it {I}... 1:51 ‘Ltfz W1 LIBRARY Michigan State University 1% ”Mess"! ||ll||ll|lll||ll|llllllve WWI HHHllH ll 3 1293 006 066 This is to certify that the dissertation entitled Transverse Layer Rigidity in Vermiculite Pillared by Mixed Alkylammonium Ions presented by Hyungrok Kim has been accepted towards fulfillment of the requirements for Ph.D. Chemistry degree in v/W [MM41ZW professor ‘F—‘m July 21 1989 Date MSU is an Affirmative Action/Equal Opportunity Institution 0-12771 PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE MSU Is An Affirmative Action/Equal Opportunity Institution TRANSVERSE LAYER RIGIDITY IN VERMICULITE PILLARED BY MIXED ALKYLAMMONIUM IONS BY Hyungrok Kim A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Chemistry 1989 @OAJZEE ABSTRACT TRANSVERSE LAYER RIGIDITY IN VERMICULITE PILLARED BY MIXED ALKYLAMMONIUM IONS BY Hyungrok Kim The mixed alkylammonium ion pillared clay system [(CH3 ) 4N+]X[ (CH3 ) 3NH+]1_X-vermiculites(x=O-O . 96) has been shown to be suitable for quantitatively describing the transverse layer rigidity of the host clay layers by observing the relationship between gallery height and the fraction of pillaring (CH3)4N+ ions occupying the gallery surfaces. The mixed ion [(CH3)4N+]X[(CH3)3NH+]1_X-vermiculites were prepared by the partial exchange of (CH3)3NH+ ions in [(CH3)3NH+]-vermiculite with the appropriate amounts of (CH3)4N+ ions. Very poor agreement. was obtained. between the X-ray diffraction peak positions of the observed and the calculated 001 reflections based on the Hendricks-Teller and Reynolds models for ion segregation. However, the observed patterns were in good agreement with the Bragg model for uniform ion mixing within the galleries. Also, a one- dimensional Patterson synthesis of the mixed ion system showed vector distributions corresponding to a unique d- Hyungrok Kim spacing. In addition, the experimentally observed x dependence of the normalized d-spacings (dn(x)=(d(x)- d(0))/(d(1)—d(0))) of the t(CH3)4N+JXI(CH3)3NH+11-X- vermiculites deviated strongly from the expected behavior for an ion-segregated system. Thus, the diffraction data are consistent with the random distribution of both ions within each. gallery, as opposed. to ion-demixed intercalates in which the two ions occupy separate galleries. A rapid rise in dn(x) with increasing x was observed near the threshold value of xt=0.2 and the maximum possible dn(x) value of 1.0 was observed near an x value of 0.6. A model which-related dn(x) to the layer rigidity parameter p, the binding energy difference between the gallery sites and the non-gallery(defect) sites, and the ratio of the two sites yielded an excellent fit to the experimental dn(x). A layer rigidity parameter p=8, a binding energy difference of 2.5 Kcal/mole, and, the defect, site ratio of 2K3 % were determined for the vermiculite host layers. ACKNOWLEDGEMENTS I would like to express sincere thanks to Professor T. J. Pinnavaia for his great personality, guidances, and encouragement throughout the short science journey. I also want to express thanks to my whole family, specially to Hyoran for their patience and love. The kindness from the people of U. S. A. and the financial supports from NSF are very appreciated. iv TABLE OF CONTENTS Page LIST OF TABLES ................ ........ ................ vii LIST OF FIGURES ....................................... viii I. INTRODUCTION ....................................... I I-1. Structure and Properties of 2:1 Type Clay Minerals 0....0......0...OOOOOOOOOOOOOOOOOOOOOO 1 I-2. Pillared Clays as Two-dimensional Microporous Material ...................................... 6 I-3. Transverse Layer Rigidity and Access in Pillared Clays ................................ 10 I-4. Approach to Characterizing the Transverse Rigidity of Clay Layers ....................... 12 I-S. Defining the Mixed-Ion Clay System ............ 17 II. EXPERIMENTAL ...................................... 18 II-l. Materials .................................... 18 11-2. Synthesis .................................... 18 II-2-1. [(CH3)3NH+]-Vermiculite ................. 18 11-2-2. [(CH3)4N+]-Vermiculite .................. 19 II-2-3. Mixed Ion [(CH3)4N+ ]X[(CH3)3NH+ 11-x Vermiculites ............................ 20 11-3. Characterization ...... ..... .................. 20 11-3-1. Exchange Ion Compositions of [(CH3)4N+ 1x [(CH3)3NH+ 11-x - vemiCUJ-ites 00000000000 20 II-3-2. X-ray Diffraction Patterns ......... ..... 21 II-3-3. Modeling X-ray Diffraction Patterns ..... 22 II-3-4. MacEwan Direct Fourier Transform of XRDPatterns OOOOOOOOOOOOOIOOOOOOOO00.... 22 Page 11-3-5. Infrared Spectroscopy ................... 23 11-3-6. Raman Spectroscopy ...................... 24 II-3-7. Chemical Analysis ....................... 24 II— 4. Modeling the Composition Dependence of Galler Height for [(CH3)4N+ 1x [(CH3)3NH ]1-x-Vermiculite ................... 25 III. RESULTS ........................ ..... ............. 27 III-1. Synthesis ................................... 27 III-2. X-ray Diffraction ........................... 31 III-3. Generation of dn(x) vs x Plots for Mixed Ion Monolayer Model ......................... 44 III-4. Infrared+ and Raman Spectroscopies of Mixed IV. DISCUSSION ........................................ 53 V. CONCLUSIONS ............ ....... .... ....... .... ..... 66 APPENDIX I ............................................ 67 APPENDIX II ........................................... 73 APPENDIX III .......................................... 76 REFERENCES 0.0..0.00000000000000000000000000000000000.0 78 vi Table LIST OF TABLES Page Idealized Structual Formulae for Representative 2:1 Phyllosilicates. In each formula the bracket and the parentheses define metal ions in octahedral and tetrahedral sites, respectively ................ 5 Compositions of Representative [Me4N+ ] fi[MegNI-I': 11-x' Vermiculites Obtained by Reaction of [ Ne3N Vermiculite with Aqueous Me4NC1 ................... 32 vii LIST OF FIGURES Figure Page 1 Schematic illustration of the tetrahedral and octahedral sheets in a 2:1 layered silicate layer. open circle: oxygen shade circle: hydroxyl group small closed circle: tetrahedral site cation Si,Al large closed circle: octahedral site cation A1,Mg, Fe, Li, vacant .............. 2 (a) The idealized Kagome net defined by the 3- membered and 6-membered oxygen rings in the basal surfaces of a 2:1 layer lattice silicate. The primitive cell (---) and lattice parameter, a=5.34 A of an undistorted lattice are also shown. (b) Distortions in the Kagome net due to in-plane rotations of the Si(Al)O4 tetrahedra .............. 4 (a) Schematic representation of the pillaring of clay. O and P are simple and robust cations, respectively. (b) Typical types of pillaring agents: tetraalkyl- ammonium; bicyclic amine cations: polyhydroxyl or oxo cations; metal chelates ......... ........... 8 Schematic representation of the layers of a pillared clay assuming (a) Rigid layers (b) Flexible layers ........ ..... ... ....... . ...... 11 (a) The concept of probing layer distortions by increasing lateral pillar separation. (b) The concept of probing layer distortions by mixing large and small ions in a mixed ion clay .................................. 13 Schematic representation of the expected relationship between dn and x in [A]X[B]1_x -Y. (a) Infinitely rigid layers (b) Infinitely flexible layers ............. ...... . 15 Plots for comparing of ion exchange reactivity of powdered +Mg2 +-vermiculite with (CH3 )3 NH+ and (CH3 )4 N+ in the absence of EDTA anion. After each 24 hour reaction period, the ion viii Figure Page exchange solution was replenished with a fresh solution containing an excess of alkylammonium ion. The ratio of alkylammonium ion : Mgz+ was 54:1 for the first four reaction periods, and 108:1 for the last two reaction periods. .................... 29 8 X-ray diffraction patterns for (CH3)3NH+-vermiculite as (a) freshly prepared wet gel, (b) a film dried 1 hr. in air, and (c) an oriented film dried 1 hr at 100 °C. All diffraction patterns were recorded under equivalent conditions of x-ray exposure ..... 33 9 X-ray diffraction patterns for oriented film samples of representative [8CH3)4N+ ]x[(CH3)3NH+ 11-x' vermiCUlites heated at 100 ....OOOOOOOOOOOOOOOOO 10 Comparison of the observed and calculated xrd pattern for homoionic (CH3)3NH+-vermiculite ....... 36 11 Comparison of the observed and calculated xrd pattern for homoionic (CH3)4N+ -vermiculite ........ 37 12 Comparisons of observed and calculated xrd patterns+ based on three +models for [(CH3)4N+ ]o 63[(CH3)3NH+ )0 37 " vermiculite BRG: Bragg model. H-T: Hendricks-Teller model REY: Reynolds model, EXP: Experimental pattern .... 39 13 Direct MacEwan transform of xrd pattern of [(CH3)4N+ ]0. 63[(CH3)3NH+ )0. 37 - vermiculite ....... 40 14 Representative Q vs 1 plots for [(CH3)4N+]x [(CH3)3NH+ 11-x -vermiculites. The broken l1nes are the linear least squares fit to the data pOints O... ...... 00...... ..... OOOOOOOOOOOOOOOO 42 15 The composition dependence of the +normalized basal spacings for [(CH3)4N+ ]x[(CH3)3NH+ 11-x -vermiculites. Solid line is the least square fit to xthe data using Equation 7 of the text with the following parameter set: p=8.0; f=0.5: A/kT=4.3 .... ..... ... 43 16 The composition dependence of the normalized basal spacing for ion segregation in [(CH3 )4 N+]X- [(CH3 )3NH+]1_X -vermiculites based on 4the XReynolds model. 3The basal spacings were calculated from least square regression of the first six 001 reflections ............................. .......... 45 ix Figure Page 17 Monolayer triangular lattice simulations (dotted lines) of d n(x) for idealized [A] [B]- vermiculites for several differeng va ues of the healing length., 1, and the layer rigidity parameter, p. The solid lines are from Equation 4 of the text with (1) p=1, A=Or (2) p=7,1=ao; (3) p=13, A=V3 a0; p=°° . l=°° Inset: the triangular lattice defined by the spherical A and B pillaring cations. The expanded region represents the distortion which occurs in the host layer when A replaced B and x=a0. In this latter case, the number of expanded sites is p=n+1=7 where n is the number of nearest neighbors......... 47 18 IR spectra of (CH )3 NH+ -vermicu1ite with different incident beam ang es. Note the change in relative peak intensity of the N+ -H stretching at 2730 cm 1 relative to other angle-independent peaks ......... 48 19 IR spectra of [(CH )4N+ ] 54[(CH3)3NH+ 10 46‘ vermiculite for differen 1ncident beam angles. Note the change in relative peak intensity of the N+-H stretching at 2730 cm’1 relative to other angle-independent peaks ...................... ..... 49 20 Low frequency Raman spectra of [(CH3 )4N+ ]X [(CH3 ) NH+ ] -x' vermiculites. The 1nsert xidenti ies ihe eigenvector of the vibration ....... ................ ............ ...... 51 21 X-dependence of the torsional mode frequency for [(CH3 ) N+ ] [(CH )3NH+ 11-x' vermiculites. The error bar 1n ica es t e uncertainty in the frequency measurements ...................................... 52 22 Schematic illustration of possible non-gallery defect sites for exchange ions on the basal surfaces (designated x) which do not contribute to the 001 basal spacing (a).Surfaces at twisted layer-planar layer interfaces, (b).Surfaces of bifuricated layers, (c).Surfaces of internal and external ledges and crevasses formed by layer termination, (d).Surfaces of layer edge-layer face interfaces. Defects (a)-(d) are in addition to non-gallery binding sites at layer edges(e) .................................... 61 Figure 23 24 25 26 Page Plot of x the dependence of gallery ion composition ) on the total (CH )4 N+ composition (x) for wgfl3)4NH ]X[(CH3)3NH ]3_ I -vermiculites. The two bin 1ng site mo e assumes p=8, f= 0.5,and A/kT=403 OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO 63 26 dependence of the LP factor, squared structure factor, interference function, and intensity in the xrd pattern calculation of (CH3)3NH+ -vermiculite ...00......0.0.000...OI0..........OOOOOOOOOOOOOOOOO 70 Comparison of observed and calculated xrd patterns for [(CH3)4N+ 10. 63[(CH3)3NH+ 10. 37 - vermiculite usingx =0.45 H-T : Hendricks-TelIer calculation EXP.: Observed xrd pattern ..... ................... 74 Comparison of observed and calculated xrd patterns for [(CH) N+ ]0 63[(CH3)3NH+ 30. 37 - vermiculite using x =3. 35 REY : Reynolds calc lation EXP : Observed xrd pattern ............ . ........... 75 xi I. INTRODUCTION I-1. Structure and Properties of 2:1 Type Clay Minerals Layered alumino-silicates minerals, generally known as clay minerals, have different forms that vary from amorphous powders to single crystals. They also show broad variations in chemical and physical properties.1I2 Even among clays with almost same layer lattice structures where the basic sub-layer building blocks are almost identical, their properties can vary' greatly. These two-dimensional layer structures consist of sheets of M04 tetrahedra and a sheet of M'O6 octahedra where M is most commonly Si4+ and, occasionally, Al3+ or Fe3+, and M’ is A13+, Fe3+, Mg2+, Li+ or a vacancy. Also, some of the oxygens of the M’O6 are replaced by hydroxy groups. Thus, the various clay layers may be considered as condensed products of the two different kinds of sheets.3 The 2:1 layer lattice structures of interest in this research contain two tetrahedral sheets fused to a central octahedral sheet.4 Fig. 1 illustrates the idealized 2:1 structure along with the interlayer cations. For simplicity, metal cations in both the tetrahedral sites and the octahedral sites of layers are shown using solid circles. The surfaces of the unit layer are composed of basal oxygen Oxygen plane Tetraheral sheet Octahedral sheet Tetraheral ‘ sheet Fig. 1 Schematic illustration of the tetrahedral and octahedral sheets in a 2:1 layered silicate layer. open circle: oxygen shade circle: hydroxyl group small closed circle: tetrahedral site cation Si, Al large closed circle: octahedral site cation Al,Mg, Fe, Li, vacant planes which define a Kagome net containing 3-membered and 6-membered oxygen rings. This net is described Fig. 2(a). Ideally, 6-oxygen rings form sites of hexagonal symmetry. In a real layer, tetrahedral rotations occur around the Si(Al)- 0 .bonds perpendicular to the basal plane, because of aa misfit between the b parameters of the individual tetrahedral and octahedral sheets. The directions of the tetrahedral Si(Al)O4 rotations are indicated by the arrows in Fig. 2(b). The rotation of the tetrahedral sites changes the cavity symmetry from hexagonal toward trigonal,l but the triangular lattice symmetry which is defined by the centers of 6-membered oxygen cavities is not affected by the rotation. The thickness of a layer is about 9.6 A. In the unit cell of 2:1-type clay minerals, twenty oxygen atoms and four hydroxy groups form a 44 e‘ anion framework which defines six octaheral sites and eight tetrahedral sites for occupancy by layer cations, along with four 6-membered. oxygen. cavities.1 The idealized. chemical formulae of various kinds of 2:1-type clay minerals are summarized in Table l. The terms "dioctahedral" and "trioctahedral" mean that two-thirds and three-thirds of the octahedral sites are occupied by metal cations, respectively. As shown in Table 1, a large portion of 2:1 type clay layers are negatively charged. These negative charges are balanced by interlayer cations. Depending on the amount of layer charge, 2:1-type clay minerals can be divided into 5 groups: the talc-pyrophyllite, smectite, Fig. 2 YVXX\\ \\/XXXXV \w/VVV (a) YXXXYY 4 Mica Muscovite: Phlogopite : K31Al41(3i6 Al2)020(OH)4 K5[M951(Sie A|2)020(OH)4 Brittle Mica Margarite : Clintonite : 2+ . Ca22iAl4](Si4 ,c\14)020(ol-l)4 Caz[Mg4Al2](Slels)Ozo(OH)4 vermiculite, mica, and brittle mica groups. In the talc- pyrophyllite group, the layers are electrically neutral and coupled through van der Waals forces.5 In the other extreme, mica and brittle mica minerals possess 2 e" and 4 e' layer charges per 020(OH)4 unit, respectively, and the counter cations, usually K+ and Ca2+, are tightly bound in the ditrigonal oxygen cavities of the. interlayer surfaces. Therefore, no significant intercalation chemistry based on cation exchange reactions is observed for both groups. On the other hand, in the smectite and vermiculite groups of clays with intermediate layer charge densities, diverse intercalation chemistry has been established due to their the appreciable swelling and cation exchange properties.6 I-2. Pillared Clays as Two-dimensional Microporous Material Two different types of intercalated clays are distinguished from the point of View of gallery(interlayer region) environments. One group is the simple hydrated metal cation exchange forms (Mn+, n=l,2), such as the smectite group clay’ minerals, typically found in nature. In these exchange forms, the interlayers collapse at high temperature (above 200 OC) due to the loss of water. This kind of system requires a swelling agent to provide access of other species to the interlayer region. The second type of intercalated clay is the pillared clays in which robust cations act as props to prevent interlayer collapse even in the absence of a swelling agent. Pillaring means here the creations of accessible micropores in the interlayer regions of the clay by chemical and physical processes based on the cation exchange property of the 2:1 type clays. Pillaring is shown schematically in Fig. 3(a).7 The first pillaring of 2:1 type clays was demonstrated by Barrer and MacLoed8 by utilizing tetraalkylammonium ions as pillaring species and a smectite clay as the layered host. Several variations in both the pillaring species and layer types have been tried.9 Among the typical pillaring agents illustrated in Fig. 3(b), organic-containing molecular pillaring species, such as alkylammonium cations and metal chelate compounds, have been used greatly because of big advantages in defining the shapes and sizes of pillars in the clay galleries.10, There has been increasing interest in the catalytic properties of pillared clays.ll Various types of inorganic pillaring species have been examined in order to achieve higher thermal stability and larger pore sizes. Al,12 Zr,13 Fe,l4 Cr,15 and Ti polynuclear hydroxy-oxo cationl6 intercalation reactions afford more thermally stable and catalytically active pillared clays. Usually, pillared clays derived from inorganic cationic species are stable above 500 OC and exhibit large surface areas (BET: 200—500 mz/g) and large pore volumes ( 0.1-0.2 cm3/g) with relatively large pore diameters ( 8-20 A). The distributions of metal oxide Fig. 3 (a) (b) (a (a (9 l — ] fl Pillaring H '4- Y N+ R/l\R [,1] R l+ H “II"' MHOHWn+ ‘A. M (chel)3 m“ (b) Schematic representation of the pillaring of clay. eiand P+ are simple and robust cations, respectively. Typical types of pillaring agents: tetraalkyl~ ammonium: bicyclic amine cations: polyhydroxyl or oxo cations: metal chelates. 9 pillars and pores in the interlayer regions of metal hydroxy lO cation pillared clays can be highly regular, even though the layer charge densities of smectites vary as much as a factor of 2 from layer to layer.17 The pores created by pillaring reactions provide chemists with a constrained reaction environment and a system of interconnecting pores for conducting chemical conversions. Chemical reactions within pores may be different from those on bulk surfaces due to differences in kinetic and shape-selective effects.18 For pores of less than 40 A diameter, theoretical calculations based on kinetic theories19 show that great reactivity increases, as much as four orders of magnitude, as well as selectivity increases of 1 or 2 orders of magnitude, can be realized for non-diffusion controlled chemical reactions when the walls of the pores are involved as catalytically active sites. High temperature reactions, such as oil cracking on zeolites or acidic pillared clays can be good systems for testing these small pore effect predictions. Another important point regarding pillared clays is the high possibility of forming large pores (above 20 A in diameter) formation which may be very useful for the diffusion of large molecular species into active surface sites.10 It is well-known that the pore size of pillared clays can be increased simply by increasing the size and charge of the pillars or by the decreasing the charge density of the clay layers.lo lO I-3. Transverse Layer Rigidity and Access in Pillared Clays with regard to the two-dimensional molecular sieving and catalytic applications of pillared clays, one of the very important factors to be considered is the interlayer access phenomena of the species to be catalyzed or to be absorbed. The interlayer access in the pillared clay system is controlled by the size of pores and the size and shape of the species to be accessed. As depicted in Fig. 4(a), the pore size of a pillared clay may be defined by two factors: d1, the pillar height and d2, the lateral separations between pillars. However, this description is accurate only if the layers are rigid against transverse distortions, such as sagging due to the interlayer interactions or other external forces. In the case of flexible layers, which are schematically shown in Fig. 4(b), the pore size is not defined well by the above two parameters. Thus, the shortest vertical distance between layers dl’ is expected to be smaller than d1. In spite of the successful pillaring of 2:1 type clays, which proves that the layers are reasonably rigid, transverse distortions of 2:1 clay layers in the range of 102 A have been directly observed by using electron microscopy. Bending of vermiculite layers and folding’ of smectite layers have been reported.20 Definitely, the 2:1 clay layers have finite rigidity.21 ll Fig. 4 Schematic representation of the layers of a pillared clay assuming (a) Rigid layers (b) Flexible layers 12 Therefore, it is very important to describe the rigidity of the 2:1 clay layers in order to obtain a better understanding of access phenomena in pillared clays. This layer rigidity argument is also relevant to the design of large pores in pillared clays. In principle, by increasing the lateral distances and pillar height, the pore size may be increased. However, it is expected that gallery collapse due to interlayer interactions also will increase with increasing lateral separation of the pillars. Thus, the extent of gallery collapse will depend on the transverse layer rigidity of the host layer. Here, the term "transverse layer rigidity" refers to the tendency of a layer to resist transverse (out-of-plane) distortions. In designing a large-pore pillared clay system, one should be aware of the importance of the transverse layer rigidity of the host layers. I-4. Approach to Characterizing the Transverse Rigidity of Clay Layers As described previously, the extent of layer distortion in a pillared clay should be a function of the lateral separation between pillars. Therefore, measuring the average basal spacing as a function of pillar density, as shown in Fig. 5(a), would be a ideal way of probing layer rigidity. However, varying the pillar density would require varying l3 j (b) Gradual increase of lateral distances between pillars induction of layer distortion by mixing cations Gradual increase of lateral distances between larger pillars Fig. 5 (a) The concept of probing layer distortions by increasing lateral pillar separation. (b) The concept of probing layer distortions by mixing large and small ions in a mixed ion clay 14 either the layer charge on the clay or the charge on the pillars. Neither of these approaches, however, are plausible over suitably broad ranges of pillar densities. Another more promising route to probing layer rigidity is to induce layer distortions by mixing two cations of different size in the same clay gallery. The term "mixed ion clay" refers to a clay system in which two cations of different kinds are randomly distributed in the same interlayer region.22 The mixed ion clay system is represented by the general formula of [A]X[B]1_X-Y, where "A" and "B" are interlayer cations and "Y" indicates the host clay system. In the above formula, the size of the "A" cation is larger than that of the "B" cation by definition. In a mixed ion clay, the lateral distances between larger cations can be changed almost continuously from the limiting value of homoionic [A]-Y, to infinity by changing the ratios of two different cations.( see Fig. 5(b)) Since the magnitude of the layer distortion should be small compared to the size of the layer, one dimensional X- ray diffraction from stacked clay layer samples could be used to detect layer distortions which change in the repeat distance of the layers along the c axis.23 Fig.6 illustrates the expected behavior for two extreme caSes of mixed cation pillared clays: the infinitely rigid limit (Fig. 6a) and the infinitely floppy limit (Fig. 6b). The expected normalized basal spacings (dn), relative responds of repeating distances in mixed ion clays, defined L0 Fig. 6 15 L 3 O 0 O L :J 0 O O L_ i3 0 0 O F fl :- ------------------ « 1.0 - x' -: an — : 1 0 x 1 l 0 LO 0 L0 x x (a) (b) Schematic representation of the expected relationship between dn and x in [A]X[B]1_X -Y. (a) Infinitely rigid layers (b) Infinitely flexible layers 16 as dn(x)=(d(x)-d(0))/(d(1)-d(0)) where d(x) is the d-spacing of the mixed ion clay at a concentration x of the larger cation, are also plotted in the lower part of the figure. For the infinitely rigid layer case (a), dn reaches 1 at a very low concentration of the "A" cation, ideally, corresponding to only three "A" cations per gallery. For the infinitely floppy limit, the situation is totally different. In this case, the layers fluctuate locally in accord with the cation size and this results in a linear relationship between dn(x) and x. Since the real 2:1-type clays with finite layer rigidities do not belong to either of the two extreme cases, it is expected that the shape of the dn(x) vs. x plot will be intermediate between these of two extreme cases. Based on the above considerations of the behavior of dn(x) vs. x, it may be concluded that layer rigidity information can be extracted from dn(x) vs. x plot, provided that an appropriate model can be developed. In order to experimentally quantify the transverse distortions of 2:1 clay layers, one must achieve the following goals: (1) the synthesis and characterization of a series of mixed-ion clays, (ii) determination of the dn(x) vs. x plot for the mixed-ion clay system, and (iii) establish a suitable model for the simulation of the dn(x) vs. x dependence and extract a layer rigidity parameter from the experimental results. l7 I—S. Defining the Mixed-Ion Clay System Llano vermiculite, which has layer charge density as high as the natural micas,24 served as the clay host for this research. The high charge density is preferred for the simulation experiments, because the symmetry and locations of the intercalated cations can be reasonably deduced. As indicated in the early part of the introduction, the location and symmetry of cations in micas are well-defined. It can be assumed that the distribution of cations in Llano vermiculite is almost the same as that of micas. Trimethylammonium and tetramethylammonium cations were utilized in synthesizing the mixed cation clays, because of the suitable swelling properties of trimethylammonium and tetramethylammonium exchange forms of Llano vermiculite. II. EXPERIMENTAL II-l. Materials Tetramethylammonium. chloride and trimethylamine were purchased from Aldrich Chemical Co. and used without further purification. Single crystals of Llano vermiculite were obtained from the Source Clay Repository at the University of Missouri, Rallah, Missouri. The crystals were ground in a mortar and pestle to a fine powder (-200 mesh). Then, the powdered form of the vermiculite was treated with 2.0 M MgClz solution for one day at room temperature to ensure complete saturation of the exchange sites by Mg2+. After the separation of the clay from suspension by repeated centrifugations and decantations, the Mg-vermiculite was dried in air for at least two days. The qualityi of vermiculite was sufficient to prepare all of the mixed ion derivatives used in this study. II-2. Synthesis II-2-1. [(CH3)3NH+]-Vermiculite 18 19 The trimethylammonium exchange form of vermiculite was obtained by ion exchange of the pristine mineral in the presence of EDTA anion as a complexant for magnesium ion. A solution containing 12.5 wt% trimethylamine and 2.7 wt% ethylenediaminetetraacetic acid was adjusted to pH=7.0 by the slow addition of concentrated HCl. This solution was mixed with powdered magnesium vermiculite in the ratio 1.0 g of clay per 200 ml of exchange solution, corresponding to a Mg2+z (CH3)3NH+: EDTA molar ratio of 1:560:50. The reaction mixture was allowed to stir vigorously for one day at room temperature. The [(CH3)3NH+]-V (V: vermiculite) was recovered by five cycles of centrifugation and washing with 200 ml quantities of distilled water. The x-ray diffraction pattern of an oriented film sample dried at 100°C exhibited a basal spacing of 12.70 A. II-2-2. [(CH3)4N+]-Vermiculite This derivative was prepared from [(CH3)3NH+]-V by ion exchange reaction with the (CH3)4N+ cation. To a 2.5 M solution of (CH3)4NC1, an equal volume of 1.0 wt% of [(CH3)3NH+]-V suspension was added at room temperature. After 24 hours of stirring at room temperature, the [(CH3)4N+]-V was washed by successive centrifugation(15,000 rev/min), decantation and redispersion into distilled water. Analysis for 'trimethylammonium. ion (see below) indicated 20 that the exchange reaction was 96% complete. An oriented film sample dried at 100 OC exhibited an x-ray basal spacing of 13.34 A. 11—2-3. Mixed Ion [(CH3)4N+]X[(CH3)3NH+]1_X Vermiculites Mixed ion exchange forms of [(CH3)4N+]X[(CH3)3NH+]1_X-V were prepared by the reaction of [(CH3)3NH+]-V suspensions with (CH3)4NC1 solutions at various (CH3)4N+ molar ratios. The concentration of the (CH3)4N+ was in the range of 0.0024-2.500 M. After allowing the reaction mixtures to be stirred for designated reaction times at room temperature, the mixed-ion vermiculites were collected by using the same isolation procedures as described previously. II-3. Characterization II-3-1. Exchange Ion Compositions of [(CH3)4N+]X[(CH3)3NH+]1_X - Vermiculites The exchange ion compositions of [(CH3)4N+]X [(CH3)3NH+]l_X —vermiculites were determined from the amount of trimethylamine released upon dispersing a fixed amount of the mixed—ion clay in 1:1 methanol water(volume ratio) mixture containing 0.05 M NaOH. The liberated trimethylamine was slowly transferred by distillation into a 21 titration flask equipped with a pH electrode and a buret containing standard HCl solution. During the distillation process, the pH of the solution in the titration flask was kept just below a value of 7 by the periodic addition of a standard HCl solution. The amounts of HCl consumed for the mixed ion clays were compared directly with. the amount needed to neutralize the trimethylamine liberated from an equivalent quantity of [(CH3)3NH+]-vermiculite. II-3-2. X-ray Diffraction Patterns Oriented film samples for X-ray diffraction analysis were prepared by drying approximately 1 ml of a 1 wt% clay suspension on glass slides in air. The films were then dried in an oven at 100 OC. The mosaic spreads of the small clay platelets in the films were 5-10 degrees.25 No swelling by water was observed upon exposure of the heated samples to the atmosphere for prolonged periods of several months. Diffraction patterns were obtained using Ni-filtered Cu-Ka radiation. The basal spacings were determined from the slopes of Q vs. 1 plots where Q is the momentum transfer, (41r sin 9/1 = 21rl/d) and l is the diffraction order. The slopes of the Q vs. 1 plots were determined by weighted linear-least square regression analysis.26 22 II-3-3. Modeling X-ray Diffraction Patterns In order to interpret the xrd patterns of samples in terms of gallery ion distributions, one-dimensional xrd patterns were generated based on the three known models: Bragg, Hendricks-Teller,27 and Reynolds models.28 In the Bragg' model, the homogeneous mixing' of two different kinds of cations in the same gallery is assumed. However, in the two other models, the segregation of two different kinds of cations into different galleries is assumed and modeled by appropriate mathematical treatments. Details of the treatment of the xrd pattern calculations are summarized. in Appendix I. Layer atomic coordinates were taken from the single crystal structure determination of vermiculite.29 The X-ray scattering factors of the clay layer atoms were taken from the values provided by Wright.30 II-3-4. MacEwan Direct Fourier Transform of XRD Patterns In order to determine the sequences of layer stacking, diffraction patterns were transformed into the function W(R)28, which is the probability of finding a given layer to layer distance R. In this equation (see Eq. 1), Is is the diffraction intensity at angle theta, LP the Lorentz- polarization factor, {Glz the squared magnitude of the layer factor, Q the momentum transfer, and R a distance in unit 23 of A. This transform is a modification of the Patterson function3l, which is used to obtain the vectorial separation of atoms within an unit cell. In the calculation, the sum was taken over the continuous diffraction profile and the structure factor of trimethylammonium vermiculite was used. Is W(R)= 23—— oos(RO) 1 e LPlGlZ ( ) II-3-5. Infrared Spectroscopy Infrared spectroscopy was used mainly to obtain information. on the orientation. of the smaller trimethylammonium ions in the galleries. The absorptions of the N+-H stretching frequency at about 2730 CM"1 were compared with changes in "the angle of incidence" between the surface of the self-supported clay film and the incident infrared beam. An IBM FT-IR spectrometer model 403 was used to obtain the spectra. 24 II-3—6. Raman Spectroscopy Laser Raman spectroscopy was used to examine the mixed- ion clays. The in-plane intralayer torsional mode of tetrahedral alumino-silicate sheets32 was examined to determine the effect of the cation mixing on this vibrational mode. Raman spectra were obtained using the 5145 A line of an argon laser at a typical power level of 150 mW. All measurements were made with the scattered light collected at 90 degrees to the direction of propagation of the laser light, which was incident at an angle of about 45 degrees to the sample plane. The experiments were undertaken at room temperature. II-3-7. Chemical Analysis Chemical analysis on independent samples of Llano vermiculite were accomplished using ICP atomic emission spectroscopy. The NBS 98-a plastic clay served as a standard clay for the analysis. With regard to the preparation of the sample solutions, about 30 mg of Mg-vermiculite was mixed with 0.3 g of LiBOZ in a graphite crucible, and the mixture was fused in a furnace at 1000 0C for 10 minutes. The molten material was poured into 30 ml of 10 % HNO3 and the solution was stirred 25 to achieve complete dissolution. The elements Mg, Al, and Si were analyzed. II-4. Modeling the Composition Dependence of Gallery Height for [(CH3)4N+]X[(CH3)3NH+]1_X - Vermiculite To help us understand the behavior of dn(x) on x in mixed ion [A]X[B]l_X-clays, we generated, in collaboration with Professor Mahanti and Solin and their students, dn(x) vs. x data for mixed ion [A]X[B]1_X-vermiculite model systems containing monolayers of pillaring ions between clay layers with finite transverse rigidity. For simplicity, we assumed that the pillaring ions are rigid spheres positioned at the centers of the hexagonal cavities defined by the Kagome lattice of the gallery surfaces. These pillar positions define a two-dimensional triangular lattice of lattice constant a0= 5.34 A. Starting with a single gallery in which each pillar position is occupied by a smaller "B" ion of height dB, we randomly’ replaced the "B" ions ‘with larger‘ "A" ions of height dA° The height of the gallery within a healing length is assumed to increase from dB to dA upon replacement of "B" by "A". Here, the healing length is a distance from the layer position of a "A" cation which is surrounded by only "B" cations to the layer positions at which the layer returns to d(0). A second "A" ion within the healing length 26 of a first "A" ion does not affect the already expanded unit cells, Inn: it expands unexpanded cells within its healing length . The process of random replacing the "B" ions continues until x=1. The dn(x) values are determined by the fraction of expanded triangular cells. III. RESULTS III-1. Synthesis Naturally occurring single crystals of Mg2+- vermiculite (Llano, Texas) served as the starting mineral for the synthesis of our mixed alkylammonium ion clays. The results of independent chemical analysis for Mg, Al, and Si indicated the unit cell formula to be: MgO.862+(SiS.87A12.13)[A10.48M95.521020(OH)4° In order to enhance the ion exchange reactivity of the parent Mg- vermiculite, the particle size was reduced by wet grinding in water to -200 mesh sizes. The reaction of the powdered Mg2+-vermiculite with excess (CH3)3NH+ and EDTA anion as a complexant of Mg2+ afforded [(CH3)3NH+]-vermiculite in quantitative yield as expressed in Equation 2, wherein the horizontal lines represent the clay layers. EDTA” a- : 2(CH3)3NH+ + MgEoTA‘" ’ (2, swollen Mg 2* + 2(CH3)3NH* The presence of the EDTA anion was essential for complete exchange. In the absence of the complexant, 27 28 complete exchange could not be achieved even after six reaction cycles of the Mg2+-vermiculite with (CH3)3NH+ solutions under the forcing reaction condition defined in Fig. 7. The complete replacement of hydrated magnesium ions by the trimethylammonium cations in the presence of the EDTA anion results in the swelling of the interlayer region by water. The degree of swelling was very extensive, as evidenced by gel formation at concentrations of 10 wt% clay and 90 wt% of water. The swelling phenomenon and associated layer exfoliation most likely leads to the further reduction of the clayplatelet size. Thus, the final [(CH3)3NH+]- vermiculite product is no longer in the form of single crystal particles. Instead, the clay layers are aggregated as turbostratic tactoids with regular stacking in the direction perpendicular to the layer but random in-plane orientations. Analogous layer aggregation mechanisms have been found for smectite clays with 2:1 layer lattice structures.33 The reaction of Mg2+-vermiculite with tetramethyl- ammonium ion in aqueous solution under conditions analogous to those used to prepare the trimethylammonium derivative (Mg2+:Me4N+:EDTA=1:560:50) yielded only partial replacement of the interlayer magnesium ions by tetramethylammonium ions because of the precipitation of the EDTA anion by tetramethylammonium cation. The exchange reaction with tetramethylammonium in the absence of EDTA anion also showed 29 60* - -60 ‘0 Q) 0 C E 40 g (CH:):NH* ‘40 I.1J .: ‘ - 0“ 2 20 R ~ ~20 OJ ..0 f I I 0 1 2 3 l 5 is Number of 24-hr Reaction Cycles < ————— 25°C ---—> <---- 55°C -—--> < ------- 3.2 M ------- > <-- 6.4 M —-> Fig. 7 Plots for comparing of ion exchange reactivity of powdered Mg2+-vermiculite with (CH )3 NH+ and (CH3 )4N+ in the absence of EDTA an1on. After each 24 hour reaction period, the ion exchange solution was replenished with a fresh solution containing an excess of alkylammonium ion. The ratio of alkylammonium ion : Mg2+ was 54:1 for the first four reaction periods, and 108:1 for the last two reaction periods. 30 a very slow exchange rate compared to that of the trimethylammonium case, as shown in Fig. 7. The slow migration of tetramethylammonium ions may be due to their lack of hydration within the clay galleries. The differences in the solvation properties of the trimethylammonium and tetramethylammonium cations in the galleries of the host clay provided the key to the synthesis of the mixed ion derivatives. The addition of controlled amounts of tetramethylammonium ion to [(CH3)3NH+]- vermiculite suspensions resulted in the flocculation of the suspension and the concomitant replacement of some (CH3)3NH+ ions by the desired (CH3)4N+ ions of larger size. Because flocculation is rapid, the segregation of the exchange ions into separate galleries is impeded. Consequently, the distribution of the two ions within the galleries is under kinetic control rather than thermodynamic control. Equation 3 summarizes the overall reaction for mixed ion synthesis. (SENT + x'lCHa)4N*—+ [(CH3)4N+]x[(CHalaNH+]1-x swollen (3) + x (CH3)3NH+ + (x'-x) (CH3)4N+ The chemical compositions of the mixed ion products were determined by displacing the alkylammonium ions with sodium cation as shown in Equation 4, and subsequent titration of the liberated trimethylamine with standard HCl solution. 31 [(CH3)4N*L.[(CH3>3NH*11-x Ali—Oi N75- . (1-x) (CH3)3N (4) + x(CH3)4N+ + (1-x) H20 Table 2 provides the compositions of some representative mixed-ion products obtained by the reactions of [(CH3)3NH+]-vermiculite with (CH3)4NC1 according to Equation 3. It is noteworthy that the (CH3)3NH+ ions are easily displaced by the (CH3)4N+ ions at low values of x; In order to achieve mixed ion products with high values of x, however, a large excess of tetramethylammonium ion was necessary . III-2. X-ray Diffraction Fig. 8 illustrates the diffraction patterns for (CH3)3N+-vermiculite as a freshly prepared wet gel, an air- dried oriented film, and an oriented film dried at 100 OC. The gel form of the sample is essentially X-ray amorphous. Only very weak reflections by a small non-clay impurity phase were detected. This means there is no long range order of layers within the x-ray correlation distances in the sample. The air-dried sample exhibits two broad 001 reflections indicative of an interstratified system containing at least two basal spacings due to different 32 mm.o mm.o om.m o.om o.om mw.o sm.o m>.o o.om o.om mm.o vm.o omo.o 0.0m o.om av.o . bv.o vao.o o.om o.om va.o ma.o vmoo.o o.om o.om +Hzeoza QE poscoum mm mcoH .0paa90fl8u0> coH ooxaz ca 53:0 1309 2.3535 quonaoE i +mzmoz x Mo msam> Mo coauomum .ocoo HMHDACH mo.&o> wu3 H mo.ao> Hozvoz moomoqm sue: ouflasofl8u0>lm+mzmmza mo cofluomom >9 cocamuQO mmuflasoflfiuoxwixia:mzmozwxrzvoa 0>Humucommummm mo mcoflufimOQEoo .N OHQMH. 33 .ousmomxo xmulx uo wcofiuwccoo ucoao>fisoo uocc: pocuoomu mumB mcumuumm COMDUMHHMMG HH< .Uo ooa up an H cmfluc Eaau cmucmfiuo cm ADV can .ufim Ca .un a Umwuc Edam m Amy .Hoo O03 nonhuman xasmouu any we muflasoweuo>|+mZmA em iii/Ill: 1 31h. l ill j ml A mov you mcumuumm cofluomuuuflc >muix m .aflm loop 1 DON l con l 00¢ 1 con l com #0 [ ALIS N31Nil 34 water content le'the interlayer regions. :Dl addition, the diffraction peak intensities are relatively weak and layer stacking is considered to be incomplete. Upon heating to 100 0C, the pattern sharpens dramatically and multiple orders of reflection are observed corresponding to a single basal spacing of 12.70 A. Re-exposing the oven-dried sample to air did not result in the re-adsorption of water and no change of d-spacing was detected. Fig. 9 illustrates the X-ray diffraction patterns for oriented film samples of representatives [(CH3)4N+]X[(CH3)3NH+]1-X-vermiculites that have been dried at 100 OC. The presence of multiple orders of 001 reflections with reasonably small band widths indicates a regular distribution of gallery heights for each alkylammonium ion composition. To obtain further verification of the mixing of two alkylammonium ions within each gallery, we compared the experimental x-ray diffraction patterns with computer- generated patterns according to three models. In order to confirm ‘the structural and compositional inputs for the calculation, the 001 reflections of homo-ionic (CH3)3NH+- vermiculite and (CH3)4N+-vermiculite were generated (see Appendix I for the details). Fig. 10 and Fig. 11 compare the observed 001 reflections with the calculated Bragg reflections for the homo-ionic (CH3)3NH+-vermiculite (x=0.0) and (CH3)4N+-vermiculite(x=l.0), respectively. Reasonable agreements are obtained for the observed and calculated E INTENSITY (ARBITRARY UNITS) (\ L K L? p Fig. 9 X-ray diffraction patterns+for oriented film samples of representative [(CH3)4N $X[(CH3)3NH 11-x’ vermiculites heated at 100 C 36 _ . _ 4 I I fill H volcano T Too... ouflH30aEum>I+m2mAmmOv Dacofioson u0u :umuuma cux cmumazoamo can co>uomno ecu mo conflquEoo oa .oam mm 3 on as 2 o “03230.00 (eAgoIaa) Kilsueiul 37 p u oufiHSUHeum>I+Z¢AMmOV OMCOHOEOS you abouumm cux coumasono cam Uo>uomno on» no comwummeou Ha .oflm pouo_:o_oo 02030 ' mm ON _ or _ (23/010193) Kilsuaiul 38 relative intensities for homo-ionic derivatives with a single basal spacing. The agreement between the observed and calculated patterns showed that the structure factor assigned to the layer could be used for the calculation of the x-ray diffraction patterns of the mixed-ion [(CH3)4N+]X[(CH3)3NH+]1_X-vermiculites. Fig. 12 illustrates the observed and calculated 001 reflections for x=0.63. Clearly, very poor agreements are observed in the positions of the observed and calculated reflections based on Hendricks-Teller and Reynolds models for ion segregation compared to those of the Bragg model for the ion mixing in the galleries. These results strongly suggest that the basic synthetic strategy used in preparing the mixed ion vermiculite samples was successful.(There are some arguments concerning the true gallery x value which should be used in the calculation of 001 intensities. See appendix II.) We have also performed a MacEwan direct Fourier transform, which is equivalent to a one-dimensional Patterson synthesis based on the whole diffraction profile. This transform provides the lattice vector distribution along the c axis. The probability function W(R) for the mixed ion sample with x=0.63 is shown in Fig. 13. Only vector distributions corresponding to a unique d(x) value were observed. Thus, segregation of the two ions into separate phases with d=l3.34 and 12.70 A characteristic of the parent homo-ionic intercalates can be precluded. The 39 :uouuma Hmucmefiuomxm "mxm .Hocoe caocaom "yum HwUOE umaamelmxofluocmm "Bum .Hmmba oomum "0mm muflasofleum> I h om+m2mammovg mm.oH+Z¢AmmOVH now mamcoa mwuzu :0 comma mcumuuom flux Umumasono can Uw>ummno uo mcomwummeoo NH 0* on ON 0— n b _ . _ {1 mxu . fiW 1+1. _\1 2.4. 4 as % :1 1 l 4 % J... ... <4 .Oflh #9. r8 18. 18. 108 ii 7 0mm I own (eAgolaa) Kigsueiul on muflflaofieum> u sm.ofi+mzmfimzovu mm.ofi+z¢AmmocH mo cumuumm tux mo Snowmcmuu cm3momz wowuflo ma .on 90¢ a 4O (am : C no 9. 2.3 and. 41 absence of peaks at certain R, which corresponds to the sum of these d values, also precludes the possibility of alternating sequences of gallery spacings containing the two ions (ABAB, AAB, BBA, etc.). Basal spacings for the mixed alkylammonium vermiculites were determined by fitting the observed 001 reflections to the Bragg relationship Q: 2 nl/d where Q=4w sin e/x. Plots of Q versus 1 are shown for some typical mixed ion vermiculites in Fig. 14. The broken lines in the figure represent the best least square fit in the data points. It is noteworthy that the Q vs 1 plots exhibit a high degree of linearity, as expected for uniformly mixed systems in which the two ions co-occupy the same galleries. If the two types of alkylammonium ions were demixed with the alkylammonium ions randomly segregated into separate galleries, we would not expect a strictly linear relationship between Q and 1. Fig. 15 illustrates the compositional dependence of the normalized basal spacing, dn(x) for 27 independent [(CH3)4N+]X[(CH3)3NH+]1_x-vermiculite samples over the composition range, x=0 to 0.96. Here dn(x)=[d(x)- d(0)]/[d(1)-d(0)]. The basal spacings clearly follow a non— Vegard’s laW' (non-linear) dependence (n1 composition. Particularly noteworthy are (i) the rapid rise in dn(x) with increasing x near the threshold value of Xt=0°2 and (ii) the value of dn(x) reaches the maximum possible value of 1.0 very near an x value of 0.6. Fig. Arbitory Q (4Trsin0/X) % \: \ o~-o x=.96 G-{EJ x=.86 b-fi x: 63 ow» x=.41 /fi9 v—a x=.14 jg/I x ;dev.pokfl ;£f// 2n 0: 1 d C{//£T//£J/// ' .91 / // «JX/ I i 7 ('3 —-4 03‘ \J 14 Representative Q vs 1 plots for [(CH3 )4 N+ ]X [(CH3 )3NH+]1 the best we1g data points. fite -vermiculites. The broken lines are d linear least squares fit to the 43 mmuflasofieum>lxl Amman cmmfiameuoc onu no mocoocmmmc coflufimomfioo one ma .mflm .m.vu8x\< um.ouu “o.mum "pom umuosmuma ocfi3oHHow mnu nufl3 uxou on» no u coflumsvm gnaw: Bump may ow uflm mumsvm ummma mnu ma mafia cwaom H+m2mxmmoclxa+zexmmoVH you mocflomam V V V V .Vfi T WV h p y p—n . p p h h— b p p p y.- i p p r pu- p p p > r- L LL ALI AALAl Ago.o mmwo num.o mmao onfia (10“? 44 For the comparison purposes, the expected dn(x) vs. x plot for a ion-segregated [(CH3)4N+]X[(CH3)3NH+]1_X-V system according to the Reynolds model is shown in Fig. 16. The observed data (Fig. 15) deviate strongly from the expected behavior for an ion—segregated system. III-3. Generation of dn(x) vs x Plots for Mixed Ion Monolayer Model A computer simulation of the dependence of dn(x) on x was carried out in order to provide a useful model for the experimental results in Fig. 15. Our model began with the definition of a transverse layer rigidity parameter "p". Consider the single mono-layer system shown in Fig. 17 in which there is one gallery ion per host layer unit cell. If we begin with all small "B" ions in the gallery and replace one of them with a larger "A" ion, then one or more unit cells in each layer defining the gallery surfaces will undergo a distortion. The distortion of one primitive layer unit cell per "A" cation would represent an infinitely floppy host layer with a layer rigidity parameter of p=l. For the distortion of all unit cells (p=m), the layer would be infinitely rigid. Intermediate values of p would quantitatively describe the degree of layer rigidity. Depending on the size of the unit cell, each value of p would be associated with a characteristic healing length A. 45 .mcofluomamon Hoo xflm umuflw ocu mo coflmmmuwmu mumsvm ummma Eouu owumasoamo wumB moCHUMQm Hommn one .HwUOE moaocsmm map no woman wmuflasofleum>n xuafl+m2mlmmocl Ixfl+ZvAmmoVH CH :OHummwummm 50H Mow ocflommw Hmmmn UwNHHwEMo: mcu uo monoccwmoo coflpflmomeoo one !---- L A oo.o mm.o om.o mv.o co.“ (XVI) 46 The simulation results are shown in Fig. 17 for several different values of p for the triangular net. In the floppy layer limit where p=1 and the healing lengthflk=0, a Vegard’s law behavior is obtained. However, in the infinitely rigid layer region where p=w,2.=m, the initial slope is infinite. For the intermediate cases: p=7, A=ao and p=13,x=J'3'ao, there are differences in the initial slopes and in the x values at which each dn reaches 1. Significantly, there is no percolation threshold even for finite values of p because dn(x) depends upon all of the large ions, not only on those belonging to the infinite percolation cluster. III-4. Infrared and Raman Spectroscopies of Mixed [(CH3)4N+]X[(CH3)3NH+]1_X-Vermiculites Two series of angle-dependent infrared spectra for trimethylammonium vermiculite(x=0.0) and for a mixed-ion [(CH3)4N+]X[(CH3)3NH+]1_X-vermiculite with x=0.54 are shown in Fig. 18 and Fig. 19, respectively. In both cases, the peak intensity of the hydrogen-bonded N+-H stretching [absorption band of interlayer (CH3)3NH+ at about 2730 cm'l,34 relative to the trapped water O-H stretching at 3280 cm’l,35 and the in-phase C-H stretching band at 2970 cm'1 and at 3040 cm"1,36 increases upon decreasing the angle between sample film and the incident infrared beam. This means that the orientations of N+-H dipoles in the galleries 1.0 0.8 0.6 dn(Xg) 0.4 ' 0.2 0.0 Fig. 17 47 V V V Y Y I Y 1 V I V V I I Y I ' l Monolayer triangular lattice simulations (dotted lines) of dn(x) for idealized [A] [B] -x' vermiculites for several different va ues of the healing length,k, and the layer rigidity parameter, p. The solid lines are from Equation 4 of the text with (1) p=l,2.=0; (2) p=7, x=a0; (3) p=13, 13.13- a0; (4) p=w’ 1:00, ' Inset: the tr1angular lattice defined by the spherical A and B pillaring cations. The expanded region represents the distortion which occurs in the host layer when A replaced B and k=a0. In this latter case, the number of expanded sites is p=n+1=7 where n is the number of nearest neighbors. 48 Transmittance(arbitary) 2970 3040 r' I I I_' T I 3500 3000 2500 Frequency (cm'1) Fig. 18 IR spectra of (CH3)3NH+-vermiculite with different incident beam angles. Note the change in relative peak intensity of the N+-H stretching at 2730 cm"1 relative to other angle-independent peaks. Transmittance0.20). However, this x-dependent ion segregation mechanism is unlikely because there is no apparent reason why the mixed ion system should be segregated at low values of x and uniformly mixed at high values of x, especially when dealing with single-crystal vermiculite wherein the layer charge is expected to be more or less uniform. The second possible mechanism invokes the presence of two types of ion-binding sites. One site is the gallery positions already defined in our monolayer model. In addition, however, there may be ion exchange positions, such as the edges of the clay platelets, which do not contribute to or influence the observed gallery heights. Such sites, which are independent of x, and which do not contribute to the magnitude of dn(x), shall be referred to as "defect sites." The threshold in the dn(x) vs x plots can then be explained qualitatively by the preferential binding of 57 (CH3)4N+ ions at the defect sites. Once the defect sites are saturated near a threshold value(xt) of x=0.2, the further addition of (CH3),,‘N+ ions results in the abrupt expansion of the gallery region. This second mechanism seems to be much more plausible than the bimodal ion- segregation/ion-mixing mechanism described earlier. We now consider a quantitative treatment of transverse layer rigidity in our alkylammonium pillared clay while at the same time allowing for the binding of pillaring ions at both ordered gallery sites and defect sites. We assume the basal spacing to be dependent on the gallery (CH3)4N+ ion concentration xg. The latter quantity is a function of the total (CH3)4N+ composition x. The dependence xg on x is determined by two parameters, f and A/kT where f=Nd/Ng is the fraction of the defect sites relative to the gallery sites and A is the effective binding energy difference between these two sites. The binding of (CH3)4N+ ions should be favored over (CH3)3NH+ ions at the defect sites, owing to the larger volume of such sites. For simplicity, we assume only one type of defect site. A statistical mechanical calculation4O relates xg to x, f, A./kT. The steps involved in the calculations are described in Appendix III. Basically, (CH3)4N+ cations are treated as the particles which follow the Fermi distribution function for the distributions of (CH3)4N+ cations in both sites. The result is Equation 5, which relates xg to x: 58 ( > (5) From this relationship we expect that for xxt, additionally ingested (CH3)4N+ ions enter the galleries and the result is a rapid increase in dn(x). On the basis of methods recently developed by Xia and Thorpe,41 the analytical solution shown in Equation (6), can be obtained for our monolayer simulation model: dn=I1-<1-xg>"1 (6) k= ¢(xfmyn) Here, p=n+1, the total number of sites expanded by the larger ion exchange in the gallery and n is the number of layer triangular sites that are transversely puckered by the insertion of a large cation. The p parameter,which is approximately' proportional to the square of the healing 59 length, is defined as the transverse layer rigidity parameter in the monolayer model. Equation 6 fits the simulated monolayer model extremely well, as shown by the solid lines in Figure 17 for four different sets of transverse layer rigidity parameters. No percolation threshold is predicted for dn(x) in this case because all of the pillaring ions occupy gallery sites which directly influence the gallery heights. Equation 6 may be readily adapted to a two-site binding model in which one site is a defect site which does not contribute to the gallery height of the mixed ion clays and the second type of site is a gallery site. By substitution of xg in Equation 6 using Equation 5, we can express xg in terms of the overall composition x, the ratio of defect to gallery sites f, and the binding energy' difference between the two sites A . Hence, Equation 6 takes the form shown in Equation 7. (7) dn(x)-_-[1-{1- (I)(x,f,A/kT)}p] 60 We have used Equation 7 to obtain a non-linear least- squares—fit to the data of Figure 15. The parameters which gave the best fit (solid line in Figure 15) are p=8.0, f=0.5, ,A/kT=4.3. Despite the limitations of our monolayer model, particularly the abruptness of the layer distortions in the vicinity of the large ion positions, the parameters obtained for the best least-squares-fit are physically plausible. The difference in binding energy for the two binding sites amounts to only about 2.5 Kcal/mole. The value of the layer rigidity parameter p=8.0 indicates that little or no transverse layer distortion occurs at lateral distances up to one unit cell from the pillar position, which in our model is about 5.3 A. This means that "sagging" does not occur at cell positions adjacent to the pillaring site, and that the galleries remain fully expanded when two pillars are separated by a distance of at least 10.6 A. The f value of 0.5 indicates that 33% of the exchange cations are at positions other than those which contribute to the observed basal spacings. It is instructive here to consider the possible defect sites other than edge sites which do not contribute to the observed basal spacings. In our synthetic procedures, the clay layers are initially in a greatly swollen state due to unique swelling property of the trimethylammonium exchange form. When the layers are reflocculated by the addition of tetramethylammonium ions, several types of defect centers may be introduced. Fig. 22 illustrates possible defect sites for exchange ions on basal 61 x x “ g / g 0 $ Fig. (e) 22 Schematic illustration of possible non-gallery defect sites for exchange ions on the basal surfaces (designated x) which do not contribute to the 001 basal spacing (a): Surfaces at twisted layer-planar layer interfaces, (b): Surfaces of bifuricated layers, (c): Surfaces of internal and extenal ledges and creverses formed by layer termination, (d): Surfaces of layer edge-layer face interfaces. Defects (a)-(d) are in addition to non-gallery binding sites at layer edges(e) 62 surfaces in reconstituted alkylammonium vermiculite tactoids. The surfaces marked by an "x" indicate the positions of the pillaring cations on basal surfaces which do not contribute to the basal spacing change. In addition to the basal positions formed by these four different types of defects, the exchange ions can occupy positions at the edges of the layers which also do not contribute to the basal spacing. Thus there is a large variety of possible defect sites in pillared clays. High resolution electron microscopy studies are currently underway in an effort to identify such non-gallery defects. In Fig. 23, the calculated gallery site concentration of (CH3)4N+ is plotted against the total (CH3)4N+ concentration. As expected, there are two regions in xg vs. x curve. For the low x value range, the (CH3),er+ cations prefer defect sites, which results iJIIa small increase of xg. Beyond a certain x value, an almost linear relationship is observed between xg and x. The initial small increase in the large cation concentration in the gallery sites results in the rapid increase of d-spacing due to a finite layer rigidity, as described by the simulation results. The infrared and. Raman spectroscopy studies on the mixed [(CH3)4N+]X[(CH3)3NH+]1_X-vermiculites (cf.,Figure 18- 21) also provided insight into the mixed ion systems. The orientation of smaller cations in the gallery affects the apparent height of the trimethylammonium cation and, also, the d-spacing of the clay. The orientation of N+-H dipoles 63 m.vHBX\< 6cm .m.oum .wHQ mmEJmmm Homoe owflm acapcfln can one .mouaH50esso>uxaefi :zmA moc_ _+2exmzow1 00w x :ofluflmomeoo +2v mmov Hmpou wcu co x :ofluHmOQEOO :ofi xuwafimo mo wocwccwawc 0:0 mo uon mm .qflm 64 of trimethylammonium ions in vermiculite galleries was determined to be perpendicular to the oxygen planes of layer, regardless the ratios of the two different cations. The specific orientation of the trimethylammonium cations in the gallery is probably attributed to the dominant van der Waals interaction between the threeimethyl groups and the surface oxygen atoms, in addition 'Uo hydrogen. bonding between N+-H and basal oxygen atoms. Such a specific orientation of smaller cations in the galleries satisfies the requirement of our monolayer model in defining the sizes of cations. The Raman shifts for the torsional mode of the surface oxygen planes in mixed [(CH3)4N+]X[(CH3)3NH+]1_X-vermiculite are well correlated with the two different cation sites of our mixed ion clay model. A comparison of xg vs. x plot (Fig. 23) with the plot of Raman shift vs. x (Fig. 21) clearly shows the dependency' of Raman shift. on :xg, 'the gallery (CH3)4N+ concentration. Qualitatively, the Raman shift of the torsional mode can be related with the collective interactions between the interlayer cation and hexagonal oxygen atoms during' the vibration. Apparently, the concentration increase of the larger (CH3)4N+ cation in the gallery results in a linear increase in the extent of the interactions. In the range of x < 0.25 in Fig. 21, there is almost no change in the Raman frequency, which might be due to the defect site effects for the real vermiculite samples. Also, in the range of 0.25< x 65 < 1.0, a linear increase in Raman frequency is observed with increasing the x value. This latter dependence can be interpreted in terms of a linear increase in the x9 value, as calculated from the two site model parameters (cf. Fig. 23). V . CONCLUSIONS The major conclusions of this study are as follows: 1. Mixed ixn1 clay systems [(CH3)4N+]X[(CH3)3NH+]1_X - vermiculites have been prepared which are suitable for quantitatively describing the transverse layer rigidity of host silicate layers by observing the relationship between gallery height and the fraction of pillaring (CH3)4N+ ion occupying the gallery surfaces. 2. A layer rigidity parameter (p=8) for vermiculite host layers was determined by analytical and two-site ion binding model simulation. This model now makes it possible to quantitatively compare the transverse layer rigidities of several different types of host systems and to explicitly determine the structural and compositional factor influencing the pillaring of lamellar solids. 66 A. 67 Appendix I Calculation of the One-Dimensional (00]) X-Ray Diffraction Pattern for [(CH3)3NH"] -Vermiculite 1. Intensity (9)=K* LP (en [cam2 * 0(0) K : constant LP : the Lorentz Polarization factor Lorentz factor : lnstmment and geometry dependent Polarization factor : due to polarized X-ray scattering |G(9)|2 : square of structure factor of unit cell (I): Interference function 2. LP factor calculation LP = (1 + 003229) w sin 29 \l/ : powder ring distribution factor for a random powder : w is proportional to . srn 9 for a single crystal : w is a constant for a clay sample : w is intermediate Reynolds method 42 is used to calculate \V 68 3. One-dimensional structure factor calculation (3 = Z nf(cos(RQ)+i sin (RQi) plane n : number of like atoms inaplane 1(6) : atomic x-ray scattering factor R : distance from origin 4ndn9 Q . —1_— R (R) 2 n* atom/plane Q < 6.35 0.86 (CH3)3NH+ <— 3.295 6.00 0 ] 2.77 2.94 Si + 1.06 AI = 1.07 4.00 O + 2 (OH') < 0 0.48 Al + 5.52 Mg < -107 4 .000 + 2 (OH-) -2.77 2.94 Si + 1.06 Al < -3.295 6.00 0 @ < 6.35 0.86 (CH3)3NH+ Unit cell formula : ((CH3)3NH+)1-72 (Sis-87 AI2-13) [No-48 M9562] O.20 (OH)4 69 4. Interference function (0(0) N-1 u0mno ”.mxm COMUMHDOHMU HOHHOBIWMCHHUCQm u Bl: s ofi+mzmxmmUVH zexmmoc_ you manouuoo oux m¢.oumx anew: owflasofleuo> 1 mm. _.. + omuoasoHoo can po>ummno mo confluomaoo mm .aflm mm 0... on ow or 0 Pi % p _ . _ L L Li 0 axm .. roe .llll l, \tllllllafi. 114 Yll from ...lI .. new. low— r. DON (eAnolea) Kirsueiul 75 cumuunm pux pm>ummno ".mxm :ofluoHDOHoomwmoc>mm " wmm mv.oumx mcflm: ouflHDOMEum> I on H+mZmAmmUVH mo.om+zexmmocu toe announce cox pmuoasoaoo can pw>uomno no comwquEou mm .mflm mg" o4 on om o. o pi — p — n — n — n 6 $0 .. roe. mm r 11 as nu ll/\i flow .6... ‘4- Gm - .A ) 8 Iowa 6... nu H 1 _A Ad row: ( loom 76 Appendix III Derivation of Equation 5 We start with the general expression for thermodynamic potential _ n . 0k =-leog 2 [e111 ski/k1] k 0 k3 Thermodynamic potential of K'" level (I : Chemical potential of particle 8k : Binding energy ofk‘h level nk : # of particles in the k'h level For the sites that have only two choices : non occupying or occupying by a particle : n, =0 0r1 (2k =- kT log [1+9 (ll-Ekwk'r] ko 9(H-8k)/RT du 1+e(ll“ek)/'(x,f,A/kT) 10. ll. 12. 78 References Bailey, S. W. In Crystal Structures of Clay Minerals and Their X-ray Identification; Brindley, G.W.; Brown, G., Eds.; Mineralogical Society: London, 1984; p 1. Theng, B. K. G. The Chemistry of Clay-Organic Reaction; Wiley: New York, 1974. Dixon, J. B. In Minerals in Soil Environments; Dixon, J. B.; Weed, S. B., Eds.; Soil Science of America: Madison, 1977. Mott, C. J. B. Catalysis Today 1988, 2, 199. Geise R. F. Clays Clay Miner. 1975, 23, 165. Thomas J. M. In Intercalation Chemistry; Whittingham M. 8.; Jacobson A. J., Eds.; Academic Press: New York, 1982. Vaughan E. E. W. Catalysis Today 1988, 2, 187. Barrer R. M.; MacLeod D. M. Trans. Faraday Soc. 1955, 51, 1290. . Vaughan D. E. W. In Perspectives in.Mblecular Sieve Science; Flank W. H.; Whyte T. B., Eds.; American Chemical Society: Washington, 1988. Barrer R. M. In Zeolites and Clay Minerals as Sorbents and Melecular Sieve; Academic Press: New York, 1978. Pinnavaia T.J. Science 1983, 220, 365. Lahav N.; Shaui U.; Shabtai J. Clays Clay Miner. 1978, 26, 107. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 79 Yamanaka S.; Brindley G. W. Clays Clay Miner. 1978, 26, 21. Tzou M. S. Ph.D Thesis, Michigan State Univ., 1983. Pinnavaia T. J.; Tsou M. S.; Landau S. D.; Raythatha R. H. J. Molecular Catal. 1984, 27. 195. Sterte J. Clays Clay.Miner. 1986, 34, 658. Stul M. S.; Mortier W. J. Clays Clay Miner. 1974, 23, 391. Setton R. Chemical Reactions In Organic and Inorganic Constrained Systems, Nato Asi Series 165, D.; Reidel Publishing Co.: Dordrecht. 1986. Freeman M. P. J. Colloid. Interfac. Sci. 1971, 37, 760. Vali H.; Kosier H. M. Clay minerals 1986, 21, 827. Solin S. A. J..Mbl. Catal. 1984, 27, 293. Barrer R. M.; Brummer K. Trans. Faraday Soc. 1963, 59, 959. York B. R.; Solin S. A.; Wada N.; Raythatha R. H.; Johnson I. D.; Pinnavaia T. J. Solid State Comm. 1985, 54, 475. Foster M. D. Clays Clay Miner. 1961, 10, 70. Lee S.; Kim H.; Solin S. A.; Pinnavaia T. J., In Chemical Physics of Intercalation, Nato Asi Series Vol. 172; Legrand A. P.; Flandrois S. Eds.; Plenum Press: New York, 1987. Dye J. L.; Nicely V. A J: Chem. Education 1971, 48, 443. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 80 Hendricks S.; Teller E. J. Chem. Phys. 1942, 10, 147. Reynolds R.C. In Crystal Structures of Clay Minerals and Their.X-Ray Identification; Brindley, G.W.; Brown, G., Eds.; Mineralogical Society: London, 1984; p 249. Mathieson A. McL.; Walker G. F. Am. Miner. 1954, 39, 231. Wright A. C. Clays Clay Miner. 1973, 21, 489. Stout G.; Jensen L. H. X—ray Structure Determination; Macmillan Publishing Co.,Inc.: New York, 1968. Ishii J.; Shimanouchi T.; Nakamura M. Inorganica Chemica Acta 1967, 1:3, 387. Norrish K. J. Colloid. Interface Sci. 1975, 50:1, 120. Huong P.V.; Schlaak M. Chem. Phys. Letters 1974, 27, 111. Bertsch A. G.; Habgood H. W. J; Phys. Chem. 1963, 67, 1621. Howe R. F.; Taylor M.J. Spectrochemica Acta 1987, 43, 77. Fink D. H.; Nakayama F. S.; McNeal B. L. Soil Sci. Soc. Amer. Pro. 1971, 35, 552. McBride M. B.; Mortland M. M. Clays Clay Miner. 1973, 21, 323. Vansant E.F.; Uyterhoeven Clays Clay Miner. 1972, 20, 47. Davison N. Statistical Mechanics; McGraw Hill Book Co.: New York, 1962. 81 41. Xia W.; Thorpe M. F., Phys. Rev. A (to be published). 42. Reynolds R. C. Clays Clay Miner. 1986, 34, 359. mfllliflifllllflils