Exploiting enhanced paramagnetic NMR relaxation for monitoring catalyst preparation using T1–T2 NMR correlation maps

Heterogeneous catalysts are widely used in the chemical industry for the production of a variety of bulk and specialty chemicals, including fuels, polymers and fine chemicals. A large family of catalysts is that of metal and/or metal salts deposited on porous solid oxides, including silica, alumina, titania and other porous solids. A variety of methods have been developed over the years to prepare heterogeneous supported metal catalysts, which can be broadly classified as precipitation, deposition and impregnation methods. Impregnation methods are among the most widely used methodologies. In this procedure, the solid support is contacted with a solution, usually aqueous solutions, containing the metal precursor, usually a salt; the support is then aged for a short time, dried and calcined. Depending on the amount of solution used, two types of impregnation are usually distinguished; in the so-called “incipient wetness'” or “dry” impregnation, the volume of the solution containing the precursor does not exceed the pore volume of the support. The other method, usually referred to as “wet impregnation” involves the use of an excess of solution with respect to the pore volume of the support, which is slowly evaporated during the preparation. Once catalysts have been prepared, their characterisation is a key step in both understanding catalyst performance in reactions as well as optimising and improving preparation procedures. A variety of tools are nowadays routinely used for catalyst characterisation, each of them aiming at elucidating different physico-chemical aspects. Textural properties such as surface area, pore volume and pore size distribution are usually characterised with gas adsorption isotherms or mercury porosimetry, the former usually suited for micro and mesoporous catalysts, the latter more suitable for macroporous catalysts. Information on crystal structure are often obtained using X-ray diffraction (XRD), which can give insights into crystallinity, unit cell dimensions, crystal size and lattice parameters. Microscopy techniques such as transmission and scanning microscopy are very useful in order to understand surface morphology of both support and metal deposition. Other techniques have been developed including temperature-programmed studies, which can be useful to quantify metal oxidation state as well as adsorbed and/or deposited species such as coke, and a variety of spectroscopic techniques, including Raman, infrared (IR), and NMR. Most of the NMR methods used to characterise solid catalysts

Heterogeneous catalysts are widely used in the chemical industry for the production of a variety of bulk and specialty chemicals, including fuels, polymers and fine chemicals. [1][2][3][4] A large family of catalysts is that of metal and/or metal salts deposited on porous solid oxides, including silica, alumina, titania and other porous solids. A variety of methods have been developed over the years to prepare heterogeneous supported metal catalysts, which can be broadly classified as precipitation, deposition and impregnation methods. 5 Impregnation methods are among the most widely used methodologies. In this procedure, the solid support is contacted with a solution, usually aqueous solutions, containing the metal precursor, usually a salt; the support is then aged for a short time, dried and calcined. Depending on the amount of solution used, two types of impregnation are usually distinguished; in the so-called "incipient wetness'" or "dry" impregnation, the volume of the solution containing the precursor does not exceed the pore volume of the support. The other method, usually referred to as "wet impregnation" involves the use of an excess of solution with respect to the pore volume of the support, which is slowly evaporated during the preparation.
Once catalysts have been prepared, their characterisation is a key step in both understanding catalyst performance in reactions as well as optimising and improving preparation procedures. A variety of tools are nowadays routinely used for catalyst characterisation, 6 each of them aiming at elucidating different physico-chemical aspects. Textural properties such as surface area, pore volume and pore size distribution are usually characterised with gas adsorption isotherms 7 or mercury porosimetry, 8 the former usually suited for micro and mesoporous catalysts, the latter more suitable for macroporous catalysts. Information on crystal structure are often obtained using X-ray diffraction (XRD), 9 which can give insights into crystallinity, unit cell dimensions, crystal size and lattice parameters. Microscopy techniques such as transmission and scanning microscopy are very useful in order to understand surface morphology of both support and metal deposition. 10 Other techniques have been developed including temperature-programmed studies, 11 which can be useful to quantify metal oxidation state as well as adsorbed and/or deposited species such as coke, and a variety of spectroscopic techniques, including Raman, infrared (IR), 12 13 C, 29 Si, 27 Al and 129 Xe nuclei, which can be very useful in zeolite studies for example, as it is able to elucidate Al and Si distribution, acidity and in some cases porosity. 6,14 Magnetic resonance imaging (MRI) techniques have also been proposed to study heterogeneous catalysts and their preparation. For example, MRI with T 1 and T 2 contrast has been used to study the impregnation step during the preparation of Ni/γ-Al 2 O 3 hydrogenation catalyst pellets and understand transport and interactions of Ni precursors as well as metal distribution across single catalyst pellets. 15 We have recently shown that NMR relaxation time measurements can be used as a non-invasive, rapid tool to characterise adsorption and molecular dynamics of species inside catalysts pores, which can be related to surface characteris-tics of solid catalysts. [16][17][18][19] With this methodology we have been able to assess a variety of important aspects, including solvent affinity, 20 water-tolerance 18 and effect of mechanical treatments. 17 Prompted by our initial work, we had the idea to see if the technique can be exploited to develop new protocols to monitor deposition of metal precursors over solid supports during catalyst preparation.
In the work reported here, we have used two-dimensional T 1 -T 2 NMR relaxation measurements to characterise the evolution of the various surface adsorption sites on catalysts prepared by wet impregnation. In particular, we investigate wet impregnation of copper sulphate (CuSO 4 ) over alumina (Al 2 O 3 ) and probe the evolution of the observed surface adsorption sites by probing changes in NMR T 1 -T 2 maps of 1-octene, the probe molecule used for this study. Choosing Fig. 1 Two-dimensional T 1 -T 2 NMR pulse sequence. The thin (red) and thick (blue) vertical bars represent 90°and 180°radiofrequency (RF) pulses, respectively. T 1 relaxation is encoded in the variable time t delay . T 2 relaxation is encoded in the train of n 180°pulses. A single data point is acquired at the centre of each echo time, τ. The grey bars (HG) represent homospoil magnetic field gradients. this system, we show that we are able to distinguish different adsorption sites over the catalyst surface and their evolution as the concentration of salt used for the preparation increases. The use of short-chain liquid hydrocarbons has shown to give rise to narrow T 1 -T 2 peaks 17 as compared with other probe molecules, particular compared to those with higher viscosities such as higher alcohols and polyols. 21 This, together with the paramagnetic nature of CuSO 4 , which is expected to shift and decrease significantly T 1 and T 2 values of molecules in very close proximity of paramagnetic sites, 22 relative to those interacting with the Al 2 O 3 surface, is able to produce T 1 -T 2 maps with clearly distinguishable and wellseparated peaks associated to different adsorption environments, which can therefore be unambiguously analysed and quantified to provide new insights into adsorption site evolution as a function of metal precursor loading. The γ-Al 2 O 3 used for the experiments was supplied by Johnson Matthey. BET and BJH analysis were carried out in order to obtain the textural properties of the porous oxide, which has an average pore size of 12 nm, a pore volume of 0.52 cm 3 g −1 and a surface area 90 m 2 g −1 . Samples of γ-Al 2 O 3 doped with CuSO 4 were prepared with a procedure similar to what has been reported in the literature for similar catalysts. 23 In more details, alumina particles were dried in the oven at 105°C for 3 hours and then added to aqueous solutions of copperĲII) sulphate pentahydrate at known composition. The particles were stirred and left within the solutions for at least 24 hours. The moist solids were removed from the solution after impregnation and then dried in an oven at 70°C for 2 hours and at 150°C for further 2 hours, stirring several times during the drying process in order to ensure a more homogeneous drying. The actual content of paramagnetic CuSO 4 salt inside the γ-Al 2 O 3 particles was estimated by knowing the CuSO 4 concentration of the aqueous bulk solution used for the impregnation and the pore volume of a known amount of solid used for the sample preparation. The deposition of CuSO 4 and associated increase in the paramagnetic nature of the samples due to the presence of paramagnetic Cu 2+ ions has been confirmed in magnetic susceptibility measurements previously reported. 24 NMR relaxation experiments were performed on a Bruker DMX 300 operating at a 1 H frequency of 300.13 MHz using a T 1 -T 2 saturation recovery pulse sequence, which comprises a saturation recovery part to encode T 1 (using a comb of 90°p ulses) followed by a Carr-Purcell Meiboom-Gill (CPMG) echo train of 180°pulses to encode T 2 . The sequence is schematically shown in Fig. 1. The T 1 recovery interval, t delay , was varied logarithmically between 1 ms and 10 s in 32 steps. The echo spacing between the 180°pulses of the CPMG was set to 250 μs.
Two-dimensional T 1 -T 2 maps for bare Al 2 O 3 and wet impregnated CuSO 4 /Al 2 O 3 are reported in Fig. 2 for samples with different amount of CuSO 4 , reported as concentration of paramagnetic ions [Cu 2+ ]. In the case of pure Al 2 O 3 , [Cu 2+ ] = 0 ppm, a single peak, denoted as peak I, can be observed, which is attributed to the 1-octene interacting with the surface of the Al 2 O 3 support. Typical values of T 1 and T 2 for this peak, shown in Fig. 3, are similar to those reported for other hydrocarbons on mesoporous oxide supports. 17 With the introduction of small amount of CuSO 4 , a new peak, denoted as peak II, appears in addition to peak I, the latter attributed to Al 2 O 3 and clearly distinguishable. This is clearly visible when [Cu 2+ ] = 15 ppm. As the CuSO 4 concentration increases, the relative intensity of peak II, compared to that of peak I, increases significantly and, for concentrations of [Cu 2+ ] = 184 ]. Peak I is associated to 1-octene interacting with Al 2 O 3 whilst peak II is that of 1-octene interacting with CuSO 4 . The paramagnetic nature of Cu 2+ ions leads to much smaller values of T 1 and T 2 NMR relaxation times of peak II. View Article Online ppm and above, peak I disappears and only peak II remains visible. It is clear from the results that the new peak II appearing upon wet impregnation with the salt can be attributed to 1-octene in close proximity to new adsorption sites being formed, the latter being created upon CuSO 4 wet impregnation, and clearly distinguishable from the peak of 1-octene interacting with the Al 2 O 3 , peak I. This assumption is strongly supported by the T 1 and T 2 values of peak II, which are reported in Fig. 3 and compared with values for peak I. Paramagnetic species are well-known to be strong relaxation sinks, which enhance significantly relaxation rate of probe molecules, hence decreasing values of relaxation times; 25 therefore, it is expected that T 1 and T 2 values of molecules interacting with CuSO 4 relaxation sinks will have much shorter T 1 and T 2 values compared to the same molecules interacting with the Al 2 O 3 surface, which is indeed the case when comparing T 1 and T 2 values in Fig. 3. Further evidence that peak II is associated to CuSO 4 sites can be obtained by analysing the single values of T 1 and T 2 of this peak as a function of Cu 2+ concentration, see Table 1. As the Cu 2+ concentration increases, single values of T 1 and T 2 of peak I, associated to Al 2 O 3 sites, remain approximately constant whereas for peak II, associated to CuSO 4 sites, such values experience a significant drop above [Cu 2+ ] = 84 ppm, which can be associated to enhanced relaxation due to an increasing amount of paramagnetic Cu 2+ sites, as expected from the theory of relaxation in the presence of paramagnetic sinks. 26 Interestingly, this concentration corresponds to the value above which peak I, associated to Al 2 O 3 sites, becomes negligible compared to peak II, associated to the presence of paramagnetic Cu 2+ ions as CuSO 4 species.
The simultaneous presence of these two peaks in the concentration range [Cu 2+ ] = 15-84 ppm also suggests that for CuSO 4 /Al 2 O 3 samples in this range of concentration there is a heterogeneous distribution of adsorption sites over the surface, those attributed to Al 2 O 3 and those due to CuSO 4 , in the form of a "patchy" catalyst surface with two clearly distinguishable regions, quantitatively comparable, as can be shown in Fig. 4, which reports the area ratio of peak I, associated to Al 2 O 3 , over peak II, associated to CuSO 4 , as a function of [Cu 2+ ]. The ratio decreases with the Cu 2+ concentration in a linear fashion, which is a further confirmation of the peak assignment made. This information can clearly be related to site accessibility, hence to the relative amount of molecules interacting with each site. From values of 184 ppm and higher, the peak associated to the Al 2 O 3 support, peak I, disappears completely and this suggests that at this point the CuSO 4 salt has covered most of alumina surface, hence the hydrocarbon has a limited access to purely Al 2 O 3 surface sites and will mostly feel the influence adsorption sites due to CuSO 4 , or at the least that the amount of probe molecules interacting with the Al 2 O 3 sites becomes negligible relatively to those interacting with the CuSO 4 sites. Indeed, at higher concentration, only peak II remains prominent.
One other point to note is on the values of the T 1 /T 2 ratio, which can be related to an adsorbate/adsorbent surface affinity, for 1-octene on both Al 2 O 3 and CuSO 4 sites. Across the whole concentration range T 1 /T 2 ∼ 3 for Al 2 O 3 and T 1 /T 2 ∼ 2 for CuSO 4 sites. This suggests that the affinity of 1-octene towards CuSO 4 is slightly lower compared to that towards Al 2 O 3 and this could be due to the fact that CuSO 4 is likely to be present in the form of hydrate, hence having a more hydrophilic character and a lower affinity with the hydrocarbon.
In summary, the finding reported here shows for the first time the ability of NMR relaxation methods to monitor surface evolution during catalyst preparation, particularly catalysts obtained by deposition of paramagnetic metal salts. We believe that these results open up new possibilities in characterisation and understanding of catalyst preparation, in particular evolution of surface sites and quantification of surface interactions of chemical species with the different surface sites. Future work on this topic will focus on quantitative aspects of this approach.

Conflicts of interest
There are no conflicts to declare.