Catalyst Manufacturing Science and Engineering Consortium (CMSEC)

Rutgers University New Jersey, U.S.A.

Rutgers Catalyst Manufacturing Science and Engineering Consortium (CMSEC)

Rutgers Catalyst Consortium since 2003 While a large segment of the US industry, including the petroleum, chemical, pharmaceutical, automotive, and energy industries makes and/or uses catalysts, there has been no academic program focusing on the operations required to make catalytic materials. Thus, catalyst manufacturing processes are often designed relying on empiricism, leading to uncertain/suboptimal processes, decreased quality, and increased cost. By combining the substantial level of expertise in particle technology, optimization, multi-scale simulation, catalysis and molecular modeling available at Rutgers, we are developing and promoting science-based methods for designing and optimizing catalyst manufacturing methods and processes such as impregnation, drying, slurry mixing, extrusion, calcination, etc. This is integrated with a number of educational activities including research training of undergraduate students, graduate students and postdoctoral fellows in the area of Catalyst Manufacturing. A combination of experiments and computer models has been used to improve the understanding and performance of the unit operations used to make catalysts. This knowledge, and the resulting methods and practices, have been transferred to Industrial partners to help transform this important segment of the global economy into a science-driven enterprise. The specific goals of the consortium are: To create a world-wide prime Center of Excellence in catalyst manufacturing research. To enhance fundamental understanding of catalyst manufacturing operations. To develop new technology for efficient catalyst manufacturing. To provide a research, development, and education resource for industry. The consortium was founded on October 1, 2003. The Rutgers faculty that have participated in this consortium and their expertise areas are: Benjamin J. Glasser (drying, fluidized beds), Johannes G. Khinast (drying, catalysis), Fernando J. Muzzio (particle technology, flow, mixing), Arthur Chester (industrial liaison, extrusion), Silvina M. Tomassone (molecular dynamics, interfaces), Troy Shinbrot (particle technology, mixing), and Alex Neimark (nanoporous materials, adsorption). The annual fee for membership in the consortium is $35,000. These are unrestricted funds for research and primarily support graduate students and postdoctoral fellows. The consortium members (companies and faculty) meet every 6 months (September and March) for project reviews. In the September meeting, the consortium decides on continuing with current projects or starting new projects. Where consensus is not possible, a vote is taken, with each company getting one vote. Copies of the slides from the talks at the consortium meeting are made available to the consortium members. In addition, each project provides a written report on progress every six months. A list of possible manufacturing operations of interest to the consortium is given below: Powder blending, Powder sampling, Calcination, Slurry Mixing, Pellet compaction, Roller Compaction, Powder segregation, Catalyst drying, NonNewtonian Liquid Mixing, Powder Flow/Handling, Multiphase Flow/CFD, Dispersion of solids, Impregnation, Mulling, Extrusion , Nano-templating, Fluidized bed coating, Granulation, Catalysis, Spray Drying, Atomization, Flash/Rotary Drying/Calcination, Plow-shear mixing, Liquid-Solid Separation (Filtration/Flocculation), Precipitation Catalysts. Projects that are currently funded by CMSEC are: Impregnation of Catalysts Drying of Supported Catalysts Powder Segregation Powder Flow, Handling and Characterization Projects that have been funded by CMSEC in the past include: Calcination Mulling/Extrusion of Catalyst Supports Slurry Mixing Continuous Powder Blending A description of projects that have been funded by CMSEC follows.

Current and Past Industrial Members: Albemarle Chevron Haldor Topsoe Johnson Matthey Exxon-Mobil CRI/Criterion Inc. Engelhard Grace UOP

Projects: Impregnation of Catalysts Drying of Supported Catalysts Powder Segregation Powder Flow, Handling and Characterization Calcination Mulling and Extrusion Slurry Mixing Continuous Powder Blending Contact:
Prof. Ben Glasser Tel: 732-445-4243 Fax: 732-445-2581 bglasser@rutgers.edu

Impregnation of Catalysts

Catalyst impregnation is a crucial process in the preparation of industrial catalysts. In this process, metal salts or complexes are dissolved in a typically aqueous solution and contacted with a porous oxide catalyst support such as alumina (Al2O3) or silica (SiO2). During a contact time of typically 30-60 minutes the metal is adsorbed from the solution onto the high surface area support. The catalyst is dried and further pretreated to transform the metal from its precursor state into its active form. Generally, the process of impregnation is performed in rotating vessels with one or more nozzles that distribute the solution with the metal precursor into the catalyst support. The process is typically referred to as dry impregnation as the amount of Figure 2: Experimental impregnation study using 3 x 6 mm liquid sprayed onto the catalyst bed is alumina pellets dyed for a concurrent mixing/impregnation study. kept at a minimum, and thus, the bed continues to operate as a granular bed. water content as a function of time in specific spray The ability to control dry impregnations and zones. In addition, mixing experiments are establish effective models for large batches can performed to better understand the relationship significantly reduce the amount of time required per between spray flow rate and its effect on the batch while simultaneously achieving a more catalysts bed distribution; an example is shown in homogeneous distribution of solution. Figure 2. The main goals of this project are to establish 1) how mixing and flow are affected when the particles have a certain degree of moisture or are In this work, a saturated with liquid, 2) whether the liquid is combination study homogeneously distributed into the powder bed and incorporating a 3) the extent and distribution of dead zones for a series of given impregnator configuration and 4) to establish experiments as well a scalable model that can be quickly applied to as particle dynamic impregnation processes for optimal mixing and simulations are liquid distribution. performed using a double cone impregnator in order to obtain a deeper Drying of Supported Catalysts understanding of the mixing and Figure 1: Particle dynamic Supported catalysts are essential components in a simulation of double cone impregnation variety of industrial processes, ranging from mechanisms. A impregnator. A fan shaped catalytic converters to production of new drugs. The spray pattern is shown. computational performance of a catalytic process is intimately study is performed using the discrete element related to the catalyst design - uniform, egg-yolk, method (DEM). Commercial software (EDEM, egg-shell and egg-white metal profiles. It is generally DEMSolutions Inc.) is used to facilitate technology believed that the metal profile is controlled by the transfer. Algorithms are being used to allow for the

spraying of water droplets on a flowing powder bed and the allowance of catalyst support particles to absorb the water, thus increasing weight and density; Figure 1 shows an example of fluid spraying onto a powder bed. Furthermore, cohesive properties have been taken into account and allow for a linear increase as a function of maximum support capacity. Experimental techniques are used to validate computational models by measuring

conditions that are applied during impregnation where the metal contacts the solid support for the first time. However, experiments have shown that drying may also significantly impact the metal distribution within the support. Therefore, to achieve a desired metal profile we need to understand both impregnation and drying. Controlling the drying conditions can enhance catalyst performance. The goal of this project is to develop a fundamental understanding of unit operations during catalyst preparation, and predict, control and optimize metal distribution and dispersion of supported catalysts. Therefore, we can provide our partners with efficient tools to monitor and control the final quality of supported catalysts. In this work we have developed a theoretical model for drying which we have validated experimentally. In this model, we have taken into account heat transfer from the hot air to the wet support, solvent evaporation in the support, convective flow towards the support external surface due to the capillary force, and metal diffusion and metal deposition due to adsorption and crystallization (see Figure 3a). In general, the convective flow is the main driving force to transport the metal component and the solvent towards the support external surface (t=500s in Figure 3b), while the back-diffusion causes metal to transport towards the support center (t=1000s in Figure 3b). Experiments have been carried out to measure the distribution and dispersion of metal in catalyst supports in order to validate our modeling work. In Figure 3c, the metal profiles after drying are plotted from the support center to the surface for Ni/Alumina catalysts, where symbols represent experimental data and solid lines represent simulation results. It is clear that a nearly uniform distribution is observed for low drying temperature while egg-shell profiles can be greatly enhanced with increasing the drying rate. In this project, we have examined the distribution and dispersion of Nickel, Copper, Barium, and/or Palladium on Alumina and Silica. We are interested in investigating the importance of the physical properties of the solid support (porosity, pore size distribution, particle size) and liquid solution (pH, ionic strength, initial metal precursor

concentration), and the nature of interactions that exist between the dissolved metal and the solid support (physical adsorption, crystallization, ion exchange, film-breakage, pore-blockage) on the distribution and dispersion of the active metal.

dispersed gas (hot gas + vapor solvent) precipitated or adsorbed solute solid support

heat supplied by the hot gas

solvent evaporation

liquid phase (solvent +metal) convective back-diffusion flow dissolved solute (salt or metal) (capillary flow)

(a)
Metal Profile after impregnation t=0s t=100s
t=500s End of drying t=1000s

(b)

(c)
Figure 3: (a) drying mechanism, (b) simulation of the evolution of the metal distribution during drying, (c) comparison of simulation and experiments for drying of Ni/Alumina catalysts.

Powder Segregation

separated material can be as much as 90% pure in either example.

In the production of catalysts, granular materials In this project, we focus on developing a predictive and powders (e.g. silica, alumina) which differ in understanding of conditions under which size, density, shape and surface properties are often segregation will and will not occur in tumblers, blended together. When such granular materials or shear cells, and discharge devices. We are powders are transported, tumbled or discharged, particularly interested in providing measurement they often exhibit a tendency to spontaneously techniques to establish in advance whether a separate. The goal of this project is to investigate material will separate, and in engineering control the segregational tendencies of granular materials approaches to modify either equipment or material and powders used in catalyst manufacturing. We choices to minimize segregation in catalysis seek especially to evaluate segregation in side-byoperations. side simulations and experiments, matched by size, interaction properties and geometry. The specific aims of this project are: 1) to generate a predictive understanding of segregation mechanisms 2) to provide strategies for deliberate control over segregation vs. mixing states 3) to generate predictive rules for when segregation or mixing will occur. More broadly, through the execution of this project, we seek to (a) (b) analyze the interplay between particle Figure 4: Spontaneous granular segregation. (a) Separation of differently sized interactions on the small- grains occurs in a double cone blender (shown at top). Depending on fill level scale and aggregate and and rotation speed, multiple segregated outcomes are possible: small-out, shown at left-bottom, small-in, at right-bottom, or half-and-half, shown at right-bottom. mixing states on the macroThe mechanism here for ideal, laboratory, materials is relatively well understood scale, thus improving the and relates to a competition between inertia of the tumbling particles and sifting control over particle of the smaller particles beneath the larger. (b) Separation of mechanically properties and additive identical particles can also occur due to minor changes in surface coatings, as choices that optimize flow illustrated in this example where differences in surface dyes on irregular silica causes strong spontaneous segregation during feeding. and mixing performance. Solids separate based on differences in size, shape, density or surface composition. As two examples, in Figure 4(a) we show separation of spherical particles tumbled in a double cone blender, and in Figure 4(b) we show separation of irregular silica particles fed from a vibratory feeder. In the latter example, the grains are identical in size, shape and bulk composition, but differ in surface coating leading to promote spontaneous electrostatic separation. In both cases, the segregation mechanism is subtle, but the result is profound and indisputable: the

Powder Flow, Handling and Characterization

The goal of this project is to develop a fundamental understanding of granular and powder flow and shear properties, so that the behavior of intermediate and final catalyst products during manufacturing and processing can be predicted and controlled. The techniques and methods investigated in this project could provide our partners with valuable tools and ideas to efficiently design and scale catalyst manufacturing processes.

techniques. This project is also set to investigate the ability of the characterization methods to predict the powders process behavior by examining the flow from bench scale hoppers. The study has found that there is a discernible correlation between the measurement techniques (shown in Figure 5) in characterization of catalyst blends. We also confirmed that the generally accepted

Powders and granular materials can be found in many processing steps in catalyst manufacturing; they exhibit a variety of flow patterns, and their state and behavior differs from application to application. (a) (b) Since there is a lack of fundamental understanding of powder behavior, Figure 5: Images of (a) Gravitational Displacement Rheometer (GDR) multiple problems can be and (b) shear cell module. encountered during production, characterization methods, such as shear cell, have such as jamming of hoppers, sub-standard blending range limitations, when evaluating extremely performance, and weight variability of final products cohesive or very free-flowing powders and such due to segregation and/or agglomeration. Scale-up methods are not always applicable in the case of can also be a challenge, since the lack of constitutive unconfined flow conditions. It is also of interest to equations for granular materials forces most scaleinvestigate how various material factors change the up efforts to follow the trial-and-error route. There flow properties of the catalytic powders, such as are numerous methods to characterize the flow presence of moisture or flow-modifying agents. properties of granular materials, such as avalanching testers, fluidizers, shear cells, indicizers, density methods, angle of repose, etc.; however, most of them are application-specific, and it is not clear, how they correlate with each other or with process performance. For this reason, the use of Calcination most of these testers is restricted to a specific application, for which they were designed, and any Calcination is a thermal treatment process applied attempts to apply the results of such experiments to to ores and other solid materials in order to bring a different application frequently result in process about a thermal decomposition, phase transition, or problems. removal of a volatile fraction. Calciners can be of In this work, we have created a family of standard catalyst materials, spanning a wide range of flow properties from very cohesive to free-flowing. We then used the characterization equipment to investigate the flow properties of these materials and to determine the correlations between the batch type as well as continuous type. Calcination is a crucial operation in catalyst manufacturing. Developing better process level understanding of this operation can significantly improve the quality of the end product as well as save energy and material costs. However due to the nature of the

operations, it is not always practical to do detailed and baffle effects on heat transfer, effect of radiation experimental studies. There are limitations due to material and device availability, high temperature requirements, sampling issues, time and personnel related costs. Computer modeling and process simulations have made great progress in recent years. It is now possible (a) (b) to design realistic models of devices and Figure 6: (a) Side view of the aluminum calciner. 10 thermocouples were inserted simulate their within the calciner through the Teflon made side-wall. (b) shows the complete performance at actual experimental set-up. Calciner rests on the computer controlled rollers. The thermocouples were connected to the Omega 10 channel data logger, which is hooked scale. The discrete to the computer. element method (DEM) heat and material wetness. Experimental studies can based force models have also received significant be performed to validate the simulation methods as enhancements which not only ensure better dynamic was done in the case of a rotary calciner as shown in modeling but also Figure 6. The corresponding simulation results are allow for heat and shown in Figure 7. humidity transfer algorithms. A In addition to rotary calciners, continuous calciners commercial DEM can also be studied using such simulations as shown software package in Figure 8. The parametric space and flow/mixing called EDEM by dynamic is partly different for a continuous calciner. DEMSolutions Inc. The device tilt angles, rotation speed, particle size has simplified such and wetness determine the flow behavior. DEM modeling and models can simulate a number of cases to arrive at Figure 7: Axial snapshot of simulation optimal values of such parameters and the associate problems. Ability to color coded particles in the effect on the output particles. use extended user calciner from simulation. (blue: low temperature ; The following are the key observations from the defined dynamics green: intermediate and studies performed in this project so far. allows complex red: high temperature) calculations such as 1) Detailed studies of heat transfer mechanisms interparticle and in rotary calciners were done using DEM particle-boundary heat and mass transfer. Such simulations validated with experimental models allow detailed studies spanning a wide data. The vessel rotation speed was found to parametric space and elucidate the effect of the have no significant impact on the heat various critical parameters such as mixing rates, Figure 8: Particle dynamic simulation of heating of particles in a rotary continuous calciner. calciner wall

transfer at both low and high Froude number regimes. The fill level had a moderate effect on heat transfer, suggesting that higher fill levels might be effective in practice. 2) Average wall-to-solid heat transfer coefficient was estimated based on experimental observations for use in simulations. 3) The addition of baffles in the rotary calciner was found to have a significant effect on the heat transfer due to improvement in the mixing. Furthermore, heated baffles enhanced the calcination significantly as compared to adiabatic baffles.

in the preparation of catalytic supports, high temperature ceramics and composites.

4) Radiation from the calciner walls was found to have a significant effect on Figure 9: Porosity analysis obtained from Non-Local calcination. Walls with higher emissivity Density Function Theory (NLDFT). affected the average bed temperature as In this project, a laboratory high shear mixer was well as the distribution of temperature in the used to prepare the pre-extrusion samples, which particle bed. Cohesion in the particle also were then extruded with a simple Carver press; had a significant effect on the temperature powders and extrudates were characterized both uniformity of the bed. In highly cohesive prior to and after calcination as a function of mixing cases, the particle drying happened in two time by a variety of techniques, including FTIR, distinct time-separated regimes as opposed Powder XRD, high resolution porosimetry and to linear drying in free flowing materials. paricle size analysis, among others. Not much A short study on continuous calciners was performed. The effect of speed was found to be significant. literature is available with regards to the fundamental chemistry occurring during alumina mulling prior to extrusion using a high shear mixer. The purpose of the high speed granulation is to improve the properties of the feed powder. It can improve handling properties of a powder, reduce dust formation, eliminate segregation of two powders, improve powder solubility, and improve the extruded product properties-porosity and strength. Powder XRD line widths suggested that crystallite size was increasing during the high shear mixing operation, even though average particle sizes were significantly decreasing. Of particular interest, porosimetry indicated that the very high shear (Fukae) mixer used in the lab produced a bimodal pore distribution, whereas a larger scale industrial mixer (Eirich) did not, as shown in Figure 9.

Mulling/Extrusion of Catalyst Supports

Mulling/Extrusion is one of the most efficient techniques for producing objects with constant cross section and with high symmetry such as rods, tubes, honeycomb structures and channels. The extruded products find applications in catalyst supports, heat exchanger tubes etc. Most catalysts used in the petrochemical and petroleum industries consist of metals supported on -alumina pellets or spheres. The -alumina is formed by calcination of boehmite (-alumina monohydrate, AlOOH) at 500-550C. Boehmite is one of the most widely used precursors

Slurry Mixing
The goal of this project is to study the mixing dynamics of slurries used to make catalyst supports. It is important to have homogeneity throughout the mixing domain so that all the catalyst supports have the same properties. To this end, it is necessary to characterize exactly what is going on inside these high solids loading, low moisture fluids. This will help optimize available systems as well as potentially help toward designing new ones. In this project, we have produced experiments and simulations to assess rates, optimal conditions, and

Newtonian fluid formulations. Figure 10 shows typical results from numerous simulations and experiments performed through the RutgersIndustry collaboration. Findings of this research include the following. Experiments reveal that zeolite and kaolin powders do not mix well in suspension, and appear to exhibit tendencies to separate under vigorous agitation. A transparent surrogate material was developed with similar rheological characteristics to those seen in slurries of interest.

(a)

(b)

(c)

(d)

Figure 10: Typical experiments and simulations of catalysis mixing. (a) Stirred tank geometry using pitched blade impellers. Both experiments and simulations share the same geometry, and both use nonNewtonian fluids exhibiting the same power law stress-strain rate dependence. (b) Computational Fluid Dynamic (CFD) simulation of dye mixing in Poincare section: i.e. intersections of fluid trajectories striking a plane through the center of the tank (hatched plane in panel (a)). Notice that a compartment of flow is present near the central axis of the tank, within which mixing is slow and contained. (c) 3D CFD of flow following injection of a bolus of material near bottom of tank. The bolus (red) is transported around the tank as the impellers turn, but vertical transport is considerably slower. (d) Multiple dye injection experiment performed to validate simulations confirms that there is a central compartment (red dye) that is separated from a second outer compartment (green dye). Mixing within either compartment is comparatively fast, however mixing between compartments is considerably slower. This compartmentalization persists across scales in both simulations and experiments. scaling relations for the mixing of slurries of - Computational Fluid Dynamic simulations materials relevant to catalysis manufacture. were performed to analyze mixing rates, Research focused on mixing in tanks stirred with dependencies on injection location (for pitched blade impellers, and materials studied example of binders or other additives), and include zeolite, kaolin, and power-law nonscale-up behaviors of slurry mixing.

Experiments and simulations both produce similar compartmentalized flow with strongly separated internal and external mixing regions. Although mixing within each compartment is comparatively efficient, a bottleneck between the compartments renders mixing between compartments nearly nonexistent. Empirical scaling relations were produced to permit relating bench-scale experiments to pilot-scale and larger mixing. It was determined that by by varying impeller speed as well as mixing time, one can achieve dynamical similarity in large as well as small mixing tanks of similar geometric ratios.

NIR analytical methods was developed as well as analytical methods suitable for online monitoring. The new continuous dry blender device used in this project is shown in Figure 11a. As material enters the mixer, the powder passes the pathway of several impellers. Convection is the primary source of mixing, since a series of impellers exist around the axial length of the horizontal cylinder. The equipment has three innovative design features: (1) The system considered in this study has the flexibility of adjusting the angle at which the mixer processes (see Figure 11a). The combination of the blender angle and blade angle controls the degree of back-mixing to axial convection, and determines the number of theoretical equivalent stirred tanks along the axis of the mixer. (2) The powder is discharged through a screen (see Figure 11a). This feature ensures that agglomerates are destroyed in the mixer. By varying the mesh of this screen, different degrees of micro-homogeneity can be accomplished. (3) The powders leaving the mixer

Continuous Powder Blending

In the catalyst industry, quantification of mixing rates and mixing performance of powders is crucial for final product properties and performance, and online monitoring is always more attractive than offline approaches. In this project a new continuous mixer was designed and implemented to examine homogenization of materials used in catalyst formulations. Experiments were carried out (1) to identify formulations of variable cohesion, consisting of mixtures of silica, alumina, and zeolites with various particle sizes, selected in collaboration with our sponsors, (2) to develop analytical methods, based on NIR spectrometry, suited for fast and inexpensive analysis of blend samples, and adaptable for on-line monitoring, and (3) to examine the performance of blending operations as a function of essential process parameters (processing flow rate, blend composition, agitator speed). Results of this project include: (1) Quantification of mixing rates and mixing performance in new processing equipment relevant to sponsor members was determined as a function of blend cohesion, fill level, blender speed, and flow rate. (2) A family of standard formulations, suitable for examining the effects of blend cohesion in other systems and supplemented by validated

(a)

(b)

(c)

Figure 11: (a) GEA Buck Systems Continuous Dry Blender with Innovative features, (b) schematic of the continuous mixer with two vibratory feeders, (c) image illustrating powder hold-up inside the continuous mixer.

may also re-enter as a recycle stream. Standard catalytic materials of varying cohesion were tested in this work. The main observations are as follows: (1) Mixing performance is largely limited by the quality of the feeders. As shown in Figure 11(b), two Eriez vibratory feeders were used to feed the powder into the mixer. (2) Mixing quality improves with residence time. (3) The number and spacing of blades plays a critical role in minimizing stagnant regions in the blender. (4) Mixing is better at higher speeds. (5) Incline angle has an effect on performance, but results at this point do not display a simple trend. (6) Electrostatic agglomeration of cohesive powders on the surface of blades was observed. the cohesive powders then accumulated in stagnant regions, strongly affecting the concentration of the discharge stream (see Figure 11c).

ContactInformation: ProfessorBenjaminGlasser,DirectorCMSEC,bglasser@rutgers.edu Dr.ArthurChester,IndustrialLiaisonCMSEC,achester@rci.rutgers.edu DepartmentofChemicalandBiochemicalEngineering RutgersUniversity 98BrettRoad Piscataway,NJ088548058,USA Phone:(732)4454243;Fax:(732)4452581 Website:http://sol.rutgers.edu/centersTrainingCatalyst.html