COMMUNICATIONS TO THE EDITOR Adsorption of Ethanol- Water Mirtures by Biomass Materialr INTRODUCTION The commercial production of biomass-derived ethanol is dependent on the energy balance which is defined as the ratio of the combustible energy obtained from the product to the energy necessary to its production. The recovery of ethanol in an anhydrous form from an 8-12% solution (wt. 70) can be energy intensive. Reports in the past have indicated that the distillation of ethanol could consume up to 50% of the overall energy used in a typical grain ethanol plant.'.2 More recent reports have indicated improved energy efficiency by using alternative pro~esses.~.~ One possible route is distillation of fermentation strength alcohol to a 92% product followed by adsorption to remove the remaining water. A variety of materials could be used as adsorbents. These include calcium chloride,s barium oxide,' metallic sodium,' silica gel.' and biomass material^.^.'^ Biomass materials such as cornmeal and cellulose are particularly interesting since batch adsorption studies have shown these to be capable of yielding anhydrous alcohol starting from an 88-9570 ethanol."." In this context. screening studies on the separation of an ethanol and water-vapor mixture by starch (corn and potato), xylan, pure cellulose. corn residue, cornmeal. wheat straw. and bagasse were carried out. A gas chromatographic elution method developed during the course of this work which expedites evaluation of potential adsorbents is also described. EXPERIMENTAL PROCEDURES A model 311 Carle (Anaheim, CA) gas chromatograph (GC) was modified to provide capability for elution studies. A simplified schematic diagram of the modified gas chromatography is shown in Figure 1. The apparatus consists of two saturators: S (stainless steel. 2.54 cm 0.d. X I ' TEMPERATURE CONTROL UNIT I L -- 1 Fig. 1. Schematic diagram of modified GC apparatus. The operation is described in the Experimental Procedures section. Biotechnology and Bioengineering, Vol. XXIV. Pp. 725-730 (1982) 01982 John Wiley & Sons, Inc. CCC 0006-3592/82/030725-06$01.OO
726 BIOTECHNOLOGY AND BIOENGINEERING VOL. XXIV (1982) 9 cm long) connected to a 0.635 cm 0.d. (0.462 cm i.d.) X 10 cm stainless-steel column C. (inside cross-sectional area of 0.168 cm2). Effluent from the column is analyzed by a thermal conductivity detector (TCD) which detects both water and ethanol. The flame ionization detector (FID) was used to confirm the identity of the ethanol peak. The output from the detector was monitored by a Hewlett-Packard recorder (model 7100B, San Diego, CA). Particle size and weights of material packed in these columns are given in Table I. The columns were packed using the standard tapping procedure.12 During operation, the column and saturators were kept at 80 C for both adsorption and desorption runs. The adsorbents were dried initially with helium until no water was detected in the effluent from the column. The saturators were then charged with 104-proof ethanol (44% by weight) and kept at a total pressure of 55.7 psi (3.79 atm) with helium to give a vapor phase having a calculated equilibrium composition of 48% (by weight) ethanol, 32% water, and 20% helium. This corresponds to 8.2% ethanol, 14% water, and 77.8% helium on a mole basis. An adsorption run was started by opening valve V1 to allow the feed stream saturated with water and ethanol to pass through the column. At the end of the adsorption cycle, the desorption phase was started by opening valve V2 and closing V1 and V3 simultaneously to allow pure helium to flow through the column. The gas flow rate which corresponded to a superficial velocity of 94 cm/min (calculated based on the empty column cross-sectional area) was the same for all runs. RESULTS AND DISCUSSION The elution profiles of water and ethanol for the adsorbents listed in Table I are shown in Figures 2-4. These profiles give a comparison of the adsorbents for a constant column length. Thus, comparison of the separation capabilities of these packing materials reflect a volume basis rather than a weight basis. In typical elution profiles (see Figs. 2-4), the TCD response to the ethanol effluent from the column is a sharp rise followed by either a plateau [see Figs. 2(a) and 2(b)] or a diffusive front and then a short plateau [see Figs. 2(c), 2(d), 3(a)-3(d), and 4(b)]. In the case where the plateau immediately follows the sharp rise, the adsorption of ethanol is indicated to be almost negligible. Those elution patterns showing a diffusive front may indicate a small amount of adsorption of ethanol may be occurring and/or may reflect mass-transfer and dispersion effects. TABLE I Packing Materials Used in Screening Studies Average Packing Column Weight particle size density No. Packing material (s) (pm) (g/cm3) 1 2 10 starch (corn) starch (potato) (Sigma Chemical St. Louis, MO) Avicel (FMC Corp., Philadelphia, PA) xylan from larch wood (Sigma Chemical) cornmeal corn residue wheat straw bagasse silica gel BioSil A (BioRad, Richmond, CA) protein (Bovine serum albumin) (Sigma Chemical) 1.41 1.42 0.66 0.97 1.09 0.35 0.37 0.35 0.72 0.46 < 27 < 27 50 105 500 125-149 125-149 104-149 37-74 NA 0.84 0.85 0.39 0.58 0.65 0.21 0.22 0.21 0.43 0.27
COMMUNICATIONS TO THE EDITOR 727 Fig. 2. Elution profiles for (A) starch (potato), (B) starch (corn), (C) Avicel, (D) xylan. Curves generated from thermal conductivity detector. Conditions are 80 C for adsorption and desorption; He as carrier gas in adsorption and eluting gas in desorption; gas velocity is 94 cm/min. Time t = 0 corresponds to elution time corrected for column void volume. An example where significant adsorption occurs is given by silica gel [Fig. qb)]. In this case ethanol requires 24 min to break through compared with air (not shown) which requires 53 s. A noticeable diffusive front is also exhibited. Significant adsorption of alcohol is indicated further by silica gel since a decrease in the ethanol profile occurs 30 min after desorption with helium (indicated by arrow) is initiated [Fig. qb)]. In comparison, the ethanol profile drops almost immediately for starch and cellulose upon the onset of desorption [arrows in Figs. 2(a) and 2(b)]. The capacity of these adsorbents for water on a volume basis is indicated by the distance between the initiation of desorption (indicated by the arrow) and the emergence of the water peak.
6o 728 BIOTECHNOLOGY AND BIOENGINEEIUNG VOL. XXlV (1982) 50 2 3 4 I 2 3 4 Fig. 3. Elution profiles for (A) cornmeal, (B) corn residue, (C) wheat straw, (D) bagasse. Conditions are the same as Fig. 2. The further the distance, the better the volumetric capacity. For the adsorbents in Figures 2 and 3, a qualitative comparison on a weight basis can be made by also refemng to Table I. Thus while the capacity of the adsorbents based on column volume is potato starch and corn starch > xylan > cornmeal > Avid > bagasse, corn residue, and wheat straw, on a weight basis the capacities appear to be the same magnitude of order. The columns packed with starch (potato and corn) [Figs. 2(a) and 2(b)] and cornmeal [Fig. 3(a)] yielded a baseline separation between the ethanol and water vapors. This confirms a prior observation that starch and cornmeal are good adsorbents which can yield anhydrous ethanol.'o.ll Xylan, which was chosen as being a representative component of corn fiber, also gave a baseline separation.
COMMUNICATIONS TO THE EDITOR 729 (8) TIME (HaR) Fig. 4. Elution profiles for (A) protein, (B) silica gel. Conditions are the same as Fig. 2. The three major components of cornmeal (starch, xylan, and protein) were examined individually so that their relative contributions to the adsorption properties of cornmeal might be estimated. The starch, protein, and xylan show little or no capacity to adsorb ethanol. However, while starch [Fig. 2(a)] and xylan [Fig. 2(d)] strongly adsorbed water, protein [Fig. 4(a)] only weakly adsorbed water (indicated by overlapping regions). Hence, starch, which is 65-75% of the weight of cornmeal, and fiber, which is 5-1070 of the weight of cornmeal, appear to be primarily responsible for the water adsorption properties of cornmeal in the presence of ethanol. Cellulose. another major component of biomass (30-40%) also has water adsorption capacity as has been reported in previous ~ ork.'~.'~ An indication of this is shown with Avicel, a microcrystalline cellulose in Figure 2(c). Although its volumetric separation capability is lower than cornmeal or xylan. Avicel again does not adsorb significant ethanol. Better separation of ethanol and water with Avicel could be achieved by using a longer column. These characteristics also make Avicel and other cellulose adsorbents suitable for analytical gas chromatography. I3.l4 The potential of agricultural residues as adsorbents for dehydrating ethanol is indicated in Figures 3(b)-3(d). Corn residue [Fig. 3(b)], wheat straw [Fig. 3(c)], and bagasse [Fig. 3(d)] show a separation of water from ethanol which is better than silica gel. The difference is even more pronounced on a weight basis when the lower packing densities of the biomass materials are considered. While these materials have a lower separation capability than starch. xylan, or cornmeal, they may have capacity to be of interest in view of current developments on the conversion of biomass to ethanol. This data also indicates that other cellulosic materials such as. for example, corncobs. wood chips, and sugarcane bagasse, might be suitable for dehydrating ethanol. CONCLUSIONS The data presented show that a variety of polysaccharide-type materials are capable of separating water from ethanol and hence are potential adsorbents for dehydrating ethanol. It appears that the capacity for cornmeal to remove water from ethanol arises from the starch and fiber components of grain while the protein component is less important. Given the adsorption properties of both starchy and cellulosic materials, grains and biomass other than those presented here should have desirable properties for separating water from alcohol. For example, grains
730 BIOTECHNOLOGY AND BIOENGINEERING VOL. XXIV (1982) such as barley, rice, and sorghum and biomass materials such as wood chips, grasses, and corncobs are likely candidates. References 1. T. K. Ghose and R. D. Tyagi, Biotechnol. Bioeng.. 21, 1387 (1979). 2. W. A. Scheller, Energy requirements for grain alcohol production, presented at 176th National ACS Meeting, Miami Beach, FL,, Sept. 10-15. 1974. 3. R. Katzen, W. R. Ackley, G. D. Moon, Jr., J. R. Messick, B. F. Brush, and K. F. Kaupisch, Low energy distillation system, presented at 180th National ACS Meeting, Las Vegas, NE, August, 1980. 4. News Features, Chem. Eng.. 88(11), 29 (1981). 5. W. A. Noyes, J. Am. Chem. Soc., 215, 857 (1923). 6. P. Pusl, Znt. Sugar J.. 266 (1933). 7. F. G. Smith, Ind. Eng. Chem. (Anal. Ed.), 7, 72 (1979). 8. H. M. Davis and C. E. Swearingen, J. Phys. Chem.. 35, 1308 (1931). 9. M. R. Ladisch and K. Dyck, Science, 205, 898 (1979). 10. M. R. Ladisch, M. Voloch, K. Dyck, and J. Allen, Energy efficient dehydration of ethanol, presented at 178th National ACS Meeting, Washington, DC, Sept. 10-15, 1979. 11. M. Voloch, J. Hong, M. R. Ladisch, and G. T. Tsao, unpublished. 12. F. W. Rowland, The Practice of Gas Chromatography. 2nd ed. (Hewlett-Packard, Avondale, PA, 1974). 13. J. Hong, M. R. Ladisch, and G. T. Tsao, J. Chem. Eng. Data. 26, 305 (1981). 14. B. Miller, H. L. Friedman, and C. H. Meiser, Text. Res. J.. 50(1), 10 (1980). Laboratory of Renewable Resources Engineering Purdue University West Lafayette, Indiana 47906 Accepted for Publication September 1, 1981 *Also with the Department of Agricultural Engineering. talso with the School of Chemical Engineering. J. HONC M. VOLOCH M. R. LADISCH*t G. T. TsAot