`
`(Received 30 June
`
`accepted in revisedform 4 December
`
`1997) Abstract-Activated carbon honeycomb structures based on synthetic precursors are described. These strong, highly durable honeycombs are continuous interpenetrating structures of activated carbon and a ceramic with adjustable broad range of carbon percentage (5-95 wt%). Dynamic adsorption performance of a particular honeycomb structure containing 18 wt% carbon is described with respect to various adsorbates, i.e. butane, toluene, formaldehyde, isopropanol etc., and the importance of space velocity in determining the adsorption performance as well as the effect of structural parameters on the performance is shown. 0 1998 Elsevier Science Ltd. All rights reserved. Key Words-A. Activated carbon, A. carbon honeycombs, D. adsorption properties. 1. INTRODUCTION Activated carbon is a very important material indu- strially, with applications in a variety of areas [l] such as adsorbers in air and water pollution control, catalysts in the chemical and petrochemical indu- stries, electrodes in batteries and supercapacitors, and purifiers in the food and pharmaceutical indu- stries. Typically activated carbon is made from natu- rally occurring materials such as wood, coal and nutshell flour [2,3] etc. via high temperature, inert atmosphere processing followed by activation to create porosity in the nanometer size range. This porosity imparts special adsorption characteristics to carbon and makes it useful in the variety of applica- tions mentioned above. As a result of the natural raw materials used as well as the necessary processing steps associated with the manufacture, the material is obtained in a finely powdered form, and is granu- lated later to make it suitable for handling on a large scale [ 11. In spite of the widespread use of this type of product, there are some drawbacks associated with it. First, the variations in the natural material make it difficult to control the properties of the carbon from batch to batch. Secondly, the granular material has to be used in traditional packed beds in many of the pollution control applications. Although func- tional, such beds have inherent drawbacks such as high pressure drop associated with the flow through the packed media, particle entrainment, channeling etc. For applications such as battery electrodes fine powders may only be used with binders and metallic current collectors, which result in poor utilization of the carbon properties. To obtain carbon with con- trolled and reproducible properties, synthetic starting materials may be used. The synthetic materials typi- cally include polymeric resins. Several studies [4-61 have been carried out on synthetic raw material- based carbon. These carbons eliminate the first draw- back of the variability in raw material source and have the added advantage of control of the carbon properties through control of the synthetic material structure. In spite of these advantages the synthetic material-based carbons have not been widely avail- able commercially presumably due to high cost. Recently a new type of product, activated carbon fibers based on synthetic raw materials has been introduced. A number of studies of the properties of these fibers and the performance advantages the fibers demonstrate, have been published [7-91. The fibers, however, have not found widespread commer- cial use. One of the drawbacks of the fibers is that these small diameter (N 10 pm diameter) fibers have to be made into a structural shape to put into a device and this requires more difficult procedures such as weaving into shapes. These procedures are expensive. Availability of activated carbon in a mono- lithic form with controlled adsorption properties is thus desirable. The main objective of the various devices in indu- strial applications is to make the special nanoporosity in the carbon accessible to a flowing stream or more accurately the components in the stream, so that these components may be adsorbed and removed from the stream. The better the efficiency of the device in doing so, the better the carbon is utilized. It is thus important that the device has as high a geometric surface area per unit volume as possible. In the case of the packed beds, although the surface area per unit volume is high, in many cases all of it may not be accessible to the fluid stream because of the preferential flow patterns that may be established. Another problem with the packed beds is that the carbon pellets in the beds are sufficiently large in diameter to prevent full utilization of carbon because of the diffusional resistance associated with diffusion through the macropores before the components are adsorbed in the meso or the microporosity. 981
`
` MAHLE-1018
`U.S. Patent No. RE38,844
`
`1991;
`
`
`982 K. P. GADKAWI. A natural structural shape that may eliminate the problems associated with the traditional products mentioned above is the honeycomb shape. This shape is known to have a very high geometric surface area to volume ratio, a feature which has resulted in widespread use of honeycomb structures in applica- tions such as automotive catalyst carriers. The high surface area provides high contact efficiencies between the substrate and the flow stream. If acti- vated carbon is made available in the honeycomb form, its adsorption capacity may be utilized better and a more efficient device design may result. The pressure drop for a honeycomb structure based system will be significantly lower than a packed bed system, with the extent of decrease in pressure drop depending on the structural parameters of the honeycombs. Although attempts have been made to form acti- vated carbon honeycombs and some industrial pro- ducts such as ozone filters in laser printers have resulted, these honeycomb products are not widely used. The main reason for the commercial honey- combs not being successful is that these honeycombs have poor durability. The honeycombs are fabricated mainly by extruding finely powdered carbon into honeycomb shapes with a polymeric binder. The bonding between the carbon particle and the binder is typically poor which results in low strength, low durability honeycombs. In liquid streams the liquid preferentially adsorbs at the interface between the particle and the binder weakening the interface still further and severe durability problems result. These honeycombs therefore have been used only in less demanding applications such as for ozone removal in laser printers. A new class of activated carbon honeycombs have been developed, which eliminate the binders associ- ated with the honeycomb structures mentioned above. These highly adsorbent honeycombs have a continuous carbon structure. Since no binders are used, these inert high strength honeycombs are highly durable and provide an attractive alternative to the traditional packed bed system. In the following one of the concepts for fabricating these honeycombs is presented. The dynamic adsorption performance of any activated carbon system is the key to determining the usefulness of the system in any application. Although the major advantage of the honeycomb structure is the very open structure which results in a low pressure drop, this same open structure may result in poor adsorption performance. Preliminary data is thus presented on dynamic adsorption perfor- mance of one type of such a honeycomb structure with various adsorbates and the effect of some of the process and structural parameters on this perfor- mance is discussed. 2. CONCEPT AND EXPERIMENTAL DETAILS The honeycombs discussed in this report were fabricated as composite structures of ceramics and carbon. In principle the process is as follows. Highly porous ceramic honeycombs are first fabricated. These honeycombs are fabricated from various ceramic compositions starting with clays. Cordierite- based compositions, which are used for fabricating low thermal expansion honeycombs for automotive catalyst support were used for this work. The honey- combs are commercially available from Corning Inc., Corning, NY. Various clays are mixed together with polymeric binders and extruded through steel dies in honeycomb shapes. The honeycombs are then fired to high temperatures ( - 1500°C) to burn out binders and to react and sinter clays to form cordierite honeycombs. These honeycombs may be fabricated with a wall thickness of anywhere from 0.1 mm and higher and with cell densities as high as 95 cells per square cm. The cells may be rectangular, triangular, hexagonal or other shapes. Another important factor that can be adjusted is the wall porosity. By utilizing appropriate compositions, the wall porosity may be adjusted from 10% to 70%. Most of the work described in this report was done with cordierite honeycombs with 62 cells per square cm, a wall thickness of 0.19 and 0.29 mm and wall porosity of - 50%. The mean pore size for these honeycombs is ten microns. Figure 1 shows the pore size distribution of the honeycombs measured by the mercury poro- simetry technique. The honeycomb is impregnated with high carbon yield polymeric resin of low viscosity. The resin is allowed to soak into the ceramic honeycomb struc- ture. The excess resin is then drained and the resin- coated honeycomb is subjected to a drying and curing cycle to crosslink the resin. The resin forms an interpenetrating network with the ceramic. The cured honeycomb is then subjected to carbonization and activation in an inert atmosphere to form the compos- ite carbon-ceramic honeycomb. The resin chosen for this work was a phenolic resole from Occidental Chemical Co., Niagara Falls, NY, because of two important characteristics. This resole has a viscosity of 100 cP. This low viscosity allows the impregnation and draining step to be carried out with ease. In addition the inexpensive phenolic resins have a very high carbon yield (- 50% of the cured weight), thus lllll.,l , I* ,,... I , IL, ,,,, , 111**1* 100 10 1 0.1 0.01 Diameter (micrometers) Fig. I. Pore size distribution of ceramic honeycomb.
`
`
`
`reducing the cost of the carbon produced. The aque- ous resole contains about 65% solids. The honey- combs are simply dipped in the resin, allowed to soak for a few minutes and then drained of the excess resin by blowing air through the cells, before being subjected to a drying and curing process. The coated honeycombs are air dried and then dried at 95°C and cured at 150°C to crosslink the resin. Most of the cured resin remains in the cell wall porosity with a thin layer of resin on the surface. The coated honey- comb is then subjected to a carbonization and activa- tion process. The carbonization is carried out at 900°C in nitrogen. The activation is carried out at the same temperature in carbon dioxide to obtain burnoff in 25-30% range. On carbonization a ceram- ic-carbon composite structure is formed. This struc- ture is monolithic with carbon forming a continuous structure inseparable from the ceramic backbone. As the SEM micrographs show, it is difficult to see the boundary between the carbon and the ceramic. The SEM’s of the ceramic honeycomb and the composite honeycomb are shown in Fig. 2. Carbon honeycomb structures for adsorption applications 983 porosity in the honeycomb wall is not strongly affected. Figure 2(b) shows a high magnification view of the fracture surface of the honeycombs. It is clearly seen that the carbon has formed on the ceramic and also occupied the small pores in the wall. The activa- tion step generates the desired porosity in the carbon for adsorption without affecting the strength of the structure significantly. The carbon-ceramic compos- ite structure has 30-40% higher strength than the precursor ceramic honeycomb. Table 1 shows the comparison of the strength of the honeycombs. The dynamic adsorption properties of the honey- combs mentioned above were measured with an apparatus, the schematic of which is shown in Fig. 3. As shown in the figure various gases such as nitrogen, butane and toluene stored in pressurized tanks are metered through flowmeters in appropriate propor- tions and mixed before passing through the sample Figure 2(a) shows the polished cross section of the ceramic as well as the composite carbon honeycomb. It is clear from the micrograph that the wall thickness of the honeycomb does not change significantly as a result of the carbon impregnation and the large Table 1. Axial crush strength of the honeycombs (2.54 mm min-’ speed) Ceramic honeycomb 6.3 MPa (k 10%) (0.19 mm wall) Composite carbon honeycomb 8.6 MPa (k6%) (0.19 mm wall) Composite carbon honeycomb 13.8 MPa (k8%) (0.29 mm wall) (4 (4 (‘4 Fig. 2. SEM of ceramic and composite carbon honeycomb.
`
`
`
`984 K. P. GAI)KhRI~I. Schematic Of Dynamic Adsorption Apparatus Detectors Flow Meters Adsorption Bed ) Bypass Valve t tt Control Valves Gas Tanks Fig. 3. Schematic of dynamic adsorption apparatus. chamber. A small fraction of the mixed gases is diverted to the hydrocarbon detector. The calibrated detectors measure the hydrocarbon levels in the gas streams. All the equipment is computer controlled and the data is automatically stored and plotted to generate the breakthrough curves. The adsorption performance of the samples is measured as a function of time. Adsorption isotherms of the carbon are measured on an Omnisorp 100 from Coulter Inc. using standard procedures. Although methods have been developed to fabricate anywhere between 5 and 100 wt% carbon honeycombs, all the data given in this report was generated on first generation honey- combs with about 18 wt% carbon. 3. RESULTS AND DISCUSSION The adsorption performance of the honeycombs depends on several material and process parameters. The material parameters include the carbon adsorp- tion porosity on the honeycomb, the carbon percen- tage as well as the wall thickness of the honeycomb. The process parameters include the flow rate and concentration of the adsorbate, the adsorption poten- tial which is dependant on the carbon structure as well as the adsorbate properties such as molecular weight. Although these carbon honeycombs have been made with a range of carbon structures which affect the adsorption performance, the carbon struc- ture used for this work is described by a type I isotherm shown in Fig. 4. The carbon in this case is thus essentially microporous although some adsorp- tion at higher relative pressure is evident. The nitro- P g 200- $ P E” 100 - 2 3 I I I I I I I I 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.6 C P/PO 9 Fig. 4. Nitrogen adsorption isotherm of carbon. gen adsorption isotherm was obtained at liquid nitrogen temperature. A TEM micrograph of the carbon structure is shown in Fig. 5. As the figure shows the carbon has a very regular structure with platelet spacing of 0.7-0.8 nm. This carbon structure is unusual and changes very significantly as a function of material and process parameters. Detailed studies of these structures will be published separately. The adsorption performance of a device is mea- sured in terms of the efficiency of removal of a certain constituent from a flowing stream. The performance is generally given in terms of a breakthrough curve, which shows the ratio of effluent concentration to influent concentration as a function of time [IO]. The breakthrough capacity, i.e. amount adsorbed until the effluent to influent concentration reaches 0.95, depends on parameters such as flow rate, concen-
`
`
`
`100
`
`Well
`
`80- Space velocity 748/mln
`
`Carbon honeycomb structures for adsorption applications Fig. 5. TEM of carbon nanostructure. tration of adsorbate and the amount of carbon in the honeycomb. In the literature, the breakthrough curve is given as a function of surface velocity calculated from volumetric flow rate divided by the empty cross-sectional area of the bed. Although honeycomb structures may have advan- tages mentioned above such as low pressure drop, an important factor may substantially affect the performance negatively. Honeycombs have straight flow paths with very small cell dimensions. Under standard conditions encountered in practice, the flow through the honeycomb cell is laminar. Laminar flow means that there is very little mixing of the fluid stream during the flow through the honeycomb. As a result the efficiency of contact between the carbon surface and the contaminant to be removed may be very low and the adsorption efficiency may be nega- tively affected, especially at the high flow rate low concentration conditions encountered in practice. Experiments were thus done to evaluate adsorption efficiencies for butane and toluene, an aliphatic and an aromatic hydrocarbon, at 80 ppm concentration. These two hydrocarbons at the specified concen- tration are used as model compounds for adsorption behavior in certain industrial applications. Formaldehyde and acetaldehyde adsorption perfor- mance was evaluated at 30 ppm concentration. These two components are present in diesel exhaust. Finally, isopropanol adsorption performance was evaluated at 80 and 300 ppm concentration. Isopropanol is a typical solvent in solvent recovery applications. The data obtained with all these compounds at the speci- fied concentrations and flow rates is thus expected to give an idea of the performance of these new struc- tures under the dynamic conditions encountered in practice. Figure 6 shows the dynamic adsorption perfor- mance of a 400 cell honeycomb with 0.19 mm wall thickness for butane adsorption at 80 ppm inlet con- centration. The honeycomb was 2.54 cm in diameter and 3.8 cm long. The adsorption performance was evaluated at a flow rate of 15 000 cm3 min-’ of the nitrogen-containing butane at 80 ppm. As the figure
`thidtnerr O.lBmm Surface velocity 29.6mImin I 0 20 40 60 80 Time (min) Fig. 6. Breakthrough curve for butane, thin-walled honey- comb (0.19 mm wall). shows, the effluent concentration is 1.6 ppm or 2% of the influent at one minute. There is thus an immediate breakthrough. As the adsorption sites are filled up, the effluent concentration begins to increase and eventually reaches 95% of the influent level, i.e. 76 ppm at 75 minutes. Integrating the area above the curve gives the total adsorption capacity of the honeycomb. In this case the capacity is 127.9 mg. Since there is immediate breakthrough, it is clear that the length of the honeycomb is not sufficient to develop the mass transfer zone fully. The surface velocity in this case is 29.61 m min-‘. The honey- comb was then removed from the apparatus and regenerated in a 150°C oven. The experiment was then repeated with the same honeycomb at the same concentration of butane (80 ppm) but at a surface velocity of 46 m mini. The data showed that there was immediate breakthrough and the initial adsorp- tion efficiency was 95% (4 ppm breakthrough) in this case. Increasing the surface velocity further to 100 m min-’ results in a further decline in initial efficiency and the initial effluent concentration is 6 ppm. Figure 7 shows a comparison of these three curves. The nature of the curves is similar to the curves obtained from packed beds and so there does not seem to be a heavy penalty associated with using an open honeycomb structure with straight flow Time (min) Fig. 7. Effect of surface velocity on breakthrough curve for butane.
`
`
`
`986
`
`0
`
`K. P. GADKARISS channels and low pressure drop in terms of adsorp- tion behavior. All the breakthrough curves are labeled in terms of space velocity. The space velocity is defined as the flow rate of the fluid stream divided by the volume of the bed. This space velocity may be visualized as the number of times the entire fluid amount in the bed is exchanged per unit time and may also be called the turnover frequency. The concept of space velocity is routinely used in catalysis literature. In adsorption literature, however, this concept is not commonly used. Data is normally presented in terms of surface velocity. As shown later in this paper. space velocity is a more fundamental parameter and is very useful when comparing data or scaling up the system even for adsorption systems. The space veloc- ity is defined here based on the empty volume of the bed. In that sense it is not the true turnover frequency, because the bed has the honeycomb in it. Only the void volume should be taken into account if the true turnover frequency or true space velocity is to be found. As the honeycomb structural parameters such as cell density or wall thickness or wall porosity change, this true space velocity will change. It is shown later that in spite of this limitation the empty bed space velocity is still a very effective indicator of the performance of the bed as long as similar systems are compared, in addition to being a much simpler parameter to calculate than the true space velocity. It is desirable to obtain high adsorption efficiencies from small bed volumes. A parameter that can be changed to improve adsorption efficiency for the honeycombs is the wall thickness. Increased wall thickness at the same porosity will result in increased amount of carbon per unit volume of the honeycomb, improving its adsorption capacity. Thicker walls reduce the open area for flow through the honeycomb thus increasing the true surface and space velocity. but also increase the contact efficiency between the contaminant to be removed and the carbon. A bal- ance between these two factors will determine whether the adsorption efficiency increases or decreases under given conditions. If the adsorbate diffuses at a high enough rate to the interior of the wall and empty adsorption sites are available for adsorption continuously, then a thicker-walled hon- eycomb should give better adsorption efficiencies and capacities. To test this hypotheses honeycombs with the same composition and cell density but with 0.29 mm wall thickness were fabricated, compared to 0.19 mm wall honeycombs used for earlier experi- ments. The thicker-walled honeycomb was treated in the same manner to form a carbon composite honeycomb as described previously and its adsorption performance tested with 80 ppm butane under identical conditions, i.e. 80 ppm butane, 15 000 cm3 min-’ flow rate over a 2.54 cm diameter, 3.8 cm long honeycomb. Figure 8 shows the data. This data may be compared with the data in Fig. 6. As the figures show the adsorption performance of 100 _ Butane 80 ppm - Space velocity 748/min
`20 40 60 60 1 Time (min) IO Fig. 8. Butane adsorption on thick-walled honeycomb (0.29 mm wall). the thicker-walled honeycomb (0.29 mm) is clearly superior to the thin-walled honeycomb (0.19 mm). The initial adsorption efficiency is 100% with no butane breakthrough for about 15 minutes. The adsorption capacity, i.e. the butane adsorbed up to 95% breakthrough is 185 mg compared to 127 mg for the thin-walled honeycomb. The thicker-walled sample picked up more resin and hence had more carbon. The amount of carbon on the thick- and thin-walled samples was 2.21 and 1.45 g, respectively. This 52% increase in carbon results in a 46% increase in adsorption capacity. In spite of the increased true space velocity (the empty bed space velocity remains the same) the thicker-walled honeycombs give sig- nificantly enhanced adsorption performance, demon- strating that the honeycomb structures may be designed to obtain desired adsorption performance. Toluene adsorption experiments were carried out on the honeycomb samples under identical conditions to those of butane experiments. Figure 9 shows the data for the thin-walled sample. Influent concen- trations were 80 ppm of toluene. The toluene adsorp- tion capacity was 427 mg, substantially higher than the butane capacity of 127 mg for the sample. Increasing the wall thickness (form 0.19 mm to 0.29 mm as described previously) increased the tolu- ene capacity to 579 mg. To evaluate the relative effect of surface velocity and space velocity on adsorption 100 - Toluene 80 ppm 1 Space velocity 748/min 60 _ Wall thickness 0.19mm Jz _ z - g 40- = - 2 - l,,,I~~~IS~SI~*# 0 20 40 60 60 100 120
`
`Time (min) Fig. 9. Toluene adsorption--breakthrough curve. 10
`
`1
`
`
`
`Carbon honeycomb structures for adsorption applications 987 performance, experiments were done with the 0.19 mm wall thickness samples as follows. The first set of experiments were done with samples 2.54 cm in diameter and 3.81 cm long at 15 000 cm3 min-’ flow rate and at 30 000 cm3 min-’ flow rate. The corresponding space velocities are 748 and 1496 min-‘. The toluene concentration was 80 ppm in both cases. Another experiment was done with a sample 2.54 cm in diameter but 7.62 cm in length with a flow rate of 30 000 cm3 mini. Since the cross- sectional area of the samples for both 30 000 cm3 min-’ flow rate experiments is the same, the surface velocity remains the same, however, the space velocity changes because the volume is different. The surface velocities are 29.56 and 59.13 m min-‘, respectively, for the 15 000 and 30 000 cm3 min- ’ flow rates. The space velocities for the 3.81 and 7.62 cm long samples at 30 000 cm3 min-’ flow rate are 1496 and 748 min-‘, respectively. Figure 10 shows the data obtained. As the figure demonstrates, in spite of doubling the surface velocity from 29.56 m min-’ to 59.13 m min- 1 the adsorption performance remains identical if the space velocity is constant. As the space velocity changes from 748 min- ’ to 1496 min-‘, however, there is significant drop in adsorption performance. It is thus clear that space velocity is a more meaningful parameter as far as adsorption performance is concerned. Surface veloc- ity on the other hand is not an independent variable affecting the adsorption performance. Formaldehyde adsorption experiments were con- ducted on the honeycomb samples to evaluate perfor- mance for the case of small molecules at very low concentration. Figure 11 shows the adsorption data at 748 min-’ space velocity for thin-walled samples. It is seen that formaldehyde is adsorbed at a high efficiency initially (no breakthrough), but break- through occurs soon after. The breakthrough is delayed with the thick-walled samples compared to the thin-walled sample as expected. The nature of the formaldehyde breakthrough curves is different from the butane and toluene curves shown earlier. 100
`0 20 40 60 60 100 120 140 Time (min) Fig. 10. Effect of the surface velocity and space velocity on toluene adsorption. 60 1 ‘1: 10 20 30 40 50 0 Time (min) Fig. 11. Formaldehyde adsorption on thin-walled (0.19 mm) honeycomb. After the initial breakthrough the curves rise rapidly, however, in both the cases shown these curves do not rise to complete breakthrough, i.e. equal influent and effluent concentrations. The curves rise to a certain value and then flatten out at a constant effluent to influent ratio or adsorption efficiency. For the thin-walled sample this effluent to influent concen- tration ratio is 68% and for the thick-walled sample the ratio is 56%. This behavior is unexpected. As in the case for butane and toluene, the curves should attain 100% breakthrough. Initially all the adsorption sites are vacant and hence there is no breakthrough. All the adsorbate is easily adsorbed. As the vacant sites fill up the breakthrough occurs and when the adsorption sites are all filled up there is 100% break- through. For the case of formaldehyde, however, this does not appear to be the case. After attaining a certain breakthrough value the curve flattens out and does not rise for a considerable period of time. A possible explanation for this behavior is as follows. For very small highly volatile molecules, adsorption potential is low. Particularly under the high flow rate low concentration conditions, it is very difficult to adsorb these compounds from a flowing stream. Adsorption efficiency will depend on how efficiently the molecules come in contact with the carbon, which is determined by wall thickness, flow rate and concen- tration as well as by the size of the pores relative to the molecules. Only the smallest size pores with pore widths close to that of the adsorbate size will be effective. Figure 12 shows the pore size distribution obtained on these samples utilizing the Micromeritics ASAP2000 equipment and DFT software. As the figure shows there is a significant distribution of pore sizes in the micropore range. The pore sizes fraction that will be effective for formaldehyde adsorption under the given conditions is difficult to determine based on the breakthrough curve data since many factors contribute to the adsorption performance. A qualitative explanation may be as follows. For form- aldehyde adsorption only a small fraction of pores are effective because of its low adsorption potential. Initially all the ejixtive adsorption sites are vacant
`
`spalx wloclty
`1498/mln
`!swfaa v*ocity
`59.lm/min
`
`.
`.
`_
`_
`
`rekcity
`surface
`28.5m/mln
`
`
`
`988 K. P.
`
`GAIJKAKW
`
`0.12 ( I Gi 2 0.10 E s 0.10 E 3 0.06 i Pore Width (nm) Fig. 12. Micropore size distribution of carbon. and so all the molecules are adsorbed. As the mole- cules are adsorbed, these molecules begin to diffuse inside the wall. The surface sites in contact with the flowing stream are filled up quicker than the mole- cules could diffuse inside the wall, and so break- through occurs and adsorption efficiency begins to drop. However, as the surface sites become available due to diffusion of the adsorbed molecules to the interior, more adsorption can take place. When a dynamic equilibrium is reached between the diffusion rate and the adsorption rate, the curve flattens out. Eventually, when all the sites are filled up the effluent concentration should rise and attain the influent concentration value. For larger molecules with higher adsorption potential, a much larger fraction of pores are effective adsorbers. There is thus sufficient time available for the molecules to diffuse into the interior. while other unfilled pores are being filled. As a result breakthrough cannot occur until a much larger pro- portion of pores is filled up and continues to increase in magnitude until 100% breakthrough is observed. The experiment with formaldehyde cannot be contin- ued very long to verify whether the curve rises to 100% breakthrough because of practical considera- tions, i.e. the limited amount of formaldehyde stored in the tanks, and the number of tanks needed etc. Once adsorbed, the molecules could diffuse into larger pores in the sample, since high flowrate stream is not a factor any more. In the thick-walled samples the curve flattens out at a lower breakthrough level because of the higher efficiency of contact discussed earlier. Experiments were also done with acetalde- hyde at 30 ppm concentration and 748 min- ’ space velocity. The breakthrough curve flattens out in this case at 88%, a higher level than 68% obtained with formaldehyde as expected. Isopropanol experiments were done at 80 and 300 ppm levels to check the suitability of the honeycomb structures in solvent recovery type of applications. The isopropanol adsorption curve for both concentrations of 80 and 300 ppm at 748 min-’ space velocity is similar to the standard breakthrough curves obtained with butane and shown in Figs. 13 and 14. At 300 ppm concen- tration of isopropanol, an experiment was carried Isopr~anol 80 ppm Space velocity 74S/min Wall thickness O.lQmm Time (min) Fig. 13. lsopropanol adsorption performance at 80 ppm. 100 - Space velocity 748/min 0 10 20 30 40 50 60 ’ Time (min) Fig. 14. Ekt of space velocity on isopropanol adsorption performance. out to evaluate the effect of variation in surface velocity at constant space velocity as described earlier for toluene. Figure 14 shows again that the perfor- mance obtained is equivalent as long as the space velocity is kept constant, even though the surface velocity doubles, again indicating that space velocity is a more fundamental parameter in dynamic adsorp- tion measurement. 4.
`
`New activated carbon honeycomb structures based on synthetic raw materials have been demonstrated. These honeycombs with continuous, uninterrupted carbon structure are strong and highly durable. A large variety of such honeycombs with controlled cell density, wall thickness and cell geometries can be fabricated. In spite of the very open structure of these honeycombs, high adsorption efficiencies may be obtained even with high flow rate, low