Optimisation of the Fermentation of Dilute Acid Hydrolyzed Pine using Saccharomyces Cerevisiae for 2 nd Generation Bioethanol Production

Transcription

1 Optimisation of the Fermentation of Dilute Acid Hydrolyzed Pine using Saccharomyces Cerevisiae for 2 nd Generation Bioethanol Production Peter Larsson Department of Chemical Engineering, Lund University Abstract The work presented in this article is aiming at investigating and improving the ethanol production from dilute acid hydrolysed pine. The Process Development Unit (PDU group) at Lund University is currently developing a novel acid hydrolysis technique which has shown good overall yields. Various fermentation techniques together with yeast adaptation were performed to investigate and optimise the fermentation of this material. Experiments were carried out in 2L bench size fermentors with PID controls. Adapted yeast (S. cerevisiae) increased the overall yield by more then five times compared to regular yeast. Fed-batches increased the total yield by more than three times and a final ethanol concentration of 23. g/l was reached, corresponding to 82 % of theoretical yield. Fed-batch fermentation where the cells were not harvested showed higher productivity and yield than corresponding that had been harvested. They also showed higher stability at low cell concentrations. The results also indicate that the absolute concentrations of the inhibitors -hydroxymethyl-2-furaldehyde (HMF) and 2-furaldehyde (Furfural) has little importance to overall fermentative performance. Low performing fermentations could still convert most inhibitors, while some good performing fermentors did not manage to convert them. Experiments with added glucose showed complete conversion of all fermentable sugars and inhibitors. This supports the suggestion that the actual amount of fermentable sugars and intergroup relation with the inhibitors seems to be very significant, and this is also promising for further improvements of the hydrolysis process. Key Words: Lignocellulose, Ethanol, Dilute acid hydrolysis, Fermentation technology, Yeast adaptation, Cultivation, HMF, Furfural Introduction The expected depletion of the world s energy supply, accelerated by the fast economical growth in China and India combined with the surging oil prices and concerns about global warming has lead to increasing worldwide interest in alternative sources of energy [1][2][3]. Utilization of biofuels provides one new way of attaining energy and decreasing net-emissions of greenhouse gases. Ethanol has been produced using yeasts throughout history. It is an energy carrier with many similarities to petrol or diesel which are commonly used as car fuels. But there are one distinct difference if produced correctly, it can be combusted without release of fossil green house gases. This makes it ideal as fuel. Ethanol has so far primarily been produced from Brazilian sugar cane and corn from the USA. These techniques are fully developed, but the increased demand and political pressure have strained current production and highlighted both technical and environmental bottlenecks. Further on there have been substantial debates on the ethics and economics of using crops that could be used as human food for ethanol production, and on the factual CO 2 reductions from ethanol produced this way. [4] [] Lignocellulose rich materials such as wood, agricultural and industry residues have been proposed as attractive alternative routes from both energetic and environmental perspectives towards second generation biofuels [6] [7]. But since the sugar content of these materials is much more difficult to extract than first generation biofuel substrates (e.g. grain, corn) the process becomes much more complicated. There are currently several strategies of producing ethanol from lignocellulose being implemented all over the world, but no one has been able to do it in a cost efficient way so far [8]. The currently most promising method for degrading the raw materials is using enzymes for the hydrolysis. However, the enzymes are still much too expensive to allow large scale production and the market is eagerly awaiting a drop in prices. A route which has been traditionally used for hydrolyzing cellulose is treatment with acid. This method is much simpler and cheaper, but the disadvantage is that the yields are low, and the acid hydrolysis forms high levels of inhibitors so fermentation without some kind of detoxification has proven difficult as best so far [9][]. 1

2 In the current work ethanol production from pine, which consist of lignocellulose biomass, was studied. The objective was to improve fermentation of dilute acid hydrolyzed pine at industrial applicable conditions without prior detoxification. The strategies used were fermentation techniques and adaptation of the yeast. Material and Method Raw material The raw material constituted of Pine (Pinus sylvestris) grown in Härjedalen, in the north of Sweden. The pine was chipped in Sveg by Härjedalen Miljöbränsle AB and sent to Lund in sealed bags. The chips were roughly 1-6 cm long and 4 8 mm thick. The composition of the raw material is given in table 1. The dry matter content was determined to 4.4 %. TABLE 1 Composition of the pine used. Compound Concentration Glucan 39 % Mannan 11 % Xylan 7 % Galactan 3 % Arabinan 2 % Lignin 38 % This corresponds to a % content of fermentable hexose sugars in the raw material. The pentose content is less then %. Hydrolysis The wood chips were impregnated with dilute sulphuric acid (H 2 SO 4 ) to a final acid concentration of 1 %. Approximately 13,2 Kg DM pine was soaked (34 kg wet). The material was treated in a novel hydrolysation process were it was heated to 2-22ºC for -9 minutes. The material was then collected in a cooled flash off vessel. The process was performed in a L reactor with approximately 7 g DM pine per batch. The reactor was heated by a steam boiler providing -2 bar of steam. Separation The hydrolysed solution was filtered in a pilot filter press (Larox Laboratory Filter PF,1 H2). Thirty liters of hydrolysate was collected, together with 4 filter cakes. The slurries from all batches were pooled together before distributing into 1L flasks and than frozen at -2 ºC. The carbohydrate composition of the hydrolysate is given in table 2. TABLE 2 Composition of the lignocellulosic hydrolysate Compound Sample 1 Sample 2 Average Glucose Mannose Xylose Galactose Arabinose Acetic Acid HMF Furfural The sum of the fermentable sugars, glucose and mannose, becomes 6. g/l. This means that the theoretical maximum ethanol concentration after fermentation will be 28.6 g/l. Cultivation and Adaptation Saccharomyces cerevisiae was obtained from Jästbolaget AB (Rotebro, Sweden). The yeast was maintained on regular non-selective agar plates made from yeast extract g/l, soy peptone 2 g/l, agar 2 g/l with D-glucose 2 g/l as additional carbon source. Inoculum cultures were grown in 2 ml cotton plugged baffled flasks on a rotary shaker at 3ºC for 2h. The liquid volume was 7 ml and the shaker speed was 7 rpm. The defined growth medium was prepared according to Taherzadeh et al [11] with the exception of the glucose concentration which was 24 g/l. Cultivations were made in 2L bioreactor (Labfors, Infors AG, Brottingen, Swizerland) at 3ºC and stirring at 7 rpm. The flow rate of air was set to approximately 1. bar and 2 units on the fermentor. The ph of the medium was controlled at. (±.1) by automatic addition of % w/w NaOH. The reactor was initially filled with ml or 7 ml (same as fed-volume) of the same defined medium as the inoculum, but with a glucose concentration of 2 g/l. The concentrations of mineral salts and metals were however doubled to compensate for the dilution. The initial OD ranged from.44 to.84. The po 2 was continuously measured and the rpm was adjusted towards a maximum of 9 rpm so that po 2 did not fall under %. When the glucose was consumed (po 2 = %) the hydrolysate feeding was started. The added volume of hydrolysate was the same as the initial filling volume (/7 ml), but 8ml/Lhydrolysate of % NaOH was also added to adjust the ph to.. The feed rate was corresponding to a feeding time of 16 h (.78 and.2 ml/min respectivesly). 2

3 This method of gradually exposing the cells for the hydrolysate is known as adaptation, and is performed to increase the fermentative performance of the yeast. The cells were harvested by centrifuging the fermentation broth in 7mL containers using a HERMLE Z 13 K centrifuge (HERMLE Labortechnik, Wehingen, Germany) at 3 rpm for minutes. The supernatant was then discharged, leaving a fairly dry (~2 % DM) suspension. The time between harvest and the addition of the harvested cells to the fermentors were kept below 3 minutes. Fermentation Fermentations were made in 2L baffled bioreactor (Labfors, Infors AG, Brottingen, Swizerland) at 3ºC with stirring at 3 rpm. The ph of the medium was controlled at. (±.1) by addition of % w/w NaOH. Both batch and fed-batch fermentations were performed, with the same amount of total hydrolysate ( ml). The hydrolysate was adjusted to ph. by addition of 8ml/L % w/w NaOH. The total final volume per fermentor then became approximately 1 8ml since no other substances were added. In all fed-batches the proportions between initial and final volume was 1:4. In the trials with fermentation without cell harvest, the initial volume was only that of the cultivation broth containing the cells. The hydrolysate volume was the same, 1 8 ml, which resulted in other final volumes. Analyses The dry matter (DM) content of the cultivated yeast was determined by drying them in ºC for 24h. The concentrations of monomeric sugars, ethanol and inhibitors were performed by means of HPLC-analysis (Shimadzu LC-AD, Kyoto, Japan). Samples of 2ml were withdrawn from the reactor, centrifuged at 3 rpm for 3 minutes, and then filtered through a.2μm filter (Dismic-13CP2AN, Toyo Roshi Kaisha, Ltd., Japan). The samples were then kept at -2 ºC. Glucose, xylose, galactose, mannose and arabinose were separated using a Pb-column (Aminex HPX-87P, Bio- Rad, USA) at 8ºC, with water as eluent, at a flow rate of. ml/min. The compounds were quantified using a refractive index (RI) detector (Shimadzu RID-A, Kyoto, Japan). Ethanol, glycerol, acetic acid, furfural and - hydroxymethyl furfural (HMF) were separated on an H- column (Aminex HPX-87H, Bio-Rad, USA) at 6ºC, with mm H 2 SO 4 as eluent, at a flow rate of. ml/min. The compounds were quantified using a refractive index (RI) detector (Shimadzu RID-A, Kyoto, Japan). Results The batch performance was examined in three batch processes (B1 B3) with different yeast concentrations. These were compared to six fed-batch trials with different yeast concentrations and fed-rates (FB1-FB3, FB-FB7). The results of the adapted yeast were evaluated by a fed-batch with regular yeast (FB4). TABLE 3 Summary of results for trials B1-3, FB1-7. Cells refers to initial yeast concentration (DM); Ethanol is maximum ethanol concentration. Yield is percent of theoretical maximum (.1 g cells per g fermentable sugars) *Regular non adapted yeast. Cells Fed-rate (ml/min) Ethanol Yield (%) B B2. - 8,9 3 B3. - 6,9 27 FB FB FB FB4.* FB FB The overall performance is low in all batch fermentations, and the one with low yeast concentration (B1) seems to stop completely within the first 24 hours (see figure 1-A). Results indicate a decrease in ethanol after ~ h. This is probably due to a combination of very low concentrations and error in sample handling. The trials with a regular fed batch were performed both with yeast that had been and that had not been adapted. The adapted yeast consistently has a much higher productivity (see figure 1-B). The non adapted yeast shows some initial conversion, but seems to loose all performance after hours. The last trials matrix were set up to investigated if it was possible to decrease the cell concentration by starting the fermentation on the cultivation broth, without harvesting the cells first (FB12-14, FB16). The trials were also designed to give a hint on what could be expected if the hydrolysis were to be further improved and yield an initial fermentable sugar content of 8 g/l (FB14-16). This was done by an addition of glucose to the hydrolysate prior to fermentation. (see Table 4) TABLE 4 Summary of results for trials FB Cells refers to initial yeast concentration; L/H indicates whether cells were added as liquid or harvested. Ethanol is maximum ethanol concentration. ID# CELL CONC. (g /L) FEED RATE (ML/MIN) SUGAR ETOH YIELD % FB g/l L FB13 4. g/l L FB g/l L FB1 1,6 g/l H FB g/l H

4 A. B FB1 6 4 B1 B2 B3 1 FB2 FB3 FB4 FB FB C. D FB12 FB FB14 FB1 FB FIGURE 1 A-D Total ethanol in fermentators (g). The final volume is 8 ml in all cases except FB12, FB13 and FB14, where it was 1.2, 1.37 and 1.2 ml respectively. For information about cell concnteration and fed rate, see table 3 and 4. Fermentations without cell harvest showed a continuous high performance and no lag phase (see figure 1-C). The yield increased by approximately 6 % compared to FB/FB6 with similar attributes. S.Cereviciae has the ability to grow at anaerobic conditions (cultivation), even though it prefers aerobic conditions. Since the cultivation solution was directly fed with a higher rate of hydrolysate, it would be likely that the some substrate would be used to generate new cells which impair the total yield. However, figure 1-C indicates that this probably not occurs in these trials since the ethanol formation starts direct and the yield is almost total (73 and 7 % respectively). The increased sugar content in FB14-16 results in ethanol concentrations close to 4 g/l. The conversion is very fast, and the non harvested cells perform better than the cells that have been harvested, as expected (see figure 1-D). Discussion The fed-batch fermentations showed that adapted yeast is more than six times more effective than the nonadapted yeast. The result also indicates that regular yeast becomes inactive and probably dies after less then six hours. The adapted yeast can readily ferment the sugars and almost completely convert all glucose and mannose into ethanol. The highest achieved yield was.42 g/g, which is close to the theoretical maximum of.1 g/g. Trials with cell concentrations ranging from 2., g/l final volume were performed. The fermentations with g/l always gave high yields and productivity, but in the fed-batches with 2 g/l a larger variation in result are shown. Some 2 g/l fermentations had almost as high productivity as the g/l, but a small change in set up reduced the yield dramatically, while the g/l performance seemed to be almost constant. When the yeast was not harvested, and instead added directly from the cultivation solution, a high yield could be achieved at cell concentrations as low as 1.6 g/l. This 4

5 shows that a initially less toxic surrounding can be of great benefit. However, some cell growth could have occurred in the anaerobe fermentation phase, which could mean that the actual cell content was higher than what was added. But since the overall yield is very high, this can not have been a major effect. The results in this work indicate that 2 g/l of cells is to low to obtain reproducible results, even at fed-batch operations. The high level which was tested here is probably higher than necessary and an optimum is probably found between 2- g/l. When comparing the techniques it is obvious that the least productive fed-batch is more than two times as efficient as the most productive batch. The most effective fed-batch is four times as efficient as the batch. The reason for this is unclear, since the most probable theory; that the total level of inhibitors becomes too high for the yeast to detoxify the surroundings, is not supported by the results. The yeast converts most HMF and furfural in the batch, but still the overall ethanol yield is low. Meanwhile some of the more successful fedbatches still contain much inhibitors at the end. The fed-batch differs from the batch since the cell initial concentration is much higher, which should mean that each cell only have to convert fewer inhibitormolecules. However, as mentioned before the final concentration is approximately the same. This further indicates that mechanism that separates the techniques is some other than the theory of inhibitor levels. One other possibility is that the cells continue to grow in the ferementation, which combined with a selection pressure yield viable cells that can ferment the hydrolysate. The high total yield in the sugar-added fermentations (FB1 and FB16) contradicts this, since the ethanol yield would be much lower if some sugars are consumed for cell formation. Even though the exact mechanism is not quite clear, the result is. The findings that it is very difficult to ferment dilute acid hydrolysates in batch operations is also concurrent with that of Taherzadeh et al. [12] It was expected that a lower feed rate would result in lower productivity but higher overall yield. This is also shown in FB3 in comparison with FB1-2. However the results showed that in the cases with lower cell concentration, a higher feed rates increased performance and that too low feed rate actually lowered the yield significantly. Since the batch fermentations were the least successful, there must be an upper optimal level of which feed rate that can be used before the performance decreases again. The equipment did however not allow this limit to be investigated. These results further support the previous theory that the detoxification is closely linked to the available sugar content. When the available energy is to low, all is directed to intracellular supporting processes and little detoxification is performed. The inhibitors can then exercise the toxic effect, which reduces the total viable cells and hence total performance. When the cells were added directly in suspension, without harvest (FB12-14) the results show higher yields and also higher final concentrations, even though the solution is more diluted. The final yield is approximately 6 % higher than similar fed-batches with harvested cells (FB/FB6). This indicates that the harvested cells suffer more severely from some form of stress when they are added directly to the hydrolysate. The additional volume when the cells are not harvested is though not negligible, since it adds approximately one fourth of the hydrolysate volume. The cultivation broth could contain some vital nutrient, salts etc which improve fermentability and may be an important In FB12-1 an additional burst of pure glucose was given to the fermentors after 98h. The results are not clear, but in the fermentors that had an initial glucose concentration of 6 g/l the added glucose is ackumelated indicating that no viable cells are present. In the fermentors with an initial glucose concentration of 8 g/l no accumulation is shown, indicating that there are viable cells. Conclusions Fermentation of dilute acid hydrolysed pine should be performed as a fed-batch operation, with > 2 g/l of adapted yeast and a rather high feed rate. The inhibitors HMF and furfural are not singly responsible for determining the fermentability. The factual amount of fermentable sugars and intergroup relation with the inhibitors seems to be very significant. Even though this study has shown that it is possible to have a high yield of ethanol from hydrolyzed pine, the absolute concentrations are to low. This causes the cost of the distillation of the fermentation broth to be substantial. The initial content of fermentable sugars in the pine is approximately kg per ton DM. The yield of fermentable sugars in the hydrolysis is approximately.2 kg/kg. When the sugars are fermented, this study indicates that at least 4 % of the sugars can be fermented in the process. This corresponds to an end yield of approximately 1 l of ethanol per ton DM substrate. In order to reach the objective of the final production facility of 2 2 l per ton, the yield in the hydrolysis step needs to be increased considerably. It will be very difficult to improve the fermentation yield more than a few units.

6 References [1] Demirbas A. (27). Progress and recent trends in biofuels, Progress in Energy and Combustion Science, 1:1-18 [2] Energimyndigheten. (26). Rapport: Kinas växande energibehov Snabb ekonomisk tillväxt påverkar den globala energimarknaden. Accessed August 8, 28: FilAtkomst/ET2_6W.pdf/$FILE/ET2_6W.pdf [3] Stern N. (27). The Economics of Climate Change. Cabinet Office - HM Treasury. ISBN 13: [4] Granda C., Zhu L. & Holtzapple M. (27). Sustainable liquid biofuels and their environmental impact. Environmental Progress 3:233-2 [] Pimentel D. & Dale B. (27). The Costs Of Biofuels. Chemical & Engineering News 1:12-16 [6] Olofsson K., Bertilsson M. & Lidén G. (28) A short review on SSF an interesting process option for ethanol production from lignocellulosic feedstocks. Biotechnology for Biofuels. 1:7-21 [7] Gray KA., Zhao L. & Emtage M. (26).Bioethanol. Curr Opin Chem Biol : [8] Schubert C. (26). Can biofuels finally take center stage? Nature Biotechnology. 24: [9] Millati R., Niklasson C., & Taherzadeh M. (22). Effect of ph, time and temperature of overliming on detoxification of dilute acid hydrolysates for fermentation by Saccharomyces cerevisiae. Process Biochemistry 38:1-22 [] Rudolf A. (27). Fermentation and cultivation Technology for Improved Ethanol Production from Lignocellulose. Thesis (PhD). Lund University. [11] Taherzadeh M., Lidén G., & Gustafsson L. (1996). The effects of pantothenate deficiency and acetate addition on anaerobic batch fermentation of glucose by Saccharomyces cerevisiae. Appl Microbiol Biotechnol 46: [12] Taherzadeh M.J., Niklasson C. & Liden G. (1999). Conversion of dilute-acid hydrolyzates of spruce and birch to ethanol by fed-batch fermentation. Bioresource Technology 1:9-66 6