several experimental studies of illite stability, one of the most recent and comprehensive being that of Saas et al. (1987). According to their findings, smectite, illite, mica and kaolinite all have substantial fields in aqueous solutions at 80~ Stoichiometric

13 Chemical reactivity of illites and implications for petroleum production 457 Control Distiled water 0.01 HCI 0.1M HCI 0.1M NaOH FIG. 13. XRD profiles for 001 reflections of I2 residues following flow-cell experiments. Lower trace: air dried; upper trace: glycolated. td 0.01M NaOH dissolution in very dilute solutions (undersaturated with respect to K, A1 and Si, and therefore aggressive) would be expected, and dilute reactant solutions would be anticipated to come to equilibrium with any of these minerals after modest reaction. The same cannot be said of the strongly alkaline and acid solutions used in this study. The apparent instability of product solutions and the short duration of experiments both make it highly unlikely that equilibrium has been attained. All these experiments were conducted under conditions far removed from those of reservoirs. The approach, however, allows many more tests to be done than could possibly be attempted under simulated reservoir conditions. Our approach was intended to establish the likely nature of possible reactions and identify particularly sensitive systems. Such information should allow the design of the most appropriate experiments under more realistic conditions. Even at relatively low temperatures and atmospheric pressures, however, we have demonstrated that illitic clays are indeed sensitive to reaction with a range of aqueous solutions. Reactions are rapid (hours) at 80~ they could confidently be predicted to occur even faster at the higher temperatures encountered in many reservoirs. This would apply to a range of drilling, production and stimulation procedures. Illite appears to behave quite differently in acidic and alkaline solutions. Acids cause diffraction peak broadening which strongly suggests interlayer hydration. The effect would be enhanced by the presence of high-hydration cations (Ca 2+, Mg 2+) in acid solution. Swelling and/or detachment are obvious possibilities with permeability reduction in consequence. Strongly alkaline solutions attack illites at least as effectively as acids, although seemingly by quite a different mechanism. An Al-rich residual phase (gel?) is left associated with

14 458 C.R. Hughes et al. unreacted clay. Possibly even more important for production operations is the indication of even greater sensitivity of kaolinite. The most dramatic clay dissolution and alteration reactions will take place in the immediate vicinity of the well-bore when fluids are injected into the formation. Here, solution/solid ratios will be very high, as simulated by the flow cells. Further out, diffusive mixing with very different fluids and progressively declining fluid velocities will favour precipitation of gels as both acid and alkaline solutions evolve towards more neutral states. Modification of clay surfaces will also influence surface reactivity for adsorption (surfactant plugs). ACKNOWLEDGMENTS This project is funded by the Commission of the Europen Communities, Non-Nuclear Energy R&D Programme 'Optimization of the production and utilization of hydrocarbons', under contract No. EN3C UK 'Improved prediction of clay mineral reactivity and distribution within reservoirs'. ICP analysis was undertaken by Dr N. Walsh at Royal Hollaway and Bedford New College, University of London. The ready and able assistance of Dr J. A. Whiteman and Mr H. Brockley with the analytical electron microscopy is gratefully acknowledged. Mr H. Lock (University of Manchester) provided invaluable assistance with drafting. The equipment was developed through financial assistance from the Natural Environmental Research Council and British Petroleum Development, Dyce. The preserved core samples for the whole programme were donated by B.P., Britoil (as was then) and Amoco. REFERENCES CLIFF G. & LORIMER G.W. (1975) The quantitative analysis of thin specimens. J. Microscopy 103, CURTIS C.D., HUGHES C.R., WHITE~N J.A. & WHITTLE C.K. 0985) Compositional variation within some chlorites and some comments on their origin. Mineral. Mag. 49, HU6nES C.R., CURTIS C.D., WnITEMA~ J.A. & HEPING S. (1989) Applications of analytical transmission electron microscopy to clay mineral geochemistry studies. In" Electron Microscopy and Microprobe Techniques in Clay Analyses. Course notes for the 1986 meeting, The City Minerals Society. HUGHES C.R. (1987) The composition and origin of layer silicates in iron-formations and ironstones: a preliminary analytical transmission electron microscopical study. Unpublished PhD thesis, University of Sheffield. PHAKEY P.P., CURTIS C.D. & OERTAL G. (1972) Transmission electron microscopy of fine-grained phyllosilicates in ultra-thin rock sections. Clays Clay Miner. 20, SASS B.M., ROSENBER6 P.E. & KITTRICK J.A. (1987) The stability of illite/smectite during diagenesis; an experimental study. Geochim. Cosmochim. Acta 51, l 15.