. v&- A .1 ' v“ 1'.»- ...x m. nwhunrvupA, a am;- y- w « nn' - w 29% . Y w 9;: a 3,3 ‘, ’ 4.7; in I' .L‘ . . "v .Lo.‘-,‘ .‘. ‘7: ‘ 2" K‘ . V ‘ ‘ I. A . “WU-3,43" ' “4 "'01 1 3mg“: -.c.‘,-_. ‘ '.' , N: 1’ '. " L Kath-{1‘ 174' “I r h ‘ n 1 .- m3- s. W nun} 3, vii 1"4 F-L’whc‘l.‘ : «.uI‘ u 'In'hO- .’.",- ,- "Qvn \: w i w" I, 1“! man n u. 73-74 '9’.“ lY I at ‘ ‘go vfizo 9“» J Jan? wMn I." , , w.“ x . . p . . ‘. - «'Nr- . ’ l_' . . ;, "n r: . 1 ! I ..m., mam". .~ LIBRARY Michigan State University This is to certify that the thesis entitled GRAVIMETRIC ANALYSIS OF HYDROGEN FLUORIDE VAPOR ABSORPTION BY BIGTOOTH ASPEN WOOD presented by William Richard Mohring has been accepted towards fulfillment of the requirements for Master of Science degree inChemical Engineering fi/fiflf& 9-7 A} ‘ Major professor Date 3;;Vi 83‘ /7?8 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution IV1ESI.J RETURNING MATERIALS: Place in book drop to LlaRARIEs remove this checkout from .—_——. ' your record. FINES wi ll be charged if book is returned after the date stamped below. GRAVIMETRIC ANALYSIS OF HYDROGEN FLUORIDE VAPOR ABSORPTION BY BIGTOOTH ASPEN WOOD BY William Richard Mohring A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemical Engineering 1988 ABSTRACT GRAVIMETRIC ANALYSIS OF HYDROGEN FLUORIDE VAPOR ABSORPTION BY BIGTOOTH ASPEN WOOD BY William Richard Mohring To obtain data necessary for evaluation of hydrogen fluoride (HF) saccharification processes, a gravimetric technique for measuring the quantity of HF vapor absorbed by lignocellulose was developed and used to measure absorption and desorption isotherms for dry bigtooth aspen wood (Populus grandidentata) for temperatures from 30 to 80°C and HF partial pressures from 0 to 1 atm. Hysteresis was observed in the absorption-desorption loop, apparently due to cellulose decrystallization and saccharification. Using sugar yield data obtained elsewhere, the minimum HF requirement to achieve the maximum sugar yield was determined to be between .32 and .60 g HF/g wood. The heat of vaporization of HF from wood was calculated from the isotherm data using the Clausius-Clapeyron equation. Kinetic data for absorption and desorption was shown to be influenced by heat transfer limitations at the conditions studied. ACKNOWLEDGEMENTS The author wishes to thank Dr. Martin Hawley, Mr. Gregory Rorrer, and Dr. Eric Grulke for their assistance in completing this project and also the Department of Chemical Engineering and the United States Department of Agriculture for financial support. ii TABLE OF CONTENTS INTRODUCTION Literature Survey of HF Saccharification Objectives of This Study EXPERIMENTAL Description of Apparatus Experimental Procedure Additional Experimental Information RESULTS AND DISCUSSION Absorption and Desorption Isotherms Relationship Between Absorption and Yield Data Thermodynamics of Absorption Kinetics of HF Absorption and Desorption CONCLUSIONS FUTURE WORK APPENDICES A TECHNICAL INFORMATION AND PROCEDURES FOR EXPERIMENTS HF Flow Measurement and Control Data Acquisition Thermobalance Operation iii 15 16 18 18 25 27 3O 34 36 38 38 38 42 S3 HF Loading Measurement HF Neutralization B EQUATIONS USED IN THERMODYNAMICS DISCUSSION Derivation of Clausius-Clapeyron Equation for the HF-Wood System Equation for the Association Factor C ISOTHERM DATA REFERENCES iv 56 59 61 61 63 65 70 II. III. IV. LIST OF TABLES Effect of HF Loading on Maximum Sugar Yield Heat of Vaporization Data HF Flowmeter Calibration Results List of Variables in Data Acquisition Program 25 28 39 44 LIST OF FIGURES Flow Diagram of Apparatus Reactor Schematic Absorption Isotherms for 30, 40, 50°C Absorption Isotherms for 60, 70, 80°C Absorption and Desorption Isotherms for 40°C Absorption and Desorption Isotherms for 60°C Clausius-Clapeyron Plots Absorption Transient Data Acquisition Program Listing vi 10 12 20 21 23 24 29 31 47. INTRODUCTION In view of the limited and somewhat unpredictable nature of the world's supply of fossil fuels, significant incentive exists to develop processes based on renewable feedstocks for the production of fuels and chemicals. Many processes based on biomass materials have been proposed, but few appear to be economically competitive, at present, with those based on petroleum. However, the combination of rising petroleum costs and improved biomass-conversion processes could make large-scale biomass utilization attractive in the near future. Wood, also called ligno- cellulose, is attractive for this purpose since large quantities can be grown economically and because large amounts are already availible as waste from various industriesl. Wood consists of cellulose, hemicellulose, lignin and minor amounts of extractives and ash. Cellulose is a crystalline polymer of glucose units (CGH 05) joined by 10 1,4-beta—glucosidic bonds and makes up 40-50 wt % of the wood. Hemicellulose is a mixture of amorphous poly- saccharides and makes up 25-35 wt % of the wood. Lignin is an amorphous, irregular, organic polymer containing methoxy-substituted propylphenol units and makes up 18-35 wt % of the woodz. One way of utilizing wood is to depolymerize the cellulosic and hemicellulosic fractions to simple sugars (a process called saccharification) which can then be fermented to various products, such as ethanol or butanol. Uses for the lignin fraction are less obvious but it may be possible to use it as a feedstock for phenol or other aromatic chemicals. Alternatively, lignin may be burned as a fuel. Saccharification of wood can be accomplished by hydrolysis using dilute acids, concentrated acids, or enzymes. With dilute acids, the hydrolysis rate for cellulose is slow, apparently due to its rigid crystal structure which hinders the glucose ring-flexure step (to form a carbonium ion) in the reaction mechanism. The hydrolysis rate is comparable to the sugar degradation rate even under optimal conditions so that the yield of sugars is limited to about 50% of theoretical for dilute-acid processesB. Yield is an important factor because the cost of the wood will be a significant fraction of the total product cost. Concentrated acids dissolve the crystalline cellulose (which is held together by hydrogen bonds between hydroxl groups along the chain) so that hydrolysis occurs rapidly, even at near-ambient conditions where sugar degradation is minimal. Thus, yields approaching 100% are possible. The main disadvantage of concentrated-acid processes is the high cost associated with acid consumption and/or recycle4. Enzymatic hydrolysis of cellulose has been demonstrated, but processes suffer from low yields and high chemical costs. The present study is part of an ongoing research program at Michigan State University investigating anhydrous (concentrated) hydrogen fluoride (HF) sac- charification of lignocellulose. Because of its strong tendency to hydrogen bond, HE is particularly effective in dissolving cellulose. HF saccharification produces the high yields common to other concentrated-acid processes and, because HF can be efficiently desorbed and recycled after reaction, an economical recovery and recycle system may be possible. However, HF recovery must be very high because HE is expensive. Literature Survey of HF Saccharification The first major studies of hydrogen fluoride saccharification were done in Germany by Helferich and BottgerS and Fredenhagen and Cadenbach6. In 1933, Fredenhagen and Cadenbach showed that HP saccharification of cellulose proceeds through cellulose dissolution and the formation of a glucosyl fluoride intermediate (C6H 05F). 11 The reaction rate was found to decrease with increasing water content of the system. In the presence of HF and H20, glucosyl fluorides were found to revert to easily-hydrolyzed sugar oligomers. They also showed that saccharification of spruce wood is feasible using liquid or gaseous HF (normal boiling point of HE is 19.50C). In the gas-phase process, a 95% yield of sugar was obtained using an HF:wood ratio of 1:1 by mass. With smaller amounts of HF, the yield was lower regardless of the reaction time. Almost 99% of the HF could be removed after reaction by evaporation at 100°C. The sugar oligomers were then washed from the lignin using H O or dilute acid. 2 As a result of the work of Fredenhagen and Cadenbach, two processes utilizing HF saccharification were proposed. In a continuous process patented by Pfleider and Koch7 in 1933, wood was moved though a long tube using a helical screw. The first section of the tube was cooled to provide for HF absorption from the gas phase and subsequent reaction. The second section of the tube was heated in order to desorb the HF. Air blown countercurrently to the movement of the wood returned the desorbed HP to the first section of the tube, where it was reabsorbed. Hoch and Bohunek also developed a process, which was described by Luers8 in 1938. In this process, wood chips with a 2-3% moisture content were exposed to gaseous HP at a reduced pressure of 30 mm Hg and a low enough temperature for 40 kg of HF per 100 kg of wood to be absorbed. After reaction, the HF was removed at low pressure by increasing the temperature to a maximum of 62°C. The HF was then re- cycled after removing acetic acid that was produced as a by-product. A pilot plant Operated in Germany for six months using this process. Apparently, work on both of these processes was discontinued due to World War II, as no further developments are present in the literature. HF saccharification was also studied in the Soviet Union9 in the 1950's, where high sugar yields were obtained using cotton linters (nearly pure cellulose). The reversion oligomers formed after HF evaporation were also studied, and the glucosidic bonds between the units were determined to be 1,6-a1pha. HF saccharification was reinvestigated starting in 1979 here at Michigan State University (MSU) and also at approximately the same time by several researchers in Europe. At MSU, the first investigations used liquid HF with a water content of less than 10% and pure cellulose or bigtooth aspen (POpulus grandidentata) chips. Yields of D-glucose (after post-hydrolysis of the reversion olig- omers) from pure cellulose approaching 100% were attained after 1 h at temperatures from 0 to 23°C. Yields of D-glucose and D-xylose from wood were around 80%. After extended evacuation at 100°C, the fluoride content of the reacted aspen wood could be reduced to as low as .4% by weight, with the residual fluoride apparently bound by metal ions in the woodlo. Also, the glucosyl fluoride intermediate was isolated and identifiedll. DeFaye12 has studied the composition of HF sol- utions of cellulose and proposed that glucosyl fluoride formation takes place through a carbonium ion; same as the accepted mechanism for acid-catalyzed hydrolysis of cellulose (although HF saccharification is not an hydrolysis because no H20 is consumed). Thus, cellulose decrystallization is probably necessary to achieve fast reaction rates for HF saccharification just as it is for acid-hydrolysis processes. The gas-phase HF process has also been investigated at MSU13,14 . From the standpoint of commercialization, it is probably superior to the liquid-phase process because less HF is needed to achieve the same conversions, making HF recycle more economical. Also, the temperature in the gas—phase process can be varied above the normal boiling point of HF while still keeping the system at atmospheric pressure. Results of kinetics studies using aspen wood showed that an 80-90% yield of D-glucose and a 70-80% yield of D-xylose could be obtained (after post-hydrolysis) in only a few minutes using pure HF vapor at atmospheric pressure and 30°C. Reaction rates were found to decrease with decreasing HF pressure and increasing temp- erature. Maximum yields decreased at low HF pressures and high temperatures, apparently because not enough HF was absorbed to decrystallize the cellulose. Several investigators have recently studied gas-phase processes at the pilot scale15'16’17. In each case, a glucose yield of 80-90% and a xylose yield of 70-80% was obtained and also at least 99% of the HF was recovered after reaction. Objectives of This Study In order to model the reaction kinetics of gas-phase HF saccharification, the rate at which HF is absorbed by lignocellulose and the quantity absorbed at equilibrium under various process conditions needs to be determined. The term "HF Loading”, which will be defined as the mass of HF absorbed as a percent of the lignocellulose mass, will be used to describe HF absorption. The HF loading can be used as a concentration term in developing a kinetic equation for the reaction, which is currently being done at MSU. HF loading data is also necessary in order to develop and analyze gas-phase HF processes. It can be used to calculate HF flows, heat loads, and times necessary to carry out the absorption-reaction-desorption sequence. HF loading data at different HF partial pressures and temp- eratures is needed to analyze schemes for absorbing and desorbing the HF in a continuous fashion. For this study, a gravimetric method for measuring HF loading under controlled conditions of HF partial pressure and temperature was developed. Because no gas-phase species are formed during HF saccharification (except for possible minor amounts of degradation products) which would affect the sample weight, the amount of HF absorbed is the same as the weight change of the sample during HF exposure. The validity of this assumption was tested by comparing the initial weight with the weight after HF desorption. A sensitive balance was used to continuously monitor the sample weight during HF exposure. This method was then used to obtain HF absorption and desorption isotherms (HF loading vs. HF partial pressure at constant temperature) and kinetic data using dry bigtooth aspen wood at HF par- tial pressures from 0 to 1 atm and temperatures from 30 to 80°C. These conditions were chosen as the range likely to be of interest for process design. Low temperature and high HF pressure are desirable for absorbing HF. Temp- eratures much below 30°C are undesirable for a process because of the need for a refrigeration system and pressures above 1 atm are unnecessary and could cause the HF to condense in the equipment. High temperatures and low HF pressures are desirable for desorbing the HF and these are represented in the range of variables. As in other studies at MSU, bigtooth aspen was used because it is representative of species being developed for possible use in "energy plantations.” It is a fast-growing species and has the ability grow out from a stump after harvesting, thus eliminating the need for replanting. The HF loading data obtained was analyzed both to gain insight into the chemical and physical processes which occur during HF absorption and also to determine parameters expected to be useful in reaction modelling and process design studies. Actual reaction modelling and process development is, however, beyond the scope of this paper. EXPERIMENTAL Description of Apparatus The flow diagram for the experimental apparatus is presented in Figure 1. A description of the system follows: HF vapor was removed from the top of a tank containing anhydrous HF liquid (min. 99.9% purity). The tank was held' in a water bath 2-4OC above the normal boiling point of HF (19.S°C) in order to cause HF flow. The tubing upstream from the HF control valve (HF was throttled to nearly atmospheric pressure at the control valve) was heated above the bath temperature using electrical heating cord in order to prevent HF condensation. The flowrate of HF was controlled using a mass flowmeter, electronic control valve and controller purchased from Matheson Company. Calibra- tion of the flowmeter was done by measuring the change in weight of the HF tank for a given time period at a given flowrate. Because the flowmeter output was found to be nonlinear, only two flowrates (.23 t .01 SLPM and .98 i .03 SLPM) were used. Extra-dry nitrogen (min. 99.7% purity) was removed from a high-pressure tank and was further dryed by passing over a molecular-sieve bed before being used to dilute the 9 10 m>~m> Loafing I >m uu>am> guano I >0 “can manxua I n “mnoum musumummawh I a «005mm whammmum I m uuuumsmuom I m uw>nm> Houucou m: I m> «umuusaoz man: u: I mm 363.55: mum: coach; I E Eamon; com MU005 Sum .033 3.35 >¢ In ”/kgmwmr. .I AU Av mu _4 “=1 _ .23... I32: m> mu ~> v) V“. v ‘ .0.J 8 .CJ..”u......Q .O. v A I. “u w> .I I "I O _ w I, a .o.o.o.n.:.oi.o.o.). . i 2...: ..... o IIIIIIIII l I; 105.043».— _enm H wk moz<4 m NP> 325.1% .5 2,2135 so: .. F .9“. 11 HF. The flowrate was controlled using a mass flowmeter from Matheson (0-3 SLPM range, 1% accuracy) and a manual metering valve. In order to prevent HF condensation which can occur during mixing with N2, the HF stream was heated prior to mixing using heater cord wrapped around the tubing. The diluted stream was preheated to the temp- erature of the reactor and then was fed to the reactor to contact the sample. Temperature control for this stream was provided by a PID controller and a thermocouple located in the feed tube to the reactor. The partial pressure of HF in the reactor was deter- mined from the HF and N2 flowrates and all the experiments were conducted under flow conditions at atmospheric pressure to maintain a constant concentration of HF in the reactor. The thermobalance (reactor and balance) used a novel design adapted from Costa and Smith18 whereby the balance mechanism and the sample are put in separate enclosures without a direct connection to avoid contacting the balance with corrosive HF vapor. A schematic diagram of the reactor is given in Figure 2. The reactor was made from a 304 stainless-steel pipe (1.5”, schedule 40). The top and bottom halves were flanged to permit sample replacement. A thin sample of bigtooth aspen wood (20 to 100 mg) was held in the bottom-half of the reactor by a .001” diameter chromel hang-down wire which was connected to a Cahn 2000 electrobalance located above the reactor. The 12 Fig. 2 - REACTOR SCHEMATIC I? <—.001” DIA. WIRE .015” DIA. ORIFICE 'buy ’. CONTfim‘EETNT N2_, ?. .oa” DIA. ORIFICE I @W ””””" "F'NZ —-» BAFFLE OUTLET m\\\\\\\\ 71/11 7111‘ ///n\ 'l’llA .I 1- HEATING ELEMENT SUPPORT PLATE .1 I I .I I. I ’ SAMPLE THERMISTOR—> ' o .5 p "oo D .5..:.'.‘."‘. 5——STEEL BALLS [0.0.0.3. IIF-N2 INLET 13 wire passed through two orifices (.015” and .03” diameter) located in the tOp-half of the reactor. A specially designed stand kept the balance and reactor aligned so that the wire did not rub on the walls Of the orifices. The weight of the sample was measured to within .1 mg. The sample was placed inside a doubled-over piece Of monel screen, with a piece Of teflon film covering the bottom of the sample to prevent particles from dropping Off the sample, which tended to become brittle after exposure to HF vapor. The HF-N2 mixture entered the bottom of the reactor, flowed up past the sample and then exited through a baffle which was inserted between the flanges. A low flowrate of N2 ("containment N ") was metered to the top of the reactor 2 using a rotameter and leaked out through the two orifices, creating a slight pressure increase which prevented the gases in the bottom of the reactor from escaping through the top of the reactor and into the atmosphere. The tOp of the reactor was heated to prevent HF condensation near the lower orifice where HF and N2 mixed. The baffle between the flanges prevented N2 that was flowing down from the tOp Of the reactor from affecting the HF concentration. The tare weight of the sample holder and wire was affected by HF adsorption and by the N flowing out through the 2 orifices, which created drag on the wire. TO eliminate these effects from the results, a correction factor was determined by running the system under the same conditions 14 as for the absorption experiments but with no sample present. The temperature in the bottom-half Of the reactor was maintained to within 1°C of the desired value by an electrical resistance heater wrapped around the reactor. Power input was controlled using a proportional controller and a thermistor probe located near the sample. The reactor was insulated for experiments above 40°C. Preheating of the feed discussed earlier drastically reduced the heat input to the reactor (because of the high heat capacity of HF vapor near saturation) and made it much easier to control the reactor temperature and reduce variations from the set point. Steel balls placed in the bottom Of the reactor also helped to heat the incoming feed. The effluent from the reactor was passed through a bed containing calcium carbonate chips to neutralize the HF before releasing the gas stream into the atmosphere. TO ensure that there were no Obstructions in the flow system, a rotameter was placed downstream from the bed so that the flow of gases leaving the bed could be detected. Because of the high toxicity Of HF, every precaution was taken tO prevent Operator exposure. Protective clothing was worn while Operating the equipment, which was situated in a laboratory fume hood. NO significant corrosion problems were encountered using the anhydrous HF. Most Of the tubes and all of the 15 valves contacting HF were made Of monel metal. Copper tubing and steel fittings were also used successfully as was the 304 stainless-steel reactor. Conditions in the calcium carbonate bed tended to be more aggressive due to the water formed in the neutralization reaction, which absorbed some HF to form hydrofluoric acid. The column was made from polyvinyl chloride and was not affected, although all Of the metals used, including monel, were corroded. All of the process variables were printed out during the course Of the experiment and saved on a floppy disk using an IBM PC-XT microcomputer equipped with an analog-to-digital converter. Experimental Procedure The following is a summary Of the procedure used to measure the absorption and desorption isotherms: A thin chip Of bigtooth aspen (weighing 20 to 100 mg, with dimensions Of 1 cm x 1 cm x .05 cm to 1 cm x 1 cm x .25 cm) was placed in the tared sample holder which was then attached to the hang-down wire. The chip was heated to 101-1050C in dry, flowing N for about 30 min in order to 2 completely dry the chip (in a separate experiment, 30 min was found to be more than adequate to dry the chips, as determined by the attainment of a constant chip weight). The reactor temperature was then adjusted to the value to be used for the experiment and the weight of the sample was 16 recorded as the initial sample weight. The HF partial pressure was stepped up and then down a predetermined set of values. Before the HF pressure was changed, the chip weight was allowed tO come to steady-state. Steady-state was assumed to be achieved after the weight remained steady for at least 5 min. Some points were run for longer times (>1 h) and did not show appreciable weight changes after the 5 min criteria was met. For the experiment in which the temperature Of the chip was measured, a .03" diameter hole was drilled .10" deep into the edge of a 1.5-cm x 1.5-cm x .2-cm chip and a small, bare Type T thermocouple probe was lodged in the hole. The chip was held in the reactor and the thermocouple lead passed through a bushing in the side of the reactor. The experiment was then carried out using the same procedure as for the gravimetric experiments except that the weight was not measured. A more detailed discussion of the equipment and procedures used is given in Appendix A. Additonal Experimental Information The average chemical composition Of bigtooth aspen wood is (by mass); 50% cellulose, which is made up Of pure glucan (CGHIOOS); 29% hemicellulose, which is made up Of 75% xylan (C5H804), 20% glucan and 5% other sugars; 16.6% lignin; 4.1% extractives; .3% ash. 17 The definitions of the variables used in the experiment are given below: Weight of HF Absorbed HF Loading 3 Weight of Dry Sample X 100% W + W - W. = WC 1 x 100% i where: W = Apparent sample weight (mg); WC = Weight correction factor; Wi = Initial (dry) sample weight (mg). The resolution of the balance limited the precision Of the HF Loading to 1.5% (20-mg sample) or better, depending on sample size. F HF P = y P = P HF HF a FHF + FN2 a where: PHF = HF partial pressure (atm); Pa = Atmospheric (system) pressure (atm); yHF = HF mole fraction; FHF = HF flowrate (Standard Liters per Minute, SLPM); FN2 = N2 flowrate (SLPM). Standard conditions, used for the gas flowrates, were 1 atm pressure and 21°C. The HF partial pressure was accurate to 1.01 atm below .5 atm and 1.02 atm above .5 atm. RESULTS AND DISCUSSION Absorption and Desorption Isotherms The absorption isotherms at 30, 40 and 50°C are shown in Figure 3 and the isotherms for 60, 70 and 80°C are shown in Figure 4. The isotherm data in numerical form is given in Appendix C. Each curve represents the results of a single experiment where the partial pressure of HF started at 0 and gradually was increased to 1 atm. The typical experiment tOOk between 1 and 2 h. The isotherms include HF which is physically absorbed by the carbo- hydrates and lignin in wood as well as the HF which has reacted to form sugar fluorides. Stoichiometrically, 10 g of HF are required per 100 g of wood to react with the cellulose and hemicellulose. The amount of HF in the form Of sugar fluorides may be less than the stoichiometric requirement, because Of incomplete reaction and because HF may be regenerated by sugar reversion. The reaction sequence is: cellulose or HF __+ sugar _. reversion hemicellulose fluorides "' oligomers DeFaye12 showed that for dilute solutions (10% w/w) Of glucan or xylan in liquid HF at 20°C, equilibrium in the 18 19 reversion reaction shown above is shifted far to the right towards reversion oligomers for xylan and far to the left towards D-glucosyl fluoride for glucan. The systems formed when HF vapor is absorbed by lignocellulose are probably very similar tO solutions of cellulose, so that reversion, especially with D-xylose, may occur at the conditions under which the absorption isotherms were measured. From equilibrium considerations, lower HF loading would tend to favor sugar reversion; temperature will also have an effect on the extent Of reversion. The absorption isotherms exhibit a characteristic shape which will be described here. The slopes appear to be relatively large near P F=0 but decrease rapidly with H HF' NO data was Obtained at PHF less than .07 atm because of equipment limitations which made it increasing P difficult to get very low, stable flowrates of HF vapor. Starting at an HF loading Of about 15% and extending to 25%, the lepes of the isotherms increase rapidly, forming "bumps" in the curves. At around 25%, the slopes decrease somewhat and the curves continue up smoothly as PHF increases to 1 atm. When the dryed aspen was immediately exposed tO HF vapor at 1 atm, the HF loading Obtained is greater than that obtained from the isotherm experiment (where the HF pressure was increased gradually). At 50°C, the difference in HF loading is 8% (.08 g HF/g wood); at 30°C, the difference is 16%. 20 Fig. 3-ABSORPTION ISOTHERMS FOR 30°, 40°, 50°C I70 , l60 '- I ISO " O '40 .. O 30°C '30 I- D 40°C ‘20 I. A 50°C 0 ”0 ' / l00 ' 90 '- / I: so - / a/ 70 - ° / 60 - o a/ so I- / / / 40 P ’0 a) A/A so .. 20° / /A/ ... ,. 1° )0 . : 1.x“ 0L I I L I I I I I I I 0 .IO .20 .30 .40 .50 .60 .70 .80 .90 l.00 PHF (ATM) H F LOADING (7.) O 21 Fig.4-ABSORPTION ISOTHERMS FOR 60°, 70°, 80°C HF LOADING(%) 50 ' 4o .. O 60°C ' O D 70°C /0/ o a / °’ 20 r o O/ /o/ u/ A I0 - ’0 ,n/ AA] 0" 0—" $.72“ 0"" agggzia’ i I I I I I J I I I I ' 0 .IO .20 .30 .40 .50 .60 .70 .80 .90 LOO PHF (ATM) 22 Desorption isotherms were measured after completing the absorption measurements. These are shown along with their associated absorption isotherms for 40 and 60°C in Figures 5 and 6, respectively. For both temperatures, the absorption and desorption isotherms are coincident above an HF loading Of approximately 25%. Below this value, the desorption isotherms lie above the absorption isotherms, thus exhibiting hysteresis. The desorption isotherm for 80°C was also measured (but not shown) and found to be coincident with the absorption isotherm over the entire range of P In all the experiments, it was possible to HF' return the wood sample to, or slightly less than, its initial weight by desorbing the HF into pure N This 2. indicates that the sugar fluorides reverted to oligomers, releasing the HF. Referring tO Figures 5 and 6, the HF loadings at P F=0 on the desorption isotherms are negative, H which is a result of the final desorbed weight of the sample being less than the initial weight. This may be due to the formation Of volatile degradation products such as acetic acid. It also may be due to some error in the measurement system, since the weight losses were not consistently observed. In all experiments, however, the final desorbed weight was within 2.5% Of the initial weight. The precision of the isotherm measurements was examined by repeating a number of them. Between HF loadings Of 15 and 25% (approximately), where the isotherms 23 Fig. 5 —ABSORPTION/DESORPTION ISOTHERMS FOR 4O°c I00 *- 9 I. A ° /° °\o 80 " O ABSORPTION 3 :5 70 El DESORPTION / 2 6° ' o/ 2 5°' I” _I 30 - /cl LI. 0 _ 0.0 :I: 2 £64 l0 f0” 0 I I I I I L I_ I I 0 .IO .20 .30 .40 .50 .60 .70 .80 .90 LOO PHF (ATM) 24 Fig.6 -ABSORPTION/DESORPTION ISOTHERMS FOR 60°C 50 '- 3 4o 0 ABSORPTION 0 ° / v U D ESORPTION b 0 30 - °/ z ./ o :7 8 20 I- / _I 7f U ’0/0 O 0 q J 1 l l I I J J l I 0 .I0 .20 .30 .40 .50 .60 .70 .80 .90 LC PHF(ATM) 25 are steep, the difference in measured HF loading between identical runs was found to be as large as 10% (.10 g HF/g wood). In other areas of the isotherms, the deviation was typically less than 2%; the largest Observed difference was 3%. Some of the discrepancies can probably be attributed to slight differences in sample composition. Relationship Between Absorption and Yield Data The minimum HF loading required for maximum sugar yield is an important quantity from a process standpoint. Table I shows maximum D-glucose and D-xylose yield data Obtained by Rorrer14 for bigtooth aspen wood at various temperatures using HF vapor at 1 atm and also the HF loading data from this study under the same conditions: Table I: Effect of HF Loading on Maximum Sugar Yield Temperature HF Loading Glucose Yield Xylose Yield 30°C 160% 83% 70% 40 89 85 60 so 60 85 63 66 32 72 63 81 13 48 46 From this data, the minimum HF loading required to Obtain the maximum yield of D-glucose is between 32 and 60%. For D-xylose, the minimum HF loading is between 13 and 32%. The lower requirement Of HF for maximum xylose yield is consistent with the fact that it is in the amorphous 26 hemicellulose fraction of wood. It is possible that by varying HF partial pressure instead Of temperature (as is done in Table I) to change the HF loading, somewhat different results for the minimum HF loading could be Obtained. These results for minimum HF loading are in the range of values determined by other investigators using different types Of lignocellulose. Fredenhagen and Cadenbach6 found that an HF loading of 100% was necessary for spruce wood. Franz15 reported between 40 and 80% was necessary, depending on the starting material and whether it was subjected to prehydrolysis. The yield data is also helpful in trying to explain the unusual bump Observed in the absorption isotherms at HF loadings Of 15-25%. Because the yields Of glucose and xylose are increasing towards their maximum values at HF loadings Of 15-25%, the bump in the isotherms may be due to chemical or physical changes related to saccharif- ication. Most apparently, the disruption of the crystal structure of cellulose may make the hydroxyl groups more accessible and cause an increase in HF absorption. The hysteresis Observed in the absorption-desorption lOOp below an HF loading of 25% is consistent with this explanation since the changes caused by saccharification or decrystal- lization do not reverse themselves during desorption. 27 Thermodynamics of Absorption From the absorption isotherms, it is possible to calculate the heat of vaporization of HP from aspen wood at various levels Of HF loading using the Clausius-Clapeyron equation. These heats Of vaporization are expected to be useful in process design studies for calculating heat loads and also for studying the effects of heat transfer on the kinetics Of the HF saccharification process. The Clausius-Clapeyron equation (derived in Appendix B for this application) is: ln PHF = -AHVf + c RT where: PHF = Partial pressure of HF (atm); AHv = Heat of vaporization Of HF from wood (cal/gmol HF): f = Association factor for HF vapor; T = Temperature (K); R = Ideal gas constant, 1.987 cal/gmol K; c = Constant. This equation can be applied tO P vs. T equilibrium data HF where the HF loading is constant. Through the association factor f, which is defined as the ratio of the actual HF vapor density to its ideal gas value, the equation can take into account the fact that HF vapor is polymerized (slightly) at the conditions studied here. In order to Obtain a linear relation between ln PHF and l/T required 28 for accurate evaluation Of AHv from the lepe Of the line, f should be approximately constant. The semi-log plots Of P vs. l/T are shown in Figure RF 7 for HF loadings of 10%, 35%, 60%, and 90%. The lines in Figure 7 were fitted to the data using linear regression. From the slopes Of the lines, AHv's were calculated. The required association factors were calculated from the equations of Smith19 (given in Appendix B), who fitted HF vapor density data by assuming a monomer-tetramer-hexamer equilibrium exists in HF vapor. Fortunately, the association factors are relatively constant for a given HF loading: at HF loadings of 10% and 35%, f varies by less than 1%: at 60%, the variation is 3%; and at 90%, the variation is 5%. These data are presented in Table II: Table II: Heat Of Vaporization Data * HF Loading f AH AH v v 10% 1.01 10700 cal/mol HF 10800 cal/mol HF 35 1.04 8200 8500 60 1.14 6300 7400 90 1.49 4500 6700 00 3.81 1800 7300 For comparison, the heat of vaporization Of pure HF (In HF loading) at its normal boiling point is included in Table II. Also included is AHV*, which is the enthalpy difference between HF in the hypothetical ideal gas state (as a monomer) and HF absorbed by lignocellulose. 29 Fig. 7 - CLAUSIUS-CLAPEYRON L00 PLOTS .90 .80 . \l‘\\ 1221\::\\ 1UI| .40 " ’2‘ I- .30 A\ S, u- A :r CL.20 - ° 83 0 IO% LOADING A 35% LOADING D 60% LOADING .I0 x 90% LOADING ° .09 .08 .07 .06 111'! 2.8 2.9 3.0 3. I 3. 2 3.3 fi-‘K' X IO3 30 Excess-enthalpy data from the literature20 was used to calculate AHv* from AHv. The Observed decrease in AHV* with increasing HF loading can be attributed to enthalpy changes in the solid phase. The decrease in AHv is a result Of enthalpy changes in both the gas and solid phases. Using the AHv data Obtained above, the adiabatic temperature rise resulting from increasing the HF loading from 0 tO 60% is calculated to be ZOO-300°C. Experimental temperatures will, Of course, be limited by the equilibrium temperature from the absorption isotherm. Kinetics Of HF Absorption and Desorption Kinetic data for HF absorption and desorption is necessary for reaction kinetics studies as well as design studies. Based on the previous discussion of thermodynamics, one would expect heat transfer limitations (internal or external to the chip) to influence the kinetics. In order to examine HF absorption kinetics and the possible influence Of heat transfer limitations, two experiments will be described. In the first, a 1-cm x l-cm x .2-cm dried aspen chip (SS-mg weight) was exposed to HF vapor at .5 atm (1 SLPM of HF and 1 SLPM Of N2) and 30°C and the HF loading was measured as a function Of time. This transient curve is presented in Figure 8. As can be seen, HF absorption is quite rapid: the equilibrium loading 31 Fig. 8 - ABSORPTION TRANSIENT o’o‘o—O—O-‘0-0 60 '- ’04“, /° ” o 1” I 50 " l4 I” A O I a: J” v 6' Q 40 - I” z ,’ f 5 ’ ° <1: ' ’ O 30 r: ° .1 :0, GAS TEMP=30°C II. :I TIME=0,Pm:=0 I 20 1° I, TIME>09 pHF=050 ATM :‘I" EQ CURVE —--- I0 . O o! I l I I 0 400 800 I200 I600 TIME(S) 32 was achieved within 30 min. The second experiment was identical to the first except that a thermocouple was inserted into the edge of the chip and the gravimetric system was not used. From this experiment, the temperature Of the chip was found to rise to 75°C within 10 s after HF exposure and then gradually decrease to the gas temperature Of 30°C. The temperature data Obtained and the isotherm data were used to construct an equilibrium curve, which represents the maximum HF loading which could occur given the temperature Of the chip. This curve is also shown in Figure 8. If heat transfer resistance-completely controlled the kinetics, one would expect the two curves in Figure 8 to be coincident. This is not quite the case, and some Of the discrepancy may have been the result Of temperature gradients within the chip or temperature measurement errors. In any case, the results Of these two experiments show that heat transfer resistance (either external or internal) affects the kinetics at the conditions of the experiment and probably for most Of the experiments that were done. Because of this, the acquisition of kinetic data was de-emphasized for this study. HF desorption proceeds at rates comparable to absorption except at low levels of HF loading. For the experiment described above where the chip temperature was measured, the chip temperature decreased to as low as 13°C when the HF was desorbed into pure N2 (1 SLPM) at 30°C. 33 This temperature decrease, due to evaporative cooling, shows that heat transfer limitations will affect desorption kinetics at the conditions Of the experiment. When desorbing HF into pure N the rate Of desorption was found 2. to slow down markedly at HF loadings Of less than about 10%. This may be due to the fact that sugar fluorides must revert to oligomers to release all of the HF from the sample, and this reversion reaction probably occurs more slowly than desorption of physically absorbed HF. Temperature has a significant effect on the rate Of desorption Of HF, especially at low levels Of HF loading. For a chip initially exposed to HF vapor at 30°C and .5 atm, it took 1 h to return the chip to a constant weight (.2% resolution) by desorbing the HF into pure N at 80°C. 2 By contrast, if the same chip was desorbed at 30°C, the HF loading would still be about 5% after 1 h and at least 8 h would be required to attain a constant weight. CONCLUSIONS Gravimetric analysis was found to be a useful method for studying HF absorption by lignocellulose. Although the sample weight was found to decrease by as much as 2.5% after HF desorption (possibly due to degradation), this error is small in comparison with the HF loading Observed under most conditions. Because of the possibility of weight loss by the sample, however, gravimetric methods do not appear to be useful for measuring the residual fluoride content Of materials after HF saccharification. The HF absorption isotherms are very steep between HF loadings of approximately 15 and 25%. This is most likely due to the decrystallization Of cellulose making it easier for HF tO interact with the hydroxyl groups on the cellulose chain. The phenomenon does not reverse itself during desorption; thus, hysteresis is observed in the absorption-desorption lOOp. Using the HF loading data and the heats of vaporization that were calculated in this study along with data on reaction yield, it will be possible to perform material and energy balances necessary for process design and analysis. The effects Of using different types Of lignocellulose with varying moisture content on HF loading 34 35 still needs to be determined so that the process can be evaluated more completely. HF absorption by lignocellulose is a rapid process at the conditions studied. HF desorption proceeds at similar rates except when the HF loading is less than 10% and the temperature is low. The data suggested that the absorption kinetics are controlled primarily by the rate Of heat transfer away from the wOOd, but this needs to be confirmed with more experimentation. Transport limitations will cause difficulties in modelling HF saccharification kinetics because the process may not be isothermal and because the absorption kinetics will be difficult to describe. For commercial processes where the wood chips will be larger than used here, transport resistances may be even more important and signicantly influence reaction rates . FUTURE WORK Many possibilities exist for additional work related to the Objectives Of this study. Some Of them will be discussed below: The effects Of absorbed H O in wood ((10%) on HF 2 absorption should be studied. The presence of water could significantly increase the quantity of HF absorbed over what would be absorbed by dry material because HF is very hydrophilic. The results Of these experiments are important because, for a commercial process, it may be uneconomical to completely dry the feed. Experiments utilizing wet samples are somewhat more difficult to interpret because H O can evaporate during HF exposure, 2 which will affect the sample weight (and hence the measured HF loading) and the H 0 concentration. 2 Also, the weight lost by the wood after HF treatment should be studied more carefully than was done here. Weight loss may be attributed tO volatile degradation by-products. The amount of volatile by-products produced would be important in analyzing HF recycle schemes, because they may have to be removed from the HF. Of course, chemical analysis for degradation products would still be necessary. However, the gravimetric experiments would be 36 37 easy to perform and could be used to confirm results from chemical analysis. These weight-loss experiments could be done with larger sample sizes (up to about 1 g) to improve resolution. Finally, the effects of transport resistances on absorption and desorption kinetics can be studied using larger sample sizes. Information from these experiments should be useful in process scale-up. APPENDICES APPENDIX A TECHNICAL INFORMATION AND PROCEDURES FOR EXPERIMENTS The purpose of this section is to provide more details on the apparatus and procedures used in the gravimetric experiments, especially for those interested in using the apparatus. The discussion of the equipment is limited to nonstandard items. The procedures described should not be attempted until one thoroughly understands the equipment and safety precautions for handling HF. Although the procedures described have been used successfully, the author assumes no responsibility for mishaps resulting from their application or misapplication. HF Flow Measurement and Control A Matheson mass flowmeter (model 8168), electronic control valve (model 8241), and flow controller (model 8250) were used to control HF flow. The flowmeter was calibrated by measuring the weight Of HF removed from the tank (change in tank weight) after removing HF at a given flowrate for a given amount of time (procedure given at end Of this discussion). The HF flowrate was calculated using the equation: 38 39 F = 1.206 x w HF t where: FHF = Flowrate Of HF in standard liters per minute (SLPM) - Standard conditions are 21 C and 1 atm pressure: w Weight of HF removed from tank (9): t Time Of HF flow (min). The flowmeter was calibrated at two flowrates and the results are shown in Table III. In this table, "Runs" is is the number of times the calibration was determined: FHF the flowrate indicated by the flowmeter readout; FHF is the actual flowrate defined above. Varying the time over which HF was allowed to flow during calibration did not affect the results, indicating the flowrate remained constant. Table III: HF Flowmeter Calibration Results * Runs FHF FHF 2 .40 1 .01 .23 1 .01 SLPM 5 2.77 1 .01 .98 1 .03 For all runs, the HF tank was held in a water bath at 22 + .5°C and the flowmeter temperature (measured using a thermistor cemented to flowmeter) was maintained between 24 and 25°C. Increasing the bath temperature to 24°C and the flowmeter temperature to between 25 and 26°C did not affect the results. The significant deviation Observed between the flowmeter readout (based on Matheson calibration) and the actual flowrate is because the flowmeter operates on a 40 heat-transfer principle and Matheson failed to take into account the effect Of HF polymerization on the heat capacity of HF. One must exercise caution when heating the HF tank to prevent the temperature from exceeding 125°F. If this limit is exceeded, the HF liquid in the tank could expand to completely fill the tank and possibly cause tank rupture. To guard against this, a device was attached to the water-bath heater to shut it Off if the temperature of the water became too great. TO prevent the same sort Of thing from happening in the line connecting the tank to the flowmeter, a relief valve was installed in this line. The procedure used tO calibrate the flowmeter is given below. Valve numbering is shown in Figure 1. HF Flowmeter Calibration Procedure Note- Within each numbered section Of the procedure, the instructions should be performed in the exact order in which they are mentioned. 1. Make sure the hood blower is on and working properly. Also make sure that the system has been purged of HF. Ideally, the thermobalance should be removed from the hood and the inlet and outlet tubes for the reactor should be connected together tO prevent HF release if mistake is made in procedure. 2. Make sure that valves v1, v2, v3, v4, v5, v8, v9 and v10 are closed. 3. Open the nitrogen cylinder nearest the hood at v6 and verify the outlet pressure Of 10-15 psig. Also make sure that v7 is Open all the way. 4. Weigh HF tank with valve outlet cap and valve cover attached and record value in notes. Attach the regulating valve to the HF tank. Before tightening the packing nut, purge out area between valves by Opening v2 and then v5 to give a low flow of N2 for a couple of minutes and then 41 close both valves. Purging Of apparatus prior to HF exposure is done to minimize corrosion. Beware of HF leaking past v1. 5. If necessary, fill the water bath. 6. Turn on HF flow controller ("dynablender"), move v4 to "to trap", and put v3 to "full Open". Adjust nitrogen flow through system to about 3 SLPM on HF flowmeter using v5. Purge for a few minutes and then Open v2 and purge for a few more minutes. Verify the flow on the exhaust rotameter. 7. Put v4 midway between ports. Close v3. Wait about 30 s and then close v5. Wait about 5 min and then open v5, Observing for appreciable N flow through the purge rotameter which could indicgte a leak in the HF pressure system. If everything is OK, move v4 to "to trap", close v5 and then Open v3 momentarily tO bleed pressure from system. Finally, close v2 and put v4 midway between ports. 8. Turn on HF line trace power sugply and adjust so that flowmeter temperature is 24 to 25 C. 9.0Turn on water bath temperature controller and adjust to 22 C. 10. Open v10.Turn on nitrogen mass flowmeter and adjust flow to about .5 SLPM using VB. This flow keeps HF from backing up in the lines. 11. Turn on and adjust power supply tO HF control valve heater. Verify that the heater is working by touch. Valve should feel warm but not really hot. Heating prevents HF condensation in valve. 12. Make sure bath and flowmeter temperatures are OK and then Open v1 and v2. 13. Adjust HF flow setpoint to 0. Leave dynablender in ”Set" mode. Move v4 to "HF to Trap", switch dynablender to control mode and then adjust HF flow setpoint. Put dynablender to "Read". Also, switch on timer. 14. Let run for desired time period. Allow enough time for at least 20 g Of HF to be removed from tank. 15. TO shut system down, close v2 and then put v3 to ”full Open." Purge system by Opening v5 and adjusting flow to several SLPM on dynablender. 16. After a few minutes, close v5, close v1, Open v2 and then Open v5 and adjust purge flow to several SLPM. 42 17. Turn off bath heater and turn Off all heater power supplies. 18. Purge with v4 in both positions. Also, purge with v2 closed. Finally, close v5, v3, v4, v8, and v10. 19. Loosen packing nut on HF tank and then Open v2 and Open v5 (CAREFULLY! HF is released into hood) to give a small flow of N , which may be increased after a little while. Purging i6 this manner eliminates HF from leaking while tank is being removed and also prevents valve corrosion. Continue purging for several minutes and then close v2 and v5. Remove valve from tank and put cap on tank. Check for leaks using ammonia. 20. Remove HF tank from bath, replace valve cover and then dry tank. Weigh tank and record value in notes. Data Acquisition An IBM PC-XT and a Data Translation 2805 interface board were used to print out and store data from the absorption experiments. The analog-to-digital converter on the board digitized the voltage or current outputs from the instruments for entry into the computer. Details on the interface board are availible in the manual from Data Translation. The data acquisition program, written in Advanced BASIC, is shown in Figure 9. Definitions Of the variables in the program are given in Table IV. This progam is stored under the file name ”dap3". In order to use this program, the following steps should be taken: 1. The 2805 board should be installed in one Of the expansion slots Of the computer and a 707-T terminal board (located outside the computer) should be connected. 43 2. The instrument outputs should be connected to the proper channels on the terminal board. 3. A printer should be connected to the computer. 4. The 2805 board should be calibrated. 5. The machine language program ”PCTHERM" (supplied by Data Translation) should be on the hard disk of the PC-XT. 6. Advanced BASIC and then dap3 should be run on the computer. The data acquisition program is interactive and contains prompts for the user. Upon running the program, the user is instructed to put a diskette in the floppy drive. Data from the experiment will be saved on this diskette. The user is then instructed to enter the experiment number and the initial sample weight. The computer Opens a data file on the data diskette using the experiment number as the file name. The initial sample weight is also stored in the file at this time. The user then starts the data acquisition mode Of the program by pressing the space bar. The computer then prints the headings on the printer and starts printing data. The data taken are: the time in seconds at which the measurements are made (time s): the weight Of the sample in milligrams from the Cahn balance using the 100 mg range (wt. mg): the flow Of nitrogen to the reactor in standard liters per 44 Table IV: List of Variables in Data Acquistion Program N TE(N) WIN) N2F(N) HFF(N) TEMP(N) ANALOG.VALUE% ADC.VALUE HIGH.VI LOW.V! RANGE! NOC! LSB! CHANBT% CHANW% CHANTPH% CHANTMJ% CHANN2% CHANHF% GAINBT% GAINW% GAINTPH% GAINTMJ% Numeric Variables Data point number Time elapsed from starting program to data point N (5) Sample weight for data point N (mg) Flowrate Of nitrogen tO reactor for data point N (SLPM) Flowrate of HF to reactor for data point N (SLPM) Tgmperature in reactor for data point N ( F) Parameter for pctherm Parameter for pctherm Parameter for pctherm Parameter for pctherm Parameter for pctherm Parameter for pctherm Parameter for pctherm Board temperature channel Electrobalance channel Reactor feed preheater thermocouple (Type T) channel HF-N2 mixing junction thermocouple channel N2 mass flowmeter channel HF mass flowmeter channel Board temperature channel gain Electrobalance channel gain Feed preheater thermocouple channel gain Mixing junction channel gain GAINN2% GAINHF8 GAINT% W0 H M 8 TI TD T2 T SUM I TBC TBF VREF VOLTAGE VS TMJC TMJF TPHC TPHF 45 Table IV (cont'd) N2 mass flowmeter channel gain HF mass flowmeter channel gain Reactor temperature channel gain Initial sample weight (mg) Hours (hr) Minutes (min) Seconds (5) Time when program is started (5) Time delay between data points (5) Time during time delay loop (5) Time at which previous data point was taken (S) Sum Of voltages for channel output averaging (v) Iterative lOOp counter Terminal board temperature (°C) Terminal board temperature (°F) Reference voltage for thermocouple measurement using type T thermocouple (v) Voltage determined from a channel (v) Thermocouple voltage corregted to the. reference temperature of 0 C (v) Temperature at mixigg junction determined from thermocouple ( C) Tsmperature at mixing junction thermocouple ( F) Temperature at feed preheater (°C) Temperature at feed preheater (°F) D$ AD$ TI$ HR$ MIN$ SEC$ TS 03 46 Table IV (cont'd) String Variables Experiment number and data file name Data file name for writing tO floppy disk Time when program is started (hr:min:s) Hours (hr) Minutes (min) Seconds (5) Time (hr:min:s) Question answer 47 O.ucoo camm OOOOHOOOAOAE. oomuxszm.3oqu>.mOHmnmmOz.3oq moH+um>.=on mum=a<>.on< onxmsq<>.OOq nofinzoefic. 0.:8903909: Qaup wasmmmb oxv co OHDAHHN>O. on :Ehonaom: anumoua onv pang can on an“ :m Spa: coonmuopcfi on canon Homm. :vaodnamhp want a van» NOHHOUOH Emhmonm 0:9 .ouzpmhoaaoo can :0wamuvzoocoo. .opnu 30am Ooaaoupcoo on N:\m£ MO Emouum msoommm mGASOHm m on Oomonxo. Canaan Caucanaaooocmaa n no omcmfio vsmdos ofianchv 03¢ mousmmos Scans. Eopnhn Oauvoaa>num m: 0:? Eoum nvnc apocaaoo .mnn.ano .Enumoum canon nfisv. mcwumwa EmumOum :Owuwmwsvoc nude um mmawmm can own Cam com 0mm omm CNN com com ovm 0mm 48 mmZHB”: amenum O.u:OO : Eszma . bzmmmd : EszmJ Roecaua :0 mmaacno: veaua. mm+om*m2+ooom*m umHB Awowqu<>nmm Aazmzvq< um! Aammva<>nmm AN.>.6HBV$GHZNaomm AN.¢.9HBV9QH2nazHZ AN.H.$HBV$QHZu9mm mazoomm Ob 9mm>zoo. 62“» AmfipAAA pom. one qum ocm zmma : «m2H9uaHe .uarmezH Em .pumpm on gun 00666 was: aszm ceapoon mcacuooon spat. 03.Hu msme 9H9HAO cw xnap camp p:m:eszm .EopnAm Oflu905a>num m: uOm Enumoum :Oapumflsdon want: eszm mqo ... ZOHB+ZDmNZDm xbmzH<0\Am>.3OA+A”mma*xm00<>.woq ANm04<>.DOa.UG< Ja monouomou onwasoano. m @809 hopes: wagon cunt pcosouocw. H+2u% owe zmma navAsImsv hH m+om*:+oomm*mnma Anommva<>um Aaszvq<>uz .amqu<>u: Am.p.aevaOH:uaomm Am.v.aevaOHznasz AN.H.OEVOOH2umm: «mzHEnaa OEHQ ouanoHno. 0N0 mmam ommfl zmms : :narmsz hm Swamp mega cudanapacfi. onab hopes: veaom swap onwanapacw. 0n: gnovn OP has ocean paszeszm :nnoumoum a“ pcoefiuonxo: Hzmmm :H snag a: snag N: we .9: m 68A» .AA 6066 . eszmq .. eszAS .ms .uoxu. upnmfioz oAaagm AaApAEA . eszma anu:n.o: POOEwuomxo : Hszmq :mcauzos .3 ”novnmflumo>ca : Bzmmmq 000 000 0pm 000 000 0cm 000 0N0 0H0 000 000 000 0b0 000 000 0G0 000 0N0 0H0 000 005 005 05» 005 00> 0¢b fiE 00> 00> 0H5 00> 000 50 U. ooofi\hmm>+m~\2=mum> H axmz m0+200uzsm untth<0\Am>.300+A”000*xm00<>.000 Axmaq<>.004.00< qa*¢~+w0>0v0.m+0*mH+wN0H00.NIn*HH+H000>0.0+¢*3000 00¢>¢N0Imnaa0.mmmnmomh+m*nmmmm.mVMS0FIm>*ummm¢0 N+3H000000~.nomm9 000~\hmm>+m~\zomnm> H exmz m0+ Sbmuzsm xzmezH.30q+aAmma*xm00<>.000 Axwaa<>.000.00< aa<0 mm 0? Hum mom chasm O~QOOOOEAOSP hognosoum anon. A.n.H ms00a .QOOH 90 m8 a“ unmdoz opnasoHno. Zamnocauazvz H mez mcfivnou mo 05“» mafiahouoc. amszuab ZNEH omuH hm mu+20mu200 N32H<0\Am>.ZOJ+A”mmq*xm04<>.ooq Axm00<>.000.00< JJn0 Hoccnso oocnannouo«e cnou. N¢¢>.Ihmhnflnmmmo.nhmm> .bmbm 00N~ ommfi 0~N~ 00N~ omaa 00H“ 0>HH Goad 00nd oeuu 0m- 0NH~ ouua 00- 000a 000a 0>0H 000a 000a 0¢0~ 000w 51 v. 00> 0900 h0:9.hmme.szmzm9.szhmx.szhmz.A203.szme.zu:uu.uuuuu = 02mm: eszma ... nopnnna no none penna. osfip Oonmnao oanasoflno. HBIBNAZVHB m+00*2+0000*mn9 Awomqu<>um A92H200<>uz Awmqu<>nm AN.>.09000HZH00HW AN.v.$vaQHZN$zHZ AN.H.09000HZNamm nonooom on mannnon mo osnp pno>coo. 0xa00Hv.q0I0*00.00N+v§00.¢0¢I0*mH.05¢+N*>0.NNNIm0*m>.N~N+n00~0.00nszmZMB XEZH<0\AN>.3OA+A”mmq*xm04<>.000 ANMDJ<>.004.00< qa*0.quvhhm XQIZH<0\AL>.SOJ+A”mmq*xmaa<>.004 Axm04<>.UOA.OQ< aa*0.Nszth xmzzH<0\Am>.3OJ+Ammmq*flm04<>.woq Axuaq<>.000.00< 04<0 Honnnno noaoESOHm w: anon. N0+0023§0.~nh029 ncoo 0cm— 000a 0NOH 0HOH 000a 000a 000a 050d 000a 000a 0vma 000a 0NmH 0ama 0*va+M0>0¢0.m+0*0n+mmmdmw.NIm*H~+m00050.0+v*8000 00¢5¢N0I0*uum.nmmmmow>+m*unmmm.nvmhmbIm>*ummmv0.bmhnN+uH000000H.nohtfi 0¢m~ 52 02% H3 H0000 H Exmz HHVHZMB.HHVHHI.HthNz.HH03.AHvHE.Hu HEHm3 z 09 HNH mom z.H: HBHmz 03.H$ mmez ameHnO n« anc anxonn uaa:Bszm 000H 0900 ZMEB :::u$0 hH acu:H:\>0 >000 nsxonn oxnszeomzH H: @0040 H Exwz HHVHZMB.Hthhm.HHVHNZ.HH03.AHVHH.H3 Hmez z 09 HRH mom z.Ha MEHmS OHHM Q.“ 6900 OHOVm. 0N0 0900 0%”:Hm0 nmcafinon :oozuon hnHoO oaapzszmzH onhn 0&00 zmmp :Ozuac EH momzpn3d on a pen npnu mcfixnu osnfivnoo OP 0 mmonnzbsmzH 000H 000H 000H obmH 000H 000— oewH ommH 0NOH 0HOH 000H ombH 00>H obe 00>H ombH 0¢>H ombH ONSH OHFH 00>H 000H 000H 0>0H 000H ommH 53 minute from the Matheson mass flowmeter (n2 slpm): the flow of HF indicated by the mass flowmeter (hf slpm): the temperature in the reactor in degrees Fahrenheit from the Omega Model 49 controller output (temp f): the temperature in the feed preheater tube in degrees Fahrenheit from a type T thermocouple (tph f): and the temperature at the mixing junction in degrees Fahrenheit from a type T thermocouple (tmj f). Data acquisition can be stopped by pressing the space bar. The user then has the Option of continuing data acquistion and also adjusting the time delay between readings or saving the data obtained on the data diskette and then quiting. Some difficulites were encountered with the data acquisition system when reading the small (0-10 mv) weight signal from the Cahn balance. First, it was difficult to eliminate Offset between the voltmeter which was used to zero the balance and the computer output. Thus, it was necessary to subtract the Offset from the computer output to get the true weight. Also, the weight from the computer was found on several occasions to be several milligrams in error. Because Of these two problems, all of the steady-state weights were taken from the voltmeter to eliminate any errors. Thermobalance Operation This section will describe some of the Operating characteristics Of the thermobalance and also give 54 information for setting up the thermobalance. Details about the Cahn 2000 electrobalance used will not be given as they are availible in the Operating manual. The Cahn 2000 electrobalance is a delicate instrument and one must be careful not to exert too much force on the balance arm. The hang-down wire should be removed from the balance while working on the equipment in case the wire is accidently knocked. Also, special hooks should be used in attaching the hang-down wire or counterweights to the balance. These hooks should have a small loop through which a forceps can be inserted to hold the hook. If too much force is exerted on the hook, the forceps will slip out Of the lOOp, preventing damage to the balance. Several factors have been found to influence the weight measured by the balance. First, the hang-down wire may rub on the walls of the orifices in the reactor and cause noise in the weight signal. This noise was reduced to as low as .1 mg by aligning the balance and reactor. Also, the air flow through the fume hood created a small force (<.5 mg) on the balance, the magnitude Of which depended on the hood sash level. Thus, it was necessary tO maintain the hood sash level at the same height while measuring weight. Also, gas flow through the upper orifice in the reactor created a small drag force on the wire. This effect was eliminated from the results by determining a correction factor. Also, HF condensation on the hang-down wire affected the weight signal. HF can liquify 55 in the vicinity Of the lower orifice of the reactor because N2 mixes with HF in this area. HF condensation can actually cause the hang-down wire to stick on the walls of the lower orifice, creating large errors in the weight. Sticking was only Observed at 30°C and could be eliminated by heating the tOp of the reactor. Less severe condensation on the wire caused a slight weight increase and this effect was included in the correction factor. Installation of a new hang-down wire requires that the two orifices be realigned with the balance. The easiest way to do this was to first align the lower orifice and balance with the upper orifice and baffle removed and a hang-down wire with small weight attached threaded through the lower orifice. The tOp-half of the reactor was slid on the support plate until the hang-down wire (connected to the balance) was centered in the orifice as Observed from below. This position was then marked by locating two points on the reactor flange relative to the edges Of the support plate. The reactor was removed from the support plate and a hang-down wire was rethreaded through both orifices and then the tOp-half Of the reactor and the baffle were placed on the support plate and the reactor positioned so as tO align the lower orifice. The tOp-half of the reactor and baffle were then bolted to the support plate. The upper orifice was aligned by sliding it until the hang-down wire was centered in the orifice as Observed from above using a magnifying glass. 56 The .001" diameter chromel wire used for the hang-down wire was purchased from Omega CO. Hooks were attached at either end by tying the wire using a forceps and a magnifying glass. Steel hooks should not be used tO attach the wire to the balance because of the magnet in the balance. HF Loading Measurement The procedure described in this section was used for the absorption measurements. It was also used to determine the weight correction factor, with minor modifications to account for no sample being used. TO prevent HF from leaking out of the reactor during the experiment, the containment N flowrate must be set 2 properly. The necessary flowrate will depend on the backpressure developed in the reactor. As a rough guide, .05 SLPM of N are required per 1 SLPM Of N fed to the 2 2 reactor. This was determined by measuring the pressure difference between the top and bottom chambers of the reactor and increasing the containment flowrate until the difference was 0. The containment flowate should be larger than this minimum value, but not so large as to affect the HF concentration near the sample. The experiments were actually run with flowrates very close to the minimum values. Because the reactor does have a hole in it, special precautions should be taken. First, the valve downstream 57 from the reactor (v10) should never be closed while HF is flowing through the reactor as it would cause the HF tO flow out into the hood. Also, one should stay away from the thermobalance when HF is being used because Of the possibility of leakage through the orifice. The procedure for the absorption experiments follows. Valve numbering is given in Figure 1. HF Absorption Measurement Procedure Note- Within each numbered section Of the procedure, the instructions should be performed in the exact order in which they are mentioned. 1. Make sure the hood is on and working properly. Also make sure that the system has been purged of HF. 2. Make sure valves v1, v2, v3, v4, v5, v8, v9, v10 and v13 are closed. 3. Zero the balance on the DVM (digital voltmeter) with the empty sample holder and teflon tape (to wrap sample) attached to hang-down wire, the reactor bolted together and the hood sash at the specified level. The balance should be on the 100 mg range with filter 2 on. 4. Set up the computer and run dap3 to Obtain zero Offset on the computer. 5. Remove sample holder from balance, place sample on teflon tape and slide sample into holder. Attach holder to balance, bolt reactor together, and record the initial weight in notes. 6. Open the nitrogen cylinder nearest the hOOd using v6 and verify the outlet pressure of 10-15 psig. Also make sure that V? is open all the way. Adjust balance purge nitrogen to full scale on rotameter using v13. Tape up opening through which hang-down wire passes to reduce Open area. 7. If necessary, attach V2 to the HF tank. Before tightening the packing, purge out area between valves by Opening V2 and then v5 to give a low flow Of N2 for a couple of minutes and then close both valves. Purging is performed prior to using HF to minimize corrosion. 8. If necessary, fill the water bath. 58 9. Turn on HF mass flow controller ("dynablender") power. Move-v4 to ”to trap" and put v3 to "full Open". Adjust nitrogen flow to 3 slpm on dynablender using VB. Purge for several minutes and then Open v10 and switch v4 to "to reactor" and purge some more. Finally, switch v4 back tO "to trap", close v10 and Open v2. 10. Put v4 midway between ports, close v3, wait about 30 s and close v5. Wait about 5 min and then Open v5, Observing for appreciable nitrogen flow through the purge rotameter which could indicate a leak in the HF pressure system. If everything is OK, move v4 to "to trap”, close v5 and then open v3 momentarily to bleed pressure from system. Close v2 and put v4 midway between ports. 11. Make sure v10 is Open and then adjust N flow to reactor to .5 slpm on mass flowmeter using 38. Verify flow in exhaust rotameter by momentarily switching v11 to "to flowmeter" and then back to direct exhaust. 12. Wrap insulation around reactor. Set the reactor temperature controller to 215 F and set the power supply to about 50 volts. After the setpoint is reached, leave for 30 min. and then adjust setpoint to temperature needed for experiment and adjust the power supply voétage accordingly. As a ggide, 10 volts is suitable for 30 C and 40 volts for 80 C. Remove the insulation and pull down hood sash to speed cooling. 13. Turn on HF line trace power sgpply and adjust so that flowmeter temperature is 24 to 25 C. 146 Turn on water bath temperature controller and adjust to 22 C. 15. After reactor is at setpoint temperature, turn Off flow to reactor using v8, put hood sash to specified level and record weight from DVM in notes. This is the dry sample weight. 16. Adjust containment N flow to reactor using v8. flow using v9 and N 2 2 17. Adjust preheater controller settings and preheater power supply. Wrap insulation around reactor. Large oscillations in preheater temperature will not affect reactor temperature so a large gain and high voltage may be used. 18. Turn on and adjust power supplies to containment chamber and HF control valve heaters to about 50 volts each. Verify that the heaters are working by touch. 59 19. Set up computer and run DAP3. 20. Make sure bath and flowmeter temperatures are OK and v10 is Open (VERY IMPORTANTII). Finally, Open v1 and v2. 21. Adjust dynablender setpoint to 0. Leave dynablender in ”Set” mode. Move v4 to “HF to Reactor", switch to control mode and then adjust HF flow setpoint. Finally, put dynablender to ”Read". 22. After steady-state is attained, record the DVM reading in notes. 23. Readjust containment N flow, N flow to reactor, dynablender setting, and pfeheater Control parameters as necessary and then wait for steady-state. Record DVM reading in notes. Keep repeating this step as necessary. 24. After experiment is complete, increase N flow to reactor and containment flow using v8 and v9. Next, put v4 to ”to trap", close v2, and Open v3. Finally, use v5 tO adjust purge flow to several slpm on dynablender. 25. After a few minutes, close v5, close v1, Open v2, and then Open v5 and adjust purge N tO several slpm. Note that purge flows will affect apBarent sample weight. 26. Turn Off bath heater and turn Off all power supplies except tO reactor. 27. Purge with v4 in both positions. Also purge with v2 closed. Finally, close v5, v3, v4, v8, v9, and v10. If another experiment is to be conducted, go to the beginning of this procedure. 28. Turn Off power at pilot box and close v6, v12 and v13. Remove hang-down wires from balance. HF Neutralization HF was neutralized by passing it over a bed Of landscaping-grade marble chunks (CaCO3) placed in a vessel made from PVC. HF reacts with CaCO to form Can, CO2 and 3 H20. Unfortunately, a significant amount Of HF was absorbed by the H20, thus forming hydrofluoric acid which remained in the bed. Thus, precautions must be taken to 60 prevent exposure to this hydrofluoric acid or its vapors. In particular, v4 and v10 should always be closed except when being used. Also, the hydrofluoric acid must be neutralized before disposing Of the material. Extreme caution should be used in neutralizing the hydrofluoric acid. APPENDIX B EQUATIONS USED IN THERMODYNAMICS DISCUSSION Derivation of Clausius-Clapeyron Equation for the HF-Wood System The situation to be considered consists of HF vapor in equilibrium with HF which is absorbed by lignocellulose. In the vapor phase, the nonidealities of HF will be 21 has shown that the considered. An earlier study properties of HF vapor at a given temperature and partial pressure are not affected by the presence of air. Thus, the equation can be derived by assuming that HF is the only gas-phase species and used for the HF-N2 system studied here. It will be assumed that the solid phase is actually one phase, although wood is heterogeneous. At equilibrium: de = dGS eq. 1 where: Gv = Gibb's free energy of HF in the vapor phase: GS = Partial Gibb's free energy Of HF in the solid phase. Substituting standard thermodynamic relations for equation 1 gives: yvdp - svd'r = VSdP - Ssd'r eq. 2 61 62 The right side Of equation 2 is valid for a constant solid-phase composition, which will be assumed for the derivation. Equation 2 can be simplified to: (5v - s5) dP ET = fizrfrjii eq. 3 where: Yv a Specific volume Of HF vapor: §v = Specific entropy of HF vapor; Vs = Partial specific volume Of absorbed HF: SS = Partial specific entropy Of absorbed HF: P = HF pressure; T = System temperature. At equilibrium, movement Of HF between the solid and vapor phase is reversible, thus: sv - SS = AHv/T eq. 4 where: AHv = Heat of vaporization Of HF from lignocellulose. VS should be Of the same order of magnitude as the specific volume of pure HF liquid so that the usual assumption that yv >> Vs can be made, thus: Xv - Vs = yv (approximately) eq. 5 Ev can be expressed using the ideal gas law: yv = RT/fP eq. 6 where: 71 II Ideal gas constant: H\ II Association factor for HF vapor. 63 Combining equations 3, 4, 5 and 6 gives: _d_P_ = AvaP dT RT2 which can be integrated to give the Clausius-Clapeyron equation: where: c = Integration constant. Equation for the Association Factor Smithl9 derived an equation for the association factor of HF vapor as a function of temperature and pressure by assuming a monomer-tetramer-hexamer equilibrium exists: _ 4 6 f-(P01+ 4K4P01 + 6K6P01 )lp eq. 7 and P’P-BPZ eq8 0 0 ' p = p + K P 4+ K P 6 eq 9 0 01 4 01 6 01 ' where: f = Association factor; P01 2 Implicit parameter (mm Hg): p0 = Implicit parameter (mm Hg): p = HF pressure (mm Hg): x4 = Tetramer equilibrium constant (mm Hg-3)i K6 = Hexamer equilibrium constant (mm Hg-5)i B = Nonideality parameter (mm Hg-l). Equations for K K6 and B vs. temperature were 4! 64 determined from tabular data in reference 19. These are: ln K 10557.90/T - 57.5864 eq. 10 4 1n K6 20639.95/T - 101.9869 eq. 11 ln B 4001.106/T - 22.1403 eq. 12 where: T = Temperature (K). These equations fit the tabular data very well, which is expected if the values are true equilibrium constants. Equations 8 and 9 were solved by successive substitution to determine P01 and then f was determined from equation equation 7. APPENDIX C ISOTHERM DATA The isotherm data used for Figures 3 through 7 are given below and on the following pages: Experiment Number : e4—29-1 Temperature ( C) : 30 Atmospheric Pressure (atm) : .968 Dry Sample Weight (mg) : 20.6 Type of Experiment : absorption PHF (atm) HF Loading (%) 0.0 0.0 0.07 10.2 0.10 12.1 0.15 19.9 0.19 30.6 0.24 33.0 0.28 41.3 0.38 50.5 0.47 59.2 0.57 71.4 0.67 84.5 0.77 97.6 0.87 120.0 0.97 149.0 65 66 Experiment Number : e4-19-1 Temperature ( C) : 40 Atmospheric Pressure (atm) : .996 Dry Sample Weight (mg) : 43.0 Type Of Experiment : absorption PHF (atm) HF Loading (%) 0.0 0.0 0.07 8.4 0.10 10.0 0.15 12.1 0.20 13.9 0.25 16.3 0.29 26.3 0.39 34.2 0.49 40.5 0.59 48.4 0.69 56.5 0.80 65.1 0.90 76.5 1.00 89.0 Experiment Number : e4-19-2 Temperature ( C) : 40 Atmospheric Pressure (atm) : .996 Dry Sample Weight (mg) : 43.0 Type of Experiment : desorption PHF (atm) HF Loading (%) 0.0 -2.3 0.10 13.5 0.25 23.5 0.38 34.9 0.59 49.0 0.80 66.3 0.90 78.8 1.00 89.0 67 Experiment Number : e4-7-l Temperature ( C) : 50 Atmospheric Pressure (atm) : .992 Dry Sample Weight (mg) : 22.8 Type Of Experiment : absorption PHF (atm) HF Loading (%) 0.0 0.0 0.07 7.5 0.10 7.5 0.15 10.1 0.20 11.1 0.25 12.3 0.29 14.5 0.39 27.2 0.49 32.5 0.59 36.8 0.69 41.7 0.79 47.8 0.89 52.6 0.99 59.2 Experiment Number : e5-22-1 Temperature ( C)‘ : 60 Atmospheric Pressure (atm) : 1.009 Dry Sample Weight (mg) : 40.5 Type of Experiment : absorption PHF (atm) HF Loading (%) 0.0 0.0 0.07 4.7 0.10 5.7 0.15 7.2 0.20 7.9 0.25 8.9 0.29 9.4 0.39 11.6 0.49 14.3 0.60 18.5 0.70 28.4 0.81 31.9 0.91 35.3 1.01 39.3 68 Experiment Number : e5-22-2 Temperature ( C) : 60 Atmospheric Pressure (atm) : 1.009 Dry Sample Weight (mg) : 40.5 Type of Experiment : desorption PHF (atm) HF Loading (%) 0.0 -2.5 0.10 7.7 0.25 11.9 0.60 23.7 0.91 35.1 1.01 39.3 Experiment Number : e5-27-1 Temperature ( C) : 70 Atmospheric Pressure (atm) : 1.012 Dry Sample Weight (mg) : 40.4 Type of Experiment : absorption PHF (atm) 0.0 0.07 0.10 0.15 0.20 0.25 0.29 0.39 0.50 0.60 0.70 0.81 0.91 1.01 27.0 HF Loading (%) OOQQO‘U’IUIIth Hk‘ N UO‘O‘O‘N‘ONQWOOWO NH #0 o Experiment Number Temperature ( C) Atmospheric Pressure (atm) 69 Dry Sample Weight (mg) Type Of Experiment P Experiment Number Temperature ( C) Atmospheric Pressure (atm) HF 0.0 0.07 0.10 0.15 0.20 0.25 0.29 0.39 0.49 0.59 0.69 0.80 0.90 1.00 (atm) Dry Sample Weight (mg) Type Of Experiment P H F (atm) 0.0 0.07 0.15 0.25 0.39 0.59 0.80 1.00 HF Loading (%) HF Loading (%) e5-30-2 80 .999 32.4 absorption HFJ e5-30-2 80 .999 32.4 desorption NwmmbWI-‘N H NHmmmflO‘fih-hubNNo oooooooooooooo \IHOO‘O‘QCD‘OWCOU‘UIO \lmmm‘OI-‘NN 10. 11. 12. 13. 14. REFERENCES Cheremisinoff, N. P. Wood for Energy Production: Ann Arbor Science: Ann Arbor, 1980; Chapters 1 and 8. Wegner, T. H.: et al. In Kirk-Othmer Encyclopedia 9f Chemical Technology, 3rd ed.: Grayson, M., Ed.; Wiley: New York, 1984; Vol. 24, p 580. Harris, J. F. Appl. Polym. Symp. 1975, 28, 131. Wright, J. D.: Power, A. J., paper presented at the Energy from Biomass and Wastes X Conference, April 1986, Washington, D.C. Helferich, B.: Bottger, S. Liebig's Ann. 1929, 476, 150. Fredenhagen, K.: Cadenbach, G. Angew. Chemie 1933, 46, 113. Pfleider, G.: Koch, E. Deutsches Reich Patentschrift 585318, 1933. Luers, H. Holz Roh und Werkstoff 1938, 1, 342. Rogovin, Z. A.: Pogosov, Y. L. Khim. i. Khim. Tekhnol. 1959, NO. , 368. Selke, 8.: et al. Ind. Eng. Chem. Prod. Res. Dev. 1982, 21, 11. Hardt, H.: Lamport, D. T. A. Biotech. Bioeng. 1982, Vol. XXIV, 903. DeFaye, J.: et al. Appl. Polym. Symp. 1983, 31, 653. Rorrer, G. L. M.S. Thesis, Michigan State University, E. Lansing, Mich., 1985. Rorrer, G. L.: et al., paper presented at the Summer National Meeting of the American Institute Of Chemical Engineers, Minneapolis, Minn., Aug. 1987. 70 15. 16. 17. 18. 19. 20. 21. 71 Franz, R.: et al. In Energy from Biomass, Second E. C. Conference: Strub, A., Ed.: Applied Science: London, 1982: p 873. Ostrovski, C. M.: et al., paper presented at the IV International Symposium on Alcohol Fuels Technology, Ottawa, Canada, May 1984. Reffstrup, T.: Kau, M. In New Approaches to Research in Cereal Carbohydrates: Hill, R., Munck, L., Eds.: Elsevier Science: Amsterdam, 1985: p 313. Costa, E.C.: Smith, J.M. AIChE J. 1971, 1_7, 947. —_' Smith, D. F. J. Chem. Phys. 1958, 28, 1040. Vanderzee, C. E.: Rodenburg, W. W. J. Chem. Thermodynamics 1970, g, 461. Simons, J.: Hildebrand, J. H. J. Am. Chem. Soc. 1924, 46, 2183. IIIIIIIIIIHIIIIIIIHIIIIIIII 3 1 6