Chapter 1: Xenon as a Probe Atom: Introduction, Characteristics, Investigation of Microporous Solids

The central idea of the pioneers¹  of this technique was to find a non-reactive molecule particularly sensitive to its environment, detectable by NMR, which could be used as a probe to determine microporous solid properties difficult to detect by classical physicochemical techniques in a new way.

The isotope ¹²⁹Xe is this ideal probe: spin 1/2, 26.4% concentration in xenon, fairly good detection sensitivity and very wide chemical shift range. The very large and extremely polarizable electron cloud of xenon makes this atom particularly sensitive to its immediate environment. Small variations in the physical interactions with the latter cause marked perturbations of the large and extremely polarizable electron cloud which are transmitted directly to the xenon nucleus and greatly affect the NMR chemical shift. The quadrupolar ¹³¹Xe isotope, with spin 3/2, can also be used, but less easily, for certain specific applications. Now, by using optical polarization techniques the sensitivity of detection can be increased by several orders of magnitude, which widens the field of applications of this technique.²  Fragmentary reviews have already been published.^3,4  In this chapter we will focus our attention mainly on ¹²⁹Xe NMR as a means of probing the properties of zeolites, adding only few words on other microporous solids.

Zeolites are crystalline aluminosilicates with open 3D framework structures consisting of SiO₄ and AlO₄ tetrahedra linked to each other by sharing all the oxygen atoms, to form regular intracrystalline cavities and channels of molecular dimensions, inside which cations, water and/or small molecules may reside. They are often also referred to as molecular sieves. This term refers to their ability to selectively sort molecules based primarily on a size exclusion process. This is due to a very regular pore structure of molecular dimensions. The limiting pore diameters are roughly between 3 and 10 Å. At present, over 130 different framework structures are known.⁵  In addition to having silicon or aluminium as the tetrahedral atom, other compositions have also been synthesised, including the growing category of microporous aluminophosphates, known as AlPOs.

Zeolites are used in a variety of applications with world production estimated to be about 3 million metric tons per year. They are used in many organic reactions (cracking, isomerization, alkylation), water softening and purification, in the separation and removal of gases and solvents, agriculture, animal husbandry and construction. They can also serve as oxidation or reduction catalysts, often after metals have been introduced into the framework. Examples are the use of titanium ZSM-5 in the production of caprolactam, and copper zeolites in NO_x decomposition.

Consider, first of all, zeolites containing only OH or cations having very weak interactions with the xenon, such as Li⁺ or Na⁺. Most information is generally obtained by analysis of the variation of the chemical shift δ with the xenon concentration at different temperatures, usually at 300 K,^1,4  according to eqn (1.1).

δ₀ is the position of the isolated atom. This corresponds to virtually zero pressure. Figure 1.1 shows that each variation is characteristic of a zeolite^4,6  and defines two parameters, the term δ_s at zero concentration and the variation δ_Xe corresponding to Xe–Xe interactions, (δ_Xe–Xe), and which increases with the concentration, ρ_Xe, of adsorbed Xe as in the gas phase.⁷  δ_s depends on the dimensions of the cages or channels and on the ease of xenon diffusion in the crystallite. So the introduction of the mean free path, , imposed by the pore structure which is a parameter taking account of these two parameters. is defined as the average distance travelled by a Xe atom between two successive collisions against the pore walls.

Exchange between adsorbed and gaseous phases being negligible for microporous solids at room temperature,⁸  the experimental chemical shift, δ_s, of one atom is the average value of the shift of xenon in rapid exchange between a position A on the pore surface, defined by δ_a (with probability N_a), and a position in the volume V of the cavity or channel, defined by δ_v (probability N_v).⁹ 

δ_a depends on Xe-surface interactions. δ_v is a function of δ_a and the distance travelled between two successive collisions against the pore wall, ℓ. In fact, δ_v=δ_a when the atom leaves the surface; then δ_v decreases during the journey between two consecutive adsorption sites, whence the need to determine a mean value 〈δ_v〉=f(δ_a, ).^8,9  The plot of δ_s=f() for some classical zeolites is hyperbolic in shape, according to eqn (1.3) (see Figure 1.2-I):

λ and μ are fitting parameters. δ_a is related to the molecular interaction energy which depends on the surface curvature, hence on .¹⁰ 

The fast exchange model easily explains how the effect of the temperature depends on the pore size. 〈δ_v〉 is slightly lower than δ_a if the mean free path, , is small or, on the contrary, much smaller than δ_a if is large. The experiment temperature changes the residence time of Xe on the pore wall, so also the influence of 〈δ_v〉. Therefore, the greater , the larger the influence of T. For example, in the range 150≤T≤390 K, the variations of δ_s are 48 and 18 ppm for Y and ZSM-5 zeolites, respectively.¹¹ 

The chemical shift δ_s depends slightly on the Si/Al ratio. For example, when this ratio increases from 1.28 to 54, δ_s decreases monotonically by 4 ppm in the case of faujasite and by about 9 ppm for ZSM-5 and ZSM-11.⁴ 

The variation of δ_s with framework composition has been also reported for SAPO-37, AlPO4-5 and SAPO-5 molecular sieves.⁴  These deviations from the simple model are responsible for the fact that many structures studied, mainly with very low cation content, do not fit the first δ_s=f() curve; they form a cloud of points limited by a second curve for 0.05< <0.4 nm (see Figure 1.2-II).

Using the fast exchange model, Cheung et al.¹²  expressed the chemical shift δ_s of xenon adsorbed in Y zeolites at temperature T as:

where b=k_a/k_d is the ratio of the rate constants for adsorption and desorption of a xenon atom on a site, n_a the number of adsorption sites in a unit volume.

Cheung also determined an expression of δ_s(T) for different structures:¹³ 

where m=1: layer-like pore (clay); 2: cylindrical pore; 3: spherical pore.

L is the distance between leaflets (m=1) or the pore diameter, and r_Xe is the radius of the Xe atom. (L−2r_Xe)/m is the mean free path, , previously introduced experimentally.⁹  The terms A and B vary with the temperature, but the variation Δδ_s(T) of δ_s increases with the pore size, as was also shown experimentally.¹¹ 

Ripmeester and Ratcliffe considered the potential energy between a xenon atom and a spherical shell of radius R representing the cage wall.¹⁴  Oxygen atoms are smeared over this shell with a density similar to that of the zeolite A cage. Cheung made analogous calculations for a xenon atom trapped between two infinite parallel layers.¹³  From their calculations these authors concluded that for small cages or interlayer distances, δ_s values can be expected to reflect the true void space. However, for large cages or interlayer distances, δ_s is a complicated function of sorption energy, void space and temperature, the δ_s variations with temperature being a means of distinguishing the two types of behavior, in agreement with experiments.¹¹ 

¹²⁹Xe-NMR spectra have as many components as there are different types of void volume in the zeolite structure, at least if exchange between xenon adsorbed in the different zones is slow on the NMR time scale. This is the case of ferrierite at 300 K whose spectra have two lines corresponding to the two types of channels.⁴ 

The spectrum of xenon adsorbed on Rho zeolite depends on the nature of the cation and on the temperature. For H-Rho at 300 K, there is rapid exchange between cavities and prisms. The two characteristic lines are then obtained only at low temperature. On the contrary, for CsRho there is only one line, since the Cs cations are located in the prisms and prevent Xe atoms from being there. The flexibility of the Rho structure as a function of the temperature has also been studied.¹⁵  In the same way, the ¹²⁹Xe NMR spectrum of mordenite has two signals corresponding to one-dimensional channels connected to the side-pockets in the perpendicular direction, at least if the temperature is low enough to prevent exchange between these two sites.^16,17 

¹²⁹Xe NMR is no longer used to determine the pore structure of a perfect zeolite, except to refine certain surface details or to check the interconnection between pores. But it yields valuable information on the structure of the gel and the mechanism of the synthesis during the crystallisation of the zeolites.^18,19  Indeed Xe atoms are being used to probe the size of the cavities under formation, their modifications and the ultimate appearance of well-defined zeolitic materials. In addition, it can be used to follow the progression of the crystallinity.

This technique is also the best for the characterization of defects, whatever their nature. In the case of a mixture of zeolites or a structure intergrowth, each zeolite component gives rise to its own NMR lines in the spectra, provided that the diffusion of Xe between monocrystalline domains is not too fast and prevents the averaging of Xe–zeolite interactions; for example: synthetic mixture of Ca–A and Na–Y, ferrierite–mordenite intergrowth, AlPO4-8 structure in a VPI-5 sample, ZSM-5-ZSM-11 intergrowth in a ZSM-8 sample have all been analysed successfully.²⁰ 

Finally, this technique can characterize the structure defects, generally distributed randomly in industrial zeolites. For example, the two signals in Figure 1.3-left correspond to xenon in the supercages of a partly dealuminated Y zeolite and in a well defined secondary porosity corresponding to the volume of two adjacent supercages whose separation has been destroyed, respectively.²⁰  With the mordenite studied (see Figure 1.3-right) the decrease in the chemical shift of the signal reveals a significant increase in the diameter of the main pore; the shoulders at lower δ correspond to the presence of various undefined but relatively large defects.

The zeolites used in various applications most often contain xenon adsorption sites stronger than the surface, in particular more or less charged or voluminous cations. In this case the NMR spectrum depends on both the Xe-surface and Xe-other sites interactions. If the exchange between these various sites is fast with respect to the NMR time-scale (generally the case at 300 K), one average signal is observed, but the δ(T)=f(ρ) variation can provide much information about the nature of these sites and their distribution in the structure. To obtain more details it is enough to reduce the experiment temperature. Generally speaking the variation of the chemical shift with the adsorbed xenon concentration is given by eqn (1.6)

where the last three terms correspond to: the Xe–cation interaction, δ_SAS, the effect of the electrical field, δ_E, and, if it exists, the magnetic field, δ_M, created by these cations. These three terms are negligible with H⁺, Li⁺ or Na⁺.^20,21  The δ(T)=f(ρ) plots for KY and RbY are parallel to the previous ones but the shift increases with their bulk: δ_N→0 is 78 and 99 ppm, respectively. But the study performed on Y zeolites shows that δ depends not only on the number, the charge and the size of the cations, but also on their electronic structure.

In this case the large positive shift and the parabolic form of the δ=f(N) curves was attributed first to the distortion of the xenon electron cloud by the strong electric fields, δ_E, created by the 2+cations;^4,6,20  but also, later, to the formation of a partial bond between these two species formed by the donation of a xenon 5p electron to the empty s orbital of the divalent cations.¹²  The corresponding chemical shift δ_N→0 is greater than in the case of an uncharged structure (influence of δ_SAS). When the xenon concentration N increases, δ decreases if there is fast exchange of the atoms adsorbed on SAS with those adsorbed on the other sites. When N is high enough, the effect of Xe–Xe interactions again becomes the most important and δ increases with N. This technique is particularly interesting for detecting cations which are really accessible to other adsorbed molecules. For example, in the case of Na⁺Mg²⁺–Y or Ca⁺Mg²⁺–Y, the extent of exchange between Na⁺ and Mg²⁺ or Ca²⁺ has to be higher than 54% to be detected in the supercages after water desorption at 400 °C (see Figure 1.4, left).^6,20,21 

When the cations have the nd^x electronic structure there is a competition between two opposed effects: the positive chemical shift due to the polarization of the xenon electron cloud by the charge of the cations; and a negative shift due to a nd^x–5d⁰ donation from the cation to xenon during the short-lived cation–xenon complex. The resulting shift depends mainly on the charge of the cations, and also on their position in the pores. This effect is particularly important for x=10 such as Ag⁺, Cu⁺, Zn²⁺ and Cd²⁺ cations in faujasite-type structures (see Figure 1.4, right).²²  For example the shift in Ag–X is distinctly lower than that for Na–X over the range of concentrations studied, exhibiting negative values in the range −40 to −50 ppm at low xenon concentration. The case of fully exchanged CuY or CuX zeolites is particularly interesting since the Cu²⁺–Xe and Cu⁺–Xe interactions give opposite upfield and downfield shift, respectively.²¹  This differentiation is important in view of the interest of Cu⁺ in catalysis.

Xe NMR is particularly useful for the localization of La³⁺, Ce³⁺, Y³⁺ and Ru³⁺ cations in Y zeolite structure: supercages, hexagonal prisms or sodalite cages.^23–25 

Partially dealuminated zeolites often contain non-framework aluminium, Al_NF, in the pores and are generally detected by ²⁷ Al NMR. Xe-NMR has shown that the average charge on each Al_NF atom depends on the amount of Al_NF relative to the total number of Al atoms.^26,27 

Finally we mention the interest in ¹³¹Xe NMR of the 131 isotope for the quantification of the electric field gradient created by all charged species in zeolite pores.²⁸ 

Metal–Xe interactions are much more important than Xe-surface interactions. At very low temperature δ_s characterizes the Pt particle size. In the case of fast exchange at 300 K, the spectrum depends on the size, the concentration and the distribution of the particles in the zones accessible to Xe.^29,30  Finally, this technique makes it possible sometimes to demonstrate the bimetallic character of certain particles, as has been shown for Pt–Ir/Y based catalysts.³¹ 

Xe NMR is sensitive to the presence of a co-adsorbate; in particular to the type of its adsorption,³²  shrinking core model³³  or non-uniform model.³⁴ 

Xe NMR is also a particularly interesting technique for the study of the chemisorption of gases on zeolite-supported metals. Indeed, the metal–xenon interaction depends greatly on the nature of the chemisorbed gas and the particle coverage by the gas, as well as the size and distribution of these particles in the zeolite.^4,6 

Another application is the possibility to follow the poisoning of zeolites by chemical reactions; in particular, the deactivation of these solids by carbonaceous deposits (coke) during industrial cracking and their regeneration. In addition, Xe NMR is an efficient tool for locating coke inside or on the external surface of the zeolite. It can also reveal the role of extra-framework aluminium in catalytic cracking.⁴ 

There have been several studies of pillared clays by Xe NMR; for example: montmorillonites pillared by cationic silicon species or by alumina, SiO₂–TiO₂ sol, or hectorite pillared by tetraethylammonium ions. All authors obtained an average interlayer space in good agreement with XRD and porosity measurements. A more detailed review of these studies can be found in Gil et al.³⁵  The characterization of clays by ¹²⁹Xe NMR will be discussed in Chapter 9.

Such compounds, used as selective oxidation catalysts, are ionic solids. The most common are those whose anions have the Keggin structure. Xe NMR revealed a microporosity resulting from a translation and/or a rotation of the Keggin anion in the crystal when the size of the cation increases.³⁶ 

Terskikh et al.³⁷  have shown that Xe NMR makes it possible to study silica-supported heteropolyacids and to follow the clustering of these molecules “step by step”.

When the MOF structures are not affected by any adsorption or temperature variation the Xe NMR spectra show a monotonic increase in the chemical shift versus the xenon concentration which is typical for small pores in a rigid structure.^38,39  But very often local structural changes of the MOF can be induced by chemisorption–desorption and/or by temperature variation. Adsorption of the xenon probe molecule was the first direct means of pointing out the structural flexibility of these materials. For example the MOF MIL-53 exhibits a structural transition between two possible pore structures, the so-called large-pore (lp) and narrow-pore (np) forms, depending on the temperature or when guest molecules are adsorbed.⁴⁰ 

The characterization of an another MOF, Ni(2)(2,6-ndc)(2)(dabco), by gas adsorption and ¹²⁹Xe NMR reveals the reversible structural transformation of this compound without loss of crystallinity upon solvent removal and physisorption of several gases. ¹²⁹Xe NMR seems to be the best technique for the detection and characterization of the so-called “gate-pressure” effect in this MOF material.⁴¹  The use of ¹²⁹Xe NMR for the characterization of MOFs is further described in Chapter 12.

Besides the ordinary adsorption techniques, ¹²⁹Xe NMR is generally accepted for the characterization of zeolites, mesoporous and amorphous silica. In contrast, ¹²⁹Xe NMR has been only rarely used to study porous carbon materials. It has been mentioned in applications for the determination of the pore properties of activated carbons and coals,^42–46  carbon black materials,^47,48  various filamentous carbons including carbon nanotubes,^49–53  polyacenic semiconductor materials⁵⁴  and graphitized carbon.²  Most of these studies used experimental strategies derived from the ¹²⁹Xe NMR of more ordered porous solids. A few of them aimed at finding some specific ¹²⁹Xe NMR data for the characterization of the carbon surfaces.

In general, the ¹²⁹Xe NMR shift provides qualitative information and trends concerning the porosity and microstructure of amorphous carbons. However, in contrast to pure silicates, the interpretation of ¹²⁹Xe NMR data obtained for porous carbon is complicated by several factors: structural disorder – distribution of the crystallite size or presence of amorphous domains; heterogeneity of surface properties – presence of various surface groups and surface structures (basal and edge); and conductivity and strong paramagnetic sites. The paramagnetic properties of porous carbon materials have several origins: structural defects naturally present in most of such materials and admixtures of paramagnetic particles of inorganic nature.

As a consequence, the relationships between δ_s and the pore dimensions of carbons were not found. Hence the idea of using another parameter to characterize the pore dimensions, one which is free of the direct solid-surface interaction. This parameter can only be the second virial coefficient, δ_Xe–Xe, in eqn (1.1); this describes the interaction between xenon atoms in pores and, therefore, depends on the pore size and shape, which influence the xenon collision frequency. To overcome problems associated with the different types of carbons, Romanenko et al. prepared a series of activated carbons obtained by successive air oxidation/pyrolysis treatments of a unique precursor.⁴⁴  In this case, the chemical shift extrapolated to zero loading, δ_s, varies by 10 ppm over the pore size range 0.6–2.3 nm and without any monotonic behavior; this despite the fact that the chemical nature of the pore surface can be considered identical for all samples. Thus, δ_s cannot be used as a pore size probe. However, the chemical shift measured for these microporous carbons is a linear function of the xenon density, ρ_Xe, with a slope, δ_Xe–Xe, which varies very linearly with the pore size according to the equation:

where δ_Xe–Xe is expressed in ppm cm³ mmol⁻¹; D is the mean pore size in nm, κ=7±0.2 ppm cm³ mmol⁻¹ nm⁻¹ and η=5.1±0.3 ppm cm³ mmol⁻¹.

Nevertheless, this equation cannot be generalized since there is no other study of microporous carbons with the same surface homogeneity.

Romanenko et al. have also used the term δ_Xe–Xe for the characterization of mesoporous carbons. In this case the number of Xe atoms in the pore volume, N_V, is negligible compared to the number of atoms on the surface, N_S. Thus, the collision frequency at the surface is much greater than that in the pore volume. Consideration of the chemical shift as a function of the surface Xe density, ρ_surface, is more logical. So eqn (1.8) is analogous to the eqn (1.1) for this case:

The coefficient δ_Xe–Xe was shown to increase with the mean pore size in the case of micropores.⁴⁴  In the case of mesopores, the opposite dynamic was reported.⁵⁰  A simple model describing these data has been proposed.⁵⁵ 

In another respect, carbon-supported metals are advantageous catalysts from an economic standpoint, since the metal recovery is rather easy. Indeed, the carbon support can be burnt off, leading to highly concentrated ashes. As with zeolites, metal particles inside the pores or, on the contrary, on their external surface are easy to locate by Xe NMR.⁵⁶  An overview of these various applications has been published.⁵⁷ 

Finally, as with aluminosilicates, ¹²⁹Xe NMR is very useful for the characterization of micro–mesoporous carbons and the interconnection of the pores. For example, parallel studies with hyperpolarized ¹²⁹Xe of purely microporous carbon and mesoporous carbon, CMK-3, have unambiguously revealed the presence of micropores inside the CMK-3 structure.⁵⁸  In addition, ¹²⁹Xe 2D-exchange NMR experiments have clearly shown direct exchange of Xe atoms between the micro and the mesoporosity. This indicates the presence of micropores inside the carbon rods which make up the mesoporous structure of CMK-3 materials.

The ¹²⁹Xe-NMR of adsorbed xenon used as a probe is a very useful technique for the characterization of the properties of microporous solids such as zeolites, clays, metal–organic frameworks and even carbons. Indeed it is possible:

• to determine the dimensions and the form of their internal free volumes;

• to reveal and characterize the structure defects, and to calculate the short-range crystallinity, as opposed to that determined by X-rays;

• to locate cations in the zeolite structure and to define their interaction with adsorbed species;

• to characterize any encumbering species: adsorbed molecules, supported metal particles, coke formed during chemical reactions, etc.;

• to point out the structural flexibility of the metal–organic framework.

The author wants to thank Taro Ito from Sapporo who started with him on this long story and all members of the “Laboratoire de Chimie des Surfaces” of the University P. and M. Curie, Paris, who were his co-workers in this field during many years.