WW l INN?" ES: ilHHlH‘HWH}NHWIHH THS ' lllll II llmllmlllllllnll‘lll ' _ 9 1293 10714 8847 LI 0 In A firm. 5.. Quinlan—‘5'. A”), c&m‘ - O .4 -" '-‘k reg-AIDE? . 11:15 ‘ . In. 0 sun-.- This is to certify that the thesis entitled REACTION RATES FOR GAS-PHASE HYDROGEN FLUORIDE SACCHARIFICATION OF NOOD presented by Gregory Lawrence Rorrer has been accepted towards fulfillment of the requirements for M. S. ' ° ' degree in Chem1cal Engmeem ng Major professor Date /0,/7:/YS‘ 0-7639 MS U is an Waive Action/Equal Opportunity Institution MSU LIBRARIES RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. REACTION RATES FOR GAS-PHASE HYDROGEN FLUORIDE SACCHARIFICATION OF WOOD BY Gregory Lawrence Rorrer A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemical Engineering 1985 ABSTRACT REACTION RATES FOR GAS-PHASE HYDROGEN FLUORIDE SACCHARIFICATION OF wooo By Gregory Lawrence Rorrer The intrinsic reaction rates for gas-phase BF saccharification of cellulosic materials in Bigtooth Aspen were determined. Experimental techniques were developed to obtain sugar yield versus reaction time data from wood chips reacted with an HF/nitrogen gas-stream of fixed HF partial pressure and temperature. Mass-transfer effects were experimentally screened out by determining the regime where the reaction rate was independent of both flowrate and wood chip size. The intrinsic glucose yield versus time data proceded by a sigmoidal profile. For pure HF flow at ambient reaction conditions (30’C, 1.0 atm). the reaction achieved maximum glucose yields of 2.7 to 2.9 mnol/g-wood within 2 minutes. The glucose production rate decreased nonlinearly with decreasing HF partial pressure from 1.0 to 0.2 atn at 30‘C. Surprisingly, the glucose production rate decreased with increasing reaction temperature from 28 to 108'C at an HP partial pressure of 1.0 atm, and this effect was not due to product degradation. Processes such as HF adsorption onto the lignocellulosic matrix may be coupled to cellulose solvolysis and may strongly influence the overall glucose production rate. ACKNOWLEDGEMENTS I gratefully appreciate the financial support provided by the Department Of Chemical Engineering, the Division of Engineering Research. the National Science Foundation, and the 3M Company. I would also like to personally thank Dr. Martin C. Hawley and Dr. Derek T. A. Lamport for their assistance. guidance, and encouragement. ii TABLE OF CONTENTS Page LIST OF TABLES ......................................... iv LIST OF FIGURES ......................................... v INTRODUCTION ............................................ 1 EXPERIMENTAL ............................................ 6 Reaction Apparatus .................................... 6 Procedure ............................................. 8 Sugar Analysis ....................................... 10 RESULTS AND INTREPRETATION ............................. 14 Elimination of Mass-Transfer Resistances ............. 14 Glucose and Xylose Yield vs. Time Profile ............ 17 Dependence of Reaction Rate on HF Partial Pressure...22 Dependence of Reaction Rate on Temperature ........... 31 DISCUSSION AND CONCLUSIONS ............................. 38 Intrinsic Reaction Sequence .......................... 39 Future Research ...................................... 43 LIST OF REFERENCES ..................................... 45 APPENDIX ............................................... 46 Sugar Yield and Reaction Rate Data ................... 46 Sugar Chromatography ................................. 55 iii LIST OF TABLES Page Summary of Reaction Conditions ...................... 11 Assumptions ......................................... 12 The Intrinsic Rate Processes Of Gas-Phase HF Saccharification of Cellulose in Wood ............... 40 Sugar Yield and Reaction Rate Data .................. 46 iv 10. 11. 12. 13. LIST OF FIGURES Page Reaction Apparatus ................................... 7 Glucose Yield versus HF Flowrate .................... 16 Glucose and Xylose Yield versus Time Profiles at 30°C ............................................. 18 Glucose Yield as a Function of Time for HF Partial Pressures from 0.6 atm to 1.0 atm at 30°C ........... 23 Glucose Yield as a Function of Time for HF Partial Pressures from 0.2 atm to 0.6 atm at 30°C ........... 24 Glucose Production Rate versus Glucose Yield as a Function of HF Partial Pressure from 0.6 atm to 1.0 atm at 30'C ..................................... 25 Glucose Production Rate versus Glucose Yield as a Function of HF Partial Pressure from 0.2 atm to 0.6 atm at 30'C .................................... 26 Glucose Production Rate as a Function of Time for HF Partial Pressures from 0.6 atm at 1.0 atm at 30'C ............................................. 27 Glucose Production Rate as a Function of Time for HF Partial Pressures from 0.2 atm to 0.6 atm at 30°C ............................................. 28 Xylose Yield as a Function of Time for HF Partial Pressures from 0.2 atm to 0.6 atm at 30°C ........... 30 Glucose Yield as a Function of Time for Temperatures from 28'C to 108'C at an HP Partial Pressure of 1.0 atm ................................. 32 Glucose Production Rate versus Glucose Yield as a Function of Time for Temperatures from 28 C to 108 C at an HF Partial Pressure of 1.0 atm ....... 33 Glucose Production Rate as a Function of Time for Temperatures from 28'C to 108'C at an HF Partial Pressure of 1.0 atm ................................. 34 14. Xylose Yield as a Function of Time from 28'C to 108'C at an HF Partial Pressure of 1.0 atm .......... 37 vi INTRODUCTION The conversion of lignocellulosic materials to simple sugars by acid hydrolysis is a well-established biomass saccharification technology. The acid hydrolysis of cellulosic materials in wood produces sugars that can be readily fermented to ethanol or other useful chemicals. However. the saccharification of wood by conventional sulfuric and hydrochloric acid hydrolysis processes suffers from a variety of drawbacks. including low sugar yields. high acid consumption. long reaction times. and high temperatures and pressures required to drive the cellulose hydrolysis (1). The use of anhydrous hydrogen fluoride (HF) to "crack" cellulosic materials in wood overcomes the aforementioned process inefficacies plaguing other conventional acid hydrolysis technologies. The "HF saccharification" of lignocellulosic materials Offers 1) high sugar yields, 2) no product degradation. 3) short reaction times, 3) ambient reaction conditions, 4) low acid consumption due to efficent recovery and recycle of HF. and 5) no elaborate substrate pretreatment except for drying. Actually. the process of "HF saccharification" is not a direct hydrolysis of the cellulose chain. The solvolysis of cellulose (a beta-1.4 linked polyglucan) proceeds by a glycosyl fluoride 1 intermediate, which is readily hydrolyzed in even a stoichiometric amount of water to yield glucose and regenerate HF (2). The solvolysis of xylan in wood procedes similarly. The first serious studies of the "HF saccharification" of wood were initiated by several German investigators in the early 1930's. Apparently. gas—phase HF saccharification of wood reached the pilot-plant stage just prior to World War 11. Most notably, Each and Bohunek (1937) developed a process to react wood chips with gaseous HF at a reduced pressure of 30 mm Hg and temperature of 35 to 40 'C (3). After about 30 minutes, the reaction was complete. The reaction was followed by excess HF desorption at 62 “C. Luers (1938) summarzied the results of a I.G. Farben pilot plant study which utilized the Hoch and Bohunek process (4). Gas-phase HF saccharification of wood was studied as a function of reaction time to maximum yield, temperature, HF concentration. and HF amount. Batch reactions that were carried out with 40 kg of pure HF per 100 kg of wood at 2 to 3X moisture gave the best results: 50s and 57% of reducing sugars were Obtained from pine and beechwood respectively. Optimal reaction times and temperatures were not specified. and many of the original results were lost. The advent of World War II drew attention away from "HF saccharification", and after World War II. the technology was essentially "forgotten". Research in "HF saccharification" was renewed following several "energy 3 crises" in the mid 1970's by several investigators interested in novel biomass conversion technologies. Early, it was evident that the advantages of "HF saccharification" lay in the unique chemical reaction engineering and chemical kinetic aspects of the process. In this light, Selke _5 _3. studied the reaction kinetics of liquid-phase HF saccharification of wood (Bigtooth Aspen) and pure cellulose in a well-stirred batch-reactor (5,6). Total. water-soluble glucose and xylose yield data were obtained as a function of reaction time (0 to 60 min), temperature (-12 to 4'C), and water content in HF (0 to 6.4% by weight). For example, at O'C, anhydrous, liquid HF readily "cracked" cellulosic materials to produce glucose and xylose yields in excess of 90% of maximum theoretical, typically within 20 to 30 minutes. The reaction rate was independent of particle size for wood chips of 0.05 to 0.2 cm thickness. The reaction kinetics were reasonably described by a homogeneous, psuedO-first order rate model (6). Also. preliminary process-design, economic, and technOlogy assessment studies by Hawley gt 3}. indicated that although HF saccharification lacks sufficient process technology development. it may possess basic advantages over the more- developed acid-hyrolysis technologies (1,7). Recently (1982), pilot-plant studies for the gas—phase HF saccharification of poplar were intiated by Ostrovski g; g}. at Canertech of Canada in the attempt to commercialize this process (8). The process used a semi-continuous test reactor of 20 kg wood (per charge) capacity that featured an HF adsorption/reaction stage, followed by an HF—desorption stage at elevated temperature. Reaction products underwent dilute-sulfuric acid post-hydrolysis at 140‘C to yield free sugars and re-convert reversion Oligosaccharides to monomeric sugars (glucose and xylose). Pentosans and hexosans were then fermented to yield a supposed 438 liters of ethanol per tonne of dry wood. Results of their initial pilot-plant studies demonstrate the feasiblility of gas- phase HF saccharification of wood. They report that maximum glucose yields of 90% and xylose yields of 70% were typically obtained. with 99% of total HF recycled. The optimum conditions for HF saccharification of poplar occurred under a 100% HF flow at ambient temperature and pressure (9). Apparently. gas—phase HF saccharification of wood possesses several advantages over the liquid-phase process. Gas-phase HF saccharification of wood offers rapid reaction rates and greater flexibility of reactor operating conditions. Gas—phase HF solvolysis of cellulosic materials in wood can be carried out at higher temperatures (the liquid phase is limited to 19.8'C, the normal boiling point of HF), and pressures. The anhydrous HF concentration can be easily diluted in inert gas. The recovery of HF is simpler and more efficient because excess HF does not have to evaporated off the solid. The HF to wood requirements are also lower, owing to the reduced density of HF in the gas-phase. 5 Gas—phase HF saccharification of wood can be more quantitatively assessed by developing reactor models that predict the rate of cellulose cracking and sugar product yield as a function of reactor operating conditions. However, such models require an intrinsic chemical kinetic model which predicts the rate of cellulose cracking and sugar yield as a function of reaction time, reaction temperature and gas—phase HF concentration (or HF partial pressure). Unfortunately, no chemical kinetic studies of gas-phase HF saccharification of wood have been reported in the literature. Therefore, a fundamental chemical reaction engineering study is needed to elucidate and model the heterogeneous rate processes which characterize the solvolysis of cellulosic materials in wood by gaseous. anhydrous HF. The purpose of this investigation is to generate intrinsic reaction—rate data for the gas-phase HF saccharification of wood so that the intrinsic rate processes and chemical kinetics that characterize this process can be elucidated. EXPERIMENTAL Reaction rates for the solvolysis of cellulosic materials in wood by gaseous HF were determined from total. water-soluble glucose and xylose yield versus reaction time data acquired at a fixed set of reaction conditions. Rate data were obtained from Bigtooth Aspen chips reacted with an HF/nitrogen gas-stream of fixed HF partial pressure and temperature. Reaction Apparatus The batch-mode reaction between a single wood chip and an anhydrous HF/nitrogen gas—stream is contained in the reaction apparatus shown in Figure 1. The reaction apparatus is designed to 1) set and monitor the gas-stream reaction conditions. 2) carry out the reaction and still allow .easy access Of the wood-chip sample, and 3) neutralize the effluent HF gas-stream. All HF-wetted parts are constructed of monel, teflon, or similar HF-resistent materials. The reaction apparatus is contained in a fume hood. HF is supplied in liquid form from Hatheson Gas Products. A heating jacket surrounding the HF tank keeps the surface Of the HF tank at 50 C (safety limit) and thus 6 ‘V////// $3282 cozumom 4 $50: wAnSOOOSEth 55.230: a: 50.. $33 hwxos. 02....(m: Dz< ¥z<._. “51:050..— .sAIfiMMWWHVIIi wmzk Oz.>mo uda2. a = xylose «'3 1,0. . ' U) m m = a) 0.5- ‘ O i J j 1 I O 50 100 150 200 250 Reaction time,seconds FIGURE 3. Glucose and xylose yield versus time profiles at 30°C. 19 for Bigtooth Aspen is 3.441 millimole per gram of wood (6). This value is approximate and based on previous material balance studies for liquid-phase HF saccharification of wood. The contribution of glucose from hemicellulose solvolysis was assumed negligible. No reaction product degradation was observed to occur for glucose yield-plateau data. If such degradation were to occur, then the glucose yield would pass through a maximum in time, as is the case with dilute-acid cellulose hydrolysis at elevated temperatures (12). Reaction products resided on the lignocellulosic matrix, and so there were no product losses to the gas-stream. Occasionally, glucose yields in excess Of 90% were obtained, consistent with results of other investigators at Canertech. Based on all of the information provided above, this slight discrepancy in glucose yield cannot be explained. The scatter in glucose yield probably arises from inconsistencies in initial wood chip sample composition, errors in reaction time during the quenching process. (particularly at short reaction times), and errors in chromatographic sugar analysis. Normally, the acid—catalyzed hydrolysis of cellulose is a random, stepwize depolymerization characterized by a monotone-decreasing time-dependent evolution of glucose. The solvolysis of cellulose in wood by gaseous HF is still considered a random, stepwize depolymerization. However. the sigmoidal profile Of the intrinsic glucose yield versus time data suggests that other heterogeneous rate processes are coupled to the cellulose solvolysis. Candidates for 20 these processes will be described later. Assume for now that the intrinsic reaction rate is determined by two sequential rate processes described by two independent, first-order rate equations governed by the intial condition of t=0, Y=O. The general solution of this system of rate equations can be used to fit glucose yield versus time data. It is given by ~ct -ft Y = a + be — de (1) where a,b,c.d and f are adjustable parameters, t is the reaction time, and Y is the normalized glucose yield. The glucose production rate is determined from the time- derivative of equation (1). given by dY —ft -ct -- = dfe - bce (2) dt Glucose yield versus reaction time data were curve— fitted to equation (1) by least-squares multiple nonlinear regression (13). For example, note from Figure 3 that the curve-fitting model described by equation (1) adequately predicted the sigmoidal profile and asymptotic leveling of the glucose yield versus time data. Because a complete rate model for gas-phase HF solvolysis of cellulose in wood has not yet been proposed, kinetic parameter estimation (i.e., rate constants, reaction orders. etc.) was not attempted. 21 In general, curve-fitted lines were not extrapolated beyond the upper reaction time limit of the data. Gas-phase HF solvolysis of xylan in the hemicellulosic fraction of wood proceeded more rapidly than cellulose solvolysis at the same reaction conditions. For comparison, glucose and xylose yield versus time data are plotted together in Figure 3. Maximum experimentally attainable xylose yields of 0.9 to 1.2 millimole per gram of dry wood (65% to 85% Of maximum theoretical) were achieved after approximately 1 minute. The maximum theoretical xylose yield in Bigtooth Aspen is 1.4 millimole per gram of dry wood, based on previous material balance studies for liquid- phase HF saccharification (6). The xylose product may have degraded to 2—furfuraldehyde during dilute-acid post- hydrolysis (10). However. the analytical procedures were not set up to detect for 2-furfura1dehyde. Inconsistency of xylose product degradation during post-hydrolysis. and possibly the non-uniformity of the initial xylan present in the wood chip sample created the scatter in the xylose yield. Additionally, the reaction time for high cellulose conversion was more than sufficient for high xylan conversion. and so kinetic data for xylan solvolysis was incomplete at low xylose yields. Therefore, curVe-fitting of xylose yield versus reaction time data was not attempted. In any event, rate data acquisition focused on glucose yield because it is the more abundant and desired end—product. 22 Dependence of Reaction Rate on HF Partial Pressure Glucose yield versus time data at 30°C for HF partial pressures from 0 to 1.0 atm (under 1 atm total pressure) are presented in Figures 4 and 5. Glucose production rate versus normalized glusose yield profiles at 30’C for each HF partial pressure are presented in Figures 6 and 7. Finally. glucose production rate versus reaction time profiles at 30 C for each HF partial pressure are presented in Figures 8 and 9. In general, the intrinsic glucose yield versus time data possessed a sigmoidal profile at all HF partial pressures from 0 to 1.0 atm. The experimental maximum glucose yield of 2.7 to 2.9 millimole per gram of wood was independent of HF partial pressure. Even though long-time , yield—plateau data were not obtained for low HF partial pressures. the glucose yield did appear to converge on the experimental maximum, and not some lowered equilbrium value. as would be expected for a reversible reaction. The glucose production rate at all reaction times uniformly and nonlinearly decreased with decreasing HF partial pressure. The glucose production rate consistently passed through a maximum between 10% and 20% of normalized glucose yield. Existence of a reaction rate maximum suggests that other intrinsic rate processes may be coupled to the cellulose solvolysis reaction. Does the use of nitrogen as a diluent for HF 23 doom am can a; 8 ed :5: 3.5395 BEE m: .8 me: .6 SEE: m an 22> 38:5 .3 $50: «vacuum 65: cozomom com oov com com cop . \n . EE omd 0 ES mud I . ES 9.? a . 65303 .2th “.1 . 0. on L o... l o.N LVN m3 pOOM-B/elowgmm ‘pmfl esoonls 24 $2 p00M-6/alommiw 'p|eul esoonlg . doom E can 3. 2 N a so: 8.55.8.5 3:8 “.2 .2 we: .6 8:25.. m mm 22» 38:5 .m “:50. m mun-000m .05: cozowwm coop 00m com 005 com com cow com CON 00.. O 1 l d a q r q i 4 b l H \ \ d L O.“ . .. md ES «6 0 ES «to m . EB and e L o a 0.0m “05303 32.2. “.1 Glucose production rate,mIlIImoleIg-wood-sec 0.06 0.05 ’ 0.04 0.03 0.02 t 0.01 1.0 atm I r 0.73 atm 0.60 atm J i 1 J 30°C 1 J i J 4 0 10 20 30 40 l 50 60 70 80 90 100 Glucose yield, percent FIGURE 6. Glucose production rate versus glucose yield as a function of HF partial pressure from 0.60 to 1.0 atm at 30°C. 26 doom Em 5m 3. 3 N: ES. 8:32.. 3:8 “E .o 8:55: m mm 22» 38:3 89...; 28 8:88: 3820 .N $50: 2505.. .203 03020 0m: 00 00 05 00 00 0? 00 0m 0w 0 - q d u q q q q - P=3 «.0 chm 00.0 1 000.0 0_.0.0 0.8 A 0—00 oas—pOOM—Blemwgmm ‘GIBJ uouonpOJd esoonua 27 0.06 - 30'C 0.05 '- 0.04 - 0°03 '- 1.0 atm Glucose production rate, mlllimole/g-wood-sec 0'02 " 0.73 atm 0.01 - 0.60 atm J 1 __l 4 0 100 200 300 400 500 Reaction time,seconds FIGURE 8. Glucose production rate as a function as time for HF partial pressures from 0.60 to 1.0 atm at 30°C. 28 doom 6 5m cod 2 «5 so... 3.5 32a 3:8 ..=._ .8 we: .5 5:55: a mm 38 c2528: 38:5 .o maze: 2:303 .05: :ozomom 000— 000 000 00m. 000 000 00v 000 00a cow 0 - 4 q q q q q q d . //I I\J. e m. .53 «d .. m s 9 . m .. O I p n Eufl VV.O 4 ”00.0 mlw. . w m as and . m . m. l w. L Sod m. I . 3.. M i O O D. .53 86 . 3.. 0.8 m M... 39¢ 29 concentration specifically effect the reaction rate? HF normally exists as a hydrogen—bonded hexomer in the pure vapor state at ambient conditions (14). The heat Of mixing of HF in nitrogen was observed to be endothermic, with the corresponding drop in temperature of the mixed gas-stream proportional to the decrease in HF to nitrogen ratio. This phenomenon is attributed to disruption of hydrogen-bonded HP in the gas-phase. The feed-stream heater on the reaction apparatus compensated for the temperature drop and isothermally maintained the gas—stream at 30 °C. If the hexomeric form of HF is preferentially adsorbed on the surface of the lignocellulosic matrix (by virtue of its larger size). then the reaction rate would have a more pronounced dependence on an HF concentration that is set by dilution of pure HF in nitrogen as opposed to an HF concentration that is set by varying the total system pressure. However, the specific effect of "HF- depolymerization" could not be quantified experimentally with the apparatus and methods used. Xylose yield versus time data for three representative HF partial pressures are presented in Figure 10. In general, xylose yield versus time data is broken up into two regimes. At short reaction times. a fraction of the hemicellulose is rapidly converted to xylose, followed by a much slower conversion of the remaining hemicellulose. Note that when the HF partial pressure was decreased, the initial fraction of hemicellulose converted to xylose also decreased. In general. decreasing the HF partial pressure 30 doom .m 5% co... 2 N... :5... 8.38.5 5:8 “.1 .5. me: .o 252:: m mm 22.. .3on .2 $52. 8:003 .05: cozomom 000.. 000 000 00s. 000 000 00* 000 com 00.. 0 u a 1 q u q q q d d EH «.0 O - Eum vvd 1 Sum 00.0 I 1 50:5on .2th n=.. o. 3 0.— m... poem-G/aloumuui “pleui esoMx '41 / decreased the overall rate Of xylose formation. Dependence of Reaction Rate on Temperature Isothermal glucose yield versus reaction time data Obtained at an HF partial pressure of 1.0 atm for reaction temperatures from 28 to 108 C are presented in Figure 11. Isothermal glucose production rate versus normalized glucose yield profiles at each reaction temperature are presented in Figure 12. Isothermal glucose production rate versus reaction profiles at each reaction temperature are presented in Figure 13. Isothermal conditions for the complete reaction system (HF vapor and wood chip) were insured by pre-heating the wood chip to the desired reaction temperature in an oven prior to reaction. Sub-ambient, pure-HF gas—stream temperatures could not be attained at an HP partial pressure of 1.0 atm, because the normal boiling point of HF is 19.8'0. The HF gas-stream had to be at least 26 'C to provide a driving force for HF vaporization and flow. The glucose yield and production rate decreased with increasing reaction temperature from 28 to 108 C. The glucose yield versus time profile retained its characteristic sigmoidal profile. The reaction rate from 28 to 50 C decreased slowly away from the maximum at 28'C. In this temperature range. the maximum glucose yield remained at 2.7 to 2.9 millimole per gram of dry wood. and so no .5... 3 .o 9539:. 3:23 m: cm E memo. o. comm 52. 33.2853 .5. me: .o 5:82 m mm 22.. 38:6 .: maze: 20.5000 .05.. cozoaom 005 000 000 00¢ 000 00a 1 1 N (I I 1 .\\IIIII 0. 00p 0 0. mu 0 0.00 I 0.0m Q ”053.0050. cozommm Sum 0.9 u “in. 0 0.0 pOOM-Blelowmgw ‘plegA esooms .5: o.. .o 9539:. 3:8 .1 cm .5 008. 2 mama :5... 839.853 5 8:25. 5 3 29.. 88:5 3: 35> 99. 8258.5 38:6 .N. £52... 20200 .20.» 00003.0 00.. 00 00 05 00 00 0? 00 ON 0... 0 a u q d 1 q q 0.00.. .05 :30 0.5 N n— 3.. 50.0 «0.0 00.0 #00 00.0 00.0 oes—pOOM-Blelowgmw 'etel uogionpOId esoonlg g (105 U) I 'U 0 O f E“ 1104 2 O .5. E 0.03 2’ E .5 (102 5 3 'U 2 0. g (101 a O O 2 <3 P HF = 1.0 atm 23‘0 50°C 75, 108'C 1 —‘ J i 3" J 100 200 300 400 500 600 Reaction time, seconds FIGURE 13. Glucose production rate as a function of time for temperatures from 28°C to 108°C at an HF partial pressure of 1.0 atm. 35 reaction product degradation occurred. From 50 to 75‘C, the reaction rate dropped sharply, indicating a shift in rate control from cellulose solvolysis to some other process. At reaction temperatures greater than 50°C, the reaction rate was so slow that glucose yield—plateau data were not obtained within a experimentally-realistic reaction time. and so evidence for possible thermal reaction product degradation was not detected. The reaction rate from 75 to 108’0 changed little and maintained a consistently low rate. Apparently, an intrinsic process other than cellulose solvolysis may have assumed control of the reaction rate. Upon first inspection, the glucose production rate data seemed to contradict the Arrhenius equation. However. these rate data suggest that equilibrium-driven, heterogeneous rate processes may be coupled to the intrinsic cellulose-cracking reaction. Such processes may assume control Of the overall intrinsic reaction rate at elevated reaction temperatures. An example of an equilibrium—driven process is surface- adsorption Of HF into the lignocellulosic matrix. It was observed that during the solvolysis reaction between cellulosic materials in wood and gaseous HF. the structure of the lignocellulosic matrix was retained and the solvolysis products appeared to "stick" to the lignin binder. Therefore, the ligoncellulosic matrix was available to provide sites for surface adsorption by HF molecules. The rate of chemisorption for a heterogeneous reaction normally increases with increasing temperature. However. 36 the physical adsorption isotherm decreases with increasing temperature. Surface adsorption of HF onto the lignocellulosic matrix prior to cellulose solvolysis could alter the overall intrinsic reaction rate. particulary if adsorption constants determined by a negative heat of adsorption are stronger functions of temperature than the rate constant for cellulose solvolysis. This speculation suggests that the overall intrinsic reaction rate could pass through an optimum in temperature. However, there is no direct experimental evidence to support such a conclusion with the data at hand. Xylose yield versus reaction time data at 1.0 atm HF partial pressure for each reaction temperature are presented in Figure 14. In general, the xylose yield versus reaction time profile followed the same trend as previously described. In general, the time-dependent evolution of xylose decreased with increasing temperature. consistent with glucose yield versus time data at the same set of reaction conditions. However, for xylose yield data at 50 C, the xylose yield appeared to pass through a maximum in reaction time, indicating that some product degradation might have occurred during solvolysis. However, as mentioned earlier, the analytical procedures were not set up to detect for xylosyl degradation products. 37 .5... 5.. .o 9538.. 53.8 .1 c: .m memo. 0. comm Ea... 0:5 .o 5.62:. m mm 22.. 083x .3 $50.. 005000 .05.. 202000.. 000 00? 000 00m 005 0.00.8 0.05 O 0.00 I Pow d l ”03.0.0060. 5:30.. . n...- Eea 0.5" t 0... 0... pOOM-0/9|Ou.ll||!ul “maul esouix DISCUSSION AND CONCLUSIONS The rapid reaction rates at ambient conditions and high sugar yields obtained from gas-phase HF saccharification of wood highlight the advantages of this process over liquid— phase HF saccharification and acid hydrolysis processes in general. It is desirable to develop reactor models for gas— phase HF saccharification of wood based on a fundamental. quantitative. intrinsic kinetic model for the solvolysis of cellulose in wood by gaseous HF. These reactor models can then be applied to reactor design and process optimization studies. In this way, gas-phase HF saccharification of wood can be more quantitatively assessed and compared to other biomass conversion technologies. The first step toward development of a reactor model for gas-phase HF saccharification of wood is to generate intrinsic reaction rate data so that the intrinsic rate processes and chemical kinetic model can be elucidated. A detailed intrinsic kinetic model requires a physically representative reaction sequence and knowledge of rate- limiting steps. Further. the set of formulated rate equations must be solved in a form which can be fitted to glucose yield versus time data. In this way, chemical kinetic parameters (e.g.. rate constants) for each rate process can be estimated. Once an intrinsic kinetic model 38 39 has been developed. it can be modified to account for particle size and convective mass-transfer effects. The complex reaction rate behavior exhibited by gas- phase HF saccharification of wood indicates that formulation of a quantitative rate model is premature if based solely on the rate data presented here. However, the rate processes which may determine the global reaction rate can be postulated from these data. The Intrinsic Reaction Sequence Reaction rate data for gas-phase HF saccharification of wood indicate that even in the absence of mass-transfer effects. heterogeneous rate processes are important. The sigmoidal profile of the glucose yield versus time data and the decrease of overall glucose production rate with increasing temperature suggest that surface interactions between vapor-phase HF molecules and cellulosic chains residing within the lignocellulosic matrix control the "cracking" of cellulose to glycosyl fluoride monomers. Logical candidates for the intrinsic rate processes of gas— phase HF saccharification of wood are summarized in Table 3. The intrinsic reaction scenario of Table 3 will now be described. HF molecules in the pure vapor state exist as hyrogen- bonded hexomers at ambient temperature and pressure. Upon dilution in nitrogen. the hydrogen—bonded HF hexomer 4O . 6+«u: .chaxomuov coaumnwuoa%H0dov mmoasaaoo .m: voucoaucmwouvmn mo acquasuwwc muwanaoo .mmoaaaawo mo :ofium>aow cam Aouwm oumwuam I mv xauuma camoazaaooocwaa «so ouco cOwunuomvm m: .Assfimusassv as u x .oooosnooon . c .Hfiunawouuqs onu cfisufia mcwmnu omoazaaoo unocua vovconncmwouvmn mo cowuasuman .c We .N2 5 :owuaauc com: w: No cowumuwuoazaonon mmmooup .coo3 :H mmoazaamo mo cowumowMfiumxuumm mm mmmnnlmmw mo mommmooua mumu camcfiuuca one ON: + Amy muon: : .muauu. m Amv m- onum + Amv mousfiuvuz annulu G k: 3 .7: + E 3-1.87: TI 3 :1 $4. 3 IL 3 is: 3 5:5 + m Allllnlllu c m m A3 5-73.: 3 moucAoT: .m 1 Amy Ammv Amy mouconum.x .nuuatv Amy xmmouaflovumw .~ va Ammv H-s .lsunur Amv Rams + Amy a: Amv efimmv .H 1 N2 cowuummm .m anmH a .msmrougnnumom 23 no.9: 3+3 T: 2.3 =ouw+ GT: g .vm A 3+: 3.. u a ._. 2:. a SWEB 2.3 m: + :3 so- GT: AH: 23 mo- ST: + ..T GT: amuse: QT? .o v?& 1 .. mmgumzoumwomgo N c+a v a 4 wESomLBms 3 .8356”. o z + :3 mo- GT: All. 2.3 no- ST: + 2.3 .5. GT: .0 .. 3.3 .5 + 5-57:4 3 p-387: .o v m 3 38:5 a o a 55 H 3 .53 a: so cofiumuwcowmu can muoavoum ummu 532.8 do 29:21: 2.3 mm + :3 mail. AIINI 3 “I; .n A: o a mmwuopm Coauuwwm .AsoacsscouV .m ”Hams 42 "depolymerizes" in the gas-phase, producing an oligomeric distribution of HF molecules of 1 to 6 monomer units in size. These HF molecules adsorb onto surfaces within the lignocellulosic matrix. Hydrogen-bonded linear cellulose chains (beta-1.4 polyglucan) make up the cellulose microfibril. which constitutes the bulk of the secondary cell wall in the lignocellulosic matrix. The hydrogen— bonded cellulose microfibril is "unraveled" by interactions with adsorbed HF molecules. These interactions may also reduce adsorbed. hydrogen-bonded HF molecules to monomeric HF. The BF monomer fixes itself onto a glucan unit site on the cellulose chain and irreversibly "cracks" the beta-1,4 linkage at that site. The products of the solvolysis (cleavage) are two polyglucan (or single glucan) fragments. one of which is fluoridated at the 1-carbon on the terminal glucan unit of that fragment. Eventually. the cellulose molecule is completely cracked to alpha-D-glucopyranosyl fluoride monomers (10), which reside or "stick" on the lignocellulosic matrix. It is observed that the lignocellulosic matrix remains intact throughout the reaction. indicating that the solvolysis products are held in place by the lignin binder. Presumably. the lignin does not decompose or depolymerize in the presence of HF. Additionally. the reaction mixture was not agitated, and so no mechanical fragmentation of the sample occurred. Finally, the chemistry of hydrolysis. reversion, and post—hydrolysis summarized by steps 5 to 7 in Table 3 are 43 described elsewhere (10). The intrinsic rate processes described by steps 1 to 4 in Table 3 may explain the complex reaction rate behavior exhibited by gas-phase HF solvolysis of cellulosic materials in wood. The nonlinear dependence of reaction rate on HF partial pressure is probably determined in part by the disruption of the HF hexomer in the vapor phase upon dilution in nitrogen. Such a phenomena could result in an effective HF partial pressure over the wood chip which is lower than the calculated RF partial pressure. The decrease in overall glucose production rate with increasing reaction temperature is probably determined by physical adsorption of HP from the gas-phase onto the lignocellulosic matrix. The sigmoidal profile of the glucose yield versus time data may be determined by processes immediately preceeding cellulose solvolysis, such as HF-adsorption and hydrogen bond-breaking of cellulose microfibrils. As indicated earlier however. it is not certain which of these processes is rate—controlling. Future Research Future research of the chemical kinetics of gas—phase HF saccharification of wood will focus on unraveling the complex reaction rate behavior of this process and developing a quantitative rate model. The effect of HF concentration on the reaction rate will be determined by 44 varying the total system pressure in the attempt to screen out the compounding effect of "HF—depolymerization" (via dilution of RF in nitrogen) on reaction rate. The effect of reaction temperature on reaction rate will be determined at a lowered HF partial pressure so that sub-ambient (less than 20 C) gas-stream temperatures can be achieved. In this expanded temperature range. it is anticipated that the reaction rate would pass through an optimum in temperature. indicating a shift from a reaction rate-controlling to an HF adsorption-controlling regime. Insitu gravimetric measurement of the HF uptake on wood will be used to study the HF adsorption isotherm on wood as a function of reaction time, temperature. and HF partial pressure. Internal mass- transfer effects will be explored and modeled by observing the effect of wood chip particle size on reaction rate. It is hoped that the results from all of these chemical kinetic experiments can be brought together to develOp a fundamental rate model for gas-phase HF saccharification of wood. LIST OF REFERENCES 10. 11. 12. 13. 14. LIST OF REFERENCES Hawley. M. C.. Selke, S. M.. Lamport. D. T. A., Energy in Agriculture. g. 219 (1983). Fredenhagen. K., Cadenbach, 6.. Angewandte Chemie. 49. 113 (1933). Luers, H., Holz Roh Und Werkstoff. l, 35 (1937). Luers, H.. Holz Roh Und Werkstoff, ;. 342 (1938). Selke. S. M.. Doctoral Disseration, Michigan State University (1983). Selke. S. M.. Hawley. M. C., Lamport. D. T. A., Wood and Agricultural Residues: Research on Use for Feed. Fuels, and Chemicals. Academic Press, Inc., 329 (1983). Selke, S. M.. Hawley. M. C., Hardt. M.. Lamport. D. T. A., Smith. G., Smith, J..I & EC Product Research and Development. g3, 11 (1982). Ostrovski. C. M.. Duckworth, H. 5.. and Aitken, J. C.. IV International Symposium on Alcohol Fuels Technology. Ottawa, Canada. May 21-25, 1984. Ostrovski, C. M.. Personal Communication (1985). Hardt. M.. Lamport. D. T. A., Biotechnology and Bio- engineering, 23. 903 (1982b). Albersheim, P.. Nevins, D. J., English, P. D.. Karr. A., Carb. Res.. g, 340 (1967). Saeman, J. F., Ind. & Eng. Chem., g3, 43 (1945). Dye, J. L.. NIcely. V. A., J. Chem. Educ.. 12, 443 (1971). Simons, J. H., Fluorine Chemistry, Vol. 1. Academic Press. N. Y. (1950). 45 APPENDIX Sugar Yield and Reaction Rate Data Table 4. Sugar yield and reaction rate data. HF partial pressure: 1.0 atm Analysis method: GLC alditol-acetates, averaged over two assays REACTION TIME TEMPERATURE GLUCOSE YIELD XYLOSE YIELD (sec) (‘C) (mmol/g-wood) (mmol/g-wood) 10 29.7 0.050 0.820 15 30.3 0.161 0.950 20 30.0 0.307 0.954 30 30.3 0.513 0.921 40 30.1 1.012 1.337 60 30.0 2.239 1.306 80 30.3 2.690 1.398 100 30.3 3.055 - 120 30.4 2.504 1.058 140 30.4 2.405 1.173 160 30.6 2.702 1.385 180 30.4 2.512 1.212 200 30.4 2.998 0.914 220 30.4 2.642 1.260 46 47 Table 4. (cont'd.). HF partial pressure: 0.73 atm Analysis method: GLC alditol—acetates. averaged over two assays REACTION TIME TEMPERATURE GLUCOSE YIELD XYLOSE YIELD (sec) (’0) (mmol/g-wood) (mmol/g-wood) 15 30.6 0.0 0.299 30 30.0 0.130 1.140 45 30.3 0.142 1.035 60 30.2 0.335 — 75 30.1 0.454 0.829 90 30.0 0.361 0.719 105 30.3 1.336 0.980 120 30.4 1.588 0.927 135 30.1 1.448 0.957 150 30.3 2.665 1.292 180 30.2 2.681 1.166 210 30.5 2.852 - 225 30.5 2.554 - 240 30.2 2.751 1.380 270 30.5 2.809 1.274 48 Table 4. (cont'd.). HF partial pressure: 0.60 atm Analysis method: HPLC REACTION TIME TEMPERATURE GLUCOSE YIELD (sec) ('C) (mmol/g-wood) 30 30.2 0.060 60 30.5 0.189 75 30.5 0.358 90 30.7 0.646 105 30.5 0.766 120 30.2 0.640 150 30.5 1.263 180 30.5 1.572 210 30.5 1.847 240 30.6 2.085 270 30.5 2.231 300 31.0 2.301 330 30.0 2.565 360 30.2 2.870 390 29.6 2.593 420 30.2 2.375 XYLOSE YIELD (mmol/g-wood) OOOOOOOOOOCOOOOO .615 .695 .532 .668 .684 .601 .779 .847 .691 .943 .875 .756 .859 .649 .792 .851 49 Table 4. (cont'd.). HF partial pressure: 0.50 atm Analysis method: HPLC REACTION TIME TEMPERATURE GLUCOSE YIELD (sec) (°C) (mmol/g-wood) 60 30.2 0.103 120 30.5 0.235 150 30.7 0.366 180 30.5 0.450 210 30.5 0.816 330 30.7 1.706 450 30.4 2.265 540 30.4 1.915 660 30.5 2.455 780 30.5 2.424 XYLOSE YIELD (mmol/g-wood) OOOOOOOOCO 50 Table 4. (cont'd.). HF partial pressure: 0.44 atm Analysis method: HPLC REACTION TIME TEMPERATURE GLUCOSE YIELD XYLOSE YIELD (sec) (’C) (mmol/g-wood) (mmol/g-wood) 60 30.6 0.040 0.516 120 30.2 0.152 0.470 180 30.4 0.334 0.485 240 30.5 0.675 0.519 300 30.3 0.998 0.615 360 30.0 0.997 0.572 420 30.0 1.487 0.631 480 30.0 1.467 0.625 600 30.3 1.555 0.690 900 30.6 1.947 0.817 51 Table 4. (cont'd.). HF partial pressure: 0.2 atm Analysis method: HPLC REACTION TIME TEMPERATURE GLUCOSE YIELD XYLOSE YIELD (sec) ('0) (mmol/g-wood) (mmol/g-wood) 60 30.3 0.042 0.399 180 30.2 0.037 0.267 240 30.2 0.045 0.359 300 30.2 0.067 0.359 360 30.1 0.089 0.361 420 29.8 0.103 0.269 480 30.4 0.211 0.400 540 30.4 0.389 0.476 600 30.4 0.446 0.504 900 31.0 0.523 0.584 52 Table 4. (cont'd.). HF partial pressure: 1.0 atm Analysis method: HPLC REACTION TIME TEMPERATURE GLUCOSE YIELD XYLOSE YIELD (sec) (‘0) (mmol/g-wood) (mmol/g-wood) 10 50.0 0.095 0.302 20 49.4 0.063 0.333 40 49.8 0.522 0.458 50 50.4 0.690 0.492 60 50.3 0.905 0.492 70 50.5 1.281 0.679 80 50.5 1.122 0.508 90 50.4 1.574 0.565 100 50.5 1.667 0.654 120 49.9 1.943 0.659 180 50.5 2.445 0.767 240 55.4 2.429 0.742 300 55.8 2.580 0.627 420 55.8 2.899 0.567 53 Table 4. (cont'd.). HF partial pressure: 1.0 atm Analysis method: HPLC REACTION TIME TEMPERATURE GLUCOSE YIELD XYLOSE YIELD (sec) (’0) (mmol/g—wood) (mmol/g-wood) 15 27.6 0.207 0.830 30 27.5 0.697 0.527 45 27.6 1.171 0.729 60 26.6 2.579 0.939 75 28.4 2.384 0.795 90 28.6 2.515 0.861 150 28.6 2.813 0.898 Table 4. (cont'd.). HF partial pressure: 1.0 atm Analysis method: HPLC 54 REACTION TIME TEMPERATURE (see) ('C) 360 75.2 420 75.2 540 74.7 660 74.7 180 108 a 240 108 9 300 108 1 GLUCOSE YIELD (mmol/g—wood) 0000 COD .194 .157 .261 .402 .055 .125 XYLOSE YIELD (mmol/g-wood) 0000 COD .180 .108 .114 .179 .035 .061 .120 Sugar Chromatography 14757 ‘ 3 g 5: N O s E A 0 E a O I 0 on O :2 H w 9 I 2 C 3239. __.Z—\ R. I l T T 1 l l l u 15 20 25 as as so 45 Analysis method: GLC-alditol acetates/Internal Standard Sample description: sugars from HF-hydrolyzed Bigtooth Aspen. Reaction conditions: 180 sec, 30.4'C, HF partial pressure of 1.0 atm. Injection volume: 1.5 .ql Carrier gas: helium at 40 cc/min Temperature program: initial temperature of 130°C; l'C/min for 50 min Column: PROS-224 Detector: flame ionization (FID) Gas-liquid chromatography of water-soluble monomer sugar yields obtained from gas-phase HF solvolysis of Bigtooth Aspen. 55 56 26489‘ COUNTS 4771. GLUCOBI INOSITOL d; TFA XYLOSI TI T T I l l T I HINUTES Analysis method: HPLC/Internal Standard Sample description: sugars from HF—hydrolyzed Bigtooth Aspen. Reaction conditions: 75 sec, 28.4'0, HF partial pressure of 1.0 atm. Injection volume: 50 .41; ca. 400,qg total sugar plus standard Mobile phase: doubly-distilled, deionized, degassed water Flowrate: 0.6 ml/min Temperature program: isothermal at 85°C for 42 min Column: Biorad Aminex 42A and 87P in series Detection: Refractive index (RI) note: TFA peak due to incomplete evacuation of post-hydrolysis acid residue. High-performance liquid chromatography of water- soluble monomer sugar yields obtained from gas-phase HF solvolysis of Bigtooth Aspen. MICHIGAN STATE UNIV. LIBRARIES 1|HIWI‘lHIIWIW‘I"WWW“I"‘IWIHHW 31293107148847