Vania Engström
A Vibrational Spectroscopy Study of the Growth of Silicalite-1 Films
on Noble Metal Surfaces
1999:65
LICENTIATE THESIS
Licentiate thesis
Institutionen för Kemi och metallurgi
A Vibrational Spectroscopy Study of the Growth of Silicalite-1 Films on Noble
Metal Surfaces
Vania Engström
Division of Chemical Technology
Department of Chemical and Metallurgical Engineering Luleå University of Technology
S-971 87 Luleå, Sweden
Abstract
Ultra-thin membranes of molecular sieves are potentially capable of separating chemical components on a molecular size basis and may therefore be used in a variety of technically important application areas. To be able to produce even better membranes of this kind a deeper knowledge about the film formation mechanisms is necessary. Therefore the work presented in the present thesis was devoted to a better understanding of the film formation process, using the preparation of silicalite-1 films as a model system. Throughout the thesis reflection absorption infrared (RAIR) spectroscopy was used to gain insight, on a molecular level, about the structures formed. The vibrational modes of the clusters considered were calculated from the dynamic matrix of the clusters using a Keating-type potential. The frequencies and the atom vector displacements were derived from the eigenvalues and eigenvectors of the cluster´s dynamic matrix.
Concerning the structures formed upon hydrolyzing the coupling agent (γ- MPT) or during the evolution of the silicalite-1 layer in the synthesis solution, the results and conclusions presented rely heavily on the comparison of experimental with calculated vibrational spectra.
From RAIR spectra and calculations performed it was possible to show that the coupling agent was oriented with its molecular axis nearly perpendicular to the substrate surface. After hydrolysis the SiO3 groups form a quasi two-dimensional monolayer built predominantly of 6- membered puckered Si-O rings. The treatment of this monolayer with a charge-reversing cationic polymer disrupts the connectivity and increases
Subsequent hydrothermal treatment in a synthesis solution resulted in a gradual deformation of the 5-membered silicon-oxygen rings implying a disruption of Si-O-Si linkages and a formation of linear defects along the crystallographic c-axis. These imperfections may be the starting point for cracks upon calcination of the films.
Larger seeds substantially enhance the formation of defects in as- synthesised films, whereas a 60 nm-seeded surface was able to produce almost defect-free silicalite-1 films with a thickness of about 180 nm.
Scanning electron microscopy (SEM) was used to verify the coverage of the substrate surface with seed crystals, and x-ray diffraction (XRD) to measure the preferred orientation of the crystallographic axes of the molecular sieve films.
List of publications
This thesis is based on the following publications, referred to in the text by Roman numerals:
I Infrared spectroscopic study of a γγγγ-mercaptopropyl- trimethoxsilane monolayer on a gold surface
B. Mihailova, V. Engström, J. Hedlund, A. Holmgren and J. Sterte, J. Mater. Chem., 1999, 9, 1507-1510.
II Vibrational Spectroscopy Study of the Structure of Silicalite-1 Films on a Gold Surface
B. Mihailova, V. Engström, J. Hedlund, A. Holmgren and J. Sterte, Microporous and Mesoporous Materials, 32 (1999) 297-304.
III Preparation of Silicalite-1 Films on NaSH-modified Metal Surfaces
V. Engström, B. Mihailova, J. Hedlund, A. Holmgren and J. Sterte Part of a manuscript in preparation.
IV Silicalite-1 films grown on gold surfaces from seeds with different size
V. Engström, B. Mihailova, J. Hedlund, A. Holmgren and J. Sterte Accepted for publication in Microporous and Mesoporous
Materials.
Contents
1 INTRODUCTION ...1
1.1 MOLECULAR SIEVE FILMS...1
1.2 METHODS FOR SYNTHESISING MOLECULAR SIEVE FILMS...2
1.2.1 Molecular sieve films on gold surfaces...3
2 SCOPE OF THIS WORK...5
3 EXPERIMENTAL ...7
3.1 SYNTHESIS OF SILICALITE-1 FILMS ON DIFFERENT SURFACES...7
3.2 CHARACTERISATION METHODS...7
3.3 DESCRIPTION OF THE THEORY FOR SIMULATION OF VIBRATIONAL SPECTRA....11
4 RESULTS AND DISCUSSION ...17
4.1 COUPLING AGENTS AND CATIONIC MODIFIER...18
4.2 THE GROWTH OF SILICALITE-1 FILMS...22
4.3 PREFERRED ORIENTATION OF CRYSTALLOGRAPHIC AXIS...30
5 CONCLUSIONS...34
6 RECOMMENDATIONS FOR FUTURE WORK...36
7 ACKNOWLEDGEMENTS ...38
8 REFERENCES ...40
1 Introduction
1.1 Molecular sieve films
The nomenclature of molecular sieves is sometimes confusing and a short clarification with focus on the molecular sieve system used in the present thesis is therefore appropriate.
Historically, silicate minerals are divided into broad categories, one of which is the tectosilicates. The tectosilicates consist of minerals containing three-dimensional frameworks of SiO4 tetrahedra and are divided into families, one of which is the zeolites.1
Molecular sieves are porous materials containing channel systems with different radius depending on the specific kind of crystalline structure. The network is built up of rings of silicon, oxygen and aluminium, but other metal ions can be present as well.
In this work we have investigated only one kind of molecular sieve, namely Silicalite-1. Silicalite-1 is a silica molecular sieve with a two channel system, one straight and one zigzag. The diameters of these channel systems are about 5.4 Å, and Silicalite-1 is therefore termed a microporous material, since the pore size is less than 20 Å. The topology is a MFI framework which is the same as for the Zeolite ZSM-5. The difference between Zeolites and Silicalite-1 is that Zeolites has a SiO2/Al2O3 ratio with at least one aluminium per unit cell, whereas silicalite-1 contains only silica and oxygen atoms. Accordingly, Silicalite-1 is not a zeolite by definition although the term zeolite has been used sporadically even for porous all-silica materials or materials containing
Molecular sieves, especially the zeolites, are currently used in a large number of important industrial application areas,2 since they have extremely good adsorption, ion-exchange and catalytic properties. Their high specific surface area and well defined pore size renders the materials suitable as selective adsorbents or support for catalysts. Due to their strong ion-exchange selectivity for calcium, molecular sieves are frequently used in laundary products softening hard water.
Besides the already mentioned application areas, it is increasingly evident that these microporous materials have a great potential in the fields of chemical sensors, membrane technology, and semiconductors.
1.2 Methods for synthesising molecular sieve films
A variety of substrates have been used to prepare molecular sieve films.3 These substrates include organic polymer supports, ceramic and metal oxides, and metal surfaces.
In principal there are three different techniques for the preparation of zeolite films on substrates:
I. Deposition of preformed or embedded crystals on the substrate.4 - 6 II. Attachment of a precusor (eg. seed crystals) to the substrate surface,
providing suitable conditions for growth and inter-growth under hydrothermal conditions in the vicinity of the surface.7 - 10
III. Direct crystallization onto the substrate surface from synthesis solution or from a gel film formed on the surface at high
The growth of zeolite films on ceramic and metal oxide supports can be accomplished by immersing the substrate into an appropriate synthesis mixture followed by hydrothermal treatment. This is an example of direct crystallization.
The crystallization of films on various metal substrates has been reported.13,14 In this case both direct crystallization and film preparation using the seed film method have been utilized.
Polytetrafluoroethylene, polysulfonated styrene,15 and teflon16 have been used to prepare free-standing zeolite films, where the support is removed from the film after crystallization. The thin films synthesized so far are however very fragile and cracks easily into several pieces.
Carbon in the form of fibers, and ion exchange materials have also been used as support for zeolite layers forming membranes and porous particles, respectively. 17
1.2.1 Molecular sieve films on gold surfaces
The seed film method18 includes a charge modification of the substrate surface by means of a cationic polymer. The polymer is attached to the surface by electrostatic attraction and a firmly attached polymer therefore requires the substrate to have a substantial negative charge in aqueous solution. This fact excludes the direct modification of noble-metal surfaces by means of cationic polymers as well as the direct crystallisation of positively charged colloidal silicalite-1 seed crystals (pH < 4) on the support surface. The deficiency of negative charge on a gold surface was
coupling agent.19 The thiol group of the coupling agent have a strong affinity for noble metal surfaces, permitting the formation of covalent bonds between sulfur atoms and Au atoms on the metal surface.20-23 After adsorption the trimethoxy groups of the silane was hydrolysed under acidic conditions to give the corresponding silanol, which ultimately condense to siloxane. The resulting negatively charged siloxane surface was subsequently charge reversed by adsorption of a cationic polymer. The method was originally demonstrated by electrostatic adsorption of negatively charged ZSM-5 crystals and the formation and a continuous zeolite film after hydrothermal treatment in the synthesis solution.19 This novel method was also demonstrated using colloidal silicalite-1 seed crystals at low pH (pH < 4, implying positively charged crystals), where the charge reversing step could be excluded from the synthesis procedure.24 The four-step procedure described above was utilized in this thesis for the preparation of silicalite-1 films on gold substrates.
2 Scope of this work
The main objective of this work was to acquire a better understanding of the mechanisms involved in the preparation of Silicalite-1 films in general and on gold substrates in particular.
A condition for the elucidation of such mechanisms is knowledge on a molecular level of each step in the preparation procedure of the films.
Since vibration spectroscopy is known to be a very powerful and sensitive technique to study molecular structure and structure changes, a further objective was to investigate whether vibration spectroscopic methods could be applied to evaluate structural features of the silicalite-1 films.
Accordingly, various vibration spectroscopic methods were tested in order to find the best method for the examination of the metal surface after treatment with coupling agent, surface-charge-reversing cationic polymer, colloidal molecular sieve seeds, and finally after the growth and intergrowth of the seed crystals into a continuous film. The structural features of the films were evaluated and correlated to the preparative conditions used in the film synthesis.
During the evaluation of these structural features gradual structural changes were discovered leading to the formation of linear defects occuring during the post-hydrothermal treatment of the sample surfaces. A further objective thus became to investigate the origin of these defects in the silicalite-1 layers. It should be emphasized that the spectroscopically observed structural features involved in the synthesis of silicalite-1 films are explained by means of theoretical calculations correlating infrared
spectra to simulated spectra obtained from the diagonalization of the structure´s dynamical matrix.
Therefore, part of this work was also devoted to the understanding of the computational technique.
3 Experimental
3.1 Synthesis of silicalite-1 films on different surfaces
Silicalite-1 films on gold surfaces were prepared using a multi-step procedure.25 The gold surface was first modified using a silane (γ- mercaptopropyltrimethoxy silane, γ-MPT, was used in this work). The silane used was hydrolysed to form a monolayer of silicon oxide on the surface. The silicon oxide surface was then modified using a cationic polymer thereby obtaining a positive surface charge which facilitates the adsorption of the negatively charged colloidal silicalite-1 seed crystals.
After adsorption, the seed crystals were grown into a continuous film with a thickness which is primarily controlled by the synthesis time. This seed- film method allows for the preparation of very thin films (<100 nm) of molecular sieves of controlled thickness on a variety of substrate surfaces, including metal oxides, silicon, noble metals and fibres of cellulose and carbon.
3.2 Characterisation methods X-ray Diffraction:
X-rays are generated by bombarding a metal with high-energy electrons. In the Siemens D5000 powder diffractometer used in this study the metal was copper and CuKα radiation was generated. The diffractometer was equipped with a thin film accessory (a Göbels mirror).
When the sample is a powder, some of the crystallites will always be orientated so as to satisfy the Bragg condition, implying recorded
intensities at certain deflection angles (2θ) depending on the d-spacing of the crystals. For a powder sample this will give rise to an X-ray diffraction (XRD) pattern typical of the crystals investigated. However, the crystals in a silicalite-1 film are not expected to be randomly orientated like they are in a powder. This is an advantage in our investigation, since it permits the calculation of preferred orientation of the silicalite-1 crystals.
Scanning Electron Microscopy:
A Philips XL 30 scanning electron microscope (SEM) equipped with a LaB6 emission source was used in the investigation of the morphology and the thickness of the synthesized films. In order to obtain sufficient conductivity, the surface of the samples was sputtered with gold.
Fourier Transform Infrared Spectroscopy (FT-IR):
Infrared spectra of γ-MPT adsorbed on gold were recorded under vacuum using a Bruker IFS 113 V spectrometer ( paper I ), whereas the spectra of silicalite-1 films were recorded on a Perkin-Elmer 2000 FT-IR spectrometer ( paper II – IV ). All data were recorded at room temperature using a SPECAC reflection accessory. The beam of the Globar source was incident at an angle of 84o and masked with a 4 mm wide iris.
Five hundred scans were combined, and the resultant interferogram was Fourier transformed to obtain a resolution of 1 cm-1 over the spectral range.
The experimental method used is described in the following section.
Reflection Absorption Infrared Spectroscopy (RAIRS):
From the work of Robert G. Greenler26 it is known that the infrared reflection-absorption spectrum of a thin film adsorbed on a metal substrate depends on the optical constants of the film and the substrate as well as on the angle of incidence and polarization of the infrared radiation. The set-up of a RAIR experiment is shown in figure 1.
Incident radiation To detector
EII,XZ
Air n1
θ1 EII,Z
Thin film d n2 + ik2
θ2
Substrate n3 + ik3
Figure 1. The set up of the RAIR experiment. Θ is the angle of incidence d is the film thickness and n + ik is the complex refractive index containing the absorption coefficient, k.
In order for a vibrational mode in a molecule adsorbed on the surface of the substrate to absorb energy from the infrared radiation this mode of vibration must have a non-zero dipole transition moment. Furthermore, the electric vector of the radiation must of course be non-zero at the position of the vibrating entity but also the projection of this vector onto the transition moment of the molecule must be non-zero.
There are two vectors from the infrared radiation present at the substrate surface, one which is in the same plane as the incident and reflecting radiation (the p-polarized component) and one which is perpendicular to that plane (the s-polarized component). At the grazing angle only the p-polarized component has an enhanced electro-magnetic field on a metallic surface, while the s-polarized component approaches zero. For molecules on the surface therefore only those vibrational modes with dipole transition moments normal to the surface can absorb infrared radiation. Accordingly, grazing angle measurements are useful for studies of the orientation of adsorbed molecules.
The enhancement of the electric field at a metallic surface may be a factor of three depending on the grazing angle while a non-metallic surface shows no enhancement. In our experiments a grazing angle of 84o was used to obtain maximum sensitivity.
The magnitude of the electric field in the vicinity of an absorbing molecule depends on the refractive index of the sample.
3.3 Description of the theory for simulation of vibrational spectra A deeper understanding of vibrational spectra of molecules or molecular aggregates requires theoretical calculations of vibrational states.
Although the vibrational states of molecules are a quantum mechanical feature frequently described by the Schrödinger wave mechanics, the treatment of molecular vibrations has evolved from classical mechanics.
Basically, the classical and the quantum theories of molecular vibrations are closely related, especially so if the simple-harmonic approximation is valid. In practice this assumption is often very good and therefore the experimental frequencies needed for allowed transitions between quantized vibrational energy states in molecules are very similar to calculated mechanical frequencies.
The methods for modelling the Raman and infrared spectra describe with different degree of accuracy the experimental data depending on the structure of the system and the character of the force field acting. The methods used can be separated into two categories: small-cluster approximations, and simulation methods using a large-number particle ensemble.
The second category is represented by molecular dynamics and Monte Carlo simulations, and is based on iterative calculations of the configuration giving the lowest energy for an ensemble of interacting particles. The method makes it possible to model the structure of the system with great accuracy. It is possible to calculate both the static and the dynamical properties, i.e. the energy-optimised geometry as well as the vibration density of states. In addition, both thermodynamic and elastic
parameters can be calculated. However, the modelling of the dynamical properties is a quite sophisticated procedure requiring a lot of computer time and the accuracy of the calculations depends strongly on the right choice of the shape of the potential near the energy minimum. For that reason small-cluster methods continue to be developed and are successfully used to examine the correlation between subtle structural modifications and spectral changes.
The quantum mechanical method using molecular orbitals is also based on the minimization of the total energy of the cluster. In principle, the method allows for determination of the optimal geometry of the cluster, the values of the force constants, as well as the vibrational and electron density of states. However, the utilization of this method is limited by several factors,27 the calculation accuracy decreases with increasing number of cluster atoms and sometimes it is not possible to choose the most representative structural unit; the results are very sensitive to the initial set of base functions and to the manner of the compensation of the peripheral cluster charge; the method does not work well enough for materials with mixed ionic-covalent character of the chemical bonds because of the difficulty to find a potential appropriate for quantum mechanical calculations.
Different small-cluster approaches have been developed for materials with covalent or mixed ionic-covalent bonds such as silicalite-1. The FG- method implies a group analysis of the eigenmodes of a structural unit with
FG - λE = 0 (1)
where E is the unit matrix and G and F represents kinetic energy and potential energy, respectively.28 The elements of the G-matrix is determined from the cluster geometry, and the elements of F from the force constants of the interatomic interactions. The values of the force constants can be refined by fitting the calculated frequencies to the measured ones.
The FG-method is used for a large number of molecules and crystalline phases, but is less conveniently applied to low-symmetry materials with a complicated unit cell, as the molecular sieves, and for solids which are partially or entirely disordered.
The method using a Keating-type potential29 in a cluster with variable geometry surrounded by a shell of the nearest neighbours of the cluster peripheral atoms is applicable for investigation of both crystalline and disordered materials. Its efficiency to model the peak frequencies and intensities has been checked for various types of complex materials.30,31 Although, the half-width and the line shape of absorption bands are not very correctly simulated in the framework of this method, this is the computational method employed in the present thesis.
To get a deeper understandig of the formation of the silicalite-1 films on gold surfaces we have been using a computer program, developed by Dr. Boriana Mihailova, which can simulate the IR-spectrum for a given structure.
The computational procedure starts with the definition of independant structural parameters of the system studied. In our case the system is a number of connected SiO4 tetrahedra. If we for simplicity choose just one such tetrahedra, it is easy to realize that nine independant structural parameters are needed. Refering to the FG - method these parameters or coordinates should be four stretching coordinates (r1….r4) and six bending coordinates (α1…α3 and β1…β3 ), of which one coordinate is redundant.
Therefore we have nine coordinates (4+6-1) representing the vibrational modes of the tetrahedra. One should remember that there is only 3N - 6 vibrational degrees of freedom or vibrational modes in a molecular structure containing N atoms.
However, in the calculational method used in this thesis the structural parameters are choosen in a somewhat different way. The four stretching coordinates are the same, but the bending coordinates are represented by two angles corresponding to the O-Si-O angles of the two triangles constituting the tetrahedral structure. In addition there are three angles defining the orientation of these two triangles. Therefore we still have nine vibrational degrees of freedom (4+2+3). Generally the number of structural parameters can be expressed as, 12n - 3, where n is the number of connected SiO4 tetrahedra.
If two tetrahedra are connected, there should be eighteen internal tetrahedral parameters (9+9). However, three angles are needed two define the orientation of the tetrahedra with respect to each other. Therefore, the
angles describes the elongation of the two tetrahedra, the other two are defining the so called puckering of the system.
The intra-tetrahedral parameters are related to the short range order of the system while the inter-tetrahedral parameters ( three angles ), and the number of connected SiO4-tetrahedra in the cluster are related to the intermediate range structure.
Because each assumed cluster is not an isolated unit, the interaction between the cluster and its surroundings must be considered. This is accounted for by introducing a shell of tetrahedra surrounding the cluster.
Additional structural coordinates are needed namely three angles for the orientations of a certain shell tetrahedra with respect to its neighbour tetrahedra in the cluster, and coordinates describing the deformation of the shell tetrahedra. The interaction between cluster and shell atoms is included in the elements of the dynamical matrix, D.
To construct the dynamical matrix, D, we need in addition to these structural parameters and the masses of the atoms constituting the structure, also the potential energy of the system. In this computational method a Keating-type potential is used accounting for interactions between first and second order neighbours. The potential energy, U, may be written as,
U = ∑ K(i,j)[(r(i) - r(j))e(ij)]2 + ∑ B(ijk)[(r(i) - r(j))e(ik) - (r(i) - r(k))e(ik)]2 (2)
where K and B are the stretching and bending force constants, and (i,j,k) are first order (i,j) and second order (i,k) neighbours.
From the structural parameters and the assumption about the potential energy the dynamical matrix D is constructed:
Dbend Dstr
rj ri
U mj
mi ,
Dij
/ = +
∂ β
∂ α
⋅ ∂
αβ =
2 2 1
1 (3)
where α and β are the Cartesian co-ordinate index defining the motional direction of i and j atoms.
To calculate the frequencies (eigenvalues) and the atom vector displacements (eigenvectors) of the vibrational modes for a given cluster geometry with a certain number of atoms, the computational program diagonalizes the dynamical matrix by using the Jacobi method. The dipole moment (or rather transition moment) of a certain vibrational mode is calculated from the atom vector displacements on the assumtion that the atoms constitute point charges. From the transition moment of a vibrational mode the intensity of the corresponding absorption band is calculated and thus a vibrational spectrum can be simulated.
4 Results and discussion
To acquire a better understanding of the mechanisms involved in the preparation of silicalite-1 films on metal substrates it became natural to first investigate the surface structure of the coupling agent and its interaction with the cationic modifier, and then to monitor the dependence of the silicalite-1 film structure on reaction time, and the size of the seed crystals. The first step in the synthesis of silicalite-1 films on gold substrates is the adsorption of a coupling agent onto the gold surface, and the subsequent one to modify the surface charge by means of a cationic polymer. This part of the synthesis procedure was investigated in paper I.
The objective was to determine the structures formed by trimethoxy-silane functions of the adsorbed γ-MPT molecules upon hydrolysis and condensation, and to study the influence of the cationic polymer on the structures formed.
A second objective was to investigate the influence of the synthesis time on the silicalite-1 structure formed when a monolayer of seed crystals of a constant size (60 nm) had been adsorbed on the modified surface (paper II).
It has been shown by others25 that a charge modification of a negatively charged substrate surface is sufficient in order to attach seed crystals to this surface and subsequently grow the crystals into a continuous film.
Consequently any other firmly attached coupling agent producing a negative charge on a gold surface for the cationic polymer to adsorb on should work as good as γ-MPT in the preparation procedure of silicalite-1
where the films grown by this new method is compared with those prepared using γ-MPT as coupling agent.
Finally, the effect of seed size on the growth and inter-growth of silicalite-1 layers was investigated, and optimal synthesis conditions for the preparation of defect-free silicalite-1 films were suggested (paper IV).
4.1 Coupling agents and cationic modifier
In this thesis the coupling agents γ-mercaptopropyltrimethoxy silane (γ- MPT) and sodium hydrosulfide (NaSH) were used. As the name implies
γ-MPT is comprised of two important functional groups. The thiol tail is able to form a covalent bond to a variety of metals through the sulfur atom.20-23 The methoxysilane head group on the other hand is capable to undergo hydrolysis and condensation reactions. In order to elucidate the structure of γ-MPT adsorbed on the gold surface reflection-absorption infrared spectroscopy is a powerful method since the surface selection rule dictate that vibrational transitions with dipole moments predominantly perpendicular to the surface are strongly enhanced relative to the transitions with dipole moments parallel to the surface. The gain in sensitivity in a reflection experiment over a corresponding transmission measurement is due to the enhancement of the electric field component normal to the surface, and the fact that the absorbance is proportional to the electric field squared.
Figure 2. Two idealized conformations of fully hydrolyzed γ-MPT molecules adsorbed on the gold surface.
where black circles represent Si atoms. According to RAIR spectra of unhydrolysed γ-MPT the methyl groups are predominantly oriented with their C3 – axis perpendicular to the substrate surface. At first sight this result may seem a bit surprising. However, it can be explained by the proximity of nearest neighbours. If the coupling agents are close enough to each other the methyl groups are forced to align with their C3-axis perpendicular to the surface. A reason for close packing of coupling agents may be a strong affinity of sulphur atoms for surface Au-atoms. A close packing would imply that the propyl chains are predominantly oriented along the surface normal. This was experimentally verified by the lack of
Au Au
CH2
CH2 CH2
CH2 CH2
CH2
S S
A B
unhydrolysed samples. Accordingly, the planes containing the atoms of a CH2 group are parallel to the surface.
X-ray diffraction data showed the crystalline gold to be oriented with the (1,1,1)-planes parallel to the silicon wafer surface. In this context it is interesting to notice that the distance between two neighbouring Au-atoms is too short to allow the adsorption of γ-MPT on each atom if the coupling agent retain all three methyl groups. This implies that neighbouring Au- atoms cannot be occupied by a γ-MPT molecule unless this molecule is partially hydrolysed. The hydrolyzation may be accomplished by small amounts of water in the pure methanol solvent used. The orientation of the methyl groups and the long time period necessary (3h) to cover the gold surface with sufficient silane for successfull seeding of the silicalite-1 colloidal crystals after surface modification supports this suggestion. After hydrolysis and condensation the possible structural units in the γ-MPT monolayer formed are chains and sheets of monomers, dimers or n- membered rings of SiO3R groups according to figure 3.
Figure 3. Top view of possible clusters of SiO3R units on a Au (111) surface: a chain of monomers (a), a chain of dimers (b), an isolated 3- membered ring (c), and a defected sheet of rings (d). Black circles represent Si atoms and grey circles are vacant sites in the hexagonal close packing.
In order to estimate the predominant type of structure in the layer, the vibrational frequencies of possible arrangements of SiO3R entities were calculated and compared with the experimental frequencies from RAIR
(c) (a)
(b) (d)
a large component of their dipole moments normal to the gold surface. The results show that the experimental IR spectrum of an adsorbed γ-MPT could be best modelled using a 2D-polymerized 6-membered ring in which two of the SiO3R groups have the B conformation (Fig. 2)
After treatment of this two-dimensional silicon-oxygen sheet with the cationic polymer, the structure of the sheet is changed. Absorption bands influenced by bridged Si-O stretching and Si-O-Si bending vibrations strongly decrease in intensity and become broader. Therefore, the cationic polymer seems to disrupt the connectivity of the Si-O sheet which increases its topological disorder. Infrared absorption bands originating from the cationic polymer could not be detected, most probably due to its low concentration.
A new method to facilitate the preparation of molecular sieve films on noble metal surfaces was introduced by using NaSH as coupling agent. As compared with γ-MPT the adsorption time could now be reduced to 5 minutes instead of 3 hours, and still give the same quality of the continuous silicalite-1 films synthesised by the seeding method. Unfortunately, the surface species formed could not be detected by RAIR spectroscopy.
4.2 The growth of silicalite-1 films
From X-ray diffraction patterns as well as RAIR spectra it was not possible to differentiate between seeds adsorbed or films grown on a
the film formation. Thus, it seems likely that any negatively charged surface modified by a charge reversing agent should have the properties suitable for the formation of continuous molecular sieve films.
The next step in the elucidation of the film formation process was to investigate the time dependence of the growth and inter-growth of the silicalite-1 films after seeding with crystals having a mean diameter of 60 nm (paper II).
So far the research on the growth of molecular sieve films has been investigated mainly by SEM and XRD since the attention has been focused on the thickness, morphology and crystallite orientation. The use of vibrational spectroscopy is rather scant although this technique offers the possibility to gain information about the structure of crystalline grains, the grain boundary interface and possible cavities in the volume of a continuous film.
The observed changes in the RAIR spectra with synthesis time were analysed on the basis of calculated vibrational modes of the SiO4-rings building the MFI structure, namely; 5-membered rings forming chains along the crystallographic c-axis, 4- and 6-membered rings linking these chains along the b-axis and 5- and 6-membered rings linking the chains along the a-axis.31
The calculated frequencies and intensities of vibrational modes involving these rings were adjusted to corresponding data from RAIR spectra.
In figure 4 below, RAIR spectra of silicalite-1 seeded Au-surfaces treated for different times in the synthesis solution are shown.
600 800 1000 1200 1400 3000 P2
P1
P3
S3 S6 S8.5 S11
S36
S24
S18
S0
Absorption
Frequency / cm-1
Figure 4. Experimental RAIR spectra of silicalite-1 seeded Au surface treated for different times in the synthesis solution ( S3 = 3 hours ).
The peak P1 is typical for the MFI-type structure with well developed long range ordering,32 whereas the band denoted P2 is characteristic of any 3D network of SiO4 tetrahedra. According to our calculations the P1-band could be modelled as originating from anti-symmetric Si-O bond stretching in the chains containing 5-membered rings.
Furthermore, our calculations show that a vibrational mode with high contribution from non-bridging oxygen atoms in 5-membered rings would appear in the frequency range 1030 cm-1 – 1070 cm-1. Non-bridging oxygens are formed from broken Si-O-Si linkages, which would disrupt the long range ordering in the structure.
After about 11 hours of hydrothermal treatment an extra band (P3) at 1068 cm-1 emerges as a shoulder on the P2-band. Since this is within the spectral range where the non-bridging oxygen atoms would appear we assume this band to originate from broken Si-O-Si bridges in chains containing 5-membered rings causing linear defects along the c-axis of the film structure. An experimental fact that supports this assumption is the P1/P2 intensity ratio. This ratio has a maximum value at about 11 hours of crystallisation time, and then decreases slightly with the time of crystallisation.
Upon calcination of the samples the intensity of the P3-band increases, most probably because of additional formation of linear defects. The assumed void spaces formed due to the interaction between grain boundaries and dislocations may be the ultimate reason for the cracking behaviour of continuous molecular sieve films.
Optimal synthesis conditions
Although the so called fingering structure of the surface of a cracked film may be an advantage in certain applications such as piezoelectric sensor devices and catalytic coatings, the absence of nano-scale defects is usually desirable in membrane preparation. It is therefore interesting to find out the optimal synthesis conditions for the preparation of defect-free films.
Our experimental approach to this objective was to examine the films formed after different times of hydrothermal treatment in the synthesis solution, and to use seed crystals of three different mean diameters ( 60 nm, 165 nm and 320 nm ). RAIR spectra from this comparative study are shown in figure 5.
600 800 1000 1200 1400 3000 N=165nm
t = 0h t = 12h t = 24h t = 36h
Frequency (cm-1)
600 800 1000 1200 1400 3000 t = 36h
t = 24h
t = 12h
t = 0h N=60nm
Absorption
600 800 1000 1200 1400 3000 t = 36h
t = 12h
t = 0h t = 24h N=320nm
The spectra are normalized to the P2-band at about 1117 cm-1. The evaluation of these spectra show that films grown for less than 18 hours ( corresponding to a film thickness of 180 nm ) from 60 nm seeds are almost free from defects. Even after 36 hours in the synthesis solution the intensity of the P3 band is low and probably the concentration of defects is large only in a few nanometers near the film surface. In this context it may be worthwhile to mention that the films obtained after 36 h with exchange of the synthesis solution each 12 h showed no observable difference from films synthesized for 36h without changing the synthesis solution.
Consequently, the emerging of defects is not due to a decreased concentration of silica in the synthesis solution during film formation.
For the larger seeds the formation of defects starts much earlier and after 36 hours the degree of defectness is considerably higher than for 60 nm seeds. The existence of these defects enhances the risk for crack formation during the subsequent calcination step.
Surface coverage and concentration of defects
RAIR spectroscopy can not only be used to study the structure of molecular sieve films but also to get information about the extent to which the seed crystals cover the substrate surface.
The O-Si-O bending mode at 465 cm-1 is most appropriate for this purpose, since it is independent of preferred axis orientation and insensitive to structural modifications.
For an incident angle of θ ≤ 84o it can be shown26,33 that the linear absorption can be approximated by,
I = 9.212 sin2θ .ε c d /n3 cos θ (4)
where ε is the molar absorbtivity, c is the concentration of absorbing species, d is the film thickness and n is the refractive index of the film.
Since I ∝ c.d the degree of coverage of the seeded surface before hydrothermal treatment is proportional to I/d2. In this case, d, is equal to the diameter of the seed crystal. The calculation showed that the surface seeded with the larger crystals ( 165 nm and 320 nm ) only occupies 70 % of the surface covered by 60 nm crystals. This value is approximately in correspondence with the SEM micrograph below showing a surface seeded with 60 nm, 165 nm and 320 nm crystals (from top to bottom in figure 6 on the next page).
Figure 6. SEM micrographs of gold surfaces with crystallites of a size of 60 nm, 165 nm and 320 nm ( from top to bottom ).
The linear absorption, I, at 465 cm-1 initially shows an expected linear increase with increasing crystallisation time. However, at times
corresponding to the formation of defects ∂I/∂t decreases probably due to an increased porosity of the film and therefore a lower concentration of absorbing species. Thus the non-linearity of I(t) is a measure of the concentration of defects.
4.3 Preferred orientation of crystallographic axis
According to XRD data films grown from 60 nm, 165 nm and 320 nm seeds all show the same tendency to change the crystallographic orientation from the b-axis being the dominating axis along the surface normal to the a-axis being the preferred orientation perpendicular to the surface. In figure 7 the XRD patterns for S(165 nm/ t=0), S(165 nm / t=24h) and the powder reference are shown.
2 2,5 2 3,0 2 3,5 2 4,0 2 4,5 reference
S(165|0) S(165|24)
(133) (303) (151) (051) (501)
Intensity
2θ / deg
Figure 7. XRD patterns of silicalite-1
(0,5,1) and (5,0,1) describes the orientation of the b- and a-axis, respectively. It can immediately be seen that the c-axis is predominantly oriented parallel to the surface whereas b starts to be perpendicular to the surface (before hydrothermal treatment) but after a continuous film has been grown the a-axis is the predominant direction perpendicular to the surface. It should be stressed that in this evaluation the intensity values originating from the silicalite-1 film are compared with the corresponding intensities measured in the diffractogram of the powder, where all directions are equally probable.
The preferred relative orientation of the a, b and c axes during the film growth can be viewed more clearly by defining a parameter, α, as,
α(x/y) = 0.5[ (ρs - ρr)/(ρs + ρr) + 1] = ρs /( ρs + ρr) (5)
where ρ = I(h,k,l)x / I(h,k,l)y and, s, and, r, stands for the sample and the powder reference.
The parameter α(x/y) = 0 when I(h,k,l)x = 0 i.e. when the normal to the (h,k,l)y plane is perpendicular to substrate surface, and when α(x/y) = 1 the (h,k,l)x plane is perpendicular to the surface. When the grains are randomly oriented, the value of α(x/y) is calculated according to equation 6.
α(x/y) = α(051/501) = I(051/501)s / ( I(051/501)s + I(051/501)r = 1/2 (6)
The dependence of α(051/501) on the crystallisation time and film
0 5 10 15 20 25 30 35 40 0,0
0,5 1,0
a b
α [(05
1)/(501)]
time / hour
100 200 300 400 500
0,0 0,5 1,0
a b α [(05
1)/(501)]
thickness / nm
Figure 8. Dependence of α[(051)/(501)] on the crystallisation time and on the film thickness. (■) represent series S(60|t), (▲) series S(165|t) and (◆) series S(320|t).
As evident from this figure the dominating orientation of the b and a
seed (60 nm) possess a somewhat stronger a-orientation than the larger seeds.
5 Conclusions
The following conclusions can be drawn from the results presented in this licentiate thesis:
• Infrared Reflection Absorption Spectroscopy (RAIR) offers a possibility to get a deeper understanding of mechanisms involved in the formation of molecular sieve films on a highly reflecting substrate like gold.
• The γ-MPT coupling agent is oriented with the methylene chain perpendicular to the sputtered gold surface, and the C3-axis of the methyl group is also oriented in that direction.
• A close packing of the γ-MPT molecules demands the methoxy groups to be at least partially hydrolysed before the actual hydrolyzing reaction at low pH. This is probably the reason why such a long time period is necessary (3h) to cover the gold surface with sufficient silane for a successful seeding of the silicalite-1 colloidal crystals after surface modification.
• The cationic polymer modifies the surface charge but also introduces structural disorder in the silane layer.
• Upon condensation the silanol groups predominantly form 6-membered puckered Si-O rings in which part of the oxygen atoms are non- bridging.
• To avoid linear defects in the as-synthesised films the time of
• After about 5 hours of hydrothermal treatment in the synthesis solution the crystallographic a-axis is the dominating direction perpendicular to the substrate surface.
• The deviation from linearity in a plot of the linear absorption intensity, I, at 465 cm-1 versus the synthesis time or the film thickness may be used as a measure of the concentration of defects in the as-synthesised films.
6 Recommendations for future work
Many ideas have arisen during the work leading to the results presented in this licentiate thesis. Among these are the use of other metal substrates in conjunction with NaSH as coupling agent to synthesize molecular sieve films by the seeding method. Another interesting coupling agent would be NaCN, since the cyanide ion is known to form strong metal cyanide complexes especially with the transition metals, and transition metals are often used in catalysis. The cyanide complexes formed are negatively charged and should be easily modified by a cationic polymer, and therefore fulfil an important requirement for the synthesis of inorganic layers for example zeolites on transition metal substrates.
The linear defects observed in the present thesis is a serious drawback if defect-free films are desirable. The seed crystals are synthesised using templates like tetrapropyl ammonium bromide (TPABr) or tetrapropyl ammonium hydroxide (TPAOH). No defects are observed in the seed crystals. It would therefore be interesting to adsorb templates on the substrate surface and try to synthesise seed crystals in situ. This modified procedure would exclude the need for a modifying agent and perhaps result in more defect-free films or films with another preferred orientation.
The difficult part here may be the synthesis of a template with a functional group able to attach on the substrate surface.
Still another development in the synthesis of inorganic films would be
with macro-pores controlled by the amphiphilic surfactants in the liquid crystalline phase (micellar or hexagonal) surrounded by nano-porous inorganic structures.
7 Acknowledgements
It is a pleasure for me to express my sincere gratitude to,
Dr. Boriana Mihailova, Central Laboratory of Mineralogy and Crystallography, Bulgarian Academy of Sciences, Sofia, Bulgaria, for all her help in this work. Her knowledge about Vibrational Spectroscopy in general, but especially the method of calculating vibrational frequencies has been invaluable to me. Also, I will never forget the talent you expressed on the dancing floor when we experienced Luleå by night.
Professor Johan Sterte and Dr. Jonas Hedlund, my supervisors.
Thank you for giving me the opportunity to perform this scientific work and for your great ability to find out new experimental ideas.
Docent Allan Holmgren, my patient fiancé who listened when I wanted to talk and talked when I was too tired to listen.
Secretary Ingrid Granberg. There are just to few people like you in this world! Thank you for pleasant discussions and your ability to show empathy.
Ms. Magdalena Lassinantti and Ms. Lubomira Tosheva. Thank you for all your support, both at work and in my leisure time, and for providing me with a lot of more or less decent jokes.
Professor Willis Forsling, for putting the spectrometers at my disposal, and Maine Ranheimer for keeping the spectrometers working, which was a prerequisite for the success of this work.
Till sist vill jag tacka mina föräldrar, Maj-Louise och Carl-Gustaf, för att ni alltid funnits till hands när jag behövt både uppmuntran och omtanke.
8 References
1. Szostak, R., Handbooh of Molecular Sieves, Van Nostrand Reinhold, New York, 1992.
2. Ozin, G., Kuperman, A., and Stein, A., Angew. Chem. Int. Ed. Engl. 28 (1989) 359
3. Bein, T. , Chem. Mater. 8(1996) 1636
4. Munoz Jr., T., and Balkus Jr., K. J., J. Am. Chem. Soc. 121 (1999) 139 5. Gimon-Kinzel, M. E., Balkus Jr., K. J., Microporous and Mesoporous
Materials 28 (1999) 113
6. Nishayama, N., Ueyama, K., and Matsukata, M., Micropor. Mater. 7 (1996) 299
7. Valtchev, V., Schoeman, B. J., Hedlund, J., Mintova, S., and Sterte, J., Microporous and Mesoporous Materials 8 (1997) 93
8. Mintova, S., Hedlund, J., Valtchev, V., Schoeman, B. J., and Sterte, J., J. Mater. Chem. 7 (1997) 2341
9. Lovallo, M. C., Gouzinis, A., and Tsapatsis, M. AIChE Journal 44 (1998) 8
10. Hedlund, J., Mintova, S., and Sterte, J., Microporous and Mesoporous Materials 28 (1999) 185
11. Jansen, J. C., Koegler, J. H., van Bekkum, H., Calis, H. P. A., van den Bleek, C. M.,Kapteijn, F., Moulijn, J. A., Geus, E. R., and van der Puil, N., Microporous and Mesoporous Materials 21 (1998) 213
12. Nakazawa, T., Sadakata, M., and Okubo, T., Microporous and
14. Tsikoyaninis, J.G., Haag, W.O., Zeolites 12 (1992) 126
15. Anderson, M.W., Pachis, K.S., Shi, J., Carr, S.W., J. Mater. Chem. 2 (1992) 255
16. Sano, T., Kiyozumi, Y., Mitzukami, F., Takaya, H., Mouri, T., Watanabe, M., Zeolites, 12 (1992) 131
17. Tosheva, L. Zeolite Macrostructures, Licentiate thesis 1999:27, Department of Chemical and Metallurgical Engineering, Division of Chemical Technology, Luleå University of Technology.
18. Hedlund, J., Schoeman, B., Sterte, J., Stud. Surf. Sci. Catal., 105 (1996) 2203
19. Mintova, S., Hedlund, J., Schoeman, B., Valtchev, V. and Sterte, J., Chem. Commun. (1997) 15
20. Nuzzo, R. G. and Allara, D. L., J. Am. Chem. Soc., 105 (1983) 4481 21. Bain, C. D., Troughton, E. B., Tao, Y.-T., Evall, J., Whiteside, G. M.,
J. Am. Chem. Soc., 111 (1989) 321
22. Nuzzo, R. G., Zegarski, B. R. and Dubois, L. H., J. Am. Chem. Soc., 109 (1987) 733
23. Carey, R. I., Folkers, J. P. and Whiteside, G. M., Langmuir, 10 (1994) 2228
24. Mintova, S., Valtchev, V., Engström, V., Schoeman, B., Sterte, J., Microporous Materials 11 (1997) 149
25. Hedlund, J., Thin Films of Molecular Sieves, Synthesis and Applications, Doctoral Thesis 1998:33, Department of Chemical and Metallurgical Engineering, Division of Chemical Technology, Luleå University of Technology.
26. Greenler, R. G., J. Chem. Phys. 44 (1966) 310
27. Hartman, M., Clark,T., van Eldik, R., J. Am. Chem. Soc. 119 (1997) 7843
28. Wilson, E. B., Decius, J. C., and Cross, P. C., “Molecular Vibrations”, McGraw-Hill, New York, 1955
29. Furukawa, T., Fox, K., and White, J., J. Chem. Phys., 75 (1981) 3226 30. Mihailova, B., Zotov, N., Marinov, M., Nikolov, J., and Konstantinov,
L., J. Non-Cryst. Solids, 168 (1994) 265
31. Mihailova, B. and Konstantinov, L., Phys. Chem. Glasses, 38 (1997) 27 32. Burkett, S, Davis, M. E., Chem. Mater. 7 (1995) 920.
33. Boerio, F. J., Schoenlein, L. H., and Greivenkamp, J. E., J. Appl.
Polym. Sci 22 (1978) 203.
