NMR studies of ceria doped alumina powder catalysts

580

Abstracts / Magnetic Resonance Imaging 19 (2001) 569 –589

The present study demonstrates that, two MRI techniques, chemical shift imaging method and phase-encoding method, are effectively combined to measure stagnant immiscible liquid (silicone-oil) distribution and water velocity profiles, respectively. Multi-slice two dimensional measurements show that water velocity profiles drastically varies due to the change of the pore structure by the stagnant silicone-oil. Quantitative estimation has been obtained by a statistical method to clarify the relation between pore structure and fluid flow. Localized flows develop at relatively large pores in the case of water single phase flow, whereas in the case that water flow in porous media containing stagnant silicone-oil, the immiscible liquid transforms the effective pore structure in porous media and varies the water velocity distributions. It is clarified by quantitative pore structure and flow rate analysis that silicone-oil blockaded some large pores and flow rates at many small pores increase contribute to reduce permeability of porous media. REFERENCES [1] Gladden LF, Hollewand MP, Alexander P. AIChE Journal 1995;41: 894 –906. [2] Feinauer A, Altobelli SA, Fukushima E. Magnetic Resonance Imaging 1997;15:479 – 87. [3] Caprihan A, Fukushima E. Physics Reports 1990;198:195–235. PII: S0730-725X(01)00338-1 1

H NMRD dispersions of porous media: a model-free analysis Fragai Ma, Luchinat Cb, Nerinovski Ka, Parigi Gb. aUniversity of Florence, Via Sacconi, 6, 50019 Sesto Fiorentino, Italy; bUniversity of Florence, P.le delle Cascine 28, 50144, Florence, Italy. A model-free analysis of the 1H NMRD profiles of water adsorbed on different diamagnetic porous media have been performed to reproduce the experimental dispersions, stretched with respect to Lorentzians. The profiles show the values of water proton relaxation rate as a function of the applied magnetic field, from 0.01 to 20 MHz of proton Larmor frequency. Water proton relaxation, when dominated by homonuclear dipole-dipole interactions, is related to the Fourier transform of the correlation function, which exponentially decays to zero with a time constant given by the correlation time ␶ c . The Fourier transform of a mono-exponential decay in the time domain is a Lorentzian in the frequency domain [1]. Non-Lorentzian profiles are thus indicative of the presence of several correlation times modulating the dipolar interactions. Diffusion processes can also cause the presence of non-Lorentzian components. The model-free approach, proposed by Halle et al. [2] for protein systems, shows a good capability to fit the experimental profiles by using physical parameters, defined in the theory. In porous materials there are at least two pools of protons to be considered: bulk water protons, freely diffusing inside the pores, and protons diffusing at the porous surface interface. The latter relax much faster as subject to restricted motions. A distribution of porous sizes is then responsible for the different correlation times ␶ c modulating the proton dipole-dipole interaction. The stretched profiles have been reproduced by a sum of a finite number of Lorentzians, as customarily done in model-free approach used to analyze relaxation data on biomolecules in the presence of internal motion. In the model-free approach the parameters used to characterize the profiles are the average motional time scale, provided by the average 具 ␶ c 典 value, the integral of the dispersion profile, physically related to the mean-square fluctuation, the inverse of the frequency of half amplitude of the Lorentzian with the same low and high frequency limits and the stretching of the dispersion. REFERENCES [1] Bertini I, Luchinat C. Coord Chem Rev 1996;150:1. [2] Halle B, Jo´hannesson H, Venu K. J Magn Reson 1998;135:1. PII: S0730-725X(01)00339-3

Pore size distribution from hydrogen and sodium NMR using the transverse relaxation at 4.7 T Rijniers L, Pel L, Huinink HP, Kopinga K. Department of Applied Physics, Eindhoven University of Technology, The Netherlands. Salts in porous building materials can crystallize during drying, which may occur at the surface, causing defacing, or just under the surface, where it may cause structural damages, e.g., delamination, surface chipping or disintegration. Therefore a detailed knowledge of the moisture and salt transport is essential for understanding the durability of building materials. Up to now only a few studies have been reported on unsaturated moisture and ion transport in porous media, because suitable measurements techniques are lacking. About the issues which still have to be studied experimentally and theoretically are: chemical and physical interaction of ions with a material, supersaturation of ions in a pore system and the growth of crystals in pores [1]. For these studies it is important to know the ion distribution within the pore system. Up to now usually only the transverse relaxation of hydrogen is used to determine the pore water distribution [2]. In this study the transverse relaxation of sodium as measured at 4.7 T was used to determine the distribution of the sodium ions in a saline solution in various porous materials. First the transverse relaxation of hydrogen and sodium were compared for Nucleosil samples of different pores sizes. These measurements show clearly that also the transverse relaxation of the sodium can be linked to the pore size. The upper limit is set by the bulk relaxation of sodium. These measurements were also done for materials having a bimodal pore size distribution like mortar and calcium silicate brick. Also in these cases the sodium measurements give a clear bimodal distribution.

References [1] Scherer GW. Cement and Concrete Research 1999;29:1347. [2] Brownstein KR, Tarr CE. Phys Rev A 1979;19:2446. PII: S0730-725X(01)00340-X

NMR studies of ceria doped alumina powder catalysts Piras Aa, Trovarelli Aa, Plavec Jb, Dolcetti G.a aDip. Scienze e Tecnologie Chimiche, Universita` di Udine, Italy bNational Inst. of Chemistry, University of Ljubljana, Slovenia. Aluminum oxide is a porous material widely used as a support in heterogeneous catalyst application, such as catalytic combustion [1]. In order to avoid loss of the specific surface area, during high temperature conditions, different metal oxides could be added to stabilize ␥ and ␪-Al2O3 phases. It is well known that the addition of ceria (CeO2) to three way catalysts improves both performances and durability [2]. Unfortunately, stabilizing effects of cerium oxide are negligible at temperature higher than 1373K under oxiding conditions [3]. In the present work, the thermal and structural stability under different atmosphere of a series of CeO2/Al2O3 powder supports was investigated. Specific surface area values and pore dimensions, measured from N2 adsorption isotherms with the B.E.T. method, have been correlated with magnetic resonance parameters. Proton T1 and T2 relaxation times and Ms/M0 magnetisation transfer ratio were determined by filling pores with liquid water. Moreover a series of solid state MAS NMR 27Al spectra were acquired to follow alumina structure evolution. Finally X-ray powder diffraction data were collected to identify the presence of different phases. Our studies clearly show that CeO2/␥-Al2O3 system, treated under hydrogen at high temperature, prevents ␣-alumina formation and retards surface-area drop in a manner similar to that shown by lanthana under oxidizing conditions. Indeed, a surface area in the order of 60 m2/g can be retained after treatment at 1473K for several hours. T1 relaxation time measurements could follow the surface area evolution. This remarkable stabilization under reductive and redox conditions is due to the formation

Abstracts / Magnetic Resonance Imaging 19 (2001) 569 –589 of Ce3⫹, present as CeAlO3. From NMR data it seems that the CeAlO3 formation deeply influences the MT process. Moreover MAS spectra revealed a tetracoordinated, and probably a pentacoordinated, Al structure generally not present in this high temperature conditions.

References [1] Groppi G, et al. Catal Today 1999;50:399. [2] Trovarelli A. Catal Rev-Sci Eng 1996;38:439. [3] Iuga D, et al. J Phys Chem B 1999;103:7591. PII: S0730-725X(01)00342-3

A new numerical procedure for solving the Nuclear Magnetic Resonance relaxometry problem Barone Pa, Ramponi Ab, Sebastiani Ga. aIst. Appl. Calcolo “M. Picone”, C.N.R., 00161, Rome, Italy bDip. Mat. Pura e Applicata, Universita` di L’Aquila, 67100, Italy. We propose a new procedure to solve in a stable way the Nuclear Magnetic Resonance relaxometry problem based on several numerical methods both deterministic and stochastic. The measured signal s(t), t ⫽ t 0 , . . . , t n⫺1 is related to the distribution density f(T1) of the NMR relaxation times T1 within the sample by s共t兲 ⫽ 共m 0 ⫺ M ⬁兲

冕

⬁

f共T1兲e ⫺t/T1 dT1 ⫹ M ⬁ ⫹ ⑀ 共t兲

0

where ⑀ (t) is a zero mean i.i.d. Gaussian error with known variance ␴2. The problem consists of recovering the unknown constants m 0 and M ⬁ and the function f(T1) with the constraints f(T1) ⱖ 0 and 兰 ⬁ 0 f(T1) dT1 ⫽ 1. From a mathematical point of view, this problem is strongly related to the inversion of the Laplace transform, hence it is an ill-posed problem. To get a stable solution we consider three classes of methods. Each class has its own advantages and drawbacks. We therefore propose the following sequential procedure: 1. Compute a rough point estimate by the non-parametric statistical Prony’s method [1]. 2. Use the rough estimate as a starting point for a regularization method based on UP [2]. The regularization method is applied many times in sequence and selects automatically model hyper-parameters. 3. A stochastic Bayesian version of this method based on a MCMC algorithm also provides a variability estimate of the solution. 4. If f(T1) can be assumed to be composed by several “peaks”, a semiparametric mixture model can be used both in a deterministic and in a stochastic setup to further improve the properties of the estimates. It turns out that this approach can cope with several different kinds of patterns of f(T1), as shown by some examples on sinthetic data sets.

References [1] Barone P, March R. IEEE Trans Signal Proc 1998;46(9):2448 –57. [2] Borgia GC, Brown RJS, Fantazzini P. J Magn Reson 1998;132:65–77. PII: S0730-725X(01)00343-5

Optimum excitation and detection of NMR signal in static magnetic field gradient Reiderman A, Itskovich G, Krugliak Z, Beard DR. Baker Atlas, USA. NMR characterization of earth formations in-situ is a substantially gradient type of measurement where the NMR excitation volume depends on the spectral bandwidth of the radio-frequency (RF) pulse. This feature of

581

in-situ measurements suggests that, unlike a traditional laboratory CPMG experiment, for a given RF pulse power a shorter refocusing RF pulse with less than a 180° flip angle may be required to maximize SNR. We explore the optimum excitation condition and detection for relaxation measurement in a constant static field gradient. Our theoretical analysis includes modeling of the NMR signal and SNR for different refocusing pulse flip angles. The experiment was performed using a laboratory 5-MHz NMR system with a strong static magnetic field gradient. Both the theoretical studies and experiment consistently show that a flip angle corresponding to the maximum SNR is about 135°. The optimum flip angle gives a 33% lower DC power (P DC ) consumption compared to that of the 180° flip angle. When it is possible to trade DC power for SNR, a relevant criterion for the flip angle optimization is to maximize the ratio SNR/ 公P DC .. This criterion leads to an optimum refocusing flip angle of less than 100°. Two methods can utilize power to derive maximum benefits from using a low flip angle mode. One is based on the linear dependence of minimum attainable inter-echo spacing on the refocusing pulse width. Another effective way to utilize DC power is a multi-volume mode of the logging instrument operation [1]. Further SNR improvement in gradient-field measurements can be achieved by optimizing the integration window used for the echo detection in order to match the echo shape. For a constant static field gradient, the echo shape is predetermined by the refocusing pulse spectrum and by the flip angle. CPMG experiments in a static field gradient can exhibit systematic error in the transversal relaxation time measurement when the ratio between longitudinal and transversal relaxation times (T1/T2) is not 1 [2]. Our numerical simulation for different T1/T2 ratios shows that T2 estimates obtained with a low flip angle do not differ from those of a 180° flip angle by more than 2%.

References [1] Prammer MG, Bouton J, et al. SPE Annual Technical Conference, 1998. [2] Goelman G, Prammer MG, J Magn Reson 1995;A113:11–18. PII: S0730-725X(01)00344-7

NMR study of tortuosity during deactivation and decoking of a naphtha reforming catalyst Ren XHa, Bartusseck Ia, Bertmer Ma, Demco DEa, Stapf Sa, Blu¨mich Ba, Kern Cb, Jess Ab. aITMC, Lehrstuhl fu¨r Makromolekulare Chemie, RWTH-Aachen, Worringerweg 1, Aachen 52074, Germany bITMC, Lehrstuhl fu¨r Technische Chemie und Petrolchemie, RWTH-Aachen, Worringerweg 1, Aachen 52074, Germany. The deactivation and decoking of porous catalyst supports, e.g. naphtha reforming catalysts during the catalytic hydrocarbon conversion processes, are of great industrial relevance in the petrochemical industry and have received research attention over the years [1–3]. But several aspects of the fundamental mechanisms in these procedures are still poorly understood, which has hindered the simulation and optimization. The self-diffusion of bulk n-heptane as well as the diffusion of fluid in a commercial naphtha reforming catalyst (Pt-Re-Al2O3) have been investigated at different stages of the deactivation and decoking processes using an echo-recalled PGSTE sequence. The diffusion of n-heptane in a cylindrical catalyst pellet was shown to be isotropic. Based on these measurements the tortuosity was computed from the ratio of the diffusion coefficients of the confined and the bulk liquid. Because deactivation (coking) is accompanied by the growth of a coke layer on the surface of the catalyst supports, one observes a reduction of the accessible pore space and an increase of the tortuosity. On the other hand, decoking the catalysts during the regeneration process leads to a decrease in the tortuosity. While the coke which clogs micropore channels is removed first, an increase in