Control of fermentation of lignocellulosic hydrolysates Anneli Nilsson Department of Chemical Engineering II, Lund University P.O. Box 124, S-221 00 Lund, Sweden In this work substrate feeding rate to a fermentation of lignocellulose hydrolysate for production of ethanol has been studied. During hydrolysis of lignocellulose inhibitors are formed that effect the microorganism (here Saccharomyces cerevisiae) that ferments the sugar to ethanol in a negative way. To make the hydrolysate fermentable the inhibitors are often removed by detoxification prior to fermentation, but an alternative to detoxification is to use fed-batch fermentation. The reason for this is that the yeast can convert the inhibitors to less toxic compounds, if high concentrations of the inhibitors are avoided. To get a successful fermentation the substrate feed rate should be controlled so that severe inhibition is avoided and the production of ethanol is maximal. In this work on-line measurement of carbon dioxide evolution rate (CER) was used to control the feed rate of hydrolysate in fed batch fermentation of wood hydrolysate. A control algorithm was created for the feed rate of the hydrolysate and the results showed that the in-situ detoxification worked for moderately inhibiting hydrolysates but not for more severely inhibiting hydrolysates. Introduction Ethanol can be regarded as a more environmentally friendly fuel than gasoline because it adds only little net carbon dioxide to the atmosphere. This is the main reason why large research efforts are done to find a cheap way of producing ethanol from renewable raw materials although also other reasons (e.g. agricultural politics, economic independence) are important. Ethanol production Ethanol can be produced in two different ways. Either chemically, by hydration of ethylene, or by fermentation of sugar-containing feeds, starchy feed materials or lignocellulosic materials. A process for ethanol production by fermentation of lignocellulose is schematically shown in Figure 1 (Olsson, 1994). In this work only fermentation of lignocellulose has been studied. Lignocellulose contains the following substances: (Danielsson, 1996) Cellulose: a polymer of β-dglucose. Hemicellulose: contains two different polymers: glucomannan (a polymer of glucose and mannose) and xylan (a polymer of xylose). Lignin: a polymer of guajacylpropane- and syringylpropane units. Extractives: for example: unsaturated fatty acids. The first steps in the process are prehydrolysis and hydrolysis. In these steps the lignocellulose is delignified and a depolymerisation takes place. Prehydrolysis can be chemical, physical or biological. There are two principally different hydrolysis methods: acid hydrolysis and enzymatic hydrolysis. (Taherzadeh, 1999)
Figure 1 Flowchart of ethanol production from lignocellulose. In this process pentos and hexos fermentation is performed separately. During the hydrolysis not only free sugars are formed, but also inhibitors. Examples of inhibitors are: furfural, 5- hydroxymethyl furfural (HMF), carboxylic acids and phenolic compounds. Two of the most important inhibitors are furfural and HMF (Figure 2) (Taherzadeh, 1999). common pentos) to xylulose. Xylulose can be fermented by Saccharomyces cerevisiae to ethanol. 2. Use a yeast that naturally ferments xylose to ethanol: Pachysolen tannophilus, Candida shehatae or Pichia stipitis. 3. Use recombinant microorganisms that can convert both pentoses and hexoses to ethanol. Figure 2 HMF and furfural. The inhibitors have a negative effect on the microorganisms used in the fermentation and therefore the hydrolysate is often detoxified to reduce the concentrations of inhibitors. Detoxification processes can be chemical, physical or biological (Taherazadeh, 1999). After detoxification the hydrolysate is fermented. To be economically profitable both pentoses and hexoses must be fermented in the process (especially if hardwood, that contains more pentoses than softwood, is the raw material). There are three different ways to ferment pentoses: (Zacchi, 1999) 1. Use the enzyme xylosisomeras that catalyze the reaction of xylose (a The well-known Baker s yeast, Saccharomyces cerevisiae, is the most frequently used microorganism in hexose fermentation. S.cerevisiae can produce ethanol from glucose and mannose if the concentrations of sugars are high or when the yeast is grown under anaerobic conditions. (Ratledge, 1991) To get an efficient fermentation severe inhibition should be avoided. There are four different strategies to do this: (Taherzadeh, 1999) 1. Modify the hydrolysis process so that less inhibitors are formed. 2. Detoxification. 3. In-situ detoxification. Furfural and HMF can be converted to less toxic compounds by S.cerevisiae if high concentrations of the inhibitors are avoided.
4. Use a microorganism that is less sensitive to the inhibitors. Fermentation alternatives In this work in-situ detoxification in a fed-batch fermentation has been studied. The fermentation can be done in a batch, fed-batch or continuous process. The advantage with a fedbatch process is that high concentrations of inhibitors can be avoided and there is no loss of substrate or microorganisms. Materials and methods In this work fermentation was done by S.cerevisiae (CBS 8066) under anaerobic conditions. First the yeast was grown for 24 hours in a 1000-ml E-flask with 400 ml 1.5 % glucose solution (with mineral salts, trace metals etc.) at 30 C. The aim of this was to get a solution with a high cell concentration to start the fed-batch with. 20 ml of this solution was then added to a bioreactor with a volume of 3 liters. The reactor contained initially 1 liter of a glucose solution (including mineral salts, trace metals etc. for the whole fermentation). Fed-batch operation was started when all the glucose was consumed (the carbon evolution rate (CER) starts to decrease). During the whole fermentation ph was controlled at 5.0 (with 2 M NaOH), temperature to 30 C and nitrogen gas (600 or 1000 ml/min) was sparged through the system to create an anaerobic environment for the yeast. (Figure 3). When 1500 ml of substrate had been pumped into the reactor the pump was stopped. (Figure 3) The composition of the gas leaving the reactor was continuously measured by a gasanalyzer that can detect carbon dioxide, oxygen and ethanol. A flow injection analysis (FIA) system was used to measure the cell concentration in the reactor at least once an hour. The analysis results were collected by a computer and used by the computer program that controls the substratefeeding rate. Samples were taken out from the reactor so that the content of the liquid phase could be analyzed on a HPLC (High Performance Liquid Chromatography) and so that the dryweight could be measured. The hydrolysate used had been produced from a two-stage dilute acidhydrolysis of spruce and the composition based on HPLC analysis is shown in Table 1. Substans Amount (g/l) Glucose 21.9 Mannose 16.9 Xylose 8.1 Galactose 6.7 HMF 3.2 Furfural 1.5 Table 1 The content of the hydrolysate. Theory The aim of this work was to construct and implement a control strategy for the substrate feed rate so as to avoid severe inhibition ( stuck fermentation ) and minimize fermentation time. At the optimal flow rate the inhibition should be minimal and the production rate of ethanol should be maximal. An idea is to base the control algorithm on CER. CER is a good measurement for ethanol production because for every mole ethanol produced 1 mole of carbon dioxide are also produced. If no biomass is formed the fermentation from glucose to ethanol can be summarized in the following reaction:
C 6 H 12 O 6 2 C 2 H 6 O + 2 CO 2 The advantage of using CER is that it can be measured continuously on-line. Figure 3 A schematic picture of the experimental setup. Results and discussion Different algorithms for the pump control were tested. One algorithm that worked well was to increase the rate of the fed-pump in proportion to z, defined by: z= (1/CER)dCER/dt where t is time [s]. z is a normalized derivate of CER. The change of the feed rate was done when the CERvalue was approximately constant. From the results (Figure 4) one can see that the algorithm worked satisfactory. The fermentation took about 28 hours to complete and the amounts of inhibitors (especially furfural) were kept low. The advantages of the fed-batch process becomes clear when the results are compared with a batch process (Figure 5). In the batch process the fermentation is slow (about 56 hours). Figure 4 Results from a fed-batch fermentation of hydrolysate by S.cerevisiae. The pump rate was increased with 25*z ml/h.
Future work should focus on how to shorten the fermentation time using different control algorithms, how to ferment severely inhibiting hydrolysates. The effects of the inhibitors on the microorganisms should be further examined. More conclusions of the inhibition could e.g. be drawn from measurements of the amounts of viable cells. Literature Cited Danielsson, N-Å. 1996. Processteknologi. LTH, Lund, Sweden. Olsson, L. 1994. Ethanol production from lignocellulosic materials. Ph.D. thesis. LTH, Lund, Sweden. Figure 5 Results from batch fermentatin of hydrolysate by S.cerevisiae. The hydrolysate was pumped into the reactor with a rate of 2400 ml/h during a short time period. An experiment where hydrolysate with added furfural was used as substrate was done to test an algorithm for decrease of feed rate. This experiment showed that it is hard to get a inhibited system to work again and this suggests that the inhibition is irreversible. Ratledge, C. 1991. Yeast physiology a micro-synopsis. Bioprocess Engineering. 6:195-203. Taherzadeh, M.J. 1999. Ethanol from lignocellulose: physiological effects of inhibitors and fermentation strategies. Ph.D. thesis. Chalmers Univ, Göteborg, Sweden. Zacchi, G. 1999. Etanol ur energiskog. Biokemisk reaktionsteknik 1999. Course notes. LTH, Lund, Sweden. Conclusions The current work has shown that, with the hydrolysate used, is was indeed possible to control the feed rate of the hydrolysate so that severe inhibition could be avoided and the fermentation was finished in a short time. However, more severely inhibiting hydrolysates could not be handled fully adequately.