Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Solid-state NMR, X-ray diffraction and neutron diffraction studies on organic guest/zeolite ZSM-5 complexes Lee, Jang Seob 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-ubc_2007-318102.pdf [ 22MB ]
Metadata
JSON: 831-1.0059774.json
JSON-LD: 831-1.0059774-ld.json
RDF/XML (Pretty): 831-1.0059774-rdf.xml
RDF/JSON: 831-1.0059774-rdf.json
Turtle: 831-1.0059774-turtle.txt
N-Triples: 831-1.0059774-rdf-ntriples.txt
Original Record: 831-1.0059774-source.json
Full Text
831-1.0059774-fulltext.txt
Citation
831-1.0059774.ris

Full Text

Solid-state NMR, X-ray Diffraction and Neutron Diffraction Studies on Organic Guest/Zeolite ZSM-5 Complexes by Jang Seob Lee B.Sc. (Hon), University of British Columbia, 2001 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Doctor of Philosophy in The Faculty of Graduate Studies (Chemistry) THE UNIVERSITY OF BRITISH COLUMBIA June 2007 © Jang Seob Lee, 2007 Abstract Zeolites are microcrystalline aluminosilicate materials that have intricate channel systems, into which molecules can be adsorbed. Important as industrial materials, they are used in gas separations, sorption and catalysis where the property of 'molecular sieving' is exhibited toward guest species. Thus, understanding the interactions between guest species and the host zeolite frameworks is central to predicting the syntheses and applications of zeolites, and this requires detailed structural information about the guest/zeolite complexes. For most zeolites, structure determination by single crystal X R D is precluded due to the microcrystalline nature of the materials. As an exception, ZSM-5, which was used exclusively for this thesis, is available in large crystals (> 100 nm). In the quest for a universal method to probe the structures of various types of guest/zeolite complexes, we have developed an alternative approach using solid-state NMR, which is not limited by crystal size, unlike single crystal X R D . The main focus of this thesis is exploring the potential of the NMR method by using it to determine various guest/zeolite complexes followed by verification by diffraction techniques where possible. The NMR method uses the dipolar coupling interactions between nuclei on the organic guest molecules and the silicon nuclei in the zeolite framework, which contain information regarding relative proximities of the protons in the guest to the silicon atoms in the rigid framework. The complexes tested were: o-xylene/ZSM-5, p-dicyanobenzene/ZSM-5, p-dinitrobenzene/ZSM-5 and benzene/p-xylene/ZSM-5 (the mixture of the two organics as guest molecules in the structure determination of the complex for the first time). Precise structures for all the complexes listed have been determined successfully by the NMR method. Other relevant NMR studies, including connectivity, dynamics and spin diffusion, are also presented to support a better understanding of the interactions between the organic guests and the zeolite frameworks. Verification of the first three complex structures determined by NMR has been done successfully by diffraction methods. For o-xylene/ZSM-5, attempts to apply single crystal X R D were unsuccessful due to the very low diffusivity of the organic. Instead, powder neutron diffraction was used. i i Table of Contents Abstract ii Table of Contents iii List of Tables viii List of Figures. xiii Symbols and Abbreviations xxi Acknowledgements xxiv Dedication xxvi Chapter 1 INTRODUCTION: General Background 1 1.1 Zeolites 1 1.1.1 Applications of zeolites c 2 1.1.2 Zeolite synthesis 3 1.1.3 Zeolite characterization 4 1.1.4 Zeolite lattice structures 5 1.1.5 Zeolite ZSM-5 5 1.2 Solid-state NMR 6 1.2.1 The dipolar coupling 7 1.2.2 The chemical shift anisotropy 8 1.2.3 The quadrupolar interaction 10 1.2.4 The cross polarization technique 11 1.2.5 'Magic-angle' spinning and high power decoupling 11 1.2.6 NMR relaxation mechanisms 13 1.2.6.1 Spin-lattice relaxation time (7i) 13 1.2.6.2 Spin-spin relaxation time (7"2) v 14 1.2.6.3 Spin-lattice relaxation time in the rotating frame (Tip) 15 1.2.6.4 Motion and temperature dependence of relaxation times 17 1.2.7 Second moments 18 1.2.8 Internuclear distance measurements from dipolar couplings 19 1.2.9 Dynamic studies by 2H solid-state NMR 20 1.2.10 Solid-state NMR and zeolites i 21 1.3 Crystallography • 22 1.3.1 Crystal system and Bravais lattices 22 iii 1.3.2 Point groups and space groups in crystallography 24 1.3.3 Diffraction and reciprocal space 26 1.3.4 Single crystal X R D 30 1.3.4.1 Structure calculation from diffraction data 31 1.3.4.2 The Patterson method 32 1.3.4.3 Direct methods • 33 1.3.4.4 Refinement of a single crystal structure 33 1.3.5 Powder Diffraction 35 1.3.5.1 Powder XRD instrumetation 36 1.3.5.2 Powder neutron diffraction 36 1.3.5.3 Rietveld method 43 1.3.6 Comments on organic/zeolite structure determinations by X-ray and neutron diffraction 44 1.4 Aims of this thesis ! 44 Chapter 2 EXPERIMENTAL DETAILS: Methods and Materials 46 2.1 Sample preparation and analysis 46 2.1.1 Deuterium Labeled samples 46 2.1.2 Highly siliceous ZSM-5 samples 46 2.1.3 Loading of guest organic molecules into zeolites 48 2.1.4 Determining the loadings of organic guests in the zeolite 49 2.2 Solid state NMR ..' • 51 2.2.1 NMR spectrometer 51 2.2.2 MAS probes 51 2.2.3 Magnetic field shimming 53 2.2.4 Variable temperature experiments 53 2.2.5 Setting up experiments using reference samples 55 2.2.6 Relaxation time measurements 56 2.2.7 2 9 S i MAS and 1 H / 2 9 S i C P M A S NMR 57 2.2.8 INADEQUATE experiments 58 2.2.9 1 H 2D-NOESY 59 2.2.10 2 H static NMR. . . 59 2.2.11 NMR data analysis and calculations 59 2.3 Diffraction methods 60 2.3.1 Single crystal X R D data collection and processing 60 2.3.2 Powder diffraction sample preparation 61 iv 2.3.3 Powder X R D data collection 61 2.3.4 Powder neutron diffraction data collection 63 2.3.5 Powder diffraction data processing 63 Chapter 3 Structure Determination Strategies 64 3.1 Introduction 64 3.2 NMR structure determinations 65 3.2.1 2 9 S i INADEQUATE assignments 66 3.2.2 Probing the dipolar couplings 66 3.2.3 NMR structure determination program 70 3.2.4 'Goodness' of the proposed structure 75 3.2.5 Verification of the proposed structure 76 3.3 Powder diffraction structure determinations ».. 76 3.3.1 Indexing the diffraction data 78 3.3.2 Ab initio structure solution by global optimization 79 3.3.3 Structure refinement 82 3.3.4 Quality of Results 84 3.4 Summary 85 Chapter 4 NMR Determination of the Structure of o-xylene/ZSM-5 86 4.1 Introduction 86 4.2 Solid State NMR experiments on o-xylene/ZSM-5 88 4.2.1 Characterization of the 2 9 S i spectra of the o-xylene/ZSM-5 complex 88 4.2.2 One-dimensional 2 9 S i NMR 89 4.2.3 Cross Polarization Experiments 93 4.2.4 C P Experiments on o-xylene-de/ZSM-5 94 4.2.5 Solving the structure of the o-xylene/ZSM-5 complex 97 4.3 CP Experiments on o-xylene-oyZSM-5. 106 4.3.1 Solving the structure of the o-xylene-ayzSM-5 complex 108 4.4 o-xylene-de/ZSM-5 CP Drain Experiments 113 4.4.1 Solving the structure of the o-xylene-cyzSM-5 complex 116 4.4.2 Comments on the Structures Determined by NMR 122 4.5 Single Crystal XRD of o-xylene/ZSM-5 123 4.6 Summary... 1 2 4 v Chapter 5 Powder Neutron Diffraction Study of the o-xylene/ZSM-5 System 125 5.1 Introduction 125 5.2 Structure determination of o-xylene-d10/ZSM-5 126 5.3 Summary 135 Chapter 6 Investigation of the Structures of p-Dicyanobenzene/ZSM-5 and p-Dinitrobenzene/ZSM-5 by NMR and Single Crystal XRD 136 6.1 Introduction 136 6.2 Study of 4 molecules per unit cell of DCNB in ZSM-5 137 6.2.1 Solid-state NMR study of 4DCNB/ZSM-5 137 6.2.1.1 Variable Temperature 2 9 S i M A S NMR Spectra 137 6.2.1.2 Peak Assignments b y 2 9 S i INADEQUATE experiment 138 6.2.1.3 C P experiments 140 6.2.1.4 Solving the structure of the p-dicyanobenzene/ZSM-5 complex 141 6.2.1.5 C P drain experiments 144 6.2.2 Single crystal X R D study of 4DCNB/ZSM-5 149 6.2.3 Comments on the NMR and single crystal X R D results 151 6.3 Study of 2 molecules per unit cell of DCNB/ZSM-5 151 6.3.1 Solid-state NMR study of 2DCNB/ZSM-5 152 6.3.2 Single crystal X R D structure of 2DCNB/ZSM-5 160 6.3.3 Comments on the NMR and single crystal X R D results of 2DCNB/ZSM-5 161 6.4 Study of 2 molecules per unit cell of p-dinitrobenzene in ZSM-5 162 6.4.1 Solid-state NMR study of 2DNB/ZSM-5 163 6.4.2 Single crystal structure of 2DNB/ZSM-5 171 6.4.3 Comments on the NMR and single crystal X R D structures of 2DNB/ZSM-5 172 6.5 NMR cf. single crystal XRD 173 6.6 Summary 176 Chapter 7 Investigation of Mixtures: Study of the Structures of the 2+2 Benzene/p-Xylene/ZSM-5 Complex by NMR 177 7.1 Introduction 177 7.2 Solid State NMR experiments on the 2+2 mixtures in ZSM-5 181 7.3 2+2 BENZ-de/PXY-de/ZSM-S 187 7.3.1 Structural study of 2+2 BENZ-dg/PXY-de/ZSM-S at 270 K 187 7.3.2 Structural study of 2+2 BENZ-cfe/PXY-de/ZSM-S at 293 K 194 7.3.3 Comments on the NMR structures of BENZ-de/PXY-de/ZSM-S 201 vi 7.4 2+2 BENZ-de/PXY-oVZSM-S 202 7.4.1 Structural study of 2+2 BENZ-d6/PXY-d 4 /ZSM-5 at 270 K 202 7.4.2 Structural study of 2+2 BENZ-dg/PXY-d^ZSM-S at 293 K 208 7.4.3 Comments on the NMR structures of BENZ-des/PXY-d 4 /ZSM-5 „ . 214 7.5 2+2 BENZ/PXY-d10/ZSM-5 215 7.5.1 Structural study of 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K 215 7.5.2 A structure study of 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 293 K 221 7.5.3 Comments on the NMR structures of B E N Z / P X Y - d 1 0 / Z S M - 5 228 7.6 2H static NMR study of 2+2 BENZ/PXY/ZSM-5 at 293 K 229 7.7 1H NOESY study of 2+2 BENZ/PXY-oVZSM-5 at 293 K 230 7.8 Summary 233 Chapter 8 Summary and Suggestions for Future Work 234 8.1 Summary 234 8.2 Suggestions for future work 235 References 237 Appendix A Supplementary information for the NMR structures 243 A.1 o-Xylene/ZSM-5 system 243 A.2 p-Dicyanobenzene/ZSM-5 system 246 A.3 p-Dinitrobenzene/ZSM-5 248 A. 4 2+2 Benzene/p-xylene/ZSM-5 249 Appendix B Supplementary information for diffraction structures 254 B. 1 Crystal structure determination for o-xylene-d10/ZSM-5 254 B.2 Crystal structure determination for 4DCNB/ZSM-5 256 B.3 Crystal structure determination for 2DCNB/ZSM-5 259 B.4 Crystal structure determination for 2DNB/ZSM-5 262 vii List of Tables Table 1.1 Typical magnitudes of nuclear spin interactions for common nulcei in a 9.4 Tesla magnetic field 7 Table 1.2 Corresponding symbols for symmetry elements in the Schonflies notation and Hermann-Mauguin notation 25 Table 1.3 List of refinement variables and their uses during the refinement 34 Table 4.1 Relaxation parameters for the 1 H and 2 9 S i nuclei in the o-xylene/MFI complex 89 Table 4.2 Average values of the six structural parameters of o-xylene-d 6 in the framework of ZSM-5 at 273 K and 315 K by C P 101 Table 4.3 Fractional atomic coordinates for the o-xylene-de/ZSM-5 complex at 273 K and 315 K from the strucuture determined by C P experiments 104 Table 4.4 Average values of the six structural parameters of o-xylene-d 4 in the framework of ZSM-5 at 273 K and 315 K from C P experiments 110 Table 4.5 Fractional atomic coordinates for the o-xylene-d 4/ZSM-5 complex at 273 K and 315 K from C P experiments 112 Table 4.6 Values of /c',s for C P and kiS from C P drain plots at 273 K 115 Table 4.7 Values of k)s for C P and kiS from C P drain plots at 315 K 116 Table 4.8 Values of ks determined from C P drain experiments 116 Table 4.9 Average values of the six structural parameters of o-xylene-d 6 in the framework of ZSM-5 at 273 K and 315 K from C P drain experiments 118 Table 4.10 Fractional atomic coordinates for the o-xylene-de/ZSM-5 complex at 273 K and 315 K from C P drain experiments 121 Table 5.1 Refined parameters for o-xylene/ZSM-5 complex at 272 K (space group Pnma) 132 Table 6.1 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step in the calculation from the C P experiments on 4 D C N B / Z S M - 5 142 Table 6.2 Values of the average six structural parameters of DCNB in the framework of ZSM-5 at 305 K from the C P experiments on 4DCNB/ZSM-5 . 142 Table 6:3 Values of /c' / s from C P and /c,s from C P drain plots for 4DCNB/ZSM-5 at 305 K 146 Table 6.4 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step in the calculation from the C P drain experiments on 4DCNB/ZSM-5 147 Table 6.5 Average values of the six structural parameters of DCNB in the framework of ZSM-5 at 305 K from C P drain experiments on 4 D C N B / Z S M - 5 . 147 Table 6.6 Values of k)s from C P and k,s from C P drain plots for 2DCNB/ZSM-5 at 330 K 156 Table 6.7 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step in the calculation for the C P data from 2DCNB/ZSM-5 156 Table 6.8 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step in the calculation for the C P drain data from 2DCNB/ZSM-5 157 Table 6.9 Average values of the six structural parameters of D C N B in the framework of ZSM-5 at 300 K with r2 > 0.93 from C P experiments for 2DCNB/ZSM-5 157 Table 6.10 Average values of the six structural parameters of DCNB in the framework of ZSM-5 at 300 K with r2 > 0.93 from C P drain experiments for 2DCNB/ZSM-5 157 Table 6.11 Values of k'ls from C P and k,s from C P drain plots for 2DNB/ZSM-5 at 330 K 166 Table 6.12 Ranges used for the parameters in the structure calculations and numbers of solutions after each step in the calculation for 2DNB/ZSM-5 from C P experiments 167 Table 6.13 Ranges used for the parameters in the structure calculations and numbers of solutions after each step in the calculation for 2DNB/ZSM-5 from C P drain experiments 167 Table 6.14 Average values of the six structural parameters of 2DNB/ZSM-5 at 330 K with r2 > 0.93 from C P experiments 168 Table 6.15 Average values of the six structural parameters of 2DNB/ZSM-5 at 330 K with r2 > 0.92 from C P drain experiments 168 Table 7.1 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 B E N Z - d y P X Y - d y z S M - 5 at 270 K by C P 190 Table 7.2 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 B E N Z - d y P X Y - d y Z S M - 5 at 270 K by C P drain 190 Table 7.3 Average values of the six structural parameters from the C P data for 2+2 B E N Z - d 6 / P X Y - d y Z S M - 5 at 270 K 191 Table 7.4 Average values of the six structural parameters from the C P drain data for 2+2 B E N Z - d 6 / P X Y - d y Z S M - 5 at 270 K 191 Table 7.5 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 B E N Z - d y P X Y - d y Z S M - 5 at 293 K by C P 197 Table 7.6 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 BENZ-de /PXY-dyzSM-5 at 293 K by C P drain 197 Table 7.7 Average values of the six structural parameters from the C P data for 2+2 B E N Z - d 6 / P X Y - d y Z S M - 5 at 293 K 198 Table 7.8 Average values of the six structural parameters from the C P drain data for 2+2 BENZ-dyPXY-de/ZSM-5 at 293 K 198 Table 7.9 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 BENZ-dyPXY-d 4 / ZSM-5 at 270 K by C P 204 Table 7.10 Ranges used for the parameters in the structure calculation and the number of solutions after each step for 2+2 BENZ-d 6 /PXY-d 4 / ZSM-5 at 270 K by C P drain 204 Table 7.11 Average values of the six structural parameters from the C P data for 2+2 BENZ-d 6 /PXY-d 4 / ZSM-5 at 270 K 205 Table 7.12 Average values of the six structural parameters from the C P drain data for 2+2 BENZ-d 6 /PXY-d 4 / ZSM-5 at 270 K 205 Table 7.13 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 BENZ-dyPXY-d 4 / ZSM-5 at 293 K by C P 210 Table 7.14 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 BENZ-dyPXY-d 4 / ZSM-5 at 293 K by C P drain 211 Table 7.15 Average values of the six structural parameters from the C P data for 2+2 BENZ-d 6 /PXY-d 4 / ZSM-5 at 293 K 211 Table 7.16 Average values of the six structural parameters from the C P drain data for 2+2 BENZ-d 6 /PXY-d 4 / ZSM-5 at 293 K..." 211 Table 7.17 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K by C P 217 Table 7.18 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K by C P drain 218 Table 7.19 Average values of the six structural parameters from the C P data for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K 218 Table 7.20 Average values of the six structural parameters from the C P drain data for 2+2 . B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K 218 Table 7.21 Ranges used for the parameters in the structure calculation and the numbers of solutions after each step for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 293 K by C P 224 Table 7.22 Ranges used for the parameters in the structure calculation and the numbers of solutions after each Step for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 293 K by C P drain 224 Table 7.23 Average values of the six structural parameters from the C P data for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 293 K 225 Table 7.24 Average values of the six structural parameters from the C P drain data for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 293 K 225 Table A.1 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-d 6 for the average location in ZSM-5 with r2 > 0.92 determined from the 1 H / 2 9 S i C P at 273 K 243 Table A.2 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-d 6 for the average location in ZSM-5 with r2 > 0.92 determined from the 1 H / 2 9 S i C P at 315 K 244 Table A.3 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-d 4 for the average location in ZSM-5 with r2 > 0.72 determined from the 1 H / 2 9 S i C P at 273 K 244 Table A.4 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-d 4 for the average location in ZSM-5 with r2 > 0.72 determined from the 1 H / 2 9 S i C P at 315 K 244 Table A.5 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-d 6 for the average location in ZSM-5 with r2 > 0.72 determined from the 1 H / 2 9 S i C P drain at 273 K 245 Table A.6 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-d 6 for the average location in ZSM-5 with r2 > 0.82 determined from the 1 H / 2 9 S i C P drain at 315 K 245 Table A.7 Atomic fractional coordinates and error ellipsoid parameters of the p-dicyanobenzene for the average location in 4DCNB/ZSM-5 with t 1 > 0.93 determined from the 1 H / 2 9 S i C P at 305 K •. '. 246 Table A.8 Atomic fractional coordinates and error ellipsoid parameters of the p-dicyanobenzene for the average location in 4DCNB/ZSM-5 with r2 > 0.93 determined from the 1 H / 2 9 S i C P drain at 305 K ; : 246 Table A.9 Atomic fractional coordinates and error ellipsoid parameters of the p-dicyanobenzene for the average location in 2DCNB/ZSM-5 with r2 > 0.93 determined from the 1 H / 2 9 S i C P at 300 K 247 x Table A.10 Atomic fractional coordinates and error ellipsoid parameters of the p-dicyanobenzene for the average location in 2DCNB/ZSM-5 with r2 > 0.93 determined from the 1 H / 2 9 S i C P drain at 300 K 247 Table A.11 Atomic fractional coordinates and error ellipsoid parameters of the p-dinitrobenzene for the average location in 2DNB/ZSM-5 with r2 > 0.94 determined from the 1 H / 2 9 S i C P at 330 K : '. 248 Table A. 12 Atomic fractional coordinates and error ellipsoid parameters of the p-dinitrobenzene for the average location in 2DNB/ZSM-5 with r2 > 0.92 determined from the 1 H / 2 9 S i C P drain at 330 K 248 Table A.13 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d 6 for the average location in 2+2 BENZ-de/PXY-de/ZSM-5 with r2 > 0.62 determined from the 1 H / 2 9 S i C P at 270 K 249 Table A.14 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d 6 for the average location in 2+2 BENZ-de/PXY-de/ZSM-5 with r2 > 0.94 determined from the 1 H / 2 9 S i C P drain at 270 K 249 Table A.15 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d 6 for the average location in 2+2 BENZ-de/PXY-de/ZSM-S with r2 > 0.99 determined from the 1 H / 2 9 S i C P at 293 K 250 Table A.16 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d 6 for the average location in 2+2 BENZ-de/PXY-de/ZSM-5 with r2 > 0.99 determined from the 1 H / 2 9 S i C P drain at 293 K 250 Table A. 17 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d 4 for the average location in 2+2 BENZ -de /PXY-d 4 /ZSM-5 with r2 > 0.93 determined from the 1 H / 2 9 S i C P at 270 K 250 Table A. 18 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d 4 for the average location in 2+2 BENZ -de /PXY-d 4 /ZSM-5 with r2 > 0.93 determined from the 1 H / 2 9 S i C P drain at 270 K 251 Table A. 19 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d 4 for the average location in 2+2 BENZ -de /PXY-d 4 /ZSM-5 with' r2 > 0.90 determined from the 1 H / 2 9 S i C P at 293 K 251 Table A.20 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d 4 for the average location in 2+2 BENZ -de /PXY-d 4 /ZSM-5 with r2 > 0.92 determined from the 1 H / 2 9 S i C P drain at 293 K 251 Table A.21 Atomic fractional coordinates and error ellipsoid parameters of the benzene for the average location in 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 with r2 > 0.82 determined from the 1 H / 2 9 S i C P at 270 K 252 Table A.22 Atomic fractional coordinates and error ellipsoid parameters of the benzene for the average location in 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 with r2 > 0.82 determined from the 1 H / 2 9 S i C P drain at 270 K 252 Table A.23 Atomic fractional coordinates and error ellipsoid parameters of the benzene for the average location in 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 with r2 > 0.92 determined from the 1 H / 2 9 S i C P at 293 K 252 Table A.24 Atomic fractional coordinates and error ellipsoid parameters of the benzene for the average location in 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 with r2 > 0.92 determined from the 1 H / 2 9 S i C P drain at 293 K 253 xi Table B.1 Interatomic distances (A) and angles (degree) for o-xylene-d 1 0 /ZSM-5 254 Table B.2 Crystal data and structure refinement for 4DCNB/ZSM-5 256 Table B.3 Atomic coordinates (x 10 4) and equivalent anisotropic displacement parameters (A 2 x 10 3) f o r4DCNB/ZSM-5 256 Table B.4 Bond lengths [A] and angles [degree] for 4DCNB/ZSM-5 257 Table B.5 Crystal data and structure refinement for 2DCNB/ZSM-5 259 Table B.6 Atomic coordinates (x 10 4), equivalent anisotropic displacement parameters (A 2 x 103) and isotropic parameters (A 2 x 103) for 2DCNB/ZSM-5 259 Table B.7 Bond lengths [A] and angles [degree] for 2DCNB/ZSM-5 260 Table B.8 Crystal data and structure refinement for 2DNB/ZSM-5 262 Table B.9 Atomic coordinates (x 10 4) and equivalent anisotropic displacement parameters (A 2 x 10 3 ) fo r2DNB/ZSM-5 262 Table B.10 Bond lengths [A] and angles [degree] for 2DNB/ZSM-5 263 List of Figures Figure 1.1 Various zeolites: (a) Zeolite A, (b) Faujasite, (c) Mordenite and (d) ZSM-5 2 Figure 1.2 An example of the shape selectivity of zeolites: selective molecular sieving by the zeolite channel in zeolite A 3 Figure 1.3 (a) Simplified representation of ZSM-5 showing the straight channels running vertically and zigzag channels running horizontally (b) A more elaborate representation of ZSM-5 showing its channel intersection with the caged structure 6 Figure 1.4 A schematic description of the dipolar interaction between I and S spins with respect to the external magnetic field B 0 in the laboratory frame 8 Figure 1.5 Powder lineshapes for chemical shift anisotropy tensors, (d) Schematic representation of orientation dependence of each resonance in 1 C NMR of a carbonyl group with respect to the external field B0 10 Figure 1.6 (a) Illustration of the magic angle spinning experiment (b) 1 3 C NMR spectra of the bisacetonide 4,4'-bis[(2,3-dihydroxypropyl)oxy]benzil showing the different effects of magic angle spinning and high power decoupling on the spectra 12 Figure 1.7 Two common ways of measuring the 7i relaxation time: (a) saturation recovery and (b) 1 inversion recovery methods 14 Figure 1.8 Measuring the spin-spin relaxation time T2 by the signal from the spin echo of the transverse magnetization in the xy plane 15 Figure 1.9 Measurement of spin-lattice relaxation in the rotating frame (T 1 p) 16 Figure 1.10 General behavior of the relaxation times, Ti, 7"1p and T2 as functions of temperature... 18 Figure 1.11 Two allowed spin transitions for the different energy levels of spin-1 quadrupolar nuclei and the resulting Pake doublet patterns 20 Figure 1.12 Crystallographic crystal systems and the 14 Bravais lattices 23 Figure 1.13 Illustration of Bragg's law to explain the constructive interference of the reflected beams 27 Figure 1.14 Comparison between the direct lattice and reciprocal lattice 28 Figure 1.15 A construction of the Ewald's sphere, shown in 2-dimensional space 28 Figure 1.16 A 3D illustration of a single crystal and its reciprocal lattice with a representation of the Ewald's sphere 29 Figure 1.17 A typical diffraction pattern generated from a crystal structure 29 Figure 1.18 Goniometer and a single crystal X-ray diffractometer setup 30 Figure 1.19 The'bir th 'of the "powder ring" pattern of an imaginary powder sample 35 Figure 1.20 A typical instrumental setup for a Bragg-Brentano geometry powder X-ray diffraction experiment 36 Figure 1.21 Illustration of the neutron travel path (as a wave) in neutron diffraction 40 Figure 1.22 The neutron scattering length of selected atoms 40 Figure 1.23 Comparison between the X-ray and neutron scattering of carbon atoms with respect to sinG/A 41 Figure 1.24 Schematic representation of the NRU reactor and C2 diffractometer, Chalk River, Ontario 41 xiii Figure 1.25 Schematic representation of a powder neutron diffractometer with a monochromator.. 42 Figure 2.1 High resolution Si NMR spectrum of the calcined ZSM-5 sample ( 'GEB 177') 47 Figure 2.2 Scanning electron microscope (SEM) picture of a typical large 'coffin' shaped ZSM-5 single crystal 48 Figure 2.3 Apparatus used to load o-xylene into ZSM-5 49 Figure 2.4 A typical T G A plot of weight% vs. temperature 50 Figure 2.5 'Magic angle' adjustment using the 1 2 7 l rotational echo 'spikes' of KI in the frequency domain .' 52 Figure 2.6 Variable temperature setup for the solid state NMR experiments 55 Figure 2.7 Reference sample Q8M8 and its 2 9 S i NMR spectrum at room temperature 56 Figure 2.8 A home-built capillary-glass ampoule for precise loading of volatile organics into a single zeolite crystal 61 Figure 2.9 A custom-built powder X R D sample holder made of polymethylmethacrylate (PMMA) for volatile organics loaded into a zeolite powder 62 Figure 3.1 Pulse sequence for a general C P experiment 69 Figure 3.2 Pulse sequence for a C P drain experiment 70 Figure 3.3 Flowchart of the structure determination procedure used by the program in this study... 74 Figure 3.4 A rigid model of a para disubstituted benzene guest molecule having its long axis defined by the two atoms which are furthest apart and its orientation in terms of Euler angles 74 Figure 3.5 A flowchart of the powder diffraction strategy from the sample preparation to the structure refinement 78 Figure 3.6 The algorithm of FOX, which uses simulated annealing in order to evaluate the cost function 81 Figure 3.7 A schematic view of parallel tempering implemented by FOX 81 Figure 4.1 Effect of N 2 purging on the 2 9 S i spectrum of o-xylene/ZSM-5 88 Figure 4.2 2 9 S i MAS NMR spectra of ZSM-5 loaded with ca. 4 molecules of o-xylene per unit cell at the temperatures indicated 91 Figure 4.3 1 H / 2 9 S i C P MAS NMR spectra of ZSM-5 loaded with ca. 4 molecules of o-xylene per unit cell (a) at 273 K (b) at 315 K 92 Figure 4.4 Two-dimensional 1 H / 2 9 S i C P INADEQUATE spectra of the o-xylene/ZSM-5 complex at 273 K and 315 K 92 Figure 4.5 The two specifically deuterated o-xylenes used 94 Figure 4.6 Variable contact time 2 9 S i C P MAS NMR experiment on ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 273 K 95 Figure 4.7 Intensities of the 2 9 S i C P MAS NMR signals indicated as functions of the contact time for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 273 K fitted with independently measured 1 H 7"1p 96 xiv y Figure 4.8 Variable contact time Si C P MAS NMR experiment on ZSM-5 loaded with ca. 4 molecules of o-xylene-d6 per unit cell at 315 K 97 Figure 4.9 A graphical representation of the Euler angles used to describe the orientation of the o-xylene-d 6 molecule 100 Figure 4.10 Distributions of the structural parameters for the solutions determined from 1 H / 2 9 S i C P data of o-xylene-d 6 in the framework of ZSM-5 at 273 K and 315 K 101 Figure 4.11 Scatter plots of o-xylene molecules in the plane of the molecule for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell 102 Figure 4.12 3D scatter plot at r2 > 0.92 of o-xylene molecules in ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 273 K 102 Figure 4.13 Plot of the measured C P rate constants against the calculated heteronuclear second moments for the average location of o-xylene in ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell with r2 > 0.92 at 273 K and 315 K 103 Figure 4.14 NMR spectra for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 273 K a n d 315 K 103 Figure 4.15 NMR determined structures of the o-xylene-de/ZSM-5 complex at 273 K and 315 K.. 105 Figure 4.16 NMR determined structure of o-xylene-de/ZSM-5 complex at 273 K showing the o-xylene with the framework silicon atoms labeled 106 Figure 4.17 Variable contact time 2 9 S i C P MAS NMR experiment at 273 K for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 4 per unit cell 107 Figure 4.18 Intensities of the 2 9 S i C P MAS NMR signals indicated as functions of the contact time at 315 K for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 4 per unit cell 107 Figure 4.19 Distributions of the solutions determined from 1 H / 2 9 S i C P data of o-xylene-d 4 in the framework of ZSM-5 at 273 K a n d at 315 K 110 Figure 4.20 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average location of o-xylene for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 4 per unit cell 110 Figure 4.21 NMR spectra for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 4 per unit cell at 273 K 111 Figure 4.22 Scatter plots of the acceptable o-xylene locations in the framework of ZSM-5 for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 4 per unit cell at 273 K and at 315 K 111 Figure 4.23 NMR determined structures of the o-xylene-d 4/ZSM-5 complex at 273 K and 315 K from the C P data 113 Figure 4.24 Intensities of the 1 H / 2 9 S i C P drain NMR signals indicated as functions of the drain contact time for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 273 K... 114 Figure 4.25 Intensities of the 1 H / 2 9 S i C P drain NMR signals indicated as functions of the contact time for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 315 K 115 Figure 4.26 Distributions of the solutions determined from 1 H / 2 9 S i C P drain data of o-xylene-d 6 in the framework of ZSM-5 at 273 K a n d at 315 K... 118 Figure 4.27 Plots of the measured C P rate constants, /c,s from the C P drain experiments against the calculated hetronuclear second moments for the average locations of o-xylene for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 273 K and at 315 K 119 Figure 4.28 NMR spectra for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 273 K from C P drain 119 X V Figure 4.29 NMR spectra for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 315 K from C P drain 120 Figure 4.30 Scatter plots of o-xylene molecules in the plane of the molecule for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 6 per unit cell at 273 K and at 315 K from C P drain data ...... '. 120 Figure 4.31 NMR determined structures of the o-xylene-de/ZSM-5 complex at 273 K and 315 K from the C P drain data. 122 Figure 5.1 Powder neutron scattering profiles for the sample containing 3.7 molecules of o-xylene per unit cell of ZSM-5 at 272 K 131 Figure 5.2 Difference Fourier map of the neutron scattering density due to the o-xylene-d 1 0 molecule at the channel intersection of the framework of ZSM-5 133 Figure 5.3 Powder neutron diffraction determined structure of o-xylene-di 0 /ZSM-5 complex at 272 K showing the two symmetry eqivalent o-xylene molecules 133 Figure 5.4 Powder neutron diffraction determined structure of o-xylene-d 1 0 /ZSM-5 complex at 272 K 134 Figure 5.5 The maximum and minimum oxygen-to-oxygen distances of the 10 member elliptical ring from the straight channel view of the o-xylene-d 1 0 /ZSM-5 complex 134 Figure 5.6 Powder neutron diffraction determined structure of the o-xylene-d 1 0 /ZSM-5 complex at 272 K superimposed on the NMR determined structure of o-xylene-d 6 in the framework of ZSM-5 at 273 K 135 Figure 6.1 Structures of p-dicyanobenzene (DCNB) and p-dinitrobenzene (DNB) 136 Figure 6.2 2 9 S i M A S NMR spectra of 4DCNB/ZSM-5 at the different temperatures indicated 138 Figure 6.3 2 9 S i M A S NMR spectrum of 4DCNB/ZSM-5 at 305 K showing the individual deconvoluted resonances : 139 Figure 6.4 Two-dimensional 2 9 S i INADEQUATE spectrum of 4DCNB/ZSM-5 at 305 K together with the quantitative 2 9 S i NMR spectrum 139 Figure 6.5 Intensities of the 1 H / 2 9 S i C P MAS NMR signals indicated as functions of the contact time for 4DCNB/ZSM-5 at 305 K 140 Figure 6.6 Plot of the measured C P rate constants against the calculated heteronuclear second moments for the average location of DCNB for 4DCNB/ZSM-5 at 305 K 143 Figure 6.7 The NMR spectra for 4DCNB/ZSM-5 at 305 K 143 Figure 6.8 Distributions of the solutions determined from 1 H / 2 9 S i C P data of 4DCNB/ZSM-5 at 305 K 143 Figure 6.9 Scatter plot of the DCNB molecule in the plane of the molecule and the 50% error ellipsoid representation for 4DCNB/ZSM-5 at 305 K 144 Figure 6.10 Structure of 4DCNB/ZSM-5 complex from the C P experiment with 50% error ellipsoidal representations of the D C N B molecule at 305 K 144 Figure 6.11 1 H / 2 9 S i C P drain experiment on 4DCNB/ZSM-5 at 305 K 146 Figure 6.12 Plot of the measured C P drain rate constants against the calculated heteronuclear second moments for the average location of D C N B for 4DCNB/ZSM-5 at 305 K 147 Figure 6.13 NMR spectra fo r4DCNB/ZSM-5 at 305 K 148 xvi Figure 6.14 Distributions of the solutions determined from 1 H / 2 9 S i C P drain data of 4DCNB/ZSM-5 at 305 K. 148 Figure 6.15 Scatter plot of DCNB molecule in the plane of the molecule and the 50% error ellipsoid representation for 4DCNB/ZSM-5 at 305 K 148 Figure 6.16 C P drain determined structure of 4DCNB/ZSM-5 complex with 50% error ellipsoidal representations of the D C N B molecule at 305 K 149 Figure 6.17 Fourier electron density difference map of the 4DCNB/ZSM-5 for the D C N B molecule found at the channel intersection of ZSM-5 150 Figure 6.18 Structure of the 4DCNB/ZSM-5 complex determined by single crystal X R D 151 Figure 6.19 2 9 S i M A S NMR spectra of 2DCNB/ZSM-5 at the different temperatures indicated 153 Figure 6.20 Qantitative 2 9 S i spectrum of 2DCNB/ZSM-5 at 300 K 154 Figure 6.21 Two-dimensional 2 9 S i INADEQUATE spectrum of 2DCNB/ZSM-5 at 300 K 154 Figure 6.22 Intensities of the 1 H / 2 9 S i C P MAS NMR signals indicated as functions of the contact time for 2DCNB/ZSM-5 at 300 K 155 Figure 6.23 1 H / 2 9 S i C P drain experiments on 2DCNB/ZSM-5 at 300 K 155 Figure 6.24 Plot of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2DCNB/ZSM-5 from C P and C P drain data 158 Figure 6.25 NMR spectra for 2DCNB/ZSM-5 at 300 K 158 Figure 6.26 Distributions of the solutions determined from 1 H / 2 9 S i C P and C P drain data on the 2DCNB/ZSM-5 complex at 300 K 159 Figure 6.27 Scatter plots of the DCNB molecule and its error ellipsoid representation for 2DCNB/ZSM-5 from the C P experiments at 305 K 159 Figure 6.28 Scatter plots of the DCNB molecule and its error ellipsoid representation for 2DCNB/ZSM-5 from the C P drain experiments at 305 K..... 159 Figure 6.29 Structures of the 2DCNB/ZSM-5 complex determined from C P and C P drain experiments at 300 K... 160 Figure 6.30 Fourier electron density difference map of the DCNB sites for 2DCNB/ZSM-5 found at the channel intersection of ZSM-5 161 Figure 6.31 Structrue of the 2DCNB/ZSM-5 complex determined by single crystal X R D 161 Figure 6.32 2 9 S i M A S NMR spectra of 2DNB/ZSM-5 at the different temperatures indicated 164 Figure 6.33 Qantitative 2 9 S i M A S spectra of 2DNB/ZSM-5 at 330 K 164 Figure 6.34 Two-dimensional 2 9 S i INADEQUATE spectrum of the 2DNB/ZSM-5 complex at 330 K 165 Figure 6.35 Intensities of the 1 H - 2 9 S i C P MAS NMR signals indicated as functions of the contact time for 2DNB/ZSM-5 at 330 K 165 Figure 6.36 1 H / 2 9 S i C P drain experiments on 2DNB/ZSM-5 at 330 K 166 Figure 6.37 Distribution of the solutions determined from C P and C P drain data of 2DNB/ZSM-5 at 330 K 168 Figure 6.38 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2DNB/ZSM-5 for C P and C P drain data .169 Figure 6.39 2 9 S i M A S NMR spectra for 2DNB/ZSM-5 at 330 K 169 xvii Figure 6.40 Scatter plot of DNB molecules and the error ellipsoid representation for 2DNB/ZSM-5 by C P at 330 K 170 Figure 6.41 Scatter plot of DNB molecules and the error ellipsoid representation for 2DNB/ZSM-5 by C P drain at 330 K 170 Figure 6.42 NMR determined structures of the 2DNB/ZSM-5 complex determined by C P and C P drain at 300 K ,-. 170 Figure 6.43 Fourier electron density difference map of the DNB molecule sites found at the channel intersection of ZSM-5 171 Figure 6.44 Structure of the 2DNB/ZSM-5 complex determined by single crystal X R D . . . 172 29 Figure 7.1 Si NMR spectra of benzene in ZSM-5 at room temperature at the loadings indicated 180 Figure 7.2 2 9 S i MAS NMR spectra of p-xylene in ZSM-5 at room temperature at the loadings indicated...- 180 Figure 7.3 2 9 S i INADEQUATE spectrum of p-xylene per unit cell in ZSM-5 at a loading of 2 molecules per unit cell at 300 K 1 181 Figure 7.4 2 9 S i M A S NMR spectra of 2+2 BENZ /PXY/ZSM-5 at the different temperatures indicated 184 Figure 7.5 Quantitative 2 9 S i MAS NMR spectra of 2+2 B E N Z / P X Y / Z S M - 5 at 270 K and 293 K 184 Figure 7.6 Deconvoluted C P spectra of the 2+2 BENZ-de/PXY-de/ZSM-S at 270 K and at 293 K.. 185 Figure 7.7 Two-dimensional 2 9 S i INADEQUATE spectrum of 2+2 B E N Z / P X Y / Z S M - 5 at 270 K 185 Figure 7.8 Two-dimensional 2 9 S i INADEQUATE spectrum of 2+2 B E N Z / P X Y / Z S M - 5 at 293 K 186 Figure 7.9 The three mixtures of the guest organics used in this study: 2+2 BENZ-d6/PXY-d 6 , 2+2 B E N Z - d 6 / P X Y - d 4 and 2+2 B E N Z / P X Y - d 1 0 in ZSM-5 186 Figure 7.10 Intensities of the 1 H / 2 9 S i C P MAS NMR signals indicated as functions of the contact time for 2+2 BENZ-de/PXY-de/ZSM-5 at 270 K 189 Figure 7.11 Intensities of the 1 H / 2 9 S i C P drain MAS NMR signals indicated as functions of the contact time for 2+2 BENZ-de/PXY-de/ZSM-S at 270 K 189 Figure 7.12 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 BENZ-de/PXY-de/ZSM-5 at 270 K for C P and C P drain data 191 Figure 7.13 The NMR spectra for 2+2 BENZ-de/PXY-de/ZSM-5 at 270 K for C P and C P drain 192 Figure 7.14 Distribution of the solutions determined from 1 H / 2 9 S i C P and C P drain data of 2+2 . BENZ-de/PXY-de/ZSM-5 at 270 K 193 Figure 7.15 Scatter plot of the P X Y - d 6 molecules and its error ellipsoid representation for 2+2 BENZ-de/PXY-de/ZSM-5 from C P at 270 K 193 Figure 7.16 Scatter plot of the P X Y - d 6 molecules and its error ellipsoid representation for 2+2 BENZ-de/PXY-de/ZSM-5 from C P drain at 270 K 193 Figure 7.17 The NMR determined locations of the P X Y - d 6 in the BENZ-de/PXY-de/ZSM-5 complex at 270 K from C P and C P drain experiments : 194 Figure 7.18 Intensities of the - 1 H/ 2 9 Si C P MAS NMR signals indicated as functions of the contact time for 2+2 BENZ-de/PXY-de/ZSM-5 at 293 K. 196 xviii Figure 7.19 Intensities of the 1 H / 2 9 S i C P drain MAS NMR signals indicated as functions of the contact time for 2+2 BENZ-de /PXY-de /ZSM-5 at 293 K 196 Figure 7.20 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 BENZ-de /PXY-de /ZSM-5 at 293 K for (a) C P and C P drain data 198 Figure 7.21 NMR spectra for 2+2 BENZ -d 6 /PXY-d6/ZSM-5 at 293 K for C P and C P drain 199 Figure 7.22 Distributions of the solutions determined from 1 H / 2 9 S i C P and C P drain data of 2+2 BENZ -d 6 /PXY-d6/ZSM-5 at 270 K 200 Figure 7.23 Scatter plot of the P X Y - d 6 molecules and its error ellipsoid representation for 2+2 BENZ -d 6 /PXY-d6/ZSM-5 from C P at 293 K 200 Figure 7.24 Scatter plot of the P X Y - d 6 molecules and its error ellipsoid representation for 2+2 BENZ -d 6 /PXY-d6/ZSM-5 from C P drain at 293 K 200 Figure '7.25 Final structures of the 2+2 BENZ-de /PXY-de /ZSM-S complex at 293 K from C P and C P drain experiments 201 Figure 7.26 Intensities of the 1 H / 2 9 S i C P MAS NMR signals indicated as functions of the contact time for 2+2 BENZ - d e /PXY - d 4 /ZSM-5 at 270 K 203 Figure 7.27 Intensities of the 1 H / 2 9 S i C P drain MAS NMR signals indicated as functions of the contact time for 2+2 BENZ - d e /PXY - d 4 /ZSM-5 at 270 K 203 Figure 7.28 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 BENZ-de/PXY - d 4 /ZSM-5 at 270 K from C P and C P drain experiments 205 Figure 7.29 NMR spectra for 2+2 B E N Z - d 6 / P X Y - d 4 / Z S M - 5 at 270 K for C P and for C P drain 206 Figure 7.30 Distribution of the solutions determined from 1 H / 2 9 S i C P and C P drain data for 2+2 B E N Z - d 6 / P X Y - d 4 / Z S M - 5 at 270 K 207 Figure 7.31 Scatter plots of the P X Y - d 4 molecules and its error ellipsoid representation for 2+2 B E N Z - d 6 / P X Y - d 4 / Z S M - 5 from C P at 270 K 207 Figure 7.32 Scatter plots of the P X Y - d 4 molecules and its error ellipsoid representation for 2+2 B E N Z - d 6 / P X Y - d 4 / Z S M - 5 from C P drain at 270 K 207 Figure 7.33 The final structures of the 2+2 BENZ - d e /PXY - d 4 /ZSM-5 complex at 270 K from C P and C P drain experiments 208 Figure 7.34 Intensities of the 1 H / 2 9 S i C P MAS NMR signals indicated as functions of the contact time for 2+2 BENZ - d e /PXY - d 4 /ZSM-5 at 293 K 209 Figure 7.35 Intensities of the 1 H / 2 9 S i C P drain MAS NMR signals indicated as functions of the contact time for 2+2 BENZ-de/PXY -d 4 /ZSM-5 at 293 K 210 Figure 7.36 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 BENZ - d e /PXY - d 4 /ZSM-5 at 293 K for C P and C P drain data 212 Figure 7.37 NMR spectra for 2+2 BENZ - d e /PXY - d 4 /ZSM-5 at 270 K from C P and C P drain experiments 212 Figure 7.38 Distribution of the solutions determined from the C P and C P drain data for 2+2 B E N Z - d 6 / P X Y - d 4 / Z S M - 5 at 293 K 213 Figure 7.39 Scatter Plot of the P X Y - d 4 molecules and its error ellipsoid representation for 2+2 B E N Z - d 6 / P X Y - d 4 / Z S M - 5 from C P at 293 K 213 Figure 7.40 Scatter plot of the P X Y - d 4 molecules and its error ellipsoid representation for 2+2 B E N Z - d 6 / P X Y - d 4 / Z S M - 5 from C P drain experiments at 293 K 214 xix Figure 7.41 Final structure of the 2+2 BENZ-d<5/PXY-d 4/ZSM-5 complex at 293 K by C P and C P drain experiments 214 Figure 7.42 Intensities of the 1 H / 2 9 S i C P M A S NMR signals indicated as functions of the contact time for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K 216 Figure 7.43 Intensities of the 1 H / 2 9 S i C P drain MAS NMR signals indicated as functions of the contact time for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K 217 Figure 7.44 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K from C P and C P drain experiments 219 Figure 7.45 The NMR spectra for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K for C P and C P drain 219 Figure 7.46 Distributions of the solutions determined from 1 H / 2 9 S i C P and C P drain data of 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K 220 Figure 7.47 Scatter plot of the B E N Z molecules and its error ellipsoid representation for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 from C P experiments at 270 K 220 Figure 7.48 Scatter plot of the B E N Z molecules and its error ellipsoid representation for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 from C P drain experiments at 270 K 221 Figure 7.49 Final structure of the 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 complex at 270 K from C P and C P drain experiments 221 Figure 7.50 Intensities of the 1 H / 2 9 S i C P MAS NMR signals indicated as functions of the contact time for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 293 K 223 Figure 7.51 Intensities of the 1 H / 2 9 S i C P drain MAS NMR signals indicated as functions of the contact time for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 293 K 223 Figure 7.52 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 293 Kf rom C P and C P drain experiments 225 Figure 7.53 NMR spectra for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 270 K for C P and C P drain 226 Figure 7.54 Distribution of the solutions determined from 1 H / 2 9 S i C P and C P drain data of 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 at 293 K 227 Figure 7.55 Scatter plot of the B E N Z molecules and its error ellipsoid representation for 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 from C P at 293 K : 227 Figure 7.56 Scatter plot of the B E N Z molecules and its error ellipsoid representation for 2+2 , B E N Z / P X Y - d 1 0 / Z S M - 5 from C P drain at 293 K 227 Figure 7.57 The final structure of the 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 complex at 293 K from C P and C P drain experiments 228 Figure 7.58 Static 2 H spectra of the three different mixtures in ZSM-5 indicated at 293 K 230 Figure 7.59 Simulated 2 H static spectra for deuterated p-xylenes and benzene... 230 Figure 7.60 1 H M A S spectrum of 2+2 B E N Z / P X Y - d 4 / Z S M - 5 at 293 K 232 Figure 7.61 2D 1 H N O E S Y spectrum of the 2+2 B E N Z / P X Y - d 4 / Z S M - 5 complex at 293 K 232 Figure 7.62 NMR structure of the 2+2 BENZ /PXY/ZSM-5 at 293 K showing the B E N Z and P X Y located along the straight channel at adjacent interactions 233 xx Symbols and Abbreviations a , b, C crystallographic unit cell dimensions (lattice constants) a . U . arbitrary unit(s) A, Ahki real part of structure factor b scattering length for an atom in neutron scattering BENZ benzene bis magnitude of dipolar coupling between / and S spins bn Fourier components of the time-dependent dipolar coupling under M A S {n = 0, ± 1 , So static magnetic field of an NMR spectrometer 61 radio frequency magnetic field applied to spins Bhki imaginary part of structure factor CCD . charge coupled device CF cost function COSY Correlation Spectroscopy CP Cross Polarization CSA chemical shift anisotropy dn n number of deuterium atoms in a molecule d(Mi) spacing between lattice planes (in Bragg's Law) d/s dipolar coupling constant between spins / and S (radians per second) DCNB p-dicyanobenzene DNB p-dinitrobenzene DS Divergence slit E-,1, E22, e t c . elements of error ellipsoid tensors Ea activation energy F the structure factor of X-ray diffraction f i , f2 frequency domains in a two-dimensional NMR experiment fj atomic scattering factor of atom / Fhki structure factor for reflection hkl FID free induction decay FWHM the full width at half maximum h hour(s) h Planck constant (6.6262 x 10 ' 3 4 J s) T) Planck constants = hlln) hkl Miller indices for a reflection (hkl) Miller indices for a lattice plane H Hamiltonian Hz Hertz / unobserved spins, magnetic quantum number associated with / spins l0 scaling factor in C P curves lhki intensity of reflection hkl i .d. inner diameter INADEQUATE Incredible Natural Abundance Double Quantum Experiment INEPT Incredible Natural Abundance Polarization Transfer Experiment k Boltzmann constant (1.3807 x 10" 3 4 J K"1) k exchange rate constant kt, ks rate constant for spin-lock relaxation for the / and S spins. kts (absolute) cross polarization rate constant obtained from C P drain experiments k'is (relative) cross polarization rate constant obtained from C P curves kIS predicted C P rate constant based on linear regression between kis and M2 L liter(s) m spin state of a nucleus (m = ±1/2 for spin-1/2 nuclei) m number of rotor cycles after coherence transfer in T E D O R experiment M matrix to convert between Cartesian and fractional coordinates M2 heteronuclear dipolar coupling second moment M2(ll) homonuclear dipolar coupling second moment Mo initial magnetization in relaxation time experiments MSL magnetization in spin lock experiment for measuring 7 i p relaxation time MX magnetization in spin echo experiment for measuring T2 relaxation time MZ magnetization in saturation or inversion recovery experiments for measuring Ti MAS magic angle spinning MC Monte Carlo MFI framework topology code for ZSM-5 , Silicalite-1 mL milliliter mol./u.C. molecules per unit cell n Fourier index, spinning sideband number (n = 0, ± 1 , ±2) NMR Nuclear Magnetic Resonance NOESY Nuclear Overhauser Enhancement Spectroscopy O.d . outer diameter 0-, m-, p-xylene ortho, meta, para- xylene OXY o-xylene PMMA polymethylmethacrylate ppm parts per million (chemical shift) PT parallel tempering PXY p-xylene r internuclear distance l2 degree of linear correlation R gas constant (8.3145 J mol" 1 K"1) R residual factor in X R D R(§,Q,\\i) three dimensional rotation matrix R1 conventional R factor for X R D refinements against F RB Bragg residual in powder diffraction Rp profile residual in powder diffraction Rwp weighted profile residual in powder diffraction r.f. radio frequency REDOR Rotational Echo Double Resonance RS Receiver slit S second(s) S observed spins, magnetic quantum number associated with S spins So reference experiment for R E D O R or C P drain Sd 'drain'experiment for C P drain AS/So normalized difference intensity for C P drain or R E D O R Sep signal intensity in C P experiment SA simulated annealing ScS Scatter slit S{l} or l/S S spin observed, / spin unobserved in C P and C P drain experiments S/N signal to noise ratio Si/AI silicon to aluminum ratio SDR standard deviation of regression xxii SiC-4/2 tetrahedral silicon SoS Soller slit t time U,h time domains in two-dimensional NMR experiments t(a/2) f-value from Student's t distribution (1-<x confidence) T temperature Ti spin-lattice relaxation time T2 spin-spin or transverse relaxation time T2* time constant for decay of FID (including inhomogeneous effect) 7v spin-lattice relaxation time in the rotating frame TEDOR Transferred Echo Double Resonance TGA thermal gravimetric analysis TMS tetramethylsilane TPA tetrapropylammonium wR2 weighted R value for refinements against F2 XRD X-ray diffraction x, y, z fractional coordinates of the ring center of sorbate molecules the observed intensity of the i t h data point in the Rietveld refinement y calc the calculated intensity of the i , h data point in the Rietveld refinement W watt(s) 2D (2-D), 3D (3-D) two-dimensional, three-dimensional O the minimization function in the Rietveld refinement a, (3 polar angles describing orientation of l-S internuclear vector a, (3, y crystallographic unit cell angles (lattice constants) phase of reflection hkl Y/. Ys magnetogyric ratios of spin / and spin S A stepsize for data collection in diffraction experiments 0"ll, CJ22, CT33 principal elements of the chemical shift anisotropy tensor ACT, CTan/So anisotropy parameter of the CSA tensor C/so isotropic chemical shift «|),e,v|/ Euler rotation angles describing the orientation of sorbate molecules dephasing angle in REDOR experiment asymmetry parameter of the CSA tensor angle between l-S internuclear vector and magnetic field wavelength permitivity of free space (4n x 10'7 kg m s'2 A2) v 0 Larmor frequency (Hz) 'magic angle' (54.74°) p(*.y.z) electron density at point x, y, z T delay time, incremented pulse time correlation time x 2 . goodness of fit in powder diffraction xxiii Acknowledgements I would like to express my perpetual gratitude to the faculty, staff and my fellow students at the University of British Columbia, who inspired me to continue my study in this field. The person, to whom I owe my achievement most notably, is my supervisor, Professor Colin A. Fyfe. He has been i -an intellectual hammer and Idea itself for my study; without him, I would have never imagined science of this kind existed in this universe; without him, I would have never sharpened, myself to get organized and prepared for upcoming attacks of constructive criticisms and alternative explanations. From his passion, knowledge and devotion, I have learned a lifetime lesson that I shall never forget. Next, I would like to thank Dr. Darren H. Brouwer, whose work has been the predecessor of a major part of this thesis. It was his enduring effort and achievement that I have been 'benchmarking' throughout my study and that I am still pursuing without seeing the end. To date, he has been one of the most efficient people I have met in performing any types of tasks, and I am a true admirer of such a quality among many of his. Dr. Andrew R. Lewis should not be coming any lesser than the two mentioned; it was his introductory teaching that got me into the field of solid-state NMR. I still keep my notebook that contains his first instruction about our solid-state NMR spectrometer, which has been behaving without a 'major' incident. Also I thank his encouragement on my study so that I can be always as optimistic as possible even during questionable hours. Over years, I have become an aficionado of his intellectual flamboyancy, which has helped me look at a bright side in my 'rainy' days in Vancouver. I thank everyone from the electronic engineering shop and mechanical engineering shop, and Mr. Brian Ditchburn from the glass shop, who have helped me to set up and maintain the spectrometer and related equipment over my study; especially Mr. Milan Coschizza and Mr. Dave Tonkin from the electronic engineering shop deserve special recognition for their knowledge and know-how that they have accumulated into a form of art, which has been proven,to be priceless. I should also pay my gratitude to Dr. Brian O. Patrick and Ms. Anita Lam from the X-ray Crystallographic Services, and Professor Mati Raudsepp from the Department of Earth and Ocean Sciences for introducing me into the divine and intricate world of X-ray crystallography. All the Fyfe group members, past and present, should be mentioned for their parts in my study and graduate life in general including Drs. Florin Marica, Franziska Scheffler, Celine Schneider, Richard Darton, Jerry Brethethon, Mr. Glenn Wong and Ms. Emily Ng, and the visitors from all over the world including Drs. Andreas Stein and Midori Kasai. I would like to express special appreciation to Professor Mark MacLachlan and Mr. Jonathan Chong for providing me their T G A instrument and instruction for the sample analyses, which has been a crucial part of my research. I also thank Professor Elliot E. Burnell for discussion on NMR in general including deuterium experiments. Among many other people outside of the U B C community, I would like to thank Mr. Lachlan Cranswick and Dr. Ian P. Swainson, who invited me to their powder neutron diffraction workshop at xxiv Waterloo, Ontario and to the C N B C at Chalk River, Ontario, where Canada's flagship neutron facility is located. People mentioned above are the bare minimum to whom I owe my sincerest gratitude. If I have forgotten any of your name, please take my apology; it is not because you have been lesser important to my research, but because I am chased by my deadlines. For my family, I would like to thank my wife, Keiko; in fact I am dedicating this work to her, as she has been my better half throughout this thesis (other than that I can not find any other way ,to thank her). Without her enduring dedication and inspiration, I would have been utterly and helplessly lost in space and time. Next, I would like to pay my tribute to my parents, who have been supporting me emotionally over years. I also thank my friends and loved ones for their parts in shaping moments in my life. I shall remember my late grandmother, who did not live long enough to see this achievement of mine. I am quite sure that she is watching this moment with God in heaven, who has guided me to become what I am today. Finally, I would like to pay my humble respects to everyone whose work has been a part in my thesis, especially, again, Professor Colin Fyfe, Drs. Yi Fang, Hiltrud Grondey, Anix Diaz, Andrew Lewis and Darren Brouwer from Fyfe group, who have worked and struggled on the very same subject that stretches over a decade. In addition to their efforts and dedications, I am honored to be a 'beta tester' of their pioneering ideas and achievements. xxv Dedication to Keiko xxvi Chapter 1 INTRODUCTION: General Background This chapter provides a general introduction to the materials and methods that were used for the studies in the thesis, including information about crystals, zeolites, solid-state NMR spectroscopy, X-ray diffraction and powder neutron diffraction. 1.1 Zeolites Zeolites are three-dimensional, crystalline microporous aluminosilicate materials, which can be of either natural or synthetic origin. The word zeolite comes from Greek "to boil" (zein) and "a stone" (lithos), thus meaning 'boiling stone'. The Swedish mineralogist Axel Fredrick Cronstedt coined the word in 1756 after discovering frothing of certain silicate minerals on heating in a borax bead. The general chemical formula of a zeolite can be expressed as: (M z + ) x / z [ (S i0 2 ) y (A I0 2 r j - n H 2 0 Equation 1.1 where M represents interchannel cations (Na + , K + , C a 2 + , B a 2 + , Sr 2 * , M g 2 \ F e 2 + , etc), z is their charge, n is the number of moles of channel molecular water, and x and y are the stoichiometric coefficients for A l 3 + and S i 4 + respectively in the corner sharing A I0 4 and S i 0 4 tetrahedral sites in a zeolite. Unique features of zeolites are their well-ordered intricate framework structures that contain pores and channels, ranging between 3 and 13 A in diameter. As depicted by the four common zeolites in Figure 1.1, there are many different types of zeolite framework structures, each designated by three letter codes. The basic building blocks of zeolites are represented as T 0 4 units where T can be silicon or aluminium tetrahedrally connected to the oxygen atoms. The positions of the T atoms in the framework of a zeolite are known as "T-sites" which are linked to the adjacent T-sites via shared oxygen atoms. 1 (a) LTA (Zeolite A) (b) FAU (Faujasite, Zeolite Y) (c) MOR (Mordenite) (d) MFI (ZSM-5, Silicalite-1) Figure 1.1 Various types of zeolites: (a) Zeolite A, (b) Faujasite, (c) Mordenite and (d) ZSM-5 (Silicalite-1). The three letter codes are commonly used to represent each type of zeolite framework and are defined by the International Zeolite Associat ion 1 . Figure adapted from reference 2. 1.1.1 A p p l i c a t i o n s of zeo l i t es The first commercial applications of zeolites were as drying agents for refrigerant gas and natural gas. Union Carbide commercialized synthetic zeolites (Zeolite A is one of those) as separation and purification agents in 1954, introducing the important industrial applications of zeolites. These 3 " 1 3 are for ion exchange, gas adsorption, industrial catalysts, molecular sieves, and recently for possible nano materials with various purposes 1 4 " 1 6 . Zeolite Y, Faujasite, ZSM-5 and mordenite are especially important in the petroleum industry for hydrocarbon cracking, isomerization, and fuel synthesis 6. A well-known example of ion exchange is zeolites as water softeners. The calcium in water is the cause of hardness that can form scale and cause other problems. By allowing the calcium ions in the hard water to exchange with sodium ions in the frameworks, zeolites "soften" the water. Another important application is for gas absorption, which exploits the porous property of zeolites. The open frameworks of zeolites are capable of taking and holding organic chemicals that give odors and toxic effects. This application is widely used in laundry detergents, pet litters and livestock feeds that require various degrees of odor and toxin control. In construction, zeolites are mixed with cement materials to yield a concrete mixture. Zeolites have high thermal stability and strong structural integrity, and yet are very light because of their internal pores. Zeolite concrete provides a strong yet light material for building construction, which is ideal especially for earthquake prone parts of the world. 2 As seen from the examples above, many applications of zeolites exploit their unique porous nature and structural integrity. In fact, the key to understanding the unique properties of zeolites comes from the interaction between the non-framework species (guest species) and the framework of the zeolite. Figure 1.2 An example of the shape selectivity of zeolites: selective molecular sieving by the zeolite channel in zeolite A, which enables n-octane to enter the channel while isooctane is blocked. Figure adapted from reference 3. 1.1.2 Zeo l i te s y n t h e s i s There are about 50 natural zeolites such as analcime, chabazite, heulandite, stilbite, mordenite, and ferrierite. The natural zeolites are formed by the reaction of mineral-rich aqueous solutions with aluminosilicates, which are abundant minerals in the earth's crust. In order for the natural formation of zeolites to occur, the three main prerequisites are the composition of the host rock species and an alkaline environment of interstitial solutions (ca. pH 10), long enough times for formation (50 - 50,000 years), and relatively high temperature (up to -100 °C). Since naturally occurring zeolites contain heavy metals and other impurities, they are not as preferable as a pure synthetic zeolite in many industrial processes. Synthetic zeolites have become popular since their industrial importance was discovered. To date, there are about 150 synthetic zeolites known, and most of those are waiting for future applications. Zeolite A, X and Y with Si/AI ratios between 1.0 and 2.5 were first synthesized by R. M. Milton and D. W. Breck at the Linde Division of Union Carbide Corporation between 1949 and 1954. In the 1960s, Mobil Oil reported the synthesis of zeolites beta and ZSM-5 , which are highly siliceous zeolites. A typical synthesis of a 3 silicious zeolite is via a sol-gel process, which involves mixing silica gel, alumina and a template material in a basic aqueous solution, and heating the mixture in a stainless steel autoclave at temperatures up to 200 °C for several days. 1.1.3 Zeolite characterization Characterization of a zeolite aims to describe its morphology, chemical composition, crystal structure, catalytic ability and adsorption capacity. Common techniques for zeolite characterization include NMR spectroscopy, X-ray diffraction, electron microscopy, IR and Raman spectroscopies 1 7, thermal gravimetric analysis (TGA), and temperature programmed desorption. Among these techniques, NMR spectroscopy and X-ray diffraction are the most powerful in determining the crystalline framework structures of zeolites. These two are complementary to each other because NMR probes the local order of the sample whereas X-ray diffraction is sensitive to long range order. Powder X-ray diffraction (XRD) is a powerful way to determine and check zeolite structures as each crystalline zeolite has its own characteristic X-ray diffraction pattern. In fact, the first reported structure of the synthetic zeolite (ZSM-5) was from powder X R D . 1 8 Through quantitative phase analysis, the composition of a mixture of crystalline materials can be verified as well. In most cases, zeolites cannot be obtained in the form of large crystals that are needed for single crystal diffraction studies. There are a few exceptions to this, and single crystal XRD studies have been done in these 19-23 cases. NMR spectroscopy has proven to be a powerful tool in the characterization of zeolites, as it is capable of providing structural information, especially the number of unique silicon T-sites, which can be used to deduce space groups, and framework connectivities in many studies 2 4" 3 3 . In characterizing chemical composion of zeolites, solid-state NMR spectroscopy can be useful to verify the Si/AI ratio in zeolites 3 4" 3 6, which is important for catalytic activity in industrial applications. 4 1.1.4 Zeolite lattice structures The structures of some zeolites are known to undergo various space group changes upon sorbate loading and temperature changes. The framework T 0 4 tetrahedra are quite rigid; however T -O - T bond angles can vary between 90 and 150 degrees in general. The best-known example of structural changes is ZSM-5 , which changes its space group from monoclinic to different orthorhombic space groups when loaded with different organic molecules and to different numbers of molecules per unit cell of the framework. 3 7 Temperature plays an important role as well, and affects space group changes from a high temperature orthorhombic form to the monoclinic form at room temperature. 1 9 , 2 0 ' 3 7 1.1.5 Zeolite ZSM-5 ZSM-5 was first synthesized by Mobil Corporation in 1969, and attracted considerable attention because of its commercial applications: for example, the gasoline synthesis (MTG or methanol-to-gasoline process) and xylene conversion (a Mobil process). The MTG process 3 8 was first discovered by Chang and Silvestry at Mobil Corporation, and was developed commercially in the midst of the Energy Crisis of the 1970s. The optimum environment for the conversion is at 371 °C for about an hour, where methanol is sequentially converted to dimethyl ether, light olefins, and then finally to a mixture of paraffins and aromatics, which is commonly known as gasoline. Although it has not been widely popularized due to its less desirable economics compared to fossil fuels, the process has potential as the raw material (methane) can be obtained from renewable sources, by fermentation of organic matter for example. The xylene conversion Mobil process 3 9 , on the other hand, has been widely used because of the enduring demands for p-xylehe, a raw material to several important industrial products, such as fibers and films. It requires the highly acidic H-ZSM-5 to protonate the aromatic nucleus. Feeding a mixture of o-, m- and p-xylenes through a bed of H-ZSM-5 results in a final product that is over 99% p-xylene. The Mobil Process takes place at a temperature of about 140°C, which is considerably lower than the temperature at which isomerization occurs without presence of a catalyst, which is ca. 400°C. 5 ZSM-5 usually has a high silica-to-alumina ratio (-99); however, the small population of aluminium in the framework plays a critical role by providing one proton per atom of aluminium, which makes ZSM-5 acidic. Another interesting characteristic of ZSM-5 is that it has two different channel systems as shown in Figure 1.3. The straight channels, which are made of 10 membered rings of silicon atoms, are 5 - 5.5 A in effective diameter. The zigzag channels are also made of 10 membered rings and have similar diameters. Compared to other similar types of zeolites, ZSM-5 is unique for having the two different channels, which provide selectivity towards certain organic molecules. For example, it is known that p-xylene can occupy both the straight and zigzag channels whereas o- and m-xylenes can only occupy the straight channels of ZSM-5 , indicating shape selectivity. Figure 1.3 (a) Simplified representation of ZSM-5 showing the straight channels running vertically and zigzag channels running horizontally. The red and green ellipsoids are the possible locations of the guest molecules from various s t u d i e s 2 5 , 2 6 , 4 0 ' 5 4 . Figure adapted from reference 3. (b) A more elaborate representation of ZSM-5 showing its channel intersection with the caged structure made by connecting the silicon atoms (oxygen atoms are omitted for clarity; the red line represents the opening of the straight channel and the blue one that of the zigzag channel). 1.2 Solid-state NMR Nuclear magnetic resonance (NMR) spectroscopy 2 , 5 5 " 6 3 is one of the most important techniques in modern chemistry for the investigation of structures and dynamics of various materials. NMR spectroscopy depends on the interactions of the nuclear spins of atoms with magnetic fields, and yields in-depth information of the local environments of nuclei, which can be used to determine structures and dynamic behavior. 6 Table 1.1 shows the list of the five main interactions which must be considered for liquid-state and solid-state NMR. In liquid state NMR spectroscopy, molecules tumble, and that effectively removes the dipolar coupling and averages the chemical shift anisotropy and the quadrupolar interaction (the latter for the spin > Vi nuclei). However, in solid-state NMR spectroscopy, these three effects can be present, leading to a broadening of the signals which can be very severe. Table 1.1 Typical magnitudes of nuclear spin interactions for common nuclei in a 9;4 Tesla magnetic field. Type of Nuclear Spin Interaction Description of Interaction Magnitude in Solid State NMR Magnitude in Liquid State NNR Zeeman with static magnetic field 5 0 - 4 0 0 MHz 50 - 400 MHz Magnetic Shielding (Chemical Shift) with local electronic environment up to several kHz isotropic value Dipolar Coupling through-space interactions with neighboring nuclei up to - 30 kHz 0 Indirect Spin-Spin Coupling (J-coupling) through-bond interactions with neighboring nuclei 0 - 200 Hz 0 - 200 Hz Quadrupolar interaction with electric field gradient for spins > Vi up to - 100 MHz 0 1.2.1 The dipolar coupling The dipolar coupling arises due to dipole-dipole interactions between the magnetic moments of the observed nucleus and those of neighboring nuclei. The interaction is inversely proportional to r3, where r is the internuclear distance and is independent of the external magnetic field B 0 . Important dipolar interactions for this study are 1 H - 1 H , 1 H - 2 9 S i , 1 H - 1 3 C and 1 9 F - 2 9 S i . The homonuclear dipolar interactions of nuclei with low natural abundance (e.g. 2 9 S i - 2 9 S i or 1 3 C - 1 3 C ) are usually neglected in unlabelled samples because of their low probability and relatively large internuclear distances. The Hamiltonian for the heteronuclear dipolar coupling between nuclear spins / and S can be expressed as: His = - d ( S ( 3 c o s 20 - 1 ) / z S 2 Equation 1.2 where the d/ S is the dipolar coupling constant of the heteronuclear dipolar coupling between spin / and S in frequency unit (Equation 1.3), G is the angle between the internuclear vector between the two 7 spins and the applied static magnetic field (Figure 1.4), and lz and S z are the operators for the z components of the spin / and S respectively: d/s = Voir/sMAnr3 (in rad s"1) or D, s = dlsl2n= vtfWshlQn2? (in Hz) Equation 1.3 where u 0 is the permitivity of free space (4TI X 10"7 kg m s"2), yt and y s are the gyromagnetic ratios of the / and S spins, h = h/2n where h (6.6262 x 10' 3 4) is Planck's constant and r is the distance between / and S spins in meters. From Equations 1.2 and 1.3, the heteronuclear dipolar coupling depends on the orientations of the spin pairs in the field as well as the interspin distance. In the liquid state, where the spins exhibit rapid isotropic tumbling, the dipolar coupling is averaged to zero. In the solid state, however, the dipolar coupling is present giving up to 30 kHz line broadening of the 1 H spectrum in many organic solids. Figure 1.4 A schematic description of the dipolar interaction between / and S spins, which are ns apart, with respect to the external magnetic field 6 0 in the laboratory frame. 1.2.2 The chemical shielding anisotropy The chemical shielding anisotropy (CSA) is due to the asymmetric electronic charge distribution around the NMR nucleus. In liquid samples the effect is averaged into a single peak due to random tumbling of the observed nuclei. In solid samples, however, the nuclei can exist in all the possible orientations, which generates many peaks of different resonance frequencies resulting in broad patterns. This interaction is linearly proportional to the external magnetic field, 6 0 . The orientation dependent C S A can be quantitatively described by Equation 1.4 where a,- is a second-rank tensor (a 3 x 3 matrix), describing the interaction of two vectors, / and B 0 : 8 = y/ • cjj • 6Q Equation 1.4 where / is the operator for the magnitude of the nuclear spin, and 6 0 is the static magnetic field of the NMR spectrometer. It is possible to choose the axis frame that a, can be defined only with the diagonal elements (o"n, O22, O33). In this so-called principal axis system, the C S A of rapid tumbling molecules in a,liquid sample becomes averaged to its isotropic value a , s o (Equation 1.5), and a single NMR resonance is observed at the isotropic chemical shift aiso, which is the average over the diagonal elements: °"iso = g(°"l1 +°22 +a33) ACT = CT33 - cr iso Equation 1.5 _ CT22 - C T ^ ACT where ACT is the anisotropy parameter (aan;So) and r\ is the asymmetry parameter of the C S A tensor. Figure 1.5 shows examples of powder patterns with different C S A tensors. The powder patterns occur due to the different tensors describing the different resonance frequency of each molecular orientation with respect to B0. In solid-state NMR, a technique called 'magic angle spinning' can remove (or reduce) the effect of C S A on the sample as described later. 9 G • Figure 1.5 Powder lineshapes for (a) isotropic (cubic symmetry; an = 022= 033), (b) axially symmetric, and (c) asymmetric (less than 3-fold symmetry) chemical shift tensors, (d) Schematic representation of orientation dependence of each resonance in 1 3 C NMR of a carbonyl group with respect to the external field So. Figures (a), (b) and (c) adapted from reference 2. 1.2.3 The q u a d r u p o l a r in te rac t ion The quadrupolar interaction in NMR spectroscopy is due to the interaction of a nuclear electric quadrupole moment with the non-spherically symmetrical electric field gradient of the nucleus that has nuclear spin quantum number / > Vt (e.g. 1 1 B , 1 7 0 , 2 3 N a , 2 7AI). The interaction can range up to several MHz and the interaction can dominate the NMR spectrum completely for quadrupolar nuclei. Of the three interactions that give line broadening in NMR spectrum, the quadrupolar interaction is often the largest and most dominant, and can range up to -100 MHz. 10 1.2.4 The cross polarization technique NMR of dilute spin V2 nuclei suffers from low sensitivity and long recycle delays (due to long Ti relaxation times) in general. Increasing the sensitivity of the NMR signal of the dilute spins can be achieved by a cross polarization (CP) experiment 6 4. C P exploits the net magnetization transfer from abundant / spins ( 1H, 1 9 F , etc.) to the dilute S spins ( 1 3 C, 1 5 N , 2 9 S i , etc.), which are observed. The maximum overall S/N enhancement is the ratio of the magnetogyric ratios of the two spins y / y s (e.g. 4 for 1 H - 1 3 C and 5 for 1 H- 2 9 S i ) . Further, the recycle delay for the C P experiment is determined by the 7, of the abundant / spins, which is often considerably shorter than that of the dilute spins. Overall, these lead to an improved S/N in the final C P spectra. More detailed discussions regarding C P experiments are presented in Chapter 3. 1.2.5 'Magic-angle' spinning and high power decoupling The 'magic-angle' spinning technique 6 5" 6 7 is applied in order to remove the effects of C S A and the heteronuclear dipolar couplings in solid samples. The technique 'mimics' the isotropic nature of the spins in the solution state by removing the orientation dependent term present in both the C S A and heteronuclear dipolar coupling (Equation 1.6). <3cos2G - 1) = 1 /2(3cos 2 9 m - 1 )(3cos2B - 1) Equation 1.6 where 0 is the angle of the spin interaction tensor with respect to the external field, S 0 , Qm is the angle between the external field and spinning axis, and B is the angle between the tensor and the spinning axis (Figure 1.6). When 0m is 54.74°, which is known as the magic angle, the whole term becomes zero, which applies to both the C S A and heteronuclear dipolar interactions. In theory, both interactions become the isotropic values when magnetic spins are placed at the magic angle with respect to the static magnetic field S 0 . In reality, in order to remove these interactions, samples have to be finely powdered and the spinning of the samples has to reach a certain speed. Otherwise, the spectrum would exhibit a manifold of spinning side bands, which are the result of incomplete removal of the chemical shift anisotropy interactions in the sample. In solid-state NMR, the dipolar coupling interaction between two (or more) different types of 11 spins can be removed by high power decoupling, i.e., application of an on-resonance r.f. field to the nucleus being 'decoupled'. In practice, high power decoupling is used to remove the heteronuclear dipolar coupling to the abundant nucleus in a sample where the rare spin ( 1 3 C, 2 9 S i , etc.) is dilute enough so that its homonuclear dipolar coupling can be neglected. The strength of the applied r.f. field should be large for the frequency of the transition between the two spin states of the abundant spin, which has to be larger than the frequency of the heteronuclear dipolar coupling. The high power needed is one of the significant differences between solid-state NMR experiments and solution NMR experiments, and solid-state NMR equipment requires larger power capacity, often several hundreds of watts, much more than solution NMR (ca. 50 watts). ( a ) (b) ^ - ^ - © - u ^ j i i_ i : 1 1 1 1 (3cos26-l)=0, if 6„ =54.74° 2 5 0 2 0 0 1 6 0 J™^., 6 0 0 - 5 0 - 1 0 0 Figure 1.6 (a) Illustration of the magic angle spinning experiment. The term, (3cos28-1), is the average of the orientation dependence of the spin interaction, which can be further divided into two other terms. When 0 m is 54.74° (so-called 'magic angle'), the (3cos28-1) term becomes zero. The figure was adapted from references 63 and 68. (b) 1 3 C NMR spectra of the bisacetonide 4,4'-bis[(2,3-dihydroxypropyl)oxy]benzil showing the different effects of magic angle spinning and high power decoupling on the spectra. Adapted from reference 69. 12 1.2.6 NMR relaxation mechanisms Knowing and assessing accurate relaxation parameters in NMR are very important in understanding nuclear spin interactions as seen in many parts of this study. The NMR relaxation occurs as a perturbed spin system returns to its equilibrium population level. Nuclear spin transitions are stimulated by fluctuations of local magnetic fields resulting from motions. Such stimulations come from the surrounding environment including shielding anisotropy, dipole-dipole interactions, quadrupolar interactions and interactions with unpaired electrons. 1.2.6.1 Spin-lattice relaxation time (7"i) The spin-lattice relaxation time, 7i, describes the time for a net magnetization to return to the original equilibrium state after an r.f. pulse in the presence of the external field S 0 . There are two main ways to measure the spin relaxation times of a sample as described in Figure 1.7. Both measure the time of the net magnetization returning to the full magnetization along the z-axis. The rate of the relaxation process can be described as a first order kinetic process (Equation 1.7): 6MJ6t = -{Mz- M0)lTi Equation 1.7 where Mz is the measured magnetization at a given time t, M0 is the magnetization at equilibrium and T-i is the spin-lattice relaxation time which is the reciprocal of the first order rate constant. 13 180°(x) 90°(x) © • © ® | © © © © M (T )=M„ (1 -e T / r ' ) _ © © Mix) (a) © © © M ( x ) = / W 0 ( 1 - 2 e T / r j „ r _ . Figure 1.7 Two common ways of measuring the 7"i relaxation time are (a) saturation recovery and (b) inversion recovery methods, (a).In saturation recovery, applying multiple (n) 90° pulses (1) makes the net magnetization = 0 (2), and after x, a 90 ° pulse on the x direction is applied to produce the FID, which is observed, (b) In inversion recovery, a 180 ° pulse pn the x direction is applied (1, 2), and after T, a 90 ° pulse on the x direction is applied to produce the FID, which is observed. A plot of the intensity curve is produced and the Ti value is determined as indicated. 1.2.6.2 Spin-spin relaxation time (7"2) T2, the "spin-spin" relaxation or "transverse relaxation" time describes the rate at which the net magnetization is lost in the xy plane after a pulse; i.e., regains its equilibrium value, zero. After a 90° pulse along the x-axis, the magnetization dephases due to spin-spin relaxation and magnetic field inhomogeneities, which results in T2' dephasing. In order to remove the effect of field inhomogeneities, a spin-echo type experiment can be employed to measure the true 7"2, which is the result of spin-spin relaxation alone (Figure 1.8). The process can again be described as a first order kinetic process (Equation 1.8): dMJ6t = -MJT2 Equation 1.8 where Mx is the measured magnetization at a given time t. 14 90°(x) 1I2 T* decay 180°(y) Spin echo T2 decay 5 g t = 0 2 direction . _ reversed * - " t = 2T M(X) Figure 1.8 Measuring the spin-spin relaxation time T2 by the signal from the spin echo of the transverse magnetization in the xy plane. After applying a 90 ° pulse in the x direction (1, 2), T is applied causing T2 decay (not observed). After applying a 180 0 pulse in the y direction (3, 4), the dephasing spins are reversed in their directions with respect to the y-axis. After another x, the spin echo is formed from the magnetization refocusing along the y-axis (5, 6), which is observed as FID. The FIDs in the gray shade are not observed throughout the experiment. The T2 is determined by measuring the intensity decay as described. 1.2.6.3 Spin-lattice relaxation time in the rotating frame (7"1p) Tip, "the spin lattice relaxation time in the rotating frame", is a relaxation time measuring dissipation of the observed magnetization in the xy plane during the presence of a 'spin-locked' r.f. field, In the spin-locking state along the y-axis, the spins precess about in the rotating frame rather than decaying due to transverse relaxation. In the rotating frame, the spin-locking B, field acts as a static magnetic field, providing the relaxation that is similar to 7V It is a valuable tool to study 15 molecular motions, as it is closer to T2 values than to T<\ values. is normally sensitive to high frequency motions, whereas the T2 is sensitive to low frequency motions. Thus,. 7"1p, which measures a magnetic dephasing with respect to the xy plane, is sensitive to much lower frequency motions than Ti and provides complementary information to 7"i and T2. Figure 1.9 shows the standard T i p measurement. The experimental technique called a 'spin-locking' pulse (Bi), which is applied along the y-axis after the initial 90 0 pulse on the x-axis. The spins are then 'spin-locked' and precess around Bi in the rotating frame x-axis and do not decay as T2 or T2. The spin-locking pulse, Bi acts as a static field in the rotating frame. Thus, the 7"1p relaxation mechanism is analogous to conventional spin-lattice relaxation (7"i); hence r i p is called 'the spin-lattice relaxation time in the rotating frame'. T 1 p is a very important parameter during a cross polarization experiment as spin-locking pulses are applied during the polarization transfer step. Figure 1.9 Measurement of spin-lattice relaxation in the rotating frame (Ti p). A 90° pulse is applied on the x-axis (1), and the B i along the y-axis is applied to make the spin precess and fan out along the y-axis (2, 3). After the x period, the net spin magnetization decay along the y-axis is measured and the FID is observed (4). The Fourier transformed decay is used to determine the T i p as indicated. 16 1.2.6.4 Motion and temperature dependence of relaxation times In general, the relaxation times (Ti, T2 and 7"1p) are strongly dependent on molecular motions. where C is a second moment parameter constant, T c is the correlation time for the motion and (o0 is the Larmor frequency of the spin. The correlation time T, the average time between reorientation motions, is assumed to obey a pseudo Arrhenius type equation of the form: where £ a is the activation energy, R is the gas constant and T is the temperature of the system. Figure 1.10 shows the behaviors of the relaxation parameters, 7V T2 and 7"1p. Variation of Ti with temperature is dependent on the spectrometer frequency, and at lower frequencies, the minimum of the 7i curve is shifted toward lower temperatures as implied in Equation 1.9. The spin-lattice relaxation time Ti represents the time between the excited and the equilibrium spin state as mentioned earlier. Like other relaxation parameters, the spin relaxation time is an empirical value that can be obtained only by measurement. Nevertheless, it has a significant role in explaining the spin behavior and the environment surrounding the sample. The 7"1p relaxation times exhibit similar tendency to 7V The T2 relaxation times, however, keep decreasing as temperature is lowered until the molecular motions freeze (at T c = oo). . In solution NMR, 7V 7"2 and 7"1p are equal in the high temperature regime due to rapid tumbling of the spins as shown in the left side of the curve in Figure 1.10. In solid-state NMR, however, different 7V 7"2 and T 1 p are usual due to different T c values for different motions. This can be shown from the relationship in Equation 1.9 for 7"i as an example. Equation 1.9 T = T 0 exp(-EJRT) Equation 1.10 17 Figure 1.10 The general behavior of the relaxation times, T\, T i p and Ti as functions of temperature. Adapted from reference 2. 1.2.7 Second moments Two experimental variables of importance in characterizing the broad peaks of the static solid-state NMR are the linewidth and the second moment. The latter quantity, defined mathematically as the mean square width of the NMR peak, was first calculated theoretically by van Vleck without knowing the energy eigenstates and eigenvalues 7 0 . The homonuclear second moment for the abundant nuclei (e.g. 1 H or 1 9 F) in a rigid lattice in a static sample is given by a pairwise summation of internuclear interactions over the lattice (Equation 1.11).56 where u 0 is the permeability constant, I is the spin quantum number, r,j is the internuclear distance between two spins /',;', and 0j is the angle formed between the internuclear distance vector and the external magnetic field S 0 . For a polycrystalline sample, the term (1 - 3cos20j)2 is replaced by 4/5, its average over all orientations, giving the following 6 8: For the heteronuclear second moment, the situation becomes much more complex. There are several ca lcu la t ions 2 , 5 6 , 6 8 , 7 1 , 7 2 which show the heteronuclear second moment of two types of spins Equation 1.11 Equation 1.12 18 in the polycrystalline material, that are all variations from the original calculation by van Vleck. For this study, a simplified expression from previous work 2 5 of the heteronuclear second moment for the / and S spins both spin V* was used (Equation 1.13). It is important to note that a second moment can be calculated theoretically once the lattice positions of the nuclei are known. 1.2.8 Internuclear distance measurements from dipolar couplings The internuclear dipolar couplings in solid-state NMR can be both a blessing and a curse. In a two spin heteronuclear coupled system, the doublet feature of the powder line shape known as a Pake doublet73 arises due to the energy level difference between the spin transitions for a static sample. The splitting of the two 'zeniths' is equal to the dipolar coupling constant d / s , which can yield a distance between the two spins. However, as mentioned in the earlier section, MAS removes the dipolar coupling which otherwise leads to broadening of the spectra meanwhile the information regarding the dipolar coupling is lost. Thus, reintroducing the dipolar coupling is a key to obtaining successful results in a distance measuring experiment. Since the dipolar coupling constant is related to the internuclear distance, it can be used to determine structures by solid-state NMR, where this structure determinations may not be possible by diffraction techniques for a variety of reasons. Among the heteronuclear dipolar 'recoupling' experiments, Rotational Echo Double Resonance ( R E D O R ) 7 4 and Transferred Echo Double Resonance ( T E D O R ) 7 5 have been shown to measure nuclear distances by solid-state NMR. The two experiments have been shown to be useful in determining the spin distances in a framework of zeolite type materials while C P can also be used to measure internuclear distances when isolated spin pairs exist. 7 6" 7 8 Equation 1.13 19 1.2.9 D y n a m i c s t u d i e s by 2 H so l i d -s ta te N M R Molecular motions in solid materials provide valuable information, as they are often responsible for many properties such as reactivities, diffusivities and structures. In a solid, all nuclear spins have anisotropic interactions, which are orientation-dependent within the NMR magnetic field. The NMR interactions of the spins are governed by the five interactions described in Table 1.1. For nuclei with spin > V2, quadrupolar interaction is present. 2 H is a spin 1 nucleus with a relatively small electric quadrupole moment (Q = 2.8 x 10"31 m2), whose spectra yields quadrupole-coupling constants*, % < 300 kHz, which can be observed with ease as they dominate the spectra. The quadrupolar Hamiltonian, H Q , affects the Zeeman energy levels, which are no longer equally spaced (Figure 1.11) leading two different energy levels. The possible spin transitions are from +1 <-> 0 and 0 <-> -1, and thus give doublet patterns, Pake doublets for quadrupolar nuclei, which are separated by % % at their peaks. The powder pattern lineshapes are easily influenced by molecular motions of the order of 104 - 106 Hz, and can be easily detected by commercial NMR spectrometers. 3y Z E E M A N Figure 1.11 Two allowed transitions for the different energy levels of spin-1 quadrupolar nuclei (left) and the resulting Pake doublet patterns (right). ' Quadrupole-coupling constant is defined as: x - o2qQlh, where eQ is the electric quadrupole moment of deuteron, and eq (= Vzz) represents the zz component of the electric field gradient (EFG) tensor in its principal axis system. 20 1.2.10 So l i d - s ta te N M R a n d zeo l i t es NMR can probe almost all the different types of atoms present in zeolite frameworks. These atoms include 1 1 B , 1 7 0 , 2 7 A l , 2 9 S i , 3 1 P and 7 1 G a , and the chemical shift in 1-D MAS NMR gives information on the local environment of these nuclei such as chemical compositions. For example, the 2 9 S i NMR spectra of aluminosilicate-type zeolites yields the fractional compositions of 2 9 S i with 0, 1, 2, 3 or 4 2 7AI nuclei in the first coordination sphere, and thus can be used to determine the framework Si/AI ratio of the zeol i te. 3 4" 3 6 High resolution 2 9 S i NMR spectra of highly siliceous zeolite reveal the number and occupancy of the crystallographically unique silicon atoms, aiding in determining the crystallographic space group of the zeolite. The crystallographic phase changes induced by changes in temperature or sorbate molecule loading have been studied 3 3 ' 3 7 in detail by high resolution 2 9 S i MAS NMR as well. 2-D NMR experiments can yield a great deal of the structural information of a microporous material by probing the connectivities between the nuclei in the framework such as the Si -O-Si connectivities in highly siliceous zeolites by 2-D 2 9 S i INADEQUATE exper iments 2 9 , 7 9 and A l -O-P connectivities in microporous aluminophosphate materials by 2-D 2 7 AI / 3 1 P INEPT or C P experiments 8 0. Guest species in zeolites have been studied by solid-state NMR; the structure and dynamics of various organic sorbate and template molecules have been investigated by 1 H , 2 H , 1 3 C and 1 5 N NMR exper iments 2 5 , 2 6 , 4 5 , 4 6 , 8 1 . The acid sites in catalytically active zeolites and adsorbed water can be studied by 1 H NMR experiments. Studies of 1 9 F , 2 3 N a , 6 L i / 7 L i and 1 3 3 C s NMR have yielded information about the locations, dynamics and degrees of hydration of ionic guest species. 1 2 9 X e NMR can be employed to probe microporous materials because the 1 2 9 X e chemical shift of adsorbed Xe gas is sensitive to pore structure, pore size and the presence of other guest species. In this study, NMR experiments depending on the dipolar coupling between two spin species were used to calculate the structures of organic/zeolite complexes. Also, J-coupling based 2-D INADEQUATE experiments 8 2 on 2 9 S i in the zeolite framework were used to determine the Si-O-Si connectivities. 21 1.3 Crystallography In chemistry, a crystal is a solid that has long range structural order in all three dimensions, and a crystalline material refers to a collection of atoms, molecules or ions in a crystal structure. A lattice refers to a set of infinitely repeating points in space, which have repeating orders that can be reduced to a minimum repeating unit of the crystal order, known as a unit cell. Crystallography has become the most important tool for the structure determination of crystalline materials since W. L. Bragg showed that a crystal structure could be determined by X-ray diffraction. . The following describes some of the basic concepts behind crystallography and structure determination by crystallographic techniques. 1.3.1 Crystal system and Bravais lattices Depending on their axial contributions to the structures, crystals can be divided into seven groups, which are triclinic, monoclinic, orthorhombic, hexagonal, trigonal (rhombohedral), tetragonal and cubic. These groups are known as crystal systems, and they can be further divided into 14 distinct space lattices, so-called Bravais lattices (Figure 1.12). According to the Bravais lattices, seven out of 14 lattices are simple lattices, four are face-centered, and three are body-centered. By introducing the face-centered and body-centered lattices, mathematical calculations involving these lattices become greatly simplified to those using the simple lattices alone. 22 (a) Triclinic a,p,y^90° (T) (b) Monoclinic atb±c a=y, (3>90° (2/m) (mmm) T (c) Orthorhombic a+btc a£b£c a=P=y=90° (d) Tetragonal a=b£c <x=p=y=90° (41 mmm) p ^ (e) Trigonal/ Rhombohedral a=b=c a=p=y^ 90° R (3/m)or (31m) (f) Hexagonal a=/#c a=p=90°,y=120° (61 mmm) p a,p,y?t90o ate 1 ^ 1 (g) Cubic a=b-c a=P=y=90c (m3m) Figure 1.12 Crystallographic crystal systems and the 14 Bravais lattices; P for primitive, C for centered, / for body centered, F for face-centered and R for rhombohedral. The letters in red indicate the lattice symmetries. 23 1.3.2 Point groups and space groups in crystallography Point groups are groups of symmetry operations that consist of rotations and/or reflections. There are infinite possibilities of symmetry operations; however, all the three-dimensional symmetry operations can be reduced to 32 crystallographic point groups. In order to describe the point groups, Schdnflies notation and Hermann-Mauguin notation are commonly used. The Schonflies notation, which is widely used for molecules in chemistry, may be categorized into the following groups: C„: groups having only an n-fold rotation axis ( 'C refers to "Cyclic"). Cnh includes an additional mirror plane perpendicular to the axis of rotation. Cnv includes an additional mirror plane parallel to the axis of rotation. Although they do not have any rotation axes, C, refers to a group that only possesses inversion centers, and C s to a group only has a mirror plane. D„: groups having an n-fold primary rotation axis in addition to n two-fold axes perpendicular to it ('D' refers to "Dihedral"). Dnh includes a mirror plane perpendicular to the n-fold axis of D„. Dnd includes mirror planes diagonal to the n-fold axis of D„. S„: groups having an n-fold rotation-reflection axis ('S' refers to "Spiegelung"). 7": groups having the four three-fold and the three two-fold axes of a regular tetrahedron ( T refers to "Tetrahedral"). Td includes diagonal mirror planes, and Th includes T with an inversion. O: groups having two symmetry of an octahedron. Oh includes an inversion. The Hermann-Mauguin notation is more commonly used in crystallography, and it uses rotations (1, 2, 3, 4 and 6 fold axes), inversion axes ( 3 , 4 and 6 ) and mirror planes (m). The symmetry of the principal axis comes first, followed by the mirror plane associated to the principal axis. There are a total of 32 point groups in the Hermann-Mauguin notation: Triclinic: 1,1 Monoclinic: 2 m 2/ ' ' /m Orthorhombic: 222, mml, mmm Tetragonal: 4, 4,4/ ,422,4mw, 42^ ,4 /m/n Rhombohedral: 3,3,32,3m, 3m 24 Hexagonal: 6, 6, *>/ , 622,6mm, 62m, $/ mm ' '/m' ' . ' '/m Cubic: 23, ml, 432,43m, mlm In addition to the rotations, inversion axes and mirror planes, a crystal system can also have two other unique symmetry operations, which are the combinations of rotation and translation (screw axes), and of reflection and translation (glide planes). There are a total of 230 possible combinations of symmetry elements in crystals, known as space groups, which can be found in many references and tables, most comprehensively, in the International Tables for Crystallography, Volume A 8 3 Table 1.2 Corresponding symbols for symmetry elements in the Schonflies notation and Hermann-Mauguin notation. Schonflies Hermann-Mauguin Rotations c2 2 c3 3 c4 4 c6 6 Inversion axes s6 3 s4 4 s3 6 Mirror planes cr m Inversion i T 25 1.3.3 Diffraction and reciprocal space Diffraction is a phenomenon due to the behavior of electromagnetic radiation as a wave. It has been a well-known phenomenon in optics that light beams are grated into bright and dark regions as they pass through narrow slits, which is due to constructive and destructive interferences of light waves. Diffraction can be observed in the motions of small particles such as electrons and neutrons: as the size and mass of a particle become small, it shows more prominent wave-like behavior following de Broglie's principle. As a typical X-ray has the index of refraction nearly 1 for all materials, it cannot be focused as visible light beams in optical microscopy. * W. L. Bragg introduced the Bragg equation in order to describe an X-ray diffraction phenomenon (Equation 1.14) showing that a simple crystal structure could be solved by its X-ray diffraction pattern. The equation explains necessary conditions to invoke constructive interference of diffracted X-ray beams. It, nonetheless, is pertinent to scattering caused by any types of radiation. Figure 1.13 shows the construction of scattered beam on two lattice planes. Since every diffracted beam can be regarded as the reflection of the incident beam by a lattice plane, the angles (8) of the incident and diffracted beam to the plane must be equal. For constructive interference between two beams, the following equation can describe the conditions: nX = 2 dhki sin 0 Equation 1.14 where n is an integer (n > 1), X is the wavelength of the beam and dhM is the inter-lattice distance. The Bragg equation is a fundamental basis that explains lattice geometry from diffracted patterns. By definition, the reciprocal space lattice is a set of imaginary points, which are the reciprocal distances apart from the direct lattice space of a crystal (Figure 1.14). Although the points in the reciprocal space are the products of mathematical definitions, they are important as some points coincide with the constructively interfered spots by the diffracted beams on the direct lattice. Thus, all the spots from the diffraction are observed at the points of the reciprocal space lattice. The diffracted spots on the reciprocal lattice space can be predicted graphically from the Ewald sphere construction (two-dimensional in Figure 1.15 and three-dimensional in Figure 1.16), which illustrates that 1 Recently, there has been a development of a new technique called x-ray holography, which uses holograms to record both the intensities and the phase information that enables a 3-D projection of a crystal in the real space in references 84, 85. 26 diffractions occur on the surface of the sphere after the constructive interference of the X-rays. The Ewald sphere can be considered as a manifestation of the Bragg equation in the three-dimensional space as the reciprocal lattice distance, d*, can be obtained from the Bragg equation. Figure 1.17 shows a typical X-ray diffraction pattern on the reciprocal lattice space. The spots, which are due to the coherent scattering of X-rays, are important because: 1. The geometry of the spot pattern is related to the lattice and unit cell geometry of the diffracted crystal. 2. The positions of equally intense spots contain the information of the crystal system and space group. 3. The intensities of the individual spots, which are different from each other, hold key information about the positions of the occupied atoms in the crystal structure. Figure 1.13 Illustration of Bragg's law to explain the constructive interference of the reflected beams. 27 Figure 1.14 Comparison between the direct lattice and reciprocal lattice, showing the conversion from one to the other. Figure adapted from reference 86. Figure 1.15 A construction of the Ewald sphere, shown in two-dimensional space. Figure adapted from reference 86. 28 R e c i p r o c a l latt ice x- ray s o u r c e Detec to r E w a l d ' s sphe re Figure 1.16 A 3-D illustration of a single crystal and its reciprocal lattice with a representation of the Ewald sphere. Figure adapted from reference 86. Figure 1.17 A typical diffraction pattern generated from a crystal. The general patterns, and locations and intensities of the spots contain the structural information of the crystal structure. 29 1.3.4 Single crystal XRD Single crystal X-ray diffraction (single crystal XRD) is the most widely used technique to determine the crystal structure, and its successful implementation is highly dependent on the quality of the sample. The efficiency of single crystal X R D depends on the size of a single crystal used, as the diffracted intensities of X-rays are dependent on the number of electrons present in the sample in general. However, the absorption of X-rays by the sample increases exponentially with the size of the crystal, which is also influenced by various factors such as the X-ray wavelength, the chemical composition of the sample and the presence of heavy atoms (the heavier elements, the more absorption). In practice, a preferred size of the sample for single crystal X R D is in the order of a few hundred microns for a laboratory single crystal X-ray diffractometer. Figure 1.18 shows simple illustrations of a single crystal X R D experimental setup along with a goniometer for sample mounting. A goniometer allows accurate mounting and precise adjustments of the sample crystal in a few tenths of millimeters. The crystal can be glued on top of a glass fiber or located in a glass capillary for a volatile sample. In the earlier days of single crystal diffraction, photographic films were used in order to take successive pictures of the diffraction patterns in a systematic fashion. With recent developments of electronics, various types of area detectors such as a charge coupled device (CCD) detector have been used for the detections of X-rays. Figure 1.18 Goniometer and a single crystal X-ray diffractometer setup. The right figure is adapted from reference 87. A crystal is usually glued to the tip of the glass fiber, which sits on the top of the goniometer head. 30 1.3.4.1 Structure calculation from diffraction data As diffracted intensities in the reciprocal lattice space contain the structural information about a crystal system, they can be resolved to construct the crystal structure in the direct space. A wave can be expressed as: F = A +\B F= |F|cos<j> + i|F|sin<)> = \F\ (cos<(> + isincp) F=\F\e'* Equation 1.15 For the diffracted wave: where F(hkl) is the structure factor of the reflection with indices h, k, and /. The X-ray diffraction patterns are results of the Fourier transform (FT) of the electron density. The structure factor, which is a combination of amplitude and phase, for reflection hkl can be expressed as: where p is the electron density at each point in the unit cell. The structure factor is determined by integrating the multiples of electron density and the complex number exp[27ii(nx + ky + lz)] over the whole cell volume V. However, the equation in this form is not very convenient for calculations because of a continuous function p(xyz) and integration. The term, p(xyz), equals to r"(0)exp(-87i2(Jsin20A.2), where r"(0)> is the atomic scattering factor (also known as the form factor), and the exponential term refers to a mean-square amplitude of vibration for the atom (U is the isotropic displacement parameter). Then the Fourier transform of the electron density becomes: F(hkl) = ^ f(0)exp(-87i2t;sin20/X2)exp[27ti(/7X ) *kyj +/zy)] Equation 1.18 The atomic scattering factor f(0) is usually shown as a function of sin0A. and the values of f{Q) are available in reference tables. The values are proportional to the atomic numbers of scattering atoms and decrease as values of sin0/X increase. The electron density is the result of the reverse Fourier transform of the diffraction pattern. Because the diffraction pattern of a crystal is a collection of discrete reflections, the Fourier transform F(hkl) = \F(hkl)\ exp[i<|>(riW)] Equation 1.16 Equation 1.17 31 of the electron density can be expressed as: p(xyz) = —J> ( / 7 W ) e x p [ - 2 7 i i ( / 7 x + kY + / z ) l or p(xyz) = — ^ \F(hkl)\exp[i<|>(/jW)]exp[-2ni(/jx + ky + lz)] Equation 1.19 The amplitude of diffraction can be measured experimentally and the relative phase shifts, exp[-27ii(hx + ky + lz)], can be calculated. However, the intrinsic phases, exp[i<|)(h/c/)], cannot be obtained. Moreover, the intrinsic phases have larger contribution to the structural aspect of X-ray diffraction than the amplitudes. The problem is commonly known as the 'phase problem', and there are several methods to overcome it. Two major approaches for the data obtained by in-lab equipment are the Patterson and direct methods. 1.3.4.2 The Patterson method As previously mentioned, the intrinsic phase changes during scattering cannot be monitored. The loss of phase information during scattering makes the electron density distribution in the unit cell virtually impossible to obtain. The Patterson method, however, gives a possibility to solve this problem. The Patterson function is defined as: where \F0(hkf)\2 is the squared scattering amplitude with all phases set equal to zero. All the information for the Patterson function is known and the Patterson map can be readily calculated. As it may look similar to an electron density map, the Patterson map is a map of vectors between pairs of atoms in the structure. For instance, a pair of atoms at (x^ y i , z ^ and (x 2, y 2 , z 2 ) , the Patterson map produces two peaks at (x, - x 2 , y^ - y 2 , Z\ - z2) and (x 2 - x 1 f y 2 - y i , z 2 - z-i). The Patterson map, therefore, shows relative locations of atoms to each other, not to the unit cell origin. The Patterson method is especially useful when the structure contains a few heavy atoms among a lot of lighter atoms, as a relatively small number of large peaks that are results of the heavy atoms show clearly on the Patterson map. h,k,l Equation 1.20 32 1.3.4.3 Direct methods The term, 'direct methods', refers to general ways to solve the phase problem by using the phase relationships from observed intensities when no other methods are available to solve the single crystal structure by diffraction. The electron density via the Fourier transform of the diffraction patterns is a result of constructive interferences of the scattered waves. This restricts the number of possible phases considerably. The methods start to select prominent reflections and calculate their phases in a trial-and-error method. The calculated phases and the observed amplitudes are used to search for known molecular characteristics. With development of powerful computers, direct methods have become increasingly popular for the methods require a large number of systematic deductions. 1.3.4.4 Refinement of a single crystal structure After most of the atoms in the initial structure have been found, the model structure has to be refined. The refinement involves finding the best agreement between the observed and calculated diffraction patterns. Because no phases can be observed, only the comparison between their amplitudes | F 0 | and |F C | can be made. Any changes made in the structural parameters will affect the |F C | while the |F 0 | values will remain constant. The refinement process uses a least-squares method, which minimizes the difference between |F 0 | and |F C | as: where w = l /o 2 (FO 2 ), a weight based on the experimental standard deviations. The least-squares refinement of crystal structures could become a complex task when several variable parameters need to be fitted. Table 1.3 represents variables commonly fitted during the least-square refinement. In the end, the lower the least-sqaures sum becomes, the better the refinement on the data. The goodness of fit is an indication of improvement in the refinement, which is related to the least-squares sum. In general, the reliability of the crystal structure is indicated by mathematical forms of Equation 1.21 33 differences between |F0| and |FC|. The most widely used residual factor or R-factor can be defined as: where w is the weight of each reflection. Both R^ and wR2 ignore signs of the differences between F 0 and Fc values and normalize the sum of differences in terms of F0 values. Upon completion of the final refinement, a good data refinement would yield typically R, < 0.05 and wR2 < 0.15. Although an incorrect structure can also meet these criteria, so any result must be closely scrutinized by means of structural representations including a Fourier electron density map, thermal ellipsoids of the atoms found and bond angles. Table 1.3 List of refinement variables for each atom and their uses during the refinement. Refinement variables Purpose to determine x, y, z structural parameters L^iso. U^, U22, L/ 3 3, U23, L/ 1 3, LA2 isotropic and anisotropic vibrations sca le factor sca le factor between the |F0| and |FC| va lues site occupancy factor number of occupying atoms in the unit cell Equation 1.22 Another variation uses F2 values instead of |F| values as shown as: Equation 1.23 34 1.3.5 Powder Diffraction In single crystal data, the 3-D structural information is preserved in reciprocal space, and a good data set should yield an unambiguous structure. In powder diffraction, however, the transformation of the 3-D structure is manifested in the form of 1-D data as illustrated in Figure 1.19. With many randomly distributed crystals, the end result of the diffraction data collection is the so-called "powder ring." In principle, the powder ring contains all the same structural information as the single crystal data; however, the information is either obscured or lost when two or more reflections are close to each other in d-spacing (as in the 29 values). Due to the random display of multiple diffraction sets in the data, powder diffraction has limitations in solving crystal structures unambiguously. In most cases, powder diffraction is used to identify the sample because the powder pattern can serve as a 'fingerprint' for unknown substances. In order to determine the structure of a guest/zeolite complex, special care must be taken in collecting and interpreting powder diffraction data. Figure 1.19 The 'birth' of the "powder ring" pattern of an imaginary powder sample. The figures show reflections (a) from a single crystal, (b) 4 grains, (c) 40 grains and (d) several hundred grains of crystals in random orientations, (e) The reflection intensities taken along the radius of the powder' ring show a typical powder diffraction pattern. 35 1.3.5.1 Powder XRD instrumentation The re are numerous instruments for modern X - r a y powder dif fract ion; a m o n g them, the Bragg-Brentano geomet ry for focus ing X - r a y (F igure 1.20) is the mos t popular . T h e orientat ion of the s a m p l e is the foca l point of the diffraction, wh ich de te rmines the gon iomete r geomet ry ; the focus ing c i rc les of the geomet r ies are gove rned by the X - r a y s o u r c e s , the s a m p l e and the detector. Both the X - r a y sou rce and detector rotate in a synch ron i zed fash ion dur ing 8 - 20 or 9 - 9 data co l lect ion. O n e d i sadvan tage of the B ragg -Bren tano focus ing geomet ry is a suscept ib i l i ty to preferred orientation and the preparat ion of the s a m p l e to ensu re its opaci ty to X - r a y s . Figure 1.20 A typical instrumental setup for a Bragg-Brentano geometry powder X-ray diffraction experiment. The angles, 9 and 29 indicate the angles of the X-ray source and the detector respectively. 1.3.5.2 Powder neutron diffraction A neutron is an electr ical ly neutral part icle with a magne t i c m o m e n t of 1.891 p w (spin 1 / 2 ) , where p w is the nuc lear magne ton . Its rest m a s s (m) is 1.675 x 1 0 " 2 7 kg , wh i ch is very s imi lar to that of a proton part ic le. A neutron b e a m c a n travel through the a t m o s p h e r e a n d is sca t te red by nuc le i and magnet ic sp ins of e lec t rons. T h e wave leng ths of neutron b e a m s are dependen t on the kinetic energy of a part icle acco rd ing to de Brogl ie 's re lat ionship (\ = hlmv), w h e r e h is the P l a n c k ' s constant , Goniometer path X-ray source SoS IT Detector 36 m is the rest mass of neutron particle, v is the velocity. A neutron particle traveling at v = 3956 m/s has its wavelength, X « 1 A, similar to the interplanar spacings in a crystal lattice, and thus, can be used as a radiation source in diffraction analysis. Usually, a wavelength of 1 - 2 A is used for diffraction studies. The neutron is scattered by the interactions with the nuclei of an atom via the strong nucleus force, or with the magnetic moments of the unpaired electrons electrostatically. Thus, neutron diffraction is capable of yielding information regarding the arrangement of the atomic nuclei by the nuclear scattering or the ordering of the unpaired electron spin density by the magnetic scattering in the sample. A plane wave of neutrons entering an isolated nucleus fixed at the origin is described by a wave function ^in = exp(kz) Equation 1.24 where K is the amplitude of the wave vector, i.e. 2n/X, and z is taken along the direction of travel of the neutron towards the nucleus. The flux of the incident neutron is u K J 2 Equation 1.25 where v is the speed of the traveling neutron. After interacting with nucleus, the scattered neutron has a spherically symmetrical wave form of ¥ s c = -(o/r) exp(kz) Equation 1.26 where b is the scattering length for the atom, and r is the distance from the nucleus (Figure 1.21). Assuming an elastic scattering (the same wavelength for the incoming and the scattered) and no energy exchange between the neutron and the nucleus, i.e. K is the same before and after scattering, the outgoing current of scattered neutrons integrated over an area is given by 47tr2u|xF s c|2. The scattering cross section, a , of the nucleus is defined as: = 4TC62 Hence, the bigger the value of b for a nucleus, the greater is the probability that an incident neutron a = outgoing current of scattered neutrons incident neutron flux Equation 1.27 37 will be scattered. The nuclear scattering strength, b, depends on the nuclei structure of the atom. As shown in Figure 1.22, the b values vary from one element to another in the fashion that appears to be random. Even for the same element, the b values are different for the isotopes (e.g. 5 8 N i , 6 2 N i and 6 4 Ni) . Compared to the X-ray scattering strength, which varies for the atomic numbers, the neutron scattering strengths show no such dependency toward the magnitude of the atomic numbers. X-rays interact with the electrons of the atom through an electromagnetic interaction, which makes X-ray scatterings sensitive to the number of the electrons and thus to the atomic number of the atom. However, neutrons interact with the nucleus through the strong nuclear force. This makes neutrons sensitive to isotopic composition. As a result, atomic scattering for X-rays is proportional to the atomic numbers, whereas for neutrons it varies in a "random" fashion. Because the range of the strong nuclear force (=» 10" 1 5 m) is small compared to the wavelength of the neutrons (« 10 ' 1 0 m), the neutrons are scattered spherically showing no angular scattering dependence unlike the X-ray scattering (Figure 1.23), which shows rapid declines of the scattering intensities with the diffraction angle, 6 (also known as form factor). The fluctuations in scattering length over an assembly of nuclei give rise to a coherent scattering and an incoherent scattering. The coherent scattering depends on the mean value of the scattering lengths, b, which is the same for all equivalent atomic sites. acoh=47t{b)2 Equation 1.28 The incoherent scattering, however, depends on the random distribution of the scattering deviations from the mean value. Such deviations are due to the wavelength increases from the energy loss of scattered neutrons, and thus the incoherent scattering does not contain any structural information of the sample. In order to minimize random background caused by the incoherent scattering, some atoms are better to be avoided in neutron diffraction studies. For instance, 1 H has the coherent cross section (<jcoh) of 1.8 x 10" 2 8 m 2 whereas its incoherent cross section (a ( n c o/,) is 79.9 x 10" 2 8 m 2 , which contributes very high background intensities that masks the coherent scattering intensities significantly. Hence, a deuterium-exchanged sample is preferred to the original proton-containing sample in the neutron diffraction study. 38 Neutron diffraction is especially useful for locating light atoms as many light atoms diffract as strongly as (or sometimes even more strongly than) heavier elements. Due to the absence of form factor, neutron diffraction is also capable of showing the disorders in the system more clearly than X-ray diffraction. Stronger high angle intensities by neutron diffraction enable the determination of small diffraction d spacings in the crystal structure, which hold information for the finer structural details than low angle intensities. Like X-rays, the images produced by neutrons are reciprocals of the lattice structures, and thus the structures can be solved similarly to the X-ray diffraction technique. One disadvantage of neutron diffraction analysis is the sources,of neutrons, which are produced by the nuclear fission in nuclear reactors (the fission of uranium) and spallation in a particle accelerator, which often require expensive and complex setups beyond the scope of any ordinary laboratory facility. The latter method, which can produce higher intensity neutron beams than nuclear fission, uses the collision of proton particles with a heavy metal target (usually uranium, tantalum or mercury) resulting in many 'knocked-out' (spalled) neutrons. The resulting neutron beams are near continuous and have about 100 times higher flux than in conventional nuclear reactors. The higher flux neutron beam allows the continuous wave (CW) and time-of-flight (TOF) experiments. In the TOF experiments, the energies and wavelengths of the diffracted neutrons are calculated in time from the neutron source to the sample and from the sample to the detector. In general, the flux of neutrons from either source is lower than a laboratory-based X-ray source. As a result, a large sample volume is required in general for a neutron diffraction experiment. In the present work, we used the N R U * reactor (Chalk River, Ontario) with C2 diffractomer (Figure 1.24) whose typical diffraction geometry is shown as schematic representation in Figure 1.25. Due to their scattering natures, X-ray and neutron diffractions are complementary to each other as X-rays map the electron density whereas neutrons can show the nuclear positions. * The N R U , National Reactor Universal, was established in 1957 and has been the single most expensive investment by the Canadian government to date. It has 200 M W output and produces over $5 billion worth electricity per year. 39 Figure 1.21 Illustration of the neutron travel path (as a wave) in neutron diffraction. The sphere represents possible directions for the scattering neutron, which can be anywhere on the surface of the sphere. Selec ted atomic weight Figure 1.22 The neutron scattering length of selected atoms. The two straight lines represent the X-ray scattering lengths of the angles/wavelength as indicated, illustrating increase of atomic scattering strength in proportional to atomic weight. The data used to produce the chart are from reference 88. 40 0 ~J—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i 0.0 0.2 0.4 0.6 0.8 1.0 sin6/X. Figure 1.23 Comparison between the X-ray and neutron scattering of carbon atoms with respect to sin0/X. The X-ray scattering of carbon shows a steep decline of the scattering strength as sinGM. increases, whereas the neutron part remains constant as it does not have form factor like X-rays. C2 diffractometer RU reactor; Figure 1.24 Schematic representation of the NRU reactor and C2 diffractometer, Chalk River, Ontario. The diagram is adapted from reference 89. 41 Neutron source Figure 1.25 Schematic representation of a powder neutron diffractometer with a monochromator, which is available typically in Be, C u , Ge , graphite and S i . The detector is made with BF3 cylinders that are wired to the console. 42 1.3.5.3 Rietveld method The determination of a crystal structure can be considered complete only after multiple pattern variables and peak and structure parameters of a model have been fully refined against the observed powder diffraction profile. The refined model should also be physically and chemically reasonable. The most common way of refining the powder data is the Rietveld refinement 9 0 , 9 1 , based on the idea by Hugo M. Rietveld in the 1960's. Although the Rietveld refinement is similar to Pawley and Le Bail methods of peak fitting, it is unique in the way that the values of the integrated intensities are included into all calculations as functions of relevant geometrical, specimen and structural parameters. The Rietveld refinement requires heavy computation for its non-linear least squares method, which requires reasonably well-defined initial approximations of the variables. These variables are peak shape parameters (profile functions), unit cell dimensions and coordinates of all atoms in the crystal structure model. Other unknowns such as background, scale factor and overall atomic displacement parameters can be guessed in order to reach a global minimum. In order to have a successful refinement, a proper structural model is needed. Other than having a model that is allowed by the laws of chemistry and physics, it should also yield correct integrated intensities and the initial knowledge of suitable peak shape and background functions. In the end, the fully refined crystal structure should yield a calculated powder pattern that closely resembles the observed one. Ideally, the difference between the calculated and the measured powder profiles would be zero, which is the core objective of the Rietveld method. Although the method is limited by the one-dimensional data and the instrumental uncertainties, it is quite powerful in its simplicity and visual outcome. However, in order for the Rietveld method to work, it requires a good starting crystal model in the beginning of its refinement. As its name implies, it is a refinement method, not a structure solution process. A detailed discussion on the method with formulism is given in Chapter 3. 43 1.3.6 Comments on organic/zeolite structure determination by X-ray and neutron diffraction The purely silicious zeolite ZSM-5 used for this study has a minimum number of 288 atoms in its unit cell. Therefore, it is challenging to solve the complete structure by any known method. However, there are a few known zeolite frameworks from previous s t u d i e s ' 9 ' 2 0 , 2 3 ' 4 0 " 4 4 , 8 1 , 9 2 , 9 3 , which provided the starting framework structure to solve the. organic/zeolite structures in this thesis. Therefore, a knowri framework structure of ZSM-5 has been used in the starting point and refined subsequently along with an included guest molecule. 1.4 Aims of this thesis Most studies in this thesis aim for the structure determination of the guest/zeolite complex by solid-state NMR. The verification of the determined structure can be made by single crystal X-ray diffraction for the unambiguous structure in some cases. However, in many cases, this option is simply not possible as the crystal of the zeolite complex exhibits complex phase changes, depending on the loadings and temperature changes of the crystals. Chapter 2 provides the experimental details for this thesis. This includes the syntheses of some materials used for the experiments as well as special instruments that were used for certain studies. Experimental conditions and detailed approaches are described. Chapter 3 explains some of theoretical aspects of the experiments for NMR and diffraction studies. It contains the flowcharts of the problem solving strategies for both NMR and diffraction experiments, especially emphasizing the Rietveld refinement methods for the powder diffraction technique. In Chapter 4, the NMR structure determination strategy is extended to a guest molecule that is less symmetrical than in previous studies. In Chapter 5, the structure determined from the NMR structure in Chapter 4 is verified by powder neutron diffraction when single crystal X R D is not possible. Chapter 6 deals with classic examples of guest/MFI systems with different loading stages, and their 44 structural determination by NMR and single crystal X R D . In Chapter 7, a mixture of two types of organic guests is loaded in the MFI channels, and their locations are probed separately by solid-state NMR for the first time. This is much closer to reality as the MFI experiences more than a single type of guest molecule in an industrial process. Currently, there is no known method to probe the locations of the multiple types of guest molecules in zeolite. Finally Chapter 8 gives some suggestions for future work and conclusions for this thesis. 45 Chapter 2 EXPERIMENTAL DETAILS: Methods and Materials 2.1 Sample preparation and analysis 2.1.1 Deuterium Labeled samples o-Xylene-d 1 0 , p-xylene-c/10 and p-xylene-d 6 are commercially available and were purchased from Aldrich and Cambridge Isotopes. The partially deuterated (or protonated) xylenes, o-xylene-of4, o-xylene-06 and p-xylene-d 4 were prepared by acid exchange with deuterium oxide (normal water for o-xylene-d 6) as described in previous s tud ies 9 4 , 9 5 . A typical synthesis is (in this case to produce o-xylene-d 4): ca . 1 mL of o-xylene (99% H P L C grade from Aldrich) and 9 mL of a solution of ca. 5% DCl in D 2 0 were mixed in a long neck, thick wall Pyrex glass tube of ca. 15 mm o.d. (ca. 8 mm i.d.). The solution was degassed repeatedly under liquid N 2 freezing and thawing, and the tube was then vacuum-sealed with a flame and placed in a stainless steel reaction bomb containing ca. 5 mL water to equalize the inner and outer pressures on the glass tube. The bomb was sealed and kept in an oven at 220 °C for 48 h. The exchanged xylene was then washed with D 2 0 and decanted using a Pasteur pipet. The progress of the exchange reaction was monitored by solution 1 H NMR spectroscopy, measuring the peak intensities of the aromatic ring protons and methyl protons. The procedure was repeated until the exchanges were over 95% for each sample, which is an intensity ratio of aromatic/methyl protons of 0.03 by 1 H NMR. Other commercially available organics were suitably purified before use. 2.1.2 Highly siliceous ZSM-5 samples The 'NMR quality' ZSM-5 samples, for which it is possible to observe the 2 9 S i peaks with minimal broadening caused by aluminum present in the framework, were first prepared by G. E. Barlow 3 7 . The ZSM-5 sample was crystallized from an alkaline hydrothermal treatment of Ludox silica 46 gel with tetrapropylammonium (TPA) cation as the template that directs the formation of the ZSM-5 framework structure. After the sample was calcined at 550 °C for several hours to remove the template and ion-exchanged with an aqueous 0.2 M ammonium bifluoride solution, the sample was "steamed" at 800 °C for several days in order to remove any traces of aluminum in the framework and heal defect framework sites. The resulting materials were highly siliceous and crystalline by powder XRD and solid state NMR measurements; one of the samples, 'GEB 177', that was used for the most of experiments in this study, gave peak widths at half maximum of < 10 Hz in the 29Si MAS NMR spectrum (Figure 2.1). For the single crystal XRD investigations, large single crystals of ZSM-5 over 100 (am in size were kindly provided by Drs. W. Schweiger and F. Scheffler. Similar to the synthesis above, they were prepared based on the hydrothermal synthesis with TPA chloride- and calcined at 550 °C to remove the organic template material. Each single crystal was examined under an optical microscope, and suitable crystals for single crystal XRD were chosen based on their sizes and the quality of their typical 'coffin' shapes which had dimensions of > 100 pm in length (Figure 2.2). -112 -114 -116 -118 2 9 S i chemical shift (ppm from TMS) Figure 2.1 High resolution 29Si NMR spectrum of the calcined ZSM-5 sample ('GEB 177'), used in the most of the NMR experiments in this study. The spectrum was obtained with a 90 0 pulse of 10 us, 512 scans, and a 5 s recycle delay. 47 Figure 2.2 Scanning electron microscope (SEM) picture of a typical large 'coffin' shaped ZSM-5 single crystal (> 100 nm), prepared by Dr. W. Schweiger. 2.1.3 Loading of guest organic molecules into zeolites The procedures used to load the organics into the zeolites mainly depended on the guest organic molecule in question. In the case of the benzene/p-xylene mixture in ZSM-5 , the two liquids were simply injected using micro-scale syringes onto the microcrystalline Z S M - 5 powder in a screw capped glass vial with a Teflon-seal. The loading was confirmed by T G A several times after the mixture/ZSM-5 complex had been stored at room temperature, which was a high enough temperature to equilibrate the mixture throughout the framework within a few hours. Samples of ZSM-5 saturated with o-xylene were prepared by introducing excess o-xylene vapor to freshly calcined microcrystalline ZSM-5 powder under vacuum (Figure 2.3). A temperature range of 8 0 - 100 °C was sufficient for the o-xylene molecules to equilibrate into the ZSM-5 pores of this sample within 72 h, as indicated by the complete absence of the low field 2 9 S i M A S NMR signals indicative of the empty monoclinic phase. Subsequent T G A experiments showed that the loadings of o-xylene were from 3.7 to 4 molecules per unit cell. The solid guest organics, p-dicyanobenzene and p-dinitrobenzene, were accurately weighed and mixed with ca. 150 - 200 mg of the empty ZSM-5 powder in 1 mL ampoules. The ampoules were vacuum-sealed using a flame and then placed in an oven at 80 ~ 100 °C for at least 48 h to 48 equilibrate. To vacuum organic guest zeolite Figure 2.3 Apparatus used to load o-xylene into Z S M - 5 . By using an excess amount of the organic guest, the maximum loading was promoted within a relatively short time period (48 - 72 h). After purging the air in the vessel , it was sealed by closing the stopcock and placed in an oven at 8 0 - 1 0 0 °C. 2.1.4 Determining the loadings of organic guests in the zeolite Determinations of the loadings were carried out using thermogravimetric analysis (TGA) by heating the sample from 30 °C to 600 °C at a rate of 10 °C/min. The molecular loadings per unit cell of zeolite can be determined from Equation 2.1, assuming no other volatile species are present. molecular loadings per unit cell = 100% Weight% — 1 MW,. MW, Equation 2.1 guest where MWzeome is the molecular weight of zeolite per unit cell, MWguest the molecular weight of the guest molecule and Weight% is the reduced weight of the sample after the T G A analysis with respect to the original sample weight in weight %. An example is given in Figure 2.4 for o-xylene in ZSM-5 . 49 According to the T G A plot and Equation 2.1, the final weight % of the sample is 93%, which corresponds to ca. 4 molecules of. o-xylene per unit cell of ZSM-5 , assuming the ZSM-5 is purely siliceous having the empirical formula Si 960 1 92 for its unit cell. 100 i 92 1 , , , , : , , , 0 100 200 300 400 500 600 700 Temperature in degree C Figure 2.4 A typical T G A plot of weight% vs. temperature. The plot shows a weight decrease of the o-xylene/ZSM-5 complex to ca . 93 % during the T G A analysis, which corresponds to a loss of ca . 4 molecules of o-xylene per unit cell of ZSM-5 . 50 2.2 Solid state NMR A number of different approaches were taken to probe the distances between the 1H and 29Si nuclei in the organic/zeolite samples used in this study. In this section, some of these experiments are described together with the expected outcomes. 2.2.1 N M R s p e c t r o m e t e r All of the solid state NMR experiments were performed on a Bruker AVANCE DSX-400 NMR spectrometer with an operating frequency of 400.13 MHz for 1H, 376.343 MHz for 19F, 79.495 MHz for 29Si, and 61.422 for 2H. The spectrometer has one amplifier capable of producing power at 1000 W for 1H or 19F, and two 300 W amplifiers for lower frequency nuclei. The r.f. pulse frequencies, phases and amplitudes were generated in a digitally controlled manner, and thus the spectrometer was capable of producing very stable and consistent pulses. Most often, the FIDs were digitally filtered, which improved the S/N significantly. The spectrometer was controlled from a Silicon Graphics workstation installed with Bruker's XWIN-NMR version 2.6 software. 2.2.2 M A S p r o b e s Throughout this study, two different triple resonance MAS probes were used for MAS and static experiments. A'modified Bruker broadband H/X 7 mm CP MAS probe was used for most experiments, including the static 2H experiments in Chapter 7. This probe was capable of tuning to 1H or 1 9F for its H channel and 1 3C, 29Si, 1 5N or 2H for its X channel. It was equipped with a cylindrical, standard-speed 7 mm Kel-F MAS stator from Doty Scientific, which was capable of spinning stably at 2 kHz. The probe also had a heating coil with thermocouple and a bearing gas inlet (vacuumed dewar glass tube) to perform variable temperature experiments (all the variable temperature experiments in this study) over a temperature range of 200 to 350 K. 51 The second M A S probe, equipped with a 4 mm triple tuned H/X/Y M A S stator and single solenoid r.f. coil, was capable of spinning the sample at frequencies up to 15 kHz with ± 5 Hz precision and an optical spin rate detector coupled with the MAS unit that actively controlled the drive and bearing gas pressures. The probe was capable of tuning to 1 H , 1 9 F , 2 9 S i and 1 3 C with several home-built inserts in order to change the tuning frequencies for each sample of interest. KI was used to set the magic angle for the probes, where the FID of the 1 2 7 l resonance (80.055 MHz), showed rotational echoes (spikes) due to the spinning sidebands of the satellite transitions, which are very sensitive to the magic angle setting (Figure 2.5). For the 7 mm probe, which does not have a spin rate detector, a small amount of KI was packed in the bottom of the rotor and separated from the sample of interest by Teflon tape. By observing the spinning sidebands or rotational echoes in the time-domain, the spinning rate could be obtained from the frequency separation of the peaks (Figure 2.5c). Figure 2.5 'Magic angle' adjustment using the 1 2 7 l rotational echo 'spikes' of KI in the frequency domain, (a) Off-magic angle (b) adjusted to on-magic angle FIDs in the time domain, (c) The peaks in the spectrum showing the satellite spinning sidebands (spin rate of 2 kHz) after Fourier transformation of the rotational echo into the frequency domain. The separations between the spinning sidebands can be used to determine the spinning rate of the sample. 52 2.2.3 Magnetic field shimming Unlike solution NMR, for solid state NMR, the magnetic field does not need to be shimmed for every experiment because the lines are intrinsically broader than those of solution NMR. However, to achieve high-resolution NMR spectra (for 2 9 S i spectra, the full width at half maximum (FWHM) of < 10 Hz), it is important to have a very homogenous field. In general, shimming of solid state NMR is more demanding than solution NMR; however, once set, it does not require another adjustment for a considerable time. The shimming can be done either on solid samples, e.g. adamantane, which, being a plastic crystal, has very narrow resonances, or on liquids (ordinary water being suitable). For these studies, all solid state NMR experiments were based on shim sets from tap water 5 as they gave the best resolution for the 2 9 S i spectra of the zeolite samples. The magic angle should be carefully set before shimming as a slightly wrong angle gives peak broadening. For details of the shimming in practice, there are many good ar t ic les 9 6 , 9 7 for further reading. 2.2.4 Variable temperature experiments As indicated previously, the 7 mm H/X probe was used for all the variable temperature experiments, and its spinning rate was stable up to 2 kHz between 200 K and 350 K. To investigate the effect of temperature on a given zeolite sample, variable temperature experiments are important. In the case of high temperatures, the heater and thermocouple unit can be attached to the probe using the regular in-laboratory air outlet for the bearing and drive gases. For the low temperature experiments, a home-built apparatus setup was used (Figure 2.6). Instead of the automated MAS controller, a manual M A S controller was used to control the drive and bearing pressures into the MAS probe. For N 2 purged experiments the paramagnetic oxygen molecules in the framework of zeolite § It is thought that impurities in tap water help the T, relaxation process of the 1 H nuclei, and thus more signals can be collected in a shorter period of time. 53 (which cause the line broadening) are removed by replacing the compressed air with N 2 for the bearing and drive gases. In the usual low temperature setup, the drive gas can be replaced by the in-laboratory compressed air as this ensures a longer experiment without refilling the 200 L self-pressurizing liquid N 2 dewar. If the 200 L dewar was used for both bearing and drive gas supply, it would last a little over 48 h whereas if only used for the bearing gas supply, it lasts 5 days. The cooling of the N 2 gas takes place in the cooling dewar (40 L) as the bearing gas passes through the metallic coil submerged in the liquid N 2 . As the liquid N 2 in the 40 L dewar is boiled off during cooling the bearing gas, it is refilled by pressurizing the 50 L dewar connected via a vulcanized rubber hose. The process of refilling is automated using a programmable digital timer and a solenoid switch, which opens the N 2 gas cylinder to pressurize the refill dewar. In the auto-mode, the refill dewar is replenished twice a day with liquid N 2 for continuous low temperature operation. 54 PRESSURE REGULATOR | y E N T (CLOSED) (OPEN, GAS PRESSURE BUILDING |lO:35| i 9 > P R O G R A M M A B L E DIGITAL TIMER 7-SELF-PRESSURIZING LIQUID NITROGEN DEWAR % 160 PSI (200 L) NITROGEN GAS (30 PSI) MAS CONTROLLER DRIVE BEARING LJ PRESSURE VENTING SOLENOID (NORMALLY OPEN) PRESSURE REGULATOR COOLING DEWAR LIQUID NITROGEN (40 L) DRIVE GAS (5 PSI) F R A M E FLUSH ( 1 PSI) SOLENOID (NORMALLY CLOSED) N 2 GAS REFILL DEWAR LIQUID NITROGEN (50 LJ T H E R M O C O U P L E 2*) K | | . . i . i • TEMPERATURE CONTROLLER Figure 2.6 Variable temperature setup for the solid state N M R experiments. The range of temperature is from 200 K to 350 K with a 7 mm rotor at a 2 kHz spinning rate. This schematic shows N2 drive and bearing in order to remove paramagnetic oxygen molecules in the framework of zeolite (Chapter 4). Usually the drive gas is provided by the compressed air supply in the laboratory. 2.2.5 Se t t i ng up e x p e r i m e n t s u s i n g re fe rence s a m p l e s In order to determine experimental parameters such as pulse length, C P match condition, chemical shift and contact times, reference samples were used for convenience. For experiments with 1 H and 2 9 S i in the system, ' Q 8 M 8 ' (cubic octameric silicate, S i 8 0 1 2 [OSi (CH3)3] 8 ) was used. The 2 9 S i chemical shifts can be referenced to the highest field peak to -109.71 ppm in the Q 8 M 8 2 9 S i spectrum, which has been referenced to the liquid tetramethylsilane (TMS). The reference sample 98 99 was prepared by Prof. C. A. Fyfe according to previous syntheses ' 55 -109. 71 ppm 'Si chemical shift (ppm) Figure 2.7 Reference sample Q8M8 and its 2 9 S i NMR spectrum at room temperature. The chemical shifts of Q8M8 were originally determined with respect to TMS, and the chemical shift at -109.71 ppm (one of the 2 9 S i signals from -SiO<t) was used as a spectral reference for the experiments in this work. 2.2.6 Relaxation time measurements Measurements of the 1 H and 2 9 S i relaxation times, Tu T2 and 7"1p were made prior to carrying out C P experiments. The T\ relaxation times are crucial for quantitative experiments and important in all cases to obtain the maximum S/N, and were measured using the saturation recovery or inversion recovery mentioned in Chapter 1. The spin-spin relaxation times, T2, which are required for efficient INADEQUATE experiments, were determined by the spin echo sequence. 7"1p, the "spin relaxation time in the rotating frame" during the spin-lock period, can provide valuable information for interpretation of the C P data. In addition, the 1 H 71p relaxation times determine the "exchange regime" of the C P curves. 7"1p was measured using the same spin-lock periods with a list of variable delays. All the relaxation time data were processed using the Bruker Xwin-NMR software. 56 2.2.7 2 9 S i M A S a n d 1 H / 2 9 S i C P M A S N M R The chemical shift references for 2 9 S i MAS and 1 H / 2 9 S i C P M A S NMR spectra were obtained from the spectra of Q 8 M 8 unless otherwise specified. All match conditions used for the 1 H / 2 9 S i C P MAS experiments were set from this reference sample as well. In order to find the Hartmann-Hahn match condition between the 1 H and 2 9 S i nuclei, the power level for the 1 H channel was fixed and that of the 2 9 S i channel was varied to find the value that gave the maximum peak intensities for Q 8 M 8 . Subsequently, the 90 ° pulse of 1 H was found by varying the pulse length (using the paraopt command in the Bruker Xwin-NMR software) to maximize the peak intensity of 1 H / 2 9 S i C P MAS NMR spectrum of the reference sample. The 2 9 S i 90 ° pulse was determined by 2 9 S i M A S NMR directly on the sample of interest in most cases. 29 Quantitative Si MAS NMR spectra were obtained using either a 90 ° pulse with a recycle 29 delay of five times the longest Si 7 or a < 90 0 pulse with shorter recycle delay of 30 to 50 seconds, especially for samples that had exceptionally long 7"i relaxation times. In most 2 9 S i M A S and 1 H / 2 9 S i C P MAS NMR experiments, when a 90° pulse is used, the optimal recycle delay time of 1.25 times the 2 9 S i or 1 H 7 value gives the largest S/N within a given experimental time according to the 'Ernst angle' relationship 6 0 (Equation 2.2). cosa = exp(-7;/7;) Equation 2.2 where a is the Ernst angle, TR is the optimum recycle delay after each pulse, and Tx is the Ti transverse relaxation time of the nuclei of interest. Variable contact time C P and C P drain experiments were collected as two-dimensional experiments in the ft dimension at different contact times, which were defined by the vplist, in the 57 initial parameters of the NMR experiments. The results were processed by Fourier transformation of the f2 dimension, which yielded successive spectra collected in the order of the contact times. While the C P experiments were done with recycle delays of 1.25 times the longest 1 H 7i relaxation time, the C P drain experiments were collected using recycle delays of 5 times the longest 2 9 S i 7, relaxation times to ensure the full recovery of the 2 9 S i nuclei before the next 90 ° pulse. 2.2.8 I N A D E Q U A T E e x p e r i m e n t s INADEQUATE experiments to probe 2 9 S i connectivities are quite demanding, as they require accurate measurements of the T2 relaxation times, and 90 ° and 180 ° pulses to optimize the initial parameters. The T2 values, often the limiting parameter, are especially important since the echo delays in the INADEQUATE experiments depend on them to yield the correlations. The echo delays were usually 10 to 20 ms. A two-dimensional INADEQUATE spectrum can be collected using either 2 9 S i MAS or 1 H / 2 9 S i C P MAS depending on which yields the better S /N . In any case, the recycle delay of the experiment is set to 1.25 times the T, of the excited nuclei. In order to ensure clear peak separations, the spectral sweep width was set to the minimum value permitted by the spectrum. The usual value for the sweep width was between 700 and 1000 Hz depending on how well the peaks were resolved, especially for the overlapping peaks in the spectra. Zero filling with at least twice the number of points collected and multiplying the time domain data in each dimension by a sine-bell apodization function were applied before Fourier transformation of both f"i and f2 dimensions. The two-dimensional spectra are presented as contour plots in this study. 58 2.2.9 1 H2D-N0ESY The 1 H 2 D - N 0 E S Y experiments were carried out with mixing times of 20, 50, 100 and 200 ms in order to probe the spin diffusion between the proton nuclei in the organic guest molecules. No decoupling of other nuclei was applied during the signal acquisition period, and the sample was spun at 15 kHz with the 4 mm H/X/Y probe at 293 K. A total of 256 experiments were collected in the ri dimension with 128 scans for each experiment. The 1 H 90 ° pulse was 5 ps, and the recycle delay 2 s. The sweep widths of both the ri and f2 dimensions were 5000 Hz. The quantitative 1 H spectra for the N O E S Y spectra were collected with a 90 ° pulse of 5 ps and a recycle delay of 5 s, five times the longest 1 H T,. 2.2.10 2 H static NMR Wide-line deuterium NMR spectra were obtained using the 7 mm H/X probe and the (90 x-t-90 y-T-acq) quadrupolar echo sequence 1 0 0 with a 90 ° pulse of 7 ps. The simulations of the 2 H static spectra were created by Weblab version 4.1.1 software 1 0 1 . 2.2.11 NMR data analysis and calculations With the exception of a few spectra, most NMR data presented in this study were processed using computer programs written as Mathematica (version 3.0) notebooks 1 0 2 . The programs were developed by Dr. D. H. Brouwer and are available from his thes is 1 0 3 . The variable temperature spectra and 2 H static data were processed using the Mestre-C software. The INADEQUATE and 1 H N O E S Y spectra were plotted using the Bruker Xwin-plot software. 59 2.3 Diffraction methods 2.3.1 Single crystal XRD data collection and processing Data sets were collected on Bruker X8 APEX and Rigaku ACF7 diffractometers using graphite-monochromated Mo K a radiation in both cases. Data collections were carried out at different temperatures (-100 + 1 °C, 0 ± 1 °C and room temperature). The single crystal X R D data for the p-dicyanobenzene/ZSM-5 complexes were collected by Dr. B. O. Patrick. The single crystal X R D data for p-dinitrobenzene/ZSM-5 complex were collected by the author. The Rigaku ACF7 diffractometer has an Area Detector System Corporation charge-coupled device (CCD) detector (size 98 x 98 mm 2). In low temperature operation, a gas flow from a liquid N 2 dewar was used to cool the crystal. In the beginning of the data collection, a background was collected with the shutter closed and subtracted during the integration of the data collected. Cell parameters were determined from a least squares fitting of the selected reflections. The reflections were indexed by D*TREK and TWINSOLVE software packages and the systematic absences analyzed to determine the crystallographic space group. Refinements used SHELXS and SHELXC0* and the Fourier electron density difference maps were constructed using Win-GXm. For studies of volatile organic guest species, special care is needed as the guest could escape during the preparation of the single crystal for X R D . A popular choice is to use a capillary containing the single crystal at the tip. For a guest/zeolite complex that requires a specific loading, a home-built capillary-glass ampoule (Figure 2.8) was used to ensure accurate loading of the guest species in the framework of zeolite. While applying flame the tip of the capillary containing the single crystal can be removed, and the ampoule can be resealed until the loading of the remaining powder can be determined by TGA. 60 Figure 2.8 A home-built capillary-glass ampoule for precise loading of volatile organics into a single zeolite crystal. A single crystal of Z S M - 5 is loaded at the tip of the capillary before connecting the capillary and the glass ampoule by glassblowing. Microcrystalline zeolite powder is placed in the ampoule, a known amount of the guest molecule added prior to sealing the ampoule and equilibrium of the guest in the zeolite framework obtained. 2.3.2 Powder diffraction sample preparation In order to collect Rietveld quality data, the samples must be packed correctly: a) The particles are randomly oriented so that preferred orientations are minimized. b) The sample should be confined to all the area of the X-ray beam at all angles accessed. c) The sample height should be thick enough not to allow X-ray to pass through. d) The sample surface should be smooth and flat. The best particle size is < 10 fim, preferable 1 to 5 ^ m . In order to ensure the small crystals in a powder sample are evenly sized, the sample should be ground either by hand using a mortar and a pestle for 20 minutes or by a vibratory McCrone Micronising Mill with corundum elements for 5 - 7 minutes. In both grindings, ethanol can be used to promote even grinding. 2.3.3 Powder XRD data collection Powder X R D data were collected using Bruker D8 Advance and D8 Discover X-ray diffractometers in the Bragg-Brentano configuration, illustrated in Chapter 1. The D8 Advance has a Nal scintillation detector and diffracted beam graphite monochromator with an 1.0 mm divergence, an 1.0 mm anti-scatter and a 0.2 mm receiving slits. Its sample holder was rotated to improve particle statistics. The D8. Discover has a General Area Detector Diffraction System (GADDS) and uses 0.5 mm collimated beam with incident beam monochromator. The diffractometer can also have a glass 61 capillary setup or a plate detector with a variable temperature setup using gas flow from a liquid N 2 reservoir and a heating platform with a thermocouple connected to the automatic temperature control unit. Both diffractometers use CuKa 1 and 2 as their radiation sources with the power generator set to 40 kV with 40 mA. In order to collect the data used for the Rietveld refinement, the step size should be < 1/5 F W H M of the sharpest peak. Figure 2.9 A custom-built powder X R D sample holder made of polymethylmethacrylate (PMMA) for volatile organics loaded into a zeolite powder. A polyvinylidene chloride (PVDC) film, which does not show any significant reflections or a large background, goes between the rubber ring and the lower half of the holder. The volatile organic is then introduced through the channel at the bottom, which is sealed with a screw cap. 62 2.3.4 Powder neutron diffraction data collection The powder neutron diffraction measurement was performed on the C2 diffractometer at the Chalk River Laboratories, Chalk River, Canada by Mr. L. M. D. Cranswick and Dr. I. Swainson. The diffractometer consists of an 800-wire B F 3 position sensitive detector spanning 80 0 29, A = 0.1 ° 29, which floats pneumatically over an epoxy "dancefloor" in order to ensure a minimum disturbance from the surrounding environment. The incident 1.33020 A wavelength used for the Rietveld refinements was calibrated with an external Si powder standard and obtained from a S i ( 5 3 1 ) reflection at a 92.7 ° take-off angle. For indexing runs, the wavelength selected was 2.37135 A from a S i ( 3 1 1 ) reflection at the same take-off angle. During the indexing runs, the beam was filtered with a pyrolitic graphite filter (Panasonic Super Graphite). The range of the collected data was between 3 and 115 ° 29 with wire spacing of 0.1 ° at 272 K. 2.3.5 Powder diffraction data processing Indexing of both powder X R D and powder neutron data was done by Crysfire su i te 1 0 6 incorporating the information from the solid state 2 9 S i NMR spectrum on the sample. The space group found from indexing the powder diffraction sample was Pnma as suggested by the previous N M R 1 0 7 and powder X R D 5 1 studies. The initial solution of the o-xylene/ZSM-5 structure was obtained using the FOX (Free Objects for Xtallography)108,109 program. The final Rietveld refinement was done using the G S ^ S 1 1 0 [General Structure Analysis System) and EXPGUIm, a graphical user interface for G S A S , programs. 63 C h a p t e r 3 Structure Determination Strategies In this chapter, general strategies to determine organic guest/zeolite structures by N M R and their verification by powder diffraction methods are described. For NMR, the theoretical background of C P type experiments is outlined along with the structure determination protocol that has been developed. For the powder diffraction section, an approach to solving the structure from powder data is illustrated followed by a discussion of the Rietveld method for the refinement of the solution. 3.1 Introduction The structure of a guest/zeolite complex can usually be unambiguously determined if a large zeolite crystal is readily available and the loading and distribution of the guest molecule do not pose any difficulties. Unfortunately, in reality, it is unlikely that a single crystal of sufficient size and quality for X R D will be available. In most cases, structure determinations of these complexes have not been done. Recently, however, there have been developments in areas other than single crystal X-ray diffraction that make these structure determinations possible. They are in solid-state NMR, which is one of the most important tools in chemistry in its own right, and in powder diffraction techniques, a relatively new approach in the field of diffraction, as many samples of interest are not available in the form of large single crystals. In this chapter, these two techniques are illustrated, as they were the structure determination methods used in this work. Solid-state NMR structure determinations, which use the dipolar coupling interaction between nuclear spins in the solid state, have been implemented successfully on several s a m p l e s 2 5 , 2 6 , 4 5 , 4 6 , 6 8 , 9 4 , 1 0 3 , 1 0 7 in. recent years. All the examples have been those of the aromatic guest/ZSM-5 complexes because: (a) they are important in industrial processes, and (b) large ZSM-5 crystals are readily available for verification of the NMR determined structure by single crystal X R D . In general, the NMR method requires prior knowledge of the zeolite framework topology, which can be obtained from powder X-ray diffraction. In addition, there have been several approaches recently 64 to using the homonuclear dipolar couplings between the Si nuclei of the framework to deduce framework structures that have considerable potential 1 1 2. The powder diffraction technique has become more popular recently due to improvements in instrumentation and the development of advanced software capable of solving complex structures (ITQ-22 1 1 3 , one of the most complex zeolite frameworks known, for instance). Several investigations of guest/zeolite complexes have been pub l i shed 4 7 , 4 9 " 5 1 , 5 3 , 1 1 4 , 1 1 5 in recent years adding to previous model-fitting determinations. In this chapter, the two techniques mentioned above are illustrated by describing the strategies developed by the Fyfe group for NMR structure determinations and the general guidelines for the powder diffraction method used for this study to verify the guest/zeolite structure in Chapter 4, focusing on specific protocols for indexing (Crysfire106), structure solution by global optimization of parameters ( F O X 1 0 8 , 1 0 9 ) , and the Rietveld refinement procedure ( G S A S 1 1 0 and EXPGUIm). 3.2 NMR structure determination In order to determine a guest/zeolite structure by NMR, it is imperative to have a good 1-D 2 9 S i MAS NMR spectrum of the sample. This sounds very simple and obvious; however, there are several factors that could affect the NMR spectra as the relaxation of the 2 9 S i spins could be further complicated by other types of spins present in the channel system, for example, from the presence of paramagnetic oxygen molecules. The experiments should be carried under a N 2 purged environment to remove any paramagnetic oxygen molecules from the zeolite framework, which cause a line broadening of 2 9 S i MAS spectra from fast T2 relaxation of the s p i n s 1 0 7 , 1 1 6 . From the 1-D spectrum, many useful things can be deduced, especially for a system that has a similar organic guest molecule to previously studied systems. For example, in all the cases of guest/ZSM-5 complexes s t u d i e d 2 5 , 2 6 , 4 5 , 4 6 , 8 1 , 1 0 7 , the two peaks at highest field have been always due to the silicon T-sites 8 and 2 for the space group Pnma. In addition, the temperature that gives the best resolution can be found by variable temperature experiments. Once an optimum temperature is found, and a well resolved 2 9 S i spectrum is obtained, the structure determination by NMR can proceed: the following are proven 65 strategies for a successful structure determination. 3.2.1 2 9 S i I N A D E Q U A T E a s s i g n m e n t s In order to solve the structure of a guest/zeolite complex, the 2 9 S i N M R peaks must be assigned to specific sites in the framework. As mentioned in Chapter 1, the 2 9 S i INADEQUATE experiment is a practical choice to probe the connectivities between the silicon T-sites as it lacks the diagonal intensities of a C O S Y experiment, especially important for systems having large numbers of silicon T-sites. For example, the NMR determined connectivities of the monoclinic form of ZSM-5, which has 24 silicon T-sites, have been determined only by INADEQUATE experiments 2 9 , 3 0 . In the beginning of the development of the NMR structure determination method, assigning peaks from the INADEQUATE experiments was quite challenging and extremely time consuming as many possible assignments of the data had to be tested manually. For example, one starts with 24! = 6.2 x 10 2 3 possible assignments for the monoclinic phase of ZSM-5 with 24 silicon T-sites. However, processing INADEQUATE data has been greatly simplified by the peak assignment program 1 1 7 developed by Dr. D. H. Brouwer. The program has an algorithm that introduces one peak at a time to the existing set of peaks that are assigned previously to test peak assignments rather than testing a whole set of a possible peak assignment simultaneously every time. This simple yet efficient change in the algorithm has been proven to be a great success, and all the peak assignments of the INADEQUATE spectra in this study were done with the aid of this program. When unambiguous peak assignments are not possible, the program can still aid the peak assignment by greatly reducing the number of possible assignments. The final assignment can be made either by attempting to solve the structure with each of the reduced peak assignment sets in order to find the one that yields a solution (for detailed description, see Chapter 4), or by using the chemical shifts from variable temperature experiments if the peaks can be assigned at a different temperature for the same sample (detailed examples are given in Chapters 4 and 7). 3.2.2 P r o b i n g the d i p o l a r c o u p l i n g s Cross polarization experiments are usually carried out using the standard spin-lock sequence 66 (Figure 3.1). During the spin-locking step, there is coherence transfer from 1 H to 2 9 S i via the heteronuclear dipolar interaction if the two spin-locking fields 6 1 ( H ) and S 1 ( S i) satisfy the Hartmann-Hahn match condit ion 1 1 8 , Equation 3.1. In this work, the two r.f. fields were matched experimentally as described in Chapter 2. Y H S 1 ( H ) = YsiSi(si) Equation 3.1 where Y H and YSI are the magnetogyric ratios of the proton and silicon nuclei respectively, and S 1 ( H ) and B^,) are the applied radio frequency magnetic fields applied to the two nuclei respectively. 29 The Si peaks change their intensities due to the distances from the protons in the organic guest in the cross polarization experiments. The variable contact time C P experiments can probe this distance relationship between the 2 9 S i and 1 H . The experimental cross polarization rate constants were determined after determining the intensity changes over contact time changes in the variable contact time C P experiments. The protons in the organic guest molecules are neither truly "isolated nuclei" nor truly "abundant spin" systems. The guest molecular motions in the framework could affect the C P process, especially when the cross polarization time constant (T C p) and the proton relaxation time in the rotating frame (r1 p ( H)) are very sensitive to even very slow molecular motions. In many cases, the increases in S ( 2 9Si) spin intensities in 1 H / 2 9 S i C P curves depend on TCP (MkiS) and the decreases in the intensities by r 1 p ( H) (1//C/). The case is known as the "fast C P regime." The value of interest is kiS (1/7"Cp), known as the C P rate constant. However, in a variable contact time C P experiment these values can only be determined along with the magnetization relaxation rate constant /c, (1/7i p of 1 H) and / 0, the theoretical maximum of the S spin signal intensity from the C P transfer without any relaxation processes. Therefore, the measured value is k',s, which is a relative C P rate constant. Quantitatively, the cross polarization process is usually described using Equation 3.2 with the assumptions that k, « kts, and the magnetization relaxation rate constant of 2 9 S i , ks (1/ 7ip(Si)) is very small compared to /c / s. In order to simplify the calculations, ks can be neglected, and Equation 3.2 becomes Equation 3.3. A further assumption is that the number of 1 H nuclei is far larger than the number of 2 9 S i nuclei. This second condition is satisfied in between the high natural abundance of 1 H 67 (100.0%) and the low abundance of Si (4.6%). The major shortcoming, however, is the low concentration of protons in most of the guest/zeolite complexes. Also in the case of T 1 p ( S i ) being relatively short, the magnetization relaxation rate constant of 2 9 S i (ks) should be accounted for as well. These could compromise the description of the behavior of the S nucleus ( 2 9Si) magnetization by Equations 3.2 and 3.3. /(0 = /o[1 +(kslkiS)-(kilkls)]~\exp(-klt)-exp(-klst-kst)] Equation 3.2 /(f) = /„[1 - (k,lkls)]-\exp(-k,t) - exp(-/c/sf)] Equation 3.3 In some cases, the initial rises of the intensities depend on kt while the falls are governed by kis + ks- This phenomenon was studied by Klur et a l . 1 1 9 and is termed the "slow C P regime." When /c, < kiS » ks, it is very difficult to obtain accurate values of /c / s at very long contact times from the conventional C P experiment. To overcome this problem, an alternative experiment is needed for an exact measurement of the k/s values. In this work, we use the 1 H / 2 9 S i C P "drain" (originally termed "double cross polarization") exper iment 1 2 0 ' 1 2 2 to measure the absolute kls values in the organic/zeolite complex. The C P drain experiment consists of two separate experiments (Figure 3.2). In the first, the S spins are excited and 'spin-locked' without a contact pulse applied to the / spins so that the decay is entirely due to ks, as in Equation 3.4: S0(f) = S 0exp(-/c sf) Equation 3.4 where S 0 represents the maximum signal intensity obtainable from the exponential decay of the S spins in the absence of any loss due to relaxation processes. In the second experiment, both the S and / spins are applied with contact pulses, and the exponential decay of the S spins due to the magnetization drain is observed; the signal intensity is now governed by Equation 3.5. Sd(t) = SQexp(-kst - k/st) Equation 3.5 The kiS values can now be obtained directly from the normalized difference plot of the two decays (Equation 3.6). [S0(f) - Sd(t)]IS0(t) = AS/S 0(f) = 1 - exp(-/c /sf) Equation 3.6 A drawback of this type of experiment is poor signal-to-noise since the final data are obtained 68 from the difference between the reference and drain spectra. Further, cases where the S spins have long 7i values, which require a longer delay to recover the equilibrium magnetization, could also pose difficulty of obtaining a reliable data of high enough S/N for structure determinations. The structure determination by solid state NMR is based on the relationship between kiS and the heteronuclear second moments (M2) of the dipolar line shapes64, shown in Equation 3.7. k,s= C(AM2)iSl(AM2)uV2 Equation 3.7 The heteronuclear second moment of the dipolar coupling (simply M2 hereafter) is a sum of the squared values of the dipolar couplings, and can be calculated once the distances between the / and S spins (both the / and S nuclei are spin - Yz) are known for all IS spin pairs: "••{illib'^^i E q u a t i o n s thus, the following relations can be assumed: kis oc M2 oc (y,2 ys2)£r,s"6 ( Equation 3.9 The linear relationship between the kts and M2 in Equation 3.9 is the basis of the NMR structure determination method. 90V (Y ,H,) y C P (Y,H.) decoup l ing t Contac t t ime Figure 3.1 Pulse sequence for a general C P experiment. 69 decoupling (b) 90, (Y sH s) y (Y,H,) decoupl ing Contact time Figure 3.2 Pulse sequence for a C P drain experiment. The experiment consists of two parts: (a) the reference and (b) drain experiments 3.2.3 NMR structure determination program The key to a successful structure determination by solid-state NMR is to have as many reliable kiS values for the silicon T-sites in the zeolite framework as possible. The results of variable contact time CP or CP drain experiments are fitted to give kiS values, which are used for the structure determination programs25,26,94,103 developed by the Fyfe group. For this study, the program written by Dr. D. H. Brouwer25,103 was used to calculate the structures of the guest/ZSM-5 systems studied. First, the program generates all possible locations and orientations of the organic molecules in the channel system. It then searches for the locations of the guest molecule in the rigidly held framework of ZSM-5 that are compatible with the /c;s values of the resolved resonances corresponding to single Si sites according to the following criteria: 1) It must be physically reasonable (no framework atom-sorbate atom distances less than 2 A by default) 2) There must be a strong linear correlation between the values of k)s or /c,s and the calculated 1H-29Si second moments for the structure as expected from Equation 3.8 (within a specified standard deviation) 3) There should also be good agreement between 70 the experimental and calculated intensities of the complete Si NMR spectrum, including all overlapping peaks. For the framework coordinates, the Si and O atomic positions in ZSM-5 were taken from the single crystal X R D study of the low loaded form of p-dichlorobenzene/ZSM-5 complex by van Koningsveld et a l 4 1 , 4 2 as a representative orthorhombic structure unless the single crystal structures of the guest/ZSM-5 complexes are available (it has been shown that the exact coordinates are not essential for the success of these distance-driven NMR calculat ions 2 5 , 1 0 3 ) . In order to illustrate this point, various sets of framework coordinates for ZSM-5 including the empty orthorhombic crystal structure at high temperature (350 K ) 1 9 were used for the structure calculation, and all gave virtually the same structure. The algorithm of the structure determination program is presented in Figure 3.3. A rigid model of the guest molecule and the known zeolite framework structure are input, a set of the physically possible solutions is calculated within the ranges set by the six structural parameters, (x, y, z, {(>, 9, v)/), which are the translations of the 'defined' center of the molecule in (x, y, z) and the orientation of the 'defined' long axis in (<(>, 9, y), which are illustrated in Figure 3.4. The physical limit, an adjustable parameter limiting the closest allowed atom-atom approach between the guest and the framework, was input to avoid 'bumping' of the guest molecule into the framework atoms. For an organic guest molecule, the limit was usually set to 2 A which reflects the minimum possible interatomic distance for the known Van der Waals radii of hydrogens (unbonded), the 'outside atoms' in o-xylene, are 1.20 A, and oxygens (unbonded), the surface atoms in the framework of ZSM-5 , are 1.52 A in general. The following example is given for a variable contact time C P experiment of the saturated loading of o-xylene-de/ZSM-5 complex at 273 K (please refer to Chapter 4 for more details). Among physically possible locations found, the location {x = 0.489, y = 0.294, z = - 0.033, <|> = 37.7, 0 = 77.4, \\i = 17} satisfies the linear correlation requirement between M2 and /c,s with r2 = 0.975 and a standard deviation of regression (SDR) of 0.309 (Equation 3.10). Equation 3.10 71 where kis is the experimentally determined C P rate constant, n represents the number of resolved NMR peaks for silicon T-sites, and kIS is the line found from the linear regression of the solutions with r2 > 0.92, which can be represented as: where M2 represents the calculated heteronuclear second moment in Hz2 (Equation 3.8). The calculated M2 values (the program calculates values for all silicon T-sites) can be used to predict the best fitted line (kIS) derived from the C P rate constants for the silicon T-sites with the range of errors within the chosen confidence level defined as: where f(a/2) is a term from the Student's t-distribution (normal distribution) with n - 2 degrees of freedom. According to Equation 3.12, the intensities of the C P peaks including the overlapping peaks are predicted within the (1 - a/2) prediction intervals. For the overlapped peaks, the intensities are determined by calculating individual intensities for each contributing silicon T-site, and the sum of the determined intensities should be within a 'permissible' limit to the observed overlapped intensity. The description above can be exemplified by the overlapped peak from the silicon 1,7 T-sites in the o-xylene-oyzSM-5 complex. The calculated M2 values for Si1 and Si7 are 93,670 and 116,890 respectively, and the predicted kls values are 6.67 ± 0.80 s"1 and 7.44 ± 0.82 s"1 respectively. The intensity of the overlapped peak at a 10 ms contact time can be calculated from Equation 3.3 for the C P (Equation 3.6 for C P drain) experiment, 33.19 a.u. (arbitrary units) (the observed intensity for the overlapped peak is 33.73 a.u.), which is within the permissible limit. Thus, the solution at this location is accepted. kls = (3.30 xl0"5)M2+3.58 Equation 3.11 Equation 3.12 72 The option of generating symmetry equivalent locations for the calculations is also available. For instance, having the Prima space group for the guest/ZSM-5, the symmetry operation on the found solutions can add 'meta' solutions to the existing solutions by the symmetry operation of the reflection at y = Vi, which effectively doubles the number of solutions found. For this study, however, in general this option was not implemented. The final output of the solutions needs to be presented as sets of solutions as there could be many solutions that satisfy the selected structure determination criteria. Suitable sets include the number of solutions with a specified r2 value, the distributions of the structural parameters (x, y, z, (j), 6 and over a set, and the average location and orientation from the set of solutions given by the structural parameters. In the next chapter, detailed illustrations of the structure determinations for the o-xylene/ZSM-5 complexes are given, using these formats of the final outputs of the solutions. 73 NMR structure determination C P data (found k i sva lues ) Exper imenta l C P spec t rum Zeol i te f ramework input Rig id guest mode l R a n g e s of (x,y,z, <p,9,i)j) are c h o s e n Min imum d is tance (dmm) between the guest and f ramework c h o s e n Determinat ion of symmet ry operators within the limit c h o s e n (e.g. D is tCu tOf f = 8 A) Phys i ca l l y poss ib le locat ions (e.g. d'm> 2 A) S e c o n d moment (M2) ca lcu la t ion for the locat ions found and the l inear corre lat ion check between the exper imenta l k isand ca lcu la ted M 2 (mm r ' c h o s e n ) N M R intensity check between the predic ted and exper imenta l spec t ra (over lapped peaks are tested here) (min. r 2 c h o s e n ) C —i CO •a > 73 > 2 m H m 73 CO > r o c o z CO Opt ion of generat ing symmet ry equivalent so lut ions F ina l solut ions o C H TJ C H CO Figure 3.3 A flowchart of the structure determination procedure used by the program 2 5 in this study. Figure 3.4 A rigid model of a para disubstituted benzene guest molecule having its long axis defined by the two atoms which are furthest apart and its orientation in Euler angles. <|> represents rotation about the long axis, whereas 9 (with respect to the z axis) and \\i (rotation on the xy plane) represent the orientation of the molecule with respect to the crystal lographic 'axes as depicted. 74 3.2.4 'Goodness' of the proposed structure The re are seve ra l pa rameters wh ich c a n be used to dec ide the reliabil i ty of the structure solut ions found from the structure determinat ion ca lcu la t ion . T h e va lues of r2, the squa re of the s a m p l e correlat ion coeff ic ient, wh ich indicate the deg ree of correlat ion be tween the ca lcu la ted kIS va lues and the exper imenta l /c / s , are the mos t important m e a s u r e s b e c a u s e they s h o w how wel l the exper imenta l and predicted va lues of the s e c o n d momen ts ag ree with e a c h other. A mathemat ica l descr ipt ion of r2 is g iven in Equat ion 3.13: r2=l_^s_. Equat ion 3.13 °"» where a 2 =—— V(v,-a-fet ,) 2 and a2„, = —!— ^ (y> -yf , x, and y,- a re the exper imenta l data , a and b are the coeff ic ients f rom the l inear equat ion, y = a + bx, and y is the m e a n va lue of y,. A l o n g with the a v e r a g e structural parameters with the ranges of the errors, the distr ibut ions of the s ix structural pa rameters are other s igni f icant ' s igns ' of the reliabil ity for the so lu t ions a s the numbers of solut ions are dependen t on the r 2 va lues ; however , if the so lu t ions are rel iable, they shou ld be found within c l o s e ranges , and the numbers of the so lu t ions shou ld d e c r e a s e a s the r 2 va lues i nc rease . A 2 D or 3 D scat ter plot is a useful w a y of present ing all the st ructures at o n c e , a s a v isua l inspect ion of the distr ibut ions of all the so lut ions is poss ib le f rom it. Plot t ing of the exper imenta l and predicted N M R spec t ra a long with their d i f ference spec t rum is ano ther powerful w a y of v isua l iz ing the d i sc repancy or similari ty be tween the exper imenta l and ca lcu la ted data after the structure ca lcu lat ion is comp le te . La rge numbers of so lut ions cou ld be u s e d to indicate reliabil ity of the da ta ; however , caut ion shou ld be taken, a s the number of so lu t ions on ly d o e s not guaran tee reliability of the so lu t ions. Final ly, the structure of the guest /zeo l i te c o m p l e x itself is an unmis takab le w a y of present ing the so lu t ions; the program m a k e s it poss ib le to plot the gues t mo lecu le in error el l ipsoid fo rms, wh ich are s imi lar to those used in X - r a y c rys ta l lography to represent the thermal factors of a toms. In this study, w h e n e v e r necessa ry , the 5 0 % error e l l ipsoid presentat ions contain ing 5 0 % of the accep ted solut ions are g iven for the structures. 75 3.2.5 Verification of the proposed structure A s large crysta ls of Z S M - 5 are readi ly ava i lab le , w h e n e v e r poss ib le , a s ing le crysta l s tudy on a g u e s t / Z S M - 5 structure c a n be an unamb iguous w a y of veri fying the structure de te rmined by N M R , a l though this will not be poss ib le for other zeo l i tes where sui tab le crys ta ls a re not ava i lab le and not a lways even for MFI a s will be i l lustrated in Chap te r s 4 and 5. A s men t ioned in C h a p t e r 2, the s ingle crystal data w a s ref ined b a s e d on the f ramework topolog ies found in prev ious s t u d i e s 4 1 , 4 2 . W h e n s ingle crysta l X R D is unable to so lve the structure, powder diffraction techn iques must be u s e d , and these st rategies are now desc r i bed . 3.3 Powder diffraction structure determination" A s ment ioned in C h a p t e r 1, structural information in powder diffraction data is ei ther obscu red or lost due to the random distr ibution of crysta ls , wh ich c a u s e s heav i ly ove r l apped ref lect ions. In the c a s e of over lapp ing ref lect ions, measu r ing their individual intensit ies is s imp ly not poss ib le and this is the major l imitation in structure ana lys is of powder da ta . In order to c i rcumvent this p rob lem, m a n y different a p p r o a c h e s have been deve loped and tes ted, and mos t of t hem rely heavi ly on computa t ions . O n c e the s p a c e group and unit cel l of a powder mater ia l have b e e n es tab l i shed , the next s tep is the solut ion of its crysta l structure to find the distr ibution of a toms in the unit ce l l . T h e crystal structure so lv ing p r o c e s s starts with ana lyz ing sys temat i c a b s e n c e s to f ind the s p a c e group symmet ry of a mater ia l or at least to identify all p robab le s p a c e g roups , if the co r respond ing diffraction c l a s s conta ins more than one group. T h e next s tep requi res determin ing the content of the unit ce l l . O n c e the unit ce l l content has been es tab l i shed , a mode l of the crystal structure is c rea ted us ing ei ther direct or rec iproca l s p a c e techn iques , or a combina t ion of both. Direct s p a c e a p p r o a c h e s d o not manda te immedia te use of the obse rved integrated intensit ies, whi le rec iproca l s p a c e methods are b a s e d on them. " The purpose of this section is to provide an introduction and background to the method for researchers who may wish to pursue the general approach to structure determinations outlined in the thesis. It is based, in large part on the references 86, 108, 109, 110, 111, 124, 125. 76 The first approach is the direct method (or the reciprocal space method), which uses relationships between the reflected intensities to estimate the phases. As described in Chapter 1, for a single crystal, measuring the reflected intensities is straightforward and the method is reliable and robust. In powder diffraction, however, extracting reflections is obstructed by the heavily overlapped peaks. One way of overcoming this problem is using the high angle data from a synchrotron source, which show fewer overlapped peaks. Another approach to the powder data is to use direct-space methods, which involve model building of framework 'molecules', polyhedra and frameworks, and the comparison of the observed and calculated diffraction patterns. Thanks to modern computers and efficient algorithms developed over the past years, direct-space methods are well automated. In principle, it is a global optimization of many parameters generating the minimum differences between the observed and calculated diffraction patterns. For this thesis, the direct-space method is the usual choice for solving powder data." Figure 3.5 illustrates the general guidelines for structure determinations from powder diffraction data. Apart from the powder sample preparation and data collection, described in Chapter 2, indexing, solving and then refining the powder data are to be completed sequentially. At each step, there are a number of choices of method and selecting the most effective programs. n Several other approaches are available for the structure solution by powder diffraction. One method to note is a 'texture method', which probes the directional information by re-introducing the preferred orientations. A careful and elaborate sample preparation is necessary in order to use this method successfully, as it depends heavily on the preferred orientations of the crystals in the sample. 77 Structure Determination by Powder Diffraction Sample preparation finely ground, microcrystalline material Obtaining diffraction data by powder XRD, powder neutron diffraction Indexing the diffraction data Obtaining unit cell, spacegroups by Crysfire, TOPAS Figure 3.5 A flowchart of the powder diffraction strategy from the sample preparation to the structure refinement. This is the general guideline followed for this study, not necessarily for every powder diffraction study. Structure solution Direct methods (extracting |F(hkl)|) Global optimization in direct space by simulated annealing (FOX, TOPAS), parallel tempering (FOX), etc. Structure refinement Least-squares refinement (Rietveld method) GSAS, Fullprof, TOPAS 3.3.1 Indexing the diffraction data N o s u c c e s s is guaran teed in an automat ic indexat ion of powder diffraction da ta , and indexing powder da ta shou ld be done a s careful ly a s poss ib le . T h e data for index ing shou ld be o f the highest qual i ty at ta inable. T h e m a x i m u m d i sc repancy be tween the o b s e r v e d and ca lcu la ted peak posi t ions of the ref lect ions shou ld be less than ± 0.02 0 20. Th i s p o s e s a cha l l enge for convent iona l powder neutron da ta , wh ich has the best F W H M s over 0.1 ° 20 in mos t c a s e s . If the index ing of neutron data 78 does not give a satisfactory answer, alternative approaches must be used, such as indexing the powder X R D data of the same sample or using NMR data to obtain space group information. It is possible that the zeolite sample could have multiple phases present; thus, for reliable results, more than one indexing program should be used. In general, using TREORU6, DICVOL*27, or / T O 1 2 8 works well for zeolite samples, and for this study, the Crysfire su i te 1 0 6 , which compiles multiple indexing programs including those mentioned, was used. 3.3.2 Ab initio structure solution by global optimization The global optimization of FOX is implemented in so-called reverse Monte Carlo (MC) 129 approach The algorithm starts with a random configuration, and its parameters are varied randomly. In the Markov process, ensuring the sampling of the parameter space, the new structure generated is kept according to the well-known Metropolis algorithm™0 as illustrated in Figure 3.6. One interesting aspect of the M C type evaluation is that the process does not require generation of true random numbers, but only 'random-enough' (so-called pseudo-random) numbers, which can be provided by computer programs. In order to converge the solution by MC, the temperature T, which is the global parameter of a state function, needs to be lowered as the calculation proceeds. In order to do this, FOX uses two approaches: simulated annealing (SA) and parallel tempering (PT). Figure 3.7 describes the schematic representation of S A and PT. SA, conceptualized by S. Kirkpatrick, C. D. Gelatt and M. P. Vecch i 1 3 1 , and'independently by V. Ce rny 1 3 2 replaces the current configuration with a randomly chosen configuration with the probability depending on a T. In metallurgy, annealing increases the size of the crystals and decreases the overall defects by lowering the temperature of a molten (thus, freely mobile) metal sample gradually. Analogous to annealing, S A seeks for a more probable configuration that gives a lower T value. This process in FOX is carried out as: 1) systematic selection of the proposed structure if the cost function (CF) is better (smaller) than the previous configuration; 2) selection with probability e ( " a C F / 7 ) if the new configuration has a worse (larger) C F when the relative probability of two independent trial configurations (P1/P2) follows the Boltzmann factor (Equation 3.14). 79 5_ = e[(CF .-c^)/r] Pi Equat ion 3.14 In the p r e s e n c e of mult iple temperature m in ima , us ing S A a lone cou ld c a u s e a dange r of fal l ing into a local m in imum w h e n only one set of the temperature eva lua t ion is m a d e at a t ime, espec ia l l y for the highly corre la ted sys tem in e a c h conf igurat ion. In order to o v e r c o m e this, FOX a l so 133 u s e s the p r o c e s s of P T , wh ich invo lves exchang ing the conf igurat ions of two or more different tempera tures that have been ca lcu la ted after the s imu l taneous paral lel opt imizat ion of mult iple conf igurat ions. T h e compar i son of conf igurat ions is b a s e d on its C F va lues , wh ich a re def ined by the molecu la r information (energy mode l , bond d is tances and phys ica l ly poss ib le locat ions) , and/or by the exper imenta l da ta (e.g. , exper imenta l powder diffraction pattern). In the current vers ion of FOX, there are two types of C F , b a s e d on the exper imenta l diffraction patterns and phys ica l ly poss ib le mo lecu la r locat ions. T h e first C F is b a s e d on the weighted profi le R (Rwp) factor over the powder diffraction pattern (Equat ion 3.15). whe re Y°bs is the obse rved and Y™'c is the ca lcu la ted intensity of the / da ta point, and w, is the weight of the Z"1 data point usual ly g iven a s wi - l/a1 = l / / ) 0 * 5 , whe re a = (N is the total number of photon counts registered by the detector). T h e s e c o n d C F , wh ich is very s imi lar to determin ing the phys ica l l y poss ib le locat ions in the N M R structure determinat ion p rog ram, is a so -ca l led 'ant i -bump' funct ion; w h e n two a t o m s are c loser than m in imum d is tance , a penal ty is a d d e d to its C F , mak ing the conf igurat ion l ess probab le . A C F is a posi t ive real number , and a sma l le r va lue of C F represents a better conf igurat ion. x l 0 0 % Equat ion 3.15 I - , ( r * ) 2 80 Keep configuration with probability -ACF P = e T No Choice of parameters initial configuration Random configuration i f change Evaluation of the cost function (CF) 3£ 3 , better configuration? Keep configuration Yes Figure 3.6 The algorithm of FOX, which uses simulated annealing in order to evaluate the cost function. Chart is adapted from reference 108. Con f i gu ra t i on B \ r Con f i gu ra t i on A - A Con f i gu ra t i on C Con f i gu ra t i on C h a n g e Figure 3.7 A schematic view of simulated annealing (SA) and parallel tempering (PT), which is implemented by FOX. Each configuration is a result of SA , and two or more configurations are exchanged to find a global minimum. 81 3.3.3 Structure refinement T h e structure c a n be further ref ined by the Rietve ld method in order to determine the final structure a s the method imp lements more parameters in order to inc lude s e v e r a l exper imenta l factors. Dur ing the Rietve ld ref inement , a se r ies of equat ions are so lved by m e a n s of a non- l inear least squa res min imizat ion. ycalc f(Y0^s ycalc ltY°^s M Equat ion 3.16 ycalc _ J^Y M ycalc _ J^Y°^S n n where Y°bs is the obse rved and Y.cah is the ca lcu la ted intensity of a point / of the powder diffraction pattern, k is the pattern s c a l e factor and n is the total number of m e a s u r e d data points. In its s imp les t fo rm, the min imized funct ion, <t> c a n be g iven a s : n $> = YJwi(Y°bs-Y'akf Equat ion 3.17 1=1 where w, is the weight a s s i g n e d to the ;*h data point and k is unity (k = 1). Fo r a s ing le -phase crystal l ine mater ia l , the equat ion b e c o m e s : n m * = YJWWS " f t W * y ) D 2 Equat ion 3.18 i=i y=i for a s ingle wave leng th exper iment or n m ® = Yuw> - ft + KYJ h {y> { X J } + ° - 5 ^ (*J + ) } ] ) 2 Equat ion 3.19 '•=i >i for dua l wave leng th (Ka^ + K a 2 ) data (a c o m m o n c a s e for an X R D exper iment ) . In Equa t ions 3.18 and 3.19, o, is the backg round parameter at the th da ta point, K is the p h a s e s c a l e factor, m is the number of B ragg ref lect ions contr ibut ing to the intensity of the ;*h da ta point, /, is the integrated intensity of the / h B ragg ref lect ion, y/xy) is the peak s h a p e funct ion, Axy is the di f ference in posi t ions of Kai and K a 2 componen ts in the doublet profi le, and = 2Qcf - 2 9 , . . O n e thing to note at this point regard ing the min imizat ion of the funct ion cp is the signif icant 82 contribution of the background parameter, o,. If b t » K^Ijy^Xj), the least square refinement depends heavily on the background function. In other words, in the presence of a heavy background profile, the refinement on a crystal structure would never be acceptable. In the absence of a background, the weight w, can be given as: w, =[Y°bsY' Equation 3.20 although the value of w, in practice is calculated without subtracting the background. In the case of the multi-phase crystalline structure, the minimization function, O can be expressed for a mixture of p number of phases as: n p m « , = E w ' ( ^ 0 t a - ^ + E / : ' Z V / j ( J C « ) D 2 Equation 3.21 /=i . /=i j=\ for a single wavelength experiment, and n p m ® = Z w ' ( r ' ° f a " f t +YaK'lLI'MM'J) + 0-5y'Mi + ^ y)}]) 2- Equation 3.22 i=i ;=i j=\ for dual wavelength (Kc^ + Ka 2 ) one, which are common for X-ray diffractometers that use copper to generate X-ray radiation. The following parameters are usually refined during the Rietveld method. a. Background parameters. b. Sample displacement, sample transparency or zero-shift corrections as needed. c. Peak profile parameters including full width at half maximum, asymmetry, and the parameters for chosen peak shape functions. d. Unit cell dimensions of the structure. e. Preferred orientation, absorption corrections or extinction parameters if necessary. f. Scale functions for each phase (K,) g. Positional parameters of all independent atoms for each crystalline phase, usually (x, y, z) in the crystallographic fractional coordinates. h. Population parameters (multiplicities and site occupancies) i. Atomic displacement parameters, which can be treated as an overall displacement parameters or individual atomic displacement parameters. 83 3.3.4 Qua l i t y o f R e s u l t s T h e qual i ty of the Rietve ld ref inement c a n be judged by seve ra l f igures of merit s u c h a s profile res idua l Rp, we igh ted profile res idua l Rwp, B ragg res idua l , RB, e x p e c t e d res idua l Rexp, and g o o d n e s s of fit, x2- T h e y are all b a s e d on c o m p a r i s o n s be tween exper imenta l and the predic ted peak s h a p e s . Al l the parameters depend on both the profi le and structural pa ramete rs with except ion of the B ragg res idua l , RB, wh ich d e p e n d s on structural pa rameters a lmost exc lus ive ly . T h e fol lowing are the appropr iate equat ions : Y^yobs _yc, -xl00% Equat ion 3.23 1=1 " 2 Yjyvi{Y"hs - Y'ak^ 1=1 ' i > , ( ^ ) 2 1=1 m \ 1 jobs jcalc 2J[ j j I ( RB = x 100% 7=1 x!00% Equat ion 3.24 Equat ion 3.25 Rexp -n- p I > , ( r ) 2 x!00% Equat ion 3.26 2 n-p Equat ion 3.27 where n is the total number of points m e a s u r e d in the exper imenta l powde r diffraction pattern, Y°bs is the obse rved and Y^h the ca lcu la ted intensity of the / h da ta point, iv, is the weight of the / h data point usual ly g iven a s wi = l /a , 2 = l/Y°bs, where a = y[N (N is a total number of photon counts registered by the detector) , m is the n u m b e r of independent B ragg ref lect ions, if* is the o b s e r v e d and lfc is 84 ca lcu la ted integrated intensity of the j B ragg ref lect ion, and p is the n u m b e r of f ree least squa res parameters . However , no matter wh ich f igure of merit is p resen ted , none c a n match the graph ica l representat ions of the s c a l e d plots of the obse rved and ca lcu la ted powder patterns with their d i f ference on an e x p a n d e d intensity s c a l e as it is a g raph ica l representat ion that m a k e s the Rietve ld method straightforward, yet powerfu l . 3.4 Summary In this chapter , certain theoret ical aspec t s of the N M R exper iments and powder diffraction method used for this thes is work have been i l lustrated and d i s c u s s e d . N M R is capab le of determin ing the structure of a guest /zeol i te comp lex as has been s h o w n in seve ra l s tud ies. In addi t ion, it is an inc is ive method for deduc ing s p a c e group information for c o m p l e x st ructures a s the numbers of N M R p e a k s reflect the number of crysta l lographica l ly inequiva lent s i l icon T-s i tes of the f ramework . However , it requi res know ledge of the f ramework topo logy in order to ca lcu la te the d ipolar coup l ing st rengths with the sp ins on the gues t mo lecu les . P o w d e r dif fract ion, on the other hand , is c a p a b l e of so lv ing the f ramework structure in the a b s e n c e of s ing le crysta l da ta . Arguably , the mos t cha l leng ing part of the structure determinat ion by powder diffraction is the ab initio determinat ion of the structure, espec ia l l y for guest /zeo l i te c o m p l e x e s a s the gues t mo lecu les in genera l a re weak l y scat ter ing. Cur ren t l y /ne i ther of the two current ly ava i lab le me thods c a n compe te with s ing le crysta l diffraction, and the two methods shou ld be c o n s i d e r e d a s complementary . However , in m a n y c a s e s whe re a s ing le crystal X R D is not poss ib le , they a re c a p a b l e of produc ing structural information of guest /zeol i te c o m p l e x e s , wh ich w a s not ava i lab le in the past . 85 C h a p t e r 4 tt NMR Determination of the Structure of o-xylene/ZSM-5 4.1 Introduction Central to the many commercial and industrial applications of zeo l i t es 1 3 4 , 1 3 5 as sorbents, catalysts and for gas separations, are the size and shape selectivities that they exhibit towards organic molecules. In order to understand the host/guest interactions that control these effects with the ultimate aim of describing them theoretically, it is important to have access to a reasonable number of reliable structures of complexes of this type. However, relatively few such structures exist. This is primarily due to the very small crystal dimensions obtained for most zeolites which preclude the use of single crystal X-ray diffraction techniques (an important exception is the zeolite ZSM -5 , MFI topology, for which large enough crystals can be grown and a limited number of single crystal X-ray structures of this system are available from the work of van Koningsveld and coworkers 1 9 , 2 0 , 2 3 , 4 0 " 4 4 , 9 3 ) . Thus, in most cases, recourse must be made to powder methods which yield much more limited data and, while it is often possible to obtain the framework structure, the locations and orientations of occluded templates or adsorbed organic molecules are much more difficult to define accurately. As an alternative to diffraction techniques, a method based on solid state NMR spectroscopy has been introduced that uses the distance-dependent through space dipolar interactions between nuclei on the organic guests and the silicon nuclei in the host framework to solve these structures. The method has been tested on the known structure of the high loaded, form of zeolite ZSM -5 with p-xylene and then used to predict the structure of the low-loaded form which was then confirmed by single crystal X-ray diffraction 1 3 6. More recently, the method has been optimized computationally, the display of the results standardized and its robustness demonstrated with respect to temperature and ** A version of this chapter has been submitted for publication. C A Fyfe and J S J Lee. Solid-State N M R Determination of the Zeolite ZSM-5 - o-Xylene Host - Guest Crystal Structure. Journal of Physical Chemistry 2007 submitted. 86 to the exact structure of the zeolite framework . Several specific examples have been investigated 4 5 , 4 6 including those of this chapter. A classic example of the selectivity towards organic molecules is the behavior of zeolite Z S M -5 towards the xylene isomers which is central to industrial xylene synthesis, isomerization and separation processes. 3 9 Adsorption) and diffusion s tud ies 1 3 7 , 1 3 8 have both been carried out. S tud ies 1 3 7 , 1 3 9 of the adsorption and diffusion of the o-, m- and p-xylenes have identified that the three have very different behaviors in ZSM-5 , and these are the basis of the industrial xylene isomerization process 3 9 . The p-xylene in ZSM-5 , which diffuses 1000 times faster than the other xy lenes 1 3 7 , has been well understood from the studies of the s t ructura l 2 6 , 4 0 , 1 3 6 and dynamic 1 4 0 " 1 4 2 aspects. The structural information for either of the other xylenes is lesser known, which could provide important information on the selective behavior of ZSM-5 . Currently, no known single crystal structure of the o-xylene/ZSM-5 complex is available. Nagy and co-workers 1 4 0 studied the 1 3 C NMR spectra of the o-xylene/ZSM-5 system to investigate the motions of the sorbate in the framework and concluded that at a loading of 3.7 molecules per unit cell there was rotation of the methyl groups around their C 3 axes but that the rings were "almost static". 1 4 0 The most informative work is a powder X-ray diffraction study by Nair and Tsapatsis from a Rietveld refinement of the data who proposed that the organic molecules were located at the channel intersection. 5 1 , 5 2 The purpose of the present work was to determine the structure of the complex formed with o-xylene by solid state N M R and, if possible, to subsequently confirm it by single crystal X-ray diffraction. 87 4.2 Solid State NMR experiments on the o-xylene/ZSM-5 4.2.1 Characterization of the 2 9 Si spectra of the o-xylene/ZSM-5 complex A preliminary 2 9 S i M A S experiment on a sample containing ca. 4 molecules per unit cell of o-xylene in ZSM-5 yielded a broad spectrum (Figure 4.1(a)). After the sample was purged with N 2 gas for 12 h and N 2 gas subsequently used as both drive and bearing gas, the 2 9 S i MAS spectrum changed drastically, giving sharply resolved peaks (Figure 4.1(b)) and illustrating the effect that paramagnetic 0 2 in the channels has on the relaxation mechanisms of both 1 H and 2 9 S i for this system. Upon removal of 0 2 by purging with N 2 gas, all relaxation times (Ti, T2 and 7 1 p of both 1 H and 2 9 Si ) become longer in the temperature range studied. In particular, the 2 9 S i spin-spin relaxation T2 becomes longer (of the order of ms), leading to narrower, more clearly resolved peaks. One of the drawbacks of N 2 purging is that the 2 9 S i spin-lattice relaxation time, Tu also becomes long (>200 s in the present case) so that the accurate measurements of Ti become difficult and relaxation delays considerably longer. The increase in Ti relaxation times (Table 4.1) is more pronounced for 2 9 S i , which generally has longer 7i values than 1 H . -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 2 9Si Chemical Shift (ppm from TMS) "SiChemical Shift (ppm from TMS) Figure 4.1 Effect of N 2 purging on the sample: (a) Si M A S spectrum of the o-xylene complex at 273 K before purging with N 2 gas (b) spectrum at 273 K after N 2 purge for 12 h at room temperature. 88 Table 4.1 Selected Relaxation Parameters for the 1 H and 2 9 S i Nuclei in the o-xylene/MFI Complex Experimental Relaxation Air driven N 2 driven temperature parameter 2 9 S i 1H(ring) 1H(CH3) 2 9 S i ca. 5xT! (s) 20 6.5 5 > 1000 273 K T 2 (ms) 4.4-5.7 37 0.3 21.8-40.6 Tip(iH) (ms) 11.5 13.1 ca. 5xT, (s) 50 5 3.5 > 1000 315 K T 2 (ms) 5.5-8.3 52 0.2 20.8-63.1 Tip(iH) (ms) 17.5 - 12.4 4.2.2 One-dimensional Si NMR The ZSM-5 framework undergoes changes in space group depending on the number of organic molecules per unit cell and the temperature. 3 7 For example, the empty ZSM-5 monoclinic structure with 24 T-sites transforms completely to an orthorhombic Pnma space group with 12 T-sites at room temperature when its channel intersections are occupied with small organic sorbates in loadings of 2 to 4 molecules per unit cell and the 2 9 S i NMR spectra show corresponding changes in the number of peaks. Similar effects are seen when the temperature is changed. 3 7 2 9 S i NMR spectra were therefore acquired to determine the range of temperatures over which the complex exhibited the orthorhombic Pnma space group. Figure 4.2 shows the 2 9 S i M A S N M R spectra at temperatures from 260 K to 330 K, and indicates that the orthorhombic Pnma space group is retained over all of this fairly wide temperature range. From the variable temperature NMR spectra in Figure 4.2, those at 273 K and 315 K had the clearest peak splitting patterns and most resolved peaks and all further experiments were carried out at these two temperatures. From the peak assignments described later, at 273 K there are ten resolved signals in the 2 9 S i NMR spectrum eight of which correspond to single 2 9 S i T-sites (Figure 4.3(a)) and at 315 K there are nine resolved signals, seven of which correspond to single 2 9 S i sites (Figure 4.3(b)). The 2 9 S i NMR resonances were assigned using INADEQUATE exper iments . 2 7 , 2 9 " 3 1 7 9 These 89 show 22 connections at 273 K and 21 connections at 315 K (Figure 4). Since the connectivities of each silicon site are known for the ZSM-5 topology143, each peak in the NMR spectrum can be assigned to a specific silicon T-site. This was done using a program that generates all possible assignments of the resonances and selects all those in agreement with the observed connectivities.103,117 From symmetry considerations, there are two possible assignments in each case and the final selection was arrived at by testing both in the initial structure determinations where it was found that only one gave a significant number of structure solutions. The experiments also confirmed that there were overlapping resonance peaks as indicated in Figure 4.3. 90 -110 -112 -114 -116 -118 -120 2 9 S i Chemical Shift (ppm from TMS) Figure 4.2 2 9 S i MAS NMR spectra of ZSM-5 loaded with ca. 4 molecules of o-xylene per unit cell at the temperatures indicated. The 2 9 S i 15 ° pulse length was 1.8 us and 128 scans were accumulated with a recycle delay of 15 s for each spectrum. 91 <a> (1.7) (b) (5,6,9) ( 1 | 7 ) -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 ppm ppm Figure 4.3 1H/29Si CP MAS NMR spectra of ZSM-5 loaded with ca. 4 molecules of o-xylene per unit cell (a) at 273 K (b) at 315 K, contact time 4 ms, recycle delay 5 s with 1024 scans and 800 scans respectively. The numbers above the resonances indicate their assignments to specific T-sites. 'Si Chemical Shift (ppm from TMS) J 9 S i Chemical Shift (ppm from TMS) Figure 4.4 Two-dimensional 1H/29Si CP INADEQUATE spectra of the o-xylene/ZSM-5 complex at 273 K (left) and 315 K (right) together with their quantitative 29Si NMR spectra (recycle delay of 200 s). 36 experiments, each with 384 scans and 5 s recycle delay, were acquired in the U dimension, the echo delay during the double quantum preparation period was 18 ms and the sweep widths in the h and U dimensions were 1400 and 2800 Hz respectively. The indicated peak assignments were determined from the observed correlations and the numbers above the resonances indicate the assigned silicon T-sites. 92 4.2.3 Cross Polarization Experiments. The structure determination method has been described in detail prev iously 2 5 , 1 0 3 and in Chapter 3. It depends on using the through-space, distance dependent dipolar interactions between the 1 H nuclei on the organic molecule and each of the different framework 2 9 S i nuclei whose resonances have now been assigned. Since the method relies on reasonably localized sources on the organic molecule for the magnetization transfer, two specifically deuterated o-xylene isotopomers were used (Figures 4.5 (a), (b)) and experiments were performed on complexes of each. In addition to improving the quality of the structures, this also provides an incisive check on their validity. In terms of the spin dynamics of the cross polarization experiment, the parameter of interest for the structure determinations is the cross polarization time 7"Cp (or 1//c/s , where k/s is the rate constant for the magnetization transfer). The C P curves were first fit in terms of three variables, using single values of / 0 and ki common to all data sets which gave the best fittings, and separate ktsvalues for each site. Subsequently, independently measured kt values (obtained by a spin-locking pulse sequence) were compared to the fitted ki values, and the two were found to be comparable to each other in magnitude. The independently determined experimental kt values were then used to fit the C P curves of some experiments. However, the fittings were found not to be as good as when k, was allowed to vary, especially at longer contact times. Because the obtained k,s values are highly correlated to l0 and the ks values are neglected in the fitting of the C P curves, a fitted kiS value is denoted as k'iS, which indicates a "relative" C P rate constant (assuming constant / 0 and ki). In C P drain experiments described later in this chapter, no such approximations of the correlated values are needed and the obtained kls values are absolute values, and in that way, in principle the C P drain experiments are superior to the conventional C P experiments although relaxation considerations may produce poorer experimental data in practice. 93 H2 H3 D2 D3 \ / . \ / C± C 5 C 4 C 5 C 2 C| C 2 / \ / \ c 8 c 7 c 8 c 7 / \ • • / \ D8 D7 H8 H7 (a) (b) Figure 4.5 The two specifically deuterated o-xylenes used (a) both methyl groups are deuterated (o-xylene-cfe) (b) protons in the benzene ring are deuterated (o-xylene-04). 67,8 and H7, 8 are the simplified representations of the methyl deuteriums and methyl hydrogens at the centers of their equilateral triangles. 4.2.4 CP Experiments on o-xylene-ayZSM-5 Figure 4.6 shows the curves from a variable contact time experiment on the o-xylene-cVMFI complex at 273 K. The general appearance of the curves, together with the low efficiency of the C P experiment compared to a fully relaxed single pulse experiment, are indicative of the system being in the slow exchange regime, described before in Chapter 3. The curves were fitted to Equation 3.3 using a non-linear least squares fitting procedure constrained by single, common values of / 0 and kL Because of the correlation of kts and / 0 and the neglect of ks, these values are relative values only and are designated as k'iS values. As seen from the figure, silicon T-site 8 (Si8) in ZSM-5 shows the highest k)s value, which indicates close proximity to the proton source. On the other hand, Si10 shows the weakest peak growth indicating it is the furthest silicon atom from the source protons. Qualitatively, these relative proximities of the silicon sites to the protons rule out the o-xylene being located in the zigzag channel of ZSM-5 . As an alternative analysis, an independently measured kt value (86.7 s"1) was used for the fitting.of the C P curves instead of that obtained from the first fitting (127 s"1). The results are presented in Figure 4.7 and are similar in quality. Fitting with the experimental value of k, gave different k)s values but the relative values are the same and both sets yielded the same final structure when carried through the complete structure analysis described below. Figure 4.8 shows the results from the C P experiment on the same sample at 315 K. The values are somewhat different from the results from the 273 K data but the relative order of the /c,s 94 values is the same. At both temperatures, the fitted C P rate constants k'iS are smaller than the kt values (both fitted and experimental), confirming that both systems are in the slow C P regime. Si8 Si2 k',s = 9.34 s"1 k',s = 7.36 s'1 •• • • 9 f f 1 Si11 k'ls = 4.99 s Si12 k'ls = 5.38 s'1 Si9 k'ls = 8.86 s"1 Si6 k'is = 5.94 s"' Si10 k-,s = 4.85 s 1 50.00 40 00 30.00 20.00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0 00 50.00 40.00 30.00 20.00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0.00 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Figure 4.6 Variable contact time 2 9 S i CP MAS NMR experiment on ZSM-5 loaded with ca. 4 molecules of o-xylene-cfe per unit cell at 273 K. The points are the experimental values of the intensities and the solid curves are calculated according to Equation 3.3. The vertical axes represent the NMR signal intensities in arbitrary units and the horizontal ones the range of contact times in ms. The constant theoretical CP maximum / 0 was 433 (a.u.) and/c, was 127 s' 1 (Tip = 7.87 ms). Si5 k'is = 5.12 s" / • ~» ; - f / Si3,4 •2 Si1,7 » k'is * = 7.01 s'1 f k ' i s ' = 4.86 s'1 95 50.00 40.00 30.00 20 00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0.00 SIS ICts = 17.3 s'1 Y Si11 k%s = 8.71 s ' Si9 k'is =16.5 s'1 Si6 k'is = 10.6 s"1 0 10 20 30 40 50 60 Figure 4.7 Intensities of the Si C P M A S N M R signals indicated as functions of the contact time for ZSM-5 loaded with ca . 4 molecules of o-xylene-cfe per unit cell at 273 K. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.3. The vertical axis represents the NMR signal intensity in arbitrary units and the horizontal one indicates the range of contact times in ms. The constant theoretical C P maximum k was 202 (a.u.) and the value of the independently measured ki used was 86.7 s" 1 ( 7 i p = 11.5 ms). 96 50.00 40.00 30.00 20.00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0.00 Si8 k'ls = 4.95 s'1 Si4 k'is 2.56 s"1 Si10 k'is = 3.24 s'1 Si2 4.34 s 1 Si11 k'ls = 3.38 s'1 Si1,7 «r're = 4.04 s"1 Si3 k'is =4.19 s"1 £ 1 Si12 k'ls = 3.30 s - ' Si5,6,9 * ' , s = 4.82 s 1 10 20 30 40 50 60 0 10 20 30 40 50 60 0 2 9 c 10 20 30 40 50 60 Figure 4.8 Variable contact time Si CP MAS NMR experiment on ZSM-5 loaded with ca. 4 molecules of o-xylene-cfe per unit cell at 315 K. The points are the experimental values of the intensities and the solid curves are calculated according to Equation 3.3. The vertical axes represent the NMR signal intensities in arbitrary units and the horizontal ones the range of contact times in ms. The constant theoretical CP maximum was 421 (a.u.) and ki was 92.7 s"1 (7ip = 10.8 ms). The independently measured 1 H T i p was 17.5 ms. 4.2.5 Solving the structure of the o-xylene/ZSM-5 complex The structure determinations of the o-xylene/ZSM-5 complex were carried out using the program mentioned following the program protocol described in Chapter 3. The calculations were simplified using a rigid body model for o-xylene and the approximation that the methyl protons of o-xylene-c/4 (or deuteriums of o-xylene-de) could be replaced by locating them all at the center of their respective equilateral triangles in the methyl groups (Figure 4.9). The lower limit of framework atom-sorbate atom distances was set to 2.5 A in order to compensate the simplification of the methyl group protons (or deuteriums). Figure 4.9 describes the location i.e. xyz of the center of the ring and the Euler angles with which the location and orientation of an organic molecule in the framework of ZSM-5 can be described, using x, y and z for translation in the fractional coordinates, and 9 and \\i for orientation in angular degrees. In order to define an arbitrary long axis as previously used for p-xylene, for o-xylene an imaginary line was chosen that bisects the line between two methyl groups and passes through the center of the benzene ring in the molecular plane. In order to ensure that this choice did 97 not exclude any possible solutions, other long axes such as the one that crosses H1 and H4 and the one that crosses H3 and D8 were tested and were found to yield the same results in the structure determinations. Initially, all, possible locations and orientations were systematically tested, and those that did not involve too close of contacts with the framework atoms were selected. After all the physically reasonable locations were found, the heteronuclear second moments, M2, between the 1 H nuclei in o-xylene and 2 9 S i nuclei in the framework of ZSM-5 were calculated for each location. The cutoff value for the heteronuclear second moment calculations was set to 8 A and the second moments for all of the single, resolved 2 9 S i resonances were calculated. The calculated M2 values were then correlated to the experimental /c',s values in order to determine if there was an acceptable linear relationship between the M2 and k)s values. The degree of linearity was expressed in terms of the least squared r2 values, with a higher r2 value indicating a better linear correlation and the particular location and orientation accepted or rejected in terms of a preselected cutoff value of r2 If the position was still acceptable, at this point, the linear regression of the above correlation was used to assign the kIS values of the overlapping resonances not used in the previous correlation and the kts values used to predict the predicted intensities of all of the peaks in the complete 2 9 S i spectrum including all the overlapping peaks. These were now used in a second correlation of M2 and k'is values and a further selection made in terms of a preselected value of r2. Initially, for the 273 K data, all the possible locations were sampled with large steps in the translations (up to 0.03 in the fractional coordinates) and orientations (up to 7 degrees in angle) for the o-xylene molecule. This involved, in the framework of ZSM-5 , ranges in translations of 0.40 < x < 0.60, 0 < y < 0.5, and -0 .15 < z < 0.15 in the fractional coordinates and in orientations of 0 < cb < 180, 0 < 0 < 360, and -180 < vj/ < 180 in degrees for the straight channel. Considering the channel intersection is located at (0.5, 0.25, 0), these limits cover the entire channel and most of the channel intersection along the straight channel in the y-direction taking framework symmetry into account. In order to confirm the absence of the organic molecules in the zigzag channel of ZSM-5 , translations of 0 < x < 0.5, 0 < y < 0.5, and -0 .5 < z < 0 and orientations of 0 < cj> < 180, 0 < 0 < 360, and -180 < \\i < 180 were tested in a separate calculation. Those ranges cover the whole zigzag channel and some 98 part of the straight channel. From these results, approximate values of the six structural parameters were found that gave solutions with r 2 > 0.7, and the o-xylene molecule was located in the channel intersection. Based on this approximate location, for the 273 K data, a more detailed calculation was carried out for translations of x (step size of 0.010) between 0.42 and 0.55 in the fractional coordinate, y (step size of 0.010) between 0.1 and 0.4 in the fractional coordinate, and z (step size of 0.016 A) between -0 .12 and 0.02 in the fractional coordinate along with orientations of § (step size of 4 degrees) between 10 and 70 degrees, 9 (step size of 5 degrees) between 25 and 150 degrees, and y (step size of 4 degrees) between - 3 0 and 60 degrees. A total of ca. 40 million possible locations and orientations were tested, and 3783 solutions were found with r 2 > 0.91. The 315 K data were tested over the same ranges and step sizes as used in the more refined calculation on the 273 data set. A total of ca. 38 million possible locations were tested, and 7077 solutions were found with r 2 > 0.91. After the overlapped peaks were included for the final peak intensity calculation, the final number of solutions was reduced to 1581 (r2 > 0.91) for 273 K and 1548 (r2 > 0.91) for 315 K. Figures 4.10(a), (b) show the distributions of the structural parameters for the acceptable solutions found for 273 K and 315 K with the values indicated in the captions. Figure 4.11 shows the spatial distribution of the acceptable solutions as the scatter plots in 2-D and Figure 4.12 in 3-D in terms of atom positions for the final solutions at 273 K and 315 K with r2 > 0.92. For 273 K, a total of 1152 solutions were found that satisfy r 2 > 0.92. At 315 K, a total of 1017 solutions were found that satisfy r2 > 0.92. The black arrows above the curves in Figure 4.10 represent the average values of the different parameters for r2 > 0.92, and are summarized in Table 4.2. From these average structural parameters, the average location of the o-xylene molecule in ZSM-5 was determined for both temperatures, and the resulting atomic coordinates are listed together with those of the framework silicon atoms of ZSM-5 in Table 4.3. Figure 4.13(a) shows the degree of linear correlation between the k'iS and M2 at 273 K for the average of the acceptable structures with r2 > 0.92 and Figure 4.13(b) the corresponding data at 315 K, with r2 > 0.92. From these linear correlation graphs, the calculated values of /c',s were obtained and used to calculate the theoretical individual Lorentzian lines for the 12 silicon sites including the 99 overlapping peaks, which were compared to the experimental NMR spectra for one specific contact time. Figure 4.14(a) shows that the experimental NMR spectrum at 273 K is very similar to the calculated NMR spectrum for the average values of the six structural parameters with r2 > 0.92. Even for the overlapping peaks, the differences between the experimental and calculated spectra are small given that there may well not be exact peak overlap. Figure 4.14(b) shows the corresponding spectra of the complex at 315 K where the differences between the experimental and calculated NMR spectra are again very small. The coordinates of the average structures (Table 4.3) can now be used to construct 3-dimensional pictorial representations. The final structures from these experiments are presented in Figures 4.15(a), (b) and a representation with the framework silicon atoms labeled in Figure 4.16 for 273 K. For the average values of the structural parameters, the o-xylene ring center is located at the channel intersection between the zigzag channel and straight channel of the ZSM-5 framework (fractional coordinates of the ring center are {0.489, 0.294, -0.033}) with the arbitrary long axis of the o-xylene molecule lying along the crystallographic b axis with a deviation of 17.0 ± 13.5 ° (Figure 4.15(a)). The structure of the o-xylene-de/ZSM-5 complex at 315 K presented in Figure 4.15(b) shows that it is almost identical to that found at 273 K, providing strong support for the solutions and indicating that the method is quite robust to temperature changes over this 40 degree range. z y Figure 4.9 A graphical representation of the Euler angles used to describe the orientation of the o-xylene-cfe molecule in space. The angle 0 refers to rotation of the 'long axis' of the molecule from the z-axis; y for the rotation of the projection on the xy plane around the z-axis; § for the rotation around the long axis of the molecule. The red dashed line indicates the arbitrary 'long axis' of the molecule. 100 (a)1000 800 Figure 4.10 Distributions of the structural parameters for the solutions determined from 1 H/ 2 9 Si CP data of o-xylene-d6 in the framework of ZSM-5 at (a) 273 K and (b) 315 K with linear correlations of r2 > 0.91 (red squares), r > 0.92 (blue circles), and r2 > 0.93 (green triangles). The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of solutions by six structural parameters for the translation (x, y, z) of the center of the o-xylene-cfe (i.e. the center of the benzene ring) in fractional coordinates and the orientation (<|>, 0, vy) of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with r2^ 0.92, which are shown in Table 2. Table 4.2 Average values of the six structural parameters of o-xylene-cfe in the framework of ZSM-5 at 273 K and 315 K with r2 > 0.92 by CP. Temperature x y z <|> (3 vj/ 273 K 0.489(7) 0.294(12) -0.033(11) 37.7(56) 77.4(108) 17.0(135) 315 K 0.486(7) 0.292(14) -0.042(11) 43.7(41) 85.7(105) 6.4(100) The fractional coordinates (x, y, z) define the center of the benzene ring and the angles (<|>, 0, cp) in degrees define the orientation of the 'long axis', which is an arbitrary line bisecting the line between the two methyl groups and passing through the center of the ring on the plane of the molecule. The uncertainties for certain parameters are large, especially those of angles. For example, cp of the 315 K shows deviation of 20 degrees (6.4 ± 10.0°). 101 Figure 4.11 Scatter plots of o-xylene molecules in the plane of the molecule for ZSM-5 loaded with ca. 4 molecules of o-xylene-cfe per unit cell as a graphical representation of solutions of acceptable o-xylene locations in the framework at r2 > 0.92. (a) 273 K and (b) 315 K. The colored dots represent the atomic locations of the hydrogen (green), carbon (red) and deuterium (blue) in o-xylene-cfe. Figure 4.12 3-D scatter plot at r2 > 0.92 of o-xylene molecules in Z S M - 5 loaded with ca . 4 molecules of o-xylene-cfe per unit cell at 273 K. The colored dots represent the atomic locations of the hydrogen (green), carbon (red) and deuterium (blue) in o-xylene-cfe. 4 102 M2(Hz2) M2(Hz2) (a) (b) Figure 4.13 Plot of the measured C P rate constants against the calculated heteronuclear second moments for the average location of o-xylene in 2 S M - 5 loaded with ca. 4 molecules of o-xylene-cfe per unit cell with ? > 0.92 at (a) 273 K and (b) 315 K. The solid lines are the lines of best fit and the dashed lines represent the 95% confidence prediction intervals. -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 2 9 S i Chemical Shift (ppm from TMS) 2 9 S i Chemical Shift (ppm from TMS) Figure 4.14 NMR spectra for ZSM-5 loaded with ca . 4 molecules of o-xylene-cfe per unit cell at (a) 273 K and (b) 315 K; the spectra, in descending order, are the experimental spectrum (contact time = 10 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra. 103 Table 4.3 Fractional atomic coordinates for the o-xylene-d6/ZSM-5 complex at 273 K and 315 K from C P experiments 3 Temp. atom X y z Si1 0.4228 0.0566 -0.3456 Si2 0.3043 0.0286 -0.2020 Si3 0.2810 0.0630 0.0202 Si4 0.1232 0.0641 0.0191 Si5 0.0712 0.0277 -0.1922 Si6 0.1813 0.0574 -0.3419 Si7 0.4215 -0.1723 -0.3332 Si8 0.3040 -0.1294 -0.1932 Si9 0.2740 -0.1725 0.0250 Si10 0.1192 -0.1735 0.0217 Si11 0.0684 -0.1302 -0.1902 Si12 . 0.1827 -0.1726 -0.3286 C1 0.4390 0.3197 -0.0960 C2 0.4814 0.3638 -0.0450 C3 0.5316 0.3385 0.0175 C4 0.5394 0.2692 0.0291 C5 0.4970 0.2251 -0.0219 C6 0.4468 0.2504 -0.0845 273 K C7 0.4022 0.2036 -0.1382 C8 0.5050 0.1518 -0.0100 H1 0.4068 0.3359 -0.1362 H2 0.4765 0.4085 -0.0524 H3 0.5589 0.3669 0.0503 H4 0.5716 0.2529 0.0693 D7 0.3929 0.1942 -0.1494 D8 0.5069 0.1368 -0.0073 C1 0.4364 0.3015 -0.1139 C2 0.4680 0.3571 -0.0711 C3 0.5179 0.3479 0.0011 C4 0.5361 0.2830 0.0304 C5 0.5045 0.2274 -0.0124 C6 0.4546 0.2366 -0.0845 315 K C7 0.4213 0.1778 -0.1295 C8 0.5236 0.1588 0.0184 H1 0.4043 0.3074 -0.1603 H2 0.4563 0.3989 -0.0900 H3 0.5382 0.3837 0.0286 H4 . 0.5682 0.2771 0.0768 D7 0.4144 0.1658 -0.1389 D8 0.5277 0.1449 0.0250 8 The o-xylene coordinates were obtained from the average solutions calculated from the NMR data at 273 K and 315 K with I 2 > 0.92. The coordinates of the silicon atoms are from the low loaded form of p-dichlorobenzene/ZSM-5 complex42. The coordinates of D7 and D8 are the centroids of the triangles formed by the deuterium atoms in the (rotating) methyl groups. 104 (a) 105 Figure 4.16 Structure determined from NMR data of o-xylene-oVZSM-5 complex at 273 K showing the o-xylene with the framework silicon atoms labeled. The left figure shows the location of o-xylene viewed along the zigzag channel of ZSM-5 and the right shows the location viewed along the straight channel. For clarity, the oxygen atoms in the framework have been omitted. 4.3 CP Experiments on o-xylene-aVZSM-5. Figures 4.17 and 4.18 show the changes in the NMR peak intensities in the variable contact time C P experiments at 273 K and at 315 K respectively for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 4 per unit cell [k\ value by fitting). For 273 K, Si6 shows the highest k'ls value, and SM1 the lowest among the eight T-sites giving resolved, single-site signals. As previously mentioned, this is an indication of the differences in the distances between the silicon T-sites and the protons in the o-xylene molecules. In order to determine the relative C P constants of silicon atoms in ZSM-5 to the 29 protons in the o-xylene-d 4, the eight (273 K) and nine (315 K) peaks that belong to the single Si atoms were used. As the fitted k,s values and k, values indicate, at both 273 K and 315 K, the o-xylene-d 4 /ZSM-5 system is in the slow C P regime, as was the case for the o-xylene-de/ZSM-5 complex. 106 50.00 40.00 30.00 20.00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0.00 50.00 40.00 30.00 20.00 10.00 0.00 SiS k'ls = 7.06 s"1 Si2 k'ls = 8.52 s'1 Si11 /t',s = 5.72 s"1 / t Si9 k'is = 8.46 s'1 Si5 k'is — 6.86 s Si10 k',s =8.52 s'1 Si12 k'iS = 6.84 s"' Si6 k'is = 8.65 s"1 { 0 10 20 30 40 50 60 Figure 4.17 Variable contact time " S i C P M A S NMR experiment at 273 K for ZSM-5 loaded with ca. 4 molecules of o-xylene-ck per unit cell. The vertical axes represent the N M R signal intensity in arbitrary units and the horizontal ones the range of contact times in ms. The constant theoretical C P maximum was 751 (a.u.) and ki was 249 s" 1 ( 7 i p = 4.02 ms). The independently determined 7 i p of 1 H was 13.1 ms. , Si8 k'is = 2.76 s 1 20.00 -0.00 ^-H r-0.00 60.00 Si4 k'm = 1.66 s"1 Si10 k'ls = 3.43 s"' Si2 k',s = 3-31 s'1 Si11 * •„ = 2.05 s"1 Si3 k'!s = 3.30 s"1 Si12 10 20 30 40 0 10 20 30 40 0 10 20 30 40 Figure 4.18 Intensities of the 2 9 S i C P M A S N M R signals indicated as functions of the contact time at 315 K for Z S M - 5 loaded with ca. 4 molecules of o-xylene-oV per unit cell. The vertical axes are the N M R signal intensities in arbitrary units and the horizontal ones indicate the range of contact times (ms). The constant theoretical C P maximum k was 495 (a.u.) and /c/was 77.7 s" 1 ( 7 i p = 12.9 ms). The experimental 7 i p of 1 H was 12.4 ms. 107 4.3.1 Solving the structure of the o-xylene-aVZSM-S complex The same procedure described above was used to determine the location of the o-xylene-d4 molecules in the framework of ZSM-5 . Initially, all the physically possible locations were sampled exactly as for the o-xylene-d6 system. From initial calculations using large step sizes, the o-xylene molecule was again found to be at the channel intersection. For the final calculations, for 273 K, translations in x (step size of 0.010 A) between 0.42 and 0.55 in the fractional coordinate, y (step size of 0.010 A) between 0.24 and 0.36 in the fractional coordinate, and z (step size of 0.015 A) between -0.15 and 0 in the fractional coordinate along with the orientations with (j> (step size of 4 degrees) between 10 and 70 degrees, 6 (step size of 4 degrees) between 50 and 100 degrees, and y (step size of 4 degrees) between - 3 0 and 35 degrees, were tested. A total of ca. 7 million possible locations were tested, and 1051 acceptable solutions were found with r2 > 0.71. For 315 K, approximately the same ranges and step sizes were used. A total of ca. 24 million possible locations were tested, and 1374 acceptable solutions found with t2 > 0.71. Much lower r2 values were needed for the o-xylene-d 4/ZSM-5 complex to obtain reasonable numbers of solutions comparable to those for the d6 complex in order to make comparisons between the results for the two systems. The value of r2 > 0.71 selected yielded approximately 10 3 acceptable solutions. There are several possible reasons for the lower quality of fit for the d4 system: First, it is not possible to specify the exact locations of the protons in order to solve the structure. Hence, the three protons in the methyl groups were approximated as one 'pseudoproton' at the center of their equilateral triangle, which also reduced the calculation times. There is also fast rotation of the methyl groups about their C 3 axes so the dipolar interactions between the protons in the d4 system and the silicon atoms are much weaker than those of d6 system, leading to larger errors. Figure 4.19 shows the distributions of acceptable solutions found for the r2 values indicated in the captions. For 273 K, a total of 928 solutions were found that satisfy r2 > 0.72. At 315 K, a total of 1070 solutions were found that satisfy r2 > 0.72. The black arrows on the curves represent the average values of the different parameters and are summarized in Table 4.4. The average values of the structural parameters were used to calculate the second moments of all of the 2 9 S i T-sites. Figure 4.20 shows the degree of linear correlation between the k)s values of 108 the resolved single site resonances and the corresponding calculated M2 values (a) at 273 K and (b) at 315 K. The higher r2 value at 273 K (0.877) than at 315 K (0.774) indicates a better correlation for the 273 K data. Calculated values of k)s for the overlapping resonances were obtained from these linear correlation graphs using the average location of o-xylene, and used to calculate the complete NMR spectra that were then compared to the experimental NMR spectra at the same contact time. Figures 4.21(a) and (b) show that there is good agreement in both cases, and even for the overlapping peaks, the differences between the experimental and calculated spectra are marginal, appearing to come from errors in the deconvolutions where the resonances contributing to the overlapped peaks were assumed to have exactly the same chemical shifts. The scatter plots for the solutions found for the both temperatures show well localized solutions in Figure 4.22. One thing to notice is that the scatter plots are not as diffuse as those of the o-xylene-oyzSM-5, possibly indicating a better orientation for the guest molecule as the two methyl groups were probed in the o-xylene-d 4/ZSM-5. The average location of the o-xylene molecule in the framework was determined, giving the atomic coordinates listed in Table 4.5. Figures 4.23(a), (b) show the 3-dimensional structures of the o-xylene-oyzSM-5 complex at 273 K and 315 K respectively, which are almost identical to each other. The o-xylene is located at the channel intersection between the zigzag channel and straight channel of the ZSM-5 framework, quite similar to the previous structures of o-xylen'e-de/ZSM-5, although its long axis deviates slightly more from the b axis than in the previous determination. 109 Figure 4.19 Distributions of the solutions determined from 1 H/ 2 9 Si CP data of o-xylene-cfo in the framework of ZSM-5 (a) at 273 K and (b) at 315 K respectively with linear correlations of r2 > 0.71 (red squares), r2 > 0.72 (blue squares), and r2 > 0.73 (green squares). The vertical axes refer the numbers of solutions, and the horizontal show the distributions of solutions by six structural parameters; for the translation, x, y and z of the center of the o-xylene-ck (i.e. the center of the benzene ring) in fractional coordinates and for the orientation, <|>, 9 and of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with r2 > 0.72 for the both data, which are shown in Table 4.4. Table 4.4 Average values of the six structural parameters of o-xylene-oV in the framework of ZSM-5 at 273 K and 315 K with r2 > 0.72 from CP experiments. Temperature X y z . • ' e 273 K .. 0.486(1) 0.299(8) -0.048(20) 40.9(64) 72.0(51) 5.9(51) 315 K 0.491(6) 0.290(12) -0.042(9) 42.0(37) 87.5(125) 5.6(112) M2 (Hz2) M2 (Hz2) Figure 4.20 Plots of the measured CP rate constants against the calculated heteronuclear second moments for the average location of o-xylene for ZSM-5 loaded with ca. 4 molecules of o-xylene-ck per unit cell with r2 > 0.72 (a) at 273 K and (b) at 315 K. The solid lines are the lines of best fit and the dashed lines represent the 95% confidence prediction intervals. 110 -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 2 9 Si Chemical Shift (ppm from TMS) 2 9 Si Chemical Shift (ppm from TMS) Figure 4.21 NMR spectra for ZSM-5 loaded with ca. 4 molecules of o-xylene-04 per unit cell at 273 K (a) experimental 1 H / 2 9 S i C P M A S NMR spectrum (contact time = 10 ms) (b) the predicted spectrum from the average solution of the structure calculation (c) difference between the experimental and calculated spectra. - 4 - 3 - 2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4 (a) (b) Figure 4.22 Scatter plots of the acceptable o-xylene locations in the framework of Z S M - 5 at r 2 > 0.72 for ZSM-5 loaded with ca. 4 molecules of o-xylene-d 4 per unit cell (a) at 273 K and (b) at 315 K. The colored dots represent the atomic locations of the hydrogen (green), carbon (red) and deuterium (blue) in o-xylene-c/4. 111 Table 4.5 Fractional atomic coordinates for the o-xylene-d4/ZSM-5 complex at 273 K and 315 K from CP experiments3 Temp. atom x V z C 1 0.4344 0.3188 -0 .1128 C 2 0.4659 0.3665 -0 .0523 C 3 0.5170 0.3469 0.0129 1 C 4 0.5367 0.2796 0.0176 C 5 0.5052 0.2319 -0.0430 C 6 0.4540 0.2515 -0.1082 273 K C 7 0.4209 0.2010 -0.1720 C 8 0.5257 0.1608 -0 .0383 D1 0.4015 0.3315 -0.1548 D2 0.4533 0.4099 -0.0552 D 3 0.5373 0.3776 0.0519 D4 0.5696 0.2670 0.0595 H7 0.4139 0.1908 . -0 .1852 H8 0.5301 0.1462 -0.0371 C1 0.4494 0.3059 -0.1161 C 2 0.4788 0.3531 -0 .0523 C 3 0.5273 0.3329 0.0169 C 4 0.5465 0.2655 0.0223 C 5 0.5172 0.2183 -0.0414 C 6 0.4687 0.2385 -0.1106 315 K C 7 0.4378 0.1884 -0.1777 C 8 0.5373 0.1471 -0.0359 D1 0.4182 0.3188 -0 .1605 D2 0.4664 0.3965 -0.0558 D 3 0.5462 0.3634 0.0579 D4 0.5777 0.2525 0.0668 H7 • 0.4314 0.1783 -0.1916 H8 0.5416 0.1325 -0 .0345 * The o-xylene-di coordinates were obtained from the average solutions that were calculated from the N M R data at 273 K and 315 K with r2 > 0.72. The coordinates of H7 and H8 are the centroids of the triangles formed by the hydrogen atoms in the (rotating) methyl groups. 112 Figure 4.23 NMR determined structures of the o-xylene-oVZSM-5 complex at (a) 273 K and (b) 315 K from the C P data. The left figures show the location of o-xylene from the zigzag channel of Z S M - 5 and those on the right the location from the straight channel. For clarity, oxygen atoms in the framework have been omitted. 4.4 o-xylene-c/6/ZSM-5 CP drain experiments. As the C P studies of o-xylene/ZSM-5 indicated, the general behavior places the spin dynamics of the system in the "slow C P regime". In this situation, it is important, if possible, to confirm the results using an alternative technique to obtain absolute values for kiS. The variable contact time C P drain experiment is very appropriate for such a study although it has disadvantages because the relaxation delay of 2 9 S i can be considerably longer than that of 1 H , and the C P drain plots are differences between reference and drain spectra, significantly reducing the final signal-to-noise, since absolute /c,s values are obtained from a single parameter fit to the final data. Figures 4.24 and 4.25 show the C P drain plots for o-xylene-d6/ZSM-5 at 273 K and at 315 K respectively where each plot of the resolved 2 9 S i T-sites of ZSM-5 is considerably poorer than those from the traditional C P experiments. However, each depends on only a single parameter, and when the kts values from the C P drain plots are compared to those from the traditional C P experiment, both the C P and C P drain values are comparable to each other in magnitude (Tables 4.6 and 4.7 for 273 K and 315 K respectively). One should note that the C P curves can be deceptive in that they are in the slow exchange regime where the rises and maxima are mainly determined by /c, while the decays, which are slow and not fitted quite as well as the overall curve, have the largest contribution from /c,s. In fact, the kis values from the C P drain data, even with larger errors, yielded the structure of the complex. 113 One of advantages of C P drain experiment is that the value of ks, which is not determined in the traditional C P experiment, is simultaneously obtained from the reference part of the C P drain data. Since the reference part of C P drain follows Equation 3.4 for the signal decay, plotting the reference signal against contact time should yield ks. The values of ks from the C P drain data are shown in Table 4.8. Although they have relatively large errors, they have comparable values to the determined kis values that in turn are shorter than the kt values in the slow exchange regime. 0.40 0.30 0.20 0.10 0.00 -0.10 0.40 0.30 0.20 0.10 0.00 -0.10 0.40 0.30 0.20 0.10 0.00 -0.10 Si8 k,s =4.55 s'1 SM1 kts = 3.88 s"' • Si5 • kis =3.75 s"1 • Si10 * , s = 3.14 s' • * • • Figure 4.24 Intensities of the HI Si C P drain N M R signals indicated as functions of the drain contact time for ZSM-5 loaded with ca. 4 molecules of o-xylene-cfe per unit cell at 273 K. The points are the experimental values of the intensities and the solid curves are calculated according to Equation 3.6. The vertical axes represent the normalized signal difference (AS/So) in the C P drain data and the horizontal ones the range of drain contact times. For the curves corresponding to the groups of overlapped peaks, the fitted values of C P drain rate constant are the averaged values, denoted as kis*. 114 -0.10 - I — — , — , — , — . — , — I - I — — , — , — , — , — , — , — I - I — , — , — . , _ ,—I 0 20 40 60 80 0 20 40 60 8( 0 20 40 60 8C Figure 4.25 Intensities of the 1 H - 2 9 S i C P drain NMR signals indicated as functions of the contact time for ZSM-5 loaded with ca . 4 molecules of o-xylene-cfe per unit cell at 315 K. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.6. The vertical axes represent the normalized signal difference (AS/So) in the C P drain data and the horizontal ones the range of drain contact times. For the curves corresponding to the groups of overlapped peaks, the fitted values of C P drain rate constant are the averaged values as denoted as kts*. Table 4.6 Values of k'is from C P and kts from C P drain plots at 273 K. CP (s°) CP drain (s1) Si8 9.34(34) 4.55(153) Si2 7.35(13) 4.39(158) Si12 5.38(36) 2.51(155) Si11 4.99(31) 3.88(140) Si9 8.86(26) 4.36(167) Si6 5.94(13) 3.93(171) Si5 5.12(19) 3.75(230) Si10 4.85(6) 3.14(137) Si3,4 4.86(23) 2.68(117) Si1,7 7.01(23) 4.34(144) 115 Table 4.7 Values of k)s from C P and kis from C P drain plots at 315 K. CP (s 1) CP drain (s 1) Si8 4.95(20) 3.52(92) Si2 4.34(12) 3.62(108) Si3 4.19(8) 3.14(109) Si4 2.56(15) 0.722(133) Si11 3.38(5) 4.27(233) SM2 3.30(7) 2.35(107) Si10 3.24(10) 2.09(112) Si1,7 4.04(7) 4.12(94) Si5,6,9 4.82(21) 3.79(61) Table 4.8 Values of ks determined from C P drain experiments. Silicon T-sites in ZSM-5 ks at 273 K (s 1 ) at 315 K(s _ 1 ) Si1 1.83(57) 1.28(38) Si2 2.58(181) 1.94(85) Si3 2.11(88) 1.25(65) Si4 2.11(88) 2.99(209) Si5 4.74(134) 1.96(57) Si6 4.57(181) 1.96(57) Si7 1.83(60) 1.28(38) Si8 1.48(142) 1.21(82) Si9 2.78(178) 1.96(57) Si10 2.87(152) 1.99(127) Si11 2.30(156) 1.26(172) SM2 1.13(192) 1.88(77) The averages of the 2 9 Si 7 1 p are 430 ± 185 ms for 273 K and 594 ± 1 7 1 ms for 315 K. 4.4.1 Solv ing the structure of the o-xylene-oVZSM-5 complex. The same software used for the previous C P experiments was also used to solve the structure of o-xylene-d 6 molecules in the framework of ZSM-5 from the C P drain data. Initially, all the physically possible locations were sampled as previously described. For the final calculations, for the 273 K data, translations in x (step size of 0.010 A) between 0.42 and 0.55 in fractional coordinate, y (step size of 0.010 A) between 0.10 and 0.40 in fractional coordinate, and z (step size of 0.016 A) between - 0.12 and 0.020 in the fractional coordinate along with the orientations with (j> (step size of 3 degrees) between 30 and 70 degrees, 0 (step size of 5 degrees) between 25 and 150 degrees, and y (step size of 3 degrees) between - 30 and 40 degrees, were used. A total of ca. 35 million possible locations were tested, and 458 acceptable solutions were found with r2 > 0.71. For the 315 K data, a 116 similar sampling regime was used. In this case, a total of ca. 33 million possible locations were tested, and 1278 acceptable solutions were found with r2 > 0.81. Figure 4.26 shows the distributions of acceptable solutions found for the r2 values indicated in the captions. For 273 K, a total of 321 solutions were found that satisfy r2 > 0.72. At 315 K, a total of 917 solutions were found that satisfy r2 > 0.82. The black arrows on the curves represent the average values of the different parameters and are summarized in Table 4.9. Figure 4.27 shows the degree of linear correlation between kts and the corresponding calculated M2 values from the average locations of the o-xylene molecule at 273 K and at 315 K. The lower r2 value at 273 K (0.802) than at 315 K (0.909) indicates that the 273 K data may yield a less reliable structure. When the two correlations are compared qualitatively in terms of their average parameters, it would appear that the better distribution of single resonance silicon T-sites over the possible range of M2 values at 315 K leads to a better correlation of the data at 315 K than that at 273 K. From the linear correlation graphs, the calculated values of kiS were obtained for the overlapping resonances and used to calculate the complete NMR spectra. The calculated NMR spectra were then compared to the experimental spectra for one specific contact time. Figures 4.28 and 4.29 show the difference spectra between the experimental and predicted N M R spectra from the average of the acceptable structures at 273 and 315 K from the C P drain experiments. The experimental data have a high noise level as expected from the nature of the experiment. Figure 4.30 shows the scatter plots for the 273 and 315 K data, which are relatively well defined over the two-dimensional space. Using these parameters, the location and orientation of the o-xylene in the framework was determined and the atomic coordinates of the atoms in o-xylene molecule calculated as listed in Table 4.10. Figure 4.31 presents the structures of the complex at 273 K and 315 K which both again indicate the o-xylene is located at the channel intersection between the zigzag channel and straight channel of the ZSM-5 framework with its long axis approximately parallel to the b axis. 117 Figure 4.26 Distributions of the solutions determined from 1 H / 2 9 S i C P drain data of o-xylene-cfe in the framework of Z S M - 5 at (a) 273 K with linear correlations of r2 > 0.71 (red squares), r2 > 0.72 (blue circles), and r2 > 0.73 (green triangles) and at (b) 315 K with linear correlations of r > 0.81 (red squares), r2 > 0.82 (blue circles), and r2 > 0.83 (green triangles). The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of solutions of the six structural parameters. The arrows indicate the 'average' values of the six structural parameters of the structures at 273 K and 315 K with r2 > 0.72~and r 2 > 0.82 respectively (Table 4.9). Table 4.9 Average values of the six structural parameters of o-xylene-cfe in the framework of Z S M - 5 at 273 K and 315 K with r2 > 0.72 and f2 > 0.82 respectively from the C P drain experiments. Temperature x y z • e V|/ 273 K 0.485(7) 0.300(9) -0.045(10) 51.1(35) 89.0(73) 9.5(90) 315 K 0.491(6) 0.290(12) -0.042(9) 42.0(37) 87.5(125) 5.6(112) 118 0 50000 100000 150000 0 50000 100000 150000 M2 (Hz2) M2 (Hz 2) (a) (b) Figure 4.27 Plots of the measured C P rate constants, kis from the C P drain experiments against the calculated hetronuclear second moments for the average locations of o-xylene for Z S M - 5 loaded with ca. 4 molecules of o-xylene-de per unit cell with (a) r2 > 0.72 at 273 K and (b) r2 > 0.82 at 315 K. The solid lines are the lines of best fit and the dashed lines represent the 99% and 95% confidence prediction intervals. Each kis value comes with its error range represented by the error bars. The 99% confidence prediction interval for the data at 273 K was chosen to obtain a number of solutions comparable to that at 315 K. 1,7 -110 -112 -114 -116 -118 -120 Si Chemical Shift (ppm from TMS) Figure 4.28 NMR spectra for ZSM-5 loaded with ca. 4 molecules of o-xylene-cfe per unit cell at 273 K: (a) experimental difference spectrum between the C P drain "reference" (So) and "drain" (Sd) spectra (contact time = 80 ms), (b) the predicted spectrum from the average solution of the structure calculations with r2 > 0.72 and (c) difference between the experimental and calculated spectra. 119 5,6,9 —i 1 1 1 1 1 1 1 1 1 1 -110 -112 -114 -116 -118 -120 2 9 S i Chemical Shift (ppm from TMS) Figure 4.29 NMR spectra for ZSM-5 loaded with ca. 4 molecules of o-xylene-d6 per unit cell at 315 K: (a) experimental difference spectrum between the C P drain "reference" (So) and "drain" (Sd) spectra (contact time = 80 ms), (b) the predicted spectrum from the average solution of the structure calculations with r2 > 0.82 and (c) difference between the experimental and calculated spectra. -< -3 -2 -1 0 1 2 3 4 4 4 4 - 1 * 1 2 3 4 Figure 4.30 Scatter plots of o-xylene molecules in the plane of the molecule for Z S M - 5 loaded with ca. 4 molecules of o-xylene-cfe per unit cell at 273 K (r2 > 0.72) (left) and at 315 K (r2 > 0.82) (right) respectively from C P drain data. The colored dots represent the atomic locations of the hydrogen (green), carbon (red) and deuterium (blue) in o-xylene-cfe 120 Table 4.10 Fractional atomic coordinates for the o-xylene-cfe/ZSM-5 complex at 273 K and 315 K from CP drain experiments3 Temp. atom X V z C1 0.4418 0.3082 -0.1265 C2 0.4734 0.3640 -0.0842 C3 0.5166 0.3558 -0.0027 C4 0.5282 0.2918 0.0365 C5 0.4966 0.2360 -0.0058 C6 0.4534 0.2442 -0.0873 273 K C7 0.4202 0.1851 -0.1318 C8 0.5087 0.1682 0.0353 H1 0.4141 0.3135 -0.1789 H2 0.4659 0.4053 -0.1094 H3 0.5369 0.3918 0.0245 H4 0.5559 0.2865 0.0889 D7 0.4133 0.1731 -0.1412 D8 0.5114 0.1544 0.0440 C1 0.4381 0.3199 -0.1037 C2 0.4680 0.3606 -0.0312 C3 0.5069 0.3317 0.0445 C4 0.5159 0.2621 0.0477 C5 0.4860 0.2214 -0.0248 C6 0.4471 0.2503 -0.1005 315 K C7 0.4156 0.2071 -0.1769 C8 0.4953 0.1479 -0.0217 H1 0.4132 0.3385 -0.1524 H2 0.4623 0.4054 -0.0332 H3 0.5261 0.3580 0.0912 H4 0.5408 0.2435 0.0964 D7 0.4091 0.1985 -0.1927 D8 0.4973 0.1328 -0.0207 8 The o-xylene-de coordinates were obtained from the average solutions calculated from CP drain data at 273 K (r2 > 0.72) and at 315 K (r2 > 0.82). The coordinates of D7 and D8 are the centroids of the triangles formed by the deuterium atoms in the (rotating) deuterated methyl groups. 121 Figure 4.31 NMR determined structures of the o-xylene-GVZSM-5 complex (a) at 273 K and (b) 315 K from the C P drain data. The left figures show the location of the o-xylene from the zigzag channel of Z S M - 5 and the right the location from the straight channel. 4.4.2 Comments on the Structures Determined by NMR. From the NMR experiments, a total of six structures have been determined using two different samples (o-xylene-oyZSM-5 and o-xylene-oyzSM-5), at two temperatures (273 K and 315 K) and by two types of experiment (CP and C P drain). In general, all of the structures are in good agreement. In order to judge the reliability of each of them, one should consider many different factors such as the t2 values for the linear correlations between the kts and M2 values, the numbers of solutions, distributions of solutions and differences between the experimental and calculated NMR spectra. Overall, the most reliable structures seem to be those of d - xy lene^de /ZSM-5 from C P experiments at the two temperatures when the parameters listed above are compared to those of the other structures. However, one should note that NMR determines the positions of the protons of the o-xylene and thus, the position of the benzene ring of the o-xylene molecule in the case of o-xylene-d e / Z S M - 5 . The positions of the methyl groups of the o-xylene molecule may well be better reflected in the experiments on o-xylene-oyZSM-5 although these yielded less well defined positions due to the inherently weaker dipolar couplings involving the protons of the methyl groups. Therefore, the results on two different samples should be viewed as complementary to each other as they probe different parts of the o-xylene molecule. In the case of C P drain structures, the structure at 315 K seems more reliable than that at 273 K as it has better fit parameters and agrees with the C P determined 122 structures. However, the importance of the C P drain experiments extends beyond the determined structures as previously discussed, and one should view the results accordingly. Overall, all six structures are in good agreement for the position of the benzene ring in o-xylene, and the orientations of the long axes vary only very slightly. 4.5 Single Crystal XRD of o-xylene/ZSM-5. To date we have been unable to obtain useful single crystal X-ray data. Although numerous attempts were made to solve the structure of the complex, they were without success. The slow diffusivity of o-xylene makes it difficult to introduce it into the framework of ZSM-5 , particularly in the case of a large single crystal (over 100 microns). For many crystals, the indexing of the data indicated the presence of the monoclinic space group, characteristic of the empty of ZSM-5 framework 2 0. A higher loading temperature (140 °C) that should promote faster loading of an organic guest molecule into the framework was also used without success. Although highly siliceous ZSM-5 (Si/AI > 99) was used in the study, any traces of residual aluminum even on the outer surfaces could make ZSM-5 an active acid catalyst. Any temperature higher than 140 °C induces isomerization of xylenes in the presence of an activated H-ZSM-5, and this type of isomerization of the o-xylene was detected by the presence of a triclinic space group of ZSM-5 during indexing of the single crystal data. High-resolution 1 H NMR on CDCI3 extracts of the powder also showed the presence of all the xylene isomers. It may be possible to determine the single crystal structure of o-xylene/ZSM-5 by using a smaller crystal together with synchrotron X-ray diffraction, which should give a stronger scattering pattern. An alternative technique would be powder neutron diffraction, which, unlike X-ray diffraction, gives strong scattering from light atoms such as protons and deuterons as well as heavy atoms such as silicon and oxygen atoms in the framework, and this would have an increased scattering contribution from the o-xylene. 123 4.6 Summary All the experiments that yielded reasonable numbers of solutions at high r2 values give consistent results for the structure of the o-xylene/ZSM-5 complex. At two different temperatures over a 40 °C range of temperature, these structures of o-xylene/ZSM-5 turned out to be very similar. The o-xylene molecule is located at the intersection between the straight and zigzag channels of the Z S M -5 framework from all data sets. The arbitrary long axis of the o-xylene is almost parallel with the b axis for all data with only slight deviations. Single crystal X R D studies of the system failed to yield any solution due to an inability to load the o-xylene uniformly in the large crystal needed; a powder neutron diffraction study on the same organic/zeolite system is described in the next chapter, and the results confirm the structure presented here. 124 Chapter 5 § § Powder Neutron Diffraction Study of the o-xylene/ZSM-5 System 5.1 Introduction In Chapter 4, the use of the solid-state NMR method to determine the structure of the o-xylene/MFI complex was described in detail. It was originally hoped that a single crystal XRD study could confirm the structure; however, as discussed, this was not possible due to the" inability to prepare large crystals with a uniform loading of the guest molecule. To date, the only structural studies available on the o-xylene/ZSM-5 complex are by solid state NMR of this study 1 0 7 and powder X R D 5 1 , 5 2 . In recent years, powder diffraction techniques have greatly improved as tools to investigate the structures of microcrystalline materials containing the guest molecules 5 0 " 5 3 , 1 1 5 . Although the techniques do not give as definitive answers as single crystal X R D as the powder data are much more limited, they have been shown to be capable of solving some structures where single crystal XRD could not be attempted due to lack of high quality large single crystals. Powder neutron diffraction has been shown to be especially an effective technique for structural studies on guest/zeolite complexes, as it is sensitive to the organic guests and not affected by atomic scattering factors of light atoms unlike any X-ray diffraction technique. Also using deuterated samples would enhance the quality of the data significantly as explained in Chapter 1. In the present chapter, the determination of the' o-xylene/MFI structure by powder neutron diffraction using perdeuterated o-xylene is described along with the structural refinement by the Rietveld method. § § A version of this chapter has been submitted for publication. C A Fyfe, J S J Lee, I P Swainson and L M D Cranswick. Powder Neutron Diffraction Study of o-Xylene/ZSM-5 Complex. Microporous and Mesoporous Materials 2007 submitted. 125 5.2 Structure determination of o-xylene-c710/ZSM-5 Purely siliceous (Si/AI > 99) microcrystalline ZSM-5 (ca. 1 g), which was pre-checked by both powder X R D and solid-state NMR for its quality, was saturated with o-xylene-d 1 0 (Sigma-Aldrich) and was left to equilibrate in an oven at ca. 80 °C for several days. The mixture was kept below 80 °C at all times to avoid isomerization, which yields almost exclusively p-xylene (this isomerization happens at a reasonably high temperature of ca. 140 °C for the o-xylene/ZSM-5 complex). The loading temperature of 80 °C was used for the previous NMR study 1 0 7 on this system, and the presence of the o-xylene in the framework of ZSM-5 was confirmed by solvent extraction and high resolution 1 H NMR using non-deuterated o-xylene. After preliminary checks using powder X R D and solid-state NMR for the presence of the guest organic in the framework, the o-xylene/ZSM-5 complex was sent to the Canadian Neutron Beam Centre (CNBC) at Chalk River, Canada, where its powder neutron diffraction patterns were collected. Two of the data sets obtained, which were collected at 272 K using neutron wavelengths of X = 1.33 and 2.37 A, are presented in Figure 5.1. The indexing of several major peaks from the 2.37 A data, which shows better detail of the low angle intensities, was done using the Crysfire program 1 0 6 . The initial indexing on the neutron data suggested two possible orthorhombic space groups, Pnma and P212121. Both space groups are possible candidates for the guest/zeolite complex, as the p-xylene/ZSM-5 and p-dichlorobenzene/ZSM-5 complexes have been known to adopt b o t h 4 0 , 4 2 , 4 3 , 1 3 6 . However, the space group P2i2i2i, which lacks the centers of symmetry in the straight channels at (0, Vi, 1/2) and the mirror planes (y = V* and %), and thus has 24 silicon sites, was ruled out since the 29 preliminary Si NMR spectrum on the powder neutron sample clearly indicated 12 silicon T-sites. In the first step of a successful structure elucidation from the powder data, P O X 1 0 8 , 1 0 9 , a program for determining ab initio structure by direct space approach, was used to obtain the starting location for the o-xylene molecule in ZSM-5 with the framework structure held rigid. For this, the 1.33 A data set was used as it contains more high angle intensities, and thus has finer structural details. The framework coordinates used were those of the low loaded form of p-dichlorobenzene/ZSM-5 4 2 as it represents the orthorhombic structure of the ZSM-5 framework (the space group Pnma) when the guest p-dichlorobenzene is loaded. It is important that the framework structure be precise at the very 126 beginning (even if incorrect in detail) for the software to work properly. Having a well defined starting point for locating the guest was found to be crucial: without the starting location, previous refinement attempts of the data had yielded a distorted structure of ZSM-5 with no organic present. The initial starting point was a rigid body of o-xylene at {0, 0, 0} in fractional coordinates of the ZSM-5 framework. The molecule was then moved randomly as the global optimization was performed, and the quality of the structure was gauged from the reduction of the parameters, ~i, Rwp and Rp. At some point, a second rigid o-xylene was introduced to account for another site or disorder but none was detected. At the end of this stage of the structure elucidation, both the peak fitting parameters, Rwp and Rp reduced to ca. 0.1, and the spatial coordinates and atomic displacement parameters obtained were taken as the starting point for the final refinement. The refinement of the data was carried out by the Rietveld method 9 0 , 9 1 ; the initial parameters obtained by FOX including the atomic coordinates for the framework of ZSM-5 and the o-xylene molecule and their isotropic temperature factors were exported to the G S / A S 1 1 0 , 1 1 1 . The scale factor, lattice parameters such as the unit cell dimensions, instrumental zero points, peak profile parameters, neutron wavelengths and atomic isotropic temperature factors were allowed to vary along with the atomic coordinates for all the atoms in the o-xylene/ZSM-5 complex. In the G S A S program, there are several options to implement gradual refinement iteration: most common ways involving using restraints and constraints, which are adjusted by different weight functions for their impacts on the refinement. Applied restraints were: bond length restraints for S i - 0 (1.57 ± 0.10 A), C - C (1.4 ± 0.10 A) and C-D (1.1 ± 0.10 A), and the bond angle restraints for O-S i -0 (109.5 ±0 .10 °). Restraints were applied to the framework and o-xylene so that they retained certain degrees of structural rigidity during the refinement. In the beginning of the refinement, the restraints were applied heavily so that the starting model of the structure was as rigid as possible. As the refinement progressed, the weights on the restraints were gradually reduced. Applied constraints were the atomic displacement factor for each type of atom, and the site occupancy factor for the framework and the o-xylene molecule so that each type of atom could be refined all together and the site occupancy of the o-xylene determined with respect to the framework. Until the end of the refinement, however, the restraints and constraints were kept as the refinement became unstable 127 upon removing them. At the end of the final refinement of the data sets, the agreement between the observed and . predicted profiles is very good for a complicated structure like the guest/zeolite complex as Rp = 0.03, Rwp = 0.04 for 1.33 A and Rp = 0.04, Rwp = 0.05 for 2.37 A. Figure 5.1 shows the observed, calculated and difference patterns of the powder neutron data at different wavelengths (1.33 A and 2.37 A) for the o-xylene/ZSM-5 complex at 272 K. Upon visual inspection, the two observed neutron diffraction patterns show excellent agreement with the respective calculated patterns. Although its difference profile looks noisier, the peak profile of the 1.33 A wavelength data, which yielded better Rp and Rwp values, is more reliable for refining the structure because it contains many more intensities at higer 29 angles from the smaller d spacing diffractions than the 2.37 A data. For both wavelengths, no large background profiles were present, which is crucial for a reliable Rietveld refinement as mentioned in Chapter 3. From the final refinement, all atomic coordinates for o-xylene-d 1 0 and ZSM-5 were determined along the atomic displacement parameters (Uiso) as well as their occupancy parameters, which are presented in Table 1. The final refinement results showed that the o-xylene molecule was at the intersection of the straight and zigzag channels of ZSM-5 with an occupancy factor corresponding to approximately 3.7 molecules per unit cell of ZSM-5 , in good agreement with the T G A desorption data. The refined framework of ZSM-5 was similar to those of single crystal s tud ies 4 1 , 4 2 with S i - 0 bonds ranging from 1.53(2) to 1.66(2), and the O-S i -0 bond angles ranging from 109.1(5) to 110.2(6). The refined isotropic atomic displacement factors for Si and O were 1.2(2) and 1.5(1) respectively. Figure 5.2 shows the difference Fourier map of the neutron scattering density of the o-xylene molecule. The difference Fourier map (a three-dimensional data set) is noisy when it is obtained from powder data, as these are one-dimentional sets. However, it is clear that the majority of the density is in the straight channel. There are several reasons for the scattering density appearing diffuse. Most importantly, the symmetry of the Pnma space group means that there will be a statistical disorder of the o-xylene with respect to the mirror plane at y = %, which bisects the channel intersection. When two symmetry 128 equivalent sites are far apart from each other, the structural view and the Fourier map can be quite clearly resolved. However, when they are close to each other, distinguishing them clearly can be rather challenging. Hence, the residual scattering is due to two o-xylene molecules, each with occupancy one half, as has been seen for other organic/ZSM-5 systems.in the Pnma space group. This is illustrated using the final results from Table 5.1 in Figure 5.3, where a second o-xylene is generated by the symmetry operation and presented with the original molecule. Depending on how easy motion would be between the two orientations, this could be a reason for the almost rigid o-xylene from the previous static 1 3 C NMR study 1 4 0 . For any diffraction technique, the data are not capable of detecting any motion directly, but only statistical disorder of the system. In general, Fourier density maps calculated from powder data are more diffuse than those from single crystal da ta 1 4 4 . In addition, the resolution of neutron data is poorer than that obtained with X-rays due to much lower intensities of the neutron beams. For most elements, the absorption coefficients of neutrons are approximately four orders of magnitude lower than those of X-rays. For these reasons, neutron Fourier maps would appear diffuse, and this becomes much more significant in the presence of motions such as the fast methyl rotation about the C 3 axis in the o-xylene at 272 K. In the methyl group rotations, there are limitless possible locations for the methyl deuteriums, contributing to the weaker scattering of these neutrons. For most atoms in the o-xylene/ZSM-5 complex the multiplicities are 8, meaning that there are 8 symmetry equivalent sites in the unit cell due to the space group Pnma. Exceptions are the oxygen atomic sites,. 0 2 3 , 24, 25 and 26, and the deuterium D71, which were refined with multiplicities of 4 indicating those atoms lie on the mirror plane at y = Vi. This indicates that in the same channel intersection, two of the symmetry equivalent o-xylenes share the deuterium D71 site, and the completed structure shows that they are very close to each other. The final refined structure in Figure 5.4 shows that o-xylene is at the channel intersection between the straight and zigzag channels of the framework of ZSM-5 with the center of the benzene ring at {0.499, 0.283, -0.035} in fractional coordinates. According to the NMR of the o-xylene-cfe/ZSM-5 complex by C P at 273 K described in the previous chapter, the ring center was at {0.480, 0.265, -0.044}, in very good agreement with the neutron study. In terms of determining the location of the 129 deuterium atoms in o-xylene, the advantage of using o-xylene-d 1 0 and neutrons becomes obvious as all the deuterium atoms are accounted for after refinement. For both NMR and powder X R D studies, the locations of the hydrogen atoms in the organic guest were not determined completely. In the NMR study, the three hydrogen atoms of the methyl groups in o-xylene were replaced by one 'pseudo-proton' at the center of their equilateral triangle in order to simplify the calculations. This approximation works for the heteronuclear second moment calculation, but the exact locations of the protons cannot be determined by this method. In the powder X R D study, the hydrogens of o-xylene could not be located conclusively by X-ray diffraction, as is common for light atoms such as hydrogen because of their weak atomic scattering strengths. Thus, neither of two techniques is capable of producing the experimental results regarding the true positions of the entire hydrogen atoms of the o-xylene. However, for neutron diffraction, a different situation exists; since the scattering strength of deuterium is very similar to carbon, it would be difficult to separate their contributions to the neutron scattering density if two are very close to each other, which could be a further reason for the noisy Fourier map in Figure 5.2 in addition to those discussed earlier. Figure 5.5 shows the maximum and minimum oxygen-oxygen distances of the elliptical 10 membered ring that contains the o-xylene, which are 8.86 A and 7.45 A respectively. The maximum distance of the ellipsoid is larger than those of the empty ZSM-5 f rameworks 1 9 , 2 0 , indicating that the presence of o-xylene distorts the framework of ZSM-5 considerably. From the previous study 6 8 by A. R. Lewis, it was found that the maximum and minimum oxygen-oxygen distances from the low loaded form of p-xylene/ZSM-5 were 9.115 A and 7.319 A respectively, indicating the p-xylene molecule has more profound effect on the framework. Figure 5.6 shows the final structure of o-xylene-d 1 0 /ZSM-5 at 272 K from the powder neutron diffraction (blue) along with the structure (red) of the o-xylene-de/ZSM-5 complex at 273 K determined by NMR, showing them to be in excellent agreement and confirming the validity of the NMR structure determination. 130 Counts 5000 H ( a ) + obs — calc — bck — diff in iintiiiiiiiiiiiiiii I 1 1 1 1 50 100 26(deg) Counts 100004 5000 ( b ) + obs — calc — bck — diff i II i MI i mi in IIimn i i i i i i i i i i i i u u i i i i i i i i B i i i i i i i i i u i u B i i i u i m r a i B M i a m i i i i i i • l U B I I H I U B I U I I " 50 100 29(deg) Figure 5.1 Observed (cross points), calculated (red line), background (green line), and difference (blue line) powder neutron scattering profiles for the sample containing 3.7 molecules of o-xylene per unit cell of ZSM-5 at 272 K: neutron wavelengths (a) at 1.33 A, used for the structure solution; (b) at 2.37 A, used for the preliminary indexing. 131 Table 5.1 Refined Parameters for o-Xylene/ZSM-5 Complex at 272 K (Space Group Pnma). Atom x/a(A) y/b (A) z/c(A) l// s o(x100A 2) Occupancy Si1 0.4253(9) 0.0546(12) -0.3530(15) 1.2(2) 1.000(0) Si2 0.3067(8) 0.0282(8) -0.2049(11) 1.2(2) 1.000(0) Si3 0.2822(13) 0.0631(15) 0.0157(19) 1.2(2) 1.000(0) Si4 0.1232(13) 0.0643(16) 0.0128(19) 1.2(2) 1.000(0) Si5 0.0736(8) 0.0229(9) -0.1915(13) 1.2(2) 1.000(0) Si6 0.1789(8) 0.0559(8) -0.3465(12) 1.2(2) 1.000(0) Si7 0.4241(10) -0.1732(10) -0.3376(18) 1.2(2) 1.000(0) Si8 0.3057(9) -0.1265(7) -0.1923(10) 1.2(2) 1.000(0) Si9 0.2719(12) -0.1705(8) 0.0279(15) 1.2(2) 1.000(0) Si10 0.1207(12) -0.1740(8) 0.0198(16) 1.2(2) 1.000(0) Si11 0.0699(8) -0.1332(8) -0.1942(13) 1.2(2) 1.000(0) Si12 0.1860(10) -0.1720(7) -0.3320(12) 1.2(2) 1.000(0) 01 0.3712(12) 0.0567(14) -0.2641(17) 1.5(1) 1.000(0) 02 0.3008(13) 0.0642(13) -0.0977(18) 1.5(1) 1.000(0) 03 0.2019(14) 0.0620(18) 0.0269(16) 1.5(1) 1.000(0) 04 0.1057(12) 0.0559(14) -0.0996(19) 1.5(1) 1.000(0) 05 0.1103(11) 0.0484(12) -0.2896(18) 1.5(1) 1.000(0) 06 0.2389(11) 0.0431(12) -0.2704(18) 1.5(1) 1.000(0) 07 0.3682(12) -0.1593(13) -0.2500(17) 1.5(1) 1.000(0)' 08 0.3021(14) -0.1548(11) -0.0801(15) 1.5(1) 1.000(0) 09 0.1958(14) -0.1478(12) 0.0312(17) 1.5(1) 1.000(0) 010 0.0988(13) -0.1704(12) -0.0963(19) 1.5(1) 1.000(0) 011 0.1084(12) -0.1599(12) -0.2905(19) 1.5(1) 1.000(0) 012 0.2381(11) -0.1449(12) -0.2503(16) 1.5(1) 1.000(0) 013 0.3147(12) -0.0481(11) -0.1895(16) 1.5(1) 1.000(0) 014 0.0808(12) -0.0550(11) -0.1830(16) 1.5(1) 1.000(0) 015 0.4254(15) 0.1246(14) -0.4113(21) 1.5(1) 1.000(0) 016 0.4072(17) -0.0026(14) -0.4269(21) 1.5(1) 1.000(0) 017 0.4039(17) -0.1333(14) -0.4396(23) 1.5(1) 1.000(0) 018 0.1843(13) 0.1312(13) -0.3926(18) 1.5(1) 1.000(0) 019 0.1829(14) 0.0015(12) -0.4365(19) 1.5(1) 1.000(0) • 020 0.1969(15) -0.1318(13) -0.4366(18) 1.5(1) 1.000(0) 021 -0.0035(11) 0.0424(15) -0.1966(20) 1.5(1) 1.000(0) 022 -0.0067(12) -0.1481(14) -0.2054(20) 1.5(1) 1.000(0) 023 0.4278(22) -0.2500(0) -0.3593(27) 1.5(1) 1.000(0) 024 0.1983(17) -0.2500(0) -0.3499(24) 1.5(1) 1.000(0) 025 0.2772(18) -0.2500(0) 0.0507(28) 1.5(1) 1.000(0) 026 0.1162(19) -0.2500(0) 0.0579(25) 1.5(1) 1.000(0) C1 0.4537(9) 0.2513(10) -0.0952(13) 14(2) 0.468(9) C2 0.4879(9) 0.2165(7) -0.0201(13) 14(2) 0.468(9) C3 0.5338(10) 0.2503(4) 0.0412(14) 14(2) 0.468(9)' C4 0.5448(10) 0.3191(6) 0.0276(16) 14(2) 0.468(9) C5 0.5111(12) 0.3537(7) -0.0482(18) 14(2) 0.468(9) C6 0.4653(12) 0.3199(11) -0.1094(16) 14(2) 0.468(9) C7 0.4063(15) 0.2161(12) -0.1594(23) 14(2) 0.468(9) C8 • 0.4755(11) 0.1452(18) -0.0043(17) 14(2) 0.468(9) D3 0.5601(11) 0.2232(8) 0.1011(16) 16(2) 0.468(9) D4 0.5812(11) 0.3456(9) 0.0750(19) 16(2) 0.468(9) D5 0.5207(15) 0.4074(8) -0.0599(22) 16(2) 0.468(9) D6 0.4380(40) 0.3473(18) -0.1680(6) 16(2) 0.468(9) D71 0.3649(24) 0.2499(0) -0.1790(6) 16(2) 0.468(9) D72 0.4319(23) 0.2000(50) -0.2280(40) 16(2) 0.468(9) D73 0.3860(40) 0.1721(15) -0.1200(40) 16(2) 0.468(9) D81 0.4960(40) 0.1301(21) 0.0685(34) 16(2) 0.468(9) D82 0.4213(12) 0.1357(22) -0.0050(70) 16(2) 0.468(9) D83 0.5000(40) 0.1160(19) -0.0640(40) 16(2) 0.468(9) Cell constants a, 6, c (A): 19.973(3), 19.944(3), 13.368(2) Calculated density (g cm"3): 1.933 R-factors for 1.33 A: R„p = 0.0414, Rp = 0.0299; for 2.37 A: Rwp = 0.0500, Rp = 0.0365 Figure 5.2 Difference Fourier map showing the silicon and oxygen atoms of the framework and the neutron scattering density due to the o-xylene-cdo molecule at the channel intersection of the framework of ZSM-5 : (a) view from the zigzag channel and (b) view down the straight channel, and the residual density is the sum of the two symmetry related sites of the o-xylene molecule. (a) (b) Figure 5.3 Powder neutron diffraction determined structure of o-xylene-dio/ZSM-5 complex at 272 K: (a) view from the zigzag channel, (b) view down the straight channel of the framework of Z S M - 5 . Two o-xylenes are due to the statistical disorder of the Pnma space group, which has a mirror plane at y = Vi. For clarity the oxygen atoms in the framework are omitted, and the size of the silicon atoms is minimized. 133 (a) (b) Figure 5.4 Powder neutron diffraction determined structure of o-xylene-di 0 /ZSM-5 complex at 272 K: (a) view from the zigzag channel, (b) view down the straight channel of the framework of Z S M - 5 . For clarity the oxygen atoms in the framework are omitted, and the size of the silicon atoms is minimized. Figure 5.5 The maximum and minimum oxygen-to-oxygen distances of the 10 member elliptical ring on the straight channel view of the o-xylene-dio/ZSM-5 complex. The maximum distance is between 01 and 0 7 , and the minimum one is between 0 5 and 011. 134 (a) (b) Figure 5.6 Powder neutron diffraction determined structure of o-xylene-dio/ZSM-5 complex (blue) at 272 K together with the N M R determined structure (red) of o-xylene-Gfe in the framework of Z S M - 5 at 273 K: (a) view from the zigzag channel, (b) view down the straight channel of the framework of Z S M - 5 . For clarity the oxygen atoms in the framework are omitted, and the size of the silicon atoms is minimized. 5.3 Summary The o-xy lene-oVZSM-5 structure has been determined by powder neutron diffraction. At a loading of 3.7 molecules per unit cell of ZSM-5 , the o-xylene molecule is localized at the channel intersection with one methyl pointing approximately along the straight channel axis. The structure by powder neutron diffraction is in excellent agreement with that previously determined by NMR. 135 Chapter 6 Investigation of the Structures of p-Dicyanobenzene/ZSM-5 and p-Dinitrobenzene/ZSM-5 by NMR and Single Crystal XRD 6.1 Introduction The single crystal X R D structure of the p-dicyanobenzene/ZSM-5 (DCNB/ZSM-5) complex was unknown before the present work. This organic guest molecule, which has two para substituted cyano groups (Figure 6.1), could exhibit interesting behavior on interaction with ZSM-5 due to its length and the presence of the electronegative substituent groups. Although it is not involved in any industrial applications, it would add valuable information towards an understanding of the general interactions between zeolites and different guest molecules of relevance to the use of zeolites as host frameworks in electronic, optical and other applications. The second structural study in this chapter, the investigation of the low loaded form of p-dinitrobenzene (Figure 6.1) in ZSM-5 is a continuation of 103 a previous study of the complex with ca. 4 molecules per unit cell of p-dinitrobenzene (DNB). This previous study found that the guest DNB was at the channel intersection both by NMR and single crystal X R D ; however, the single crystal X R D structure revealed a second site for the DNB, which was ca. half populated compared to the first site. The present interest was to characterize the behavior at lower loadings. Both studies were done by the same NMR method used in Chapter 4, and also by single crystal X R D . N III c c III N Figure 6.1 Structures of (left) p-dicyanobenzene (DCNB) and (right) p-dinitrobenzene (DNB) 136 6.2 Study of 4 molecules per unit cell of DCNB in ZSM-5 D C N B is considerably longer (ca. 8 A between the terminal N's) than any other aromatics we have previously investigated. Also, the cyano (-C=N) group is electronegative, and is thus expected to show mutual repulsion and repulsion with the framework oxygens when they are in close proximity. In order to investigate its interaction with the framework of ZSM-5 , a sample of ZSM-5 saturated with DCNB (4DCNB/ZSM-5 hereafter) was prepared and a T G A desorption study showed ca. 4.2 molecules of DCNB per unit cell. There was no prior knowledge of the structure except that the 2 9 S i . MAS spectrum indicated the space group was orthorhombic, with 12 independent T-sites, presumably space group Pnma. This information was used in the NMR structure determination. 6.2.1 So l i d - s ta te N M R s t u d y o f 4 D C N B / Z S M - 5 6.2.1.1 Variable Temperature 2 9 S i M A S NMR Spectra 29 A series of variable temperature Si MAS NMR experiments were conducted and the results are presented in Figure 6.2. The range of temperatures was from 250 to 350 K, and over this temperature range, the 4DCNB/ZSM-5 complex retained the space group Pnma, showing 12 2 9 S i peaks in its NMR spectra although the spectra changed somewhat over this temperature range. Constancy in the space group over the wide temperature range was taken as an indication of the structural integrity of the 4DCNB/ZSM-5 complex even as high as 350 K. After several above-room temperature experiments, 305 K was chosen for the NMR experiments on the 4DCNB/ZSM-5 complex as it had the largest number of single resolved peaks (8). 137 .108 -110 -112 -114 .116 -118 -120 -122 ' ' •- " . ~> ' ' ' - • •106 -110 -112 -114 -116 -118 -120 -122 Si Chemical Shift (ppm from TMS) M S i Chemical Shift (ppm from TMS) Figure 6.2 2 9 S i M A S N M R spectra of 4 D C N B / Z S M - 5 at the different temperatures indicated. The numbers above the selected resonances indicate their assignment to specific T-sites in the zeolite framework. The 2 9 S i 90° pulse length was 9 us and 3 2 - 5 1 2 scans were accumulated for each spectrum with a recycle delay of 5 - 6 s. 6.2.1.2 Peak Assignments by 2 9 Si INADEQUATE experiment Among the 12 possible peaks, assuming the space group Pnma for the 4DCNB/ZSM-5 complex, 10 peaks were resolved. After a quantitative experiment, two peaks (at ca. -113 ppm and at ca. -114 ppm) were found to be doubly overlapped peaks after deconvolution of the spectra (Figure 6.3). From a 2 9 S i INADEQUATE experiment, we assigned the peaks to individual Si T-sites as shown in Figure 6.4. A total of 18 connections were observed out of the 22 possible connections. Although two overlapped peaks were present, the assignment was made with the help of the peak assignment program previously descr ibed 1 1 7 . Among the 18 connections, three could be interpreted as double connections which could come from either of two connectivities. They are the connectivities of Si7-8 and SH2-8, Si7-11 and Si12-11, and Si4-1 and Si4-5. Two doubly overlapped peaks were assigned to be Si T-sites 7, 12 for the peaks at -113 ppm, and Si T-sites 1, 5 for the peaks at -114 ppm. There are two possible assignment sets, and one of the two can be ruled out 138 from its inconsistency with the correlation between Si chemical shifts and the mean Si-Si distances from the subsequent single crystal X R D work. -110 -112 -114 -116 -118 -1 : p p m Figure 6.3 (Top) ^ S i M A S N M R spectrum of 4 D C N B / Z S M - 5 at 305 K, 90 ° pulse of 7.5 us, recycle delay 35 s with 32 scans. The numbers above the resonances indicate the assignments of the peak to specific T-sites in the zeolite framework, found by a 2 9 S i INADEQUATE experiment. The bottom spectrum shows the individual deconvoluted resonances. .7,12 1,5 -110 -112 H14 H i 6 H i 8 Z 9 S i C h e m i c a l Shi f t (ppm from T M S ) Figure 6.4 Two-dimensional Si INADEQUATE spectrum of 4 D C N B / Z S M - 5 at 305 K together with the quantitative 2 9 S i N M R spectrum (recycle delay of 35 s). 32 experiments, each with 480 scans and 9 s recycle delay, were acquired in the U dimension. The echo delay during the double quantum preparation period was 12.5 ms and the sweep widths in the h and U dimensions were 800 and 1600 Hz. The indicated peak assignments were determined from the observed correlations using the program mentioned in the text. 139 6.2.1.3 CP experiments As described in Chapter 4, a 1 H / 2 9Si C P NMR experiment on the 10 resolved peaks was performed at 16 different contact times over the range of 0 to 70 ms giving the variable contact time C P plots are shown in Figure 6.5. From the C P plots, silicon T-site 8 in the framework, which has the largest k)s, can be assumed to be closest to the protons in the organic guest molecule. For the same reason, silicon T-sites 4 and 10, which have the smallest constants are the farthest from the protons. These trends are similar to the results of C P studies of the o-xylene-oyzSM-5 complex and indicating that the organic guest molecule is probably at the channel intersection. The interpretation assumed from the low efficiency and general shapes of the curves that the system was in the slow exchange regime, as described in previous chapters. 0-00 0.06 0.04 0.02 0.00 0.06 Si8 Si2 Si11 k'is = 9.32 s"1 k',s = 7.95 s"1 k',s = 6.59 s ' • r / — / • r / Si10 Si9 A', s = 4.92 s 1 k;s = 7.84 s' 1 .•''* • — f ' . * * * — . . • " • . • • :' si(i,5) / Si(7,12) * A-, s*=6.S5s- 1 f k'is' =8.19s"' Si3 Si4 Si6 k'is =7.17s"' k',s= 4.68 s"' . k'iS = 6.05 s'1 f r V Figure 6.5 Intensities of the 1 H / 2 9 S i C P M A S N M R signals indicated as functions of the contact time for 4 D C N B / Z S M - 5 at 305 K. The intensities are normalized to those of a quantitative 2 9 S i spectrum with an equal number of scans. The points are the experimental values of the intensities at chosen contact times and the solid curves are fitted according to Equation 3.3. The fitted theoretical C P maximum was 0.504 (a.u.) and ki was 118 s"1 (7<p = 8.47 ms). 0 10 20 30 40 50 60 70 10 20 30 40 50 60 70 140 6.2.1.4 Solving the structure of the p-dicyanobenzene/ZSM-5 complex. The details of solving the structure of a guest/zeolite complex by the NMR method were described in Chapters 3 and 4. In the case of D C N B , the imaginary line that passes through the two terminal nitrogen atoms was defined as the 'long axis'. The cutoff range for the heteronuclear second moment calculation was 8 A as previously. After starting with broad ranges in the channels of ZSM-5 , for the final calculation of the location of DCNB in the framework of ZSM-5 , the calculated ranges were, 0.45 < x < 0.55, 0.15 < y <, 0.25 and -0.075 < z S 0.075 in translations, and 20 < <)> < 60, 50 < 9 < 130 and - 4 0 < y < 40 in orientations. A total of ca. 5.4 million possible locations were tested, and 1,733 solutions were found for r2 > 0.92. The number of solutions after each step of the structure determination procedure is given in Table 6.1 along with the calculated ranges. As described in Chapter 4, only single peaks were used for structure determination and the two overlapped peaks subsequently added for a comparison of the whole profile of the NMR spectrum to the calculated ones. The average structural parameters are listed in Table 6.2 for r2 > 0.93. As a general indication of the validity of the solutions, the linear correlation between the measured C P rate constants and the calculated heteronuclear second moments from the average structure shows excellent linearity (Figure 6.6). Figure 6.7 shows the experimental and predicted NMR profiles and the difference between the two using the average structure. The difference between the two profiles was small in general, indicating that the average structure predicts the experimental NMR peak intensities relatively well; however, some peaks show measurable discrepancies with the experimental signals. The distributions of the six structural parameters, which are presented in Table 6.2, show that the location of the ring center is fairly well defined; however, the orientation of the long axis, which is expressed by the three Euler angles {(j>, 9, VJ;}, is very broad, especially for the angle 0, the rotation of the long axis of the D C N B from the z-axis, whose range is ca. 75 degrees. As another way of indicating the distribution of the solutions, a 2D plot of the scattered solutions in the plane of the molecule is presented along with the 50% error ellipsoidal representations of the atoms in D C N B 2 5 , 1 0 3 in Figure 6.9. The scatter plot also shows a wide distribution of solutions, which could possibly 141 suggest a disordered structure for the complex. The structure of 4DCNB/ZSM-5 determined from the C P data is presented in Figure 6.10 in the views from the zigzag and straight channels of ZSM-5 . According to the structure, the DCNB molecule is at the channel intersection of the framework with its ring center slightly off from the mirror plane at {0, V*, 0}. Despite the wide angular ranges of the long axis, which can be seen from the distribution of the solutions, the molecule is oriented fairly parallel to the straight channel, at least in its average location. Table 6.1 Ranges used for the parameters in the structure calculation and numbers of solutions after each step in the calculation from C P experiments on 4 D C N B / Z S M - 5 . Structural Parameter Step size Minimum* Maximum9 X 0.010 A 0.45 0.55 y 0.010 A 0.15 0.25 z 0.015 A -0.075 0.075 * 4 degrees 20 60 e 4 degrees 50 130 V 4 degrees -40 40 Initial number of locations tested ca.5.4x 106 Physically possible locations (dmi„b > 2 A) ca. 1.0 x106 k'iS-M2 linear correlation checked locations r2 > 0.92 3111 r2 > 0.92 1733 Locations found after peak intensity check including overlapped peaks at a given r2 r2 > 0.93 1291 r2 > 0.94 897 ' in fractional coordinates for (x, y, z} and in degree for {<)>, 9 , vj/} b the distance between the framework atoms and the atoms of D C N B Table 6.2 Values of the average six structural parameters of D C N B in the framework of Z S M - 5 at 305 K with r2 £ 0.93 from C P experiments on 4DCNB/ZSM-5 . x y z <(> 0 Average 0.486(6) 0.167(12) -0.008(8) 45.4(43) 94.9(164) -9.7(137) values (r2 > 0.93) 142 12.0 Figure 6.6 Plot of the measured CP rate constants against the calculated heteronuclear second moments for the average location of DCNB for 4DCNB/ZSM-5 with i2 > 0.93 at 305 K. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured k'is values, which were determined from the previous fittings of the CP curves, are represented by the error bars. 0 100000 200000 M2 (Hz2) -110 -112 -114 -116 -118 -120 "S i Chemical Shift (ppm from TMS) Figure 6.7 The NMR spectra for 4DCNB/ZSM-5 at 305 K; the spectra, in descending order, are the experimental spectrum (contact time = 20 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra. 25 35 45 55 55 75 95 115 135 Figure 6.8 Distributions of the solutions determined from 1 H/ 2 9 Si CP data of 4DCNB/ZSM-5 at 305 K with linear correlations of r2 > 0.92 (red squares), i2 > 0.93 (blue circles), and i2 > 0.94 (green triangles). The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of the six structural parameters for the translation (x, y, z) of the center of the DCNB (i.e. the center of the benzene ring) in fractional coordinates and the orientation (ij), 9, \y) of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with r > 0.93, given in Table 6.2. 143 Figure 6.9 (a) Scatter plot of the D C N B molecule in the plane of the molecule and (b) the 50% error ellipsoid representation for 4 D C N B / Z S M - 5 at 305 K at f > 0.93. < ~ X fcds—fc a i * . / \ 0 l—( . \ 7 C V Figure 6.10 Structure of 4DCNB/ZSM-5 complex from the C P experiment with 50% error ellipsoidal representations of the D C N B molecule at 305 K. The left figure shows the location of D C N B viewed from the zigzag channel of Z S M - 5 and the right shows viewed from the straight channel. For clarity, oxygen atoms in the framework have been omitted. 6.2.1.5 CP drain experiments As described in Chapter 4, C P drain experiments have several advantages when used to determine the C P rate constants, kiS. According to the fitted values of k)s and k,, the DCNB/ZSM-5 system is in the slow C P regime, which poses a difficulty of determining kls values accurately due to fast 1 H 7"ip relaxation by a conventional C P experiment. Figure 6.11 shows the data and fitting of a C P drain experiment on the 4DCNB/ZSM-5 system. The range of the contact times was up to 90 ms, and the fittings of the C P drain intensity (AS/S 0 ) growth correspond well to the theoretical solid curves. Once the fitted k/s values were obtained, the same approach was taken for the structure calculation as for the previous C P experiment on the same system. Table 6.4 represents the ranges of the structural parameters tested for the final calculation and numbers of the solutions after each step of the calculation. The average of the six structural parameters are given in Table 6.5, and the values are very close to those of the C P data. The experimental k/s values fit well to the calculated 144 kIS curve, which was derived from the second moments of the average location found with r2 > 0.93 (Figure 6.12). The predicted peak intensities including the overlapped peaks show good agreement with the experimental peak profile of the C P drain (better than the C P data) despite the low S/N of the C P drain difference spectrum in Figure 6.13. The distributions of the structural parameters are shown in Figure 6.14. Noticeable features of the results are the wide distributions of the angles 9 and (seen previously in the C P experiment), which could indicate two or more orientations of the DCNB in ZSM-5 . The 2D scatter plot (Figure 6.15) along with the 50% error ellipsoidal representation of the DCNB molecule on the molecular plane shows a very similar distribution of the solutions found to that from the C P data. Having similar average structural parameters and distributions of the solutions consequently yielded a similar structure of the DCNB/ZSM-5 complex, as shown in Figure 6.16. The final structures from the C P and C P drain data are almost identical to each other. The values of the average structural parameters show a good agreement between the two sets of solutions (CP and C P drain sets) found. The next section of this chapter presents the results of a single crystal X R D study on 4DCNB/ZSM-5 carried out to complement the NMR determined structures. 145 • Si11 y ' SiS .->•' Si2 %f k,s= 9.59 s 1 x*^ *,s = 8.98 s 1 0 60 0.40 0.20 0 00 0 60 0.40 0.20 0.00 0.60 Si3 * , s = 7.75 s'1 «..-J 0 10 20 30 40 50 60 70 80 90 Si10 * „ = 5.84S 1 .yi'' . >•*• Si 9 r ' *,s= 9.34 s'1 » j . ' * k, s * = 8.05 s'1 Figure 6.11 2 9 S i - 1 H C P drain experiment on 4 D C N B / Z S M - 5 at 305 K. The points are the experimental values of the intensities at chosen contact times and the solid curves are fitted according to Equation 3.6. The vertical axes represent the normalized signal difference (AS/So) in the C P drain data and the horizontal ones the range of drain contact times. For the curves corresponding to the groups of overlapped peaks, the fitted values of C P drain rate constant are the averaged values denoted as kis*. 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90 Table 6.3 Values of k'is for C P and kts for C P drain plots for 4 D C N B / Z S M - 5 at 305 K. k ' l s (s-1) k,s (s"1) Si8 9.32(18) 9.59(40) Si2 7.95(18) 8.98(38) Si11 6.59(12) 7.84(32) Si3 7.17(8) 7.75(26) Si4 4.68(34) 5.34(41) Si6 6.05(11) 7.29(27) Si10 4.92(11) 5.84(31) Si9 7.84(25) 9.34(33) Si1,5 6.55(12) 8.05(25) Si7,12 8.19(25) 9.02(33) 146 Table 6.4 Ranges used for the parameters in the structure calculation and numbers of solutions after each step in the calculation from CP drain experiments on 4DCNB/ZSM-5. Structural Parameter Step size Minimum" Maximum" x 0.010 A 0.45 0.55 y 0.010 A 0.15 0.25 z 0.015 A -0.075 0.075 tj> 4 degrees 20 60 0 4 degrees 50 130 y/ 4 degrees -40 40 Initial number of locations tested ca. 5.4 x 10 6 Physically possible locations (d m i n b > 2 A) ca. 1.0 x 10 6 k|S-M2 linear correlation checked locations r2 > 0.92 1857 r2 > 0.92 941 Locations found after peak intensity check including overlapped peaks at a given r2 r2 > 0.93 587 r2 > 0.94 357 a in fractional coordinates for {x, y, z) and in degree for {<(>, 0, vy} b the distance between the framework atoms and the atoms of D C N B Table 6.5 Average values of the six structural parameters of DCNB in the framework of ZSM-5 at 305 K with r2 > 0.93 from CP drain experiments on 4DCNB/ZSM-5. x y Z * 9 v Average 0.484(6) 0.177(12) -0.021(8) 47.0(45) 98.0(192) . -11.7(153) values (r2 £ 0.93) 15.0 10.0 5.0 0.0 -J--10 —4' k,s = 2.35x10-* M 2 +5.44 r2 = 0.918 100000 M2 (Hz2) -i—i— 200000 Figure 6.12 Plot of the measured CP drain rate constants against the calculated heteronuclear second moments for the average location of DCNB for 4DCNB/ZSM-5 with r > .0.93 at 305 K. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured kis values are represented by the error bars. 147 7,12 1,5 6 11 -110 -112 -114 -116 -118 -120 29Si Chemical Shift (ppm from TMS) Figure 6.13 NMR spectra for 4 D C N B / Z S M -5 at 305 K; the spectra, in descending order, are the experimental C P drain difference spectrum (AS/So) (contact time = 95 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra. 55 75 95 115 135 Figure 6.14 Distributions of the solutions determined from 1 H / 2 9 S i C P drain data of 4 D C N B / Z S M - 5 at 305 K with linear correlations of r2 > 0.92 (red squares), r2 > 0.93 (blue circles), and r 2 £ 0.94 (green triangles). The vertical axes and the horizontal axes are as indicated in Figure 6.8. The arrows indicate the 'average' values of the six structural parameters with ? > 0.93, shown in Table 6.5. i .$4 , , , , , 1 i • 1 1 • < < - J - J -1 0 1 2 J 4 -2 -1 0 1 2 J Figure 6.15 (a) Scatter plot of D C N B molecule in the plane of the molecule and (b) the 50% error ellipsoid representation for 4DCNB/ZSM-5 at 305 K at r2 > 0.93. 148 Figure 6.16 C P drain determined structure of 4 D C N B / Z S M - 5 complex with 50% error ellipsoidal representations of the DCNB molecule at 305 K. The left figure shows the location of D C N B viewed from the zigzag channel of Z S M - 5 and the right viewed from the straight channel. For clarity, oxygen atoms in the framework have been omitted. 6.2.2 Single crystal XRD study of 4DCNB/ZSM-5 In order to check the structures of 4DCNB /ZSM -5 determined by NMR, a single crystal diffraction experiment was carried out at 253 K, a temperature where the DCNB /ZSM -5 complex is confirmed to be in the orthorhombic Pnma space group according to the NMR study. The single crystal used had dimensions of hundreds of microns (ca. 400x250x100 pm 3). The initial indexing indicated the space group was the orthorhombic Pnma, and instead of solving the structure by direct methods, existing framework coordinates (the low loaded p-dichlorobenzene/ZSM-5 4 2) were used in the beginning to account for the electron density due to the Si and O atoms in the framework. Without applying any structural restraints for the framework atoms, this method usually works well for an organic/zeolite single crystal because of the weaker atomic scattering strengths of the atoms (H, C and N) in the organic molecule and the structural integrity of the zeolite framework topology. The initial refinement on the single crystal data of 4DCNB /ZSM -5 complex was done using the framework coordinates only, i.e., using only the S i and O atoms of the framework, and gave R t factor of ca. 0.10 and wR2 of ca. 0.22. At this point, a 3D Fourier electron density difference map was constructed, and a significant amount of unaccounted electron density was located in the channel intersection as expected from the NMR experiments. After slowly introducing the atoms of the D C N B molecule, atom by atom to the existing framework in order to construct a rigid molecule of D C N B using various structural restraints such as atomic bond length, bond angle and planar restraints, the refinement converged to R, = 0.06 and wR2 = 0.18, which is fairly good for a routine structure refinement. Another Fourier electron density difference map with respect to the molecular plane was now 149 constructed, and it now revealed a disordered electron density at the location of the channel intersection (Figure 6.17). The map shows four sites because of the mirror plane at {0, V*, 0}, which bisects the electron density vertically. Projection of the D C N B structure onto the Fourier map led to the conclusion that there were two unique sites for the organic guest molecule. When the second molecule was introduced to the refinement, the final refinement locations converged to = 0.05 and wR2 = 0.16. The occupancy factors of the two guest molecule locations yielded a relative occupancy ratio of ca. 2 to 1 between the sites DCNB1 (red) and DCNB2 (yellow). The final structure of the DCNB/ZSM-5 complex (Figure 6.18) shows two D C N B sites are quite close in location of their benzene rings (ca. 2 A) but their orientations are quite different. From the final structures, it is clear that the DCNB1 is directed more parallel to the straight channel indicating that the DCNB2 could be the result of "tight" packing of the electronegative C s N groups in the framework of ZSM-5. At this point, it is worthwhile to compare the NMR and X R D results in order to discuss the different features of the two methods and the structures obtained by both. Figure 6.17 Fourier electron density difference map of the 4 D C N B / Z S M - 5 , where the D C N B molecule found at the channel intersection of Z S M - 5 . The population ratio between the two sites was 0.35 to 0.18 for the red to yellow figures of the D C N B (ca. 2:1). The projection of the electron density shows four molecules due to the mirror plane of the Pnma space group, which bisects the map vertically. u v Relative occupancy 2 150 (b) < _ ^ Relative occupancy 1 Figure 6.18 Structure of the 4DCNB/ZSM-5 complex determined by single crystal XRD showing two sites for the DCNB molecule in the channel intersection. 6.2.3 Comments on the NMR and single crystal XRD results. The structure of 4DCNB/ZSM-5 has been determined by both NMR and single crystal X R D , which both show the guest DCNB molecule is at the channel intersection. The angles, 0 and y, in the NMR determination show large distributions, which indicate a wide distribution of the DCNB's orientation. The single crystal X R D result shows that the D C N B is at two possible sites, both of which are at the channel intersection, and the centers of the benzene rings are quite close to each other. One of the sites shows better coincidence with the straight channel than the other, which shows a substantial deviation of ca. 30 ° with respect to the straight channel. The relative occupancies for two sites are ca. 2 to 1, the site that is more coincident to the straight channel being twice as populated. 6.3 Study of 2 molecules per unit cell of DCNB/ZSM-5 Is there a loading of DCNB at which the D C N B / Z S M - 5 complex prefers one site for the. organic? What is the loading mechanism of DCNB in ZSM-5? In order to answer these questions and broaden our understanding of the structure determination method, we conducted the same NMR and single X R D experiments on a lower loaded form of the D C N B / Z S M - 5 (ca. 2 molecules per unit cell or 2DCNB/ZSM-5 hereafter) complex. The sample was prepared and its loading was checked by T G A which showed ca. 2.5 molecules per unit cell of D C N B . 151 6.3.1 Solid-state NMR study of 2DCNB/ZSM-5 The variable temperature experiments (Figure 6.19) show that the 2DCNB/ZSM-5 shows 11 resolved peaks at 300 K and the INADEQUATE experiment at 300 K shows the connectivities between Si T-sites, which yielded a very similar T-site assignment as previously found for the 4DCNB/ZSM-5. The variable contact time CP experiment was done and the intensity changes with respect to the contact time were plotted for the single Si T-sites and overlapped peak, which results from the NMR intensities of the Si7 and 12 T-sites. The fitting was done with a single value of fitted /0 and kt values, and the fitted k'iS and kt (both fitted and experimental) values were used for structure determination. This system is also in the slow CP regime, where initial intensity decays are due to the relatively fast kt value. The structure solution was done as previously: The structural parameters and solutions after each structure determination step are listed in Tables 6.7 and 6.8 for the CP and CP drain data respectively and the average structural parameters are listed in Table 6.9 and 6.10 for the CP and CP drain data. When the errors in the parameters are compared to those of 4DCNB/ZSM-5, their ranges are smaller, possible indications of more defined structures than those Of 4DCNB/ZSM-5. This is also true for the structural parameters for the CP drain data, which show consistency in both methods. The linear correlations between the k'ls for CP (kis for CP drain) and calculated second moments show good agreement (Figure 6;24). When the experimental and predicted spectra are compared, they agree well except for the overlapped peak and the Si4 T-site (Figure 6.25). The distributions of the six structural parameters show similar patterns for the CP and CP drain data in Figure 6.26. Compared to those of the 4DCNB/ZSM-5 sample, the distributions are better defined and have relatively smaller ranges. Neither set of distributions shows any clear signs of disorder, such as the two maxima in the distribution of 0, seen for the higher loaded system although the 0 and \y distributions are still quite large. The 2D scatter plots (Figure 6.27 and 6.28) in the molecular plane of the DCNB molecule show better defined ranges of solutions than those of 4DCNB/ZSM-5 for both CP and CP drain data. The final structures from the CP and CP drain data in Figure 6.29 agree well with each other; one 152 noticeable feature of both structures is the benzene ring is located closer toward the mirror plane than previously found for the higher loading. 7,12 300K Figure 6.19 2 9 S i M A S N M R spectra at the different temperatures indicated. The numbers above the selected resonances indicate their assignment to specific T-sites in the zeolite framework. The 2 9 S i 90 ° pulse length was 7.75 us and 128 scans were accumulated with a recycle delay of 7.5 s for each spectrum. -110 -112 -114 -116 -118 2 9 S i Chemical Shift (ppm from TMS) -120 153 (7,12) -110 -112 -114 -116 ppm -118 -120 Figure 6.20 Quantitative 29Si spectrum of 2DCNB/ZSM-5 at 300 K, 90 0 pulse was 7.35 us, recycle delay 35 s with 32 scans. The numbers above the resonances indicate the assignments to specific T-sites in the zeolite framework. The spectra in the bottom show the deconvoluted resonances. Figure 6.21 Together with the quantitative 29Si NMR spectrum (recycle delay of 35 s), Two-dimensional 29Si INADEQUATE spectrum of 2DCNB/ZSM-5 at 300 K is shown. 34 experiments, each with 480 scans and 6 s recycle delay, were acquired in the fi dimension. The echo delay during the double quantum preparation period was 16 ms and the sweep widths in the h and /i dimensions were 800 and 1600 Hz. The numbers over the resonance peaks are the assigned silicon T-sites. H10 H12 -114 H16 H18 2 9 S i Chem ica l Shift (ppm from TMS) H20 154 Si8 * ' , s =2.23: Si1 k',s = 1.69 s Si9 * ' « * = 1.97 Si3 k',s = 1.32 s 1 Si4 • k ' ,s = 0.75 s 1 Si2 * ' , s =1-84 s"1 Si11 A'„ =1.56 »•' Si5 * ' , s = 0.94 s"1 Si6 k',s = 0.89 s 1 Si10 =1.178-' 0 10 20 30 40 50 60 70 Si(7,12) = 1.94 s' Figure 6.22 Intensities of the 1 H/ 2 9 Si CP MAS NMR signals indicated as functions of the contact time for 2DCNB/ZSM-5 at 300 K. The intensities are normalized to those of a quantitative 2 9 Si spectrum with an equal number of scans. The points are the experimental values of the intensities at chosen contact times and the solid curves are fitted according to Equation 3.3. The fitted theoretical CP maximum k was 1.19 (a.u.) and kt was 67.6 s"1 (Ti p = 14.8 ms). 0 10 20 30 40 50 60 70 10 20 30 40 50 60 70 Si8 ' /c, s. = 2.44 s' Si3 " * I S = 1.54s 1 Si1 k,s = 2.10 s 1 SiZ k.s = 1.95 s' Si11 k,s = 1.69 s 1 Si4 k,, =0.76 i' Si5 * , s = 1.13 s 1 Si6 *,s = 1.23 s 1 Si10 /<,S- = 1.45S-1 0 20 40 60 80 Figure ,6.23 1 H/ 2 9 Si CP drain experiments on 2DCNB/ZSM-5 at 300 K. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.6. The vertical axes represent the normalized signal difference (AS/So) in the GP drain data and the horizontal ones the range of drain contact times. For the curves corresponding to the groups of overlapped peaks, the fitted values of CP drain rate constant are the averaged values denoted as kis*. 0 20 40 60 80 0 20 40 60 80 155 Table 6.6 Values of k)s for C P and k,s for C P drain plots for 2 D C N B / Z S M - 5 at 330 K. k'lsfs"1) kis (s"1) Si8 2.23(2) 2.44(10) Si2 1.64(2) 1.95(10) Si11 1.56(5) 1.69(14) Si3 1.32(2) 1.54(9) Si4 0.75(6) 0.76(16) Si5 0.94(3) 1.13(12) Si1 1.69(2) 2.10(14) Si6 0.89(4) 1.23(11) Si10 1.17(2) 1.45(10) Si9 1.97(3) 2.29(10) Si7,12 1.94(2) 2.26(7) Table 6.7 Ranges used for the parameters in the structure calculation and numbers of solutions after each step in the calculation from C P experiments on 2DCNB/ZSM-5 . y Structural Parameter Step size Minimum Maximum x 0.010 A 0.45 0.55 y 0.010 A 0.15 0.25 z 0.015 A -0.075 0.075 <p 4 degrees 20 60 0 4 degrees 50 130 if/ 4 degrees -40 40 Initial number of locations tested ca. 5.4x106 Physically possible locations (dm,n > 2 A) ca. 1.0 x 106 k'iS-M2 linear correlation checked locations r2 > 0.92 1733 r2 > 0.92 927 Locations found after peak intensity check including overlapped peaks at a given r2 r2 > 0.93 688 r2 > 0.94 486 156 Table 6.8 Ranges used for the parameters in the structure calculation and numbers of solutions after each step in the calculation from CP drain experiments on 2DCNB/ZSM-5. Structural Parameter Step size Minimum Maximum x 0.010 A 0.45 0.55 y 0.010 A 0.15 0.25 z 0.015 A -0.075 0.075 <j> 4 degrees 20 60 0 4 degrees 50 ' 130 IJ/ 4 degrees -40 40 Initial number of locations tested ca. 5.4 x 106 Physically possible locations (d m i „ > 2 A) ca. 1.0 x 106 k l s-M 2 linear correlation checked locations r2 > 0.92 2145 r2 > 0.92 1111 Locations found after peak intensity check including overlapped peaks at a given r2 r2>0.93 792 r2 > 0.94 463 Table 6.9 Average values of the six structural parameters of DCNB in the framework of ZSM-5 at 300 K with r2 > 0.93 from CP experiments for 2DCNB/ZSM-5. x y z # 0 w Average 0.480(7) 0.234(13) -0.030(2) 47.3(35) 92.4(116) -3.4(105) location (r2 ^ 0.93) Table 6.10 Average values of the six structural parameters of DCNB in the framework of ZSM-5 at 300 K with r2 > 0.93 from CP drain experiments for 2DCNB/ZSM-5. . x y z <p 6 y/ Average 0.479(7) 0.233(14) -0.036(7) 47.3(35) 91.9(140) -3.4(103) values (1^^0.93) ' 157 Figure 6.24 Plot of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2 D C N B / Z S M - 5 with r2 > 0.93 from the (a) C P and (b) C P drain data. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured k)s and kis values are represented by the error bars. -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 2 9 Si Chemical Shift (ppm from TMS) 29SI Chemical Shift (ppm from TMS) Figure 6.25 N M R spectra for 2 D C N B / Z S M - 5 at 300 K; the spectra are (a) the experimental spectrum (contact time = 20 ms), the predicted spectrum from the average solution from the structure calculation and the difference between the experimental and predicted from the CP, and (b) the experimental C P drain difference spectrum (AS/So) (contact time = 95 ms), the predicted spectrum from the average solution from the structure calculation and the difference between the experimental and predicted spectra from the C P drain in descending order. 158 (a) (b) 60 80 100 120 140 -35 -25 -15 Figure 6.26 Distributions of the solutions determined from 1 H / 2 9 S i (a) C P and (b) C P drain data on the 2DCNB/ZSM-5 complex at 300 K with linear correlations of r2 > 0.92 (red squares), r2 > 0.93 (blue circles), and r 2 > 0.94 (green triangles). The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of the six structural parameters for the translation (x, y, z) of the center of the D C N B (i.e. the center of the benzene ring) in fractional coordinates and the orientation (+, 0, vy) of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with 0.93, which are shown in Tables 6.9 and 10. ^ ^ ^ ^ ^ ^ ^ ^ - (b Figure 6.27 Scatter plot of the D C N B molecules (a) in the plane of the molecule and (b) its 50% error ellipsoid representation for 2 D C N B / Z S M - 5 from the C P experiments at 305 K at r2 > 0.93. -5 -2 0 1 2 J -} - J -1 0 1 2 3 Figure 6.28 Scatter plot of the D C N B molecules (a) in the plane of the molecule and (b) its 50% error ellipsoid representation for 2 D C N B / Z S M - 5 from the C P drain experiments at 305 K at r2 > 0.93. -3 -2 -1 0 1 2 3 159 Figure 6.29 Structures of the 2DCNB/ZSM-5 complex determined from (a) CP and (b) CP drain experiments at 300 K. (Left) view from the zigzag channel and (right) from the straight channel. 6.3.2 S i n g l e c rys ta l X R D s t ruc tu re of 2 D C N B / Z S M - 5 It took several attempts to obtain the single crystal X R D structure of 2DCNB/ZSM-5 . In most cases, the loaded crystal which was obtained was heterogeneous, i.e. the indexing the crystal revealed two or more phases, predominantly monoclinic and orthorhombic ones. When the guest molecule is relatively large, its loading into a large single crystal (> 100 pm) will take a much longer time than for a microcrystalline powder, which are small particles of < 10 pm. A similar phenomenon has been seen in Chapter 4 in the case of OXY/ZSM-5 . This makes obtaining a specific low loading of the guest molecule very difficult. Measuring the loading in a single large crystal (dimensions of -100 microns) of ZSM-5 is a challenge for any T G A instrument. After five or six failures with several different single crystals of ZSM-5 loaded with DCNB, a single crystal structure was obtained from data collected at 273 K. The final R, and wR2 values were 0.045 and 0.129 respectively assuming a single site for the D C N B molecule. A second site was added during the refinement process, but this time there was no improvement in the quality of the refinement. A Fourier electron density difference map was created after removing the DCNB site found, in order to check the residual electron density directly (Figure 6.30). The map clearly shows a 160 single site for the D C N B with its symmetry equivalent occupancy. The population factor for the organic molecule was 0.289, which is a total of ca. 2.3 molecules per unit cell of DCNB in ZSM-5 using the multiplicity of 8. The final refined structure of the 2DCNB/ZSM-5 is presented in Figure 6.31. The D C N B molecule is at the channel intersection of ZSM-5 and in the direction of the straight channel with a deviation, and its benzene ring center is slightly off of the mirror plane at {0, V*, 0} in fractional coordinates. Figure 6.30 Fourier electron density difference map of the D C N B sites for 2 D C N B / Z S M - 5 found at the channel intersection of Z S M - 5 . The projection of the electron density shows two molecules due to the mirror plane of the Pnma space group, which bisects the map vertically. A—4 ¥ r \ / V T / M ' S r\ I \ A. j s v . . Figure 6.31 Structure of the 2 D C N B / Z S M - 5 complex determined by single crystal X R D showing the D C N B molecule in the channel intersection 6.3.3 C o m m e n t s o n the N M R a n d s i n g l e c rys ta l X R D resu l t s o f 2 D C N B / Z S M - 5 In contrast to the 4DCNB/ZSM-5 system, both the NMR and single crystal structures show one site for the guest. The benzene ring centers of the D C N B for both are in similar locations. The orientation of the D C N B for the NMR structure, however, is almost parallel to the straight channel while that of the single crystal X R D structure shows a deviation of ca. 20 ° with respect to the straight 161 channel. The NMR structure still shows large distributions for the angles, 0 and but somewhat less than the 4 D C N B / Z S M - 5 structure. In the NMR structure, the D C N B seems to be a little closer to the mirror plane than in the single crystal X R D structure. However, this may be due to the symmetrically equivalent two sites for the DCNB in the orthorhombic Pnma ZSM-5 . When the symmetry of the space group Pnma mandates two equivalent sites for the D C N B molecule as a result of the mirror plane at {0, VA, 0}, the NMR method would find some locations from the other site due to this type of disorder. When the sites are close to each other, the NMR structure shows close to the average structure of the two. From the studies of 4DCNB/ZSM-5 and 2DCNB/ZSM-5 , although the NMR results indicate possibilities of static disorder for both complexes, they can be verified only by single crystal X R D if possible. In the next section of this chapter, another disordered system is investigated and compared to the DCNB/ZSM-5 systems. 6.4 Study of 2 molecules per unit cell of p-dinitrobenzene in ZSM-5 The structure of the ca. 4 molecules per unit cell of DNB/ZSM-5 complex (4DNB/ZSM-5 hereafter) has been studied previously, and its structure determined by NMR and single crystal X R D 1 0 3 . The DNB molecule similar to DCNB, which has para substituted bulky nitro groups, was found in the channel intersection of the framework of ZSM-5 according to this study. While the NMR structure showed a single molecule parallel to the straight channel, the structure determined by single crystal X R D showed two sites, which were almost orthogonal to each other, one parallel to the straight channel and the other facing the zigzag channel. The population ratio between those two sites was ca. 2.5 to 1.4 molecules per unit cell of ZSM-5 respectively. The location of DNB determined by NMR was similar to one of the guest sites determined by single crystal X R D , which is parallel to the straight channel at the channel intersection, but with relatively large 50% error ellipsoids for the guest. The study of ca. 2 molecules per unit cell of DNB in ZSM-5 (2DNB/ZSM-5 hereafter) was continued to explore any indication of detecting static disorder by NMR and to find possible correlation between the structures determined by NMR and single crystal X R D by comparing the studies of both DCNB systems. 162 6.4.1 Solid-state NMR study of 2DNB/ZSM-5 The lower loaded form of DNB/ZSM-5 (2DNB/ZSM-5) was prepared as described in Chapter 2 and its loading checked by TGA, showing it to have ca. 2.4 molecules per unit cell of DNB in ZSM-5. 29 Si variable temperature experiments show that the 2DNB/ZSM-5 has 10 resolved peaks at 330 K (Figure 6.32). The INADEQUATE experiment at 330 K (Figure 6.34) shows the connectivities between Si T-sites, which were assigned as previously described and are in agreement with the previous study. The order of the Si T-sites is quite similar to that of the previous study on the higher loaded form of DNB/ZSM-5 at 285 K with the exception of peaks 1, 4, 5 and 7. Variable contact time C P and C P drain experiments were carried out and the intensity changes with respect to the contact time plotted for the resolved single Si T-sites as previously described. These results are presented in Figures 6.35 and 6.36 with the fittings using a single value of fitted / 0 and k, values for the C P curves. This system is also in the slow C P regime. The structure solution was carried out as previously described and the structural parameters and solutions after each of the structure determination steps are listed in Tables 6.12 and 6.13. The average structural parameters are given in Tables 6.14 and 6.15. The parameters themselves (Figure 6.37) show no large distributions except for 8 in the C P drain experiment. The linear correlation between the /c' / s for C P data (kts for C P drain) for the average location of DNB and the calculated second moments shows good agreement for both the C P and C P drain data in Figure 6.38. The experimental and predicted spectra, which were calculated based on the average location of DNB, were compared, and they also show good agreement for both C P and C P drain (Figure 6.39). The 2D scatter plots (Figures 6.40 and 6.41) show similar distributions of the solutions found for both C P and C P drain data; however, the solutions for the C P data seem to be a little more scattered than those of the C P drain. The final C P and C P drain structures (Figure 6.42) show similar orientations to each other being in the channel intersection of the framework of ZSM-5 ; however, for the location of the molecular center, the C P structure locates the molecular center farther away from the mirror plane at {0, %, 0} than the C P drain. 163 9,10 1.5 /A/VXA AA° -110 -112 -114 -116 -118 -120 M S i Chemical Shift (ppm from TMS) Figure 6.32 **Si M A S NMR spectra of 2DNB/ZSM-5 at the different temperatures indicated. The numbers above the selected resonances indicate their assignment to specific T-sites in the zeolite framework. The 2 9 S i 90 ° pulse length was 12.5 us and 256 scans were accumulated with a recycle delay of 5 s for each spectrum. Figure 6.33 Quantitative Si M A S spectra of 2DNB/ZSM-5 at 330 K, 90 0 pulse of 14 us, recycle delay 50 s with 64 scans. The numbers above the resonances indicate the assignments to specific T-sites in the zeolite framework. The spectra at the bottom show the deconvoluted resonances. i 1 1 1 i i -110 -112 -114 -116 -118 -120 ppm 164 -122 H -\20-\ -116 H -112 -108 Figure 6.34 Two-dimensional Si INADEQUATE spectrum of the 2DNB/ZSM-5 complex at 330 K. 36 experiments, each with 1280 scans and 3 s recycle delay, were acquired in the ri dimension. The echo delay during the double quantum preparation period was 18 ms. and the sweep widths in the fe and U dimensions were 800 and 1600 Hz. The indicated peak assignments were determined from the observed correlations. -112 -114 -116 2 9 S i Chemical Shift (ppm from TMS) ppm o.io o.oe 0.06 0.04 0.02 0.00 Si6 ft',. = 3.78 s' Si12 ft',.. = 3.22 s1 10 20 30 40 50 60 Figure 6.35 Intensities of the 1 H / 2 9 S i C P M A S N M R signals indicated as functions of the contact time for 2DNB/ZSM-5 at 330 K. The intensities are normalized to those of a quantitative 2 9 S i spectrum with an equal number of scans. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.3. The fitted theoretical C P maximum k was 1.56 (a.u.) and ki was 130 s" 1 (T i p = 7.70 ms). 10 20 30 40 50 10 20 30 40 50 60 165 SiS jSi2 Si11 * / S = 2.25 s' ^ < * , S = 2.24 s"'' * , s = 1.39 s 1 % • ' SiS Si4 Si7 / * , s = 1.65 s ' * , S = 0.82 s 1 kls = 2.56 s'1 . . • * - • — r -•-T » 1 1 1 , , Figure 6.36 1 H / 2 9 S i C P drain experiments on 2DNB/ZSM-5 at 330 K. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.6. The vertical axes represent the normalized signal difference (AS/So) in the C P drain data and the horizontal ones the range of drain contact times. For the curves corresponding to the groups of overlapped peaks, the fitted values of C P drain rate constant are the averaged values denoted as kis*. Si12 Si6 - * , s = 1.58 s"1 k,s = 1.16 s"1 . . .»-• v-r' * • • Si1,5 Si9,10 k,s' = 1.41 s 1 *,.,* = 1.65 s 1 • • • 0 20 40 60 0 20 40 60 60 Table 6.11 Values of k'is for C P and kis for C P drain plots for 2DNB/ZSM-5 at 330 K. k',s (s"1) k,s (s"1) Si8 5.20(18) 2.25(17) Si2 4.78(11) 2.24(18) Si11 3.98(27) 1.39(17) • Si3 3.90(13) 1.65(16) Si4 3.16(30) 0.82(18) Si7 4.71(15), ' 2.56(19) SM2 3.78(20) 1.58(24) Si6 3.22(22) 1.16(20) Si1,5 3.89(13) 1.41(7) Si9,10 4.10(15) 1.65(9) 166 Table 6.12 Ranges used for the parameters in the structure calculations and numbers of solutions after each step in the calculation from C P experiments on 2DNB/ZSM-5 . Structural Parameter Step size Minimum Maximum jc 0.013 A 0.46 0.55 y 0.012 A 0.02 0.24 Z 0.017 A -0.075 0.01 ^ 3 degrees 30 70 0 5 degrees 50 150 y/ 5 degrees -40 40 Initial number of locations tested ca. 4.6 x 10 6 Physically possible locations (d m l n > 2 A) ca. 1.4 x 10 6 k'i S-M 2 linear correlation checked locations r2 > 0.92 6561 r2>0.92 1409 Locations found after peak intensity check , including overlapped peaks at a given r2 r2>0.94 518 Table 6.13 Ranges used for the parameters in the structure calculations and numbers of solutions after each step in the calculation from C P drain experiments on 2DNB/ZSM-5 . Structural Parameter Step size Minimum Maximum 0.55 0.25 0.03 60 130 40 Initial number of locations tested ca. 3.9 * 10" Physically possible locations (dm i„ > 2 A) ca. 4.3 * 10 5 k|S-M2 linear correlation checked locations r2 > 0.90 1593 Locations found after peak intensity check including overlapped peaks at a given r2 x 0.010 A 0.45 y 0.010 A 0.15 Z 0.015 A -0.075 ^ 4 degrees 20 9 4 degrees 50 {j/ 4 degrees -40 r2 > 0.90 991 r2 > 0.91 652 r2 > 0.92 399 167 Table 6.14 Average values of the six structural parameters of the lower loaded form of DNB/ZSM-5 at 330 K with r2 > 0.93 from C P experiments. x y z tj> 9 V Average 0.505(8) 0.122(27) -0.008(15) 52.2(45) 97.2(146) -0.9(119) location (r2 Z 0.93) Table 6.15 Average values of the six structural parameters of the lower loaded form of DNB/ZSM-5 at 330 K with r2 > 0.92 from C P drain experiments. x y z $ 0 V Average 0.484(5) .0.189(19) -0.027(13) 46.9(32) 96.7(139) -7.2(106) values (r2 ^ 0.91) Figure 6.37 Distribution of the solutions determined from (a) C P and (b) C P drain data of 2DNB/ZSM-5 at 330 K with linear correlations of r2 > 0.92 (red squares), r2 > 0.93 (blue circles) and r2 2: 0.94 (green triangles) for CP, and r2 > 0.90 (red squares), r2 £ 0.91 (blue circles) and r2 > 0.92 (green triangles) for C P drain. The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of solutions of six structural parameters for the translation (x, y, z) of the center of the DNB (i.e. the center of the benzene ring) in fractional coordinates and the orientation (<j>, 9, vy) of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with ? > 0.93 for C P and r*> 0.92 for C P drain, given in Tables 6.14 and 6.15 respectively. 168 8.0 M2 (Hz2) M2 (Hz2) Figure 6.38 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2DNB/ZSM-5 with r2 > 0.93 and r 2 > 0.91 for (a) C P and (b) C P drain data respectively. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured k)s and kis values are represented by the error bars. -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 2 9 Si Chemical Shift (ppm from TMS) M S i Chemical Shift (ppm from TMS) Figure 6.39 Si M A S N M R spectra for 2DNB/ZSM-5 at 330 K; the spectra are (a) in descending order, the experimental spectrum (contact time = 40 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted from the C P experiment with r2 > 0.93, and (b) in descending order, the experimental C P drain difference spectrum (AS/So) (contact time = 75 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted from the C P drain experiment with r2 > 0.91. 169 • Figure 6.40 (a) Scatter plot of DNB molecules in the plane of the molecule and (b) the 50% error ellipsoid representation for 2DNB/ZSM-5 by C P at 330 K at r2 > 0.94. - 3 - 2 - 1 0 1 2 3 -a ^ - i o 1 2 3 170 6.4.2 S i n g l e c r ys ta l s t ruc tu re of 2 D N B / Z S M - 5 The single crystal and microcrystalline powder of ZSM-5 were loaded with DNB as described in Chapter 2. The loading of the single crystal was indirectly confirmed by a T G A desorption study of the microcrystalline powder, which showed ca. 2.4 molecules of DNB per unit cell of ZSM-5. The single crystal XRD data were collected at 273 K and refined as described earlier in this chapter. The final refinement yielded Ri = 0.036 and wR2 = 0.12 with two sites for the DNB molecule. The two sites were in similar locations to those previously found for the loading of 4 molecules of DNB in ZSM-5: each DNB is oriented almost orthogonal to the other according to the Fourier difference map (Figure 6.43) and the final structures (Figure 6.44). The population for each site was: 1.92 and 0.51 molecules per unit cell for DNB1 and DNB2 respectively. The DNB1 site is occupied ca. 3.8 times more than the DNB2, which was ca. 1.8 times for the 4-molecule DNB/ZSM-5 system, indicating the DNB1 site is preferred to the DNB2 site by the guest molecule, and it would be occupied first during the loading (as for the DCNB system). - 3 - 2 -1 Figure 6.43 Fourier electron density difference map of the DNB molecule sites found at the channel intersection of Z S M - 5 , displayed in the molecular plane. The population ratio between the two sites was 1.92 to 0.51 for the red and yellow (DNB1) to blue and green (DNB2) figures of the DNB (ca. 4:1). The projection of the electron density shows four molecules due to the mirror plane of the Pnma space group (please note that the blue and green figures of DNB2 sites are very close to each other), which bisects the map vertically. 171 Relative occupancy 4 Relative occupancy 1 Figure 6.44 Structure of the 2DNB/ZSM-5 complex determined by single crystal X R D showing two sites for the DNB molecule, (a) DNB1 and (b) DNB2, in the channel intersection. 6.4.3 C o m m e n t s o n the N M R and s ing le crysta l X R D s t ruc tu res of 2 D N B / Z S M - 5 The structures determined by both NMR and single crystal X R D show that the DNB molecule is at the channel intersection of the framework. The structures from the C P and C P drain experirnents show quite similar orientations for the two; however, the location of the DNB benzene ring centers are somewhat different, showing the DNB from the C P experiment is farther from the mirror plane at y = %. The single crystal X R D results show that the DNB molecules are at two sites, very similar to the previously determined structure of ca . 4 molecules per unit cell of DNB/ZSM-5. However, the occupancy of the DNB2 site, where the molecular long axis is perpendicular to the straight channel, of 2DNB/ZSM-5 is much lower than that of 4DNB/ZSM-5 ; for 4DNB/ZSM-5 , the ratio between the DNB1 and DNB2 sites is ca. 2 to 1 whereas it is ca. 4 to 1 between the DNB1 and DNB2 sites for 2DNB/ZSM-5. As a minor orientation, DNB2 would contribute less to the heteronuclear dipolar coupling between the organic and framework than DNB1. Because the DNB molecules in 172 ZSM-5 are not abundant, especially at a low loading of the organic, the DNB2 site (ca. 20% of the total DNB sites in 2DNB/ZSM-5) would be even less occupied. A further complication comes from the fact that the dipolar couplings from DNB2 (using the location and orientation found by the single crystal X-ray refinement) are much lower than DNB1 as no solution for the DNB2 site was found by NMR and thus will make a smaller contribution to the total effect. Since we are detecting the overall effect of the organic-framework interactions, the majority contribution comes from the DNB1. 6.5 NMR cf. single crystal XRD Is NMR capable of predicting the structure of a statically disordered guest/zeolite complex such as 4DCNB/ZSM-5 or DNB/ZSM-5? In order to answer the question, one should have a clear understanding on how the NMR structure is determined from the solid-state N M R data. The key of 1 29 the NMR structure determination is based on discrimination of the C P sources ( H for C P and Si for C P drain) with respect to the framework and guest organics as mentioned in Chapter 3. Selective discrimination of the protons in the guest organic molecule on the silicon atoms in the framework by the heteronuclear dipolar coupling strength can yield the relative proximities between the two. Currently, the NMR structure determination p rogram 2 5 , 1 0 3 uses one rigid molecule to find the locations that satisfy the linear correlation between the C P constants and heteronuclear second moments. Therefore, in theory, if there were two distinctive sites for the organic guest molecule in an organic/zeolite complex, they could show up as two localized distributions of the solutions found after the structure determination calculation by NMR data. However, in practice, when two sites are very close to each other, a clear separation between two different sets of solutions may not be possible, and the distributions of the solutions would show as one group of broad distributions of solutions, which may seem continuous. 173 For NMR, disorder in the guest/zeolite system can have a significant effect on the results. When the system is dynamically disordered, e.g. in the case of the benzene ring flipping motion, it is possible to detect such a motion by 2 H or 1 3 C static NMR. When the system is statically disordered, however, the disorder may be difficult to recognize, as NMR is not capable of probing long-range order like diffraction techniques. Especially for the multiple sites and orientations for the guest in the framework, other parameters in the NMR method may be used to detect such disorders; that is, the numbers of solutions, r2 values, scatter plots and the linearity of kiS vs. M2- Also, if possible, probing the dipolar coupling between the substituent groups and the framework could give complementary information about the guest orientation along with the aromatic- 1 H/framework- 2 9Si C P experiments. As shown in Chapter 4, the location of the benzene ring of o-xylene can be determined better in the o-xylene-de/ZSM-5 experiments whereas the orientation of the organic was better defined by o-xylene-d 4 /ZSM-5 where the methyl groups are the source of CP. A further challenge is that the DNB molecule does not have cross-polarizing nuclei in its substituent group unlike the xylenes. This limits the structure determination to the protons in the benzene ring. For the reasons mentioned, distinguishing the disorder present in DNB/ZSM-5 may not be trivial by NMR. In addition, the weaker dipolar coupling of DNB2 for 2DNB/ZSM-5 as well as the lower occupancy could impede the detection of the second site by NMR. There are many advantages to structure determinations by NMR, which is robust and insensitive to the motions of the guest molecules 1 0 3 . Even in the presence of certain motions of the guest, NMR should give an answer for the structure of a guest/zeolite complex. As seen in the case of DCNB/ZSM-5 , NMR is capable of determining the guest/zeolite structure even at 330 K, which is a challenge for diffraction methods due to the increased temperature factors. On the other hand, single crystal X R D "shows" a time averaged crystal structure, which is derived from a collection of 'snapshots' on the crystal lattice. Once determined, the single crystal 174 X R D structure is unambiguous. However, its success depends on the immobile nature of the structure, i.e. any motion or disorder present would affect reliable structure determination. Often, single crystal X R D is performed at a low temperature (< 173 K) in order to minimize any possible motions; however, this may not always be possible, for example, in the case of ZSM-5 whose space group depends on temperature as well as the guest loadings. 3 7 Another aspect of single crystal X R D to consider is the generation of the crystal lattice structure from the reciprocal space. The determined atoms in the single crystal X R D results are those in the unit cell and its content. This assumes that every unit cell has the same content. A guest organic molecule, even in the same framework, however, could have different distributions and orientations depending on its loading and shape. The loading and shape dependency of the guest molecule in the framework can be well observed in this chapter for the cases of D C N B and DNB as well as previous s tud ies . 8 1 , 1 0 3 Different distributions and orientations of the guest molecules in the framework of zeolite cause structural disorders. If such a disorder of a guest molecule exists, it would be expressed as a 'double exposure' to the overall guest molecules in the guest/zeolite structure from single crystal X R D . This, 'time-space-averaged' representation of a structure is intrinsic to single crystal X R D and can add uncertainty to the final structure of the disorder where orientations are closely related. In order to understand the system fully, therefore, N M R and single crystal X R D should be used as complementary techniques whenever possible. 175 6.6 Summary In this chapter, the locations and orientations of guest p-dicyanobenzene in two different loadings and low loaded p-dinitrobenzene were determined by solid-state NMR, and subsequently by single crystal X R D . The structures determined by NMR show single sites for all the guests in the framework whereas the single crystal X R D results indicate multiple sites for both 4 molecules per unit cell loadings of p-dicyanobenzene and 2 molecules per unit cell loadings of p-dinitrobenzene in Z S M -5. The NMR structures, however, show relatively wide distributions of the guest molecules, which can be seen in the structural parameters. This could indicate that the guest molecule in the system is disordered to a certain degree. The robustness is one of the advantages of the NMR method, which can determine structure well above room temperature. Perhaps the limitation in detecting some disorder in a guest/zeolite system by NMR is an inevitable sacrifice to maintain this attribute and both techniques should be used whenever possible. 176 Chapter 7 Investigation of Mixtures: Study of the Structures of the 2+2 Benzene/p-Xylene/ZSM-5 Complex by NMR 7.1 Introduction In real industrial processes, zeolites interact simultaneously with more than one type of 39 adsorbed guest molecule. For example, in the famous xylene conversion by the Mobil process , gasoline production from methanol 3 8 , and p-xylene (PXY) production from benzene (BENZ) and methanol involving ZSM-5 , all the reactions require more than one single chemical. The intent of the present chapter is to investigate the potential of the NMR technique to determine the structures of zeolite/sorbate complexes involving mixtures. The basic idea that will be explored and tested in detail in this chapter is that the structure of a mixture of two organics may be determined by the solid state NMR method by studying two complementary mixtures, where one is completely deuterated and the other not. The premise is that the NMR method will determine the position and orientation of the non-deuterated organic, the other being 'transparent' in the structure determination. Two experiments of this type should give the complete structure of the mixture. Benzene and p-xylene were chosen as guests because of their involvement in the xylene production from benzene and because there is a fair amount of background information on their behaviors as single species. To date there is no information at all in the literature on the structures of complexes of mixtures. Several NMR studies on benzene/ZSM-5 (BENZ/ZSM-5 ) 1 4 5 and p-xylene/ZSM-5 (PXY /ZSM-5)26,37,145,146 h g v e b e e n d Q n e j p t h e p g s t A | o a d i n g s t u c | y 1 4 5 of BENZ /ZSM-5 by 2 9 S i NMR showed a complex behavior. The monoclinic form persisted up to a loading of 4.2 molecules per unit cell; between 4.2 and 5.4 molecules per unit cell a second form exists, perhaps a mixture of multiple phases; above 5.4 molecules per unit cell, the framework is in the orthorhombic space group (Figure 177 7.1). Many of the guest/ZSM-5 complex systems s t u d i e d 2 5 , 2 6 , 3 2 , 3 J , 4 5 , 4 6 , 8 1 ' 1 ° 7 ' 1 4 5 showed that the space group changed from the monoclinic space group (P2i/n) of the empty framework of ZSM-5 to the orthorhombic space group (Pnma) of the guest/ZSM-5 complex at loadings of ca. 2 to 4 molecules per unit cell Comparatively, benzene in ZSM-5 appears to have weaker interactions with the ZSM-5 framework than these other organics, assuming the space group change is mainly the consequence of the presence of the guest. To date, the only structural studies of the BENZ/ZSM-5 complex are by powder dif fract ion 4 8 , 5 0 , 5 3 , 5 4 . Sacerdote-Peronnet and Mentzen have investigated four molecules per -unit cell loading of benzene in MFI 5 4 , and determined that it is located in the channel intersection of the framework from powder neutron diffraction data. In order to fit the data, two different sites for the benzene molecules were required, indicating complexity of the guest/zeolite structure. In another study by Fitch and coworkers 5 3, the benzene molecules were found in three sites at a loading of eight molecules per unit cell The arrangement of the benzene molecules is rather complicated; two of these are in the channel intersection, and the third is at the center of symmetry in the straight channel. The orientations showed considerable disorder, and a total of six independent molecules was required to fit the powder neutron diffraction pattern. The PXY/ZSM-5 complexes, on the other hand, have been well s t u d i e d 2 6 , 3 2 , 4 0 , 1 3 6 , 1 4 0 " 1 4 2 , 1 4 6 as they have industrial significance. To date, the structures of PXY/ZSM-5 in the low loaded and high loaded forms have been determined by both solid-state NMR and single crystal diffraction 1 . The PXY/ZSM-5 complex adopts the orthorhombic Pnma space group from ca. 1.6 to 4 molecules per unit cell, and subsequently changes to another orthorhombic P212121 space group at higher loadings of the organic molecule up to the maximum loading of ca. 8 molecules per unit cell. In the space group P212121, the PXY molecules occupy the zigzag channel of ZSM-5 as well as the straight channel, and that the loading into the zigzag channel of ZSM-5 instigates the space group change from Pnma to P2-i2i2i. This type of space group change is also seen in the p-dichlorobenzene/ZSM-5 system. Figure 7.2 shows 2 9 S i NMR spectra as a function of the loading of PXY in ZSM-5. The changes in the spectra reflect the changes in the framework of ZSM-5 to accommodate the guest molecules. From 0 to 1.6 loadings of PXY, the 2 9 S i NMR spectra indicate that the system is in the monoclinic space group, P2 1 /n, as seen from the 24 peaks in the spectrum. The change from the 178 monoclinic space group to the orthorhombic Pnma, however, is gradual, involving a mixture of the two phases as shown in the spectra. When the loading becomes 2 molecules per unit cell, the 2 9 S i NMR shows only 12 peaks indicating the complete transformation to the orthorhombic Pnma space group. A similar transition can be seen at higher loadings from 4 to 8 molecules per unit cell loadings. Between 4 and 6 molecules per unit cell, there are two orthorhombic phases present and above 6, only the P2Jn space group is present, as reflected in the 24 resonances in the 2 9 S i spectrum due to the loss of the mirror plane at y = %. In summary, the PXY /ZSM-5 complex shows two space group changes at loadings between 0 and 8 molecules per unit cell, which is known to be the maximum possible loading of PXY. As discussed in Chapter 3, the key starting point in a guest/zeolite structure determination by NMR is the assignment of the 2 9 S i NMR peaks according to their known connectivities. Thus, the combination of loading and temperature of the guest/zeolite complex should be carefully chosen to avoid any complications due to contributions from two phases or space groups to the 2 9 S i spectra. The 2 9 S i NMR peaks of the P X Y / Z S M - 5 complexes have been previously ass igned 2 9 , 9 4 by INADEQUATE experiments; to date, assignments are available for the low loaded P Y X / Z S M - 5 (ca. 2 to 3 molecules of P X Y per unit cell of ZSM-5) at 273 K 9 4 and at 300 K 2 9 , and for the maximum loading of P X Y in Z S M - 5 2 9 , one of the most complex assignments, as it has 24 2 9 S i N M R peaks. Figure 7.3 shows the peak assignment for the low loaded P X Y (2 molecules per unit cell) in ZSM-5 at 300 K, where 21 out of 22 possible connectivities are observed. 179 benzene per unit cell Figure 7.1 Si N M R spectra of benzene in ZSM-5 at room temperature at the loadings indicated. Data are from reference 37. J _ I i i i I I I I I I I i -110 -115 -120 "Si Chomical Shift (ppm Ifpm TMS) i i JJ p-xylene per unit cell V X _ A & L _ 2.0 1.6 1.4 1.2 U V A ^ _ _ _ I.C IA_XJI O.J 0.6 0.4 0.0 I ' l l l i ' - 1 1 0 - 1 1 5 - 1 2 0 **Si Chemical Shift (ppm from TMS) p-xylene ,A per unit cell 8.0 7.0 6.0 5.0 4.0 2.0 i i i I i i i i | i i i i | -110 -115 -120 '^ Si Chemical Shift (ppm from TMS) Figure 7.2 2 9 S i M A S N M R spectra of p-xylene in ZSM-5 at room temperature at the loadings indicated. The arrows indicate the disappearance (downward) and the appearance (upward) of the peaks as the loading of p-xylene increases. Data are from reference 94. 180 Figure 7.3 2 9 S i INADEQUATE spectrum of p-xylene per unit cell in Z S M - 5 at a loading of 2 molecules per unit cell at 300 K with a 1D M A S N M R spectrum above. The INADEQUATE data are from reference 29. Si Chemical Shift (ppm from TMS) 7.2 Solid State NMR experiments on the 2+2 mixtures in ZSM-5 Based on the previous studies on various organic/ZSM-5 complexes, the combination of 2+2 molecule loadings'" of B E N Z and P X Y into the framework of ZSM-5 was chosen as it has equal numbers of molecules of the two components and the total loading corresponded to the low-loaded orthorhombic form (Pnma) for PXY/ZSM-5 . The loadings were checked by T G A desorption study, which indicated ca. 4 to 5 molecules per unit cell for the overall mixture of the B E N Z / P X Y in ZSM-5 . The exact ratio of B E N Z to P X Y can be determined from the intensities of the 1 H M A S peaks for some combinations of mixtures (see later), but not for the sets having completely deuterated organics, and confirmed ca. 1:1 loadings. From both the T G A and 1 H NMR results, more precise estimations of the B E N Z and P X Y loadings were made. For the completely deuterated mixtures, the loadings were checked by T G A only, which found that the total amount of the initial guest materials added to the *** The designation "2+2 B E N Z / P X Y " will be used to indicate mixture composition of two molecules of benzene plus two molecules of p-xylene. For example, 2+2 BENZ-de/PXY-de/ZSM-S represents 2 benzene-de and 2 p-xylene-d6 molecules per unit cell of Z S M - 5 . 181 empty ZSM-5 powder was roughly equal to the amount determined by T G A . A preliminary 2 9 S i NMR spectrum on the 2+2 B E N Z / P X Y / Z S M - 5 complex yielded a somewhat surprising (and puzzling at the same time) result, as the spectrum was almost identical to that of the 2 molecules per unit cell loading of PXY /ZSM-5 complex. This raised the possibility that the benzene had been lost from the mixed system as it is quite volatile by itself and interacts less strongly with the framework. Thus, a second TGAdesorpt ion study was performed on the same sample, which yielded an identical result to the first one carried out before the spectrum was recorded. A later 1 H MAS NMR (see Figure 7.60) on 2+2 BENZ/PXY-d 4 /ZSM-5 showed two equal intensity peaks at ca. 2 and 7 ppm clearly confirming the presence of the two different molecules in the complex. Compared to BENZ/ZSM-5 , the benzene in the 2+2 BENZ /PXY/ZSM-5 complex is much more strongly held. The 2 9 S i NMR spectrum at 293 K (room temperature) showed 7 resolved peaks, and variable temperature 2 9 S i NMR spectra indicated the 2+2 B E N Z / P X Y / Z S M - 5 complex system possesses the orthorhombic Pnma space group up to 240 K (Figure 7.4). In order to prevent the volatile guest molecules from escaping the framework of ZSM-5 , experiments above 293 K were not attempted. Figure 7.5 compares two quantitative 2 9 S i NMR and their deconvoluted spectra at 270 and 293 K. At 270 K, 2 9 S i NMR spectrum of the complex shows 9 resolved peaks whereas 7 peaks are resolved at 293 K. Because the structure determination steps of the NMR experiments involve CP, the qualities of the C P spectra were carefully examined. Figure 7.6 shows two C P spectra and their deconvolutions for 2+2 BENZ-de/PXY-oyzSM-S at the two different temperatures indicated. All of the twelve peaks from the orthorhombic Pnma space group can be seen; however, there are minor peaks appearing in other parts of the spectra. These minor signals in C P are due to a very small percentage of the high-loaded form, which cross-polarizes much more efficiently than the low-loaded form. These minor peaks are much more visible when the contact times are short reflecting the efficiency of their cross-polarization. Hence, CP, which can be used for efficiency, can possibly have small errors from these peaks showing as residuals in difference spectra. C P drain, on the other hand, will be much less affected as these signals will be much smaller and .will disappear much faster in the drain experiment and not affect C P drain curve. 182 At 270 K and 293 K, the assignments of the Si NMR peaks could be made from previous INADEQUATE studies 2 7 ' 9 4 of PXY /ZSM-5 , assuming the chemical shifts of the 2 9 S i NMR peaks of the B E N Z / P X Y / Z S M - 5 framework were induced primarily by the P X Y molecule. However, in order to eliminate any possible incorrect assignments or errors, 2 9 S i INADEQUATE experiments on ca. 2+2 B E N Z / P X Y / Z S M - 5 were carried out at both 270 K and 293 K. The INADEQUATE experiment at 293 K (room temperature) yielded 19 connectivities out of a possible 22, and the 2 9 S i NMR peaks were subsequently assigned (Figure 7.7) using the peak assigning program described previously 1 1 7. The assignments on the 2 9 S i N M R peaks of the 2+2 B E N Z / P X Y / Z S M - 5 complex were identical to those of the previous assignments 2 9 on 2 molecules per unit cell loading of P X Y / Z S M - 5 at 300 K. The INADEQUATE experiment at 270 K showed 16 strong connectivities and 4 weak ones that are involved with the Si 3 T-site (Figure 7.8). Because of the presence of these weak connectivities, the program did not give one conclusive assignment, rather yielded many possible assignments. The final assignment was made from the room temperature assignment of the same sample with help of the peak shift patterns from the variable temperature experiments. The assignment was confirmed to be one of the possible assignments proposed by the program. In order to determine the structure of the 2+2 B E N Z / P X Y / Z S M - 5 complex from the NMR experiments, 3 different samples were made up using different combinations of the guest organics: 2+2 BENZ-d<5/PXY-d 6/ZSM-5, 2+2 BENZ-de/PXY-oVZSM-S and 2+2 B E N Z / P X Y - d 1 0 / Z S M - 5 (Figure 7.9). These samples were used for variable contact time C P and C P drain experiments at two different temperatures, room temperature and 270 K. At the end of the N M R study, there were a total of 12 sets of NMR data available for the complete structure determination. 183 1,5,6,7,9 Room Temp. 280K 270K 260K 250K 240K -108 -110 -112 -114 -116 -118 -120 -122 ^Si Chemical Shift (ppm from TMS) Figure 7.4 Si M A S N M R spectra of 2+2 B E N Z / P X Y / Z S M - 5 at the different temperatures indicated. The numbers above the selected resonances indicate their assignment to specific T-sites in the zeolite framework. The 2 9 S i 90 ° pulse length was 12.5 us and 64 scans were accumulated with a recycle delay of 5 s for each spectrum. A JftAA AA AA A A A A A AA -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 ppm ppm Figure 7.5 Quantitative 2 9 S i M A S N M R spectra of 2+2 B E N Z / P X Y / Z S M - 5 with a 90 ° pulse length of 14 us, recycle delay 30 s with 128 scans at (a) 270 K and (b) 293 K. The numbers above the resonances indicate the assignments to specific silicon T-sites in the zeolite framework, found from a 2 9 S i INADEQUATE experiment. The spectra at the bottom represent the deconvoluted resonances. 184 —r" • 1 • 1 • 1 • "i — -110 -112 -114 -116 -118 PPm 2 9 S i Chemical Shift (ppm from TMS) Figure 7.7 Along with quantitative Si NMR spectrum (recycle delay of 30 s), two-dimensional 2 9 S i INADEQUATE spectrum of 2+2 B E N Z / P X Y / Z S M - 5 at 270 K is shown. 36 experiments, each with 1280 scans and 3 s recycle delay, were acquired in the U dimension. The echo delay during the double quantum preparation period was 20 ms, and the sweep widths in the h and U dimensions were 800 and 1600 Hz. The indicated peak assignments were determined from the observed correlations as previously. The numbers over the resonance peaks are the assigned silicon T-sites for Z S M - 5 . 185 -110 -112 -114 -116 -118 PPm 2 9 Si Chemical Shift (ppm from TMS) Figure 7.8 Along with quantitative Si NMR spectrum (recycle delay of 30 s), two-dimensional 2 9 S i INADEQUATE spectrum of 2+2 B E N Z / P X Y / Z S M - 5 at 293 K is shown. 36 experiments, each with 1280 scans and 3 s recycle delay, were acquired in the ft dimension. The echo delay during the double quantum preparation period was 15 ms, and the sweep widths in the h and ft dimensions were 800 and 1600 Hz. The indicated peaks assignments were determined from the observed correlations as previously. The numbers over the resonance peaks are the assigned silicon T-sites for ZSM-5 . Figure 7.9 The three mixtures of the guest organics used in this study: (a) 2+2 BENZ-de/PXY-oe (b) 2+2 B E N Z -d6 /PXY-d 4 and (c) 2+2 BENZ/PXY-d io in Z S M - 5 . Each mixture contains a fully deuterated guest and a selectively protonated guest in order to discriminate the dipolar coupling between the protons of the different organic guest molecules and 2 9 S i atoms of the framework. The enlarged structures in each pair indicate the guest molecules whose locations were probed by the N M R experiments. 186 7.3 2+2 BENZ-ayPXY-c/6/ZSM-5 7.3.1 Structural study of 2+2 BENZ-de/PXY-de/ZSM-S at 270 K The fitted C P and C P drain curves for the NMR data at 270 K are shown in Figures 7.10 and 7.11. The curve fittings, based on Equations 3.3 (CP) and 3.6 (CP drain), yield the experimental /c',s (CP) and kls (CP drain) values. According to the experimentally measured 1 H 7"1p values (ca. 5 ms) and the fitted k)s and kiS values, the system is in the slow C P regime, where the C P drain experiment has advantages over CP. The ranges of the structural parameters tested and the numbers of solutions found are presented in Tables 7.1 and 7.2 for the C P and C P drain data, respectively. For the chosen r2 values, the structure calculations yielded large numbers of solutions and the location of the guest P X Y - d 6 is generally well defined, although its orientation is not, as seen in the Euler angles, v|/, from the C P data, and 9 and from the C P drain data, which show wide ranges. For the average structural parameters (Tables 7.3 and 7.4), the calculated second moments and the k'ts values for the C P experiment correlate with r2 = 0.613 (Figure 7.12a), which is lower than the values normally found due mainly to Si 10 which has very weak polarization transfer (Figure 7.10). A better correlation (r2 = 0.902) for the C P drain data indicates that it is a more reliable experiment as expected (Figure 7.12b). The solid lines, k'IS and kIS, are derived from the average values in Tables 7.3 and 7.4 for the C P and C P drain data respectively. The experimental C P spectra in Figure 7.13a show 'bumpy' base lines, more prominent at a shorter contact time, which cannot be accounted for exactly with the predicted spectra. This is due to a small amount of an extra set of peaks, as discussed at the beginning of the section, which start to show along with the twelve main peaks as space group change as seen in Figure 7.2 for p-xylene/ZSM-5 for CP, and these 'pseudo peaks' are results of this complication. This contributes to the low r2 value of the linear correlation between the k'iS and M2 values at 270 K although the fittings of the C P curves seem to agree well with the experimental /c' / s values. As expected, this effect is hardly observed in either the quantitative 2 9 S i or C P drain spectra: the former because the percentage is very low and the latter because the dephasing is very fast, as 187 can be seen in the experimental and predicted spectra for the C P drain data (Figure 7.13b). This indicates that this 'accentuated' growth of the peaks due to the P X Y in the zigzag channel is limited to the C P experiments. Although the sensitivity of the C P process is considered advantageous in general, it could act as a 'double edged sword' in some cases. In this case, the sensitivity of C P can in fact hamper the correct determination of the organic/zeolite structure as it favors a small concentration of the minor contribution from a "somewhat" impure state to the region of interest. Thus, C P data should contain errors involving the few extra P X Y molecules in the zigzag channel, and they would not yield as reliable structures reflected in low r2 values. The C P drain results are not affected as much as the C P data, and the structure from the C P drain data seems to be more reliable with higher r2 values. The distributions of the structural parameters are shown in Figure 7.14 for both C P and C P drain experiments. The distribution of the solutions with respect to the y-axis is very symmetric about y = Vi for C P drain as well as other parameters, indicating the coexistence of two symmetry related structures. Especially, the parameter 9 shows bimodality in the graph supporting the presence of symmetrically distributed solutions. The scatter plots of the solutions in the planes of the molecules are presented in Figures 7.15 and 7.16 for the C P and C P drain experiments respectively. These plots also show differences in the distributions of the solutions between the C P and C P drain data. The scatter plot of the C P drain solutions in the molecular plane shows a more symmetrical distribution, which reflects symmetrical distributions of the structural parameters. The differences become more obvious in the final structures from the C P and C P drain data in Figure 7.17, which are quite different; the C P structure shows a rather 'disoriented' P X Y molecule that is off the center of the channel intersection whereas the C P drain structure shows that the P X Y is at the center of the mirror plane at y = Vi with its orientation parallel to the straight channel. The drain structure is considered the more reliable of the two as discussed earlier. 188 0.50 0.40 0.30 0 20 0.10 0 .00* 0 5 0 T 0 40 0 30 0.20 0.10 0.00 Si10 *",s = 4.16 s Si3 k\s = 14.4 s 1 V Si12 '•k'ls = 11.6 s' Si11 = 11.2 S 1 r • / Si1,7(5,6,9 f k;s = 10.2 s' Figure 7.10 Intensities of the 1 H / 2 9 S i C P M A S N M R signals indicated as functions of the contact time for 2+2 BENZ-cfe/PXY-de/ZSM-5 at 270 K. The intensities are normalized to those of the quantitative 2 9 S i N M R spectrum with an equal number of scans. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.3. The fitted theoretical C P maximum /o was 8.23 (a.u.) and ki was 71.3 s ' 1 ( 7 i p = 14.0 ms). 1.00 O.flO 0 60 0.40 0.20 0.00 Si3 k,s =5.31 s ' Si2 * „ =4.60 s' 1 Si1,7,5,6,9 fc,s" = 5.48 s'' Si4 = 2.52 s'' Si12 Si11 kls= 4.47 s'1 'kls = 4.74 s'1 . . . " — * Figure 7.11 Intensities of the 1 H / 2 9 S i C P drain M A S N M R signals indicated as functions of the contact time for 2+2 BENZ-de /PXY-oyZSM-S at 270 K. The points are the experimental values of (AS/So) at chosen contact times and the solid curves are calculated according to Equation 3.6. For the curves corresponding to groups of overlapped peaks, the fitted values of C P drain rate constant are the average values as denoted as kis*. 20 40 60 20 40 60 189 Table 7.1 Ranges of the structure calculation and numbers of solutions after each step for the C P data. Structural Parameter Step size Minimum Maximum x 0.01 A 0.45 0.52 y 0.01 A 0.1 0.25 Z 0.015 A -0.075 0.09 if, 3 degrees 0 60 0 5 degrees 40 150 y/ 5 degrees -60 50 Initial number of locations tested ca. 4.9 x 10 6 Physically possible locations (d m i n > 2 A) ca. 1.1 x 10 6 k'is-M2 linear correlation checked locations r2 > 0.60 11562 r2>0.60 4667 Number of acceptable locations after peak intensity check including overlapped peaks at a r2 > rj.61 3722 given r r2>0.62 2913 Table 7.2 Ranges of the structure calculation and numbers of solutions after each step for the C P drain data. Structural Parameter Step size Minimum Maximum jc 0.014 A 0.45 0.55 y 0.015 A 0.1 0.4 Z 0.021 A -0.075 0.075 ^ 7 degrees 0 . 6 0 0 8 degrees 0 180 ij/ 9 degrees -90 120 Initial number of locations tested ca. 7.1 x 10 6 Physically possible locations (d m i n > 2 A) ca. 1.1 x 10 6 . k | S -M 2 linear correlation checked locations r2 > 0.92 27784 r2>0.92 10215 Number of acceptable locations after peak intensity check including overlapped peaks at a r2 > 0.93 7048 given r tr2 > 0.94 4569 190 Table 7.3 Average structural parameters from the C P data 0 location 0.483(10) 0.190(27) 0.028(19) 29.4(111) 107.0(164) -12.8(241) (/S:0.61) Table 7.4 Average structural parameters from the C P drain data x y z <j> 0 V Average location {r2 2: 0.93) 0.501(11) 0.250(44) -0.019(19) 30.7(141) 89.5(334) 1.4(470) Figure 7.12 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 BENZ-cyPXY-cyzSM-5 at 270 K for (a) C P with r2 > 0.61 and (b) C P drain data with r2 s 0.93. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured k'is and kis values are represented by the error bars. 191 -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 2 9Si Chemical Shift (ppm from TMS) Si Chemical Shift (ppm from TMS) Figure 7.13 The NMR spectra for 2+2 B E N Z - d e / P X Y - d e / Z S M - S at 270 K; the spectra are (a) the experimental spectrum (contact time = 7.2 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for CP, and (b) the experimental C P drain difference spectrum (AS/So) (contact time = 20 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for C P drain. 0.45 0.47 0.49 0.51 0.1 0.15 0.2 0.25 0.015 0.065 0 20 40 60 40 60 80 100 120 140 -60 -40 -20 0 20 40 192 0 20 40 60 0 60 120 180 -90 -30 30 90 Figure 7.14 Distribution of the solutions determined from 1 H / 2 9 S i (a) C P with linear correlations of r 2 > 0.60 (red squares), r 2 > 0.61 (blue circles), and r2 > 0.62 (green triangles) and (b) C P drain data with linear correlations of r 2 > 0.92 (red squares), r 2 > 0.93 (blue circles), and r 2 > 0.94 (green triangles) of 2+2 BENZ-de /PXY-de /ZSM-S at 270 K. The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of solutions of the six structural parameters for the translation (x, y, z) of the center of the PXY-d6 (i.e. the center of the benzene ring) in fractional coordinates and the orientation (+, 0, vy) of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters from Tables 7.3 and 7.4. Figure 7.15 (a) Scatter plot of the P X Y - d 6 molecules at r2 > 0.62 in the plane of the molecule and (b) its 50% error ellipsoid representation for 2+2 BENZ -de /PXY -de /ZSM-5 from C P at 270 K. Figure 7.16 (a) Scatter plot of the P X Y - d 6 molecules r2 > 0.94 in the plane of the molecule and (b) its 50% error ellipsoid representation for 2+2 BENZ -de /PXY -de /ZSM-5 from C P drain at 270 K. 193 Figure 7.17 The N M R determined locations of the P X Y - d 6 in the B E N Z - o y P X Y - c V Z S M - S complex at 270 K from (a) C P and (b) C P drain experiments 7.3.2 Structural study of 2+2 BENZ-de/PXY-de/ZSM-S at 293 K Figures 7.18 and 7.19 show the fittings of C P and C P drain data respectively. From the fitted kiS and k, values, the system is in the slow C P regime. The fittings for the C P data at short contact times are relatively good with exceptions of the overlapping peaks. Since the fitting of C P does not account for the ks values of 2 9 S i , minor discrepancies in the C P data fitting may be expected. Also, there are minor contributions from the higher loaded form of the complex. The C P drain data also show more discrepancies in the fittings than the 270 K data. Thus, it can be assumed that cross-polarization between 1 H and 2 9 S i at 293 K is not as efficient as at 270 K. This can be explained by the increase of molecular motions when the temperature is increased, which will weaken the dipolar coupling interactions. However, the k'iS (CP) and kiS (CP drain) values found show similar relative orders for the silicon T-sites, and therefore, similar structures can be expected from the two data sets. Peaks Si8, 2, 4, 3, 10 were used to determine the location of P X Y - d 6 for both C P and C P drain data; the number of peaks is the smallest set ever attempted using the structure determination 194 program. Both C P and C P drain data yielded reasonable numbers of solutions with high r2 values (Tables 7.5 and 7.6). The linear correlations between the experimental C P rate constants (/c,s) and the second moments (M2) found from the average locations are presented for both the C P and C P drain results (Figure 7.20). The correlation for the C P drain shows good agreement between the k/s values and the M2 found from the average structure with i2 = 0.966. The linear correlation for the C P data (r2 = 0.744, which is lower than the cut-off r2) means that the average structure is not well defined. The predicted spectra from these average locations, which agree well with the experimental spectra are presented along with the experimental spectra in Figure 7.21. While the predicted C P spectrum seems to replicate the experimental one reasonably well, the multiple peaks for the overlapping region seem to show differences in chemical shift to the experimentally determined peaks for the same sites for the C P data (Figure 7.21a). This could be due to differences in the values of the chemical shift between the preliminary 2 9 S i quantitative and the C P spectra as the chemical shift values of the quantitative spectrum were used to simulate the predicted spectrum. The ranges found for the structural parameters show good agreement between the C P and C P drain results, and the distributions of the structural parameters with respect to the numbers of solutions found are given in Figure 7.22. The solutions found from the C P data show fairly large distributions in the six structural parameters, especially y, which shows a sign of bimodality, which indicates presence of multiple sites for the P X Y molecule in the framework. For the C P drain result at 293 K, however, this type of range separation is not so clear although the ranges of the angles, 0 and vy, seem rather large (Figure 7.22b). The scatter plots for the C P and C P drain solutions found are shown in Figures 7.23 and 7.24, and are reasonably well localized for both the C P and C P drain results. The plot for the C P shows a more symmetric distribution of the solutions found than that for the C P drain. The final structures (Figure 7.25) are very similar to each other; for both C P and C P drain, the P X Y is at the center of the channel intersection with its long axis parallel to the direction of the straight channel. 195 1.20 1.00 0.80 0.60 0.40 0.20 0.00 1.20 1.00 o.eo 0.60 0.40 0.20 0.00 1.20 1.00 0.80 0.60 D.40 0.20 0.00 Si8 /(•,, = 10.42 s' 1 Si12,11 k'K = 6.43 s ' 1 Si2 k'is =6.27 s' 1 Si4 k\s = 2.63 s"1 0 10 20 30 40 50 60 Si3 Si10 *',s = 10.23 s 1 k',s = 3.36 s 1 Figure 7.18 Intensities of the 1 H / 2 9 S i C P M A S N M R signals indicated as functions of the contact time for 2+2 BENZ-de/PXY-de/ZSM-S at 293 K. The intensities are normalized to those of the quantitative 2 9 S i spectrum with an equal number of scans. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.3. The fitted theoretical C P maximum lo was 5.83 (a.u.) and ki was 177.4 s" 1 (Tip = 5.6 ms). 0 10 20 30 40 50 0 10 20 30 40 50 60 Figure 7.19 Intensities of the 1 H / 2 9 S i C P drain M A S N M R signals indicated as functions of the contact time for 2+2 BENZ-de/PXY-de/ZSM-5 at 293 K. The points are the experimental values of (AS/So) at chosen contact times and the solid curves are calculated according to Equation 3.6. For the curves corresponding to groups of overlapped peaks, the fitted values of C P drain rate constant are the average values as denoted as kis*. 0.00 f-'-196 Table 7.5 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the C P data. Structural Parameter Step size Minimum Maximum x 0.014 A 0.45 0.52 y 0.015 A 0 0.5 z 0.02 A -0.075 0.01 ip 5 degrees 0 60 0 7 degrees 20 180 [j/ 7 degrees -60 60 Initial number of locations tested ca. 5.7 x 106 Physically possible locations (d m i n > 2 A) ca. 1.2x106 k'|S-lvl2 linear correlation checked locations r2 > 0.97 27898 r2>0.97 4817 Number of acceptable locations after peak ' intensity check including overlapped peaks at a r2 > 0.98 2328 given r r2>0.99 614 Table 7.6 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the C P drain data. Structural Parameter Step size Minimum Maximum x 0.01 A 0.45 0.55 y 0.015 A 0.1 0.4 z 0.021 A -0.075 0.075 ip 5 degrees 0 60 0 7 degrees 40 150 \j/ 7 degrees -60 50 Initial number of locations tested ca. 6.9 x 106 Physically possible locations (d m i n > 2 A) ca. 1.6 x 106 k i S -M 2 linear correlation checked locations r2 > 0.97 44833 Number of acceptable locations after peak intensity check including overlapped peaks at a given r r2 > 0.97 r2 > 0.98 r2 > 0.99 8883 4282 1210 197 Table 7.7 Average structural parameters from the C P data J L location 0.493(11) 0.248(61) -0.023(19) 23.5(137) 91.3(289) 0.6(243) (f2 7> 0.98) Table 7.8 Average structural parameters from the C P drain data 0 JL. Average location (r2 2 : 0.98) 0.488(9) 0.252(38) -0.022(11) 27.8(96) 92.7(248) -3.8(312) (a) 16.0 14.0 12.0 10.0 8.0 6.0 0.0 ^2 ^ i 1 0 f A '3.57x10"5M2 + 3.35 4 r2 = 0.744 100000 200000 300000 (b) ! 2 10.0 i 8.0 6.0 -4.0 2.0 -0.0 k'is = 2.09x10"3M2+ 1.55 r 2 = 0.966 . 100000 200000 M2 (Hz2) M2 (Hz2) Figure 7.20 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 BENZ-cfe/PXY-d6/ZSM-5 at 293 K with i2 > 0.98 for (a) C P and (b) C P drain data. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured k)s and kis values are represented by the error bars. 198 (1,5,6,7,9) -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 2 9Si Chemical Shift (ppm from TMS) 2 9Si Chemical Shift (ppm from TMS) Figure 7.21 NMR spectra for 2+2 BENZ-cfe/PXY-oyzSM-S at 293 K; the spectra are (a) the experimental spectrum (contact time = 7.2 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for CP, and (b) the experimental C P drain difference spectrum (AS/So) (contact time = 20 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for C P drain. 199 Figure 7.22 Distributions of the solutions determined from 1 H / 2 9 S i (a) C P and (b) C P drain data of 2+2 B E N Z -de/PXY-de/ZSM-S at 270 K with linear correlations of r2 > 0.97 (red squares), f2 > 0.98 (blue circles), and r2 > 0.99 (green triangles). The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of solutions of the six structural parameters for the translation (x, y, z) of the center of the PXY-d6 (i.e. the center of the benzene ring) in fractional coordinates and the orientation {§, G, y) of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with r2 > 0.99 for the C P and C P drain data. Figure 7.23 (a) Scatter plot of the P X Y - d 6 molecules in the plane of the molecule at r 2 > 0.99 and (b) its 50% error ellipsoid representation for 2+2 BENZ-de/PXY-de/ZSM-5 from C P at 293 K. •5 4 , , , , . 1 -I , 1 1 , , 1 -3 -2 -1 0 1 2 3 .J .2 -1 0 1 2 3 Figure 7.24 (a) Scatter plot of the P X Y - d 6 molecules in the plane of the molecule at r 2 > 0.99 and (b) its 50% error ellipsoid representation for 2+2 BENZ-de/PXY-de/ZSM-5 from C P drain at 293 K. 200 7.3.3 C o m m e n t s o n the N M R s t r u c t u r e s of B E N Z - a y P X Y - o V Z S M - 5 All final structures for the both temperatures, 270 K and 293 K, show that the P X Y - d 6 molecule is at the channel intersection of ZSM-5 . The 270 K C P drain, 293 K C P and C P drain structures are quite similar to each other, which indicates marginal structural changes between two temperature ranges. As the C P spectra at shorter contact times show evidence of extra P X Y molecules in the zigzag channel and the C P data yielded less well defined average structures, the C P data are considered somewhat less reliable than the C P drain. 201 7.4 2+2 BENZ-cfVPXY-ctyZSM-S 7.4.1 Structural study of 2+2 BENZ-crVPXY-ctyZSM-S at 270 K The C P and C P drain experiments were performed on the sample of 2+2 BENZ-oVPXY-ayzSM-5 at 270 K. As seen in Figures 7.26 and 7.27, fits of both C P and C P drain data show excellent agreement between the experimental points and the fitted curves, determined by Equations 3.3 and 3.6. The C P fittings show that the system is in the slow C P regime, but this did not present any problems in determining the structures. Both C P and C P drain systems were tested with similar ranges of the structural parameters, and a good number of solutions were found with high r2 values (r2 > 0.91) as listed in Tables 7.9 and 7.10. The average structural parameters determined from the C P and C P drain data are similar (Tables 7.11 and 7.12). Figure 7.28 shows the linear correlations between the experimentally determined kiS values and the calculated M2 values. For both C P and C P drain results, the correlations seem quite good although the orders of the M2 values are not exactly the same. The predicted spectra for both C P and C P drain data are in good agreement with the experimental C P and C P drain spectra, respectively, as seen in Figure 7.29 with no residual signals due to the high-loaded form. The distributions of the structural parameters in Figure 7.30 show no evidence of multiple sets of solutions as they are confined to fairly narrow ranges. Figures 7.31 and 7.32 show the distributions of the solutions in the molecular planes as scatter plots, which are well localized, a good indication of well-defined structures. Figure 7.33 displays the final determined structures of 2+2 BENZ -d6 /PXY-d 4 /ZSM-5 from the average values for the C P and C P drain data at 270 K. As mentioned, the two data sets yielded very similar structural parameters, and as expected, their final structures, thus, are similar to each other, with the center of the PXY (the benzene ring center) slightly off-center from the mirror plane at y = Vi. The orientations of the molecules deviate slightly from the axial line parallel to the straight channel (the b axis). They also agree well with the C P drain structure of the 2+2 BENZ-de /PXY-de/ZSM-S system discussed previously. 202 1.00 0.80 0.60-0.40-0.20-0.00 1.00 0.80 0.60 0.40 0.20 0.00 1.00 0.80 0.60 0.40 0.20 o:oo Si8 ft'ls -5.58 s' Si3 ft',, = 10.1 s 1 Si2 Si4 k'ts = 9.60 s 1 ft',s = 2.48 s 1 Si1 Si10 ft',s = 6.73 s 1 "fc',s = 2.92 s 1 ir""'' .-:. " ' | , - • • • * • — • • — Si12 k'is ~ 5.38 s 1 Si11 k;s = 6.07 s 1 - • • • • • — • • • 10 20 30 10 20 30 Si7,5,6,9 ft'(s = 7.06 s'1 10 20 30 40 Figure 7.26 Intensities of the 1 H / 2 9 S i C P M A S N M R signals indicated as functions of the contact time for 2+2 BENZ-cfe/PXY-c V Z S M - 5 at 270 K. The intensities are normalized to those of the quantitative 2 9 S i N M R spectrum with an equal number of scans. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.3. The fitted theoretical C P maximum /o was 5.17 (a.u.) and ki was 203.0 s" 1 (Tip = 4.9 ms). 1 .UV 0.80 0.60 0.40 0.20 0.00 1.00 0.80 0.60 0.40 0.20 0.00 1.00 0.80 0.60 0.40 0.20 0.00 Si8 \ k,s = 4.75 s 1 Si3 k,s = 10.3 s'1 SM ft,, = 6.80s1 Si2 fc,s = 8.07 s' Si12 ft,s = 5.49 s 1 Si10 ft,s = 2.29 s'1 Si4 ft,s = 3.00 s 1 Si11 ft,s = 5.83 s' Si7,5,6,9 ft,,* = 5.82 s' Figure 7.27 Intensities of the 'HrSi C P drain M A S N M R signals indicated as functions of the contact time for 2+2 BENZ-d6 /PXY-d 4 / ZSM-5 at 270 K. The points are the experimental values of (AS/So) at chosen contact times and the solid curves are calculated according to Equation 3.6. For the curves corresponding to groups of overlapped peaks, the fitted values of C P drain rate constant are the average values as denoted as kis*-0 20 40 60 0 20 40 60 0 20 40 60 80 203 Table 7.9 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the C P data. Structural Parameter Step size Minimum Maximum x 0.016 A 0.44 0.55 y 0.015 A 0.1 0.4 z 0.021 A -0.075 0.03 <p 6 degrees -30 120 0 6 degrees 40 140 yz 5 degrees -10 50 Initial number of locations tested ca. 6.1x10 6 Physically possible locations (dm i n > 2 A) ca. 1.3 x 106 k'|S-M2 linear correlation checked locations r2 > 0.91 10847 r2>0.91 1502 Number of acceptable locations after peak intensity check including overlapped peaks at a r2 > 0.92 972 given r r^O.93 627 Table 7.10 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the C P drain data. Structural Parameter Step size Minimum Maximum x 0.014 A 0.45 0.55 y 0.015 A 0.1 0.4 Z 0.021 A -0.075 0.075 tfi 5 degrees -10 100 0 7 degrees 30 150 y 7 degrees 0 80 Initial number of locations tested ca. 6.7 x 106 Physically possible locations (dm,n > 2 A) ca. 1.1 x 106 k|S-M2 linear correlation checked locations r2 > 0.91 7656 r2 > 0.91 2024 Number of acceptable locations after peak intensity check including overlapped peaks at a r2 > 0.92 1498 given r r2 > 0.93 820 204 Table 7.11 Average structural parameters from the C P data 0 (r2 a: 0.92) JL locaHon 0.490(7) 0.192(24) -0.030(14) 44.2(281) 84.4(63) 15.0(41) Table 7.12 Average structural parameters from the C P drain data 0 JL locaHon 0.482(10) 0.213(28) -0.037(14) 41.4(263) 88.2(88) 12.1(53) (r2 2.0.92) (a) 14.0 12.0 10.0 8.0 6.0 4.0 -I 2.0 0.0 . . , • - 2 . ' iv::'-. •1Ci> k'is = 1.28x10" M2 + 2.11 r 2 = 0.979 ) 20000 40000 60000 M2 (Hz2) (b) 14.0 12.0 10.0 8.0 a 6.0 Je 4.0 0.0 / * 3 ^ •fi" > " . . - - 1 1 .12. • 8. -•'4 iov- 1.15x10-" M 2 r 2 = 0.964 + 2.25 20000 40000 M2 (Hz2) 60000 80000 Figure 7.28 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 BENZ-cfe/PXY-cVZSM-S at 270 K with r 2 > 0.93 from (a) C P and (b) C P drain experiments. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured k)s and kis values are represented by the error bars. 205 - 1 1 0 - 1 1 2 - 1 1 4 - 1 1 6 - 1 1 8 - 1 2 0 - 1 1 0 - 1 1 2 - 1 1 4 - 1 1 6 - 1 1 8 - 1 2 0 2 9 Si Chemical Shift (ppm from TMS) 2 9Si Chemical Shift (ppm from TMS) Figure 7.29 2 9 S i NMR spectra for 2+2 BENZ-oyPXY-oyzSM-5 at 270 K; (a) in descending order for CP, the experimental spectrum (contact time = 7.2 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order and (b) for C P drain, the experimental C P drain difference spectrum (AS/S 0 ) (contact time = 20 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra. 206 0.45 0.47 0.49 0.51 0.13 0.18 0.23 0.28 -0.075 -0.05 -0.025 -10 10 30 50 70 90 65 75 85 95 105 115 0 10 20 30 Figure 7.30 Distribution of the solutions determined from 1 H / 2 9 S i (a) C P and (b) C P drain data for 2+2 B E N Z -de/PXY-cVZSM-S at 270 K with linear correlations of r2 > 0.92 (red squares), r2 > 0.93 (blue circles), and r2 > 0.94 (green triangles). The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of solutions of the six structural parameters for the translation (x, y, z) of the center of the D C N B (i.e. the center of the benzene ring) in fractional coordinates and the orientation (ij>, 0, w) of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with r > 0.94, which are given in Tables 6.9 and 10. 5 Figure 7.31 (a) Scatter plot of the P X Y - d 4 molecules in the plane of the molecule at r2 > 0.93 and (b) its 50% error ellipsoid representation for 2+2 BENZ-d6 /PXY-d 4 / ZSM-5 from C P at 270 K. Figure 7.32 (a) Scatter plot of the P X Y - d 4 molecules in the plane of the molecule at r2 > 0.93 and (b) its 50% error ellipsoid representation for 2+2 BENZ-d6 /PXY-d 4 / ZSM-5 from C P drain at 270 K. 207 Figure 7.33 The final structures of the 2+2 BENZ-d6 /PXY-d 4 /ZSM-5 complex at 270 K from (a) C P and (b) C P drain experiments. 7.4.2 Structural study of 2+2 BENZ-de/PXY-oVZSM-S at 293 K Figures 7.34 and 7.35 show the plots of the experimental intensities and their fits to Equation 3.3 and 3.7 in order to obtain the /c' / s (CP) and kiS (CP drain) values, respectively, for 2+2 BENZ-d6 /PXY-d 4 /ZSM-5 at 293 K. Overall, the fits are in excellent agreement with the experimental results. The k'is and kiS values found are close to each other with similar orders suggesting sjmilar structures will be obtained. Tables 7.13 and 7.14 contain the parameter ranges used for the calculations and the numbers of solutions found. One thing to note is that the numbers of solutions are relatively large in general. This could perhaps indicate that the locations of the P X Y - d 4 are diffuse as more sites satisfy the criteria of the NMR data. The average structural parameters for the solutions found from the C P and C P drain data are listed in Tables 7.15 and.7.16 respectively. The values are almost identical to each other, indicating the final structures from the average solutions will be the same. Figure 7.36 shows the linear correlations between the measured k'ts and kts values and the respective M2 values, as the solid lines, the k'IS and kis, being derived from the average values in Tables 7.15 and 7.16 for 208 the C P and C P drain data respectively. The predicted NMR spectra based on these correlations are almost identical to the experimental ones (Figure 7.37). Figure 7.38 shows the distributions of the average structural parameters for both C P and C P drain; at a glance, the values are relatively large compared to the previous 270 K data. Figures 7.39 and 7.40 are the scatter plots of PXY-d 4 molecules in the molecular plane from the C P and C P drain data, respectively, which require a larger range ( 5 x 7 A 2 ) of the plane than usual ( 3 x 5 A 2 ) in order to contain most solutions found. The final structures are shown with the average PXY locations in the framework of ZSM-5 in Figure 7.41. As predicted from the average structural parameters, the structures are almost identical to each other. For both structures, the PXY-d 4 molecule is located in the middle of the mirror plane at y = Vi, and its long axis is parallel to the crystallographic b axis. I . W 0.80 0.60 0.40 0.20 0.00 1.00 0.80 0.60 0.40 0.20 0.00 1.00 0.80 0.60 0.40 0.20 0.00 Si8 k;s = 2.22 s 1 Si2 k',s" 3.64 s 1 • i< Si4 k;s = 1.12 s 1 Si3 *- , s = 4.10 s 1 • —» »— Si1 k'iS = 2.80 s 1 SMO J(',S = 1.20S1 • • • • • • ' , Si12,11 k'is = 2.33 s 1 • " II Si7,5,6,9 * ' / s =2.71 s 1 y • / Figure 7.34 Intensities of the 1 H / 2 9 S i C P M A S N M R signals indicated as functions of the contact time for 2+2 BENZ-de/PXY-d 4 / ZSM-5 at 293 K. The intensities are normalized to those of the quantitative 2 9 S i spectrum with an equal number of scans. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.3. The fitted theoretical C P maximum k was 9.56 (a.u.) and ki was 183.2 s" 1 ( T i p = 5.5 ms). 209 1.00 0.80 0.60 0.40 0.20 0.00 1.00 0.80 o.eo 0.40 0.20 0.00 Si8 k,s = 3.26 s 1 •""""*~~' Si2 k,s = 5.76 s 1 * ' Si4 fc,s = 2.07 S'1 ^f——— Si3 k,s = 6.75 s' 1 r<, ' Si1 " * / s = 4.95 s ' Si10 * , s = 1.70 s"1 •—•—• m.f=*=l—, , , ^-Figure 7.35 Intensities of the 1 H / 2 9 S i C P drain M A S N M R signals indicated as functions of the contact time for 2+2 B E N Z -d6 /PXY-d 4 /ZSM-5 at 293 K. The points are the experimental values of (AS/So) at chosen contact times and the solid curves are calculated according to Equation 3.6. For the curves corresponding to groups of overlapped peaks, the fit values of C P drain rate constant are the average values as denoted as kis*. Table 7.13 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the C P data. Structural Parameter Step size Minimum Maximum x 0.019 A 0.4 0.55 y 0.02 A 0 0.42 z 0.028 A -0.15 0.075 ^ 9 degrees" -30 120 0 11 degrees 25 150 if/ 11 degrees -60 90 Initial number of locations tested ca. 5.8 x 106 Physically possible locations (dm i n > 2 A) ca. 4.3 x 105 k'|S-M2 linear correlation checked locations r2>0.88 17090 r2 > 0.88 5944 Number of acceptable locations after peak intensity check including overlapped peaks at a r2 > 0.89 5177 given r r2 > 0.90 4215 210 Table 7.14 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the C P drain data. Structural Parameter Step size Minimum Maximum x 0.019 A 0.4 0.55 y 0.02 A 0.1 0.4 Z 0.029 A -0.125 0.05 4> 7 degrees -30 120 9 9 degrees 40 140 if/ 9 degrees -60 60 Initial number of locations tested ca. 3.7 x 106 Physically possible locations (dm i n > 2 A) ca. 4.8 x 105 kiS-fv12 linear correlation checked locations r2 > 0.90 19476 r2>0.90 7216 Number of acceptable locations after peak intensity check including overlapped peaks at a r2 > 0.91 5916 given r r2 > 0.92 4738 Table 7.15 Average structural parameters from the C P data 9 location 0.480(21) 0.247(72) -0.044(26) 45.1(328) 89.8(118) -0.1(216) (r2 Z 0.89) Table 7.16 Average structural parameters from the C P drain data x y z . ^ 9 V Average location (r2^ 0.91) 0.476(22) 0.25(6) -0.052(25) 43.1(311) 90.0(111) 0.0(174) 211 M2{Hz2) M2(Hz2) Figure 7.36 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 B E N Z - d 6 / P X Y - a y Z S M - 5 at 293 K for (a) C P with r2 > 0.89 and (b) C P drain data with r2 > 0.91. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured k)s and kis values are represented by the error bars. (7,5,6,9) -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 2 9Si Chemical Shift (ppm from TMS) J 9 Si Chemical Shift (ppm from TMS) Figure 7.37 NMR spectra for 2+2 BENZ -d6 /PXY-d 4 /ZSM-5 at 270 K; the spectra are (a) the experimental spectrum (contact time = 7.2 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for CP, and (b) the experimental C P drain difference spectrum (AS/So) (contact time = 20 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for C P drain. 212 -30 10 SO 90 45 65 85 105 125 -60 -20 20 60 -30 10 50 90 50 70 90 110 130 -60 -20 20 60 Figure 7.38 Distribution of the solutions determined from the (a) C P data with linear correlations of r2 > 0.88 (red squares), r2 > 0.89 (blue circles), and r2 > 0.90 (green triangles) and (b) C P drain data with linear correlations of r2 > 0.90 (red squares), r2 > 0.91 (blue circles), and r 2 > 0.92 (green triangles) for 2+2 BENZ-d6 /PXY-d 4 /ZSM-5 at 293 K. The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of solutions of the six structural parameters for the translation (x, y, z) of the center of the D C N B (i.e. the center of the benzene ring) in fractional coordinates and the orientation (§, 0, vy) of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with r 2 ^ 0.89 and ? > 0.91 for the C P and C P drain data respectively. Figure 7.39 (a) Scatter Plot of the P X Y - d 4 molecules in the plane of the molecule and (b) its 50% error ellipsoid representation for 2+2 B E N Z -d6 /PXY-d 4 /ZSM-5 from C P at 293 K with r 2 > 0.90. In order to include most data points, a larger area was used (5x7 A 2 ) . 213 7 •T I I I I -j -4 -s -a •) o i a 3 « ! .] .4 .3 .a .1 o i a a 4 5 Figure 7.40 (a) Scatter plot of the P X Y - d 4 molecules in the plane of the molecule and (b) its 50% error ellipsoid representation for 2+2 BENZ-d6 /PXY-d 4 /ZSM-5 by C P drain at 293 K with ? ^ 0.92. In order to include most data points, a larger area was used (5x7 A 2 ) . 7.4.3 Comments on the NMR structures of BENZ-aVPXY-aVZSM-S The structures determined by the C P and C P drain data at the same temperature seem to agree well to each other. In comparison to the 270 K results, the solutions determined from the 293 K data are more diffuse as can be seen from the scatter plots. The final structures presented from the 293 K data are the averages of multiple sets of solutions that satisfy the experimental data. As the 214 numbers of the solutions found are compared for the 270 K and 293 K data, the 270 K results show considerably fewer solutions than at 293 K; i.e., the solutions found from the 270 K data are more localized indicating that at the lower temperature, the P X Y molecules are more highly localized than at the higher temperature where there are larger distributions of acceptable orientations although the average values are the same. 7.5 2+2 BENZ/PXY-c/10/ZSM-5 7.5.1 Structural study of 2+2 BENZ/PXY-d1 0/ZSM-5 at 270 K The location of the B E N Z molecule in the mixed system was probed by both C P and C P drain experiments at 270 K using the 2+2 BENZ/PXY-c / 1 0 /ZSM-5 complex. The C P and C P drain results are presented in Figures 7.42 and 7.43 along with their fits according to Equations 3.3 (for CP) and 3.6 (for C P drain). The experimental points and their fits are in excellent agreement in both cases. In order to determine the structure of the complex, an arbitrary long axis through two opposite hydrogen atoms in the B E N Z molecule (one of the six C 2 axes), was chosen. Due to the nature of the long axis, the B E N Z molecule was expected to show many possible orientations. Tables 7.17 and 7.18, which show the ranges used for the calculations and the solutions found, indicate that there are solutions found in large ranges of the Euler angles, (<(>, 0, y) although locations of the benzene ring center are well defined in both cases. There are good numbers of the solutions found with the r2 values used. The average structural parameters found are listed in Tables 7.19 and 7.20, and are almost identical. The three Euler angles, which are the orientation parameters of the B E N Z molecule, show fairly large uncertainties compared to any previous systems studied. This wide range of orientations could be an indication of possible molecular reorientations of the benzene molecule with respect to its C 6 axis along with its C 2 axes. The linear correlations between the experimental /c',s (/c,s for C P drain) values and the second moments, M2, are presented for both C P and C P drain data in Figure 7.44. The linear correlations for the average structures for C P and C P drain are poorer (r2 values are 0.741 and 0.599 for the C P and C P drain data respectively), indicating many different related orientations are possible for the B E N Z 215 molecule in the framework. However, the predicted spectra, which were calculated from the line functions, ic'ls and icls, are very close to the experimental ones for the both C P and C P drain data as shown in Figure 7.45. The distributions of the six structural parameters for the C P and C P drain results are shown in Figure 7.46. The distributions of the translation parameters, {x, y, z}, show fairly well confined locations for the benzene ring centers for both C P and C P drain. However, as previously mentioned, the Euler angles, which cover all possible orientations for the benzene molecule as - 9 0 ° < <|> < 90° , 0° < 0 < 90° and 0° < \\i < 180°, vary greatly indicating the B E N Z molecule can exist in many different rotationally related orientations according to the same figure: The scatter plots for the solutions in Figures 7.47 and 7.48 also show broad distributions, which look almost continuous thick circular bands, of the protons found as results of the C P and C P drain structure determinations. The final structures of the complex are shown in the average of the found solutions in Figure 7.49. The two locations of the BENZ molecules from the C P and C P drain data are in excellent agreement with both the orientations of the molecules are fairly in large ranges reflecting the Figure 7.42 Intensities of the 1 H / 2 9 S i C P M A S N M R signals indicated as functions of the contact time for 2+2 B E N Z / P X Y -dio/ZSM-5 at 270 K. The intensities are normalized to those of the quantitative 2 9 S i spectrum with an equal number of scans. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.3. The fitted theoretical C P maximum lo was '8.45 (a.u.) and ki was 170.3 s" 1 (T i p = 5.9 ms). wide distributions of the orientational parameters found. 1.00 0.80 0.60 0.40 0.20 0.00 1.00 0.80 0 60 0.40 0.20 0.00 1.00 0.80 0.60 0.40 0.20 0 00 Si8 k'ls = 4.69 S 1 Si2 '•k'iS =4.44 S"1 _-•-•*—• «: 1 Si4 • k'is = 2.49 S 1 i f V , . , _ Si3 k'm =4.25S 1 Si12 - * ' , s = 4.55 S 1 Si11 -/t',s = 4.16 s'1 Si1 k'ls = 5.27 S 1 Si10 ' k'is =4.12 4* . . . . i — Si7,5,6,9 '.{ * ' , s = 4.72 S 1 216 1.00 0.80 0.60 0.40 0.20 0.00 Si2 kls = 4.85 s1 Si10 * , s = 4.49 a'1 Si4 * , s = 2.93 s1 Si3 Si12 k/s = 5.19 s1 k,s = 5.37 s' Si11 k,s = 4.76 s' Si7,5,6,9 «r,s* = 5.68 s' Figure 7.43 Intensities of the 1H/29Si CP drain MAS NMR signals indicated as functions of the contact time for 2+2 BENZ/PXY-oWZSM-5 at 270 K. The points are the experimental values of (AS/So) at chosen contact times and the solid curves are calculated according to Equation 3.6. For the curves corresponding to groups of overlapped peaks, the fit values of CP drain rate constant are the average values as denoted as kiS*. 0 20 40 60 0 20 40 60 20 40 60 Table 7.17 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the CP data. Structural Parameter Step size Minimum Maximum x 0.016 A 0.44 0.55 y 0.015 A 0.15 0.35 z 0.023 A -0.125 0.01 <j> 8 degrees -90 90 9 8 degrees 0 90 y/ 8 degrees 0 180 Initial number of locations tested ca. 4.6 x 10 6 Physically possible locations (d m i n > 2 A) ca. 1.2 x 10" k ' i S -M 2 linear correlation checked locations r2 > 0.80 3317 r2 > 0.80 2609 Number of acceptable locations after peak intensity check including overlapped peaks at a r2>0.81 2046 given r r2 > 0.82 1620 217 Table 7.18 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the C P drain data. Structural Parameter Step size Minimum Maximum x 0.018 A 0.44 0.55 y 0.02 A 0.1 0.4 z 0.032 A -0.15 0.01 $ 7 degrees -90 90 0 8 degrees 0 180 \j/ 8 degrees 0 180 Initial number of locations tested ca. 8.5 x 106 Physically possible locations (dm i n > 2 A) ca. 1.5 x 106 k|S-M2 linear correlation checked locations r2 > 0.80 3400 r2>0.80 2028 Number of acceptable locations after peak intensity check including overlapped peaks at a r2 > 0.81 1622 given r r2>0.82 1379 Table 7.19 Average structural parameters from the C P data 0 JL locator) 0.491(12) 0.250(25) -0.063(13) 1.2(391) 46.5(202) 87.4(452) (^£0 .81) Table 7.20 Average structural parameters from the C P drain data $ 9 y locYuon 0.491(16) 0.250(31) -0.067(16) 1(43) 41.5(212) 86.4(467) ( / a 0.81) 218 (a) 8.0 6.0 - ~ 4.0 2.0 0.0 • 1 1 - V 2 3 1 . - " - " l O 4a A k'is = 1.39x10-5M2 r 2 = 0.745 + 2.98 50000 100000 150000 200000 M2 (Hz 2) (b) 10.0 in 0.0 kis = 1.27x10° M 2 +3.66 r 2 = 0.599 60000 120000 180000 240000 M2 (Hz 2) Figure 7.44 Plots of the measured CP rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 BENZ/PXY-dio/ZSM-5 at 270 K with r2 & 0.93 from (a) CP and (b) CP drain experiments. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured k)s and kis values are represented by the error bars. (7,5,6,9) (7,5,6,9) -110 -112 -114 -116 -118 -120 -110 -112 -114 -116 -118 -120 2 9 S i Chemical Shift (ppm from TMS) 2 9 S i Chemical Shift (ppm from TMS) Figure 7.45 The 2 9 Si NMR spectra for 2+2 BENZ/PXY-di0/ZSM-5 at 270 K; the spectra are (a) the experimental spectrum (contact time = 7.2 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for CP, and (b) the experimental CP drain difference spectrum (AS/So) (contact time = 20 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for CP drain. 219 Figure 7.46 Distributions of the solutions determined from 1 H / 2 9 S i (a) C P and (b) C P drain data of 2+2 BENZ/PXY-d io /ZSM-5 at 270 K with linear correlations of r2 > 0.80 (red squares), r2 > 0.81 (blue circles), and r2 > 0.82 (green triangles). The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of solutions of six structural parameters for the translation (x, y, z) of the center of the D C N B (i.e. the center of the benzene ring) in fractional coordinates and the orientation (<|>, 0, y) of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with r2 > 0.81, which are shown in Tables 6.9 and 10. Figure 7.47 (a) Scatter plot of the B E N Z molecules in the plane of the molecule at r2 > 0.82 and (b) its 50% error ellipsoid representation for 2+2 BENZ/PXY-d io /ZSM-5 from C P at 270 K. 220 Figure 7.49 Final structure of the 2+2 BENZ/PXY-dio/ZSrvl-5 complex at 270 K. The structures are the results of (a) C P and (b) C P drain experiments. 7.5.2 A s t ruc tu re s t u d y of 2+2 B E N Z / P X Y - c f 1 0 / Z S M - 5 at 293 K The structure determination of the 2+2 BENZ /PXY-d 1 0 / ZSM-5 complex at 293 K was carried as before using C P and C P drain experiments. The fits for the C P and C P drain data are in good agreement as seen in Figures 7.50 and 7.51. The experimental kiS and the fitted k, values suggest 221 that the system is in the slow C P regime. Six single Si resonance peaks were resolved in the NMR experiments and used for the initial structure determinations as previously mentioned. The ranges used for the structure calculations and the numbers of the solutions found at different stages are shown in Tables 7.21 and 7.22 for the C P and C P drain data respectively. The solutions found are distributed over wide ranges of the Euler angles indicating the BENZ molecule can adopt many different orientations as shown previously by the 270 K data. The average structural parameters found from the structure determination calculations are listed in Tables 7.23 and 7.24. Figure 7.52 represents the linear correlations between the experimental k'is (kis for C P drain) and the second moments, showing good degrees of linearity for both the C P and C P drain data. The experimental NMR spectra match well in general with the predicted ones that are derived from the average structures for both C P and C P drain (Figure 7.53). The distributions of the structural parameters with respect to the numbers of solutions found are represented in Figure 7.54. The distributions of the translation structure parameters, (x, / , z), and the long axis rotational parameters, (<(>, 0, show relatively good agreement between the C P and C P drain data. The values of the cj) and vy for both the C P and C P drain data are symmetrically distributed, indicating large degrees of symmetrically equivalent solutions involved in the orientations determined. The scatter plots in Figures 7.55 and 7.56 show mirrored distributions of the solutions found confirming that the solutions are symmetry related. The final structures derived from the averages of the structural parameters are presented for the both C P and C P drain results in Figure 7.57. The B E N Z molecules are found in the channel intersections of the framework of ZSM-5 from both the C P and C P drain data. The two structures determined are in excellent agreement with each other, and are also very similar to those previously determined at 270 K. 222 1.00 0.80 0.60 0.40 0.20 0.00 1.00 0.80 0.60 0.40 0.20 0.00 1.00 0 80 0.60 0.40 0.20 0.00 SiS k',s - 2.98 s' Si2 • k;s = 2.84 s1 Si4 -*('« = 1.60 s' SiS A',s = 2.73 s1 jxT Si1 A',5 = 3.36 s1 Si10 k'is = 2.62 s1 Si12,11 fc',5 = 2.76 s' Si7,5,6,9 7 k'ls = 3.07 s1 Figure 7.50 Intensities of the 1 H/ 2 9 Si CP MAS NMR signals indicated as functions of the contact time for 2+2 BENZ/PXY-c/io/ZSM-5 at 293 K. The intensities are normalized to those of the quantitative 2 9 S i spectrum with an equal number of scans. The points are the experimental values of the intensities at chosen contact times and the solid curves are calculated according to Equation 3.3. The fitted theoretical CP maximum h was 10.0 (a.u.) and kt was 170.9 s"1 (7i p = 5.9 ms). 0.30 0.20 0.10 0.00 0.40 . 0.30 0.20 0.10 0.00 0.40 0.30 0.20 0.10 0 00 Si8 k,s =4.35 s'1 Si12,11 kls =4.33 s'1 Si2 Si4 * , s =4,38 s'1 ^ k,s = 2.60 s'1 • ,' Si7,5,6,9 k,s =4.61 s' Si3 Si1 Si10 * , s = 3.89 s"1 * l s = 5.34 s ' k,s= 4.05 s ' . *>' V*' * 0 20 40 60 80 Figure 7.51 Intensities of the 1 H/ 2 9 Si CP drain MAS NMR signals indicated as functions of the contact time for 2+2 BENZ/PXY-dio/ZSM-5 at 293 K. The points are the experimental values of (AS/So) at chosen contact times and the solid curves are calculated according to Equation 3.6. For the curves corresponding to groups of overlapped peaks, the fitted values of CP drain rate constant are the average values as denoted as kis*. 20 40 60 0 20 40 60 223 Table 7.21 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the C P data. Structural Parameter Step size Minimum Maximum x 0.018 A 0.44 0.55 y 0.02 A 0.1 0.4 Z 0.032 A -0.15 0.01 ^ 7 degrees -90 90 0 8 degrees 0 90 y/ 8 degrees 0 180 Initial number of locations tested ca. 4.6 x 10 6 Physically possible locations (dm,„ > 2 A) ca. 8.6 x 10 5 k'i S-M 2 linear correlation checked locations r2 > 0.90 2245 r2 > 0.90 807 Number of acceptable locations after peak intensity check including overlapped peaks at a r2 > 0.91 578 given r r2 > 0.92 392 Table 7.22 Ranges used for the parameters in the structure calculation and numbers of solutions after each step for the C P drain data. Structural Parameter Step size Minimum Maximum 0.55 0.4 0.01 90 90 180 Initial number of locations tested ca. 4.6 x 10 Physically possible locations (d m i n > 2 A) ca. 8.6 x 10 5 k| S -M 2 linear correlation checked locations r2 > 0.90 3854 r2>0.90 1523 Number of acceptable locations after peak intensity check including overlapped peaks at a r2 > 0.91 1069 given r r2 > 0.92 723 x 0.018 A 0.44 y 0.02 A 0.1 z 0.032 A -0.15 $ 7 degrees -90 0 8 degrees 0 if/ 8 degrees 0 224 Table 7.23 Average structural parameters from the C P data $ 9 v tocaUon 0.492(12) 0.251(29) -0.066(16) 1.9(376) 49.9(193) 86.3(450) (/;>0.91) Table 7.24 Average structural parameters from the C P drain data x y z $ 6 Average location ( /£:0.91) 0.488(12) 0.250(25) -0.060(13) 1.6(414) 45.1(215) 87.2(465) 0 50000 100000 150000 200000 0 100000 200000 M2 (Hz2) M2 (Hz2) Figure 7.52 Plots of the measured C P rate constants against the calculated heteronuclear second moments for the average structural solution of 2+2 BENZ/PXY-d io /ZSM-5 at 293 K with r2 > 0.93 from (a) C P and (b) C P drain experiments. The solid line is the line of best fit and the dashed lines represent the 95% confidence prediction intervals. The errors for the measured k)s and kis values are shown by the error bars. 225 -110 -112 -114 -116 -118 -120 ' ' ' 1 1 1 ' ' ' 1 1 , 9 -110 -112 -114 -116 -118 -120 Si Chemical Shift (ppm from TMS) 2 9 S i Chemical Shift (ppm from TMS) Figure 7.53 N M R spectra for 2+2 B E N Z / P X Y - d i 0 / Z S M - 5 at 270 K; the spectra are (a) the experimental spectrum (contact time = 7.2 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for CP , and (b) the experimental C P drain difference spectrum (AS/So) (contact time = 20 ms), the predicted spectrum from the average solution of the structure calculation and the difference between the experimental and predicted spectra in descending order for C P drain. 226 z 0.45 0475 0.5 0.525 0.15 0.2 0.25 0.3 0.35 -0.15 -0.1 -0.05 0 -90 -60 -30 0 30 60 90 0 20 40 60 80 0 60 120 180 Figure 7.54 Distribution of the solutions determined from 1 H / 2 9 S i (a) C P and (b) C P drain data of 2+2 BENZ/PXY-d io /ZSM-5 at 293 K with linear correlations of r2 > 0.90 (red squares), r2 > 0.91 (blue circles), and r2 > 0.92 (green triangles). The vertical axes refer the numbers of solutions and the horizontal axes show the distributions of solutions of the six structural parameters for the translation (x, y, z) of the center of the B E N Z in fractional coordinates and the orientation (o), 0, of the 'long axis' in degrees. The arrows indicate the 'average' values of the six structural parameters with r2 > 0.91, which are shown in Tables 7.23 and 7.24. Figure 7.55 (a) Scatter plot of the B E N Z molecules in the plane of the molecule at r2 > 0.92 and (b) its 50% error ellipsoid representation for 2+2 BENZ/PXY-d io /ZSM-5 from C P at 293 K. -3 -8 -1 0 1 i 3 .j .1 0 1 2 3 Figure 7.56 (a) Scatter plot of the B E N Z molecules in the plane of the molecule at r2 > 0.92 and (b) its 50% error ellipsoid representation for 2+2 BENZ/PXY-d io /ZSM-5 from C P drain at 293 K. 227 Figure 7.57 Final structure of the 2+2 BEN27PXY-d i 0 /ZSM-5 complex at 293 K. The structures are the results of (a) C P and (b) C P drain experiments. 7.5.3 Comments on the NMR structures of BENZ/PXY-c-WZSM-5 Nearly identical structures have been found from the four data sets for BENZ/PXY-d 1 0 /ZSM-5 at two different temperatures. The B E N Z is almost at the center of the channel intersection where the mirror plane lies at y (crystallographic b axis) = %. The solutions are distributed over fairly wide ranges of the structural parameters, especially for the orientational parameters. The orientations of the molecules found are also quite similar to each other with large ranges of uncertainties, indicating possibilities of various molecular orientations of the B E N Z molecule. 228 7.6 2 H static NMR study of 2+2 BENZ/PXY/ZSM-5 at 293 K In order to probe molecular motions in the mixture complex, static 2H NMR experiments were carried out on the three different mixed complex samples, i.e., 2+2 mixtures of BENZ-oyPXY/ZSM-5 , BENZ/PXY-d 4 /ZSM-5 and BENZ/PXY-de/ZSM-5. All the static 2 H quadrupolar echo experiments were performed at room temperature (ca. 293 K). The main purpose of these experiments was to determine translational and other motions of the organic guest molecules. The experimental spectra are presented in Figure 7.58. The possible molecular motions for p-xylene-d 4 are n-flips (180 ° flip) or fast rotation of its benzene ring with respect to its long axis. The p-xylene/ZSM-5 system has been studied previously by Vega and coworkers 1 4 1 , 1 4 2 and more recently by the Fyfe group 2 6 , 9 4 , and this provides a good starting point for understanding the present system. When compared to the various simulated spectra (Figure 7.59), the 2 H NMR spectra of the BENZ/PXY-d 4 /ZSM-5 and BENZ/PXY-de/ZSM-5 systems show that the molecular motions present in the systems are very similar to those of the 2 H static NMR study 9 4 of a low loaded (ca. 3 molecules per unit cell) p-xylene-d 4 /ZSM-5 and p-xylene-de/ZSM-5 in which the p-xylene molecule shows rr-flips and fast methyl rotation at room temperature. From the 2 H results, the PXY molecule in 2+2 BENZ/PXY/ZSM-5 exhibits n-flip motions around the long axis and fast methyl group rotation. The BENZ molecule undergoes rotational motion around its C 6 axis. The results also indicate that there are no translational motions (i.e., free molecular diffusion in the framework) of either component in the mixture that can be detected by the deuterium NMR, at least qualitatively. 229 135 kHz (a) (b) (c) Figure 7.58 Static 2 H spectra of the three different mixtures in Z S M - 5 indicated at 293 K. The spectra were collected using the quadrupolar echo sequence (Chapter 2). The 90 ° pulse was 7 us and the echo time T = 2.5 us. The total number of scans collected was 5120 for each spectrum. No diffusional motion (a sharp isotropic peak at the center of the spectrum) was observed for any mixture at 293 K. Figure 7.59 Simulated 2 H static spectra for (a) static (d4) and (b) TC flipping ring (d 4), (c) fast rotation (d6) of the methyl groups of the partially deuterated p-xylene molecule with respect to the C3 axis, and (d) CB rotation of benzene-d6. The spectra were synthesized as described in Chapter 2. 7.7 1H NOESY study of 2+2 BEN27PXY-C/4/ZSM-5 at 293 K The previous sections of this chapter have provided information about the structures and dynamics of the individual components of the mixture. As each guest component has been probed separately, the correlations between the two components within the framework need to be probed in order to obtain a complete description of the structure as a 'big picture'. A s the 2 H static experiments show no evidence of diffusion or of free translation of the guest molecules within the channel, the 230 guest components are well 'lodged' in the channel system. This is very difficult to obtain from diffraction experiments as they provide an 'average' structure over the whole lattice. However, it is possible to probe this by solid-state NMR spectroscopy. The 1 H NOESY experiment, which measures magnetization transfer between like-spins with time via dipolar interactions (spin exchange), can be used to determine the spatial arrangement of two different spins. In order for 1 H NOESY to be successful, the chemical shift must show a clear separation of the proton resonance peaks. Figure 7.60, the 1 H MAS (spinning at 15 kHz) NMR spectrum of 2+2 BENZ/PXY-d4/ZSM-5 at 293 K shows such a clear separation. Four 2D 1 H NOESY experiments (at the same spinning rate) were carried out at 293 K with different mixing times (20, 50, 100 and 200 ms) and the spectrum at 50 ms is presented in Figure 7.61. Even at a mixing time of 20 ms, spin exchange is detected, indicating a relatively strong interaction between the two sites of 1 H spins, corresponding to the benzene (at ca. 7 ppm) and methyl protons (at ca. 2 ppm). In general, the effective range of NOESY is 5 to 6 A (up to 7 A with methyl protons), and the dipolar coupling between the two protons becomes almost nil beyond that range. Thus, the distribution of the mixture must have the two different molecules close to each other. Figure 7.62 shows the average locations for the guest molecules in the same straight channel, which were determined from the previous structural studies of this chapter. The closest distance between the neighboring protons of BENZ and PXY, determined from these locations, is ca. 4.9 A as indicated. The distances to other guest molecules in the adjacent straight channels can be calculated as well, suggesting that the two molecules in other straight channels are not likely to contribute to the spin diffusion significantly as they are much further away (> 11 A ) . The preceding static 2 H NMR results overrule the possibility of the two components being in direct contact from their translational motions.. Furthermore any motion present would weaken the dipolar-coupling interactions between the nuclei, detrimentally affecting the spin diffusion. The overall results indicate that the two guest components in the 2+2 BENZ/PXY-ayzSM-5 complex are at least not in separate volumes, but rather close to each other, close enough to show the spin diffusion between the two components, and these data fit to models between random and alternating of the two guest distributions in the framework. In order to have a better understanding of the molecular 231 distribution in the mixed system, further modeling of the system and 1 H / 1 H C P experiments may provide complementary structural information between the two components. B E N Z P X Y - d 4 I (I 10 8 6 4 2 0 1H Chemical Shift (ppm) Figure 7.60 1 H M A S spectrum of 2+2 B E N Z / P X Y - d 4 / Z S M - 5 at 293 K. The sample was spinning at 15 kHz, and the two resonance peaks were clearly resolved. The origin of each peak is denoted by the letter codes, B E N Z and P X Y - d 4 . ppm o -2 - 4 - 6 -10 10 8 6 4 2 0 ppm Figure 7.61 2D 1 H N O E S Y spectrum of the 2+2 B E N Z / P X Y - c V Z S M - 5 complex at 293 K with 50 ms mixing time. The sample was spun at 15 kHz. The right figure shows the intensity growth of the methyl N M R peaks (ca. 2 ppm) due to spin diffusion with the different mixing times indicated. 232 Straight channel Distances to the molecules in the adjacent straight channels are > 11A Figure 7.62 NMR structure of the 2+2 B E N Z / P X Y / Z S M - 5 at 293 K showing the B E N Z and P X Y located along the straight channel at adjacent interactions. The structure is viewed from the zigzag channels, and the red dotted lines refer to the path of the straight channel. The structure is the combined results of two independently determined structures from 2+2 BENZ-de/PXY-de/ZSM-S and 2+2 B E N Z / P X Y - d i 0 / Z S M - 5 from the C P drain data at 293 K. The distance (4.9 A) is measured between the closest atoms of the two guest molecules, P X Y and B E N Z . For PXY, it is the C H 3 groups (the blue balls) with the hydrogens located at the center of an equilateral triangle. 7.8 Summary In this chapter I have shown that the NMR structure determination method can be applied successfully even for a mixture of guest organic molecules in ZSM-5. The structures of the mixed system of 2+2 benzene/p-xylene/ZSM-5 were determined at 270 K and 293 K using both C P and C P drain experiments. The results show that all the guest molecules are at the channel intersections in the framework of ZSM-5. According to the 2 H static NMR, the mixed systems are reasonably stable and no fast diffusional motions of the guest molecules are present. The distribution of the two guest components was determined by 1 H N O E S Y spin diffusion experiments, which show clearly that the components are not present in phase-separated volumes and that they are in intimate contact with each other. Further modeling studies should provide more information on the degree of randomness of the distribution. 233 Chapter 8 Summary and Suggestions for Future Work 8.1 Summary In this thesis, the solid-state NMR structure determination strategy developed by our group has been shown to be capable of determining complete and detailed structures for various guest/ZSM-5 complexes, including o-xylene/ZSM-5, where the guest molecule does not have its long axis unlike the previously tested p-xylene type guest molecules. The basis of the structure determination strategy is the use of the distance-dependent dipolar couplings between nuclei on the guest sorbate and the silicons in the framework of the zeolite host. The structures determined by NMR have been verified by diffraction methods. In order to verify the obtained guest/MFI structure, single crystal XRD and powder neutron diffraction were employed as appropriate. The structure of o-xylene/MFI shows that the guest is located at the channel intersection of the framework, the most common location for other guest molecules that have been previously studied. Verification of the structure could not be done by single crystal X R D due to the very low diffusivity of the o-xylene molecules into the large crystals that must be used, preventing a homogeneous distribution of the guest. As an alternative approach to verify the determined structure, we used powder neutron diffraction with success, yielding a structure almost identical to that determined by NMR spectroscopy. Disorder of guest molecules in the framework of ZSM-5 raises important questions on the determination of guest/zeolite complex structure. As seen in two of the systems, the saturated p-dicyanobenzene/ZSM-5 and p-dinitrobenzene/ZSM-5 complexes, NMR did not immediately detect the disorder detected by single crystal XRD. However, NMR is still capable of yielding the structure of the major guest occupancy and gives indications that another occupancy may be present. It is best if as 234 low a loading of guest as possible be used because the more favored site will predominate in the experiment. The structure of a mixture of guests adsorbed into ZSM-5 showed that NMR is a very potent tool of probing these more complex structures. It is very important to note that the NMR spectra of the mixed system are mainly influenced by the presence of the (larger) p-xylene. By using permutated mixtures of pairs of deuterated guest organics, the NMR method was able to determine the structures of each guest in the mixed system in terms of its position and orientation. Further the distribution of the two components in the mixture in the framework can be probed by spin exchange type experiments, giving information about the distribution of the organics within the framework of the mixed system. A diffraction technique, which yields a time and space averaged structure, cannot provide this information. 8.2 Suggestions for future work In order to completely describe the locations of xylenes in MFI, the m-xylene/ZSM-5 structure should be determined by NMR. It is known that m-xylene also has a low diffusivity, which could again pose difficulties in determining the structure by single crystal XRD. If this is the case, powder neutron diffraction could be used to verify the NMR structure. More examples of possible candidates for disordered systems should be explored in order-to collect quantitative information regarding guest disorder in ZSM-5. This would require studies of relationships between the NMR spectra and degrees of disorder as well as comparing single crystal XRD and the NMR structure determinations. Possible examples would be a guest/zeolite complex system with its guest organic having strong electronegative substituents such as cyano and nitro groups. It would also be of interest to extend the NMR method to different zeolite frameworks to determine guest/zeolite complex structures. A suitable candidate would be purely siliceous highly crystalline ferrierite, which is also available as large crystals so it should be possible to verify the NMR 235 structures by single crystal XRD. It is important to gather more reliable structures of guest/zeolite complexes so that the validity of the NMR method can be firmly established in order to apply it to different systems, where single crystal XRD is not viable. In the case of mixed guest/zeolite systems, a higher loaded mixture of benzene/p-xylene in ZSM-5 should be studied to probe the interactions between the two guest components in an even more complex environment. Diffraction studies of the system could be attempted; however, to minimize the effects of motions and disorders of the guests, it would be logical to lower the temperature considerably (170 K) during the data collection, which could lead to a phase change of the complex crystal. A modeling study of the complex has potential to understand the interactions between the two guests and with the framework to explain the stability of the mixed system studied here. Other combinations of mixtures could also be tested with the NMR structure determination strategy. Such a stability of a mixture/zeolite complex can be useful for several industrial applications. Especially, for the procedures that involve materials that are volatile or hazardous, understanding the interactions could result in novel methods and applications. One such candidate would be a gas storage material, which is required to hold a large amount of gas in a stable and safe environment. With development of a stable gas/zeolite and its combinatorial complexes, zeolites are ideal media in many aspects as they are mostly empty and chemically stable with considerable physical integrity. 236 References 1. Baerlocher, C ; Olson, D. H.; Meier, W. M., Atlas of Zeolite Framework Types. 5 t h Edition ed.; Elsevier: Amsterdam, 2001; 5 t h Edition. 2. Fyfe, C. A., Solid State NMR for Chemists. C .F .C. Press: Guelph, Canada, 1983; 3. van Bekkum, H.; Flanigen, E. M.; Jansen, J . C , Introduction to Zeolite Science and Practice. Elsevier: Amsterdam, 1991. 4. Jansen, J . O ; Stocker, M.; Karge, H. G ; Weitkamp, J . , Advanced Zeolite Science and Applications. Elsevier: Amsterdam, 1994. 5. Chon, H.; Woo, S. I.; Park, S. E., Recent Advances and New Horizons in Zeolite Science and Technology. Elsevier: Amsterdam, 1996; Vol. 102. 6. Scott, J . , Zeolite Technology and Applications: Recent Advances. Noyes Data Corporation: Park Ridge, N J , 1980. 7. Pujabo, P. R.; Rabo, J . A.; Antos, G. J . ; Gembicki, S. A., Catal. Today 1992, 73, 113. . 8. Venuto, P. B., Microporous Mater. 1994, 2, 297. 9. Breck, D. W., Zeolite Molecular Sieves. Wiley Interscience: 1974; 10. Barrer, R. M., Nature 1949, 164, 112. 11. Vaughan, D. E. W., The properties and Applications of Zeolites. Spec. Publ. Chem. S o c : London, 1980; p 33 12. Smith, J . V., Zeolites Chemistry and Catalysis. In Zeolite Chemistry and Catalysis, Rabo, J . A., Ed . American Chemical Society: Washington DC, 1976; Vol. 171. 13. Chen, N. Y ; Degnan, T. R; C M . , S. , Molecular Transport and Reaction in Zeolites. Design and Application of Shape Selective Catalysis. V C H : New York, 1994; 14. Wong, L. W.; Sun, W.; Lam, K. F ; Yeung, K. L. In Technical Proceedings of the 2006 NSTI Nanotechnology Conference and Trade Show, Volume 1, 2006; Nano Science and Technology Institute 2006; pp 31. 15. Mintova, S. ; Olson, N. H.; Valtchev, V.; Bein, T , Science 1999, 283, 958. 16. Ozin, G A. , J. Am. Ceram. Soc. 1992, 4, 612. 17. Havenga, E. A.; Huang, Y , Langmuir2002, 78, 6907. 18. Kokotailo, G. T ; Lawton, S. L ; Olson, D. H.; Meier, W. M., Nature 1978, 272, 437. 19. van Koningsveld, H., Acta Cryst. 1990, B46, 731. 20. van Koningsveld, H.; Jansen, J . C ; van Bekkum, H., Zeolites 1990, 10, 235. 21. Lewis, J . E.; Freyhardt, C. C ; Davis, M. E., J. Phys. Chem. 1996, 700, 5039. 22. Wiegel , S . J . ; Gabriel, J . - C ; Peublo, E. G ; Bravo, A . M.; Henson, N. J . ; Bull, L. M.; Cheetham, A . K., J. Am. Chem. Soc. 1996, 778, 2427. 23. van Koningsveld, H.; van Bekkum, H.; Jansen, J . C , Acta Cryst. 1987, B43, 127. 237 24. Brouwer, D. H.; Kristiansen, P. E.; Fyfe, C. A.; Levitt, M. H., J. Am. Chem. Soc. 2005, 727, 542. 25. Fyfe, C . A. ; Brouwer, D. H., J. Am. Chem. Soc. 2006, 128, 11860. 26. Fyfe, C. A ; Diaz, A. C ; Grondey, H.; Lewis, A. R.; Forster, H., J. Am. Chem. Soc. 2005, 127, 7543. 27. Fyfe, C. A.; Feng, Y ; Grondey, H.; Kokotailo, G. T.; Mar, A., J. Phys. Chem. 1991, 95, 3747. 28. Fyfe, C. A.; Gies, H.; Kokotailo, G. T.; Marler, B.; Cox, D. E., J. Phys. Chem. 1990, 94, 3718. 29. Fyfe, C. A . ; Grondey, H.; Feng, Y ; Kokotailo, G. T , J. Am. Chem. Soc. 1990, 112, 8812. 30. Fyfe, C. A.; Grondey, H.; Feng, Y ; Kokotailo, G. T , Chem. Phys. Lett. 1990, 173, 211. 31. Fyfe, C. A.; Grondey, H.; Feng, Y ; Kokotailo, G. T.; Ernst, S. ; Weitkamp, J . , Zeolites 1992, 12, 50. 32. Fyfe, C. A.; Kennedy, G. J . ; Schutter, C. T. D.; Kokotailo, G. T., J. Chem. Soc, Chem. Commun. 1984, 541. 33. Fyfe, C. A . ; O'Brien, J . H.; Strobl, H., Nature 1987, 326, 281. 34. Klinowski, J . ; Ramdas, S. ; Thomas, J . M.; Fyfe, C. A.; Hartman, J . S. , J. Chem. Soc, Faraday Trans. 1982, 78, 1025. 35. Engelhardt, G ; Lohse, U.; Lippmaa, E.; Tarmak, M.; Magi, M., Z. Anorg. Allg. Chem. 1981, 482, 49. 36. Ramdas, S. ; Thomas, J . M.; Klinowski, J . ; Fyfe, C. A.; Hartman, J . S. , Nature 1981, 292, 228. 37. Fyfe, C. A.; Strobl, H.; Kokotailo, G. T ; Kennedy, G. J . ; Barlow, G. E., J. Am. Chem. Soc. 1988, 110, 3373. 38. Chang, C. D.; Silvestri, A . J . , J. Catal. 1977, 47, 249. 39. Haag, W. O.; Olson, D. H. US Patent no. 3856871. 1974. 40. van Koningsveld, H.; Tuinstra, F ; van Bekkum, H.; Jansen, J . C , Acta Cryst. 1989, B45, 423. 41. van Koningsveld, H.; Jansen, J . C ; De Man, A . J . M., Stud. Surf. Sci. Catal. 1995, 98, 61. 42. van Koningsveld, H.; Jansen, J . C ; de Man, A . J . M., Acta Cryst. 1996, B52, 131. 43. van Koningsveld, H.; Jansen, J . C ; van Bekkum, H., Acta Cryst. 1996, S52 , 140. 44. van Koningsveld, H.; Jansen, J . C , Microporous Mater. 1996, 6, 159. 45. Fyfe, C. A.; Brouwer, D. H., Can. J. Chem. 2006, 84, 345. 46. Fyfe, C. A.; Brouwer, D. H.; Tekely, P., J. Phys. Chem. A 2005, 709, 6187. 47. Mentzen, B. F ; Gelin, P., Mater. Res. Bull. 1995, 30, 373. 48. Mentzen, B. F.; Lefebvre, F , Mater. Res. Bull. 1997, 32, 813. 49. Mentzen, B. F ; Lefebvre, F , J. Chim. Phys. 1998, 95, 1052. 50. Taylor, J . C , J. Chem. Soc, Chem. Commun. 1987, 1186. 51. Nair, S. ; Tsapatsis, M., J. Phys. Chem. B 2000, 704, 8982. 52. Nair, S. ; Tsapatsis, M., J. Phys. Chem. B 2001, 705, 1276. 238 53. Goyal , R.; Fitch, A. N.; Jobic, H., J. Phys. Chem. B 2000, 104, 2878. 54. Sacerdote-peronnet, M.; Mentzen, B. F , Mater. Res. Bull. 1993, 28, 767. 55. Abragam.A. , The Principles of Nuclear Magnetism. Oxford Universi tyPress: 1961; 56. Slichter, C. P., Principles of Magnetic Resonance. 2nd ed.; Springer-Verlag: New York, 1983; 2nd. 57. Ernst, R. R.; Bodenhausen, G.; Wokaun, A., Principles of Nuclear Magnetic Resonance in One and Two Dimensions. Clarendon Press: Oxford, 1987; 58. Harris, R. K., Nuclear Magnetic Resonance Spectroscopy: A Physicochemical View. 3rd revised ed.; Longmann Scientifc and Technical: United Kingdom, 1987; 3rd revised. 59. Sanders, J . ; B., H., Modern NMR Spectroscopy, A Guide for Chemists. Oxford University Press: 1987; 60. Derome, A. E., Modern NMR Techniques for Chemistry Research. Pergamon Press: New York, 1987; 61. Gunther, H., NMR Spectroscopy - Basic Principles, Concepts, and Applications in Chemistry. 2nd ed.; John Wiley and Sons: New York, 1995; 2nd. 62. Levitt, M. H., Spin Dynamics. Basics of Nuclear Magnetic Resonance. Wiley: Chichester, UK, 2001; 63. Duer, M. J . , Introduction to Solid-State NMR Spectroscopy. Blackwell Publishing Ltd: Oxford, 2004; 64. Pines, A.; Gibby, G.; Waugh, J . S. , J. Chem. Phys. 1973, 59, 569. 65. Andrew, E. R.; Bradbury, A.; Eades, R. G , Nature 1958, 182, 1659. 66. Andrew, E. R.; Bradbury, A.; Eades, R. G , Nature 1959, 783, 1802. 67. Lowe, I. J . , Phys. Rev. Lett. 1959, 2, 285. 68. Lewis, A . Location of guest species in zeolites by solid-state NMR. Ph.D. Thesis, University of British Columbia, Vancouver, 1998. 69. Yannoni, C. S., Acc. Chem. Res. 1982, 15, 201. 70. van Vleck, J . H., Phys. Rev. 1948, 74, 1168. 71. Gutowsky, H. S.; Pake, G. E., J. Chem. Phys. 1950, 18, 162. 72. Michel, J . ; Drifford, M.; Rigny, P., J. Chim. Phys. 1970, 67, 31. 73. Pake, G. E., J. Chem. Phys. 1948, 16, 327. 74. Gullion, T.; Schaefer, J . , J. Magn. Reson. 1989, 81, 196. 75. Hing, A . W.; Vega, S. ; Schaefer, J . , J. Magn. Reson. 1992, 96, 205. 76. Fyfe, C. A.; Lewis, A. R.; C h e z e a u . J . M., Can. J. Chem. 1999, 77, 1984. 77. Fyfe, C. A.; Brouwer, D. H.; Lewis, A. R.; Chezeau, J . M., J. Am. Chem. Soc. 2001, 123, 6882. 78. Hediger, S . Improvement of Heteronuclear Polarization Transfer in Solid-State NMR. Ph.D. Dissertation, Eidgenossishe Technische Hochschule, Zurich, Switzerland, 1997. 79. Fyfe, C. A.; Feng, Y ; Gies, H.; Grondey, H.; Kokotailo, G T , J. Am. Chem. Soc. 1990, 112, 3264. 239 80. Bretherton, J . Ph.D. Thesis, University of British Columbia, Vancouver, B C , 2002. 81. Fyfe, C. A.; Brouwer, D. H., Microporous Mesopdrous Mater. 2000, 39, 291. 82. Bax, A.; Freeman, R.; Kempsell , S. P., J. Am. Chem. Soc. 1980, 102, 4849. 83. International Tables for Crystallography. 4th revised edition ed.; Reidel Publishing Company: Dordrecht, Boston, 1996; Vol. A, 4th revised edition. 84. Tegze, M.; Faigel, G., Nature 1996, 380, 49. 85. Fitzgerald, R., Phys. Today 2001, 54, 21. 86. Pecharsky, V. K.; Zavalij, P. Y , Fundamentals of Powder Diffraction and Structural Characterization of Materials. Springer Science+Business Media, Inc.: New York, 2005; 87. Clegg, W., Crystal Structure Determination. Oxford University Press Inc.: New York, 1998; 88. Materials Science and Engineering Laboratory, the Special Feature section of neutron scattering lengths and cross sections of the elements and their isotopes. In Neutron News, National Institute of Standards and Technology: 1992; Vol. 3, pp 29. 89. C N B C / N R C http://neutron.nrc-cnrc.qc.ca/specs e.html. Accessed on 2007-01-15, (Date modified: 2002-11-01). 90. Rietveld, H. M., Acta Cryst. 1967, 22, 151. 91. Rietveld, H. M., J. Appl. Crystallogr. 1969, 2, 65. 92. van Koningsveld, H., J. Mol. Catal. A: Chem. 1998, 734, 89. 93. van Koningsveld, H.; Koegler, J . H., Microporous Mater. 1997, 9, 71. 94. Diaz, A. Investigation of the three-dimensional structures of zeolite molecular sieves by high resolution solid state NMR. Ph.D. Thesis, University of British Columbia, Vancouver, 1998. 95. Werstiuk, N. H. K., T , Can. J. Chem. 1973, 51, 1485. 96. Chmurny, G ; Hoult, D., Concepts in Magnetic Resonance 1990, 2, 131 97. Sodickson, A.; Cory, D. G., J. Magn. Reson. 1997, 128, 87 98. Agaskar, P. A., Inorg. Chem. 1990, 29, 1603. 99. Hoebbel, D.; Wieker, W., Z. Anorg. Allg. Chem. 1971, 384, 43. 100. Bloom, M.; Davis, J . M.; Valic, M. I., Can. J. Phys. 1980, 58, 1510. 101. Macho, V.; Brombacher, L.; Spiess, H. W., Appl. Magn. Reson. 2001, 20, 405. 102. Wolfram, S. Mathematica: A System for Doing Mathematics by Computer, 3.0; Wolfram Media: Champaign IL, 1996. 103. Brouwer, D. H. Location, Disorder, and Dynamics of Guest Species in Zeolite Frameworks Studied by Solid State N M R and X-ray Diffraction. Ph . D. Dissertation, University of British Columbia, Vancouver, Canada, 2003. 104. Sheldrick, G. M. SHELX-97 Programs for Crystal Structure Analsysis, 97-2; Institute fur Anorganische 240 Chemie der Universitat, Tammanstrasse 4: D-3400 Gottingen, Germany, 1998. 105. Farrugia, L. J . , J. Appl. Crystallogr. 1999, 32, 837. 106. Shirley, R. "The Crysfire 2002 System for Automatic Powder Indexing: User's Manual", The Lattice Press: 41 Guildford Park Avenue, Guildford, Surrey G U 2 7NL, England, 2002. 107. Fyfe, C. A.; Lee, J . S . J . , J. Phys. Chem. B 2007, submitted. 108. Favre-Nicolin, V.; Cerny, R., J. Appl. Crystallogr. 2002, 35, 734. 109. Favre-Nicolin, V.; Cerny, R., Z. Kristallogr. 2004, 219, 847. 110. Larson, A. C ; Dreele, R. B. V., Los Alamos National Laboratory Report LAUR 1994, 86-748. 111. Toby, B..H., J. Appl. Crystallogr. 2001, 34, 210. 112. Brouwer, D. H.; Darton, R. J . ; Morris, R. E.; Levitt, M. H., J. Am. Chem. Soc. 2005, 127, 10365. 113. Corma, A.; Rey, F.; Valencia, S. ; Jorda, J . L.; Rius, J . , Nat. Mater. 2003, 2, 493. 114. Fyfe, C. A.; Lee, J . S . J . ; Swainson, I. P.; Cranswick, L. M. D., The o-Xylene/MFI Complex: Structure determination by Solid State N M R and its Verification by Powder Neutron Diffraction Experiments. In The 47th ENC Conference, Pacific Grove, California, 2006. 115. Kaszkur, Z. A.; Jones, R. H.; Bell, R. G.; Catlow, C. R. A. ; Thomas, J . M., Mol. Phys. 1996, 89, 1345. 116. Fyfe, C. A.; Brouwer, D. H., J.Am. Chem. Soc. 2004, 126, 1306. 117. Brouwer, D. H., J. Magn. Reson. 2003, 164, 10. 118. Hartmann, S. R.; Hahn, E. L , Phys. Rev. 1962, 728, 2042! 119. Klur, I.; Jacquinot, J . - F ; Brunet, F ; Charpentier, T ; Virlet, J . ; Schneider, C ; Tekely, P., J. Phys. Chem. B 2000, 704, 10162. 120. Schaefer, J . ; McKay, R. A.; Stejskal, E. O., J. Magn. Reson. 1979, 34, 443. 121. Schaefer, J . ; Stejskal, E. O.; Garbow, J . R.; McKay, R. A., J. Magn. Reson. 1984, 59, 150. 122. Stejskal, E. O. S. , J . ; McKay, R. A., J. Magn. Reson. 1984,, 57,, 471. 123. Chase, W.; Brown, F , General Statistics. 2nd ed.; John Wiley & Sons: 1992; 2nd. 124. McCusker, L. B., Stud. Surf. Sci. Catal. 2004, 754, 41. 125. Wilkinson, A.; von Dreele, R.; Proffen, T ; Swainson, I.; Cranswick, L. In 6th Canadian Powder Diffraction Workshop (May 8 - 1 0 , 2006), University of Waterloo, Waterloo, Ontario, 2006; University of Waterloo, Waterloo, Ontario, 2006. 126. Werner, P.-E.; Eriksson, L ; Westdahl, M., J. Appl. Crystallogr. 1985, 78, 108 127. Boultif, A.; Louer, D . J . , J. Appl. Crystallogr. 1991, 24, 987 128. Visser, J . W., J. Appl. Crystallogr. 1969, 2, 89 129. McGreevy, R. L ; Pusztai, L., Mol. Simul. 1988, 7, 359. 241 130. Metropolis, N.; Rosenbluth, A . W.; Rosenbluth, M. N.; Teller, A. H.; Teller, E., J. Chem. Phys. 1953, 21, 1087. 131. Kirkpatrick, S. ; Gelatt, C. D.; Vecchi, M. P., Science 1983, 220, 671. 132. Cerny, V., Journal of Optimization Theory and Applications 1985, 45, 41. 133. Falcioni, M.; Deem, M. W., J. Chem. Phys. 1999, 110, 1754. 134. Breck, D. W., Zeolite Molecular Sieves. Academic Press: London, 1978; 135. Barrer, R. M., Zeolites and Clay Minerals as Sorbents and Molecular Sieves. Wiley: New York, 1978; 136. Fyfe, C. A.; Lewis, A . R.; Schweiger, W., private communication. 137. Olson, D. H.; Kokotailo, G. T ; Lawton, S. L ; Meier, W. M., J. Phys. Chem. 1981, 85, 2238. 138. Richards, R. E.; Rees, L. V. C „ Zeolites,1988, 8, 35. i 139. Gorring, R. L., Unpublished studies. In Mobil Research and Development Corp.: Princeton, NJ 08540. 140. Nagy, J . B.; Derouane, E. G.; Resing, H. A.; Miller, G. R., J. Phys. Chem. 1983, 87, 833. 141. Kustanovich, I.; Fraenkel, D.; Luz, Z.; Vega, S. ; Zimmermann, H., J. Phys. Chem. 1988, 92, 4134. 142. Kustanovich, I.; Vieth, H. M.; Luz, Z.; Vega, S. , J. Phys. Chem. 1989, 93, 7427. 143. Fyfe, C. A.; Gies, H.; Feng, Y.; Kokotailo, G. T , Nature 1989, 341, 223. 144. McCusker, L. B.; Von Dreele, R. B.; Cox, D. E.; Louer, D.; Scardi, P., J. Appl. Crystallogr. 1999, 32, 36. 145. Fyfe, C. A.; Strobl, H.; Gies, H.; Kokotailo, G. T , Can. J. Chem. 1988, 66, 1942. 146. Fyfe, C. A.; Diaz, A . C , J. Phys. Chem. B 2002, 106, 2261. 242 Appendix A Supplementary information for the NMR structures The following tables contain the atomic fractional coordinates and error ellipsoid parameters of the organic guests found from NMR structure determination. The error ellipsoid parameters are analogous to thermal ellipsoid parameters in diffraction data, illustrating the uncertainty of the found locations. Further detailed descriptions on how to generate the error ellipsoid can be found in the references: Fyfe, C. A.; Brouwer, D. H., J. Am. Chem. Soc. 2006, 128, 11860; Brouwer, D. H. Location, Disorder, and Dynamics of Guest Species in Zeolite Frameworks Studied by Solid State N M R and X-ray Diffraction. Ph . D. Dissertation, University of British Columbia, Vancouver, Canada, 2003. A.1 o - X y l e n e / Z S M - 5 s y s t e m Table A.1 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-cfe molecule for the average location in Z S M - 5 with r2 £ 0.92 determined from the 1 H / 2 9 S i C P at 273 K X y z e« E22 E33 En E13 E23 m 0.4068 0.3359 -0.1362 0.057 0.335 0.072 0.087 -0.005 0.078 H2 0.4765 0.4085 -0.0524 0.188 0.077 0.166 0.038 0.135 0.032 H3 0.5589 0.3669 0.0503 0.087 0.291 0.064 -0.079 0.047 -0.091 H4 0.5716 0.2529 0.0693 0.075 0.344 0.059 0.099 0.034 0.129 D7 0.3929 0.1942 -0.1494 0.291 0.466 0.259 -0.202 0.006 -0.230 D8 0.5069 0.1368 -0.0073 0.514 0.073 0.432 0.039 0.238 0.089 C1 0.4390 0.3197 -0.0960 0.032 0.158 0.042 0.036 -0.010 0.034 C2 0.4814 0.3638 -0.0450 0.070 0.065 0.065 0.021 0.049 0.020 C3 0.5316 0.3385 0.0175 0.034 0.146 0.026 -0.020 0.021 -0.020 C4 0.5394 0.2692 0.0291 0.042 0.163 0.034 0.043 0.014 0.065 C5 0.4970 0.2251 -0.0219 0.134 0.056 0.115 0.012 0.040 0.036 C6 0.4468 0.2504 -0.0845 0.087 0.131 0.084 -0.037 -0.012 -0.035 C7 0.4022 0.2036 -0.1382 0.247 0.390 0.222 -0.166 0.001 -0.187 C8 0.5050 0.1518 -0.0100 0.431 0.068 0.363 0.032 0.193 0.077 2 4 3 Table A.2 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-cfe molecule for the average location in ZSM-5 with r2 £ 0.92 determined from the 1 H / 2 9 S i C P at 315 K X y z E22 E M E12 E13 E23 H1 0.4043 0.3074 -0 .1603 0.036 0 .235 0.042 0.009 -0 .028 0 .025 H 2 0.4563 0.3989 -0 .0900 0.075 0.103 0 .093 0.054 0.045 0.035 H 3 0.5382 0.3837 0.0286 0.073 0.153 0.089 -0.014 0.011 -0 .057 H 4 0.5682 0.2771 0.0768 0.019 0.239 0.053 0.073 -0 .009 0.020 D7 0.4144 0.1658 -0 .1389 0.324 0.200 0.373 -0.134 -0.112 -0.081 D8 0.5277 0.1449 0.0250 0.301 0.111 0.399 0.102 -0 .016 0.086 C1 0.4364 0 .3015 -0 .1139 0.027 0.134 0 .035 0.006 -0.021 0.016 C 2 0.4680 0.3571 -0.0711 0.027 0.087 0.037 0.030 0 .017 0.017 C 3 ' 0 .5179 0.3479 0.0011 0.024 0.106 0.037 0.012 0 .007 -0 .017 C 4 0.5361 0.2830 0.0304 0.017 0.136 0.042 0.045 -0 .009 0.013 C 5 0 .5045 0.2274 -0.0124 0.091 0.080 0.130 0 .035 -0 .025 0.028 C 6 0.4546 0.2366 -0 .0845 0.098 0.097 0.123 -0.022 -0 .047 -0.004 C 7 0 .4213 0.1778 -0 .1295 0.276 0.176 0.320 -0 .110 -0 .099 -0.064 C 8 0 .5236 0.1588 0.0184 0.256 0.103 0.342 0.087 -0 .020 0.073 Table A.3 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-c/4 molecule for the average location in ZSM-5 with r2 2 0.72 determined from the 1 H / 2 9 S i C P at 273 K X y z E„ E22 E33 E12 E13 E23 D1 0 .4015 0 .3315 -0.1548 0.107 0.048 0.098 0.034 -0 .034 0.014 D 2 0.4533 0 .4099 -0 .0552 0.152 0.026 0.177 -0 .002 -0.031 0.016 D 3 0 .5373 0.3776 0.0519 0.105 0.045 0.166 . -0.051 -0.031 -0 .009 D4 0 .5696 0.2670 0.0595 0.037 0.057 0.116 -0.040 -0 .035 0.001 H 7 0 .4139 0.1908 -0 .1852 0.051 0.051 0.026 0.013 -0.011 -0 .019 H 8 0.5301 0.1462 -0.0371 0.025 0.030 0.077 -0 .015 -0 .014 0.008 C1 0.4344 0.3188 -0.1128 0.075 0.031 0.078 0 .019 -0 .015 0 .005 C 2 0 .4659 0.3665 -0 .0523 0.099 0.024 0.123 -0.002 -0 .017 0 .006 C 3 0 .5170 0.3469 0.0129 0.073 0.032 0.122 -0 .029 -0 .017 -0 .005 C 4 0 .5367 0.2796 0.0176 - 0.032 0.037 0.089 -0 .026 -0 .016 -0 .003 C 5 0 .5052 0.2319 -0 .0430 0.015 0.025 0.050 -0 .008 -0.001 -0 .003 C 6 0.4540 0.2515 -0.1082 0.032 0.027 0.037 0 .010 -0.001 -0 .006 C 7 0 .4209 0.2010 -0.1720 0.045 0.045 0.025 0.014 -0 .008 -0 .016 C 8 0 .5257 0.1608 -0 .0383 0 .019 0.029 0.069 -0 .014 -0 .010 0 .005 Table A.4 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-04 molecule for the average location in Z S M - 5 with r2 £ 0.72 determined from the 1 H / 2 9 S i C P at 315 K X y z E„ E22 E33 E,2 E13 E23 D1 0.4182 0.3188 -0 .1605 0.191 0.064 0.132 0.047 -0 .098 -0 .036 D 2 0.4664 0.3965 -0 .0558 0.229 0.079 0.181 -0 .013 -0.061 -0 .037 D 3 0.5462 0.3634 0 .0579 0.158 0.189 0.125 -0 .097 -0.048 -0 .010 D 4 0 .5777 0.2525 0.0668 0.144 0.233 0.088 -0.082 -0 .076 0 .076 H 7 0.4314 0.1783 -0 .1916 0.088 0.070 0 .038 0.030 -0.032 -0 .059 H 8 0.5416 0.1325 -0 .0345 0.146 0.138 0.071 -0.021 -0 .020 0.073 C1 0.4494 0 .3059 -0.1161 0.121 0.064 0.081 0.026 -0 .050 -0 .038 C 2 0.4788 0.3531 -0 .0523 0.145, 0.085 0.110 -0.011 -0 .039 -0 .035 C 3 0 .5273 0.3329 0.0169 0.114 0.145 0.085 -0.058 -0 .032 -0.011 C 4 0 .5465 0.2655 0.0223 0.092 0.167 0.053 -0 .053 -0 .037 0.030 C 5 0,5172 0.2183 -0.0414 0.062 0.112 0.023 -0.014 -0 .010 0.013 C 6 0.4687 0.2385 -0 .1106 0.058 0.070 0.024 0.018 -0 .016 -0.032 C 7 0.4378 0.1884 -0 .1777 0.079 0.067 0.032 0.030 -0.027 -0.054 C 8 0.5373 ' 0.1471 -0 .0359 0.124 0.133 0.057 -0 .020 -0 .016 0.061 Table A.5 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-cfe molecule for the average location in ZSM-5 with r2 > 0.72 determined from the 1 H / 2 9 S i C P drain at 273 K X y z E« E33 E12 E13 E23 Hi 0.4145 0.3131 -0.1789 0.040 0.122 0.034 0.012 -0.027 0.015 H2 0.4664 0.4049 -0.1094 0.054 0.046 0.052 0.030 0.038 0.018 H3 0.5373 0.3914 0.0245 0.044 0.076 0.056 -0.031 0.017 -0.030 H4 0.5564 0.2861 0.0889 0.027 0.126 0.017 -0.005 -0.010 0.040 D7 0.4137 0.1727 -0.1411 0.295 0.102 0.204 -0.048 -0.120 -0.071 D8 0.5118 0.1541 0.0441 0.292 0.052 0.166 0.040 -0.053 0.062 C1 0.4423 0.3078 -0.1265 0.030 0.063 0.027 0.007 -0.020 0.010 C2 0.4739 0.3637 -0.0842 0.018 0.036 0.025 0.013 0.016 0.011 C3 0.5170 0.3555 -0.0027 0.013 0.047 0.024 -0.012 0.011 -0.004 C4 0.5286 0.2914 0.0365 0.021 0.065 0.016 -0.004 -0.009 0.025 C5 0.4971 0.2356 -0.0058 0.095 0.032 0.056 0.006 -0.034 0.025 C6 0.4539 0.2438 -0.0873 0.098 0.041 0.066 -0.008 -0.050 -0.005 C7 0.4206 0.1847 -0.1317 0.253 0.088 0.175 -0.039 -0.106 -0.056 C8 0.5092 0.1679 0.0354 0.250 0.047 0.143 0.032 -0.051 0.054 Table A.6 Atomic fractional coordinates and error ellipsoid parameters of the o-xylene-d6 molecule for the average location in Z S M - 5 with r2 s 0.82 determined from the 1 H / 2 9 S i C P drain at 315 K X y z £« £ 22 £33 E13 £ 2 3 H1 0.4128 0.3389 -0.1521 0.043 0.157 0.057 0.067 -0.005 0.025 H2 0.4619 0.4059 -0.0328 0.078 0.054 0.122 0.021 0.052 -0.001 H3 0.5256 0.3583 0.0917 0.044 0.187 0.044 . -0.016 0.013 -0.051 H4 0.5403 0.2439 0.0969 0.072 0.210 0.083 0.089 -0.006 0.062 D7 0.4086 0.1989 -0.1925 0.407 0.251 0.194 -0.104 -0.128 -0.049 D8 0.4967 0.1332 -0.0204 0.520 0.084 0.385 0.004 -0.054 0.108 C1 0.4377 0.3203 -0.1034 0.037 0.087 0.039 0.034 -0.010 0.015 C2 0.4676 0.3610 -0.0308 0.024 0.053 0.053 0.021 0.023 0.001 C3 0.5064 0.3321 0.0449 0.015 0.109 0.027 0.009 0.009 -0.013 C4 0.5153 0.2625 0.0482 0.055 0.120 0.055 0.047 -0.010 0.038 C5 0.4854 0.2218 -0.0244 0.174 0.065 0.120 0.013 -0.038 0.047 C6 0.4466 0.2507 -0.1002 0.147 0.088 0.079 -0.011 -0.052 0.010 C7 0.4152 0.2076 -0.1766 0.354 0.213 0.170 -0.084 -0.113 -0.035 C8 0.4947 0.1483 -0.0213 0.447 0.080 0.328 0.006 -0.052 0.096 A.2 p-Dicyanobenzene/ZSM-5 system Table A.7 Atomic fractional coordinates and error ellipsoid parameters of the p-dicyanobenzene molecule for the average location in 4 D C N B / Z S M - 5 with r2 > 0.93 determined from the 1 H / 2 9 S i C P at 305 K X y z E„ E22 £ 3 3 £ « E13 E23 H2 0.4052 0.2075 -0.1254 0.071 0.664 0.128 0.174 -0.017 0.020 H3 0.4254 0.0902 -0.1102 0.167 0.411 0.176 -0.229 0.135 -0.233 H5 0.5676 0.1268 0.1096 0.030 0.525 0.082 0.143 0.008 0.053 H6 0.5475 0.2442 0.0945 0.161 0.337 0.185 -0.199 0.120 -0.190 C1 0.4747 0.2354 -0.0167 0.109 0.092 0.177 0.046 0.059 0.014 C2 0.4392 0.1906 -0.0762 0.037 0.280 0.057 0.059 -0.008 -0.008 C3 0.4509 0.1224 -0.0674 0.066 0.186 0.066 , -0.085 0.048 -0.094 C4 0.4981 0.0989 0.0009 0.089 0.054 0.144 0.010 0.082 0.008 C5 0.5336 0.1437 0.0605 0.013 0.199 0.030 0.041 0.007 0.011 C6 0.5219 0.2119 0.0517 0.062 0.143 0.071 -0.068 0.040 -0.070 N1 0.4530 0.3616 -0.0330 0.744 0.236 1.265 0.320 0.527 0.211 N2 0.5198 -0.0273 0.0172 0.687 0.128 1.173 0.219 0.592 0.195 C7 0.4626 0.3061 -0.0258 0.392 0.160 0.661 0.171 0.264 0.102 C8 0.5103 0.0282 0.0101 0.351 0.083 0.595 0.099 0.310 0.090 Table A.8 Atomic fractional coordinates and error ellipsoid parameters of the p-dicyanobenzene molecule for the average location in 4 D C N B / Z S M - 5 with r2 > 0.93 determined from the 1 H / 2 9 S i C P drain at 305 K X y z e« E22 E33 E12 En E23 H2 0.4048 0.2100 -0.1460 0.074 0.951 0.127 0.210 -0.030 -0.031 H3 0.4289 0.0941 -0.1211 0.238 0.567 0.282 -0.361 0.227 -0.334 H5 0.5640 0.1433 0.1033 0.035 0.604 0.096 0.189 0.024 0.095 H6 0.5399 0.2592 0.0783 0.159 ' 0.345 0.238 -0.231 0.131 -0.194 C1 0.4704 0.2440 -0.0359 0.104 0.117 0.207 0.090 0.047 0.019 C2 0.4381 0.1960 -0.0939 0.040 0.402 0.054 0.070 -0.015 -0.040 C3 0.4521 0.1287 -0.0793 0.100 0.257 0.109 -0.142 0.090 -0.144 C4 0.4984 0.1093 -0.0069 0.128 0.044 0.214 0.002 0.135 0.011 C5 0.5307 0.1573 0.0511 0.017 0.201 0.037 0.058 0.016 0.034 C6 0.5167 0.2247 0.0366 0.054 0.128 0.083 -0.066 0.034 -0.063 N.1 0.4445 0.3686 -0.0627 0.801 0.342 1.613 0.539 0.598 0.285 N2 0.5244 -0.0153 0.0200 0.867 0.135 1.635 0.289 0.847 0.262 C7 0.4559 0.3138 -0.0509 0.410 0.224 0.828 0.298 0.281 0.138 C8 0.5130 0.0395 0.0082 0.457 0.076 0.843 0.119 0.459 0.121 246 Table A.9 Atomic fractional coordinates and error ellipsoid parameters of the p-dicyanobenzene molecule for the average location in 2 D C N B / Z S M - 5 with r2 £ 0.93 determined from the 1 H / 2 9 S i C P at 300 K X y z E „ E22 E M E , 2 E , 3 E 2 3 H2 0.4063 0.2869 -0.1475 0.102 0.257 0.067 0.087 -0.050 -0.030 H3 0.4134 0.1672 -0.1401 0.064 0.205 0.059 -0.086 0.022 -0.076 H5 0.5532 0.1819 0.0872 0.040 0.196 0.062 0.040 -0.004 0.053 H6 0.5461 0.3017 0.0798 0.068 0.173 0.062 -0.046 -0.011 -0.068 C1 ' 0.4756 0.3041 -0.0345 0.096 0.078 0.080 0.031 -0.034 -0.038 C2 0.4371 0.2650 -0.0984 0.053 0.139 • 0.024 0.035 -0.023 -0.034 C3 0.4412 0.1953 -0.0941 0.033 0.118 0.020 -0.034 0.011 -0.040 C4 0.4839 0.1648 -0.0258 0.058 0.062 0.075 -0.020 0.013 0.005 C5 0.5225 0.2039 0.0381 0.017 0.104 0.021 0.008 0.004 0.015 C6 0.5183 0.2735 0.0338 0.035 0.100 0.022 -0.011 -0.008 -0.035 N1 0.4680 0.4328 -0.0425 0.562 0.109 0.628 0.120 -0.149 -0.053 N2 0.4915 0.0361 -0.0178 0.454 0.062 0.615 -0.025 -0.017 0.072 C7 0.4714 0.3762 -0.0390 0.306 0.094 0.322 0.076 -0.090 -0.050 C8 0.4882 0.0927 -0.0213 0.229 0.060 0.314 -0.027 0.005 0.040 Table A. 10 Atomic fractional coordinates and error ellipsoid parameters of the p-dicyanobenzene molecule for the average location in 2 D C N B / Z S M - 5 with r2 £ 0.93 determined from the 1 H / 2 9 S i C P drain at 300 K X y z . E „ EjJ E M E „ E»3 E M H2 0.4054 0.2867 -0.1525 0.098 0.361 0.100 0.115 -0.043 -0.024 H3 0.4124 0.1669 -0.1465 0.079 0.287 0.096 -0.143 0.055 -0.115 H5 0.5522 0.1802 0.0810 0.042 0.292 0.095 0.104 0.017 0.064 H6 0.5452 0.3000 0.0751 0.060 0.262 0.092 -0.100 0.004 -0.077 C1 0.4747 0.3031 -0.0392 0.087 0.088 0.120 0.018 -0.020 -0.031 C2 0.4361 0.2644 -0.1036 0.052 0.178 0.041 0.035 -0.021 -0.032 C3 0.4402 0.1947 -0.1001 0.041 0.148 0.039 -0.059 0.026 -0.057 C4 0.4829 0.1638 -0.0322 0.066 0.062 0.120 -0.014 0.044 -0.002 C5 0.5215 0.2025 0.0322 0.020 0.138 0.038 0.029 0.014 0.019 C6 0.5174 0.2722 0:0287 0.030 0.134 0.037 -0.033 -0.004 -0.035 N1 0.4672 0.4319 -0.0456 0.519 0.133 0.907 0.116 -0.002 -0.032 N2 0.4904 0.0350 -0.0258 0.458 0.059 0.905 0.024 0.180 0.051 C7 0.4705 0.3753 -0.0428 0.280 0.111 0.468 0.065 -0.019 -0.035 C8 0.4871 0.0916 -0.0286 0.237 0.058 0.467 -0.001 0.112 0.024 A.3 p-Dinitronbenzene/ZSM-5 Table A. 11 Atomic fractional coordinates and error ellipsoid parameters of the p-dinitrobenzene molecule for the average location in 2DNB/ZSM-5 with r2 s 0.94 determined from the 1 H / 2 9 S i C P at 330 K X y z E„ £ 2 2 E33 E « E13 E23 H2 0.4397 0.1698 -0.1450 0.086 0.662 0.099 -0.011 -0.002 0.052 H3 0.4406 0.0515 -0.1167 0.095 0.537 0.162 -0.218 0.059 -0.192 H5 0.5678 0.0783 0.1260 0.069 0.548 0.074 -0.019 0.018 -0.003 H6 0.5670 0.1966 0.0977 0.076 0.424 0.229 -0.267 0.060 -0.069 C1 0.5033 0.1929 -0.0259 0.089 0.289 0.168 -0.079 > 0.035 0.015 C2 0.4665 0.1507 -0.0883 0.044 0.422 0.062 -0.063 0.005 -0.007 C3 0.4670 0.0819 -0.0719 0.048 0.380 0.072 -0.128 0.028 -0.111 C4 0.5042 0.0552 0.0069 0.090 0.289 0.115 -0.056 0.046 -0.089 C5 0.5410 0.0974 0.0693 0.034 0.356 0.047 -0.068 0.017 -0.039 C6 0.5405 0.1663 0.0529 0.037 0.314 0.111 -0.157 0.029 -0.039 N1 0.5028 0.2652 -0.0432 0.310 0.324 0.528 -0.015 0.119 0.099 N2 0.5047 -0.0171 0.0242 0.312 0.323 0.418 0.033 0.141 -0.114 01 0.5364 0.3013 0.0144 0.446 0.291 0.832 -0.171 0.213 0.025 02 0.4688 0.2870 -0.1146 0.443 0.542 0.654 0.191 0.128 0.244 03 0.5388 -0.0389 0.0956 0.436 0.480 0.533 0.233 0.160 0.005 04 0.4711 -0.0532 -0.0334 0.458 0.350 0.688 -0.098 0.233 -0.250 Table A.12 Atomic fractional coordinates and error ellipsoid parameters of the p-dinitrobenzene molecule for the average location in 2DNB/ZSM-5 with r2 s 0.92 determined from the 1 H / 2 9 S i C P drain at 330 K X y z £ 2 2 E33 En E,3 E23 H2 0.4056 0.2310 -0.1493 0.053 0.405 0.136 0.078 -0.008 0.003 H3 0.4219 0.1130 -0.1296 0.084 0.266 0.091 -0.132 0.069 -0.134 H5 0.5615 0.1489 0.0947 0.018 0.509 0.040 0.064 0.006 0.034 H6 0.5452 0.2669 0.0750 0.087 0.414 0.180 -0.233 0.067 -0.117 C1 0.4741 0.2586 -0.0388 0.072 0.151 0.184 -0.049 0.034 -0.009 C2 0.4382 0.2139 -0.0983 0.028 0.211 0.075 -0.001 -0.003 -0.014 C3 0.4477 0.1452 -0.0868 0.034 0.159 0.037 -0.061 0.024 -0.058 C4 0.4931 0.1213 -0.0159 0.049 0.126 0.076 0.001 0.043 0.000 C5 0.5289 0.1661 0.0436 0.008 0.272 0.018 -0.009 0.005 0.005 C6 0.5194 0.2347 0.0322 0.036 0.245 0.089 -0.120 0.023 -0.048 N1 0.4641 0.3307 -0.0509 0.248 0.187 0.578 -0.013 0.149 0.025 N2 0.5030 0.0493 -0.0038 0.202 0.135 0.356 0.091 0.168 0.043 01 0.4973 0.3691 0.0039 0.388 0.224 0.838 -0.202 0.274 -0.071 02 0.4231 0.3500 -0.1153 0.320 0.336 0.767 0.216 0.166 0.137 03 0.5441 0.0299 0.0606 0.256 0.340 0.498 0.310 0.192 0.171 04 0.4699 0.0108 •-0.0586 0.341 0.094 0.573 -0.046 0.293 -0.062 A.4 2+2 B e n z e n e / p - x y l e n e / Z S M - 5 Table A. 13 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-cfe molecule for the average location in 2+2 BENZ-d6/PXY-cyzSM-5 with r2 z 0.62 determined from t h e 1 H / 2 9 S i C P a t 2 7 0 K X y z E33 E12 E13 E23 H2 0.3855 0.2104 -0.0712 0.121 0.735 0.225 -0.065 -0.020 0.091 H3 0.4120 0.0988 -0.0181 0.461 0.422 0.236 -0.209 0.166 -0.341 H5 0.5790 0.1707 0.1299 0.202 1.361 0.150 0.092 -0.042 0.281 H6 0.5525 0.2822 0.0768 0.361 1.046 0.430 -0.284 0.001 -0.389 C1 0.4669 0.2554 -0.0015 0.232 0.298 0.271 -0.093 0.009 -0.034 C2 0.4260 0.2021 -0.0291 0.055 0.347 0.125 -0.080 -0.010 -0.004 C3 0.4414 0.1372 0.0017 0.193 0.242 0.096 -0.100 0.070 -0.121 C4 0.4976 0.1257 0.0602 0.337 0.299 0.115 0.041 0.091 0.105 C5 0.5385 0.1790 0.0878 0.103 0.711 0.081 0.012 -0.023 0.107 C6 0.5231 0.2438 0.0570 0.134 0.604 0.208 -0.144 -0.025 -0.149 CD1 0.4513 0.3207 -0.0326 0.928 0.401 0.756 -0.064 0.145 0.055 CD2 0.5132 0.0603 0.0913 1.139 0.403 0.442 0.206 0.311 0.335 Table A.14 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d6 molecule for the average location in 2+2 BENZ-d6/PXY-d6/ZSM -5 with r2 2 0.94 determined from t h e 1 H / 2 9 S i C P d r a i n a t 2 7 0 K X y z E11 E22 E33 E12 E13 E23 H2 0.4137 0.3140 -0.0958 0.942 2.414 0.610 0.741 0.490 0.515 H3 0.4098 0.1940 -0.0988 0.836 2.549 0.561 -0.705 0.412 -0.415 H5 0.5882 0.1863 0.0577 0.681 3.358 0.729 0.700 0.447 0.936 H6 0.5921 0.3063 0.0607 0.615 3.482 0.669 -0.664 0.391 -0.892 C1 0.5032 0.3199 -0.0173 0.619 0.897 0.596 0.004 0.380 -0.280 C2 0.4502 0.2873 -0.0637 0.381 1.190 0.232 0.254 0.165 0.122 C3 0.4479 0.2175 -0.0654 0.340 1.237 0.217 -0.245 0.136 -0.083 C4 0.4987 0.1804 -0.0208 0.596 0.903 0.602 -0.044 0.367 0.242 C5 0.5517 0.2131 0.0256 0.229 1.739 0.301 0.230 0.140 0.367 C6 0.5540 0.2828 0.0273 0.212 1.780 0.280 -0.221 0.124 -0.360 CD1 0.5056 0.3902 -0.0155 2.330 1.382 2.224 -0.027 1.548 -0.598 CD2 0.4963 0.1101 -0.0225 2.284 1.394 2.237 -0.122 1.522 0.451 Table A. 15 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-cfe molecule for the average location in 2+2 BENZ-d6 /PXY-d6 /ZSM-5 with r2 £ 0.99 determined from t h e 1 H / 2 9 S i C P at 293 K X y z e 2 2 E 1 2 E 1 3 E 2 3 H2 0.3954 0.3113 -0.0778 0.195 2.267 0.336 0.545 0.046 0.116 H3 0.3967 0.1913 -0.0823 0.217 2.253 0.373 -0.607 0.068 -0.134 H5 0.5788 0.1908 0.0647 0.195 1.512 0.347 -0.067 -0.020 0.305 H6 0.5775 0.3108 0.0692 0.187 1.533 0.349 0.024 -0.008 -0.305 C1 0.4863 0.3208 -0.0040 0.205 1.288 0.353 0.355 0.029 -0.109 C2 0.4337 0.2861 -0.0480 0.085 1.686 0.137 0.253 0.017 0.015 C3 0.4345 0.2163 -0.0506 0.096 1.677 0.154 -0.288 0.026 -0.025 CA 0.4878 0.1812 -0.0092 0.222 1.267 0.373 -0.367 0.035 0.101 C5 0.5404 0.2160 0.0349 0.085 1.247 0.143 -0.103 -0.021 0.125 C6 0.5397 0.2858 0.0375 0.079 1.259 0.140 0.079 -0.019 -0.125 CD1 0.4856 0.3911 -0.0013 0.754 1.380 1.331 0.729 0.152 -0.222 CD2 0.4886 0.1109 -0.0118 0.790 1.340 1.372 -0.722 0.164 0.199 Table A. 16 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d6 molecule for the average location in 2+2 BENZ-d6 /PXY-d6 /ZSM-5 with r2 £ 0.99 determined from t h e 1 H / 2 9 S i C P d r a i n a t 2 9 3 K X y z £« £ 2 2 E 3 3 E , 2 E 1 3 E 2 3 H2 0.3918 0.3052 -0.0961 0.378 1.746 0.249 0.361 0.147 0.429 H3 0.3987 0.1854 -0.0912 0.540 1.604 0.291 -0.371 0.221 -0.523 H5 0.5816 0.1989 0.0523 0.370 2.083 0.296 0.375 0.115 0.546 H6 0.5746 0.3187 0.0473 0.497 1.874 0.361 -0.441 0.192 -0.665 C1 0.4826 0.3217 -0.0248 0.434 0.535 0.312 -0.004 0.201 -0.047 C2 0.4315 0.2830 -0.0651 0.145 0.883 0.093 0.120 0.055 0.131 C3 0.4355 0.2133 -0.0622 0.204 0.843 0.104 -0.117 0.080 -0.159 C4 0.4907 0.1824 -0.0190 0.454 0.574 0.298 0.045 0.200 0.103 C5 0.5419 0.2211 0.0212 0.140 1.079 0.119 0.129 0.037 0.199 C6 0.5378 0.2908 0.0183 0.178 1.001 0.145 -0.157 0.063 -0.242 CD1 0.4785 0.3919 -0.0277 1.696 0.669 1.178 0.031 0.801 -0.039 CD2 0.4948 0.1122 -0.0161 1.737 0.746 1.150 0.130 0.798 0.263 Table A. 17 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d4 molecule for the average location in 2+2 BENZ-d6/PXY-d</ZSM-5 with r2 £ 0.93 determined from t h e 1 H / 2 9 S i C P a t 2 7 0 K X y z E„ E22 E 3 3 E 1 2 E 1 3 E 2 3 D2 0.4363 0.2743 -0.1287 0.489 0.223 0.486 0.108 -0.357 -0.028 D3 0.4051 0.1591 -0.1476 0.457 0.243 0.421 0.055 -0.372 -0.046 D5 0.5431 0.1053 0.0688 0.475 0.187 0.436 0.020 -0.373 -0.084 D6 0.5743 0.2205 0.0877 0.472 0.193 0.524 0.010 -0.415 -0.082 C1 0.5079 0.2568 -0.0189 0.034 0.170 0.098 0.011 -0.013 0.000 C2 0.4586 0.2389 -0.0874 0.180 0.194 0.192 0.048 -0.124 -0.006 C3 0.4405 0.1719 -0.0984 0.166 0.203 0.151 0.024 -0.126 -0.014 C4 0.4715 0.1228 -0.0410 0.017 0.178 0.009 -0.014 0.002 -0.013 C5 0.5208 0.1406 0.0275 0.172 0.174 0.163 -0.003 -0.133 -0.039 C6 0.5389 0.2076 0.0385 0.174 0.174 0.211 -0.001 -0.151 -0.035 cm 0.5261 0.3242 -0.0079 0.059 0.168 0.206 0.016 -0.013 0.000 CH2 0.4533 0.0553 -0.0520 0.025 0.185 0.028 -0.035 0.018 -0.024 Table A.18 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-d 4 molecule for the average location in 2+2 BENZ-cfe/PXY-oyzSM-5 with r2 2 0.93 determined from t h e 1 H / 2 9 S i C P d r a i n a t 2 7 0 K X y z E22 E33 E12 E13 E23 D2 0.4219 •0.2825 -0.1383 0.389 0.409 0.536 0.217 -0.341 -0.052 D3 0.3975 0.1650 -0.1410 0.385 0.442 0.436 0.061 -0.349 -0.111 D5 0.5476 0.1314 0.0663 0.400 0.175 0.469 0.029 -0.350 -0.097 D6 0.5719 0.2490 0.0690 0.407 0.174 0.514 0.059 -0.381 -0.101 C1 0.4989 0.2753 -0.0344 0.037 0.219 0.100 0.071 -0.003 -0.015 C2 0.4482 0.2509 -0.0955 0.146 0.318 0.206 0.109 -0.118 -0.031 C3 0.4340 0.1826 -0.0971 0.143 0.333 0.155 0.034 > -0.118 -0.058 C4 0.4706 0.1386 -0.0376 0.030 0.239 0.016 -0.037 0.010 -0.047 C5 0.5213 0.1630 0.0235 0.152 0.182 0.167 0.000 -0.123 -0.057 C6 0.5354 0.2314 0.0251 0.156 0.177 0.200 0.032 -0.136 -0.052 cm 0.5131 0.3442 -0.0328 0.070 0.212 0.243 0.118 0.017 -0.007 CH2 0.4563 0.0698 -0.0392 0.057 0.252 0.074 -0.099 0.043 -0.071 Table A.19 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-oV molecule for the average location in 2+2 BENZ-d6 /PXY-d 4 /ZSM-5 with r2 2 0.90 determined from t h e 1 H / 2 9 S i C P a t 2 9 3 K X y z E„ E 22 E33 E12 . E„ E 23 D2 0.4081 0.3045 -0.1516 0.827 3.303 0.702 0.232 -0.354 0.482 D3 0.4088 0.1844 -0.1508 0.816 3.293 0.724 -0.414 -0.380 -0.276 D5 0.5552 0.1860 0.0680 0.891 1.742 0.805 -0.089 -0.325 0.002 D6 0.5545 0.3060 0.0672 0.920 1.731 0.809 0.071 -0.308 -0.114 C1 0.4813 0.3150 -0.0423 0.392 2.085 0.176 0.193 0.127 0.202 C2 0.4389 0.2797 -0.1057 0.377 2.680 0.294 0.090 -0.103 0.230 C3 0.4393 0.2099 -0.1052 0.368 2.677 0.304 -0.226 -0.117 -0.104 C4 0.4821 0.1754 -0.0414 0.369 2.085 0.187 -0.276 0.102 -0.172 C5 0.5245 0.2108 0.0220 0.414 1.772 0.354 -0.097 -0.086 -0.049 C6 0.5240 0.2806 0.0216 0.428 1.769 0.353 0.056 -0.075 -0.010 CH1 0.4808 0.3854 -0.0427 1.078 2.115 0.404 0.430 0.393 0.397 CH2 0.4825 0.1051 -0.0409 1.032 2.116 0.425 -0.512 0.342 -0.353 Table A.20 Atomic fractional coordinates and error ellipsoid parameters of the p-xylene-04 molecule for the average location in 2+2 BENZ-d6 /PXY-aVZSM-5 with r2 2 0.92 determined from t h e 1 H / 2 9 S i C P d r a i n a t 2 9 3 K X y z Eu E22 E33 En En E23 D2 0.4006 0.3100 -0.1550 0.719 2.311 0.677 0.252 -0.315 0.262 D3 0.4006 0.1900 -0.1550 0.719 2.311 0.677 -0.252 -0.315 -0.262 D5 0.5534 0.1900 0.0539 0.821 1.094 0.761 -0.088 -0.360 0.037 D6 0.5534 0.3100 0.0539 0.821 1.094 0.761 0.088 -0.360 -0.037 C1 0.4770 0.3198 -0.0506 0.336 1.388 0.161 0.198 0.082 0.131 C2 0.4326 0.2849 -0.1113 0.352 1.845 0.283 0.126 -0.074 0.116 C3 0.4326 0.2151 -0.1113 0.352 1.845 0.283 -0.126 -0.074 -0.116 C4 0.4770 0.1802 -0.0506 0.336 1.388 0.161 -0.198 0.082 -0.131 C5 0.5214 0.2151 0.0102 0.412 1.137 0.331 -0.071 -0.100 -0.015 C6 0.5214 0.2849 0.0102 0.412 1.137 0.331 . 0.071 -0.100 0.015 cm 0.4770 0.3901 -0.0506 0.797 1.403 0.359 0.397 0.204 0.263 CH2 0.4770 0.1099 -0.0506 0.797 1.403 0.359 -0.397 0.204 -0.263 Table A.21 Atomic fractional coordinates and error ellipsoid parameters of the benzene molecule for the average location in 2+2 B E N Z / P X Y - d i 0 / Z S M - 5 with r2 2 0.82 determined from the 1 H / 2 9 S i C P at 270 K X y z E« E 2 2 E33 £« £ » £73 M 0.5378 0.3326 0.0539 0.842 0.911 0.487 -0.324 0.163 -0.473 H2 0.4567 0.3668 -0.0671 1.210 0.903 1.004 0.126 0.119 -0.052 H3 0.4127 0.2869 -0.1832 0.711 1.391 0.304 0.170 -0.160 -0.120 H4 0.4498 0.1728 -0.1783 1.160 0.954 0.600 -0.527 -0.081 0.087 H5 0.5309 0.1386 -0.0573 1.352 0.920 1.151 -0.150 -0.122 0.711 H6 0.5749 0.2185 0.0588 0.535 1.364 0.338 0.097 -0.157 0.083 C1 0.5194 0.2991 0.0053 0.305 0.572 0.183 -0.100 0.059 -0.208 C2 . 0.4722 0.3190 -0.0650 0.451 0.573 0.353 0.061 0.043 -0.090 C3 0.4466 0.2726 -0.1326 0.321 0.743 0.131 0.051 -0.081 -0.045 C4 0.4682 0.2062 -0.1297 0.490 0.597 0.249 -0.218 -0.083 0.118 C5 0.5153 0.1864 -0.0594 0.533 0.583 0.439 -0.100 -0.097 0.354 C6 0.5409 0.2328 0.0082 0.218 0.727 0.150 0.008 -0.079 0.073 Table A.22 Atomic fractional coordinates and error ellipsoid parameters of the benzene molecule for the average location in 2+2 B E N Z / P X Y - d i 0 / Z S M - 5 with r2 s 0.82 determined from the 1 H / 2 9 S i C P drain at 270 K X y z E« E 22 E33 El2 E13 E23 HI 0.5702 0.2551 0.0657 0.382 1.446 0.264 -0.009 -0.156 -0.071 H2 0.5249 0.3565 -0.0019 1.608 0.860 0.959 -0.168 0.047 -0.716 H3 0.4452 0.3517 -0.1349 1.601 0.962 1.025 0.586 -0.100 -0.142 H4 0.4108 0.2454 -0.2003 0.621 1.511 0.431 -0.030 -0.369 -0.043 H5 0.4561 0.1440 -0.1327 1.755 0.869 1.066 -0.605 -0.092 0.236 H6 0.5358 0.1489 0.0003 1.509 0.907 0.965 0.170 -0.025 0.783 C1 0.5368 0.2531 0.0101 0.166 0.728 0.099 -0.002 -0.065 -0.025 C2 0.5105 0.3121 -0.0292 0.591 0.537 0.341 -0.005 -0.006 -0.356 C3 0.4641 0.3093 -0.1066 0.618 0.579 0.383 0.247 -0.081 -0.158 C4 0.4441 0.2475 -0.1446 0.304 0.766 0.195 -0.015 -0.189 -0.008 C5 0.4705 0.1885 -0.1053 0.677 0.542 0.403 -0.260 -0.086 0.199 C6 0.5168 0.1913 -0.0280 0.564 0.547 0.349 0.005 -0.038 0.380 Table A.23 Atomic fractional coordinates and error ellipsoid parameters of the benzene molecule for the average location in 2+2 B E N Z / P X Y - d i 0 / Z S M - 5 with r2 s 0.92 determined from the 1 H / 2 9 S i C P at 293 K X y z E« E22 E33 E , 2 E « E « H1 0.4810 0.3696 -0.0763 1.520 0.820 2.305 0.367 1.219 -0.149 H2 0.3980 0.2990 -0.1502 0.910 2.524 1.005 0.841 0.164 0.444 H3 0.4086 0.1796 -0.1403 1.078 2.305 1.161 -0.967 0.341 -0.503 H4 0.5024 0.1308 -0.0566 1.575 0.842 2.279 -0.100 1.242 0.337 H5 0.5854 0.2014 0.0172 0.850 2.407 1.071 0.655 0.113 0.617 H6 0.5747 0.3208 0.0074 0.963 2.168 1.252 -0.686 0.267 -0.816 C1 0.4855 0.3197 -0.0722 0.544 0.503 0.811 0.181 0.389 -0.109 C2 0.4372 0.2786 -0.1151 0.352 1.096 0.361 0.307 0.041 0.130 C3 0.4434 0.2092 -0.1094 0.415 1.025 0.410 -0.361 0.103 -0.131 C4 0.4979 0.1808 -0.0608 0.576 0.516 0.797 -0.091 0.402 0.174 C5 0.5462 0.2219 -0.0178 0.317 1.028 0.399 0.199 0.011 0.230 C6 0.5400 0.2913 -0.0235 0.348 0.945 0.463 -0.198 0.060 -0.313 Table A.24 Atomic fractional coordinates and error ellipsoid parameters of the benzene molecule for the average location in 2+2 B E N Z / P X Y - d i 0 / Z S M - 5 with r^  s 0.92 determined from the 1 H / 2 9 S i C P drain at 293 K X y z E11 E22 E33 E„ E,3 E23 M 0.5397 0.3269 0.0556 0.518 0.770 0.424 -0.347 0.137 -0.406 H2 0.4685 0.3705 -0.0725 0.800 0.328 0.953 0.246 0.207 -0.091 H3 0.4151 0.2957 -0.1873 0.407 '1.072 0.205 0.293 -0.098 -0.004 H4 0.4330 0.1773 -0.1740 0.658 0.811 0.471 -0.530 0.036 -0.020 H5 0.5042 0.1338 -0.0459 0.809 0.436 1.046 -0.010 0.093 0.470 H6 0.5575 0.2086 0.0689 0.276 1.139 0.251 0.219 -0.110 0.170 C1 0.5174 0.2956 0.0075 0.194 0.411 0.154 -0.107 0.043 -0.167 C2 f 0.4760 0.3210 -0.0670 0.305 0.254 0.328 0.103 0.068 -0.081 C3 0.4449 0.2775 -0.1337 0.189 0.510 0.080 0.096 -0.048 -0.005 C4 0.4553 0.2086 -0.1260 0.275 0.435 0.181 -0.213 -0.016 0.058 C5 0.4967 0.1833 -0.0515 0.311 0.317 0.382 -0.046 0.002 0.245 C6 0.5278 0.2268 0.0153 0.113 0.549 0.107 0.053 -0.055 0.096 Appendix B Supplementary information for diffraction structures The followings are the supplementary data for the powder neutron diffraction of o-xylene-d 1 0 /ZSM-5 in Chapter 5 and the single crystal XRD data in Chapter 6. The tables of interatomic distances and angles and atomic coordinates and equivalent displacement parameters were generated using CIF files from either G S A S (powder neutron diffraction) or WIN-GX (single crystal XRD). B.1 Crystal structure determination for o-xylene-di0/ZSM-5 Table B.1 Interatomic distances (A) and angles (degree) for o-xylene-dio/ZSM-5 S i ( 1 0 ) - 0 26) 1 6023 1) S i 1) - 0 ( 2 1 ) #1 1 5873 1) S i ( 1 0 ) - 0 9) 1 5952 1) S i l ) - O ( l ) 1 6074 1) S i ( 1 1 ) - 0 14) 1 5805 1) S i 1 ) - 0 ( 1 6 ) 1 5529 1) S i ( l l ) - 0 10) 1 6112 1) S i 1 ) - 0 ( 1 5 ) 1 5990 1) S i ( l l ) - 0 22) 1 5660 1) S i 2 ) - 0 ( 6 ) 1 6389 1) S i ( l l ) - 0 11) 1 5910 1) S i 2 ) - 0 ( l ) 1 6157 1) S i ( 1 2 ) - 0 24) 1 5930 1) S i 2 ) - 0 ( 1 3 ) 1 5428 1) S i ( 1 2 ) - 0 12) 1 6043 1) S i 2) - 0 ( 2 ) 1 6071 1) S i ( 1 2 ) - 0 20) 1 6263 1) S i 3 ) - 0 ( 2 ) 1 5613 1) S i ( 1 2 ) - 0 11) 1 6640 1) S i 3 ) - 0 ( 1 9 ) # 2 1 5987 1) 0 ( 1 5 ) - S i 10)#3 1 6329 1) S i 3 ) - 0 ( 2 0 ) # 2 1 5680 1) 0 ( 1 6 ) - S i 4)#3 1 5907 1) S i 3 ) - 0 ( 3 ) 1 6091 2) 0 ( 1 7 ) - S i 4)#3 1 6106 1) S i 4 ) - 0 ( 4 ) 1 5505 1) 0 ( 1 8 ) - S i 9)#3 1 5843 1) S i 4 ) - 0 ( 3 ) 1 5855 2) 0 ( 1 9 ) - S i 3)#3 1 5987 1) S i 4 ) - 0 ( 1 7 ) # 2 1 6106 1) 0 ( 2 0 ) - S i 3)#3 1 5680 1) S i 4 ) - 0 ( 1 6 ) # 2 1 5907 1) 0 (21 ) - S i 1)#4 1 5873 1) S i 5) - 0 ( 2 1 ) 1 5899 1) 0 ( 2 2 ) - S i 7)#4 1 5774 1) S i 5 ) - 0 ( 1 4 ) 1 5661 1) 0 ( 2 3 ) - S i 7)#5 1 5608 1) S i 5 ) - 0 ( 4 ) 1 5342 1) 0 ( 2 4 ) - S i 12) #5 1 5930 1) S i 5 ) - 0 ( 5 ) 1 5860 1) 0 ( 2 5 ) - S i . 9)#5 1 6176 1) S i 6 ) - 0 ( 1 8 ) 1 6270 1) 0 ( 2 6 ) - S i 10)#5 1 6023 1) S i 6) - 0 ( 6 ) 1 5928 1) C ( l ) - C ( 2 1 4006 1) S i 6) - 0 ( 1 9 ) 1 6214 1) C ( 1 ) - C (6 1 4002 1) S i 6 ) - 0 ( 5 ) 1 5743 1) C ( l ) - C ( 7 1 4575 1) S i 7 ) - 0 ( 7 ) 1 6406 1) C ( 2 ) - C ( 3 1 4015 1) S i 7 ) - 0 ( 2 3 ) .1 5608 1) C ( 2 ) - C ( 8 1 4585 1) S i 7 ) - 0 ( 1 7 ) 1 6302 1) C ( 3 ) - C ( 4 1 4011 1) S i 7 ) - 0 ( 2 2 ) # 1 1 5774 1) C ( 4 ) - C ( 5 1 3992 1) S i 8 ) - 0 ( 8 ) 1 6047 1) C ( 5 ) - C ( 6 1 4001 1) S i 8 ) - 0 ( 7 ) 1 6075 1) C ( 3 ) - D ( l 1 1001 1) S i 8 ) - 0 ( 1 3 ) 1 5740 1) C ( 4 ) - D ( 2 1 0989 1) S i 8 ) - 0 ( 1 2 ) 1 5988 1) C ( 5 ) - D ( 3 1 1002 1) S i 9 ) - 0 ( 8 ) 1 5965 1) C ( 6 ) - D ( 4 1 0994 1) S i 9 ) - 0 ( 2 5 ) 1 6176 1) C ( 7 ) - D ( 5 1 0995 1) S i 9 ) - 0 ( 1 8 ) # 2 1 5843 1) C ( 7 ) - D ( 6 1 1010 1) S i 9) - 0 ( 9 ) 1 5861 1) C ( 7 ) - D ( 7 1 0998 (1) S i 1 0 ) - 0 ( 1 5 ) # 2 1 6329 1) C ( 8 ) - D ( 8 1 1000 (1) S i 10) -O(IO') 1 6141 1) C ( 8 ) - D ( 9 1 0991 1) 254 C(8)-D(10) 1 0998(1) 0(21)#1-Si(1)-0(16) 111 189 5) 0(21)#1-Si(1)-0(1) 107 316 6) 0(16) - S i (1) -0(1) 109 484 4) 0(21)#1-Si(1)-0(15) 109 707 5) 0 (16) -S i (1 ) -0 (15) 109 345 7) 0 (1 ) -S i (1 ) -0 (15) 109 772 5) 0 (6 ) -S i (2 ) -0 (2 ) 109 584 4) 0 (6 ) -S i (2 ) -0 (1 ) 109 463 6) 0 (2 ) -S i (2 ) -0 (1 ) 109 672 4) 0 (6 ) -S i (2 ) -0 (13) 109 609 6) 0 (2 ) -S i (2 ) -0 (13) 109 234 6) 0 (1 ) -S i (2 ) -0 (13) 109 266 6) 0 (2 ) -S i (3 ) -0 (3 ) 109 193 5) 0(2) -Si (3) -0(20)#2 108 609 5) 0(3) -Si (3) -0(20)#2 • 103 851 6) 0(2) -Si (3) -0(19)#2 107 166 4) 0(3) -Si (3) -0(19)#2 112 686 6) 0(20)#2-Si(3)-0(19)#2 11.5 192 7) 0 (4 ) -S i (4 ) -0 (3 ) 109 595 5) 0(4) -Si (4) -0(17)#2 113 576 5) 0(3) -Si (4) -0(17)#2 108 058 6) 0(4) -Si (4) -0(16)#2 108 765 4) 0(3) -Si (4) -0(16)#2 107 234 6) 0(17)#2-Si(4)-0(16)#2 109 424 7) 0 (21) -S i (5 ) -0 (5 ) 109 543 5) 0 (21) -S i (5 ) -0 (14) 109 483 7) 0 (5 ) -S i (5 ) -0 (14) 109 667 6) 0 (21) -S i (5 ) -0 (4 ) 109 569 5) 0 (5 ) -S i (5 ) -0 (4 ) 109 382 4) 0 (14) -S i (5 ) -0 (4 ) 109 183 6) 0 (18) -S i (6 ) -0 (6 ) 109 847 5) 0 (18) -S i (6 ) -0 (19) 109 508 6) 0 (6 ) -S i (6 ) -0 (19) 109 265 4) 0 (18) -S i (6 ) -0 (5 ) 109 124 5) 0 (6 ) -S i (6 ) -0 (5 ) 109 302 6) 0 (19) -S i (6 ) -0 (5 ) 109 781 5) 0 (7 ) -S i (7 ) -0 (23) 109 292 5) 0 (7 ) -S i (7 ) -0 (17) 110 235 4) 0 (23) -S i (7 ) -0 (17) 109 536 6) 0(7) -Si (7) -0(22)#1 106 417 6) 0 (23) -S i (7 ) -0 (22)# l 109 752 (6) 0(17)-Si (7) -0(22)#1 111 548 5) 0 (13) -S i (8 ) -0 (8 ) 109 472 5) 0 (13) -S i (8 ) -0 (7 ) 109 153 (6) 0 (8 ) -S i (8 ) -0 (7 ) 109 840 (4) 0(13)-S'i (8)-0(12) 109 603 (6) 0 (8 ) -S i (8 ) -0 (12) 109 493 (4) 0 (7 ) -S i (8 ) -0 (12) 109 267 (6) 0 (8 ) -S i (9 ) -0 (25) 109 712 (5) 0(8) -Si (9) -0(18)#2 107 523 (4) 0(25)-Si (9) -0(18)#2 108 815 (6) 0 (8 ) -S i (9 ) -0 (9 ) 109 388 (5) 0 (25 ) -S i (9)-0(9) 109 755 (6) 0(18)#2-Si(9)-0(9) 111 608 (6) 0(15)#2-Si(10)-0(10) 111 241 (4) 0(15)#2-Si(10)-0(26) 111 093 (6) 0 (10) -Si (10) -0(26) 109 438 (5) O(15)#2-Si(10)-O(9) 106 177 (6) 0 (10) -S i (10) -0 (9 ) 109 433 (5) O(26)-Si (10) -O(9) 109 390 (6) O ( 1 4 ) - S i ( l l ) - O ( 1 0 ) 109 127 (6) 0 ( 1 4 ) - S i ( l l ) - 0 ( 2 2 ) 109 312 (6) O ( 1 0 ) - S i ( l l ) - O ( 2 2 ) 109 897 (5) 0 (14) -S i (11) -0 (11) 109 919 5 0 (10) -Si (11) -0(11) 109 258 4 0 (22) -Si (11) -0(11) 109 316 5 0 (24) -Si (12) -0(12) 109 333 6 0 (24) -Si (12) -0(20) 109 355 6 0 (12) -S i (12) -0 (20) 109 396 4 0 (24) -S i (12) -0 (11) 109 587 6 0 (12) -S i (12) -0 (11) 109 200 6 O ( 2 0 ) - S i ( 1 2 ) - O ( l l ) 109 955 5 S i ( 2 ) - 0 ( 1 ) - S i ( l ) 152 765 5 S i ( 2 ) - 0 ( 2 ) - S i ( 3 ) 151 265 5 S i ( 4 ) - 0 ( 3 ) - S i ( 3 ) 167 557 7 S i ( 4 ) - 0 ( 4 ) - S i ( 5 ) 156 374 5 S i ( 5 ) - 0 ( 5 ) - S i ( 6 ) 146 336 5 S i ( 2 ) - 0 ( 6 ) - S i ( 6 ) 172 546 6 S i ( 7 ) - 0 ( 7 ) - S i ( 8 ) 159 965 6 S i ( 9 ) - 0 ( 8 ) - S i ( 8 ) 158 693 5 S i ( 1 0 ) - 0 ( 9 ) - S i ( 9 ) 143 496 7 S i (11 ) -0 (10 ) -S i (10 ) 153 976 5 S i (11)-0(11)-Si (12) 140 204 5 S i (12 ) -0 (12 ) -S i (8 ) 162 917 6 S i ( 8 ) - 0 ( 1 3 ) - S i ( 2 ) 164 599 8 S i ( l l ) - 0 ( 1 4 ) - S i ( 5 ) 163 699 8 Si (10)#3-0(15) -Si (1) 143 238 6 S i (1 ) -0 (16 ) -S i (4 )#3 168 471 6 S i (7 ) -0 (17 ) -S i (4 )#3 146 211 5 S i (6 ) -0 (18 ) -S i (9 )#3 138 368 6 S i (6 ) -0 (19 ) -S i (3 )#3 149 033 6 S i (3)#3-0(20) -Si (12) 144 677 5 S i (1 )#4-0(21) -S i (5 ) 152 531 7 S i (11) -0(22.) - S i (7) #4 151 785 7 S i (7 )#5-0(23) -S i (7 ) 157 857 8 S i ( 1 2 ) - 0 ( 2 4 ) - S i (12)#5 .155 159 8 S i (9 ) -0 (25 ) -S i (9 )#5 157 O i l 8 S i (10) -O(26) -Si (10)#5 142 315 7 C ( 6 ) - C ( l ) - C ( 2 ) 120 018 7 C ( 6 ) - C ( l ) - C ( 7 ) 119 917 7 C ( 2 ) - C ( l ) - C ( 7 ) 120 062 5 C ( l ) - C ( 2 ) - C ( 3 ) 119 994 5 C ( l ) - C ( 2 ) - C ( 8 ) 120 272 (6 C(3) -C(2) -C(8) 119 729 (7 D ( l ) -C (3 ) -C (4 ) 120 046 (8 D ( l ) -C (3 ) -C (2 ) 120 046 (6 C(4) -C(3) -C(2) 119 892 (7 D(2) -C(4) -C(5) 119 880 (6 D(2) -C(4) -C(3) 120 015 (9 C(5) -C(4) -C(3) 120 081 (7 D(3) -C(5) -C(4) 119 984 (7 D(3) -C(5) -C(6) 120 010 (8 C(4) -C(5) -C(6) 120 005 (5 D(4) -C(6) -C(5) 119 985 (6 D (4 ) -C (6 ) -C ( l ) 120 009 (8 C ( 5 ) - C ( 6 ) - C ( l ) . 120 001 (7 D(5)-C(7)-D(7) 109 544 (8 D(5)-C(7)-