Magnetic Resonance Microimaging of Petroleum Coke - Energy

Conversion of petroleum coke into valuable products using oxy-cracking technique ... Magnetic Resonance Microimaging Studies of Porous Petroleum Coke...
1 downloads 0 Views 64KB Size
Energy & Fuels 1999, 13, 1109-1110

1109

Magnetic Resonance Microimaging of Petroleum Coke Eric B. Brouwer,*,†,‡ Igor Moudrakovski,‡ Keng H. Chung,§ Gerald Pleizier,† John A. Ripmeester,‡ and Yves Deslandes† Institute for Chemical Process and Environmental Technology, National Research Council, Ottawa, Ontario, Canada K1A 0R6, Steacie Institute for Molecular Sciences, National Research Council, Ottawa, Ontario, Canada K1A 0R6, and Syncrude Research Centre, 9421-17 Avenue, Edmonton, Alberta, Canada T6G 2G6 Received May 24, 1999. Revised Manuscript Received July 28, 1999 The structure of petroleum coke is fundamental to the understanding of its formation, role, and behavior in the operability and utilization of coking processes. Coke is heterogeneous in structure, varying widely according to petroleum feed and coking conditions. Coke structural studies have focused on diffraction, scanning and transmission electron microscopy (SEM, TEM), porosimetry, optical microscopy, nuclear magnetic resonance (NMR), and thermogravimetric studies.1 We present the first magnetic resonance microimaging (MRM) studies of coke, and discuss the impact of MRM in characterizing coke structure, formation, and in situ behavior. Void space images obtained of cyclohexane from density, T2-relaxation, and diffusion-contrast mechanisms show a significant interconnected interior pore system demonstrating both the permeability of the exterior shell layer and suggesting an agglomeration formation mechanism. Biological structures have been characterized by MRM with resolution of the order of 4-10 µm.2 In materials science, catalyst deactivation has been studied with resolution as low as 40 µm.3 In the application of MRM to coke materials, the ability to probe the 3-dimensional spatial characteristics of void spaces with both dynamic and chemical sensitivity is exploited. While scanning electron and optical spectroscopy give much higher resolution, these techniques are limited to static 2-dimensional images of the coke framework, and are not chemical-specific. Techniques such as NMR and vibrational spectroscopy are chemically and dynamically sensitive, but fail to convey the spatial distribution of these properties. MRM provide various images weighted according to the physical characteristics of the probe molecule: T2-relaxation, density, diffusion, and spatial * Corresponding author. Institute for Chemical Process and Environmental Technology, National Research Council, Bldg. M-12, Room 111, Montreal Road, Ottawa, Canada K1A 0R6. Phone: (613) 991-6347. Fax: (613) 991-2384. E-mail: [email protected]. † Institute for Chemical Process and Environmental Technology, National Research Council. ‡ Steacie Institute for Molecular Sciences, National Research Council. § Syncrude Research Centre. (1) (a) Jimenez Mateos, J. M.; Romero, E.; Gomez de Salazar, C. Carbon 1993, 31, 1159-1178. (b) Pruski, M.; Gerstein, B. C.; Michel, D. Carbon 1994, 32, 41-49. (c) Fortin, F.; Rouzaud, J. Fuel 1994, 73, 795809. (d) Pradhan, A. R.; Wu, J. F.; Jong, S. J.; Tsai, T. C.; Liu, S. B. Appl. Catal. A 1997, 165, 489-497. (2) (a) Callaghan, P. T. Principles of Nuclear Magnetic Resonance Microscopy; Oxford University Press: New York, 1991. (b) Kuhn, W. Angew. Chem. 1990, 29, 1-112. (3) (a) Hollewand, M. P.; Gladden, L. F. Chem. Eng. Res. Des. 1992, 70, 183-185. (b) Cheah, K. Y.; Chiaranussati, N.; Hollewand, M. P.; Gladden, L. F. Appl. Catal. A 1994, 115, 147-155. (c) Timonen, J.; Alvila, L.; Hirva, P.; Pakkanen, T. T.; Gross, D.; Lehmann, V. Appl. Catal. A 1995, 129, 117-123. (d) Manz, B.; Chow, P. S.; Gladden, L. F. J. Magn. Reson. 1999, 136, 226-230.

Figure 1. SEM image (75× magnification) of 4.1 mm diameter chunky coke broken in half to reveal cross-section. The internal structure reveals an agglomeration of small (50-150 µm) particles surrounded by a ∼100 µm thick shell.

distribution. In addition, a wide range of void-space molecular probessincluding model hydrocarbons to mimic in situ conditionssmay be used to generate complementary structural and dynamic information. The fluid coking process for upgrading Athabascaderived bitumen is optimized for fluid coke particles of 150 µm nominal diameter. Larger coke materials with diameters of up to 10 000 µmsthe so-called “peas and beans” and “chunky coke”sdefluidize the coke bed. The larger, unwanted chunky coke material appears to form via agglomeration of smaller particles with a radial increase in density distribution. The SEM image of a 4100 µm diameter chunky coke indicates the core is surrounded by a thin (∼100 µm) but dense shell (Figure 1). Questions of behavior, i.e., whether the shell is penetrable by hydrocarbons and if so, the interconnectivity of the inner pore spaces, are unanswered by this technique. Figure 2 shows a series of MRM images of cyclohexane in a chunky coke (similar to that in Figure 1) weighted according to different physical properties of the probe hydrocarbon.4 Cyclohexane is an excellent probe molecule since it exhibits a single 1H NMR chemical shift value, is a hydrocarbon (mimicking reactor conditions), has a relatively short T1-relaxation value, and gives high signal intensity. Each factor contributes to a high-resolution cyclohexane image. Both H2O and CH4 were used as imaging molecules, although with lower sensitivity and resolution.

10.1021/ef990096m CCC: $18.00 © 1999 American Chemical Society Published on Web 08/28/1999

1110 Energy & Fuels, Vol. 13, No. 5, 1999

Communications

Figure 2. MRM images of 4.1 mm diameter chunky coke obtained with cyclohexane as the probe molecule. The identical 300 µmthick slice is shown with 24 µm/pixel acquisition resolution to illustrate chemical and dynamic sensitivity: (a) intensity from multiple spin-echo experiment, (b) diffusion-weighted, (c) single spin-echo, (d) T2- relaxation-weighted images. Gray scales on the right of (b) and (d) correspond to linear change in diffusion constant D and relaxation time T2, respectively. The lightest regions correspond to the highest value of the respective parameters. D changes between approximately 1.4 × 10-9 m2 s-1 (white) and 5 × 10-10 m2 s-1 (medium of the scale) for the image (b), and T2 changes between 65 ms (white) and 9 ms (1/4 from the bottom of the scale) for the image in (d).

The contrast mechanism in Figure 2a arises from contributions from density, diffusion, and T2-relaxation. The contribution from each physical feature is shown in Figure 2b-d, demonstrating the power of the MRM technique to image according to different chemical characteristics. The 24 µm/pixel image resolution was obtained using a 9 h acquisition time, and is currently limited by the T1-relaxation time of 2 s. The image in Figure 2a most clearly shows the coke features determined by SEM. The inner core is composed of 50-150 µm particles with a pore system that is both large and interconnected. The dense shell (∼100 µm) shows very little C6H12 intensity, although the outer layer is obviously permeated by the liquid. Similarly, the smaller constituent particles show little cyclohexane intensity. Again, this (4) Experimental details: Samples for MRM experiments were prepared by impregnating approximately spherical chunky coke with water or cycloxehane, immersion (5 min) in an ultrasonic bath to minimize the amount of trapped air bubbles, and then loaded into a 5 mm NMR tube. Space in the tubes around the pellets was filled with wet gypsum in order to prevent evaporation of the probe liquids from the pellets and to minimize the signal arising from outside the coke. MRM images were acquired using a Bruker DSX-400 NMR spectrometer (9.4 T magnetic field, 400 MHz 1H resonance frequency) equipped with a Bruker MR2.5 microimaging probe. All images were acquired with spin-echo pulse sequences optimized for shortest spin-echo time (TE). Diffusion-weighting was introduced by a pair of diffusion gradient pulses located between the excitation and refocusing pulses applied along the slice-selection direction. For each image, 32 scans were acquired and averaged with a repetition time of 5 s for a total of ∼9 h. Slice selection was obtained using truncated Gaussian pulses, with a slice gradient of 32 G cm-1. Images were acquired in a square field of view of 5 mm in a 208 × 208 matrix, which was expanded to 256 × 256 during processing. Intensityweighted and T2-weighted images are results of fitting the images acquired with different TE with the single-exponential function I ) Io exp(-TE/T2). Intensities in the resulting images are proportional to spindensity Io and spin-spin relaxation time T2, respectively. Intensity in the diffusion-weighted image is proportional to the diffusion coefficient of cyclohexane in the coke framework.

structure is consistent with an agglomeration formation mechanism. The diffusion-weighted image (Figure 2b) shows distribution of diffusion coefficients of cyclohexane in the sample of coke (Dmax ) 1.4 × 10-9 m2 s-1). The average diffusion coefficients are approximately between one-half and two-thirds of that of the pure C6H12 liquid. The density-weighted image (Figure 2c) shows distribution of cyclohexane concentration throughout the lattice, and it essentially characterizes the amount of available void space. The T2-relaxation-weighted image (Figure 2d) indicates regions that show the most liquid-like behavior, with the maximum T2-value ) 61 ms. Comparison of the three images shows similarities in the distribution of the physical features. Notably, the intensity maxima show the same spatial distribution. The images indicate that there are significant pore volumes distributed throughout the inner core, and that these pockets of cyclohexane are essentially liquid-like. Work is currently in progress to analyze the pore size and volume distributions. In conclusion, magnetic resonance microimaging provides both chemical and dynamical images of hydrocarbons in the void spaces of petroleum cokes. The coke lattice clearly affects the physical characteristics (D, T2) of the probe molecule. The pore system shows a random pattern of interconnection, and the internal agglomerated structure shows little distribution in density. This technique shows promise in characterizing, with 10-20 µm spatial resolution, the structure and dynamics of different stages of coke formation, different types of cokes, and also petrochemical catalysts. EF990096M