|
1. |
X-Ray photoelectron diffraction studies of the micas lepidolite and biotite |
|
Dalton Transactions,
Volume 1,
Issue 4,
1988,
Page 859-879
Lynne A. Ash,
Preview
|
PDF (1921KB)
|
|
摘要:
J. CHEM. SOC. DALTON TRANS. 1988 859X-Ray Photoelectron Diffraction Studies of the Micas Lepidolite and Biotitet?$Lynne A. Ash, (in part) Stephanie L. Clark, Stephen Evans," and Anthony G. HiornsEdward Davies Chemical Laboratories, University College of Wales, Aberystwyth, D yfed SY23 7 NEComprehensive X-ray photoelectron diffraction (x.p.d.) data for a Norwegian 1 M lepidolite and a 1 Mbiotite are reported and discussed. Improved techniques for the collection and processing of x.p.d.patterns are described, including a new procedure for correcting for the variation of instrumentalresponse with the electron take-off angle. Comparison of x.p.d. patterns from rotation about near-equivalent axes at 120" to each other reveals that two major features in the interlayer ion patternscannot be attributed t o scattering by atoms nearer than 10-1 3 A from the emitting site.Enhancedanisotropy in the Rb 3d patterns relative to those for K 2p is discussed. It is shown that x.p.d.experiments involving polar rotations about an axis parallel to the unit-cell vector a can be used t odetect octahedral cation ordering involving differentiation of the two sites with cis-OH groups; asimilar method for detecting tetrahedral cation ordering is suggested. Rotation about axes at 120"to a is valuable for the identification of elements segregating equally into the two cis octahedralsites, while experiments involving rotation about axes parallel t o and at 120" to the unit-cell vectorb enable octahedral cation ordering amongst all three octahedral sites t o be characterised.By thesemeans octahedral Li and Mn in the lepidolite are shown to be concentrated in the trans M ( 1 ) site,and Al in M ( 3 ) (one of the two cis sites equivalent in C2/m symmetry). This mica thus has onlyC2 symmetry. The 1.2% of Ti in the biotite is segregated uniformly into the two cis octahedral sites,unlike the Mg and Fe which are distributed randomly amongst all three octahedral sites.Tetrahedral ordering was not detected in either mica.X-Ray photoelectron diffraction (x.p.d.), first described some17 years ago,' has in recent years been exploited, with the aid ofsingle-scattering cluster calculations, to characterise manyadsorbate structures;' it is now a rival to low-energy electrondiffraction (1.e.e.d.) for the determination of surface structure.Inparallel with these developments, it was realised, whiledeveloping methods of quantifying the X-ray photoelectronspectra of al~minosilicates,~ that, because of the differences inlocal environment, atoms occupying different sites within asingle crystal gave rise to different anisotropies in photoelectronemission which could yield structural information aboutisomorphous substitution even at low concentration^.^ Themicas are ideal for this type of study, being readily available aslarge single crystals, easily cleaved to yield clean surfaces, andchemically unreactive, but other materials for which analogousinformation is valuable, such as the spinel^,^ can also be studied.In this paper we describe first the semi-automated methodswhich have superseded our earlier, more primitive, data-collection procedures.The results of a comprehensiveinvestigation, facilitated by these techniques, of x.p.d. in two ofthe micas previously studied, a lepidolite from Harydaler,Telemark, Norway and a biotite from an unknown location 7 9 8are then presented and discussed.Compensation for the variation of the instrumental responsewith the electron take-off angle (or polar angle), 8, was notformerly possible, and results were thus necessarily reportedin the form of graphs showing normalised ratios between X-ray photoelectron spectroscopy (x.P.s.) peak intensities as afunction of 8. Such plots are independent of the responsefun~tion.~ A correction procedure for the response functionhas now been developed, and we report for the first time theangular dependence of both ratios and individual peakintensities.The former remains the most easily interpretedt Presented at a meeting of the Polar Solids Discussion Group of theRoyal Society of Chemistry, 15th December, 1986.$ N o n - U . units employed: eV M 1.60 x J; Torr M 133 Pa.format for some applications," and can often be determinedwith greater precision than the angular dependence of theintensity for an individual peak, but the physical origin of theanisotropy is more readily appreciated in the latter format.A principal aim of the present work was to examine thetentative suggestion that differences between the A1 and Mn, Lipatterns in the lepidolite were due to octahedral cation orderingand to investigate whether differences between the Ti andMg,Fe patterns in the biotite7 could also be explained in thisway. One approach to this task would be to develop aquantitative theoretical model within which the x.p.d.patternscould be calculated. Progress is being made l 1 in this but resultsfor solids as complex as the micas are not yet practicable. Thealternative adopted here is to collect a comprehensive set ofx.p.d. patterns for both micas (the previous data related torotation about only one axis) and apply qualitative arguments,largely based on symmetry, to the resulting data. In addition toproviding more definitive evidence relating to cation ordering inthese micas, an improved understanding of the x.p.d.phenomenon itself was also anticipated.The Physical Origin of X-Ray Photoelectron D$fraction.-Because of the electron wavelengths (h = ca.0.3 A) and anglesinvolved, the positions of x.p.d. peaks on the angular scalecannot be predicted by application of Bragg's law. Thesepositions can, however, for simple crystals, be correlateddirectly with densely packed planes within the crystal, the Braggangles then being related to the widths of the peaks. In this'electron channelling' or 'Kikuchi band' description,' theangular anisotropy is thus associated with structural features ofan (implicitly infinite) crystal. Recently, however, x.p.d.experiments on nickel single crystals before and after depositionof epitaxial overlayers of Cu l 3 have shown that the x.p.d.effectcan be convincingly described in terms of enhanced forwardelastic scattering of the outgoing photoelectron wave by atomsnear to the emitting site. These experiments suggest that x.p.d.effects are relatively short-range, scatterers beyond ca. 9 860 J. CHEM. SOC. DALTON TRANS. 19880000Si0OHOctahedralionFigure 1. (a) Axes of rotation in relation to the mica structure: the c axis is directed into the plane of the paper. T and C indicate an octahedral ionwith trans and cis OH,F groups respectively. The cis sites labelled 2 and 3 are equivalent in C2/m symmetry. (b) Axes of rotation in relation to theexperimental geometry. e Represents the electron beam detected by the analyser.The analyser entrance slit is located vertically above the sample; a, 6,and c represent the unit-cell vectors of the mica. The crystal is shown in the orientation denoted by a, i.e. the a vector lies parallel to the axis of rotationof the probe. The angle between the incident X-rays and the detected electron beam is constant at ca. 90". (c) Symbols applicable to Figures 3-15having only a small influence on the anisotropy of electronemission. The electron attenuation length in mica, however, isca. 34 A at 1 144 eV,4 rather longer than in typical metals (ca.14 A 14), and so longer range effects may be more significant inthe micas. The 'enhanced-forward-scattering' concept, unlikethe Kikuchi-band approach, allows a distinction easily to bemade between different emitting sites in a complex crystal, andboth approaches have qualitative utility in appropriatecircumstances.The angular anisotropy is fundamentally related todeviations from plane-wave character of the final state of thephotoelectron, resulting from the presence of neighbouringatoms.Such a description emphasises the close relationshipbetween x.p.d. and EXAFS (extended X-ray absorption finestructure), another new technique for solid-state structuredetermination,15 and is the starting point for the developmentof a quantitative theory. l 6The Structures and Compositions of the Micas.-Each layer inan ideal mica structure is built up from upper and lower sheets,each consisting of M 0 4 tetrahedra linked by their corners intoapproximately hexagonal rings, enclosing a central sheet ofoctahedrally co-ordinated cations, which share oxide ions withthe tetrahedral The co-ordination shell of theoctahedral cations is completed by OH- and F- anions: twothirds of the cations [usually designated as M(2)] have a pair ofcis OH,F neighbours and the remainder [M(l)], trans OH,Fgroups.The layer is characterised by a C2 rotation axis and amirror plane; the a unit-cell vector (5.3 A) lies in this plane,perpendicular to the C, axis, b (9.2 8,) lies parallel to this axis,and the planes containing both a and b lie parallel to thecleavage planes. At 120" to both a and b directions, 'pseudo-a'and 'pseudo-b' directions can be identified. The similarity ofenvironment for atoms in corresponding positions is howeverlimited to neighbours within a few A perpendicular to the layersbecause of the -a/3 offset between the upper and lowertetrahedral sheets in each layer, necessary to accommodate theinclined octahedra in the central sheet.The distance traversedbefore non-equivalence is first encountered increases progres-sively as the direction sampled approaches a plane parallel tothe layer. We show below that both the micas studied here are1M polytypes, in which the a/3 offset occurs in the samedirection in every layer. The mirror plane then ideally extendsthroughout the crystal, which has C2/m symmetry.In our x.p.d. experiments, a range of ca. 90" in 8 (-10< 8 < 80", specified throughout with respect to the normalto the surface) is the maximum practicable, whereasconceptually a range of 180" exists for each rotation axis.Rotation of the sample through 180" azimuthally, followed byrepetition of the experiment through the same range of 8,permits ca.160" to be investigated experimentally. For 8 > 80"electron refraction within the crystal becomes significant andultimately internal electron reflection prevents examination ofthe last few degrees.' Throughout, we shall denote directionsalong axes parallel to a and b and their associated x.p.d. patternsby a, a,, and b,b,, the subscript r denoting 'reversal' (180"azimuthal rotation) of the crystal. Experimental axes of rotationparallel to the pseudo-a and -b vectors and their associatedpatterns are then designated a', a:, a'', a: and b', b:, b", and b:.Allhave been investigated experimentally for both micas. Theexperimental axes of rotation are related to the crystal structurein Figure l(a) and (b); the directions of a and a, were establishedby Laue photography as described in the next section. Figurel(c) shows the symbols used in later figures.The near-surface regions of these micas have compositionsclose to the rational formulae637 (K0~81RbO~11)(Li1.50A11.19-MnllO. 3 1 3 5 1 ~ ~ 0 . 6 1 lo 1 0(0H)0.4F 1.6 (lepidolite) and(KO.8 ,Na0.02)(Fe111 .24Mg0.99A10.46Ti0. 1 1 Li0.02)(si2.93A11 .07)-010(OH)1.6~0.4 (biotite).ExperimentalDetermination of the Polytype and Orientation of the Micas.-The c repeat distances, which are ca. 10 A for the 1M micas, 20 8,for the 2M1 and 2M,, and ca.30 A for a 3T mica,18 weredetermined from oscillation photographs of small samples ofeach mica. Initially, mechanical cutting or sawing caused somuch damage to the edges of the flake that very indistinctphotographs were obtained. Larger crystals (2-3 mm) weretherefore cut, and suspended briefly in molten NaOH: thJ. CHEM. SOC. DALTON TRANS. 1988 86 1damaged regions were thereby dissolved, and after washing withdilute HCl and water the (now much smaller) crystals weremounted in the usual way. For the biotite, all the specimens soexamined had a single-layer repeat. The lepidolite crystalsactually studied by x.p.d. also had a one-layer repeat, but three-layer (3T) and 20-A crystals were also found.It is necessary also to demonstrate the presence of a mirrorplane in a one-layer structure to identify the polytype as lM,because disordered stacking of the layers is not uncommon.' 7918A close approximation to a mirror plane was easily located intransmission Laue photographs of both micas (camera-to-sample distance 2.5 cm), and tilting the crystals so that the cdirection lay parallel to the X-ray beam then enabled the a and bdirections to be identified both on the crystals previouslystudied by x.p.d.and on those studied here. Laue photographswere taken at 2 4 mm intervals over the entire surface of thelepidolite crystals after x.p.d. study, to confirm their structuralhomogeneity. Small regions of 3T and 2M structures were foundnear one edge, but because of their small size and position nearthe crystal periphery they are most unlikely l9 to have affectedthe x.p.d.data significantly.A number of Laue photographs were also taken fromdifferent portions of the original large crystal of biotite: themirror plane was always present and in the same orientationeven for samples taken as much as 20 cm laterally and severalmm vertically from each other. The layer orientation is thuspreserved over very long distances.Collection of' X . P. D. Data.-The mica samples were cut fromcleaved flakes < 1 mm thick to hexagonal shapes about 10 mmacross. In the lepidolite (as with other micas we have studied) aclearly visible striation (a slight ridge) observed at the edge ofthe x.p.d.crystal was found to be accurately aligned parallel to b;for the biotite one edge of the crystal was cut as accuratelyparallel to b as possible (ca. 1"). The crystals were mounted bystainless-steel screws and clips, coated with Aquadag colloidalgraphite to ensure that no peaks other than C 1s could originatefrom the mountings. Alignment of the rotation axis wasachieved visually using a protractor in conjunction with theabove-mentioned striation (lepidolite) or the crystal edge(biotite).Each data set comprised ca. 24 groups of X-ray (Mg-K,)photoelectron spectra, recorded at 3.75" intervals in 8 using anAEI/Kratos ES200A electron spectrometer with a multichannelanalyser (MCA) and microcomputer data system.20 Thisangular interval is comparable with the instrumental angularresolution: diagrams of the electron emission geometry havebeen given el~ewhere.",~' The data were collected in 0.25-eVchannels during multiple sweeps at 0.5 eV s-' and acccumulatedin the MCA: considerably more sweeps were required for theweaker peaks to secure acceptable signal-to-noise ratios.Typical automated data collection sequences are shown inTable 1.Each group of spectra required ca. 60 min to collect;any (small) time-dependent drift in overall sensitivity was, bysuch sequencing, largely averaged out.Adventitious carbon contamination was removed (usuallydaily) from the lepidolite crystals by exposure to microwave-excited nitric oxide (a potent source of active oxygen 22), at ca.70 "C for < 300 s at ca.5 x lt5 Torr, in the sample preparationchamber of the spectrometer. This process had a negligible effecton the spectrum of the underlying mica. Small progressivechanges were noted in the 0 1s profile, well away from the peakmaximum, probably arising from surface oxygen-containinggroups on the Aquadag coating the sample mounting clips. Thesame lepidolite surface could therefore be used for all the 6-group data (12 sets) and the first four sets of a-group data.However, because (as described below) the a and a, patternswere found (unexpectedly) to be significantly different, a thirdTable 1. Data-collection sequences for the typical data shown in Figure2(a)-(d) Energyscannedper sweepBiotite Initial kinetic energies (eV) of sections (eV)779 1 183 528 709 944 1086 1 120 * -* -* -* -* -* -* --*4 3 + 174- 3 - 54Sweep? t 18 + 1554 sequence - 3 - 4 3 * 174Energyscannedper sweepLepidolite Initial kinetic energies (eV) of sections (ev)1184 597 708 945 1087 1120 1168 * -* -* -* -* -* -* -.*4 13922i 4 + 139Sweep? isequence c 40 -,t Number of sweeps is indicated.pair of patterns was collected using a different surface cleavedfrom the same original crystal.Data for the four remaining axeswere then collected using the latter surface.The effect of oxidative cleaning on the biotite was moresevere. Although no major changes were observed in thespectra, the surfaces appeared slightly 'hazy' on removal fromthe instrument, and after prolonged exposure stresses induced inthe surface (presumably as a consequence of oxidation ofoctahedral Fe" to Fe"') caused very thin layers to curl up andcleave spontaneously from the crystal on exposure to theatmosphere.Consequently, exposure of the biotite to thiscleaning process was minimised, and new surfaces were exposedby cleavage for almost every data set (i.e. group of 24 spectra).Each distinct pattern was collected at least twice to establishreproducibility, so that for the lepidolite 22 complete data setswere ultimately collected. The b, b', and 6" data from the biotitewere, however, in general, not significantly different; nor were b,,b:, and 6:. Similarly, a and a, were essentially identical, as werea' and a:, and a" and a:. Consequently 12 data sets sufficed forthis mica.The total data-collection time was in excess of 1 000 h:with a more sensitive modern instrument either the sample sizeor the collection time could have been substantially reduced.Duplicate data sets were never collected immediately follow-ing the initial set; the crystals were reoriented for every set. Noeffects persisting through both sets are thus likely to be due torandom azimuthal alignment errors, which we assess as c2".Data Processing.-First, the Mg-K,,,, X-ray satellite peakswere subtracted 23 and the data smoothed by convolution witha Gaussian function.24 The optimum results [assessed by thequality of the straight-line x.p.d. patterns subsequently obtainedfor two shells of the same element close in kinetic energy (k.e.)]were given (as expected) by using the broadest Gaussian whichdid not cause adjacent peaks from different elements to overlap(full width at half maximum = 2 eV).Initially, backgrounds were subtracted using the Shirleyapproximation 2 5 except when (unusually) the background wasgreater to high k.e.(in which case a linear function wasassumed), and peak areas obtained by numerical integration.However, although peak areas are essential for obtainingreliable elemental compositions from x.P.s., it was found thatareas are not as precise an indication of relatiue intensity as ar862 J. CHEM. SOC. DALTON TRANS. 1988(a, x 3 . 8 x 8 . 0 x 2 . 8Ti 2p Fe 3p Mg 2p Fe 2p L--L&x 4 . 1- x 5 . 6JLPIK 2p C 1s - 994Li 1s Mn 3p Mn 2p 0 I S K 2p C 1sI I I I I I l l I l l I I I I J1184 597 708 945x 3 .3 IA1 1 1 1 1 1 1 1 1 1 1 1x 3 . 5x 6 . 7 I x 3 . 8 I x l lI I I I l 1 I I I Ix 7 . 3 x 4.7 x 14Si 2s A l 2s Rb 3d Si 2p A l 2pI I I I I I I I I I1087 1120 1168Kinetic energy + 40 eV -4L m I d , , , , , , , , ,C 0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70 8001 , I , I I I , , Ic -Polar angle 0 / OFigure 2. (a) Typical data for the biotite, as collected; (6) after removal of Ka3,4 satellites and smoothing; (c) and ( d ) typical data for the lepidolite,as in (a) and (b); (e) and (f) typical normalising functions for biotite and lepidolite respectively, see textpeak heights. Errors in the level of the background intensityaffect areas far more than heights because peaks are broadest atthe base.previously determined limits approximating to background tolow and high k.e.by means of a least-squares quadratic fit to thehighest five data points.26 The background was then taken asthe height at a preselected energy 3-6 eV to higher k.e., asEach peak maximum was therefore located between b e f ~ r e , ~ . ~ and the net intensity recorded as the differencebetween the height at the turning point of the quadratic and thaJ. CHEM. SOC. DALTON TRANS. 19880.68631 . . , , . . . %.a and a,( a ) Si/K 2p00a and a,t ( a ) Ti 2p1 . 20.81 . 4 1 ( b ) S i / O 1s1.0( c ) Si/Mg + Fe 3p t1.20.81.41.0r- ( d ) T i 2p/Mg + Fe 3Pd1.20.80 20 40 60 80I # m a . * * * *0 20 40 6 0 80e l o( i 1Figure 3.(i) Biotite x.p.d. patterns, for the a and a, rotations. Ordinate, normalised X.P.S. peak intensity ratio; abscissa, polar angle, 8. The solidlines indicate the mean values of the normalised ratios. Mg = Mg 2s + Mg 2p; A1 = A1 2s + A1 2p; Si = Si 2s + Si 2p. Symbol types 1 and 3(see Figure 1) identify data obtained by rotation about the a and a, axes respectively. (ii) x.p.d. curves corresponding with the patterns in (i).Ordinate, normalised peak intensities (corrected for instrumental response function); abscissa, electron take-off angle, 8. The mean values of thepeak intensities are indicated by the solid linesat the defined background energy. Linear interpolation between two close peaks, that at higher k.e. arising from the Aquadag ondata points in the background region was used to keep the the sample mounting clips.Two sets of spectra at one value of 8energy separation (peak to background) constant for each peak. were sometimes recorded, the second after standing in wcuo forThe application of these processing steps, and the limits used, > 15 h. These usually showed a significantly larger contaminantare illustrated in Figure 2(a)--(4. C 1s peak and lower intensities for the mica peaks. ComparisonThe C 1s region was recorded to allow a correction for of the two allowed an attenuation factor (effectively exp[-attenuation due to surface contamination. This region shows (contamination layer thickness/h cos e)] '1 to be estimated fro8641.61 . 2a ' and a;'IIJ.CHEM. SOC. DALTON TRANS. 19880 ' ' and a;2 . 0( a ) S i / K 2 p1 . 21 (9) S i / F e 2 p -0-8t ( b ) S i l o 1s( a ) S i / K 2p1 . 20.81 ( b ) S i / O 1s1-2 1 . 21.00.8 (c) Si/Mg + Fe 3p f1.20.8.o 1 . 21.0+C( f ) S i / T i 2 p e1 . 60.8 l*'/asPMp$(9) Si/Fe 2p2 . 0 .1 . 6 -0 20 40 60 806 / O( 1 )( c ) Si/Mg + Fe 3 p 1 . 61 . 20.80 00 1 . 2 -8 I ~- . . . . . o . .( e ) S i / A l 1 . 21 . 0 1 ;-( f ) S i / T i 2 p01 0.8;; 0 001 D . . . . . . .1.2 k d0 20 40 60 806 / 0( i i )Figure 4. Biotite x.p.d. patterns for the pseudo-a rotations: (i) for the a' and a:' rotations (symbol types 1 and 3 respectively), and (ii) for the a" anda: rotations (symbol types 1 and 3 respectively)the contaminant C 1s intensity for any other similar data set.1s (lepidolite). Magnesium and Fe in the biotite, and Mn and LiThese corrections were never greater than a few per cent. in the lepidolite, were previously found 6 3 7 to yield essentiallyX.p.d. patterns (plots of peak intensity ratio/mean ratio us. 0) equivalent x.p.d. patterns. These summations improved thewere then computed and plotted as b e f ~ r e . ~ , ~ Since all peaks reproducibility of the patterns significantly. Ultimately, using afrom a given site which occur close in k.e. have indistinguishable mainframe computer, the x.p.d. patterns could be obtainedx.p.d. Si 2s and Si 2p data were summed, as were A1 within an hour of completion of data collection.2s and A1 2p, Mg 2s, Mg 2p and Fe 3p (biotite) and Mn 3p and Li When plotting sets of patterns on common axes, lateraJ.CHEM. SOC. DALTON TRANS. 1988 865a' and a;' a" and a;( a ) Ti 2p t (a) Ti 2p1.20.8I 1.000 0 .. 1.0 /I O0 00.6 -, %, 0.61 , , , , , , , hp c ( C ) 0 Is(e) Mg + Fe 3p E,l . O / p,0 20 40 60 80e / o( i ) t0 20 40 60 80e / *( ii )Figure 5. X.p.d. curves corresponding to the patterns of Figure 4(angular) alignment was achieved by minimising the root-mean-square difference between the different data sets as a function ofangular offset. This was necessary because the original visualalignment was less precise than the width of the sharpestfeatures of the patterns.Finally, the symmetry properties of the crystal were used toestablish the correct absolute scale for each plot.In C2/msymmetry, a pair of experiments differing only by 180"azimuthal rotation and carried out with the rotation axisparallel to a should be equivalent, whereas for a rotation axisparallel to b they correspond to rotating the crystal in oppositesenses about 8 = 0". For sets of a and a, data obtained withoutdisturbing the angular relationship between the crystal surfaceand the scale on the probe, the true scale was therefore obtainedby dividing the angular offsets required to align the two sets ofdata equally between the two sets: the presence of mirrorsymmetry in the regions of the patterns above and below 0"confirmed the accuracy of the resultant scale. For similarlymatched b-group patterns, offsets were applied so that theregion of the patterns for - 10 < 8 < 10" in 6, coincided with10 > 0 > - 10" in b, again dividing the required offset equallybetween the two sets.Absolute scales for the remaining datawere assigned by analogy. We estimate the accuracy of thesescales as f 1" in 8.Correction of Individual Peak Intensities for the InstrumentalResponse Function.-Any X-ray photoelectron spectrumconsists of the primary photoelectron peaks, the intensities ofwhich are modulated by x.p.d. in single crystals, superimposedon a background of inelastically scattered photoelectrons. For 866 J. CHEM. SOC. DALTON TRANS. 1988b groupt ( a ) S i I K 2p1 . 20.81 . 4 ( b ) Si /O 1s0.8 1 . 2 i cP( b ) Si/O 1s1 .21 . 00.8 I ( c ) Si/Mg + Fe 3p II ( c ) Si/Mq + Fe 3p81 . 2 1 . 20.8 0.81 ( d ) Ti 2p/Mg+ Fe 3p1 . 21 . 01 . 20.8c ( h ) S i / T i 2p( d ) Ti 2p/Mg + Fe 3pt R(€9 Ti 2p/Mg + Fe 3p t 01 . 21 * 20.80.8t ( i ) S i / F e 2~0.80 20 40 60 80e 10( i )0.8 11 . 4 ( i ) S i / F e 2p20 40 60 80 00/0( i i )Figure 6. Biotite x.p.d. patterns for the b-group rotations. In (i) the three symbol types (1-3) relate to the b', 6, and 6" rotations respectively, in(ii) to the b:, b,, and b: rotations respectivelygiven incident photon flux, the total number of primary scattered a number of times before emerging from the solid, andphotoelectrons produced should be independent of 8, since are likely to be deflected to some extent on each collision, thediffraction of the incident photons (h z 10 A) can be anisotropy of the background photoelectron angular distribu-negle~ted.~ Because most of the photoelectrons are inelastically tion should become progressively less marked as the energJ.CHEM. SOC. DALTON TRANS. 1988 867(a) Ti 2 p (true) L. . . . . . . .t ( b ) Ti 2 p (pseudo)0l . O I v 0L U( f ) Mg + Fe 3p(a) Ti 2 p (true) 1 . 4 tL \0.6 - b( b ) Ti 2 p (pseudo)1 . 01 . 4 1 ( d ) 0 1s0.8 > t ( h l Si. . . I( f ) Mg + Fe 3p 1 . 4 .000 2 0 4 0 60 80e 10( i )t ( h ) Si1 . 20.8Figure 7. X.p.d. curves corresponding to the patterns in Figure 6, for the b-group rotations in the biotitesampled is decreased, i.e. the number of inelastic scattering superimposed on it, but from peaks at much higher k.e.s: thisevents suffered by the electrons is increased.It is thus reasonable background in total thus should be negligibly modulated byto expect that the background at energies well below the x.p.d. The background is however still proportional to theprimary energy should not be significantly modulated by x.p.d. integrated photon flux, and to the detection efficiency (aeffects. Now, in each set of spectra (Figure 2), most of the complicated function of experimental geometry').recorded background does not originate from the peak Consequently, by integrating the background after subtrac868 J. CHEM. SOC. DALTON TRANS. 1988ac ( a ) S i / K 2 p0.8 1.21 I , , v , , , ,( b ) Si/O 1s1 .21 . 01 . 21 . 0( b ) S i / O 1s1 . 2 881 . 0 1t . . . - . I . I I(c) Si/AI2 1 . 2ECCI0.8.- S L ( d ) Si/Mn 3 p + Li I scr1 . 20.8t o 01 . 20.8(e) S i / R b 3 d801 . 21 . 20.80.80 2 0 40 60 80e / o( i )0 2 0 40 60 80e 1 0( i i )Figure 8. Lepidolite x.p.d. patterns for (i) the a and (ii) the a, rotations. Symbol types 1 and 2 relate to data from the crystal used to obtain the b-group results shown in Figures 12-15; symbol type 3 denotes data from a second crystal, that used when collecting the data shown in Figures 10and 11. The two crystals were cleaved from a common parent, and the duplicate data sets were not collected immediately following the initial settion of the elastic peak intensities a normalising factor for agroup of spectra as a whole can be obtained.This factor isplotted as a function of 8 for two typical data sets in Figure 2(e)and (f’); the x.p.d. modulation is, as expected, essentially absent.Individual peak intensities were therefore calculated by dividingthe net peak intensity by the normalising factor, and thendividing by the mean corrected intensity (over all angles). Plotsof corrected intensity us. 8 then reveal the extent of x.p.d.modulation independently of the mean intensity of the peakconcerned relative to the others.One small error in the correction process arises from thecontribution to the spectra from the carbon coating of thesample mounting. The extent of this contribution varied tosome extent with the precise positioning of the probe inrelation to the analyser entrance slit, a high Aquadagcontribution leading to a spuriously high normalising factor.Variations in the extent of carbonaceous contamination of thesample surface itself will also affect normalising factors obtainedby this procedure: high levels of contamination lead to both anincreased normalising factor and a reduced count rate from themica, and hence to a low value for the corrected peak intensity.No systematic effects arising from either source were identified,but such effects nevertheless probably contributed to thesignificantly poorer reproducibility of plots of normalisedintensity against 8 (in comparison with plots of ratios betweenintensities) evidenced in the data reported below.Results and DiscussionWe shall throughout be making deductions from quite subtledifferences between x.p.d.patterns. We have found it mosthelpful to use transparent overlays of one diagram on another,and to aid this, all the data are reproduced on the same scale.Because the observation of resemblances between x.p.d.patterns tends to be subjective, we also quote, whereappropriate, correlation coefficients between the relevantpatterns. Thirty points taken at equal intervals over the central60” of the mean curves shown in the diagrams were used in eachsuch comparison. However, such coefficients must beinterpreted carefully, as, in effect, the points in any pattern mostdeviant from unity are the most heavily weighted.They are thusunreliable as a quantitative measure of resemblance whereverthe anisotropy represented by the x.p.d. pattern is small relativeto the (random) uncertainties in the data. Correlationcoefficients are thus a supplement to, rather than a substitutefor, a study of the patterns themselves.Results and Initial Symmetry Considerations.-Figure 3 showsthe results from the biotite for the two rotation axes parallel to a:the data for a and a, were found not to be significantly differentand so were plotted together. The data for the four pseudo-aaxes (shown in Figures 4 and 5 ) were very similar, but thepatterns for a’ and a: were not quite identical with those for a”and a: [compare (i) and (ii) in Figures 4 and 5 respectively]J. CHEM.SOC. DALTON TRANS. 1988 869atar1 . 2 1 . 20.8 0.81 . 21.0I ( c ) Rb 3d( d ) Mn 3 p + Li 1s z I,1 . 20.8c ( c ) Rb 3d 1 . 2v)0.8.- S 0.80 . 8 ” jt.I V0 2 0 40 60 80810( i )- . t ( d ) Mn 3p + Li 1s1 * 41.00.60 . 8 l ” j hI . . . . . . .0 2 0 40 60 80( i i 10 loFigure 9. X.p.d. curves corresponding to the patterns in Figure 8Plots of normalised individual peak intensities against 8, whichwe refer to as ‘x.p.d. curves’ to distinguish them from the plots ofpeak intensity ratios previously described as ‘x.p.d. patterns,’ areshown in Figures 3(ii) and 5.Identity of the u and a, data is expected in C2/m symmetryand inspection of Figure l(a) or a model shows that theobserved ‘pairing’ of the pseudo-a data is also a reflection of thecrystal symmetry. The slight but distinctive differences between(i) and (ii) in Figures 4 and 5 arise because planes perpendicularto the cleavage and parallel to a pseudo-a axis are notcrystallographic mirror planes.The marked differences in thetitanium data will be discussed later. The general similarity ofthe major-element a and pseudo-a data confirms that short-range effects are usually dominant, because the immediateenvironments of corresponding emitters are identical. However,there are significant differences between the a and pseudo-a datafor the interlayer ion (K’): the a and a, K 2p x.p.d. curves showpronounced peaks near 41 and 52” (arrowed, Figure 3) whichare absent from all the pseudo-a curves.These peaks also featurein the lepidolite data: they must arise from effects due to moredistant neighbours, present in relation to the a axis but not thea‘ or a” axes. However, if these features are associated withindividual scatterers the nearest possible atoms (Si) are as muchas 10.7 and 13 A from the emitting sites. Clearly, long-rangeeffects cannot always be neglected.The b-group data for the biotite are reproduced in Figures 6and 7. As expected, azimuthal rotation through 180” now causesgross changes in all the patterns (such pairs of rotations in the h870 J. CHEM. SOC. DALTON TRANS. 1988a ' and 0;'( a ) Si/K 2p ta" and a;c ( a ) Si/K 2r,0.8 1 . 2 i1 . 20.8( d ) Si/Mn 3p + Li 1s01 . 20.8t (el S i / R b 3d1 . 20.81 .21 . 01 . L . --c ( b ) S i / O 1s1 . 0 1 p,1 . 20.8I ( d ) Si/Mn 3p + Li 1s(e) Si / Rb 3d01 . 00 20 40 60 800 lo( i i )0 2 0 40 60 800 l o( i )Figure 10. Lepidolite x.p.d. patterns for the pseudo-a rotations. In (i) symbol types 1 and 3 relate to the a' and a: axes respectively, in ( i i ) torotation about a" and a: respectivelygroup are not symmetry-equivalent) but in the major-elementpatterns the true- and pseudo-b rotation axes cannot bedistinguished within each set. The recorded differences betweenthe pairs of pseudo-b and -b, data, which are in each casesymmetry-equivalent, are as great as those between these pairsand the true-b and -b, data respectively. For example, the meancorrelation coefficient between 6' and b" for (a), (b), (c), and (i) inFigure 6 ( i ) is 0.87 whereas that for the same comparisonsbetween b and b', 6" is 0.92.The b-group patterns must thus beessentially determined by short-range effects, to an even greaterextent than the a-group patterns. For Ti, however (see below),we shall show that certain of the differences between the true-and pseudo-b axes are significant; for this element only, the plotshave been separated.The Fe 2p, Ti Zp, and K 2p b-group patterns show nosystematic drift with angle (suggestive of a near-surfaceconcentration gradient 4,8); the trends in Figure 3 which mighthave been held to suggest surface depletion of Fe, and, to a lesserextent, Ti and K, must therefore actually be due to x.p.d. effects.The lepidolite data contrast quite strongly with those for thebiotite, although the basic structures of biotite and lepidolite arevirtually identical.I7 The a-group data are shown in Figures 8-11.Again there are close parallels both between the data for thedifferent near-equivalent rotation axes, and between these dataand those for the corresponding rotations for the biotite(Figures 3-5). However, there are now marked differencesbetween a and a,: compare Figure 8(i) with (ii), (a) and (b)especially. The lepidolite thus cannot have C2/m symmetry: themirror plane is absent. The peaks mentioned above for K inbiotite near 41 and 52' are nevertheless again present in both aand a, for K and Rb (arrowed, Figure 9), but not in any of thepseudo-a data. As for the biotite, differences within the pseudo-agroup are not large.The two non-equivalent rotationsrepresented by a' or a: and a" or a: are nevertheless againdifferentiated. The distinction is subtle, but successivesuperposition of sections of Figures 4 and 10 reveals the parallelsclearly, especially the consistent differences in the profiles of theSi/K patterns between 0 and 50° and in those for Si/O between50 and 60'. The mean correlation coefficient between the x.p.d.patterns from biotite and lepidolite for 'matching' orientationsare 0.80 for Si/K 2p and 0.92 for Si/O Is, cJ: 0.61 and 0.81respectively for comparisons in the opposite sense.The b-group data for the lepidolite are illustrated in Figures12-15. The b and b, data here are throughout dearlydifferentiated from the corresponding pseudo-b data.Since thebiotite data were so similar for each trio of b-group axes, thisdifferentiation indicates that in the lepidolite, unlike the biotite,there are differences in environment very close to the emittingsites. Specifically, the only plausible cause is ordering of the Li,Mn, and A1 cations in the octahedral sites, thus removing thJ. CHEM. SOC. DALTON TRANS. 1988a' and a;'87 1a" and a ;t ( a ) 0 1s1 . 21 . 0(c) Rb 3dt o 0( c ) Rb 3d E c1 . 21 . 20-8> 0.8t ( d ) Mn 3p + Li 1sc \e0-611 .ol( d ) Mn 3p + L i 1s1.00 20 40 60 800 lo( i i )0 2 0 40 60 8001"( i )Figure 11. X.p.d. curves corresponding to the patterns in Figure 10three-fold symmetry of the octahedral sheet.This is discussedin detail below.This difference between the micas is not due simply to poorercrystal quality in the biotite. The consistent presence of the twosharp features in the K, Rb patterns at 8 = 41 and 5 2 O (discussedabove) which distinguish all eight sets of true-a from the eightsets of pseudo-a data confirms the basic structural integrity ofboth crystals.The aluminium curves in biotite resemble the correspondingsilicon curves, because 70% of the A1 occupies tetrahedral (Si)In the lepidolite, only 34% of the A1 is tetrahedral andthe similarities between the curves for A1 and Si are much lessmarked. The aluminium curves here are not, however, merelyweighted sums of those for Si and Mn,Li: the aluminium sitesare clearly distinct, in part, and we shall return to this point later.Even a tentative identification of particular peaks in thex.p.d.curves with particular scatterers is rendered difficult bythe inherent complexity of the mica structure. UsingGuggenheim's2' X-ray data for a lepidolite of similar com-position, we have generated histograms representing the nearneighbours of each element as a function of 8. Atoms more than7.5 8, away were neglected and the contributions of the neareratoms weighted inversely with internuclear separation.Neighbours centred > 1 8, off the normal to the rotation axispassing through the emitting atom were also neglected. Typicalhistograms, for the b and b, rotations, are compared with thex.p.d.curves in Figure 16. Some features show a correlation; theangular gaps in the histograms match the principal minima inthe x.p.d. curves quite consistently, but the x.p.d. maxima arenot well predicted. Overall, the correlation is not remarkableand it is clear that this simplistic approach is inadequate.Preliminary calculations l6 confirm that atoms within a 60"cone from the electron emission site affect the intensitysignificantly.Comparison with Previous X . P. D. Studies.-The present Lauephotographs show that the lepidolite x.p.d. data reportedpreviously6 were obtained by rotation about the true b axisbr ( a ) S i l K 2p0 1.4 '1.0 .0.6 . . . . . ( b ) Silo 1sb'( a ) SilK 2p0 L1-41.00.61.4 . . . . .. . .( b ) S i l o 1s1.6 I. ( a ) S1 - 20-8( b ) S L1.21.20 0.82f.--4-0.8I ( d ) SilMn 3p+Li 1sc ( c ) S i l A lc0U.0.8E '0 1 . 2 . .-Y0.81.20-8. .1.0 .0.6 .1.20.81.4 '1.00.6t . , , , . , . . b0 2 0 40 60 8081"( i i )c ( c ) SilAl0.8 "l".akpgd( d ) S1.20.82,0[ (el S i1.61.20.80 20Figure 12. Lepidolite x.p.d. patterns for the b-group rotations: (i), (ii), and (iii) denote data from rotation about the b, b', and b" axes respectively. In each casesets of data from the same cleavage surface; duplicate data sets were not collected immediately following the initial setJ. CHEM. SOC. DALTON TRANS. 1988 873QcIW0070 L e ' t0 - c ?TQ NYQnYI . .. . . . . . I . . . . . . . ,Q0?7Y op0 c"p (v0 c ? ?0 c. . . . . . . . I . . . . . . . . .?0? c 0 ? ?0 - -0CD0W0 rs0 50 N874 J. CHEM. SOC. DALTON TRANS. 1988CD0Yc0 c ?c90 N O a0 N0 0 - - cNcL-Q0W0N00 . . . . . . I . . . . . . . . . . ,w0 ? - ?c"p N0 c0 0 Aa>;= * - .-0cu00 J. CHEM. SOC. DALTON TRANS. 1988 875- L90 a0u)0 - - .- a) ::v0N0rnE* +.aa0 \Q)I . . . . . . . . . . . I . . . .NT : "0 cv0 876 J. CHEM. SOC. DALTON TRANS. 19881 . 2 *0.8 - c (6) Mn 3p + Li 1s t (6) Mn 3p + Li 1s0.8 '.21-L262n1.4 1 ( d ) Si1.00.620 40 60 80 001"( i )c ( d ) Si20 40 60 80 00 1 "( i i )Figure 16. Comparison of (i) b and (ii) b, x.p.d.curves for the lepidolite [mean lines from Figures 13(i) and lS(i)] with near-neighbour histogramsderived from Guggenheim's 27 X-ray data. Wherever two scatterers were within 2' of the same direction the bar heights have been added as an aidto claritywhile in the experiments on biotite7'* rotation was about apseudo-b axis. The oscillation photographs taken in theprevious studies to determine crystal orientation did not enablethe b and pseudo-b axes to be distinguished. For both micas, thepresent data, where they duplicate previously reported data[Figures 6(i) and 12(i) only], confirm the original results. Theimproved data-processing techniques have enabled the inter-ference between Rb 3d and the A1 2s C C ~ , ~ satellites (see Figure2) to be eliminated so that the present Rb 3d data match theprevious Rb 3p patterns quite closely.Interlayer Ions: Rb + sites in Lepido1ite.-It was notedpreviously that the absence of any progressive increase in theRb/K ratios at high take-off angles demonstrates that the Rb'ions are not segregated into rubidium interlayers, but areuniformly distributed amongst the K ' ions.The presentdata for all rotation axes confirm this, but the additional dataalso reveal that the Rb 3d patterns are consistently moreanisotropic than those for K 2p (see especially Figures 9 and 12).The Rb 3d photoelectrons, at ca. 1 137 eV k.e., have a slightlyshorter de Broglie wavelength (0.37 A) than the K 2pphotoelectrons (0.40 A), and hence a smaller Bragg angle.Usingthe Kikuchi-band approach, Rb 3d x.p.d. peaks should then besomewhat sharper than those for K 2p. We attribute theenhanced anisotropy partly to this effect. However, Rb 3p, veryclose in k.e. to K 2p, also yielded6 a pattern showing smalldifferences from the K 2p pattern. Some of the differencebetween the Rb 3d and K 2p must therefore be due to otherfactors, such as the increased mass (ca. x 2) of Rb' relative toK', which presumably reduces the amplitude of thermalvibration for the heavier ion. The greater size of Rb' may alsobe important.Cation Ordering in the Micas.-There is an increasing interestin the characterisation of both octahedral and tetrahedralcation ordering in the m i ~ a s . " , ' ~ , ~ ~ , ~ ~ Ordering is ofconsiderable geochemical importance, but it is only relativelyrecently that crystallographic and other techniques capable ofproviding reliable results have become available.X-Raymethods, which only detect long-range order, have been usedextensively to characterise the distribution of ions between thethree octahedral cation sites, M( 1) with trans-OH groups lyingin a mirror plane of the ideal C2/m structure, and M(2) andM(3), both with cis-OH groups. In the ideal structure the latterare crystallographically equivalent, located symmetrically oneither side of the mirror plane. Proven examples of octahedralordering in biotite are rare, but in lepidolite segregation of thelarger cations, principally Li, into the M( 1) trans sites is quitecommon.28729 Sometimes, but not always, this is accompaniedby a further segregation of one of the remaining ions, generallyA13+, into one of the two cis sites.27p29 In the latter case, themirror plane is lost, and the space group reduces to C2.Studies of tetrahedral ordering by X-ray methods, unlikethose by neutron-diffraction technique^,^' are hindered by theclosely similar scattering powers of A1 and Si, and much recentwork 1p36 has used magic-angle spinning nuclear magneticresonance (m.a.s.n.m.r.).The information obtained differs fromthat given by X-ray methods in that it relates only to short-range order, so that domain models may be considered; but as itis also limited to distinguishing the numbers of A1 and Si nearestneighbours, unambiguous determination of the completestructure of the tetrahedral sheet is not usually possible.X.p.d.can in principle be applied in the study of bothtetrahedral and octahedral cation ordering. While not givinglong-range information in the sense that X-ray diffraction doesJ. CHEM. SOC. DALTON TRANS. 1988 877Table 2. Distribution of octahedral cations between ‘cis-like’ (C) and ‘trans-like’ (T) sites in lepidolite in comparison with correlation coefficientsbetween x.p.d. patterns. R, Rr = Correlation coefficients for indicated pairs of b-group and b,-group patterns respectively% Sites % Sitesb’,b: common, b, br common, b , b;Pattern * rotation R, Rr rotation R, Rr rotationSi/Mn,Li 1 .ooc 44%, 0.56T + 0.44C 44 + 44 = 88%, 0.44T + 0.56C0.60, 0.66 0.89, 0.72loo%, 0.63,0.8973%, 0.87,0.5661%, 0.85,0.6756%, 0.72,0.82SiJAl 0.83T + 0.17C 17%, 1 .ooc 83%, 0.17T + 0.83C0.69, 0.45 0.86, 0.82“(,Sites common, 17%,R,Rr 0.72, 0.56a%,0.66, 0.6417 + 56 = 73%,0.88, 0.75* The patterns for the octahedral ions relative to Si are used rather than the x.p.d.curves themselves to eliminate the contribution from tetrahedral Al.it is nevertheless necessary that the emitting sites be similarlyoriented over a very large distance (up to several mm). Domainstructures over which the long-range average is effectively theideal (disordered) structure will not be detected.X.p.d. has been shown to be sensitive to nearest-neighbourdirections in relation to the rotation a x i ~ , ’ ~ , ~ ~ and thus the twosets of cations at the centres of alternate tetrahedra round eachsix-cation ring should in principle give rise to different a-axisx.p.d.patterns. Ordering of A1 into one of these sets is relativelydifficult to study by X-ray crystallography, because of thenecessarily increased size of the unit cell (all the tetrahedra aresymmetry-equivalent in C2/m). The m.a.s.n.m.r. results suggestthat such ordering does sometimes occur in the 2: 1 layersilicates.31 36 If present in the micas studied here, it would bemost obvious in comparison of the a and a, aluminium patterns:the 180” azimuthal rotation would effectively interchange the A1between the two orientations distinguished by x.p.d. (cJ Zn andSe in ZnSe, where, for the appropriate axis, the patterns for Znand Se interchange on 90” rotation 37).A pronounced differencebetween the two aluminium curves would thus be expected. Nosuch effect is evident in any of the data reported here. Moreover,tetrahedral ordering could only significantly affect A1 and Siphotoelectron angular distributions. The scattering powers ofA1 and Si are so similar that effects on the angular distribution ofphotoelectrons originating from any other element in thestructure would be negligible. The lack of equivalence betweenthe a and ( I , data for lepidolite, much more obvious in the Si/Kand Si/O ratios than in Si/Al, cannot therefore be due totetrahedral ordering.This absence of the expected mirror plane must thereforedemonstrate, instead, an inequivalence of the cis M(2) and M(3)sites, i.e.that the octahedral cations order in three sites in thislepidolite, as in, cJ.g., the Tanakamiyama specimen studied byGuggenheim.” Analogous segregation in our lepidolite wouldimply an octahedral composition of M(l) Lio,83Mno,17, M(2)Li0.66Mn” 13A10.21, and M(3) with a consequentdifference in scattering power between M(2) and M(3). The twocis sites are not x.p.d.-equivalent for the a-axis rotations, the180” azimuthal rotation between a and a, interchanging M(2)and M( 3). Consequently, somewhat larger differences shouldoccur between the x.p.d. data for Mn,Li than for A1 (see Figures8 and 9). A priori, A1 could be concentrated in either cis site: theabove assignment, with M(3) defined as in Figure l(a), wasdeduced from the interpretation of the b-group data given below.As for the biotite, long-range effects necessitate the pairing ofthe pseudo-a data (a’ with a: and a” with a:) to accord with thebasic symmetry of the structure.(In other pairings the internalconsistency is much less satisfactory and no distinctionsbetween any of the pseudo-a rotations are then possible.) Thepatterns for A1 and Mn,Li in each plot are different from eachother, and features of the a’,a: Si/Al pattern (notably thepronounced dip near 47”) transfer to the a‘‘,a; Si/Mn,Li pattern,while the broader minimum centred at 42” interchangesbetween a’,ar Si/AI and a”,ag Si/Mn,Li. Azimuthal rotationsthrough 120” in effect sequentially exchange the three sites, thetrans M(l) site in a transforming into a near-equivalent of a cissite in the pseudo-a rotations and one of the cis sites similarlybecoming ‘trans-like;’ the differences are relatively small sincethe immediate oxygen co-ordination shell is similarly orientedin each case.Cation ordering thus provides a simpleexplanation of the observed interchange of features. If theoctahedral cations were randomly distributed amongst the threesites the patterns for A1 and Mn,Li would have been virtuallyindistinguishable throughout. However, to produce such effectsit is necessary only for the occupancy of the trans site to bedistinctly different from that of the cis sites, and not for the latteralso to be differentiated from each other. The only confirmatoryevidence from these plots for the differentiation of M(2) fromM(3) would be an imperfect matching of the pairs of pseudo-adata sets plotted together; in the absence of such differentiationthey would be rigorously symmetry-equivalent.In fact, nosignificant differences were observed.However, confirmatory evidence for the segregation of A1 intoone of the two cis sites can be derived from the b-group patterns.The close similarity of the 6’ data with those for b” (and 6: withbr), contrasting with b and b, respectively, indicates that themajor effect of cation ordering on these x.p.d. patterns also isdue to the differentiation of the two cis sites from the trans;several X-ray studies 1 7 , 1 8 3 2 7 - 2 9 have shown that the trans M( 1)site is enlarged to accommodate the larger cations, with loss ofthe three-fold symmetry of the octahedral sheet.The resultantdifference in mean scattering power between the ions occupyingthe cis and trans sites is probably also significant. Ordering ofthe cations additionally between the two cis sites CM(2) andM(3)] causes more subtle effects, which manifest themselvesprincipally in the residual differences between b’ and b” (and b:and br). These effects can be detected only in the patterns for theoctahedral ions themselves; even there they are not dramatic878 J. CHEM. SOC. DALTON TRANS. 1988because the six nearest-neighbour oxygen atoms are similarlyoriented around all three sites. Nevertheless, there is evidencehere not only to confirm that the A1 segregates into one of thetwo cis sites, but also to show that it is the M(3) site whichis aluminium-rich.The contribution of ‘cis-like’ (C) and ‘trans-like’ (T) sites tothe patterns for Mn,Li and A1 in the b’, bi and b”, b; orientationsin comparison with b,b, can be deduced immediately (Table 2)from Figure l(a) and the octahedral ion distribution givenabove. Also shown in Table 2 are the correlation coefficientsbetween each pair of patterns, to compare with the percentage ofapproximately common sites. Note that, because of thedistortion induced by the increased size of the M( 1) ions, it is notunreasonable that the correlation coefficient between, e.g., A1 inb and Mn,Li in b’ should be only 0.63 for an assumed 100% ofcommon sites: correlation coefficients between b,b, and b’,b: orb”,b:( patterns respectively for the non-segregated elements Kand 0 range from 0.46 to 0.82 with a mean of 0.70.Three significant groups of correlations may be identified.First, the patterns for Si/Mn,Li and Si/Al should be much moreclosely correlated for b” and b: than for b‘ and b: or b and b,, asindeed is the case.Secondly, both Si/AI and Si/Mn,Li patternsfor b,b, should correlate more closely with those for b”,b:respectively than with those for b‘,bi; this is found for all fourcomparisons of this type. Finally, b,b, %/A1 should correlatebetter with b’,bi Si/Mn,Li than with by,& Si/Al, etc.; thisexpectation is fulfilled in six of the eight such ‘diagonal’comparisons between b,b, and b’,b: in the Table.Many of theseobservations can also be made qualitatively (but moresubjectively) by inspection of Figures 12-1 5.If A1 were concentrated in M(2) rather than M(3) thecontributions under b’,bi and b,bc in Table 2 would beinterchanged. This alternative assumption of aluminiumconcentration in M(2) leads to a strong negative correlationbetween R and the percentage of common sites. Overall, theevidence seems conclusive: A1 in this lepidolite is stronglyconcentrated in the M(3) site. The octahedral ion distributiongiven above was therefore adopted when calculating thehistograms for Figure 16. However, the quantitative extent ofsegregation cannot yet be confirmed by x.p.d.The lack of distinction within each set of three 6-grouppatterns for biotite is readily explained on a similar basis.If Fe2 +and Mg2+ do not order, possibly because they are similar insize and have the same charge, there is then no distortion and novariation in mean scattering power amongst the three sites. Thepresence of the mirror plane characteristic of C2/m in the a anda, data confirms that M(2) and M(3) are not differentiated in thebiotite.Finally, we consider the titanium data for the biotite. Thedifference between the patterns for Ti 2p and Mg,Fe in the 25-55” region reported previously for a pseudo4 rotation is againobvious in the b’ and b” data sets. It had been suggested thatthe large photoelectron wavelength difference between Ti 2p(A = 0.44 A) and Mg 2s,2p or Fe 3p (h = 0.36 A) was not thecause of this difference.However, throughout the present data,Fe 2p, at even longer wavelength (0.54 A), was also recorded,and the Fe 2p patterns show the same effect, a marked change inthe intensities of the features in the 4&-55” region of the x.p.d.pattern relative to Si, even more clearly. The major part of thisdifference should thus be attributed simply to wavelength effects.However, there is one major feature of the new titanium datanot common to any of the other octahedral ion data: the largedifference between the two sets of pseudo-a data [compareFigure 4(i) and (ii), (d) and (f), and Figure 5 ( i ) (a) with (ii) (a)].Within each pair of pseudo-a axes, the trans site remainsunaffected by the change of axis whereas the two cis sites areinterchanged.Because throughout the a group of rotations theoctahedral ion patterns are made up of different contributionsfrom each of the three sites, this large difference, unique to Ti,shows that Ti must be segregated either in the two cis sitesequally or in the trans site. Either possibility preserves themirror plane, so that the good agreement between the titaniumdata for a and a, [Figure 3 ( 4 (a)] is expected. However, thedifferences between the curves for Fe, Mg, and Ti here (Figure 3)are not great, suggesting that the Ti has two sites (i.e. the cissites) in common with Fe and Mg rather than just one.To confirm this hypothesis, we compare the profiles ofrelevant x.p.d. curves with those for lepidolite.Because theordering pattern there is different, with Li and Mn occupyingboth cis and trans sites, unambiguous comparisons can only bemade with the b and 6, rotations, in which (uniquely) the two cissites are x.p.d.-equivalent. In these orientations, parallels shouldexist between the cis A1 in the lepidolite and any cis-segregatingion in the biotite. The most characteristic feature of cis A1 for theb rotation is that its peak intensity is markedly lower at 12” thanis that for Mn,Li [Figure 13(i), (e), cf: ( d ) ] , leading to a muchlarger peak in the Si/AI ratio than in Si/Mn,Li at this angle[Figure 12 (i), (c), cf: (d)]. Comparison with Figures 6 and 7shows that this is also a feature of the titanium data in relationto Fe and Mg.Similarly, for the b, rotation, a majorcharacteristic of A1 in lepidolite is a large peak at 53” resulting ina broad minimum in the Si/Al ratio at this angle (Figures 14 and15). This feature, too, is paralleled in the titanium data (Figures6 and 7 ) . The effects are less pronounced for Ti because of thelow k.e. of Ti 2p in comparison with A1 2s and 2p (cf: K and Rb,discussed earlier, for which the k.e. difference is smaller).We conclude that Ti is indeed segregated in the cis sites of thebiotite; at such a low concentration (1.2%) this would beimpossible to demonstrate with other techniques.ConclusionsImproved data-collection and analysis techniques, enabling theangular variations of individual X.P.S. peak intensities fromcomplex solids to be separated from the instrument responsefunction, have enabled a significant advance in understanding indetail the origin of the characteristic x.p.d.patterns obtainedfrom the micas. The existence of relatively long-range effects(> 11-13 A) has been demonstrated unambiguously for thefirst time, but most features of the data clearly result from short-range interactions.The value of x.p.d. methods for the detection andcharacterisation of cation ordering in 1M micas has also beendemonstrated. Study ofjust the two possible rotations (a and a,)about an axis parallel to the a unit-cell vector rapidly enablesdifferentiation of octahedral M(2) from M(3) to be detected, andshould also reveal any long-range tetrahedral ordering. Boththese phenomena involve a reduction in symmetry from C2/mto C2 and are thus relatively difficult to investigate byconventional crystallography. Comparison of x.p.d.patterns forrotation about the three b-group axes at 120” enables orderingwithin the three octahedral cation sites to be characterised.Rotation about the pseudo-a axes is helpful when cationssegregate equally in the two cis octahedral sites. X.p.d. methods,unlike conventional crystallography, can be applied to cationspresent in concentrations as low as 1%.In the 1M lepidolite previously studied by x.p.d.,6 the M(1),M(2), and M(3) sites are all differentiated. Lithium and Mnoccupy the M(l) trans site while A1 is concentrated in the cisM(3) site, as in several other recently studied trioctahedralmica^.^'-^^ The 1.2% of octahedral Ti in the previously studied1M biotite is shown to be segregated into the two equivalent cissites, whereas Fe and Mg are essentially randomly distributedamongst the three octahedral sites.Long-range tetrahedralordering does not occur in either micaJ. CHEM. SOC. DALTON TRANS. 1988 879AcknowledgementsWe thank the S.E.R.C. for support, and Drs. J. M. Adams, J. M.Maud, and D. E. Parry for valuable discussions.References1 K. Siegbahn, U. Gelius, H. Siegbahn, and E. Olson, Phys. Lett. A,1970, 32, 221; Phys. Scr., 1970, 1, 272; C. S. Fadley and S. A. L.Bergstrom, Phys. Lett. A, 1971, 35, 375.2 C. S. Fadley, Appl. Surf: Sci., 1985, 22, 193.3 J. M. Adams, S. Evans, P. I. Reid, J. M. Thomas, and M.J. Walters,Anal. Chem., 1977, 49, 2001.4 S. Evans, J. M. Adams, and J. M. Thomas, Philos. Trans. R. Soc.London, Ser. A, 1979, 292, 563.5 L. A. Ash and S. Evans, unpublished work; presented at the RSCPolar Solids Group Meeting, 15th December, 1986.6 S. Evans and E. Raftery, Clay Miner., 1982, 17, 443.7 S. Evans and E. Raftery, Clay Miner., 1980, 15, 209.8 S. Evans and E. Raftery, J. Chem. Res., 1982, ( S ) 170.9 C. S. Fadley, Prog. Solid State Chem., 1976, 11, 265; C. S. Fadley, in‘Progress in Surface Science,’ ed. S. G. Davison, Pergamon, NewYork, 1984, vol. 16, p. 275.10 S. Evans and E. Raftery, Solid Sfate Commun., 1980, 33, 1213.1 I D. E. Parry, paper presented at the Quantum Theory Conference,University of Exeter, September 1986.12 S. M. Goldberg, R. J. Baird, S. Kono, N. F. T. Hall, and C. S. Fadley,J. Electron Spectrosc. Relat. Phenom., 1980, 21, 1.13 W.F. Egelhoff, Phys. Rev. B, 1984, 30, 1052; R. A. Armstrong andW. F. Egelhoff, Surf Sci., 1985, 154, L225; E. L. Bullock and C. S.Fadley, Phys. Rev. B, 1985, 31, 1212.14 S. Evans, R. G. Pritchard, and J. M. Thomas, J. Phys. C, 1977, 10,2483.15 P. A. Lee, P. H. Citrin, P. Eisenberger, and B. M. Kincaid, Rev. Mod.Phys., 198 1, 53, 769.16 D. E. Parry, unpublished work.17 See, for example, S. W. Bailey, in ‘Crystal Structures of Clay Mineralsand their X-Ray Identification,’ eds. G. W. Brindley and G. Brown,Mineralogical Society, London, 1980, p. 1.18 A. Baronnet, in ‘Current Topics in Materials Science,’ ed. E. Kaldis,North-Holland, Amsterdam, 1980, vol. 5, p. 447.19 D. T. Clark and D. Shuttleworth, J. Electron Spectrosc. Relat. Phenom.,1979, 17, 15.20 S. Evans and D. A. Elliott, Surf: Znterface Anal., 1982, 4, 267.21 S. Evans and C. E. Riley, J. Chem. Soc., Faraday Trans. 2, 1986,541.22 S. Evans, Proc. R. SOC. London, Ser. A, 1978,360,427.23 J. M. Adams, S. Evans, and J. M. Thomas, J. Phys. C, 1973,6, L382.24 S. Evans and A. G. Hiorns, Surf: Interface Anal., 1986, 8, 71.25 D. A. Shirley, Phys. Rev. B, 1972, 5, 4709.26 A. Savitsky and M. J. E. Golay, Anal. Chem., 1964, 36, 1627.27 S. Guggenheim, Am. Mineral., 1981,66, 1221.28 S. W. Bailey, in ‘Reviews in Mineralogy,’ ed. S. W. Bailey,Mineralogical Society of America, Washington D. C., 1984, vol. 13, p.13.29 S. W. Bailey, Clays Clay Miner., 1984, 32, 81.30 See, for example, W. Joswig, H. Fuess, R. Rothbauer, Y. Takeuchi,and S. A. Mason, Am. Mineral., 1980, 65, 349.31 J. G. Thompson, Ph.D. Thesis, James Cook University of NorthQueensland, Australia, 1985.32 J. G. Thompson, Clay Miner., 1984, 19, 229.33 C. P. Herrero, J. Sanz, and J. M. Serratosa, Solid State Commun.,34 J. Sanz and J. M. Serratosa, J. Am. Chem. Soc., 1984, 106,4790.35 N. C. M. Alma, G. R. Hays, A. V. Samoson, and E. T. Lippmaa, Anal.Chem., 1984, 56, 729.36 M. Lipsicas, R. H. Raythatha, T. J. Pinnavaia, I. D. Johnson, R. F.Giese, P. M. Costanzo, and J-L. Robert, Nature (London), 1984,309,604.37 E. S. Crawford, S. Evans, E. Raftery, and M. D. Scott, Surf: InterfaceAnal., 1983, 5, 28.1985, 53, 151.Received 2nd July 1987; Paper 7166
ISSN:1477-9226
DOI:10.1039/DT9880000859
出版商:RSC
年代:1988
数据来源: RSC
|
|