Two-Stage, Dilute Sulfuric Acid Hydrolysis of Hardwood for Ethanol Production
Total Page:16
File Type:pdf, Size:1020Kb
TWO-STAGE, DILUTE SULFURIC ACID HYDROLYSIS OF HARDWOOD FOR ETHANOL PRODUCTION John F. Harris, B.S. Chemical Engineer Andrew J. Baker, B.S. Chemical Engineer John I. Zerbe, Ph.D. Program Manager Energy Research, Development, and Application Forest Products Laboratory, Forest Service, USDA Madison, Wisconsin53705 ABSTRACT The Forest Products Laboratory has developed fundamental kinetic relationships for the hydrolysis of lignocellulose dating back to World War II. Recent work at the Laboratory has provided additional informa tion on hydrolysis of red oak by a two-stage process at higher tempera tures with lower liquid-to-solid ratios. The complications of working with mixtures of sugars and acetic acid (6 percent of oak) are such that effective fractionation is necessary. This is best accomplished by prehydrolysis using dilute sulfuric acid. Dilute acid hydrolysis is also currently the best method for conversion of the residual prehydro lyzed lignocellulose to glucose. The course and yield of reactions are well predicted by kinetics data, but more attention needs to be given to the effects of ash constituents on the catalyst acid and the rever sion reaction, and in reducing the water-to-wood ratio in hydrolysis. The purity of the solutions generated by the two-stage process is considerably better than that obtained by the previous percolation process, but this advantage is offset somewhat by the higher yields of the percolation process. Projections on the economics of the two-stage process, although better than for processes based on prior technology, are still considered high risk. 1151 TWO-STAGE, DILUTE SULFURIC ACID HYDROLYSIS OF HARDWOOD FOR ETHANOL PRODUCTION INTRODUCTION Interest in wood hydrolysis dates to 1819 when Braconnot dis covered that cellulose could be dissolved in concentrated acid solutions and converted to sugar. Glucose, the principal sugar pro duced, could then be quite readily fermented to ethanol. Other carbo hydrates in wood can also be hydrolyzed to sugars. Next to glucose, xylose is the most important sugar from hardwoods. Mannose is obtained from softwoods in significant amounts, and galactose and arabinose may also be produced. This is a synopsis of a large report (6) soon to be released by the Forest Products Laboratory (FPL). The original report was devel oped in cooperation with the Tennessee Valley Authority (TVA). The report presents information on a two-stage, dilute acid hydrolysis process and its application to the production of ethanol from hard woods. The process, in its simplest outline, is shown in Figure 1. Wood chips, impregnated with a dilute sulfuric acid solution and drained of all interstitial liquid, are charged to the first stage. Here they are heated with direct steam, resulting in the hydrolysis of most of the hemicelluloses, and then discharged to washers. After being washed free of the material solubilized in the first stage, the lignocellulose is reimpregnated with acid and charged to the second stage. As with the first stage, the liquid content of the second-stage charge is kept to a minimum. Conditions here are sufficient to hydro lyze the resistant cellulose. The resulting mixture is discharged to washers where the glucose solution is separated from the lignin residue. This process was selected for investigation because it was thought that: 1) The separate stages for hydrolyzing the hemicelluloses and cellulose would result in high yields and high-purity products. 2) The energy consumption would be minimized since much of the liquid is removed before each of the hydrolysis steps. 3) The resulting sugar solutions would be more concentrated. 4) The information available on the use of dilute sulfuric acid as a hydrolysis medium would be a great advantage if the process were to be carried rapidly to commercial operation, which was the underlying motivation for the work. Some development work on a similar process was carried out in Sweden during World War II and a short description of the results presented at a United Nations conference in 1952 (2). Although the results were encouraging, the process was clearly considered to be a wartime expedient, and work was discontinued when the war ended. No attempt had been made to deal with effluents, to complete the design, or to optimize the process. The wood used was spruce (Piceaexcelsa ), whereas the present study deals with hardwoods--southern red oak (Quercus falcata Michx.) in particular. The large amount of fundamental data available from previous re search at FPL (Madison) shaped the approach taken in this research. Existing information was believed sufficient to permit a fairly accurate design of the process; that is, the more important components 1152 Figure 1.--Two-Stage hydrolysis process could be modeled and brought into a harmonious whole. Our purpose was to recommend processing conditions pertaining to each unit, and to estimate yields, energy requirements, and other pertinent process information. Additional experimental work was planned to validate or modify the assumptions and data used in modeling. The project spawned several related fundamental studies not essential to the current process design. They include investigations of xylose metabolism, cellulose hydrolysis, sugar degradation, and deacetylation kinetics. We plan to continue these studies and release the results through technical journals. PREHYDROLYSIS (FIRST STAGE) The studies on the first-stage hydrolysis or prehydrolysis were intended to simulate, as nearly as possible, the performance of a continuous digester with direct-steam heating. Successful modeling of the system would allow prediction of yields as a function of time as shown in Figure 2. This figure shows the removal of xylose from the wood and the amount of solubilized xylose and furfural produced. It applies to a particular sample of wood being reacted at a specific set of hydrolysis conditions. These curves are for 9-mmsouthern red oak chips reacting in a sulfuric acid solution at pH 1.7 and having a liquid-to-solid ratio of 1.35 when charged to the reactor. After charging to the digester, they are heated to 170°C with direct steam. It was found that complete modeling of the system was not possible. The difficulty lies in predicting the rate of removal of the xylan from the wood. The lower curve of Figure 2 must be experi mentally determined for particular wood samples at each set of hydrolysis conditions. Continuing research on the kinetics of hemi cellulose hydrolysis will undoubtedly remove this restriction, but at present it is not possible to estimate the removal curve with suffi cient accuracy for process calculations. However, having established the removal curve, the yields of xylose and furfural can be predicted satisfactorily. Figure 3 shows experimental data for the removal of xylan and yield of total xylose in solution (for southern red oak chips con tacted with a 1.45 percent H2SO4 and reacted at 170°C). The lower curve through the xylan yield data is the empirical functional relationship: bt ct Xylan Yield = a e - + (100 - a)e- where a, b, and c are constants determined from the experimental data. As noted previously, it is necessary to determine this relation ship experimentally for each substrate and each particular set of hydrolysis conditions. However, once having established the rate of xylan hydrolysis, it is possible to predict the soluble xylose yield. This is the upper curve in Figure 3, which is seen to agree satisfac torily with the experimentally determined points. Since the xylose in solution is simply the difference between xylose released to the solu tion and the xylose loss, one need only know the amount of xylose 1154 1155 Figure 3.--Xylanremoval and soluble xylose yields versus hydrolysis time. Experimental values compared to predicted southern red oak chips (9 mm), 170oC, pH - 1.7, direct steam heating. 1156 destroyed during the hydrolysis to construct the upper curve. Root (13) studied the kinetics of xylose decomposition in aqueous H2SO4 solu tions, and his data were used in this calculation. However, the calculation is quite involved because of the difficulty of determining the acidity of the hydrolyzing solution. The acidity is dependent on the concentration and amount of the acid solution used, the neutralizing capacity of the wood, and the move ment of solution during heating. These complicated relationships have been discussed in detail in other publications (7,15). It was found that, at the low liquid-to-solid ratios employed in this work, the neutralizing effect of the substrate was significant; in some instances as much as 50 percent of the applied acid was neutralized. It was determined that not all of the inorganic cations present in the wood were solubilized, and further, that they were released slowly as hydrolysis proceeded. The water content of the hydrolyzing solution is determined by the amount of water charged with the chips, the amount of steam condensed in heating the chips, and the subsequent movement of the solution. It was found that, on heating with direct steam, chips lost solution from the interior while steam was condensing on the surface. All of the aforementioned factors must be considered to determine the actual acidity of the hydrolyzing solution, which is then used to calculate the xylose yield in solution, the upper curve of Figure 3. Furfural yields were also estimated using Root's data. From this information it was determined that at the conditions of prehydrolysis they should be nearly solely dependent on the amount of xylose decom posed and independent of the particular conditions of hydrolysis employed. Figure 4 compares the experimental values for a large number of runs with that predicted. The wide scatter in the data is believed due to the difficulties and differences of the experimental method. To summarize: It was determined that complete modeling of the first-stage hydrolysis was not possible, that it was necessary to determine, for each substrate and set of hydrolysis conditions, the xylan removal- time relationship.