Selective Short Chain Carboxylates Production by Mixed

Selective Short Chain

Carboxylates Production

by Mixed Culture

Fermentation

Doğa Arslan

Doğ

a Arslan

Selective Sh

ort C

hain

Carb

oxy

lates Pro

du

ction

by

Mix

ed C

ultu

re Ferm

entatio

n

VK ST thesis omslag def1.indd 1 25-09-2014 23:48:44

SELECTIVE SHORT CHAIN CARBOXYLATES

PRODUCTION BY MIXED CULTURE FERMENTATION

DO A ARSLAN

Thesis committee

Promotor

Prof. Dr C.J.N. Buisman

Professor of Biological Recovery and Re-use Technology

Wageningen University

Co-promotors

Dr H. De Wever

Project Manager

VITO, Flemish Institute for Technological Research, Mol, Belgium

Dr K.J.J. Herben-Steinbusch,

Researcher, Sub-department of Environmental Technology

Wageningen University

Other members

Prof. Dr J.H. Bitter, Wageningen UniversityProf. Dr B. Calli, Marmara University, TurkeyDr R.A. Rozendal, Paques b.v., Balk, the NetherlandsDr R. Kleerebezem, Delft University of Technology, the Netherlands

This research was conducted under the auspices of the Graduate School SENSE (Socio-Economic and Natural Sciences of the Environment)

SELECTIVE SHORT CHAIN CARBOXYLATES

PRODUCTION BY MIXED CULTURE FERMENTATION

DO A ARSLAN

Thesis

submitted in fulfillment of the requirements for the degree of doctor

at Wageningen University

by the authority of the Rector Magnificus

Prof. Dr M.J. Kropff

in the presence of the

Thesis Committee appointed by the Academic Board

to be defended in public

on Tuesday 21 October 2014

at 1:30 p.m. in the Aula.

Do a Arslan

Selective short chain carboxylates production by mixed culture fermentation,

214 pages.

PhD thesis, Wageningen University, Wageningen, NL (2014)

With references, with summaries in Ducth and English

ISBN: 978-94-6257-116-7

Babam ve annem için

For my parents

Table of Contents

Chapter 1 Introduction............................................................................................................ 1

Chapter 2 Selective short chain carboxylates production: control mechanisms to direct mixed

culture fermentations............................................................................................................ 17

Chapter 3 Effect of hyrogen and carbon dioxide on carboxylic acid patterns on mixed culture

fermentation ........................................................................................................................ 71

Chapter 4 Selective carboxylate production by controlling hydrogen, carbon dioxide and substrate

concentrations in mixed culture fermentations ....................................................................... 99

Chapter 5 In-situ carboxylate recovery and simulatenous pH control by tailor-configured bipolar

membrane electrodialysis during continuous mixed culture fermentation ................................. 133

Chapter 6 General results and Discussion ............................................................................. 169

Summary …………………………………………………………………………………………………………………………196

Samenvatting…………………………………………….……………………………………………………………………..201

Acknowledgements.…………………………………………………………………………………………………………..209

About the author….……………………………………………………………………………………………………………211

What is life?

It is the flash of a firefly in the night. It is the breath of a buffalo in the wintertime. It is the little shadow which runs across the grass and losses itself in the sunset.

Blackfoot

1

Chapter 1

Introduction

2

1.1. Bio-based Economy

The term ‘bio-based economy’ relates to the transformation of biomass resources into

energy and bio-refinery. Due to the recent improvements in bio-based technologies, worldwide

bio-based chemical and polymer production (excluding biofuels) has already increased to 50

million tons making 10% of the global production of chemicals (KBBPPS, 2013; IEA Bioenergy,

2012). Several countries such as China, USA, Canada and The Netherlands already declared that

they are ambitious to replace even 25 - 30 % of petroleum based chemicals with bio-based

alternatives by 2030 (Patel et al., 2006; IEA Bioenergy, 2012). This indicates that biomass is

becoming a very important resource for the large-scale production of a broad spectrum of non-

food chemical products such as plastics, paints and constructing materials (Koops et al., 2012).

The biggest challenge for a bio-based economy is to provide sufficient and stable

feedstock. Around 73 % of the biomass worldwide is used for food and feed production (Bos-

brouwers, et al., 2012). It is inevitable that the demand for food and feed will even increase more

with the growing global population. Establishing a stable bio-based economy while keeping a

sufficient food supply, will require on the one hand efficient use of limited natural biomass

resources and on the other hand improved technologies for waste recycling. Both are becoming

key issues for the bio-based economy. (Ostergard et al., 2012). The latter is the main concept of

this thesis.

1.2. Carboxylate platform in the bio-based economy

In bio-refinery concept, there are three main platforms used to convert biomass into

valuable products (Figure 1.1). These platforms are syngas platform, sugar platform and

carboxylate platform. In syngas platform organics are gasified by thermochemical processes into

CO2, CO, H2 and H2O. The resulting gases are later used in secondary processes to produce

3

alcohols or other valuable products. Hydrogen gas can be used as direct energy source after

purification. In sugar platform biomass is hydrolyzed to sugars for example by using acids or

enzyme catalysts. Sugar can then be further fermented for production of valuable compounds such

as specialty chemicals like lubricants, food ingredients (e.g. enzymes), health products (e.g.

penicillin) and bulk chemicals (e.g. acids and alcohols).

Figure 1.1. Bio-refinery platform concept and possible intermediate and end products. Modified from(Danner & Braun, 1999; Cherubini et al., 2009).

‘Carboxylate platform’ concept is based on the principle of acidification (fermentation).

Biomass in the form of a mixture of carbohydrates, proteins and lipids is hydrolyzed to

corresponding monomers such as sugars, amino acids, long chain fatty acids and glycerol. These

monomers are converted into short chain carboxylates mainly consisting of acetate, propionate, n-

butyrate in varied concentrations and a small portion of valerate, n-caproate and ethanol by mixed

4

culture fermentative organism. Together with carboxylates, H2 and CO2 gasses are produced in the

headspace. In anaerobic conditions, it is thermodynamically and kinetically more feasible that

organics are reduced to methane rather than to desired carboxylates. Hence, methane producers

needs to be inhibited strictly. Moreover the process needs to be steered to towards operational

conditions favoring acid production, e.g. pH below 6 and short retention times. Inhibition of

methane production, e.g. by 2-bromoethanesulphate or heat treatment can be either expensive or

toxic to environment. Yet, if one of these conditions fails, short chain carboxylates production is

limited by methane producers.

Principally, all three platforms use biomass to produce valuable products but the method

used in each platform differs in practice. Economics and efficiency determine which platform is

most suitable for which type of biomass. A comparison of three platforms was made by Holtzapple

& Granda, 2009 in which ‘ideal biomass’ defined as 68% polysaccharides and 32% lignin was

converted to ethanol. Lignin is not biologically reactive under anaerobic conditions and can only be

processed thermochemically. Polysaccharides can be processed thermochemically or biologically

under anaerobic conditions. These platforms were investigated according to aspects of efficiency

and conversion capacity in detail. Apparently the highest product yield among the three was

achieved in carboxylate platform where the conversion of biomass rounded up to 480 L

ethanol/ton ash free biomass. The yield in sugar platform was 314 L/ton and in syngas (or

thermochemical) platform was 302 L/ton. Furthermore, carboxylate platform proved less energy

intensive compared to the thermochemical platform and did not require septic conditions or

expensive enzymes like in sugar platform. These enzymes are naturally produced in the mixed

culture environment. Compared to sugar platform, carboxylate platform route also turned out to

be more tolerant to instabilities such as variable substrate composition and ion concentrations.

5

The major advantage of carboxylate platform is elimination of expensive sterilization and

high investments in stainless steel fermenters. Hence, low grade, cheap and readily available

biomass can be cost effectively converted to short chain carboxylates.

Short chain carboxylates

Carboxylic acids or carboxylates are organic compounds which contains one or more

carboxyl groups. According to the carbon number in the chain, they grouped as short (C2-C6),

medium (C7-C12) and long (C13-C21) chain carboxylates. This thesis focuses on the production of

short chain mono-carboxylates or so-called ‘fatty acids’, which are acetate, propionate and n-

butyrate. Short chain carboxylates are versatile bulk chemicals which can be directly used in

industry or can be used as building blocks in several industrial applications such as production of

dyes, textile, cosmetics, pesticides, pharmaceuticals, food additives, lubricants, resins, soap

cleaning agents, plastics, surfactants, medium and long chain carboxylates and alcohols (Figure

1.1).

1.3. Current carboxylate production applications and market

Today, petro-chemical processes are well-established and have so far always provided a

cheaper option to produce carboxylates compared to biological systems. This was mainly due to an

abundance of oil resources, leading to relatively low crude oil prices. However, achievable crude oil

extracted from conventional petroleum reservoirs is rapidly diminishing (Head, 2011). Moreover,

petroleum based sources create a serious threat for environment because of massive damage to

environment during extraction and an adverse impact on the global climate due to emission of

greenhouse gases (GHG) e.g. CO2, N2O. According to the last IPCC report (2013), the recorded

global mean temperature rise is 0.6 C/year and it will continue to rise over the 21st century if GHG

emissions continue unabated. It is known that only few degrees of warming causes a climate

6

change, drinking water shortages, sea level rise, species reduction and many other adverse effects

in global (Hughes, 2000). The use of biomass as an oil-replacing feedstock could decrease the

negative impacts of global warming.

Table 1.1 shows the annual global production capacities and current prices of short chain

fatty acids. Despite the negative impacts, the majority of carboxylates is still produced from fossil

resources by chemical synthesis. Biological production of carboxylates is limited compared to

petrochemical production. Among the short chain carboxylates, biological production of acetate is

well established in industry though it still has a small market share. Almost all acetate produced by

biological fermentation is used for vinegar (food industry), polyethylene terephthalate (plastic),

vinyl acetate (additive for dye industry) and cellulose acetate (textile industry) production

(Wisbiorefine, 2009). For n-butyrate and propionate, bio-based production technologies are not in

pilot plant stage yet (Patel et al., 2006; Xu & Jiang et al., 2011).

Current acetate production is achieved by one step batch or fed-batch fermentation of

sugar based feedstock with pure culture under sterilized conditions. After fermentation, acetate is

harvested from fermentation by downstream application such as crystallization. Pure culture

processes are preferable in industry for two reasons. Firstly, they result in a single product at high

concentration and purity. Secondly, they can provide almost complete conversion of biomass. For

instance, biological acetate production in industry can generate product streams of 50 g/L with

90% purity and substrate conversion efficiency is 0.8-0.9 g acetate produced/g glucose consumed

(Rogers et al., 2006). The most important advantage of a high product yield and purity is that it

lowers the costly downstream processes.

7

Table 1.1. Properties, annual market share of major short chain carboxylates after fermentation ofbiomass

Chemical MolecularFormula

pKa at25 °C

Annualmarket (Mt)

Biologicalproduction inmarket (Mt)

Market Price(petrochemical)(€/t)

Acetic acid C2H4O2 4.76 3.51 0.192 5003

Propionic acid C3H6O2 4.88 0.356 / 10006

n-Butyric acid C4H8O2 4.82 0.54 / 8005

1Yang et al., 20072Patel et al., 20063Tecnon OrbiChem, acetic acid fact sheet, 20134Straathof, 20145Zhu, 20036Icis, 2007

1.4. Carboxylates from organic waste streams

Carboxylates can be produced from a wide range of biomass resources: (1) plants from

agricultural areas e.g. corn, sugarcane; (2) plants grown in lands abandoned from agriculture e.g.

grasslands (3) crop residues e.g. corn stover, straw from rice, (4) forest residues e.g. wood, (5)

marine vegetation e.g. algae and (6) organic waste substances e.g. food waste, waste streams

from industrial organic processes (Tilman et al., 2009; Long, Li, Wang, & Jia, 2013).

Latest legislations about food security concerns has limited the bio-chemical and bio-fuel

production from food-competing feedstock (World Economic Forum, 2010). For suitable green

chemistry, it is desired to obtain feedstock from agricultural areas, not competing for food or feed

or grown solely for energy purposes and harvested sustainably (Brehmer, 2008). In this respect,

organic waste streams are the most suitable raw materials for the generation of industrial bulk

chemicals.

8

Organic waste streams

Organic wastes are cheap, sometimes available at a gate fee. Unlike other biomass

resources, massive amounts of bio-waste is unconditionally produced on a daily basis, and it is

regionally available. Considering the increasing demand towards biomass and food security

requirements, and the lack of biomass in some regions, organic waste streams or bio-wastes are

excellent feedstock for fermentative production of carboxylic acids.

Organic fraction of municipal solid waste in Europe is 90 million tons each year (Siebert,

2013). Around 30% of the bio-waste in Europe is recycled for composting and the rest is disposed

on landfills. A small fraction of industrial organic waste and organic fraction of municipal solid

waste is treated by anaerobic digestion.

Food-related (together with kitchen) wastes represent around 64 % of bio-waste collected

daily. Food waste, as it has little fraction of lignocellulose content, can be very suitable for

fermentation (Zwart, 2013). Food wastes contain however a diverse composition of complex

polymers carbohydrates, proteins, lipids, fats and lignin. This, together with daily and seasonal

variations in the composition can be reflected as instability in the product spectrum. In this thesis,

food wastes collected from specific industrial activities was used as substrate, being rich in one

type of polymer and having a stable and constant composition throughout the year.

Another major challenge of using waste material is the high water content. The water

content of food and kitchen wastes reaches up to 80 %. This leads to a considerable amount of

energy needed for heating up the feed stream to the fermenter and concentrate the products.

9

1.5. Technological limitations of using organic waste

Having feedstock with a complete sprectrum of polymers and solid particles makes it

necessary to use a diverse microbial community to achieve complete conversion of all polymers.

The mixed microbial communities survive in symbiotic relationship and can deal with complex

molecules separately by degrading them to harvest energy. Due to the complexity of mixed culture

environment, unfortunately it is difficult to predict the product types and concentrations in mixed

culture fermentation. The most expected reactions and end products from a fermentation of

glucose by mixed culture is presented in Figure 1.2.

The major problem is that the diversified composition of organic waste leads to a

combination of fermentation products in various amounts and in diluted concentrations. This

eventually turns into applying energy-intensive and costly downstream processing to concentrate

and purify the carboxylates. Thus, rather than obtaining a maximal total concentration of mixed

Box 1. Organic waste potential for short chain carboxylate demand

Around 90 million tons of municipal solid waste collected every year in Europe is organic1. Organic

waste collected from retail, food services and manufacturing contributes to 54 million tons2. This

adds up to 144 million tons of fermentable waste available every year in European countries.

Usually, conversion efficiency of mixed food waste to carboxylates is reported in literature around

60% depending on protein and lipid concentration and availability of lignocelluloses types of

polymers. If we assume 60% of the collected waste in Europe being converted to mixed

carboxylates by biological fermentation this amounts to 86 million tons carboxylates. The dry

matter of this waste type is often too high for continuous stirred fermentation, thus it should be

diluted with industrial wastewater or sewage. The total acetate, n-butyrate and propionate

demand of the world is around 4.3 million tons (see Table 1.1). Around 5 % of the organic waste

from Europe is sufficient to fulfill this demand with large amounts mixed short chain carboxylates.

1European Compost Network, 2013 2European Commission DG Health & Consumers, 2010

Box 1. Organic waste potential for short chain carboxylate demand

Around 90 million tons of municipal solid waste collected every year in Europe is organic1. Organic

waste collected from retail, food services and manufacturing contributes to 54 million tons2. This

adds up to 144 million tons of fermentable waste available every year in European countries.

Usually, conversion efficiency of mixed food waste to carboxylates is reported in literature around

60% depending on protein and lipid concentration and availability of lignocelluloses types of

polymers. If we assume 60% of the collected waste in Europe being converted to mixed

carboxylates by biological fermentation this amounts to 86 million tons carboxylates. The dry

matter of this waste type is often too high for continuous stirred fermentation, thus it should be

diluted with industrial wastewater or sewage. The total acetate, n-butyrate and propionate

demand of the world is around 4.3 million tons (see Table 1.1). Around 5 % of the organic waste

from Europe is sufficient to fulfill this demand with large amounts mixed short chain carboxylates.

1European Compost Network, 2013 2European Commission DG Health & Consumers, 2010

10

products, the fermentation process needs to be tuned in such a way that a single product can be

generated in a maximal concentration. In this thesis this ‘tuning’ will be addressed by investigating

the impact of waste composition, concentration in the fermenter and the partial pressures and

types of headspace gases on the product concentration and spectrum.

At certain levels carboxylates in the fermenter can have inhibitory impact on the organisms.

The toxic effect of products, the competition for carbon and electron sources and the continuous

intermediate conversion reactions inside and outside the cell lowers the final product concentration

(Buque-Taboada et al., 2006). Integrating separation technology with the fermentation, the overall

process can be intensified. This approach is called in-situ product recovery (ISPR) and provides a

substantial driving force for biological fermentation process since it shifts the reaction equilibrium

to the product side by continuously reducing the product concentration in the fermenter.

Removing carboxylates from the fermentation broth can be performed by several

separation technologies: crystallization, evaporation, extraction, adsorption and electrodialysis

(ED). In this thesis, ED was integrated in-situ to the mixed culture fermentation to assess the

impact on carboxylate productivity. On top of this in-situ recovery is also expected to selectively

enrich a single compound due to differences in concentration between the different carboxylates.

Finally, ISPR, also allows to use higher organic loadings because of the improved substrate

conversion efficiency.

11

Figure 1.2. Thermodynamically feasible reactions and end products from a glucose fermentation bymixed culture (Temudo et al., 2007)

1.6. Objectives of the thesis

The aim of this thesis is to gain more insight on how fermentation processes can be

steered towards production of a single compound or group of compounds at the highest possible

yield. Manipulating the operational conditions during mixed culture fermentation process and

and/or integration of separation technology in the fermentation reactors are considered to be the

key parameters to drive the fermentation reactions towards the targeted products. The

technological challenges set for this thesis were;

- To determine to what extent the composition, constitutes and concentration of organic

waste impact the fermentation product type and yield.

12

- To investigate the impact of headspace composition on the formation of fermentation

products.

- To develop and study a system with in-situ product recovery to increase productivities

and product yields of fermentation process.

1.7. Thesis outline

Chapter 2 reviews the mixed culture fermentations studies dealing with conversion of

organic substrates towards carboxylates in batch, fed-batch and continuous reactor configurations.

In mixed culture fermentation, the concentration and the speciation of carboxylates can vary

considerably, depending on the substrate type, inocula, headspace composition, process

parameters like hydraulic and solid retention time, organic load and pH, the substrate

concentrations. Thus it is crucial to know how to use these key elements to obtain a desired

product or mixture of products in the highest concentration(s) possible. In Chapter 3, the aim

was to observe to what extent the operational parameters like headspace composition in elevated

partial pressures, together with organic waste type can impact the fermentation process, product

type and yield in batch-fermentation. In Chapter 4, the study on headspace composition was

further investigated at increasing substrate concentration in batch reactor configuration. The final

target was to define operational conditions from the outcome of batch experiments in order to a

perform continuous process with an organic waste as feed material and a mixed bacterial culture

for stable production of the targeted carboxylate. In Chapter 5, the process parameters were

overviewed for the improvement of carboxylate formation and spectrum with integrated in-situ

recovery technology. Here, a novel ‘direct contact’ ED configuration with bipolar membranes was

integrated with the continuous, mixed culture fermentation. Additionally, it was successfully

demonstrated that the pH of the continuous reactor was self-regulated by the use of bipolar

13

membranes and external base addition was eliminated completely during the operation. Finally, in

Chapter 6 carboxylate production from organic waste with mixed culture fermentation process

was evaluated according to the results obtained in this thesis.

14

References

Bos-brouwers, H., Langelaan, B., Sanders, J., van Dijk, M., van Vuurer, A., (2012). Chances for biomass Integrated valorisation of biomass resources.Wageningen University, Project number: BO-08-016-001.

Brehmer, B. (2008). Chemical Biorefinery Perspectives The valorisation of functionalised chemicals from biomass resources compared to the conventional fossil fuel production route.Ph.D. thesis, Wageningen University.

Buque-Taboada, E., Straathof, A., Heijnen, J., & van der Wielen, L. (2006). In situ product recovery (ISPR) by crystallization: basic principles, design, and potential applications in whole-cell biocatalysis. Applied Microbiology and Biotechnology, 71(1), 1–12.

Cherubini, F., Jungmeier, G., Wellisch, M., Willke, T., Skiadas, I., van Ree, R. and de Jong, E. (2009). Toward a common classification approach for biorefinery systems. Biofuels Bioproducts Biorefinery 3(5):534-546.

Danner, H.R.B., & Braun, R. (1999). Biotechnology for the production of commodity chemicals from biomass. Chemical Society Reviews, 28(6), 395–405.

ECN, (2013). http://www.compostnetwork.info/

Head, I.M., (2011). Crystal Ball - Energy Climate in Environmental Biotechnology.Microbial Technology, 4, 109-137.

Holtzapple, M., & Granda, C. (2009). Carboxylate Platform: The MixAlco Process Part 1: Comparison of Three Biomass Conversion Platforms. Applied Biochemistry and Biotechnology.,156, 95-106.

Hughes, L. (2000). Biological consequences of global. Trends in Ecology and Evolution, 15(2), 56–61.

ICIS, (2007). Chemical Profile Databank,Propionic acid. www.icis.com

IEA, Bioenery, (2012). Biobased chemicals Value Added Products From Biorefineries. Task, 42, Biorefinery Report.

IPCC, (2013). The Physical Science Basis Working Group I Contribution to the Fifth Assessment Report of theIntergovernmental Panel on Climate Change. Stocker, T.T., Qin, D., Platter, G.K., Neauels, A., Tigno, M.M.B., Xia, Y., Allen, S.K., Bex, V., Boschung. J., Midgley, P.M. Cambridge, United Kingdom.

KBBPPS, (2013). Knowledge Based Bio-Based Product Pre-Standardization. Stakeholder workshop report.

Koops, A.,J., Brumbley, S., Poirier, Y., de Laat, A., Scott, E., Sanders J. P. M., & van der Meer I. M. (2012). The biobased economy biofuels, materials and chemicals in the post-oil era.Plant Production of Chemical Building Blocks, p 131-146.

Long, H., Li, X., Wang, H., & Jia, J. (2013). Biomass resources and their bioenergy potential estimation: A review. Renewable and Sustainable Energy Reviews, 26, 344–352.

Ostergard H., Markussen M.V. and Jensen, E.S., (2012). The biobased economy biofuels, materials and chemicals in the post-oil era: Challenges for Sustainable Development, p 33 - 49.

Patel.M., Dornbur, V., Hermann, B., Roes., L. (2006). The BREW Project: Medium and long term opportunities and risks of biological production of bulk chemicals from renewable resources -the potential for white biotechnology. Utrecht, Final report Prepared under the European Commission’s GROWTH Programme.

15

Rogers, P., Chen, J.S.,Zidwick, M.J. (2006). The prokaryotes: Organic Acid and Solvent Production: Acetic, Lactic, Gluconic, Succinic, Polyhydroxyalkanoic Acids Propionic Butyric acid and aLcohols. Applied Bacteriology and Biotechnology,

Siebert, S.,(2013).Biowaste: Key role in waste management, resource efficiency and bioeconomy. EEB Annual Conference, Brussels.

Straathof, A.J.J., (2014). Transformation of Biomass into Commodity Chemicals Using Enzmes or Cells. Chemical REviews, 114(3), 1871-1908.

TechnoOrbiChem, (2013). Acetic acid fact data sheet. www.orbichem.com.

Temudo, M.F., Kleerebezem, R., & van Loosdrecht, M. (2007). Influence of the pH on (open) mixed culture fermentation of glucose: A chemostat study. Biotechnology and Bioengineering, 98(1), 69–79.

Tilman, D., Socolow, R., Foley, J. A., Hill, J., Larson, E., Lynd, L., Williams, R. (2009). Beneficial Biofuels The Food, Energy, and Environment Trilemma. Science, 325(5938), 270–271.

Xu, Z., Juang, L. (2011). Bio-based chemicals-Butyric acid. p 207-215.

Yang, S.T., (2007). Bioprocessing Value Added Products from Renewable Resources New Technologies and Applications. Extractive Fermentation for the Production of Carboxylic Acids, p 420-447. 1st edition, Elsevier, Amsterdam, The Netherlands.

Zhu, Y. (2003). Enhanced Butyric Acid Fermentation By Clostridium Tyrobutyricum Immobilized In A Fibrous-Bed Bioreactor. Ph.D. thesis, Department of Chemical Engineering, The Ohio State University.

Zwart, K., (2013). Pyrolysis in the Countries of the North Sea Region Potentially available quantities of biomass waste for biochar production. A publication of the Interreg IVB project Biochar. www.biochar-interreg4b.eu.

Wisbiorefine, (2013). Energy Center of Wisconsin. http://www.wisbiorefine.org.

World Economic forum, (2014). Future Industrial Biorefineries. King, D., Hagan, A.

16

I’ve been looking for truth at the cost of living, I’ve been afraid of what’s before mine eyes, every answer I found begs another question, the further you go, the less you know, the less I know.

Five-O, James

17

Chapter 2

Selective short chain carboxylates production: control mechanisms to direct mixed culture fermentations

This chapter has been submitted:

D. Arslan, K.J.J. Steinbusch, L. Diels, H.V.M. Hamelers, C.J.N. Buisman, H. De Wever (2014). Selective short chain carboxylates production: a review on control mechanisms to direct mixed culture fermentations

18

Abstract

Anaerobic digestion of organic residual streams can be directed to produce carboxylates such as acetate,

propionate and n-butyrate, which can be either directly used in industry or converted into other valuable

compounds. This paper reviews the studies working with mixed culture conversion of organic substrates

towards carboxylates. It draws connections between standard fermentation parameters and the carboxylate

product concentrations and composition. The use of more concentrated carbohydrate-rich substrates, at

longer residence times and at neutral pH ranges stimulates total acid production. When increasing pH to the

neutral range, acetate and propionate fractions are elevated. High propionate concentrations and fractions

are infrequently reported and mainly appear on high protein containing feedstock. High n-butyrate fraction

> 70% is usually found when pH < 6, at longer retention times or organic loading rates, under CO2

atmosphere or on substrates with high lactate concentrations. The review concludes with prospects for

further developments related to the carboxylate platform.

19

2.1. Introduction

Short chain carboxylates and their derivatives can be used as building blocks to produce

bulk industrial chemicals. Those carboxylates or volatile fatty acids (VFAs) such as acetate,

propionate and n-butyrate can be directly used in a wide variety of industrial applications such as

fertilizers, plastics, man-made fibers, flocculants, and detergents (Danner & Braun, 1999).

Carboxylates can also be converted to more reduced and also higher energy density chemicals.

The use of carboxylates for different applications represents a carboxylate platform, which was

first mentioned by Holtzapple & Granda (2009). In recent years the carboxylate platform received

more interest from different researchers (Kleerebezem & van Loosdrecht, 2007; Li & Yu, 2011).

Agler et al., (2011) gave an overview of pathways for carboxylate conversion. Examples are

biohydrogenation to alcohols (Steinbusch et al., 2008), chain elongation to longer chain fatty acids

with higher carbon/oxygen ratios (Agler et al., 2012; Grootscholten, et al., 2012), esterification to

polymers (Peralta-Yahya et al, 2012), or transformation into polyhydroxyalkanoates, used in the

production of bioplastics (Lemos et al., 2006). In most of these processes, the carboxylate

concentration and composition has significant importance.

Carboxylates are produced as intermediates in anaerobic digestion, in which four phases

can be distinguished (Figure 2.1) (Metcalf & Eddy, 2003; De Mes, 2003).

1. In the hydrolysis phase, complex organic material composed of different polymers such as

carbohydrates, proteins and lipids is broken down and solubilized into monomers such as

sugars, amino acids, long chain fatty acids and glycerol. It is a combination of an

extracellular biological process allowing hydrolysis of polymers catalysed by secreted

enzymes, and a non-biological process, allowing lysis of cell structures and disintegration of

particulates.

20

2. In the acidogenesis phase, these monomers are converted to a mixture of carboxylates,

alcohols, H2 and CO2. Sugars are converted by different organisms to carboxylates like

acetate, propionate and n-butyrate but also ethanol and lactate, while proteins are

converted to a broader range of carboxylates varying from 2 to 6 carbon atoms, isomeric or

saturated. During acidogenesis also inorganic compounds are produced like CO2, H2, NH3

and reduced sulfur.

3. Acetogenesis consists of further conversion of acidogenesis products into acetate.

Conversion of higher organic acids like propionate, n-butyrate and valerate to acetate is an

oxidation step that requires an additional electron donor like protons and CO2. The main

thermodynamic reactions for carboxylate oxidizing organism are listed in Table 2.1. The

standard Gibbs free energy of the oxidation reactions is close to zero and requires mainly

low hydrogen partial pressure to make the reaction exergonic. In case of a high hydrogen

partial pressure the acetogenic reactions are reversed to reduction reactions.

4. Methanogenesis is the last step in anaerobic digestion, but has to be avoided to obtain high

levels of carboxylates.

21

Table 2.1. Thermodynamics of reactions for fatty acid oxidizing organisms under acidifying conditions

Product Reaction Equation Gr0 *

kJ mol-1Gr

’ **kJ mol-1

H2, HCO3- OH3CHHCOH4 24

-32 -175 -46.5

Propionate -32

-2322

-253 HCOH3OHCOH3OHC 188 -1.0

n-Butyrate2

-2322

-274 H3OHC2OHOHC 89 -4.0

Caprylate224222168 H6OHC4O6HOHC 266 -14.6

* r0) is calculated at 273 K with concentrations and pressures of all components

1 M or 1bar; r’) is calculated at pH 6, 303 K and carboxylate concentrations of 0.01M, H2 pressure of

10 Pa, CH4 pressure of 1000 Pa and HCO3- concentration of 0.01M

Unfortunately, mixed culture conversion creates mixtures and variations in carboxylate

composition which may not be desirable for precise industrial applications. One of the biggest

challenges of mixed culture fermentation is to control the acidification and acetogenic reactions

occurring during the fermentation and direct these reaction flows towards a specific end product at

a high conversion rate. This paper compiles the standard operational parameters for mixed culture

fermentation processes and correlates their use for efficient production of a single desired

carboxylate or composition of carboxylates. Operational parameters include the type of substrate

and substrate concentration, hydraulic retention time (HRT), organic loading rate (OLR), pH,

temperature, headspace composition and inocula.

22

Figure 2.1. Schematic representation of the AD process (Adapted from Metcalf and Eddy, 2003; Reith et al., 2003)

The assessment is evidently complicated by the fact that several parameters interact with

each other. OLR is for instance calculated from the substrate concentration and HRT but we

considered it important to also assess these parameters individually. To be able to make a good

comparison between the researches, all studies presented in this review used continuous stirred-

tank reactor type, so sludge retention time (SRT) is assumed to be equal to HRT. Reactor designs

for which SRT is independent of HRT e.g. up-flow anaerobic filter reactors, were not included in

this review but the effect of SRT will be briefly discussed.

Earlier studies have reported significant increases in both hydrogen yields and hydrogen

production rates in mixed cultures with improved process control (De Gioannis et al., 2013; Wang

23

& Wan, 2009). Carboxylate production presents additional challenges because the products are

soluble and there is no automatic phase separation. Furthermore, the range of soluble products is

much broader than that of gaseous products. For this review we selected mixed culture studies

that provide not only the fermentation conditions but also details on carboxylate concentrations

and product spectrum and their evolution during the tests. The majority of studies use organic

waste materials which are abundant and cheaper than pure substrates. One third of the

referenced studies worked with pure carbohydrate substrates, avoiding interference of substrate

variability, and thus giving a more clear indication about the impact of an operational parameter

on acid production. The selected studies were grouped and evaluated per operation mode: batch

reactors in Table 2.2 (25 studies), fed-batch in Table 2.3 (6 studies) and continuous reactors in

Table 2.4 (22 studies). In addition to the studies compiled in the tables, others have been

referenced in the text that either only discussed total carboxylate concentrations or only

emphasized certain aspects of the product spectrum without giving a comprehensive overview.

2.2. Effect of substrate type and pre-treatment

The type of substrate is one of the most important factors that affects the total product

concentrations and also has an influence on the product type. Substrates can be grouped into

three categories according to their polymer composition: rich in carbohydrate, protein or lipid,

each with different hydrolysis rates. In general it can be said that the higher the hydrolysis rate,

the higher the carboxylates concentrations. Direct comparison of these three types of substrates

has only been attempted in batch studies (Table 2.2). Dong et al. (2009) prepared synthetic

wastes from various food materials, and obtained the highest total acid production on the

carbohydrate-rich substrates which have also the highest hydrolysis constant. Similarly, on real

organic waste streams from food industry the highest total acid production was obtained on a

starch-based substrate (Arslan et al., 2012).

24

Summarizing all reviewed studies, carboxylate levels were found ranging between 2 and 20

g/L (Figure 2.2) at yields varying between 0.1 and 0.6 g total carboxylate/g COD feed added

(Figure 2.3). Concentrations higher than 10 g/L were only reported occasionally. In batch studies

(Table 2.2), they were obtained at yields of 0.1 to 0.7 g carboxylates/g COD feed on

carbohydrate-rich food waste (Dong et al., 2009; Kim et al., 2009; Kim et al., 2011a; Kim et al.,

2011b) In continuous studies (Table 2.4), they were found at yields of 0.55 g carboxylates/g COD

feed on sucrose (Kyazze et al., 2005) and of 0.45 on glucose (Hafez et al., 2010), and 0.26 to 0.47

g total acids/g VS sugarcane bagasse fed (Fu & Holtzapple, 2011). For single protein- and lipid-

based streams, the total carboxylate concentrations are on average below 3 g/L at yields below

0.1. The only exceptions are two batch studies where either much longer incubation times were

applied for the protein-rich substrates than for the others (Okamoto et al., 2000) or where acetic

acid was produced through homoacetogenesis from H2 and CO2 in the headspace (Arslan et al.,

2012).

Figure 2.2. Total carboxylate levels as a function of substrate concentration. Only those studies for which substrate concentrations could be calculated into g COD/L are represented.

0,0

0,5

1,0

1,5

2,0

2,5

0 50 100 150 200 250Tota

l car

boxy

late

pro

duct

ivity

(g/L

.h)

OLR ( g COD/L.d)

25

Figure 2.3. Total carboxylate productivity as a function of OLR. OLR data > 100 g COD/L.d are from a small number of studies. Only those studies for which OLR could be calculated into g COD/L.d are represented.

Food wastes generally contain different types of polymers and their composition can vary in

time. Elevated carboxylate levels > 10 g/L are usually found when the food has been cooked,

which accelerated the hydrolysis step (Kim et al., 2009; Kim et al., 2011a in Table 2.2, Lim et al.,

2008 in Table 2.3, Ueno et al., 2007 in Table 2.4). Mixing different types of wastes can increase

also total acid concentration and the efficiency of the conversion. Increasing the contribution of

primary sludge to food waste from 10 to 25%, dramatically increased the total carboxylate level at

24 h retention time in the study of Min et al. (2005) (Table 2.4). The combination of food waste

and sewage sludge (Kim et al., 2011b in Table 2.2) and corn stover and manure (Thanakoses et

al., 2003 in Table 2.4) produced higher carboxylate levels of 20 g/L compared to single substrate

fermentation. A mixture of sugarcane bagasse and manure even resulted in 28 to 40 g

carboxylate/L (Fu & Holtzapple, 2011 in Table 2.4), the highest levels identified in this review. The

most likely reason is that the in-situ air-lime pre-treatment of sugarcane bagasse provided enough

alkalinity to avoid system failure at high acid concentrations. Next to a more balanced pH, mixing

0

5

10

15

20

25

0 20 40 60 80 100 120

Tota

l car

boxy

late

(g/L

)

Substrate concentration (g COD/L)

26

different streams can lead to a more balanced nutrient composition or can increase the inoculum

concentration or variation.

There is clear impact of substrate type on the product type, though it is difficult to make

the link between a specific type of feed and a product. In studies with total carboxylate

productions > 1 g/L on carbohydrate substrates, n-butyrate fractions surpassed 60% (Okamoto et

al., 2000 in Table 2.2), (Ueno et al., 1996; Fang & Liu, 2002; Hussy et al., 2003; Hussy et al.,

2005; Kyazze et al., 2005; Yu & Mu, 2006; Temudo et al., 2007; Arslan et al., 2013 in Table 2.4)

and even 70% (Hussy et al., 2003; Fang et al., 2006; Lee et al., 2008a; Kim et al., 2009; Kim et

al., 2011a in Table 2.2), (Lin & Jo, 2003 in Table 2.3), (Horiuchi et al., 2002; Liu & Fang, 2002 in

Table 2.4), usually at pH < 5 and at higher organic loading rates. At higher pH levels > 6 (Oh et

al., 2003 in Table 2.2) (Fang & Liu, 2002; Horiuchi et al., 2002 in Table 2.4) or during prolonged

periods of batch incubations (Arslan et al., 2012 in Table 2.2) or at lower organic load (Hafez et

al., 2010 in Table 2.4), shifts may occur towards high fractions of acetate, again going up to levels

higher than 70%. Propionate proportions with carbohydrate substrates are only found occasionally

up to 30% (Feng et al., 2009; Khanal et al., 2004 in Table 2.2) (Lim et al., 2008 in Table 2.3) (Min

et al., 2005 in Table 2.4). In summary, mixed waste streams may yield one major carboxylate

product, which is usually acetate or n-butyrate, and only infrequently propionate. It is, however,

still difficult to predict the exact product distribution according to the substrate type considering

the fact that even two similar carbohydrate-rich waste streams can produce different product

types. For example, Hussy et al. (2005) (Table 2.4) showed under similar process conditions that

sucrose yielded 67% n-butyrate while sugar beet extract dosed at the same loading rate yielded

equal fractions of acetate and n-butyrate.

Protein-rich waste streams seem to enhance the production of odd numbered and/or

longer carboxylates like i- and n-valerate and n-caproate (Okamoto et al., 2000; Chen et al., 2007;

27

Feng et al., 2009; Morgan-Sagastume et al., 2011; Yuan et al., 2011) Primary or activated sludge

containing proteins yielded up to 78% of propionate (Min et al., 2005; Zhu et al., 2008). Waste

activated sludge yielded up to 10% n-valerate (Yuan et al., 2011) or even 35% of i-valerate in

batch tests (Chen et al., 2007). Gelatin enriched wastewater resulted in around 10% i- and n-

valerate and 10% n-caproate (Yu & Fang, 2003 in Table 2.4). Morgan-Sagastume et al. (2011)

found isomers of n-butyrate and n-valerate fractions up to 20 – 30% with amino acids rich

substrate (Table 2.3). For lipid waste streams, there are too few studies to draw conclusions.

Substrates with high lactate concentrations promote n-butyrate production (Agler et al.,

2011). Komemoto et al. (2009) indeed detected up to 80% of n-butyrate with food waste

containing 30 g/L lactate (Table 2.2), even when the pH was controlled at 7 which is unfavourable

for n-butyrate production (see section 8). Similarly, n-butyrate was the major product in batch

reactors where lactate was consumed at pH > 6 (Lee et al., 2008a). Kim et al. (2012b)

investigated the effect of lactate addition at pH 4.5. In the tests where lactic acid was added and

consumed, n-butyrate concentration was higher comparing to the tests where lactic acid was not

added or consumed.

Medium chain fatty acids (MCFAs) such as n-caproate as an end product of a mixed culture

fermentation can be advantageous over short chain acids. Because MCFAs have a longer

hydrocarbon tail, they are less soluble in water which makes their separation easier than

carboxylates (Steinbusch et al., 2011). Although high n-caproate production next to short chain

acids is not commonly reported in mixed culture fermentation, major fractions of n-caproate up to

64% were found in one study on sucrose-rich wastewater, at the expense of n-butyrate (Yu & Mu,

2006). Because not all studies measured valerate or MCFAs, it is difficult to draw conclusions from

the studies reviewed here.

28

2.3. Effect of inocula pre-treatment

Biomass from acidogenic reactors, anaerobic digesters and composting systems is rich in

acid producing organisms together with methane producers. Thus, inocula may need a pre-

treatment to avoid methanogenic activity unless at least one operational parameter is selected in a

range where methanogens do not naturally assimilate. Although pre-treatment methods are not

cost efficient and technically difficult to implement in industry, they help to increase the total acid

concentration if there is a risk of methanogenic consumption of the acids. The most common 3

pre-treatment methods are (1) heat shock treatment, (2) acid/alkali conditioning and (3) chemical

inhibitors.

Among the pre-treatment methods, heat treatment is the most widely used one in

literature. The principle is based on inactivation of non-spore forming micro-organisms like

methanogens while maintaining spore forming acid producers. In general, heat treatment is

achieved by boiling the sludge for 15 min (Han & Shin, 2004) or keeping the sludge at 75 °C

(Chang et al., 2002) to 100 °C (Wang et al., 2008) for 1 to 7 hours (Tao et al., 2007). Half of the

batch studies, represented in Table 2.2, used heat treated inocula. For the fed-batch and

continuous experiments in Table 2.3 and 2.4, only 6 studies out of 25 used heat treatment as a

pre-treatment method. The reason might be that pre-treatment method may only have a short

term effect, as demonstrated in long term continuous experiments (Luo et al., 2010), Also, in

continuous systems, other parameters such as HRT can be used to avoid methane production.

Heat treatment can selectively eliminate non-spore formers like methanogens but propionate

producers as well. So in general, heat treatment promotes selective production of acetate and n-

butyrate (Hawkes et al., 2002; Hussy et al., 2003; Kyazze et al., 2005, Kyazze et al., 2008). Cohen

et al. (1985) also indicated that sporulation results in n-butyrate type product. Conversely, a study

by Wang et al. (2011) did not observe any changes in carboxylate distribution when the inoculum

29

was heat treated at 80 °C and 90 °C. The carboxylate distribution was similar to the control

reactor which was not heat treated. In all reactors, acetate was around 70% and n-butyrate was

20%. When the temperature of pre-treatment was increased to 100 °C, n-butyrate disappeared

and only acetate was detected.

In acid/alkali treatment, the inoculum is kept for a certain period under acidic (pH 2 to 4)

or basic (pH 9 to 11) conditions. Acid and alkali treatment mainly inhibits methanogenic activity

but there is not a clear effect on the carboxylate distribution. According to the study of Ren et al.

(2008) one could say that less propionate was found when the inoculum was pre-treated with

alkali solution. Ren et al. (2008) observed the highest carboxylate production of 3.5 g/L with

alkaline pre-treatment. Acetate and n-butyrate were found to be the dominant products reaching

each around 50% in alkali pre-treated samples while propionate was observed only in acid pre-

treatment. Wang et al. (2011) reported acetate as main product in acid pre-treatment, while the

secondary product varied. When the same inoculum was alkali pre-treated, n-butyrate was the

main secondary product. Wang & Wan (2008) also observed selective production of acetate (60%

– 70%) after alkali and heat treatment, albeit in low concentration ranges. When inoculum was

acid treated, n-butyrate, acetate and propionate were detected in equal proportions (Wang &

Wan, 2008)

Using chemical inhibitors such as 2-bromoethanesulfonate (BES), iodoform, chloroform,

ethylene, ethane, methyl chloride or acetylene is another option for inhibition of methanogenesis.

It is, however, not widely applied because these chemicals are expensive and toxic to the

environment. The mechanism of these inhibitors is based on blocking the functions of specific

enzymes of methanogens (Valdez-Vazquez & Poggi-Varaldo, 2009). With respect to acid formation,

BES was not found effective for regulation of fermentation pathways, while chloroform inhibited

acetate formation by homoacetogens, and increased production of propionate and n-butyrate

30

(Conrad & Klose, 2000). On the contrary, high selectivity together with high concentrations of

acetate was obtained with a chloroform treated inoculum by Wang & Wan (2008). Mohan et al.

(2008) found that substrate utilization and total carboxylate concentration after 12h was the

highest in batch vials pre-treated with BES combined with acid (at pH 3) or heat-treatment (100

°C for 1 hour) among the variable single and combined pre-treatment methods evaluated. Pham

et al. (2012) reported that the addition of 70 ppm iodoform lead to an undesired decreased

carboxylate productivity. This is in line with the findings of Chan & Holtzapple (2003). They

reported that iodoform provided a fourfold decrease in methane production, but also reduced

acetate production.

2.4. Effect of gas composition

Hydrogen and carbon dioxide are involved in many fermentation reactions in terms of both

production and consumption. Their partial pressures and relative ratio in the headspace affect in

many cases the carboxylate distribution. Thermodynamically, H2 and CO2 partial pressures in the

headspace can affect the direction of acetogenic reactions (listed in Table 2.1). Literature studies

on this topic cover two approaches (1) removal of produced gases by sparging or flushing with N2,

or (2) the replacement of the headspace by applying external H2 and/or CO2.

Sparging and flushing with N2 is used to remove H2 from the liquid and headspace to avoid

that H2 becomes inhibitory for certain fermentation pathways (Angenent et al., 2004). The studies

in this review (Table 2.4), that addressed this aspect, show that the effect of sparging on total

carboxylate production is negligible (Hussy et al., 2003; Kim, et al., 2006) or even negative

(Kyazze et al., 2005; Massanet-Nicolau et al., 2010). Neither sparging nor flushing with N2 was

found to affect the carboxylate composition. Intermediate sparging with N2 negatively affected

reactor performance in a study by Kyazze et al. (2005), decreasing total carboxylate concentration

31

from 17 to 15 or from 14 to 12 g/L (Table 2.4). It should be noted that all tests were performed at

pH 5.3 (or lower) where 20% of the carboxylic acids is in the undissociated form, and gas sparging

may lead to stripping of carboxylate. Furthermore, severe sparging was reported to create stress

on the biomass leading to decreased acid production levels (Bastidas-Oyanedel et al., 2012). To

avoid carboxylate stripping and such problems in biomass activity, flushing of the headspace can

be a better option. Even then, the equilibrium between headspace and liquid phase is disturbed

and losses might occur.

Addition of extra CO2 to the headspace increased the n-butyrate fraction relative to the

other fermentation products (Kim et al., 2006; Arslan et al., 2013). A possible explanation is that

CO2 decreases the pH of the system down to around 4.8 leading to pH ranges favourable for n-

butyrate production. Another possible explanation is that CO2 causes damage on the cell

membrane leading to functioning problems in especially acetogenic and lactic acid bacteria (Kim et

al., 2006). One important remark is that the impact of CO2 in the headspace on selective n-

butyrate formation was less when substrate concentration was increased (Arslan et al., 2013).

No obvious trends can be derived for the impact of elevated H2 partial pressure on the

short chain carboxylate levels and composition. In some studies, a high partial pressure of

hydrogen favoured the production of longer chain carboxylates and more reduced compounds

(Smith & McCarty, 1989; Arslan et al., 2012 in Table 2.2) (Yu & Mu, 2006 in Table 2.4). This can

be explained based on thermodynamics. At high hydrogen partial pressure (> 102 Pa), the

oxidation of propionate and n-butyrate or even longer n-caproate to acetate is limited. Ding et al.

(2010) reported that high hydrogen partial pressures are not necessarily needed for the n-

caproate production pathway,

32

Changing the headspace composition by applying proper portions of H2 and CO2 instead of

N2 increases the acid concentration and also regulates the product spectrum. When H2/CO2

(50/50) was applied up to 2 bar, acetate fraction reached 87% in mixed culture fermentation

(Arslan et al., 2012 in Table 2.2), mainly as a result of homoacetogenesis. Selectivity up to 87% is

promising when a mixed culture is used, but acetate production was achieved mainly from H2 and

CO2 consumption instead of waste reduction. Thus, this system can only be interesting if these

gases are available as wastes.

2.5. Effect of initial substrate concentration

In principle, product concentration increases with the substrate concentration in the

reactor, but at feed levels above 40 g COD/L a further increase of carboxylate concentration was

limited due to overloading or to inhibition (Figure 2.2). This general principle was indeed observed

in most studies that systematically investigated this effect on one substrate (Fang et al., 2006;

Arslan et al., 2013; Lee et al., 2008a in Table 2.2), (Lim et al., 2008; Gomez et al., 2009 in Table

2.3) (Liu & Fang, 2002; Thanakoses et al., 2003; Kyazze et al., 2005; Yu & Mu, 2006; Ueno et al.,

2007; Fu & Holtzapple, 2011 in Table 2.4). Only in the work of Coats et al. (2011) (Table 2.2) and

Kim et al. (2012a) (Table 2.3) various substrate levels above 35 g VS/L or 25 g COD/L

respectively, resulted in comparable carboxylate concentrations. In two other studies (Table 2.4),

the total carboxylate concentration levels off at 12 g/L at substrate concentrations > 30 g COD/L

(OLR > 90 g COD/L.d) (Hafez et al., 2010) or at 20 g/L at substrate concentrations > 40 g VS/L

(OLR > 5 g VS/L.d) (Thanakoses et al., 2003). Within the referenced studies, when those

continuous investigations are grouped that can be expressed in g COD/L, the substrate

concentrations applied vary between 5 and 107 g COD/L, the most typical range being 5 to 40 g

COD/L. As shown in Figure 2.2, this results in carboxylate concentrations of maximal 20 g/L, which

33

do not increase further at higher feed levels above 40 g COD/L. This may point to overloading or

product inhibition (see section 10).

Carboxylate yields (expressed as g total carboxylate/g COD feed added) vary between 10

and 60%. Surprisingly, yields are sometimes as low as 15% on synthetic sugar based substrates.

Only in one tenth of the studies, 45-55% of conversion is reached (Hafez et al., 2010; Kyazze et

al., 2005 in Table 2.4). It is not clear what can explain this, because substrate type and

concentrations, product concentrations and operational parameters were similar in many other

studies. Maybe too short retention times, occurrence of methane production, and an incomplete

product spectrum due to unanalysed compounds such as valerate and n-caproate may explain the

low yields.

It can be predicted that increasing substrate concentration per time unit increases the

amount of available electrons per cell in the reactor, and an excess of electrons drives reactions

more towards reduced compounds such as ethanol, n-butyrate or n-caproate than acetate and

propionate. The increase of substrate concentration indeed shifted the product type from acetate

to n-butyrate or even n-caproate ( Hafez et al., 2010; Arslan et al., 2013 in Table 2.4). In one

batch study (Fang et al., 2006) testing a broad substrate concentration range, the increase of

substrate concentration led to more n-butyrate and not n-caproate. The explanation could be that

the test was performed at a rather low pH of 4.5 which is too low for n-caproate production.

2.6. Effect of HRT in fed-batch and continuous systems

HRT is a key parameter that can affect the fermentation process in different aspects. HRT

can determine the dominating microbial population through washout of slower growing organisms.

To avoid methanogenic growth, HRTs longer than 2 days are not preferable for fermentation

purposes (Ghosh & Pohland, 1974). From an economical aspect, a short HRT is desirable because

34

a fast conversion results in smaller reactor volumes (De Mes et al., 2003). Yet, HRTs should be

long enough to allow optimal hydrolysis and fermentation of the substrate.

In the referenced studies, HRT varied between 3 and 200 h for continuous fermentations

(Table 2.4) and between 4 and 290 h for fed-batch systems (Table 2.3). In practice, most

continuous systems were operated at a HRT between 3 and 50 h. This is mainly due to the fact

that acid production can start within 2 h of incubation, while some methanogens can already

assimilate at an HRT of 30 h (Ten Brummeler et al., 1991; De Mes et al., 2003). Hydrolysis is the

slowest step in the anaerobic digestion and thus the hydrolysis rate of the substrate determines

the HRTs needed to maximize carboxylate levels. Because glucose-rich substrates do not require

hydrolysis, retention times of a few hours would be enough for acid production (Horiuchi et al.,

2002 in Table 2.4). For more complex and non-soluble substrates, the retention time should be

higher. In some studies, a higher total carboxylate concentration was found with increasing HRT

under the same substrate concentration (Lin & Jo, 2003; Lim et al., 2008 in Table 2.3), (Liu &

Fang, 2002; Ueno et al., 1996 in Table 2.4), while in others the carboxylate concentrations did not

vary (Horiuchi et al., 2002; Yu & Mu, 2006 in Table 2.4) or even decreased (Min et al., 2005 in

Table 2.4). This is mainly due to the fact that these studies use different types of substrates which

imply different kinetics. Thus, according to the substrate type, the limiting HRT level varies.

By regulating retention time specific microbial populations can be selectively favoured. In

single culture fermentation on glucose, acetate producers have a doubling time around 2.5 h

(Drake et al., 2006) n-butyrate producing ones around 10 h (Michel-Savin et al., 1990) propionate

producers around 7 h (Lewis & Yang, 1992) and lactate producers 3 h (Mercier et al., 2007). In

mixed cultures fermentation, HRT alone is not expected to be a determining factor for selective

production of a single compound, thus operational parameters other than HRT would play a role to

increase the population of certain organisms. In theory, acetate is a major product under short

35

HRT conditions because the energy yield is higher (4 ATP) by producing acetate from glucose than

by producing n-butyrate (3 ATP) (Kleerebezem & Stams, 2000). Indeed, because acetate

producing bacteria have shorter doubling times than n-butyrate producing ones, acetate may

accumulate at shorter HRTs. At long HRT conditions, acetate might still be produced in high

concentrations. However, as the retention time gets longer, the product type shifts from more

oxidized compounds such as acetate or lactate towards more reduced compounds such as n-

butyrate or n-caproate (Smith & McCarty, 1989; Angenent & Kleerebezem, 2011). To give more

examples, Yu & Mu (2006) (Table 2.4) reported an increase in n-caproate fraction up to 64% at 30

h HRT whereas it dropped to half at 3 h. Liu & Fang (2002) (Table 2.4) detected n-caproate and

ethanol at a HRT longer than 14 h with sucrose-rich wastewater, whereas at a shorter HRT the

reduced products were not observed at all. Propionate and lactate production was higher at 9 h

HRT than at 15 h in a fed batch reactor on corn starch substrate (Arooj et al., 2008). In the study

of Wu et al. (2005) decreasing HRT from 6 h to 0.5 h led to decreasing n-butyrate and increasing

acetate levels at all tested substrate concentrations.

2.7. Effect of OLR in fed-batch and continuous systems

The OLR combines the effect of substrate concentration and HRT and determines the food

to microorganism ratio (F/M). The OLRs applied in the studies of this review vary from 5 to 240 g

COD/L.d.

As we concluded in the previous two chapters, the OLR can be affected both ways. When a

high OLR is achieved by increasing substrate concentration the product concentration goes up and

the type of product becomes more reduced. However, at a high OLR by decreasing HRT the

hydrolysis can become less effective, resulting in lower total carboxylate concentrations and/or

shifts towards more acetate because only fast growing bacteria such as acetate or lactate

36

producers are retained in the reactor. So multiple effects are seen in the studies that vary OLR and

this gives a rather confusing view when they are considered as one parameter.

For all studies where OLR could be calculated to a value expressed in g COD/L.d, the

productivity was plotted versus the OLR. Figure 2.3 indicates that in the most applied OLR range

of 5-50 g COD/L.d. the productivity is rather scattered but seems to increase with increasing

loading rate and amounts to 1 g carboxylate/L.h at maximum. At higher OLRs, the productivity

seems to level off towards a value of 1.5 g/L.h. The reason of limited productivity might be due to

substrate overloading. These studies were performed either with glucose or sucrose type substrate

under maximum 15 h HRTs. Thus, increasing HRT under high OLR might help to increase the

productivities due to better acidification performance.

From our evaluation of continuous studies, particularly non-controlled pH conditions or pH

< 6 lead to high n-butyrate fractions, provided that OLR is high. At increasing substrate

concentration and hence OLR, the product spectrum can shift from acetate to n-butyrate (Gomez

et al., 2009) or remain stable with n-butyrate as the major compound (Kyazze et al., 2005; Wu et

al., 2005 in Table 2.4). Though no inhibition was mentioned in these studies, Kyazze et al. (2005)

indicated instability in the system with increasing OLR.

2.8. Effect of pH

The fermentation and hydrolysis steps of acid fermentation are significantly influenced by

pH. Like temperature, pH is an operational parameter that has a direct effect on growth conditions

or the biological activity. The optimum of the parameter is different for each species or enzyme.

Enzymes which are involved in hydrolysis have an optimum pH between 5 and 7 (Veeken et al.,

2000; Jonke & Michal, 2007). Gallert & Winter (2005) indicated that hydrolysis of proteins and

lipids requires neutral or weakly alkaline pH whereas for carbohydrates a slightly acidic pH would

37

be preferable. An increase in hydrolysis degree is important to mention here since it raises the

total carboxylate concentrations. The pH also selects for the organism to be most active and

therefore the product spectrum.

The hydrolysis is optimum between pH 5 and 7, while pH close to neutral pH is favorable

for methanogenic growth. Therefore, pH interval 5 to 6 might be the best condition to obtain

highest hydrolysis and acidification together (Jonka & Michal, 2007). However, if the substrate

cannot easily be hydrolyzed and HRTs longer than 2 days are required for complete hydrolysis and

fermentation, it is recommended to control the pH below 6 or above 8 to avoid methanogenic

growth (Chen et al., 2002; Lay et al., 1999). During acidogenesis the pH naturally drops even as

low as 3.5. For every carboxylate that is produced a proton is released. There are two things that

can prevent the pH from dropping naturally: first by methanogenesis or second by the release of

ammonium in case of protein-rich waste streams. In all other cases, the pH must be controlled by

addition of alkalinity. The effect of pH was evaluated per reactor operation mode and differences

were found in product spectrum.

In batch-wise operated reactors (Table 2.2), an increase in pH decreases the relative n-

butyrate fraction with a concomitant increase in acetate and propionate fraction (Oh et al., 2003;

Shin et al., 2004; Fang et al., 2006; Chen et al., 2007). The opposite finding was obtained by Kim

et al. (2011a). The n-butyrate fraction on food waste increased with a pH increase from 5 to 9.

The highest fraction was 73% at pH 8. Lee at al. (2008a) did not observe a change in product

type with increasing pH.

In continuous mode (Table 2.4), the n-butyrate fraction is usually high when pH is lower

than 5.5 like observed by Liu & Fang (2002) and Hussy et al. (2003). The highest selective n-

butyric acid production reported was 70% at pH 4.5 (Hussy et al., 2003). An explanation of

increasing n-butyrate production can be a decrease in NADH/NAD+ ratio in the fermenter. Under

38

low pH conditions where an excess of protons is available, the NADH/NAD+ ratio increases. Thus,

NADH consuming reactions become more favourable and electrons flow towards more reduced

compounds such as n-butyrate or ethanol from glucose (Temudo et al., 2007 in Table 2.4). That is

also the reason why n-butyric acid production shifted towards n-butanol when pH was kept lower

than 4.5 as observed by Kim et al. (2004). Although selective n-butyrate production can be

favoured at a pH lower than 5.0, concentrations often remain lower due to the toxic effect of the

free acid on the organisms (see section 10). When evaluating only the studies that explicitly

investigated the effect of pH, the conclusions are less clear, probably due to the variety of

substrates tested. On glucose, the data from Horiuchi et al. (2002) and Fang & Liu (2002) show

decreasing n-butyrate fractions when pH is raised from 5 to 7, while Temudo et al. (2007) only

found n-butyrate predominance below pH 5.5. In contrast, Yu & Fang (2003) always found acetate

as predominant carboxylate in pH range 4-7, on a gelatine based synthetic wastewater.

Propionate fractions exceeding 30% are rather exceptional (Table 2.2, 2.3 and 2.4). Only

results from Horiuchi et al. (2002) and (Min et al., 2005) showed that propionate fractions reached

higher than 30% under alkali pHs. Wang et al. (2006) avoided propionate production in a

continuous system by keeping the pH at 4.5. Although propionate production can be ceased by

decreasing the pH, it seems that selective propionate production with a mixed culture is not

common. The highest propionate fraction of 80% was observed at pH 6.5 in a continuous reactor

(Min et al., 2005 in Table 2.4). In batch fermentations, the highest propionate fraction observed

was around 50% at neutral pH (Khanal et al., 2004; Fang et al., 2006). There was an attempt by

Chen et al. (2003) for selective propionate production from food waste in a batch reactor by

keeping the pH at 8. Its fraction remained around 42% with concentrations of only 2 g/L.

The effect of pH on n-caproate production was studied by Steinbusch et al. (2011). It was

observed that its production rate was higher (8.17 g/L.d) at pH 7 than at pH 5.5. Yu & Mu (2006)

39

(Table 2.4) and Grootscholten et al. (2012) also had relatively high n-caproate fractions keeping

the influent pH at 7. Increasing pH from 7.5 to 9.5, led to a n-caproate fraction decrease (Zhao &

Yu, 2008). Agler et al. (2012) showed that n-caproate production rate could be kept at around 2

g/L.d at pH 5.5 if the product was continuously extracted from the reactor.

2.9. Effect of temperature

Hydrolysis rate and most of the fermentation reaction rates increase at increasing

temperature to an optimum of 40°C for mesophilic conditions (Veeken & Hamelers, 1999). In the

studies summarized in Tables 2.2 and 2.4, the impact of temperature on total carboxylate

concentration is rather negligible in the range of 25 to 55 °C. However, Valdez-Vazquez et al.

(2005) found higher acid production at 55°C because of a better hydrolysis degree of proteins at

this temperature. Due to the limited number of systematic investigations, it is even more difficult

to detect links between product type and temperature. For instance, Shin et al. (2004) (Table 2.2)

found at 55°C a significantly higher n-butyrate fraction than at 35°C, while the opposite was true

for Valdez-Vazquez et al. (2005). Yu & Fang (2003) showed similar product spectra in a

temperature range of 20 to 55°C.

2.10. Product toxicity and iin-situ product recovery

It was mentioned in section 5 that the highest carboxylate concentrations were around 20

g/L, even at increasing substrate concentrations, which seems to point to inhibition (Figure 2.2).

Veeken et al. (200) indicated that no negative impact of increasing total and undissociated acid

concentrations up to 30 g COD/L was found on the hydrolysis of substrate at pH values of 5 - 7.

Yet, elevated acid concentrations can cause inhibition of fermentative organism functions and can

stop the further production of acids (Pratt et al., 2012). It is well know that only free acids also

called undissociated acids can pass through the cell membrane and will dissociate in the cell.

40

When bacteria consume energy to maintain ion gradients and regulate pH inside the cell instead of

using the energy for growth, acid production may finally stop (Zhang, et al., 2009).

Inhibitory concentrations of undissociated n-butyric acid were reported as 4.4 g/L for a

mixed culture at pH 4.8 (van den Heuvel et al., 1992). This is generally in line with the n-butyrate

concentrations found in batch and continuous studies in Tables 2.2 and 2.4. Acetate concentration

was found inhibitory at higher concentrations >12 g/L at pH 6.6 (Van Den Heuvel et al., 1988).

Yet, acetate production with homoacetogens in mixed culture reached 18 g/L, even at pH 5.0

(Arslan et al., 2012 in Table 2.2), or 25 to 36 g/L, albeit at pH 7 at which the acids are 88%

present in dissociated form (Fu & Holtzapple, 2011 in Table 2.4). The reported propionic acid

inhibitory levels are much lower than for acetic and n-butyric acid. Producing propionate at high

concentrations will thus be a lot more challenging than acetate or n-butyrate. Inanc et al. (1999)

indicated that propionate production stopped when it reached 1 g/L and 50% of total acids at pH

5.0. Ma et al. (2009) mentioned that propionic acid concentration needs to be kept lower than 1.5

- 2 g/L to avoid inhibition, but did not specify at which pH. It seems indeed that the general trend

is that propionate levels do not reach higher than 3 g/L (Table 2.2 and Table 2.4). From our

literature screening, we could find one study reaching 6 g/L contributing 76% of total acids

produced at pH 6.5 (Min et al., 2005 in Table 2.4) and one reaching even 8 g/L contributing 45%

at pH 5.5 (Lim et al., 2008 in Table 2.3).

In biological processes, reaching high target product concentrations can be limited by the

toxic effect of the product on the organism or its conversion into another product. Such limitations

can be overcome by continuously removing the desired products before they reach inhibitory

concentrations or are degraded in the reactor. This approach is called in-situ product recovery

(ISPR), and relies on the integration of the most suitable separation technology with the

fermentation process (Buque-Taboada et al., 2006). ISPR may then allow reaching higher

41

carboxylate yields on a substrate. So far, investigations on separation of organic acids have mainly

focused on single culture fermentations for production of e.g. lactic, citric or succinic acid and not

on mixed carboxylates. Separation of carboxylate mixtures from waste streams or selective

separation of a single carboxylate in a mixed product spectrum has hardly been studied. In fact,

only one study describes the continuous extraction of n-caproate from a mixed culture

fermentation on waste streams (Agler et al., 2012).

2.11. Outlook

The ‘carboxylate platform’ is a potential way to produce fuels and chemicals. Its feasibility

will largely depend on a good understanding of the carboxylate production step, in terms of

parameters that affect total carboxylate levels as well as the carboxylate composition. From the

studies reviewed in this paper, some general directions can already be derived to steer mixed

culture fermentation towards a desired product spectrum. The use of more concentrated

carbohydrate-rich substrates, at longer residence times and at neutral pH ranges stimulates total

acid production. When substrate concentrations become too high in short retention times, or

under conditions of feed overloading, product inhibition may occur. When increasing pH to the

neutral range, not only total acid concentration, but also acetate and propionate fractions are

elevated, though high propionate concentrations and fractions are only reported infrequently. A

high n-butyrate fraction > 70% is usually found when pH < 6, at somehow longer incubation or

retention times or OLRs, under CO2 atmosphere or on substrates with high lactate concentrations.

Increasing residence times create a shift from shorter to longer chain fatty acids such as valerate

or n-caproate. Such more reduced compounds were also found more often on protein-rich waste

streams than on carbohydrate ones. As most of the studies referenced in this review addressed

hydrogen rather than carboxylate production, it can be anticipated that the set-ups and

operational conditions were not optimized for carboxylate production. To maximize the potential of

42

the carboxylate platform as a means to convert bio-based and organic waste materials into added

value products, more systematic and fundamental investigations are needed on the bioconversion

of waste streams into carboxylates, and the parameters affecting the process.

Most of the referenced studies used batch and continuous stirred tank reactors. It seems

worthwhile to also evaluate alternative reactor concepts. To improve product concentrations and

fractions and also decrease negative effects such as acid inhibition a “counter-current”

fermentation technique was for instance proposed (Holtzapple et al., 1999). In this technique,

fresh substrate is transferred from one to another fermenter in 4 to 5 sequencing steps while the

liquid phase is transferred on the opposite direction. This allows for optimal hydrolysis of the

substrate, and minimal inhibitory effects of the end-products. The concept indeed resulted in the

highest carboxylate concentrations found in this review. References on the use of alternative

approaches, such as high rate systems, immobilized systems or membrane bioreactors which allow

to uncouple SRT from HRT seem to be scarce. A higher SRT can increase carboxylate levels, but it

can also introduce slow growing methanogens that accelerate oxidation of carboxylate (Lee et al.,

2014). It was reported that SRTs up to 12 days increase the carboxylate production due to the

increase in available soluble feed concentration in the reactor. However further increase e.g. up to

16 days increased methane growth and hence depletion of carboxylates. Product yields may also

be further improved through novel integrated concepts, consisting of biological, bioelectrochemical

and/or catalytic steps.

So far, the concentrations of carboxylate that have been reached in mixed culture

fermentation are in general rather dilute compared to pure culture fermentations. This presents a

major challenge for carboxylate recovery and purification. The integration of a proper separation

technology with the fermentation process leads to process intensification. On the one hand, the

first step of the downstream processing is integrated with the bioconversion stage and the product

43

may be recovered in a concentrated form. On the other hand, higher conversions can be reached

by maintaining lower product concentration in the reactor, therefore removing potential inhibition.

This in turn allows using higher substrate concentrations or more concentrated feedstocks with

maximal yield and efficiency. Separation of carboxylate mixtures from complex fermentation broths

presents as yet a largely unexplored and challenging area of research, since the focus of studies

on recovery of organic acids has mainly been on pure culture fermentation systems. Furthermore,

the use of ISPR technology may introduce selectivity in the product removal stage.

44

Tabl

e 2.

2.O

verv

iew

of

proc

ess

para

met

ers

and

prod

uct

dist

ribut

ion

of b

atch

rea

ctor

typ

e fe

rmen

tatio

n

Subs

trat

eSu

bstr

ate

Con

c.In

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

Artif

icia

l ce

llulo

se

was

tew

ater

10 g

ce

llulo

se/

LD

iges

ted

slud

ge12

0N

C160

1.8

650

35-10

--

-33

Uen

o et

al

., 19

95W

AS2

357

043

--

--

58

OFM

SW3

blen

ded

with

se

wag

e sl

udge

1/5

(w:w

)Ac

idog

en

reac

tor

and

Dig

este

d sl

udge

(Hea

t tr

eate

d)

190

NC1

374.

422

1464

--

-1

66La

y et

al

., 19

99

Carb

ohyd

rate

ba

sed

MSW

5% T

S/re

acto

rD

iges

ted

slud

ge50

7 (NC1 )

375.

738

062

--

-3

-O

kam

oto

et a

l.,

2000

Prot

ein-

rich

MSW

4% T

S/re

acto

r18

813

577

2313

--

0-

Lipi

d ba

sed

MSW

8% T

S/re

acto

r(H

eat

trea

ted )

160

3.6

2527

48-

--

Neg

4-

Whe

at s

tarc

h10

g/

LD

iges

ted

slud

ge(H

eat

trea

ted )

504.

530

-35

1.7

24-

76-

--

0.5

50H

ussy

et

al.,

2003

Glu

cose

3 g

COD

/ L

Dig

este

d sl

udge

(N

on

heat

tr

eate

d)

180

7.5

253

7119

10-

--

Neg

475 0

Oh

et

al.,

2003

6.2

2.3

7013

17-

--

Neg

455

0

Hea

t tr

eate

d7.

51.

354

388

--

-N

eg4

550

6.2

2.2

6814

18-

--

Neg

455 0

45

Subs

trat

eSu

bstr

ate

Con

c.In

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

Sucr

ose

6 g

/LCo

mpo

st(H

eat

trea

ted )

806 N

C137

7.5

1347

40-

--

-40

Khan

al

et a

l.,

2004

Star

ch3.

26

5341

--

--

35

Din

ing

Hal

l Fo

od w

aste

2 g

VS/

LAc

idog

en

reac

tor

140

4.5

350.

628

1557

--

--

4Sh

in e

t al

., 20

045.

50.

848

3121

--

--

16.

50.

954

3511

--

--

0.5

4.5

550.

610

090

--

--

235.

50.

620

080

--

--

2115

6.5

0.6

390

61-

--

-14

3 g

VS/

L11

5.5

0.7

460

54-

--

-23

6 g

VS/

L17

1.2

381

62-

--

-55

8 g

VS/

L12

1.7

460

54-

--

-64

10 g

VS/

L18

237

162

--

--

69

Glu

cose

2 g/

LSl

udge

fr

om

WW

TP7

(Hea

t tr

eate

d)

50N

C130

1.3

914

5-

--

0.1

700

Park

et

al.,

2005

0.6

8710

3-

--

0.5

87

Ric

e sl

urry

(Hea

t tr

eate

d an

d ye

ast

extr

act

adde

d)

5.5

g ca

rboh

ydra

te/

L

Dig

este

dsl

udge

(Hea

t tr

eate

d)

400

437

3.1

300

70-

--

-45

-55

Fang

et

al.,

2006

4.5

3.2

290

71-

--

-62

.65

3.6

374

60-

--

-45

-55

5.5

3.4

394

57-

--

-45

-55

63.

439

457

--

--

45-

556.

53

471

52-

--

-45

-55

73.

452

246

--

--

45-

55

46

Subs

trat

eSu

bstr

ate

Con

c.In

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

3 g

carb

ohyd

rate

/ L

4.5

1.7

330

591

8-

-45

-55

5.5

g ca

rboh

ydra

te/

L

3.7

444

520

0-

-45

-55

8 g

carb

ohyd

rate

/ L

4.6

300

640

6-

-45

-55

11 g

ca

rboh

ydra

te/

L

6.3

300

680

3-

-45

-55

14 g

ca

rboh

ydra

te/

L

7.3

350

650

0-

-45

-55

22 g

ca

rboh

ydra

te/

L

10.5

290

710

0-

-45

-55

WAS

2(p

rote

in-

rich)

13 g

TCO

D/

LN

A519

04

210.

312

219

35-

--

Chen

et

al.,

2007

50.

833

2110

23-

--

60.

635

2812

14-

--

70.

846

235.

515

--

-8

1.4

3724

618

--

-9

1.9

3424

718

--

-10

2.7

4616

1116

--

-11

0.8

7317

26

--

-N

C10.

748

223

16-

--

Vege

tabl

e ki

tche

n w

aste

75 g

CO

D/

LCo

mpo

st

of f

ood

was

te &

raw

dus

t

485.

555

110

0-

--

-2

-0

Lee

et

al.,

2008

b

47

Subs

trat

eSu

bstr

ate

Con

c.In

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

64.

45

595

--

-1-

50 206.

55.

20

595

--

-3-

50 10

74

05

95-

--2

-90 20

Fodd

er m

aize

20 g

TS

/L

reac

tor

Dig

este

d sl

udge

(Hea

t tr

eate

d)

705.

235

4.2

36-

65-

--

075

10Ky

azze

et

al.,

20

08W

ilted

gra

ss20

g T

S/L

reac

tor

403

15-

85-

--

-4.5

->0.

275

10Pe

renn

ial-

ryeg

rass

10 g

TS/

L re

acto

r42

1.1

344

62-

--

-60

<

Food

was

te

100

g su

bstr

ate/

LAc

id

reac

tor

NA5

7 NC1

350.

69

586

--

-0.

5-

Zhu

et

al.,

2008

PS9

2.6

5831

12-

--

0-

PS9 +

WAS

22.

168

2012

--

-0

-Fo

od W

aste

+

PS (

1:1)

3.4

2424

52-

--

0.5

-

Food

Was

te +

W

AS2

(1:1

)2.

438

2142

--

-0.

25-

Food

Was

te +

W

AS2

+ P

S (1

:1:1

)

339

1644

--

-0.

3-

Chee

se w

hey

perm

eate

dai

ry

WW

6

4.6

mg

TCO

D/L

Anae

r.di

gest

er17

-57

days

637

1.6

5414

32-

--

--

Beng

tss-

on e

t al

., 20

08Pa

per

mill

WW

67.

7 m

g TC

OD

/L2.

414

773

--

--

-

Pulp

and

pa

per

mill

WW

62.

4 m

g TC

OD

/L0.

298

02

--

--

-

48

Subs

trat

eSu

bstr

ate

Con

c.In

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

Ther

moc

hem

ical

Pu

lp m

ill

WW

616

.7 m

g TC

OD

/L1.

551

2227

--

--

-

Cass

ava

star

ch16

g C

OD

/LSl

udge

fr

om

WW

TP7

(Hea

t tr

eate

d)

155

372.

622

276

0-

-0.

3-

Lee

et

al.,

2008

a

63.

222

177

0-

-0.

5-

6.5

5.5

211

780

--

0.1

-7

525

174

0-

-N

e g4

-N

C137

1.4

285

670

--

1.4

-55

0.3

272

710

--

0.4

-8

g C

OD

/L25

637

2.3

221

770

--

Neg

4-

16 g

CO

D/L

5.5

211

780

--

0.1

-24

g C

OD

/L6.

925

075

0-

-N

eg4

-

32 g

CO

D/L

9.5

220

780

--

Neg

4-

Food

was

te

(Hea

t Tr

eate

d)30

g

COD

ca

rboh

ydra

te/L

No

inoc

ula

485

3512

1511

74-

-0

2.9

-Ki

m e

t al

., 20

09U

ntre

ated

foo

d w

aste

110

00

0-

-16

0-

Food

was

te

from

res

taur

ant

(Ble

nded

. si

eved

<2m

m)

93 g

TCO

D/L

No

inoc

ula

530

715

2.5

2040

40-

-20

->

5-

50Ko

mem

-ot

o et

al

., 20

0925

729

1457

--

-25

-45

356.

411

1178

--

-30

-45

457

147

79-

--

30-

4055

6.7

1510

75-

-25

->

15-

55

654

3813

50-

--

20-

4

49

Subs

trat

eSu

bstr

ate

Con

c.In

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

Ric

e20

g V

S/L

Anae

robi

c di

gest

er16

85.

5N

C137

14.4

4919

274

--

0.2

7025

Don

g et

al

., 20

09Po

tato

5.3

554

410

--

0.5

6020

Lett

uce

(Hea

t tr

eate

d)5

686

260

--

0.2

7020

Oil

2.2

97

813

--

0.1

20 0

Fat

184

151

0-

-0.

10

Bany

an le

aves

2.2

7510

141

--

0.1

20 0Le

an m

eat

12.5

g V

S/L

2.3

3220

3216

--

00

WAS

2

(Add

ed w

ith

rice.

ble

nded

an

d si

eved

< 2

m

m)

18 g

VSS

/ L

Slud

ge

from

W

WTP

7

190

421

1.1

5823

118

-N

eg4

--

Feng

et

al.,

2009

53.

351

2118

10-

Ne g

4-

-6

4.5

4152

44

-N

eg4

--

75.

240

515

3-

Ne g

4-

-8

6.5

4648

41

-N

eg4

--

95.

944

494

4-

Neg

4-

-10

2.8

7814

44

-~

1-

-11

2.7

8411

23

-~

1-

-N

C11.

978

148

0-

Neg

4-

-

Food

was

te

(Hea

t tr

eate

d)30

g

carb

ohyd

rate

/L

No

inoc

ula

200

535

1732

1058

--

Neg

40.

6-

Kim

et

al.,

2011

a6

1924

1165

--

Ne g

40.

4-

719

380

62-

-0.

90.

5-

818

270

73-

-N

e g4

0.3

-9

1629

071

--

00.

5-

50

Subs

trat

eSu

bstr

ate

Con

c.In

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

WAS

211

g T

COD

/LN

A548

0N

A54

0.5

6718

87

--

--

Yuan

et

al.,

2011

141.

654

2712

8-

--

-25

1.8

5226

128

--

--

40.

162

248

5-

--

-14

0.9

797

68

--

--

251.

357

1816

10-

--

-

Dai

ry m

anur

e35

g V

S/ L

No

inoc

ula

192

NC1

203.

870

1711

1-

--

-Co

ats

et

al.,2

011

40.7

g V

S/ L

3.7

7017

121

--

--

48.7

g V

S/ L

4.1

7016

131

--

--

Food

w

aste

:sew

age

slud

ge

(Hea

t tr

eate

d)

61 g

TCO

D/

L (F

W:S

S 1

0:1

on C

OD

bas

is)

No

inoc

ula

366

3520

2213

65-

-0.

22.

6-

Kim

et

al.,

2011

b

Carb

ohyd

rate

-ric

h W

W6

8 g

COD

/LD

iges

ted

slud

ge

(Hea

t tr

eate

d)

168

5 NC1

303.

351

241

05

-0.9

170

100

Arsl

an e

t al

., 20

128

(7 d

ays)

1.2

256

670

0-0

.850

010

0%

CO2

1.6

277

631

0-0

.9N

eg4

502.

856

3113

00

-0.9

Ne g

42%

Prot

ein-

rich

WW

65

0.5

469

233

8-0

.1N

eg4

100

0.7

624

180

2-0

.158

610

0%

CO2

3.8

921

50

0-0

.10

501

654

170

0-0

.10

1%Li

pid-

rich

WW

65

0.6

3614

286

0N

eg4

010

0N

C10.

10

1441

130

Neg

40

100%

CO

20.

337

922

70

Neg

40

500.

0561

919

30

Neg

40

Neg

4

51

1 NC:

Not

con

trol

led

2 WAS

: W

aste

act

ivat

ed s

ludg

e3 O

FMSW

: O

rgan

ic f

ract

ion

of m

unic

ipal

sol

id w

aste

4 Neg

: N

eglig

ible

5 NA:

Not

ava

ilabl

e6 W

W:

Was

tew

ater

7 WW

TP:

Was

tew

ater

tre

atm

ent

plan

t8 H

eads

pace

d m

odifi

ed b

y fil

ling

H2

and

CO2

9 PS:

Prim

ary

slud

ge10

-: N

ot p

rodu

ced

or n

ot in

dica

ted

in t

he s

tudy

Subs

trat

eSu

bstr

ate

Con

c.In

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

Carb

ohyd

rate

-ric

h W

W6

13.5

g C

OD

/LD

iges

ted

slud

ge

(Hea

t tr

eate

d)

168

5 NC1

301.

99

980

00

-1.4

0.2

100

Arsl

an e

t al

., 20

138

(7 d

ays)

2.6

285

640

0-1

.40.

510

0%

CO2

2.9

417

490

0-1

.4N

eg4

504

472

450

0-1

.3N

eg4

6%

23 g

CO

D/L

7.3

464

450

3-2

0.6

100

7.3

484

440

2-2

0.6

100%

CO

28.

252

342

02

-20.

550

8.2

315

650

0-2

0.2

27%

52

Tabl

e 2.

3. O

verv

iew

of

proc

ess

para

met

ers

and

prod

uct

dist

ribut

ion

of

fed-

batc

h re

acto

r ty

pe fe

rmen

tatio

n

Subs

trat

eO

LRIn

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

Sucr

ose

40 g

CO

D/L

.dW

AS1

126.

735

7.3

399

52-

--

0.3

15Li

n &

Jo,

20

0348

g

COD

/L.d

107.

938

1052

--

-0.

420

60 g

CO

D/L

.d8

5.6

226

72-

--

0.5

29

80 g

CO

D/L

.d6

5.8

2814

59-

--

0.5

33

120

g CO

D/L

.d4

5.5

2710

63-

--

0.5

32

Food

w

aste

fr

om

cafe

teria

(I

nitia

lly

gluc

ose

adde

d)

9 g/

L.d

Dig

este

d sl

udge

190

535

16.5

174

170

13-

--

Lim

et

al.,

2008

(8 d

ays)

5.5

23.5

3227

2013

3-

--

625

4924

207

0-

--

5 g/

L.d

965.

535

5.5

3631

1716

--

--

190

13.5

2828

2221

--

--

290

(12

days

)21

2441

1717

--

--

9 g/

L.d

190

5.5

2517

2045

1518

0-

--

3523

.532

2721

133

--

-45

2049

418

524

--

5 g/

L.d

190

5.5

3513

.528

2822

210

0-

-9

g/L.

d23

.532

2620

133

0-

-13

g/L

.d29

379

90

1118

--

Food

w

aste

20 g

TS

/L.d

Slud

ge

slau

ghte

r-ho

use

WW

TP6

72~

5.5

345.

721

018

-61

-0.

127

Gom

ez e

t al

., 20

09

53

1 WAS

: W

aste

act

ivat

ed s

ludg

e2 W

W:

Was

tew

ater

3 Neg

: N

eglig

ible

4 10%

to

20%

i-bu

tyra

te a

nd i-

vale

rate

det

ecte

d

Subs

trat

eO

LRIn

ocul

aIn

c.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%

Pr

%B

u%

Va

%C

a%

La (g /L)

EtO

H

(g /L)

H2

(%)

Ref

.

Dig

este

d sl

udge

1217

066

-17

-1

14

20 g

TS

/L.d

PS+

W

AS2

119

261

-28

-1.

230

10 g

TS

/L.d

3.5

499

21-

21-

130

Palm

oil

mill

WW

25

g CO

D/L

.dAn

aero

bic

slud

ge(H

eat-

trea

ted)

966.

837

134

065

-0

-N

eg3

19Ba

diei

et

al.,

2011

7 g

COD

/L.d

721.

422

077

-0

-N

eg3

50

10 g

CO

D/L

.d48

1.5

180

82-

0-

Neg

334

13 g

CO

D/L

.d36

1.4

340

65-

0-

Neg

314

WAS

142

g

COD

/L.d

NA

48N

C42

1550

1614

45

Neg

3N

eg3

Neg

3M

orga

n Sa

gast

ume

et a

l.,

2011

4W

AS 2

49 g

CO

D/L

.d37

9.4

4117

179

3N

eg3

Neg

3N

eg3

WAS

316

g

COD

/L.d

379.

537

1916

76

Neg

3N

eg3

Neg

3

WAS

419

g

COD

/L.d

429.

742

1515

62

Neg

3N

eg3

Neg

3

Food

w

aste

(T

reat

ed

with

bas

e)

27 g

CO

D/L

.dN

o in

ocul

a24

5.3

3513

3915

46-

-1.

51.

223

Kim

et

al.,

2012

a30

11.5

3913

48-

-1.

90.

930

54

55

Tabl

e 2.

4. O

verv

iew

of

proc

ess

para

met

ers

and

prod

uct

dist

ribut

ion

of c

ontin

uous

rea

ctor

typ

e fe

rmen

tatio

n

Subs

trat

eO

LRIn

ocul

aR

eten

.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%P

r%B

u%V

a%C

a%La (g /L

)

Et-

OH

(g/l

)

H2

(%)

Ref

.

Suga

r-ric

h W

W1

64 g

CO

D/L

.dAn

aero

bic

slud

ge12

6.8

603.

335

065

-10-

--

65U

eno

et

al.,

1996

32 g

CO

D/L

.d24

3.7

360

64-

--

-65

16 g

CO

D/L

.d48

4.5

390

61-

--

-65

11 g

CO

D/L

.d72

5.5

420

58-

--

-65

Prim

ary

slud

ge

Pota

to W

W1

(1:1

)

7 g

TS/L

.dAn

aero

bic

slud

ge18

NC2

220.

556

1916

3-

--

-Ba

nerj

ee

et a

l.,

1999

4 g

TS/L

.d30

0.6

5712

221

--

--

4 g

TS/L

.d30

300.

768

1214

1-

--

-35

0.5

7612

52

--

--

Prim

ary

slud

ge7

g TS

/L.d

1822

0.2

6125

53

--

--

7 g

TS/L

.d18

0.3

6116

96

--

--

4 g

TS/L

.d30

0.4

6127

25

--

--

4 g

TS/L

.d30

300.

268

222

4-

--

-35

0.1

6721

26

--

--

Glu

cose

14 g

/L.d

Anae

robi

c sl

udge

145

373.

818

380

--

-0.

230

Hor

iuch

i et

al.,

2002

16 g

/L.d

123.

921

078

--

-N

eg3

3321

g/L

.d9

426

173

--

-N

eg3

3315

g/L

.d12

.56

418

478

--

-0.

235

19 g

/L.d

103.

921

673

--

-0.

242

24 g

/L.d

83.

617

677

--

-0.

232

27 g

/L.d

73.

816

1371

--

-0.

232

38 g

/L.d

54

2510

65-

--

0.3

3621

g/L

.d9

74.

329

566

--

-0.

245

38 g

/L.d

53.

925

174

--

-0.

148

38 g

/L.d

54

312

67-

--

0.1

4748

g/L

.d4

3.9

304

67-

--

Neg

346

15 g

/L.d

138

4.7

6135

4-

--

0.3

2119

g/L

.d10

559

365

--

-0.

418

56

Subs

trat

eO

LRIn

ocul

aR

eten

.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%P

r%B

u%V

a%C

a%La (g

/L

)

Et-

OH

(g/l

)

H2

(%)

Ref

.

24 g

/L.d

84.

756

349

--

-0.

524

32 g

/L.d

64.

357

3310

--

-0.

222

35 g

/L.d

5.5

4.8

5336

11-

--

0.1

16

Glu

cose

28 g

/L.d

Anae

robi

c sl

udge

64.

036

1.6

322

64-

2N

eg0.

240

Fang

&

Liu,

200

24.

5 2.

239

156

-4

0.1

0.2

505.

01.

942

150

-7

0.1

0.2

605.

51.

945

251

-3

0.1

0.3

606.

0 1.

853

442

-1

0.2

0.3

606.

52.

354

738

-1

0.1

0.2

507.

02.

348

1832

-2

Neg

0.1

40

Sucr

ose

rich

W

W1

25 g

su

cros

e/L.

dAn

aero

bic

slud

ge5

5.5

260.

919

180

-0

-N

eg3

70Li

u &

Fa

ng,

2002

71.

421

179

-0

-N

eg3

6510

2.3

261

73-

0-

Ne g

365

143.

927

172

-0

-N

eg3

6517

4.5

221

75-

2-

Ne g

363

297.

112

183

-4

-0.

160

Sucr

ose

37 g

CO

D/L

.dSl

udge

fr

om

aera

tion

tank

13N

C235

4.6

446

50-

--

0.8

35Ch

en e

t al

., 20

0348

g

COD

/L.d

105.

921

772

--

-0.

341

60 g

CO

D/L

.d8

5.6

234

73-

--

0.3

42

80 g

CO

D/L

.d6

4.8

278

65-

--

0.4

41

96 g

CO

D/L

.d5

528

467

--

-0.

441

120

g CO

D/L

.d4

3.7

2811

62-

--

0.3

46

160

g CO

D/L

.d3

3.9

213

76-

--

0.2

47

240

g CO

D/L

.d2

2.9

251

74-

--

0.3

42

57

Subs

trat

eO

LRIn

ocul

aR

eten

.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%P

r%B

u%V

a%C

a%La (g

/L

)

Et-

OH

(g/l

)

H2

(%)

Ref

.

Whe

at s

tarc

h(G

luco

se a

dded

1

g/L)

13 g

/L.d

Anae

robi

c sl

udge

185.

230

3.7

484

47-

--

0.2

50H

ussy

et

al.,

2003

13 g

/L.d

184.

535

2.3

310

69-

--

055

13 g

/L.d

(Hea

t-tr

eate

d )18

5.2

304.

452

1730

--

-N

eg3

4820

g/L

.d12

3.6

4812

39-

--

Neg

348

16 g

/L.d

153.

332

761

--

-0.

1Sp

. N

24

Corn

sto

ver

80%

pig

man

ure

20%

(Hea

t tr

eate

d)

7.5

g VS

/L.d

Rum

en

192

6.3

4020

4214

1811

11-

--

Than

ako-

ses

et a

l.,

2003

6 g

VS/L

.d(8

day

s)21

3620

2311

8-

--

4 g

VS/L

.d17

.542

1519

1010

--

-3

g VS

/L.d

1436

1515

1115

--

-2

g VS

/L.d

1235

1218

1016

--

-

Gel

atin

Sy

nthe

tic W

W1

8 g/

L.d

Anae

robi

c sl

udge

125.

520

1.2

2119

1111

9-

Neg

380

Yu &

Fan

g 20

0325

1.3

2418

1210

9-

Neg

380

301.

427

1512

119

-N

e g3

8037

1.4

2613

1312

8-

Neg

360

451.

428

2511

1210

-N

e g3

7050

1.5

2414

1311

9-

Neg

365

551.

522

1615

149

-N

e g3

604

370.

815

3311

117

-N

eg3

604.

51

1929

119

6-

Ne g

335

51.

324

2012

1010

-N

eg3

205.

51.

426

1313

128

-N

e g3

06

1.5

2913

1610

9-

Neg

30

6.5

1.6

3211

219

6-

00

71.

635

922

106

-0

0

Sucr

ose

16 g

/L.d

Anae

robi

c sl

udge

145.

232

4.5

321

670

-0

Neg

350

Hus

sy e

t al

., 20

054.

134

165

0-

0N

eg3

Sp.

N24

Suga

r be

et

wat

er e

xtra

ct3.

542

251

0-

00.

250

3.3

400

600

-0

0.4

Sp.

N24

58

Subs

trat

eO

LRIn

ocul

aR

eten

.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%P

r%B

u%V

a%C

a%La (g

/L

)

Et-

OH

(g/l

)

H2

(%)

Ref

.

Food

was

te f

rom

ca

fete

ria

(Ble

nded

. sie

ved

<2m

m)+

Prim

ary

slud

ge

10 %

FW

:90%

PS (3

0 g

TCO

D/L

.d)

NA6

24

NC2

18

1.8

00

100

--

NA6

-0

Min

et

al.,

2005

(10

g TC

OD

/L.d

)72

2.2

6418

17-

--

0

(6 g

TC

OD

/L.d

)12

01.

40

5248

--

-0

25 %

FW

:75%

PS

2413

1440

46-

--

072

1.6

055

45-

--

012

00.

46

4945

--

-0

10 %

FW

:90%

PS

246.

518

0.9

00

100

--

-0

720.

70

5050

--

-0

120

0.4

00

100

--

-0

25 %

FW

:75%

PS

248

076

24-

--

072

1.1

078

22-

--

012

00.

0412

4444

--

-0

10 %

FW

:90%

PS

24N

C235

4.3

066

34-

--

072

0.9

047

53-

--

012

00.

40

010

0-

--

0

25 %

FW

:75%

PS

2410

055

45-

--

072

0.7

010

00

--

-0

120

0.2

1534

51-

--

0

10 %

FW

:90%

PS

246.

535

1.5

00

100

--

-0

720.

80

5149

--

-0

120

0.3

00

100

--

-0

59

Subs

trat

eO

LRIn

ocul

aR

eten

.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%P

r%B

u%V

a%C

a%La (g

/L

)

Et-

OH

(g/l

)

H2

(%)

Ref

.

25 %

FW

:75%

PS

243.

50

010

0-

--

072

0.5

00

100

--

-0

120

0.2

941

50-

--

0

Sucr

ose

20 g

su

cros

e/L.

dAn

aero

bic

slud

ge(H

eat-

trea

ted)

125.

235

6.1

28-

72-

--

-60

Kyaz

ze e

t al

., 20

0540

g

sucr

ose/

L.d

8.4

24-

76-

--

-56

80 g

su

cros

e/L.

d13

28-

72-

--

-55

100

g su

cros

e/L.

d16

28-

72-

--

-54

96 g

su

cros

e/L.

d10

1730

-70

--

--

No

S p.

69 g

su

cros

e/L.

d14

1528

-72

--

--

Sp.

N24

120

g su

cros

e/L.

d10

1429

-71

--

--

No

S p.

200

g su

cros

e/L.

d6

1230

-70

--

--

Sp.

N24

Sucr

ose

rich

syth

etic

W

W1

7 g

COD

/L.d

WW

TP5

slud

ge18

738

1.5

154

6312

6-

0.2

60Yu

& M

u,

2006

11 g

CO

D/L

.d2.

412

362

176

-0.

350

14 g

CO

D/L

.d3.

216

460

147

-0.

552

26 g

CO

D/L

.d5.

69

250

929

-0.

740

30 g

CO

D/L

.d6.

210

155

528

-0.

840

77 g

CO

D/L

.d6.

612

258

622

-0.

740

85 g

CO

D/L

.d3

2.6

253

353

34-

0.3

40

60

Subs

trat

eO

LRIn

ocul

aR

eten

.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%P

r%B

u%V

a%C

a%La (g

/L

)

Et-

OH

(g/l

)

H2

(%)

Ref

.

43 g

CO

D/L

.d6

2.6

182

363

41-

0.4

45

26 g

CO

D/L

.d10

2.6

161

332

48-

0.4

45

18 g

CO

D/L

.d14

2.8

172

145

60-

0.3

45

12 g

CO

D/L

.d22

2.8

115

4024

20-

0.3

45

9 g

COD

/L.d

302.

515

213

664

-0.

343

Sucr

ose

40 g

CO

D/L

.dW

WTP

5

slud

ge(H

eat-

trea

ted)

125.

335

6.8

513

46-

-3.

60.

863

.2

(C7 )

Kim

et

al.,

2006

6.8

482

50-

-3.

30.

7In

t.

R8

7.3

512

47-

-3.

10.

7Sp

. N

24

5.6

173

80-

-0.

70.

6Sp

. CO

29

Gar

bage

slu

rry

mix

ed w

ith

pape

r w

aste

89 g

CO

D/L

.dAc

tivat

ed

slud

ge

com

post

145.

955

545

549

1-

3.8

-50

-60

Uen

o et

al

., 20

0795

.5 g

CO

D/L

.d5.

93.

451

538

6-

4-

50-6

0

138

g CO

D/L

.d5.

96.

237

260

1-

4.7

-50

-60

54 g

CO

D/L

.d26

65.

735

460

1-

2.8

-50

-60

97 g

CO

D/L

.d6

1819

180

0-

6.9

-50

-60

61

Subs

trat

eO

LRIn

ocul

aR

eten

.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%P

r%B

u%V

a%C

a%La (g

/L

)

Et-

OH

(g/l

)

H2

(%)

Ref

.

Glu

cose

5 g/

L.d

WW

TP5

slud

ge+

Ac

idog

enic

re

acto

r

204

301.

732

068

--

0.2

0-

Tem

udo

et

al.,

2007

4.75

1.6

3911

51-

-0.

30.

1-

51.

632

068

--

0.2

0.1

-5.

51.

828

568

--

Neg

30.

1-

12 g

/L.d

85.

51.

634

561

--

0N

eg3

-6.

251

8016

3-

-0

0.6

-7

1.1

8515

0-

-0

0.7

-7.

751.

360

1327

--

0.2

0.6

-8.

51

8710

4-

-0.

10.

6-

Cellu

lose

ric

h fo

od w

aste

+

corn

sta

rch

20 g

CO

D/L

.dAc

idog

enic

re

acto

r 96

5.5

355

570

43-

--4

.7-

36Li

et

al.,

2008

Cellu

lose

ric

h fo

od w

aste

20 g

CO

D/L

.d1.

340

2040

--

-5.6

-1.

5

Food

was

te f

rom

re

stau

rant

26 g

CO

D/L

.d3.

330

070

--

-8.1

-33

50 g

CO

D/L

.d48

440

555

--

-6.8

-35

Glu

cose

6 g/

L.d

Dig

este

d sl

udge

W

WTP

5

8~

637

1.1

603

371

-0

071

Haf

ez e

t al

., 20

1024

g/L

.d4.

259

338

0-

00.

173

46 g

/L.d

7.3

572

410

-0

0.1

6594

g/L

.d13

592

390

-0

0.1

6714

0 g/

L.d

1238

557

0-

00.

443

187

g/L.

d12

368

541

-0

0.6

39

Suga

rcan

e ba

ggas

e+

man

ure

(pre

-tre

ated

air

lime)

5 VS

/L.d

Anae

robi

c bi

omas

s(io

dofo

rm

trea

ted)

192

755

4088

39

0.4

--

--

Fu &

Hol

tzap

ple,

20

112.

5 g

VS/L

.d(8

day

s)28

873

100.

5-

--

-

62

1 WW

: W

aste

wat

er

2 NC:

Not

con

trol

led

3 Neg

: N

eglig

ible

4 Sp.

N2:

Spa

rged

with

N2

5 WW

TP:

Was

tew

ater

tre

atm

ent

plan

t

6 NA:

Not

ava

ilabl

e7 C

: Co

ntro

l 8 I

nt R

: In

tern

al r

ecirc

ulat

ion

9 Sp.

CO

2: S

parg

ed w

ith C

O2

10-:

Not

pro

duce

d or

not

indi

cate

d in

the

stu

dy

Subs

trat

eO

LRIn

ocul

aR

eten

.ti

me

(h)

pHT (°

C)

TVFA

(g/L

)A

c%P

r%B

u%V

a%C

a%La (g

/L

)

Et-

OH

(g/l

)

H2

(%)

Ref

.

Carb

ohyd

rate

-ric

h W

W1

23 g

CO

D/L

.dD

iges

ted

slud

ge

(Hea

t tr

eate

d)

484

306.

529

467

00

-0.7

0.1

45Ar

slan

et

al.,

2013

Mol

asse

s19

g

COD

/L.d

WW

TP5

slud

ge10

NC2

350.

7553

1136

--

-0.

7053

Wan

g et

al

., 20

1324

g

COD

/L.d

80.

8054

937

--

-0.

7055

32 g

CO

D/L

.d6

1.10

547

39-

--

1.30

63

38 g

CO

D/L

.d5

1.20

527

42-

--

1.80

58

48 g

CO

D/L

.d4

1.20

516

43-

--

1.00

52

Food

was

teM

ixed

with

sew

age

10 g

VS

/L.d

Anae

robi

cco

mpo

st72

4.5

300.

891

09

00

3.7

0.5

10-1

5Pa

nt e

t al

., 20

135.

56.

729

1444

130

0.9

0.4

10-1

55.

06.

932

1042

140

1.6

0.5

10-1

510

725.

56.

635

2819

180

-0.1

010

-15

321

67.

545

2319

130

-0.3

0.1

10-1

520

365.

06.

631

1340

360

0.4

0.2

10-1

510

726.

920

1336

320

-0.4

0.7

10-1

5

63

References

Agler, M. T., Spirito, C. M., Usack, J. G., Werner, J. J., & Angenent, L. T. (2012). Chain elongation with reactor microbiomes: upgrading dilute ethanol to medium-chain carboxylates. Energy & Environmental Science, 5(8), 8189 - 8192.

Agler, M. T., Wrenn, B. a, Zinder, S. H., & Angenent, L. T. (2011). Waste to bioproduct conversion with undefined mixed cultures: the carboxylate platform. TRENDS in Biotechnology, 29(2), 70–78.

Angenent, L. T., Karim, K., Al Dahhan, M. H., Wrenn, B. A., Dominguez Espinosa, R. (2004). Production of bioenergy and biochemicals from industrial and agricultural wastewater. TRENDS in Biotechnology, 22(9 ), 477–485.

Angenent, L. T., & Kleerebezem, R. (2011). Crystal ball. Microbial Biotechnology, 4,109-137.

Arooj, M. F., Han, S.-K. K., Kim, S.-H. H., Kim, D.-H. H., & Shin, H.-S. S. (2008). Effect of HRT on ASBR converting starch into biological hydrogen. International Journal of Hydrogen Energy, 33(22), 6509–6514.

Arslan, D., Steinbusch, K. J. J., Diels, L., De Wever, H., Buisman, C. J. N., & Hamelers, H. V. M. (2012). Effect of hydrogen and carbon dioxide on carboxylic acids patterns in mixed culture fermentation. Bioresource Technology,118, 227–234.

Arslan, D., Steinbusch, K. J. J., Diels, L., De Wever, H., Hamelers, H. V. M., & Buisman, C. J. N. (2013). Selectivecarboxylate production by controlling hydrogen, carbon dioxide and substrate concentrations in mixed culture fermentation. Bioresource Technology, 136, 452–460.

Badiei, M., Jahim, J. M., Anuar, N., Sheikh Abdullah, S. R. (2011). Effect of hydraulic retention time on biohydrogen production from palm oil mill effluent in anaerobic sequencing batch reactor. International Journal of Hydrogen Energy 36(10), 5912–5919

Banerjee, A., Elefsiniotis, P., Tuhtar, D. (1999). The effect of addition of potato-processing wastewater on the acidogenesis of primary sludge under varied hydraulic retention time and temperature. Journal of Biotechnology,72, 203–212.

Bastidas-Oyanedel, J.-R. R., Mohd-Zaki, Z., Zeng, R. J., Bernet, N., Pratt, S., Steyer, J.-P. P., & Batstone, D. J. (2012). Gas controlled hydrogen fermentation. Bioresource Technology, 110, 503–9.

Bengtsson, S., Hallquist J., Werker, A., Walender, T. (2008). Acidogenic fermentation of industrial wastewaters: Effects of chemostat retention time and pH on volatile fatty acids production. Biochemical Engineering Journal, 40, 492–499.

Buque-Taboada, E., Straathof, A., Heijnen, J., & van der Wielen, L. (2006). In situ product recovery (ISPR) by crystallization: basic principles, design, and potential applications in whole-cell biocatalysis. Applied Microbiology and Biotechnology, 71(1), 1–12.

Chang, J. S., Lee, K. S., & Lin, P. J. (2002). Biohydrogen production with fixed-bed bioreactors. International Journal of Hydrogen Energy, 27, 1167–1174.

Chen, C., Lin, C. (2003) .Using sucrose as a substrate in an anaerobic hydrogen-producing reactor. Advances in Environmental Resources, 7(3), 695–699

Chen, C. C., Lin, C. Y., & Lin, M. C. (2002). Acid-base enrichment enhances anaerobic hydrogen production process. Applied Microbiology and Biotechnology, 58(2), 224–228.

64

Chen, Y., Jiang, S., Yuan, H., Zhou, Q., & Gu, G. (2007). Hydrolysis and acidification of waste activated sludge at different pHs. Water Research, 41(3), 683–689.

Chen, Y., Li, X., Zheng, X., & Wang, D. (2013). Enhancement of propionic acid fraction in volatile fatty acids produced from sludge fermentation by the use of food waste and Propionibacterium acidipropionici. Water Research, 47, 615–622.

Chan, W., Holtzapple, M. (2003). Conversion of municipal solid wastes to carboxylic acids by thermophilic fermentation. Applied Biochemical Biotechnology, 111, 93 - 112.

Coats, E. R., Gregg, M., & Crawford, R. L. (2011). Effect of organic loading and retention time on dairy manure fermentation. Bioresource Technology, 102(3), 2572–2577.

Cohen, A., Distel, B., van Deursen, A., Breure, A. van Andel, J. (1985). Role of anaerobic spore-forming bacteria in the acidogenesis of glucose: Changes induced by discontinuous or low-rate feed supply. Antonie van Leeuwenhoek, 51, 179 - 192.

Conrad, R., & Klose, M. (2000). Selective inhibition of reactions involved in methanogenesis and fatty acid production on rice roots. FEMS Microbiology Ecology, 34(1), 27–34.

Danner, H. R. B., & Braun, R. (1999). Biotechnology for the production of commodity chemicals from biomass. Chemical Society Reviews, 28(6), 395–405.

Drake, H.L., et al. Acetogenic prokaryotes. In The prokaryotes; Dworkin , M., Falkow, S., Rosenberg, E., Schleifer, K. H., Stackebrandt, E., Eds.; Springler-Verlag, New York 2006; pp 354 – 420.

De Gioannis, G., Muntoni, A., Polettini, A., Pomi R. (2013). A review of dark fermentative hydrogen production from biodegradable municipal waste fractions. Waste Management, 33, 1345 – 1361.

De Mes, T. Z. D., et al. (2003). Methane Production by Anaerobic Digestion of Wastewater and Solid Wastes. In Bio-methane and bio-hydrogen; Reith, J.H., Wijffels, R.H., Barten, H., Eds.; Dutch Biological Hydrogen Foundation: Petten pp. 58–102.

Ding, H. B. B., Tan, G. A Y., Wang, J. Y. Y. (2010). N-caproate formation in mixed-culture fermentative hydrogen production. Bioresource Technology, 101(24), 9550–9559.

Dong, L., Zhenhong, Y., Yongming, S., Xiaoying, K., & Yu, Z. (2009). Hydrogen production characteristics of the organic fraction of municipal solid wastes by anaerobic mixed culture fermentation. International Journal of Hydrogen Energy, 34(2), 812–820.

Fang, H. H. P., Li, C., & Zhang, T. (2006). Acidophilic biohydrogen production from rice slurry. International Journal of Hydrogen Energy, 31(6), 683–692.

Fang, H. H. P., & Liu, H. (2002). Effect of pH on hydrogen production from glucose by a mixed culture. Bioresource Technology, 82(1), 87–93.

Feng, L., Chen, Y., & Zheng, X. (2009). Enhancement of waste activated sludge protein conversion and volatile fatty acids accumulation during waste activated sludge anaerobic fermentation by carbohydrate substrate addition: The effect of pH. Environmental Science and Technology, 43(12), 4373–4380.

Fu, Z., & Holtzapple, M. (2011). Anaerobic thermophilic fermentation for carboxylic acid production from in-storage air-lime-treated sugarcane bagasse. Applied Microbiology and Biotechnology, 1–11.

Gallert, C., & Winter, J. (2005) Bacterial metabolism in wastewater treatment systems. In Environmental Biotechnology: Concepts and Applications; Jördening, H. J., Winter, J., Eds.; Wiley-VCH: Weinheim, pp. 1–48

65

Ghosh, S., & Pohland, F. G. (1974) Kinetics of substrate assimilation and product formation in anaerobic digestion. J. Environ. Fed., 46, 748 – 759.

Gomez, X., Cuetos, M. J., Prieto, J. I., & Moran, A. (2009). Bio-hydrogen production from waste fermentation: Mixing and static conditions. Renewable Energy, 34, 970 – 975.

Grootscholten, T. I. M., Steinbusch, K. J. J., Hamelers, H. V. M., & Buisman, C. J. N. (2012). Chain elongation of acetate and ethanol in an upflow anaerobic filter for high rate MCFA production. Bioresource Technology, 135, 440 – 445.

Hafez, H., Nakhla, G., El. Naggar, M. H., Elbeshbishy, E., Baghchehsaraee, B., & Naggar, M. H. (2010). Effect of organic loading on a novel hydrogen bioreactor. International Journal of Hydrogen Energy, 35(1), 81–92.

Han, S.-K. K., & Shin, H.-S. S. (2004). Biohydrogen production by anaerobic fermentation of food waste. International Journal of Hydrogen Energy, 29(6), 569–577.

Hawkes, F. R., Dinsdale, R., Hawkes, D. L., & Hussy, I. (2002). Sustainable fermentative hydrogen production: challenges for process optimisation. International Journal of Hydrogen Energy, 27(11-12), 1339–1347.

Holtzapple, M., Davison, R., Ross, M., Aldrett Lee, S.; Nagwani, M., Lee, C. M., Lee, C., Adelson, S., Kaar, W., Gaskin, D.,Shirage, H., Chang, N. S., Chang, V.S., Loescher, E.M. (1999). Biomass conversion to mixed alcohol fuels using the MixAlco process. Applied Biochemistry and Biotechnology, 79(1), 609–631.

Holtzapple, M., & Granda, C. (2009). Carboxylate Platform: The MixAlco Process Part 1: Comparison of Three Biomass Conversion Platforms. Applied Biochemistry and Biotechnology. 156 (1), 95–106.

Horiuchi, J. I., Shimizu, T., Tada, K., Kanno, T., & Kobayashi, M. (2002). Selective production of organic acids in anaerobic acid reactor by pH control. Bioresource Technology, 82(3), 209–213.

Hussy, I., Hawkes, F. R., Dinsdale, R., & Hawkes, D. L. (2003). Continuous fermentative hydrogen production from a wheat starch co-product by mixed microflora. Biotechnology and Bioengineering, 84(6), 619–626.

Hussy, I., Hawkes, F. R., Dinsdale, R., & Hawkes, D. L. (2005). Continuous fermentative hydrogen production from sucrose and sugarbeet. International Journal of Hydrogen Energy, 30(5), 471–483.

Inanc, B., Matsui, S., & Ide, S. (1999). Propionic acid accumulation in anaerobic digestion of carbohydrates: An investigation on the role of hydrogen gas. Water Science and Technology, 40, 93 - 100.

Jonke, A., Michal, G. Catalytic activity of enzymes. In Enzymes in Industry Production and Application; Wolfgang A., Ed.; Wiley VCH: Weinheim 2007; pp 13 – 33.

Khanal, S. K., Chen, W.-H. H., Li, L., & Sung, S. (2004). Biological hydrogen production: effects of pH and intermediate products. International Journal of Hydrogen Energy, 29(11), 1123–1131.

Kim, D. H., Kim, S. H., & Shin, H. S. (2009). Hydrogen fermentation of food waste without inoculum addition. Enzyme and Microbial Technology, 45, 181 – 187.

Kim, D. H., Shin, H. S., & Kim, S. H. (2012a). Enhanced H2 fermentation of organic waste by CO2 sparging. International Journal of Hydrogen Energy, 37, 15563 – 15568.

Kim, D.-H. H., Han, S.-K. K., Kim, S.-H. H., & Shin, H.-S. S. (2006). Effect of gas sparging on continuous fermentative hydrogen production. International Journal of Hydrogen Energy, 31(15), 2158–2169.

Kim, D.-H. H., Kim, S.-H. H., Jung, K.-W. W., Kim, M.-S. S., & Shin, H.-S. S. (2011a). Effect of initial pH independent of operational pH on hydrogen fermentation of food waste. Bioresource Technology, 102, 8646 – 8652.

66

Kim, D.-H. H., Kim, S.-H. H., Kim, H.-W. W., Kim, M.-S. S., & Shin, H.-S. S. (2011b). Sewage sludge addition to food waste synergistically enhances hydrogen fermentation performance. Bioresource Technology, 102(18), 8501–8506.

Kim, I. S., Hwang, M. H., Jang, N. J., Hyun, S. H., & Lee, S. T. (2004). Effect of low pH on the activity of hydrogen utilizing methanogen in bio-hydrogen process. International Journal of Hydrogen Energy, 29(11), 1133–1140.

Kim, T.H., Lee, Y., Chang, K. H., Hwang, S. J. (2012b). Effects of initial lactic acid concentration, HRTs, and OLRs on bio-hydrogen production from lactate-type fermentation. Bioresource Technology, 103, 136–141.

Kleerebezem, R., & Stams, A. J. M. (2000). Kinetics of syntrophic cultures: A theoretical treatise on n-butyrate fermentation. Biotechnology and Bioengineering, 67(5), 529–543.

Kleerebezem, R., & van Loosdrecht, M. C. M. (2007). Mixed culture biotechnology for bioenergy production. Current Opinion in Biotechnology, 18(3), 207–212.

Komemoto, K., Lim, Y. G., Nagao, N., Onoue, Y., Niwa, C., Toda, T. (2009). Effect of temperature on VFA’s and biogas production in anaerobic solubilization of food waste. Waste Management, 29, 2950–2955.

Kyazze, G., Dinsdale, R., Hawkes, F. R., Guwy, A. J., Premier, G. C., & Donnison, I. S. (2008). Direct fermentation of fodder maize, chicory fructans and perennial ryegrass to hydrogen using mixed microflora. Bioresource Technology, 99(18), 8833–8839.

Kyazze, G., Martinez-Perez, N., Dinsdale, R., Premier, G. C., Hawkes, F. R., Guwy, A. J., & Hawkes, D. L. (2005). Influence of substrate concentration on the stability and yield of continuous biohydrogen production. Biotechnology and Bioengineering, 93(5), 971–979.

Lay, J. J., Lee, Y. J., & Noike, T. (1999). Feasibility of biological hydrogen production from organic fraction of municipal solid waste. Water Research, 33, 2579 –2586.

Lee, K.-S. S., Hsu, Y.-F. F., Lo, Y.-C. C., Lin, P.-J. J., Lin, C.-Y. Y., & Chang, J.-S. S. (2008a). Exploring optimal environmental factors for fermentative hydrogen production from starch using mixed anaerobic microflora. International Journal of Hydrogen Energy, 33(5), 1565–1572.

Lee, Z. K., Li, S. L., Lin, J. S., Wang, Y. H., Kuo, P. C., Cheng, S. S. (2008b). Effect of pH in fermentation of vegetable kitchen wastes on hydrogen production under a thermophilic condition. International Journal of Hydrogen Energy,33, 5234–5241.

Lee, W. S., Chua, A. S. M., Yeoh, H. K., & Ngoh, G. C. (2014). A review of the production and applications of waste-derived volatile fatty acids. Chemical Engineering Journal, 235, 83–99.

Lemos, P. C., Serafim, L. S. L. S., & Reis, M. A. M. (2006). Synthesis of polyhydroxyalkanoates from different short-chain fatty acids by mixed cultures submitted to aerobic dynamic feeding. Journal of Biotechnology, 122(2), 226–238.

Lewis, V., & Yang, S. T. (1992). Propionic acid fermentation by Propionibacterium acidipropionici: effect of growth substrate. Applied Microbiology and Biotechnology, 47, 615–622.

Li, S. L. L., Kuo, S. C. C., Lin, J. S. S., Lee, Z. K. K., Wang, Y. H. H., Cheng, S. S. S. (2008). Process performance evaluation of intermittent continuous stirred tank reactor for anaerobic hydrogen fermentation with kitchen waste. International Journal of Hydrogen Energy, 33 (5), 1522–1531.

Li, W.-W. W., & Yu, H.-Q. Q. (2011). From wastewater to bioenergy and biochemicals via two-stage bioconversion processes: A future paradigm. Biotechnology Advances, 29(6), 972–982.

67

Lim, S.-J. J., Kim, B. J., Jeong, C.-M. M., Choi, J. dal-rae, Ahn, Y. H., & Chang, H. N. (2008). Anaerobic organic acid production of food waste in once-a-day feeding and drawing-off bioreactor. Bioresource Technology, 99(16), 7866–7874.

Lin, C.-Y. Y., & Jo, C.-H. H. (2003). Hydrogen production from sucrose using an anaerobic sequencing batch reactor process. Journal of Chemical Technology & Biotechnology, 78(6), 678–684.

Liu, H., & Fang, H. H. P. (2002). Hydrogen production from wastewater by acidogenic granular sludge. Water Science & Technology, 47(1), 153–158.

Luo, G., Xie, L., Zou, Z., Wang, W., & Zhou, Q. (2010). Evaluation of pretreatment methods on mixed inoculum for both batch and continuous thermophilic biohydrogen production from cassava stillage. Bioresource Technology, 101(3), 959–964.

Ma, J., Carballa, M., Van De Caveye, P., & Verstraete, W. (2009). Enhanced propionic acid degradation (EPAD) system: Proof of principle and feasibility. Water Research, 43(13), 3239–3248.

Massanet-Nicolau, J., Guwy, A., Dinsdale, R., Premier, G., & Esteves, S. (2010). Production of hydrogen from sewage biosolids in a continuously fed bioreactor: Effect of hydraulic retention time and sparging. International Journal of Hydrogen Energy, 35(2), 469–478.

Mercier, P., Yerushalmi, L., Rouleau, D., & Dochain, D. (2007). Kinetics of lactic acid fermentation on glucose and corn by Lactobacillus amylophilus. Journal of Chemical Technology & Biotechnology, 55(2), 111–121.

Metcalf, Eddy, I., 2003. Wastewater Engineering: Treatment and Reuse. 4th Edition. McGraw-Hill, Boston.

Michel-Savin, D., Marchal, R., & Vandecasteele, J. P. (1990). Control of the selectivity of butyric acid production and improvement of fermentation performance Clostridium tyrobutyricum. Applied Microbiology Biotechnology, 32,387–392.

Min, K. S., Khan, A. R., Kwon, M. K., Jung, Y. J., Yun, Z., & Kiso, Y. (2005). Acidogenic fermentation of blended food-waste in combination with primary sludge for the production of volatile fatty acids. Journal of Chemical Technology & Biotechnology, 80(8), 909–915.

Mohan, V.S., Lalit B., V., Sarma, P. N. (2008) Effect of various pretreatment methods on anaerobic mixed microflora to enhance biohydrogen production utilizing dairy wastewater as substrate. Bioresource Technology, 99(1), 59–67.

Morgan-Sagastume, F., Pratt, S., Karlsson, A., Cirne, D., Lant, P., & Werker, A. (2011). Production of volatile fatty acids by fermentation of waste activated sludge pre-treated in full-scale thermal hydrolysis plants. Bioresource Technology, 102, 3089–3097.

Oh, S.-E. E., Van Ginkel, S., & Logan, B. E. (2003). The Relative Effectiveness of pH Control and Heat Treatment for Enhancing Biohydrogen Gas Production. Environmental Science & Technology, 37(22), 5186–5190.

Okamoto, M., Miyahara, T., Mizuno, O., Noike, T. (2000). Biological hydrogen potential of materials characteristic of the organic fraction of municipal solid wastes. Water Science and Technology, 41(3), 25–32.

Pant, D., Arslan, D., Van Bogaert, G., Alvarez G. Y., De Wever, H., Diels L., Vanbroekhoven, K. (2013). Integrated conversion of food waste diluted with sewage into volatile fatty acids through fermentation and electricity through a fuel cell. Environmental Technology, 34, 1935 – 1945.

Park, W., Hyun, S. H., Oh, S. E. E., Logan, B. E., Kim, I. S. (2005). Removal of headspace CO2 increases biological hydrogen production. Environmental Science Technology 39(12), 4416–4420.

68

Peralta-Yahya, P. P., Zhang, F., del Cardayre, S. B., & Keasling, J. D. (2012). Microbial engineering for the production of advanced biofuels. Nature, 488(7411), 320–328.

Pham, T.N., Nam, W.J., Jeon, Y.J., Yoon, H.H. (2012).Volatile fatty acids production from marine macroalgae by anaerobic fermentation. Bioresource Technology, 124, 500–503.

Pratt, S., Liew, D., Batstone, D. J., Werker, A. G., Morgan-Sagastume, F., & Lant, P. A. (2012). Inhibition by fatty acids during fermentation of pre-treated waste activated sludge. Journal of Biotechnology, 159(1-2), 38–43.

Ren, N., Guo, W., Wang, X., Xiang, W., Liu, B., Ding, J., Chen, Z. (2008) Effects of different pretreatment methods on fermentation types and dominant bacteria for hydrogen production. Int. J. Hydrogen Energy, 33 (16), 4318–4324

Shin, H.-S. S., Youn, J.-H. H., & Kim, S.-H. H. (2004). Hydrogen production from food waste in anaerobic mesophilic and thermophilic acidogenesis. International Journal of Hydrogen Energy, 29(13), 1355–1363.

Smith, D. P., & McCarty, P. L. (1989). Reduced product formation following perturbation of ethanol- and propionate-fed methanogenic CSTRs. Biotechnology and Bioengineering, 34(7), 885–895.

Steinbusch K. J. J., Hamelers, H. V. M., Buisman, C. J. N. (2008). Alcohol production through volatile fatty acids reduction with hydrogen as electron donor by mixed cultures. Water Research, 42, 4059–4066.

Steinbusch, K. J. J., Hamelers, H. V. M., Plugge, C. M., & Buisman, C. J. N. (2011). Biological formation of n-caproate and caprylate from acetate: fuel and chemical production from low grade biomass. Energy & Environmental Science, 4(1), 216–224.

Tao, Y., Chen, Y., Wu, Y., He, Y., & Zhou, Z. (2007). High hydrogen yield from a two-step process of dark- and photo-fermentation of sucrose. International Journal of Hydrogen Energy, 32(2), 200–206.

Temudo, M. F., Kleerebezem, R., & van Loosdrecht, M. (2007). Influence of the pH on (open) mixed culture fermentation of glucose: A chemostat study. Biotechnology and Bioengineering, 98(1), 69–79.

Ten Brummeler, E., Horbach, H. C. J. M., & Koster, I. W. (1991). Dry anaerobic batch digestion of the organic fraction of municipal solid waste. Journal of Chemical Technology & Biotechnology, 50(2), 191–209.

Thanakoses, P., Black, A. S., & Holtzapple, M. T. (2003). Fermentation of corn stover to carboxylic acids. Biotechnology and Bioengineering, 83, 191 – 200.

Ueno, Y., Fukui, H., & Goto, M. (2007). Operation of a Two-Stage Fermentation Process Producing Hydrogen and Methane from Organic Waste. Environmental Science & Technology, 41(4), 1413–1419.

Ueno, Y., Kawai, T., Sato, S., Otsuka, S., Morimoto, M. (1995). Biological production of hydrogen from cellulose by natural anaerobic microflora. Journal Fermentation Bioengineering 79(4), 395–397.

Ueno, Y., Otsuka, S., & Morimoto, M. (1996). Hydrogen production from industrial wastewater by anaerobic microflora in chemostat culture. Journal of Fermentation and Bioengineering, 82(2), 194–197.

Valdez-Vazquez, I., & Poggi-Varaldo, H. M. H. M. (2009). Hydrogen production by fermentative consortia. Renewable and Sustainable Energy Reviews, 13(5), 1000–1013.

Valdez Vazquez, I., Riosleal, E., Esparzagarcia, F., Cecchi, F., Poggivaraldo, H. (2005). Semi-continuous solid substrate anaerobic reactors for H production from organic waste: Mesophilic versus thermophilic regime. International Journal of Hydrogen Energy, 30(13-14), 1383–1391.

69

Van Den Heuvel, J. C., & Beeftink, H. H. (1988). Kinetic effects of simultaneous inhibition by substrate and product. Biotechnology and Bioengineering, 31(7), 718–724.

Van den Heuvel, J., Verschuren, P., Beeftink, H., & de Beer, D. (1992). Determination of the critical concentration of inhibitory products in a repeated fed-batch culture. Biotechnology Techniques, 6(1), 33–38.

Veeken, A., Kalyuzhnyi, S., Scharff, H., & Hamelers, B., A. (2000). Effect of pH and VFA on hydrolysis of organic solid waste. Journal of Environmental Engineering, 126(12), 1076–1081.

Veeken, A., & Hamelers, B. (1999). Effect of temperature on hydrolysis rates of selected biowaste components. Bioresource Technology, 69(3), 249–254.

Wang, B., Wan, W., & Wang, J. (2008). Inhibitory effect of ethanol, acetic acid, propionic acid and butyric acid on fermentative hydrogen production. International Journal of Hydrogen Energy, 33(23), 7013–7019.

Wang, B., Li, Y., Ren, N. (2013). Biohydrogen from molasses with ethanol-type fermentation: Effect of hydraulic retention time. International Journal of Hydrogen, 38(11), 4361–4367.

Wang, J., & Wan, W. (2008). Comparison of different pretreatment methods for enriching hydrogen-producing bacteria from digested sludge. International Journal of Hydrogen Energy, 33(12), 2934–2941.

Wang, J., & Wan, W. (2009). Factors influencing fermentative hydrogen production: A review. International Journal of Hydrogen Energy, 34(2), 799–811.

Wang, L., Zhou, Q., & Li, F. T. (2006). Avoiding propionic acid accumulation in the anaerobic process for biohydrogen production. Biomass and Bioenergy, 30(2), 177–182.

Wang, Y.-Y. Y., Ai, P., Hu, C.-X. X., & Zhang, Y.-L. L. (2011). Effects of various pretreatment methods of anaerobic mixed microflora on biohydrogen production and the fermentation pathway of glucose. International Journal of Hydrogen Energy, 36(1), 390–396.

Wu, S.-Y. Y., Hung, C.-H. H., Lin, C.-N. N., Chen, H.-W. W., Lee, A.-S. S., & Chang, J.-S. S. (2005). Fermentative hydrogen production and bacterial community structure in high-rate anaerobic bioreactors containing silicone-immobilized and self-flocculated sludge. Biotechnology and Bioengineering, 93(5), 934–946.

Yu, H. Q., & Fang, H. H. P. (2003). Acidogenesis of gelatin-rich wastewater in an upflow anaerobic reactor: influence of pH and temperature. Water Research, 37(1), 55–66.

Yu, H.-Q. Q., & Mu, Y. (2006). Biological hydrogen production in a UASB reactor with granules. II: Reactor performance in 3-year operation. Biotechnology and Bioengineering, 94(5), 988–995.

Yuan, Q., Sparling, R., & Oleszkiewicz, J. A. (2011). VFA generation from waste activated sludge: Effect of temperature and mixing. Chemosphere, 82(4), 603–607.

Zhang, C., Yang, H., Yang, F., & Ma, Y. (2009). Current Progress on Butyric Acid Production by Fermentation. Current Microbiology, 59(6), 656–663.

Zhao, Q.-B. B., & Yu, H.-Q. Q. (2008). Fermentative H2 production in an upflow anaerobic sludge blanket reactor at various pH values. Bioresource Technology, 99(5), 1353–1358.

Zhu, H., Parker, W., Basnar, R., Proracki, A., Falletta, P., Béland, M., & Seto, P. (2008). Biohydrogen production by anaerobic co-digestion of municipal food waste and sewage sludges. International Journal of Hydrogen Energy, 33(14), 3651–3659.

70

I can believe anything provided that it is quite incredible.

Oscar Wilde

71

Chapter 3

Effect of hydrogen and carbon dioxide on carboxylic acids patterns in mixed culture fermentation

This chapter has been published as:

D. Arslan, K.J.J. Steinbusch, L. Diels, H. De Wever, C.J.N. Buisman, H.V.M. Hamelers (2012). Effect of Hydrogen and Carbon Dioxide on Carboxylic Acids Patterns in Mixed Culture Fermentation. Bioresource Technology, 118, 227 – 234.

72

Abstract

This study investigated the carboxylate spectrum from mixed culture fermentation of 3 organic

waste streams after supplying 2 bar hydrogen and carbon dioxide or a mixture of these 2 gases to

the headspace. Under any modified headspace, propionate production was ceased and n-butyrate,

n-caproate and the total carboxylate concentrations were higher than in the reactors with N2

headspace (control). Production of one major compound was achieved under hydrogen and

carbon dioxide mixed headspace after 4 weeks of incubation. Both the highest acetate

concentration 17.4 g COD/l and the highest fraction 87% were observed in reactors with mixed

hydrogen and carbon dioxide headspace independent of the substrate used. In the control reactor,

acetate made up maximum 67% of the total products. For other products, the highest

concentration and fraction were seldom observed together. Selective n-butyrate production

reaching 75% fraction was found under the carbon dioxide headspace on the carbohydrate rich

waste.

73

3.1. Introduction

The chemical industry needs to consider alternative, renewable sources to reduce both

dependency on oil reserves and to decrease carbon dioxide emission into the environment

(Ragauskas et al., 2006). Biomass is an important resource for the production of industrial bulk

chemicals. Among the variety of biomass resources, organic waste streams are the most

interesting ones because they are abundantly available, cheap and they do not compete with food

(Tilman et al., 2009). Organic wastes can be biologically converted into carboxylates, such as

acetate, propionate and n-butyrate under anaerobic conditions without need of costly sterilization

operations as in pure culture fermentations. Organic waste materials generally contain a mixture

of carbohydrate, protein and lipid polymers. Because of this complexity, organic wastes can only

be transformed into the desired products by using a mixed culture.

The first step in a mixed culture anaerobic process of organic wastes is hydrolysis of

carbohydrates to monosaccharides, proteins to amino acids and lipids to glycerol and long chain

fatty acids. After hydrolysis, these monomers are converted to carboxylates, hydrogen and carbon

dioxide during acidogenesis (De Mes et al., 2003). Carboxylates are individually valuable products

and they can be used in industry after separation from the fermentation broth and further

purification. Carboxylates can however also be used in a secondary fermentation step to generate

biofuels and poly- -hydroxyalkanoates (Agler et al., 2011; Albuquerque al., 2011; Holtzapple &

Granda, 2009).

Because of the diverse polymer composition of organic waste materials and the use of

mixed cultures, anaerobic processes generate a mixture of various compounds rather than one

single product. Furthermore, because of a high water content of the waste source, the final

products are generated in diluted concentrations. Under these conditions, downstream processing

74

of carboxylates requires a considerable amount of energy that negatively affects the production

costs of the desired products. As a result, current biological production from waste materials

cannot yet compete with chemical production from fossil oil in the market. Processes, which

produce chemicals from organic waste materials, can only become competitive with production

from fossil sources if the process can generate a single chemical and in a high concentration.

Up to now, many studies have been applying fermentation with different organic waste

streams, such as artificial food waste (Komemoto et al., 2009), manure (Coats et al., 2011), dairy

wastewater (Bengtsson et al., 2008), paper waste (Ueno et al.,2007), organic fraction of municipal

solid waste (Cavdar et al., 2011). These tests were conducted under various operational

conditions, such as pH, temperature and organic load in continuous or batch-wise mode. Among

the parameters which have been studied up to today, the effect of headspace composition on

selective production of one carboxylate in high concentrations was not systematically investigated,

while we have indications that certain fermentation pathways can be energetically suppressed or

favored by hydrogen and carbon dioxide levels in the headspace (Thauer et al., 1977). Indeed,

some studies investigated the effect of headspace, particularly hydrogen partial pressure, on

anaerobic process, but these studies only focused on biological hydrogen production (Zhang et al.,

2011; Kraemer & Bagley, 2007). To the best of our knowledge the effect of hydrogen and carbon

dioxide on carboxylate formation has not been investigated yet.

In this study, the objective was to direct mixed culture fermentation of organic waste

towards a single product in sufficiently high concentrations by manipulating the headspace. This

study aimed to reach a production of one major compound making up more than 50% of the total

products. To this end, carboxylate production from 3 organic waste streams of different polymeric

compositions, was monitored in terms of total fermentation product concentration as well as

75

relative composition after supplying 2 bar hydrogen and carbon dioxide or a mixture of these two

gases to the headspace.

3.2. Materials and Methods

3.2.1. Inocula and waste streams

Granular sludge from an anaerobic digester of the potato processing industry was selected

as inoculum (Opure, Ede, the Netherlands). The granular sludge was washed with 20 mM

potassium phosphate buffer (at pH 5) and sieved with a mesh of 500 μm for three times. After the

last washing step, the granules were left in the same buffer solution overnight at room

temperature. The next day, the granular sludge, with a 9% dry matter content and 11% ash

content, was heat treated by boiling in water for 15 min to avoid methanogenesis. As carbon

source, instead of pure polymers, real organic waste streams from industry were used. Organic

waste streams were collected from potato, meat and oil processing companies.

3.2.2. Experimental set-up

PVC water-jacketed reactors (5 L) were filled with 750 ml basal medium. The basal medium

was prepared according to Phillips et al. (1993) without sulfate to prevent sulfate reduction.

Ammonium concentration in the medium was adjusted according to the carbon source used to

keep the carbon to nitrogen ratio at 10. The pH of the medium was set to 5 using 4 mM

monopotassium phosphate and 4 mM dipotassium phosphate buffers without changing the total

phosphorus concentration in the medium. After substrate addition to a concentration of 8 g COD/l,

each reactor was inoculated with 8 g of pre-treated granular sludge (wet weight). Then, the pH

was corrected to 5 with 2 M NaOH or HCl. The experiment with meat processing waste was

controlled with 2 M NaOH and HCl at pH 5 throughout the experiment to avoid pH increase by

ammonia release from proteins. All reactors were adjusted to a final headspace pressure of 2 bar

76

by adding hydrogen, carbon dioxide or a 50:50 (v:v) mixture of these two gases. Pressure of the

headspace of each reactor was monitored on-line with a manometer. During the experiment,

pressure drops were noted in the reactors with mixed hydrogen and carbon dioxide headspace. In

such case, headspace pressure was readjusted to 2 bar by adding an appropriate amount of 50:50

(v:v) hydrogen and carbon dioxide mixture. A control reactor was prepared under the same

conditions, but was flushed with nitrogen at the start of the experiment and left under

atmospheric pressure. All reactors were kept at 30°C by hot water circulation and mixed at 100

rpm by a magnetic drive stirrer. The incubation period was 30 days. Gas and liquid samples were

taken weekly. The headspace to liquid sampling ratio was chosen such that no underpressure was

caused until the end of the experiment.

3.2.3. Analysis

The headspace was sampled through a rubber septum using a gas-tight plastic syringe and

analyzed in a gas chromatograph (Trace GC Ultra) equipped with a thermal conductivity detector

(TCD). H2 content was analyzed with a 2 m stainless steel column packed with molecular sieve 5A

(80/100 mesh) using nitrogen as carrier gas at a flow rate of 20 ml/min. Methane, carbon dioxide,

oxygen and nitrogen were analyzed with a 2 m stainless steel HayeSepQ (80/100 mesh) and

molecular sieve 5A (80/100 mesh) column connected in series with a valve. Helium was used as

carrier gas at a flow rate of 15 ml/min.

Before being analyzed liquid samples were centrifuged for 5 min at 10000 rpm to remove

particles. Volatile fatty acid (VFA) samples were acidified with 1:1 (v/v) H2SO4 solution and then

extracted with diethyl ether. VFAs were analyzed in a GC (CE Instruments-Thermoquest) equipped

with a flame ionization detector (FID) and a 15 m AT-1000 filled capillary column (0.53 mm x 1.2

μm). Helium was used as the carrier gas at a constant flow of 6 ml/min. Acetone, ethanol and

propanol were analyzed in the headspace of the samples at 60°C with the same GC fitted in split

77

mode with a split/splitless injector and AT-WAX capillary column (60m x 0.32mm x 1.00μm).

Carrier gas was helium with a gas flow rate of 1.6 ml/min. Lactate was analyzed with a GC

(Interscience, CE Instruments-Trace GC) equipped with FID detector in split mode with

split/splitless injector and AT-1 capillary column (30m x 0.53mm x 5μm column). Each sample was

treated with Ce(SO4)2 and kept at 60°C for 10 min to convert lactic acid into acetaldehyde which

was then analyzed in the headspace.

Each waste was analyzed with the Phenol-sulfuric method (Dubois et al., 1956) for its

glucose content, the Total Kjeldahl Nitrogen (TKN) method with conversion factor of 6.25 (Pierce

& Haenisch, 1948) for its protein content; and the Soxhlet extraction method (Pomeranz & Meloan,

1994) for its lipid content. Total and volatile solid contents were analyzed as described in Standard

Methods (APHA, 1995). Total Chemical Oxygen Demand (CODt) was determined with Dr. Hach

Lange cuvette tests, soluble COD (CODs) and NH4-N concentrations of each sample were

determined with Dr. Hach Lange cuvette tests after centrifugation.

3.2.4. Calculations

Part of the initial CODt added to the reactor is converted to products. The remaining part of

CODt either stays in the same form of the initially added COD or is liquefied. In a balanced system,

the initial amount of CODt, added to a reactor at start-up, should be recovered in various forms at

the end of the experiment. COD recovery of the system was calculated with

CODrecovery (%) = (CODt_final/CODt_initial) * 100 (3.1)

where CODt_final and COD_initial is the sum of measured CODt in the liquid (with solid

particles) and in the gas phase at the end and start of the experiment (g COD).

The hydrolysis degree represents the portion of initially added solid particles transformed

into a soluble form. The hydrolysis degree of the system was calculated with

78

Hydrolysis degree (%)= (CODsfinal-CODsinitial)-(|CODH2initial-CODH2final|)CODsolidinitial

*100 (3.2)

where CODs_final and CODs_initial is the measured CODs at the end and start of the

experiment (g COD), CODH2_initial and CODH2_final is the measured COD of H2 at the start and at

the end of the experiment (g COD) and CODsolid_initial is the difference between the measured

values of CODt and CODs at the start of the experiment (g COD).

Fraction of single product produced in the system was calculated with

Fraction (%) = (COD_product/COD_all products) * 100 (3.3)

where COD_product is the COD calculated from the measured concentration of the single

product produced in the system (g COD) and COD_all products is the sum of all calculated product

COD values in the system (g COD).

3.3. Results and Discussion

The objective of this study was to direct mixed culture fermentation towards a single

product in sufficiently high concentrations by supplying 2 bar of 100% hydrogen, 100% carbon

dioxide or a 50:50 (v:v) mixture of these gases in the headspace. These conditions were

considered to represent an adequate range for studying the impact of headspace composition on

the fermentation process and product composition. As a carbon source, 3 real organic waste

streams different in carbohydrate, lipid and protein composition were selected to observe the

individual effect of polymer compositions on product concentration and type. Each headspace

composition was run with 3 different waste streams.

The compositions of the 3 waste streams are presented in Table 3.1. Conform expectation,

the waste stream from the potato processing company is rich in carbohydrates, the one from the

meat processing company in proteins and the one from the oil processing company in lipids.

79

Therefore, they were considered suitable for the test purposes. They are denominated

carbohydrate waste, protein waste and lipid waste from now on.

Table 3.1. Characteristics of waste streams*

Waste streams

Analysis PotatoProcessing

MeatProcessing

OilProcessing

TS (mg/g) 36.7 (1.2) 70.0 (0.8) 971 (2.8)

VS (mg/g) 30.8 (4.5) 65.5 (1.6) 971 (0)

Carbohydrate (g/kg) 22 (16) 1.8 (0.08) 4.3 (0.3)

Protein (g/kg) 5.9 (0.2) 51.5 (4.6) 0

Lipid (g/kg) 0.2 (0) 13.3 (0) 996 (0.5)

Total COD (g/kg) 26.4 (0.7) 111 (10.2) 1729 (325)

*Concentrations are mean values and standard deviations are shown in brackets. Results are presented on wet weight basis.

A COD recovery close to 100% of a system shows a good balance of the process. In this

study, the COD recoveries of the reactors were determined by equation 3.1. The COD recoveries

for the carbohydrate waste experiment were between 96% and 111%. Figure 3.1 illustrates the

COD balance in the carbohydrate waste reactor with hydrogen headspace. COD recoveries for the

protein waste experiment were between 80% and 115%. Only in the reactors with lipid waste,

COD recoveries were lower than 85%. The low COD recovery in the lipid waste experiment is

probably caused by the difficulties in handling the viscous waste and taking nonhomogenous

samples, as well as poor hydrolysis of lipids during fermentation.

80

Figure 3.1. COD balance in the reactor with a modified hydrogen headspace for carbohydrate

waste.

For all conditions, carboxylate concentrations were higher than alcohol concentrations.

Acetate, n-butyrate and propionate were the main carboxylates depending on the headspace

composition and the substrate type. In few cases, the COD fraction of i-valerate and n-caproate

was above 10% of the total COD of products. Since this study focuses on production of a single

compound, products fractions lower than 10% will not be discussed further. In the experiments

with the carbohydrate and protein wastes, lactate was completely consumed within 1 week

incubation period. On the lipid waste, lactate was produced, but in negligible concentrations.

Because the fractions of the products varied in time, the results are presented after 1 week and 4

weeks incubation periods. Methane was not detected in any of the experiments.

3.3.1. Effect of substrate on the anaerobic process in the control reactors

First, the effect of substrate on the fermentation product type and concentration was

investigated. The highest total carboxylate production was obtained with the carbohydrate waste

and amounted to 4.9 g COD/l. For the protein waste, total carboxylate concentration was 2.0 g

COD/l and for the lipid waste only 0.1 g COD/l. The relatively low carboxylate concentration in the

Anaerobicreactor

H2

CODt_final = 11.9 g CODCODt_initial = 13 g COD

CODin gas = 5.8 g COD

CODin solid = 4.9 g COD

CODin liquid = 2.3 g COD

CODout gas = 4.4 g COD

CODout liquid = 6.6 g COD

CODout solid = 0.9 g COD

Recovery = 92 %

81

lipid waste compared to other waste types is caused by the low hydrolysis constant for lipids

compared to proteins and carbohydrates (Christ et al., 2000). Since hydrolysis is often the rate

limiting step of the overall anaerobic process, poor hydrolysis rates yield low product

concentrations (Mata-Alvarez et al., 2000).

Figure 3.2 shows the 3 major product concentrations of the control reactors in the

carbohydrate, protein and lipid waste experiments. From the produced carboxylates, acetate was

the major product for all the control reactors after 4 weeks of incubation. However the acetate

concentrations and fractions varied with substrate type. The maximum acetate concentration was

2.2 g COD/l at 43% of the total carboxylates on the carbohydrate waste. Although the

concentration of acetate, 1.3 g COD/l, on protein waste was lower than on carbohydrate waste, its

relative fraction was higher reaching 67%. With the lipid waste, concentrations of acetate were

lower than 80 mg COD/l.

The results show that although carbohydrate waste produced higher carboxylate

concentrations compared to protein and lipid wastes, product formation was not oriented towards

one major carboxylate under a regular headspace, while the protein and lipid waste generated

mainly acetate. On the carbohydrate waste, the second major carboxylate was propionate.

Propionate concentration reached 1.7 g COD/l constituting a 33% fraction. n-Butyrate

concentration was lower than propionate concentration and amounted to 0.9 g COD/l constituting

a 17% fraction. Propionate and n-butyrate concentrations obtained on the protein waste were

negligible.

82

Figure 3.2. Concentrations of acetate, n-butyrate and propionate as a function of time under different headspace conditions for the (a) carbohydrate, (b) protein and (c) lipid rich waste experiments. Scales are different in each graph.

83

The variable product formation on the carbohydrate waste and the single type of product

formation on the protein waste might be caused by pH differences in these experiments. In the

carbohydrate waste tests, the reactors were run without pH control and the pH dropped to 4.2 for

the control reactor (Table 3.2).This shift in pH may have lead to variable products formation. In

the test with the protein waste, the reactors were kept at pH 5.0 to avoid alcalinization because of

ammonia release from proteins. Keeping the pH at a fixed value of 5.0 may have led to production

of a single major carboxylate, namely acetate. It is worth to note here that without pH control in

the reactors with the protein waste, a pH increase above 5 would probably have occurred leading

to a similar kind of discussion. Some references indicate that the fermentation process is more

directed to acetate production at pH higher than 5 and to n-butyrate production at pH lower than

5 (e.g. Fang & Zhang, 2006) while others rather point to the opposite (e.g. Rodriguez et al.,

2005). Still, it is difficult to draw a certain conclusion about the relationship between pH and

product type from former studies. The reason is that these studies worked with different organic

substrates with different inocula.

Gas production was negligible in the control reactors. Therefore, the effect of produced

hydrogen and carbon dioxide on the process can be accepted as being minor. Hydrogen

production in the carbohydrate waste test was 28 mg COD/l-headspace and in the protein waste

test was 8 mg COD/l-headspace and production only occurred during the 1st week of incubation.

After the 1st week, the produced hydrogen was completely consumed in both experiments. Carbon

dioxide production continued until the end for both the carbohydrate and protein waste tests

reaching approximately 4 mM. Hydrogen was not produced at all during fermentation of the lipid

waste and carbon dioxide production was only 1.5 mM.

84

Tabl

e 3.

2.Co

ncen

trat

ion

chan

ge o

f pr

oduc

ts a

nd f

ract

ions

(in

bra

cket

s) f

or e

ach

head

spac

e co

nditi

on in

the

exp

erim

ent

with

car

bohy

drat

e w

aste

af

ter

1 w

eek

and

4 w

eeks

of

incu

batio

n.

H2

H2

& C

O2

CO2

Cont

rol (

N2)

H2

H2

& C

O2

CO2

Cont

rol (

N2)

Prod

ucts

(m

g CO

D/l)

Afte

r 1

wee

kAf

ter

4 w

eeks

Acet

ate

1468

(31

%)

114

(5%

)-3

7a

1410

(38

%)

2563

(33

%)

9230

(66

%)

1850

(34

%)

2197

(43

%)

Prop

iona

te55

(1%

)13

7 (5

%)

86 (

4%)

1291

(34

%)

217

(3%

)28

2 (2

%)

360

(7%

)16

58 (

33%

)

n-Bu

tyra

te24

33 (

51%

)18

34 (

73%

)14

53(7

5%)

655

(17%

)31

40 (

41%

)24

34 (

18%

)25

37 (

47%

)86

5 (1

7%)

n-Ca

proa

te39

8 (8

%)

5 (<

1%)

2 (<

1%)

083

4 (1

1%)

262

(2%

)27

3 (5

%)

39 (

1%)

Tota

l VFA

4430

(93

%)

2148

(85

%)

1577

(81

%)

3379

(90

%)

7205

(93

%)

1226

2 (8

9%)

5356

(98

%)

4898

(97

%)

Etha

nol

208

(4%

)21

1 (8

%)

187

(10%

)35

8 (1

0%)

368

(5%

)10

62 (

8%)

29 (

1%)

129

(3%

)

Lact

ate

-927

a-9

55a

-954

a-9

34a

-938

a-9

04a

-964

a-9

53a

Hyd

roly

sis

degr

ee (

%)

3915

1038

6153

4951

COD

rec

over

y (%

)92

9097

106

pH4.

55.

05.

04.

24.

43.

74.

54.

2

aN

egat

ive

valu

es in

dica

te c

onsu

mpt

ion

of c

ompo

unds

.

Prod

ucts

nev

er o

ccur

ring

in fr

actio

ns lo

wer

tha

n 10

% a

re n

ot s

how

n.

85

3.3.2. Effect of headspace composition on anaerobic processes

We can distinguish two effects of headspace composition on anaerobic processes: (i) the

effect on the hydrolysis step and (ii) the effect on the acidogenesis step. The increase in

concentration of fermentation products depends on the efficiency of hydrolysis and the type of

products formed depends on the conditions of acidogenesis. Both effects will be discussed

separately in the following sections.

Effects of headspace on hydrolysis

Hydrolysis is the first step of the anaerobic digestion process and it is often considered as

the rate limiting step of the anaerobic process (Mata-Alvarez et al., 2000). A high hydrolysis

degree can therefore enhance the overall process and increase total carboxylate production. The

hydrolysis rate has been described before as a function of factors such as pH, temperature,

substrate type and loading, particle size of substrate and concentrations of intermediate products

(Veeken and Hamelers, 1999). In this study, the effect of hydrogen, carbon dioxide or a mixture of

these two gases on hydrolysis of the carbohydrate and the protein wastes was studied. The

hydrolysis degree was calculated with equation 3.2. Since bottlenecks in the hydrolysis of lipids

were discussed previously, the results with the lipid waste are not discussed further in this section.

Hydrolysis degree on the carbohydrate waste after 1 week and 4 weeks of incubation are

presented in Table 3.2. The rates of hydrolysis degree were clearly lower in the reactors under a

modified carbon dioxide headspace at 1 week incubation. After 1 week, the hydrolysis degrees on

the carbohydrate waste were the lowest in the reactors with the mixture of hydrogen and carbon

dioxide and only carbon dioxide. To the best of our knowledge, such an effect has never been

described in literature before. A maximum hydrolysis degree of the carbohydrate waste was

obtained in the reactor with hydrogen headspace at the end of the 4 weeks incubation period. For

86

the headspace conditions of the mixture of hydrogen and carbon dioxide, only carbon dioxide

headspace and the control reactor hydrolysis degrees reached similar values at the end of the

experiment.

Hydrogen or carbon dioxide had no effect on the hydrolysis degree of protein waste. In all

reactors fed with the protein waste, the hydrolysis degrees were approximately 30% after 1 week

(data not shown). At the end of the experiment, the hydrolysis degrees increased slightly in all

headspace conditions including the control reactor, to a maximum of 34% in the reactor with

carbon dioxide headspace.

Effect of headspace composition on product type

It was anticipated that product formation could be steered by manipulating hydrogen and

carbon dioxide in the headspace and that the effect would be different for different substrate

compositions.

Addition of hydrogen and carbon dioxide to the headspace had a clear effect on the

product type and concentration. Table 3.2 shows the total VFA and main carboxylate

concentrations at each headspace condition for the carbohydrate waste. It is observed that in the

reactors with hydrogen and carbon dioxide or a mixture of these gases supplied in the headspace,

the total VFA concentration increase was higher than in the controls. Since the final hydrolysis

degree was similar in all cases, the difference in total VFA production seems to indicate different

fermentation kinetics with variable headspace compositions.

For each substrate type, the total carboxylate concentration in the reactor with hydrogen

plus carbon dioxide headspace was higher than in those with solely hydrogen or carbon dioxide or

in the control. The highest carboxylate concentration increase was 18.8 g COD/l and was obtained

on protein waste (data not shown).

87

For each substrate type, the main carboxylate in the reactors with 2 bar hydrogen and

carbon dioxide headspace was acetate. The highest acetate concentration was 17.4 g COD/l with a

87% fraction obtained on the protein waste (Figure 3.2). Acetate production in these specific

reactors continued throughout the experiment and its concentration exceeded the CODt amount

which was added initially as substrate. The additional acetate thus had to be produced from

hydrogen and carbon dioxide present in the headspace by homoacetogens (equation 3.1 in Table

3.3). The consumption of hydrogen and carbon dioxide from the headspace in these reactors also

proves the homoacetogenic activity (Figure 3.3). In the lipid waste experiment, headspace

consumption was negligible (not shown). Homoacetogens produce acetate from hydrogen and

carbon dioxide available in the gas phase (Drake et al., 2002). Acetate production with a mixed

inoculum via the homoacetogenesis process has previously been shown by Nie et al. (2007). This

case will thus not be discussed further.

On carbohydrate waste both acetate and n-butyrate concentrations could be increased by

supplying 2 bar hydrogen to the headspace compared to the control reactor. However, selective

production towards a single compound was not found under this headspace condition (Figure 3.2

and Table 3.2). n-Butyrate and acetate reached similar concentrations under 2 bar hydrogen

during 1 month of experiments for every tested waste type. The highest acetate concentration of

2.6 g COD/l and the highest n-butyrate concentration of 3.1 g COD/l were both obtained with the

carbohydrate waste. Acetate concentrations in the experiments with carbohydrate and lipid waste

were slightly higher than in the control reactors while the opposite was true for the protein waste

(Figure 3.2). Apparently, production of acetate from protein waste was not favorable under

elevated hydrogen pressure in headspace. Here, it should be reminded again that only for protein

waste, the pH was controlled at 5.

88

Figure 3.3. Concentration change of hydrogen and carbon dioxide in the reactors with hydrogen and carbon dioxide mixed headspace for the (a) carbohydrate and (b) protein rich waste experiments. Arrows indicate when additional 50:50 (v:v) hydrogen and carbon dioxide mixture was introduced into the headspace to a final concentration of 45 mM for each gas.

89

Tabl

e 3.

3. E

quat

ions

for

diff

eren

t fe

rmen

tatio

nre

actio

ns

Reac

tion

Reac

tion

equa

tion

Gr°

'

(kJ/

mol

)a

Equa

tion

no.

Acet

ate

prod

uctio

n by

hom

oace

toge

ns

2CO

2+

4H

2ac

etat

e-+

H+

+ 2

H2O

-138

.23.

1

n-Bu

tyra

te p

rodu

ctio

n fr

om la

ctat

e2

lact

ate-

+ H

+n-

buty

rate

-+

2CO

2+

2H

2-7

1.8

3.2

n-Bu

tyra

te p

rodu

ctio

n fr

om la

ctat

e an

d ac

etat

ela

ctat

e-+

ace

tate

+ H

+

n-bu

tyra

te-+

CO

2+

H2

-21.

33.

3

Acet

ate

prod

uctio

n fr

om n

-but

yrat

en-

buty

rate

- + H

++

2H

2O2

acet

ate-

+ 2

H+

+ 2

H2

+48

.13.

4

Prop

iona

te p

rodu

ctio

n fr

om g

luco

seC 6

H12

O6

+ 2

H2

2 pr

opio

nate

-+

2H

++

2H

2O-3

08.5

3.5

Acet

ate

prod

uctio

n fr

om g

luco

seC 6

H12

O6

+ 2

H2O

2 ac

etat

e-+

2H

++

4H

2+

2CO

2-2

92.3

3.6

n-Bu

tyra

te p

rodu

ctio

n fr

om g

luco

seC 6

H12

O6

n-bu

tyra

te-+

H+

+ 2

H2

+ 2

CO2

-309

.43.

7

aFr

ee e

nerg

y fo

r th

e re

actio

n (

Gr°

') va

lues

is c

alcu

late

d at

30°

C an

d a

pH o

f 7

unde

r 1

atm

pre

ssur

e. C

once

ntra

tions

of

reac

tant

s an

d pr

oduc

ts in

aqu

eous

so

lutio

n ar

e 1

M.

90

In fact, in the reactor with hydrogen headspace, the n-butyrate concentration and fraction

was higher than acetate after the 1st week of incubation on the carbohydrate waste (Table 3.2).

After the 1st week, the acetate production rate apparently became faster than the n-butyrate

production rate. Hydrogen consumption from the headspace started at the end of the 2nd week of

incubation. From that point on, n-caproate production, albeit in negligible concentrations,

continuously increased while acetate and n-butyrate concentrations were stable. The highest

concentration and fraction of n-caproate was found under hydrogen modified headspace. In these

reactors, hydrogen consumption was 0.5 bar in the carbohydrate waste and 0.25 bar in the protein

waste experiment. Hydrogen consumption was not detected in the reactor with lipid waste.

The presented results indicate that the final product type and concentrations were not only

determined by either the effect of hydrogen and carbon dioxide added to the headspace or the

effect of substrates, but also by the combined effect of headspace and polymer source in the

substrate.

For the reactors with the carbon dioxide headspace, our findings show that a carbon

dioxide headspace favors the production of n-butyrate from the carbohydrate type of waste

whereas acetate is the main product for the other two waste streams (Figure 3.2). The highest n-

butyrate concentration increase was 2.5 g COD/l and it was 3 times higher than in the control

reactor. In fact, it is important to note that while the n-butyrate concentration obtained in this

reactor was in absolute terms lower than in the reactor with hydrogen headspace, a higher n-

butyrate fraction of 47% was obtained under a carbon dioxide headspace. Carbon dioxide pressure

in the headspace was stable during all experiments. For protein and lipid wastes, both acetate and

n-butyrate concentrations were higher than in the control reactors (Figure 3.2). However,

concentration levels were quite low (< 1.6 g COD/l).

91

The n-butyrate concentrations were higher under all modified headspace compared to the

control reactor. Furthermore fractions reached above 70% in the reactors with carbon dioxide and

with mixed hydrogen and carbon dioxide headspace after a 1 week test period for the

carbohydrate waste while it was only 17% in the control reactor. High n-butyrate concentrations

relative to other fermentation products might be linked to the lactate consumed during the 1st

week of the experiment. Several intermediate fermentation reactions such as lactate oxidation to

n-butyrate (equation 3.2 in Table 3.3) or n-butyrate production via consumption of acetate and

lactate (equation 3.3 in Table 3.3) were previously presented in Agler et al. (2011). However, in

the reactors showing lactate consumption, the amount of n-butyrate produced exceeded the

amount of lactate consumed. Hence, n-butyrate production cannot only be explained by oxidation

of lactate. Similarly, the amounts of acetate and lactate consumed in the carbon dioxide

headspace reactor did not correspond to equation 3.3 in Table 3.3. Because also no carbon dioxide

consumption or production was observed in the headspace of the reactors, carbon dioxide must

have affected the fermentation reaction mechanism in a different way. Previously, effects of

carbon dioxide sparging on a continuous fermentation process fed with sucrose were studied by

Kim et al. (2006). They reached a n-butyrate concentration of 8.2 g COD/l with a fraction of 71%

of total carboxylates produced in reactor sparged with carbon dioxide compared to 6.1 g COD/l

with a fraction of 40% in reactor sparged with nitrogen. According to Kim et al. (2006), a carbon

dioxide partial pressure higher than 0.7 bar decreases the biomass activity of acetogens which

consume n-butyrate for acetate production (equation 3.4 in Table 3.3). This conclusion is

consistent with the observation of Hansson & Molin (1981). Under 1 bar of carbon dioxide partial

pressure they found that degradation of n-butyrate to acetate by acetogens was slightly limited

compared to 0.2 bar. A possible reason of the carbon dioxide inhibition is the damage on the cell

membrane and on cell division (Dixon & Kell, 1989). In our study, the effects of limitation on n-

92

butyrate degradation to acetate might be even higher for the reactor with 2 bar of carbon dioxide

in the headspace. Because of the inhibition effect of carbon dioxide on acetogens, n-butyrate

consumption is reduced and its proportion may become higher in reactor with carbon dioxide

headspace.

From the findings it can be concluded that propionate production can be ceased by

hydrogen or carbon dioxide (Table 3.2). The propionate concentration increase in the control

reactor was 1.3 g COD/l with a fraction of 33% on the carbohydrate waste while it remained in

minor concentrations (< 0.4 g COD/l) in all other headspace manipulated conditions (Table 3.2).

This is in contrast to our expectation that propionate production could be more favorable

especially under 2 bar of hydrogen since the reaction requires 2 mols of hydrogen with 1 mol of

glucose (equation 3.5 in Table 3.3) (Angenent et al., 2004).

Results obtained in this research showed that initial hydrogen partial pressure up to 2 bar

did not limit acetate and n-butyrate production reactions from the carbohydrate, protein and lipid

rich substrates. This is in contradiction with the fact that hydrogen can thermodynamically limit

fermentation reactions when it accumulates in the headspace (Rodriguez et al., 2005). In mixed

cultures, syntrophic communities are required to avoid hydrogen partial pressure increase in the

headspace and to allow continued production of acetate and n-butyrate (Stams & Plugge, 2009).

Similar mechanisms also apply for amino acid fermentation. Amino acid pairs can be degraded

when they are coupled with hydrogen producing and hydrogen consuming reactions (Stickland

reactions) to avoid hydrogen accumulation. Single amino acids can be fermented in a process that

requires the presence of hydrogen-utilizing bacteria such as methanogens (Ramsay &

Pullammanappallil, 2001). Degradation of long-chain fatty acids is also only possible at low

hydrogen partial pressure (Sousa et al., 2007).

93

Hydrogen is involved in many reactions and it can shift the reaction equilibrium. Because

acetate and n-butyrate formation from glucose are accompanied by hydrogen production

(equation 3.6 and 3.7 in Table 3.3), our expectation with the high hydrogen partial pressure was

that monomer degradation would be towards more reduced compounds such as alcohols or n-

caproate. It is not clear why n-butyrate and acetate production reactions were more favorable

under a headspace of 2 bar hydrogen in headspace. This raises the question whether hydrogen

serves as electron donor in our experiments in the electron carrier system together with protons as

electron acceptor. Studies with single culture fermentation showed that depending on

environmental conditions, bacteria can use other electron carrier systems apart from the

proton/hydrogen couple, such as proton/formate (Müller et al., 2010). Mixed cultures, apparently,

can tolerate high hydrogen partial pressure if not produced via fermentation reactions and can

continue to produce acetate and n-butyrate in higher concentrations than under conditions having

no headspace manipulation. However, as observed by Steinbusch et al. (2011), an even longer

incubation period (>30 d) might have led to additional adaptation of microorganisms and reactions

towards the more reduced C6 or C8 carboxylates. As a matter of fact, we did observe the onset of

H2 consumption in the carbohydrate fermenter under increased H2 pressure after 2 weeks of

incubation, with the concomitant production of n-caproate.

The experiments showed various effects of the presence of hydrogen and carbon dioxide

on fermentation product concentration and distribution:

- Adding 2 bar of hydrogen to the headspace increases the total carboxylate production,

but does not result in selective product formation. Acetate and n-butyrate productions

were not limited by elevated hydrogen partial pressures in the headspace. On the

contrary, higher concentrations of acetate and n-butyrate could be generated, albeit in

equal amounts.

94

- Adding 2 bar of hydrogen and carbon dioxide 50:50 (v:v) directed the fermentation to

acetate production due to homoacetogenic activity.

- Adding 2 bar of carbon dioxide to the headspace favored the n-butyrate production up

to 70% fraction, after 1 week of incubation.

- Adding hydrogen and/or carbon dioxide to the headspace ceased propionate production.

This is contrary to expectation since propionate production from glucose is accompanied

by hydrogen consumption.

- Even when carbon dioxide is not consumed, it can shift fermentation equilibria when

present at sufficiently high pressures.

- Carbon dioxide can thus be used to direct the fermentation to n-butyrate while a mixed

hydrogen-carbon dioxide headspace yields acetate as major product. Still, it is important

to note that a compromise often needs to be found between the highest product

concentration and the highest fraction. While a high fraction may facilitate further

product purification steps, a minimal concentration level is equally important to develop

cost efficient processes.

Although the presented results showed clear effects of headspace modification on

fermentation reactions, this work should be considered a first step. Further research is required to

explore the details of the processes concerned and confirm the suggested explanations for some

of the experimental observations.

3.4. Conclusions

This study showed that concentrations and fractions of carboxylates can be directed by

supplying additional hydrogen and carbon dioxide in the headspace in mixed culture fermentation.

n-Butyrate, n-caproate and total product concentration increases were higher in reactors with

95

modified headspace than in the control reactor while propionate production was ceased. Carbon

dioxide was not consumed from the headspace but it directed the fermentation process towards

selective production of n-butyrate within 1 week. Acetate production was selective and in high

concentrations within 4 weeks under mixed hydrogen and carbon dioxide headspace. A major

single compound was not found under hydrogen headspace.

96

References

Agler, M.T., Wrenn, B.A., Zinder, S.H., Angenent, L.T., (2011). Waste to bioproduct conversion with undefined mixed cultures: the carboxylate platform. Trends in Biotechnology, 29, 70 – 78.

Albuquerque, M.G.E., Martino, V., Pollet, E., Avérous, L., Reis, M.A.M., (2011). Mixed culture polyhydroxyalkanoate (PHA) production from volatile fatty acid (VFA)-rich streams: Effect of substrate composition and feeding regime on PHA productivity, composition and properties. Journal of Biotechnology, 151, 66 – 76.

Angenent, L.T., Karim, K., Al-Dahhan, M.H., Wrenn, B.A., Domiguez-Espinosa, R., (2004). Production of bioenergy and biochemicals from industrial and agricultural wastewater. Trends in Biotechnology, 22, 477 – 485.

APHA, (1995). Standard Methods for the Examination of Water and Wastewater, 19th ed. APHA, Washington, DC.

Bengtsson, S., Hallquist, J., Werker, A., Welander, T., (2008). Acidogenic fermentation of industrial wastewaters: Effects of chemostat retention time and pH on volatile fatty acids production. Biochemical Engineering Journal, 40, 492 –499.

Cavdar, P., Yilmaz, E., Tugtas, A.E., Calli, B., (2011). Acidogenic fermentation of municipal solid waste and its application to bio-electricity production via microbial fuel cells (MFCs). Water Science and Technology, 64, 789 – 795.

Christ,O., Wilderer, P.A., Angerhöfer R.R., Faulstich, M., (2000). Mathematical modeling of the hydrolysis of anaerobic processes. Water Science and Technology, 41, 61 – 65.

Coats, E.R., Gregg, M.C., Ronald L., (2011). Effect of organic loading and retention time on dairy manure fermentation. Bioresource Technology, 102, 2572 – 2577.

De Mes, T.Z.D., Stams, A.J.M., Reith, J.H., Zeeman, G., (2003). Bio-methane and Bio-hydrogen, Reith, J.H., Wijffels, R.H., Barten, H. (Eds.), Methane Production by anaerobic digestion of wastewater and solid wastes, Dutch Biological Hydrogen Foundation, Petten, pp. 58 – 102.

Dixon, N.M., Kell, D.B., (1989). The inhibition by CO2 of the growth and metabolism of micro-organisms. Journal Applied Bacteriology, 67, 109 – 136.

Drake, H.L., Küsel, K., Matthies, C., (2002). Ecological consequences of the phylogenetic and physiological diversities of acetogens. Antonie van Leeuwenhoek 81, 203–213.

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F., (1956). Colorimetric method for determination of sugars and related substances. Analytical Chemistry 28, 350 – 356.

Fang, H.H.P., Li, C.L., Zhang, T., (2006). Acidophilic biohydrogen production fromrice slurry. International Jourbnal of Hydrogen Energy 31, 683–692.

Hansson, G., Molin, N., (1981). End product inhibition in methane fermentations: Effects of carbon dioxide on fermentative and acetogenic bacteria. Applied Microbiology Biotechnology, 13, 242 – 247.

Holtzapple, M., Granda, C., (2009). Carboxylate Platform: The mixAlco process part 1: Comparison of three biomass conversion platforms. Applied Biochemical Biotechnology, 156, 95 – 106.

Kim, D-H., Han, S-K., Kim, S-H., Shin, H-S., (2006). Effect of gas sparging on continuous fermentative hydrogen production. International Jourbnal of Hydrogen Energy, 31, 2158 – 2169.

Komemoto, K., Lim, Y.G., Nagao, N., Onoue, Y., Niwa, C., Toda, T., (2009). Effect of temperature on VFA's and biogas production in anaerobic solubilization of food waste. Waste Management, 29, 2950 – 2955.

97

Kraemer, J., Bagley, D., (2007). Improving the yield from fermentative hydrogen production. Biotechnology Letters, 29,685 – 695.

Mata-Alvarez, J., Mace, S., Llabres, P., (2000). Anaerobic digestion of organic solid wastes. An overview of research achievements and perspectives. Bioresource Technology, 74, 3 – 16.

Müller, N., Worm, P., Schink, B., Stams, A.J.M., Plugge, C.M., (2010). Syntrophic butyrate and propionate oxidation processes: from genomes to reaction mechanisms. Environmental Microbiology Rep, 2, 489 – 499.

Nie, Y.Q., Liu, H., Du, G.C., Chen, J., (2007). Enhancement of acetate production by a novel coupled syntrophic acetogenesis with homoacetogenesis process. Process Biochemistry, 42, 599 – 605.

Phillips, J.R., Klasson, K.T., Clausen, E.C., Gaddy, J.L., (1993). Biological production of ethanol from coal synthesis gasmedium development studies. Applied Biochemistry Biotechnology 39, 559 – 571.

Pierce, W.C., Haenisch, E.L., (1948). Quantitative Analysis, third ed. J. Wiley, New York.

Pomeranz, Y., Meloan, C.E., (1994). Food analysis: Theory and Practice, third ed. Chapman and Hall, New York.

Ragauskas, A.J., Williams, C.K., Davison, B.H., Britovsek,G., Cairney, J., Eckert C.A., Frederick Jr., W.J., Hallett, J.P., Leak, D.J., Liotta, C.L., Mielenz, J.R., Murphy, R., Templer, R., Tschaplinski, T., (2006). The path forward for biofuels and biomaterials. Science 311, 484 – 489.

Ramsay, I. R., Pullammanappallil, P. C., (2001). Protein degradation during anaerobic wastewater treatment: derivation of stoichiometry. Biodegradation, 12, 247 – 257.

Rodriguez, J., Kleerebezem, R., Lema, J.M., van Loosdrecht, M.C.M., (2005). Modeling Product Formation in Anaerobic Mixed Culture Fermentations. Biotechnology Bioengineering, 93, 593 – 606.

Sousa, D.Z., Pereira, M.A., Stams, A.J.M., Alves, M.M., Smidt, H., (2007). Microbial communities involved in anaerobic degradation of unsaturated and saturated long chain fatty acids. Applied Environmental Microbiology, 73, 1054 –1064.

Stams, A.J.M., Plugge, C.M., (2009). Electron transfer in syntrophic communities of anaerobic bacteria and archaea. Nature Review Microbology, 7, 568 – 577.

Steinbusch, K.J.J., Hamelers, H.V.M., Plugge, C.M., Buisman, C.J.N., (2011). Biological formation of caproate and caprylate from acetate: fuel and chemical production from low grade biomass. Energy Environmental Science, 4,216 – 224.

Thauer, R.K., Jungermann, K., Decker, K., (1977). Energyconservation in chemotrophic anaerobic bacteria. BacteriologyReviews 41, 100–180.

Tilman, D., Socolow, R., Foley, J.A., Hill, J., Larson, E., Lynd, L., Pacala, S., Reilly, J., Searchinger, T., Somerville, C., Williams, R., (2009). Beneficial Biofuels-The Food, Energy, and Environment Trilemma. Science 325, 270 – 271.

Ueno, Y., Fukui, H., Goto, M., (2007). Operation of a two-stage fermentation process producing hydrogen and methane from organic waste. Environmental Science Technology, 41, 1413 – 1419.

Veeken, A., Hamelers, B., (1999). Effect of temperature on hydrolysis rates of selected biowaste components. Bioresource Technology, 69, 249 – 254.

Zhang, K., Ren N-Q., Cao, G-L., Wang, A-J., (2011). Biohydrogen production behavior of moderately thermophile Thermoanaerobacterium thermosaccharolyticum W16 under different gas-phase conditions. International Journal of Hydrogen Energy, 36, 14041 – 14048.

98

When you do something, you should burn yourself up completely, like a good bonfire, leaving no trace of yourself.

Shunryu Suzuki

99

Chapter 4

Selective carboxylate production by controlling hydrogen, carbon dioxide and substrate concentrations in mixed culture fermentation

This chapter has been published as:

D. Arslan, K.J.J. Steinbusch, L. Diels, H. De Wever, H.V.M. Hamelers b, C.J.N. Buisman (2013). Selective carboxylate production by controlling hydrogen, carbon dioxide and substrate concentrations in mixed culture fermentation. Bioresource Technology, 136, 452 – 460.

100

Abstract

This research demonstrated the selective production of n-butyrate from mixed culture by applying

2 bar carbon dioxide into the headspace of batch fermenters or by increasing the initial substrate

concentration. The effect of increasing initial substrate concentration was investigated at 8, 13.5

and 23 g COD/L with potato processing waste stream. Within 1 week of incubation, n-butyrate

fraction selectively increased up to 83% by applying 2 bar hydrogen or 78% by applying carbon

dioxide into the headspace whereas it was only 59% in the control reactor. Although the fraction

of n-butyrate was elevated, the concentration remained lower than in the control. Both the highest

concentration and fraction of n-butyrate were observed under the highest initial substrate

concentration without headspace addition. The concentration was 10 g COD/L with 73% fraction.

The operational conditions obtained from batch experiments for selective n-butyrate production

were validated in a continuous process.

101

4.1. Introduction

The ‘carboxylate platform’ concept has been introduced by Holtzapple et al. (1999) to

generate energy intense chemicals, specifically carboxylic acids, in single stage bioconversion

processes by using undefined mixed cultures. Since then, a variety of chemical and biological

conversions in which carboxylates can be used as building block has been shown by different

researchers. The principle of the platform is a partial anaerobic digestion of organic matters using

mixed cultures in which mainly acetate (C2), propionate (C3) and n-butyrate (C4) in the aqueous

phase and hydrogen in the gas phase are produced. It is, however, important to selectively

produce one end compound or a favorable ratio of a carboxylates mixture in the first stage, since

this will be the precursor for the generation of the target compounds in the second stage (Li & Yu,

2011). For example, in the polyhydroxyalkanoate (PHA) production, the type of carboxylates and

mixture composition produced in the carboxylate platform determines the type of PHA produced

(Serafim et al., 2008).

Organic waste streams are attractive candidates to use as a substrate for generation of

biobased chemicals. The main advantage of using waste streams is that they are cheap materials

(sometimes with no cost or with a gate fee). Anyhow, large amounts of organic waste are being

generated and need to be treated. Fortunately, fermentation of organic waste materials provides

both treatment of waste matters and simultaneous generation of carboxylates like acetate,

propionate and n-butyrate (Gajdos, 1998).

Organic waste materials are mainly composed of carbohydrate, protein and lipid type

polymers. Given this complex composition of the starting material, a mixed consortium is required

to reduce each polymer to valuable end products (Angenent & Wrenn, 2008). Further, using mixed

cultures makes the fermentation process economically more interesting because of elimination of a

102

costly sterilization process. The overall process becomes more stable and flexible without the risk

of contamination (Ueno et al., 1995). However, using mixed consortia degrading a complex

substrate leads to an effluent containing a mixture of various acids and alcohols in different

concentrations. As a result downstream processing is more challenging which increases the

production costs of the target products. It is thus crucial to maximize both the concentration and

the fraction of a single compound in total products. When more than one product is required the

fermentation process needs to be steered towards a certain ratio of desired compounds.

Increasing concentrations and tuning the product distribution among fermentation products

can be achieved in several ways. In literature, substrate type, concentration of substrate in the

fermenter, retention time, reactor design, temperature, and pH were found to alter the

concentrations and also the type of products. Table 4.1 presents reported product concentrations

and distributions from different studies in literature. In our previous work, it was shown that

headspace manipulation with externally added hydrogen and/or carbon dioxide in variable ratios

can increase product concentrations and result in the selective production of a single compound

depending on the polymer type in mixed culture fermentation (Arslan et al., 2012).

Previously, the effect of feed concentration in batch and continuous mode operation on

fermentation products was investigated by Lim et al. (2008), Gomez et al. (2009), Coats et al.

(2011) and Badiei et al., (2011) (Table 4.1). These studies aimed to maximize the total

concentration of carboxylates rather than targeting selective generation of a single compound. Yet,

in their results it was shown that carboxylate concentrations and distributions were affected by

changing the substrate concentration in the fermenter.

103

The aim of this work was to investigate the combined effect on the carboxylate product

spectrum by applying only hydrogen or only carbon dioxide at 2 bar to the headspace at increasing

substrate concentrations.

104

Tabl

e 4

.1.P

rodu

ct c

once

ntra

tions

and

dis

trib

utio

ns r

epor

ted

in t

he s

tudi

es o

n m

axim

izin

g ta

tol V

FA o

r hy

drog

en p

rodu

ctio

n w

ith

mix

ed c

ultu

re

Subs

trat

eTV

FA (

g/L)

Ac (

%)

Pr (

%)

Bu (

%)

Va (

%)

Ca (

%)

Reac

tor

type

Ref.

Mun

icip

ial

solid

was

te5.

738

062

--

Batc

hO

kam

oto

et a

l.,

2000

Man

ure

3.8

7017

111

-Ba

tch

Coat

s et

al.,

201

1

Cafe

teria

Fo

od w

aste

13.5

2828

2221

-Fe

d-Ba

tch

Lim

et

al.,

2008

Palm

oil

mill

134

065

-0

Fed-

Batc

hBa

diei

et

al.,

2011

Glu

cose

1.6

322

64-

2Co

ntin

uous

Fang

&Li

u,20

02Su

cros

e8.

424

-76

--

Cont

inuo

usKy

azze

et

al.,

2005

Food

was

te3.

549

921

-21

Cont

inuo

usG

omez

et

al.,

2009

105


4.2.1. Inocula and waste streams

As inoculum, granular sludge from a potato wastewater anaerobic digester (Opure, Ede,

the Netherlands) was used. To avoid methanogenic growth during fermentation, the granules

were first washed with 20 mM potassium phosphate buffer at pH 5 and sieved through 500 μm for

three times. They were left in potassium phosphate buffer solution overnight at room temperature.

The next day, the granules were heat-treated by boiling in water for 15 min.

Carbohydrate rich waste streams were collected from a potato processing company.

4.2.2. Batch experiments

Batch reactors (5L) with 750 ml working volume were run as described in Arslan et al.

(2012). The organic waste stream was diluted to 3 different initial substrate concentrations: 8,

13.5 and 23 g Chemical Oxygen Demand (COD substrate/L) through the addition of basal medium

solution. The basal medium was prepared according to Philips et al. (1993). To avoid sulfate

reduction during fermentation, sulfate salts were omitted. The reactors were inoculated with 8 g of

wet weight (0.6 g Volatile Solids (VS) content) of pre-treated granular sludge. Pressures in the

reactors were increased to 2 bar by applying only hydrogen or only carbon dioxide. With a

manometer, the headspace pressure of each reactor was monitored on-line. If the headspace

pressure dropped during the experiment, it was adjusted back to 2 bar with the same original

headspace gases. A control reactor was prepared in the same way. However, it was flushed with

N2 only at the beginning of the experiment and connected to a volumetric gas counter

(Milligascounter, BnC-Ritter) working under atmospheric pressure. Reactors were stirred at 100

rpm by a magnetic drive stirrer (Büchi-cyclone 075) and kept at 30°C with hot water circulation.

106

The experiments lasted 30 days with weekly gas and liquid sampling. The sampling amount was

calculated in a way that the liquid to headspace ratio was not disturbed or no underpressure was

occurring.

4.2.3. Continuous experiment

A continuous reactor (5L) with 3L working volume was prepared in the same way as the

batch reactors. The initial wet weight of inocula in the continuous reactor was kept at the same

concentration as in the batch experiments. The headspace was flushed with N2 at the beginning of

the experiment and then connected to a volumetric gas counter under 1 bar of atmospheric

pressure. The reactor was kept at 30°C with hot water circulation. The reactor pH was initially not

controlled. When the pH spontaneously dropped to 4, it was kept above 3.9 by dosing 2 M NaOH.

The hydraulic retention time (HRT) of the reactor was 2 days. The substrate load to the reactor

was 23 g COD/L.d. Waste was diluted with the same basal medium solution as in the batch

experiments. The feed tank was kept at 4°C to avoid that fermentation already started inside the

tank and was continuously stirred to prevent settling of particles. A grinder pump (Cat DK 40, drive

unit X 1740) was connected in a closed loop to the influent tank and continuously circulated the

influent solution to reduce the size of the particles coming from the waste material. A peristaltic

pump (Watson Marlow 520 Du) was used for regular feeding from the loop to the reactor.

4.2.4. Analysis

Gas samples were collected by using a gas-tight plastic syringe and analyzed in a gas

chromatograph (Interscience, Thermo Scientific, Trace GC Ultra) equipped with a thermal

conductivity detector (TCD). H2 content was analyzed with a 2 m stainless steel column packed

with a molecular sieve 5A (80/100 mesh) using nitrogen as carrier gas at a flow rate of 20 ml/min.

107

Carbon dioxide, oxygen, nitrogen and methane gases were analyzed with a 2 m stainless steel

HayeSepQ (80/100 mesh) and molecular sieve 5A (80/100 mesh) column connected in series with

a valve. Helium was used as carrier gas at a flow rate of 15 ml/min.

Liquid samples were collected from the stirred bioreactors and influent tank and

centrifuged for 5 min at 10000 rpm to remove particles. Volatile fatty acid (VFA) samples were

acidified with 1:1 (v/v) H2SO4 solution and then extracted with diethyl ether. VFAs were analyzed

in a GC (Interscience, CE Instruments, Focus GC) equipped with a flame ionization detector (FID)

and a 15 m AT-1000 filled capillary column (0.53 mm x 1.2 μm). Helium was used as carrier gas at

a constant flow of 6 ml/min. Acetone, ethanol and propanol were analyzed in the headspace of the

samples at 60°C with the same GC fitted in split mode with a split/splitless injector and AT-WAX

capillary column (60m x 0.32mm x 1.00μm). Helium was used as carrier gas with a flow rate of 1.6

ml/min. Lactate was analyzed with a GC (Interscience, CE Instruments- Trace GC) equipped with

FID detector in split mode with split/splitless injector and AT-1 capillary column (30m x 0.53mm x

5μm column). Samples were first treated with Ce(SO4)2 at 60°C for 10 min to convert lactic acid

into acetaldehyde which was then analyzed in the headspace.

Waste streams were analyzed with the Phenol-sulfuric method (Dubois et al., 1956) at 485

nm wavelength to determine the carbohydrate concentration. Protein content was measured with

the Total Kjeldahl Nitrogen method, by multiplying NH4-N concentrations with a factor of 6.25

(Pierce and Haenisch, 1948). Total and volatile solid contents were analyzed as described in

Standard Methods (APHA, 1995). Total Chemical Oxygen Demand (CODt) was determined with Dr.

Hach Lange cuvette tests. Soluble COD (CODs), NH4-N and PO4-P concentrations of samples were

analyzed with Dr. Hach Lange cuvette tests after filtration of the samples at 0.45 μm.

108

4.2.5. Calculations

Hydrolysis degree, COD recovery and fractions of products were calculated with the

formulas mentioned by Arslan et al. (2012). The reaction Gibbs free energy values ( G'r) under the

actual environmental conditions were calculated by equation. (4.1).

G'r = ( G°f_products - G°

f_reactants) + R.T. i (vi ln ci) (4.1)

where vi is the stoichiometric coefficient and ci is the concentration for each reactant and

product. Standard Gibbs free energy values ( G°f) of components were taken from Amend and

Shock (2001). For the calculations, it was assumed that the reactions reached thermodynamic

equilibrium. Glucose, acetate, propionate and n-butyrate concentrations were set to 1M, hydrogen

and carbon dioxide pressure to 1 bar and 5 bar, pH to 5 and temperature to 30°C. The

calculations were set up assuming unlimited hydrogen supply and with the substrate as sole

carbon source.

4.3. Results and Discussions

The objective of this study was to investigate the impact of applying 2 bar of 100%

hydrogen and 2 bar of 100% carbon dioxide in the headspace on the fermentation product

spectrum at increasing substrate concentration. Firstly, thermodynamics of the fermentation

reactions were determined to evaluate the impact of hydrogen and carbon dioxide on reactions

from glucose to acetate, propionate and n-butyrate. Glucose was used as substrate in the

reactions as it is the monomer of the starch type substrate which constituted the waste streams

used in the subsequent tests. Next to the thermodynamic calculations, batch and continuous

experiments were performed to evaluate fermentation product spectrum.

109

4.3.1. All carboxylate production reactions from glucose are thermodynamically feasible

There are several factors that regulate fermentation reactions such as microbial

consortium, kinetics and thermodynamics (Lee et al., 2008). Thermodynamic calculations were

thus used to demonstrate under which headspace conditions the VFA generating reactions are

feasible. The changes in G'r values of reactions from glucose conversion to acetate, propionate

and n-butyrate production were studied under 1 and 5 bar hydrogen partial pressures as a

function of carbon dioxide partial pressure.

Figure 4.1 shows that all reactions were below the threshold value of -20 kJ/mol under all

the selected headspace conditions. Hence, they occur spontaneously under the selected actual

conditions. From a thermodynamic point of view, it can thus be anticipated that each product has

a similar potential to be formed since each reaction yields approximately the same amount of

energy. This also explains why in a non-sterilized environment mixtures of carboxylates will be

generated in similar concentrations.

In general, the effects of hydrogen and carbon dioxide partial pressures on the studied

reactions were negligible. Because hydrogen and carbon dioxide are produced in reactions R1 and

R3, increasing the hydrogen and carbon dioxide partial pressure has an adverse effect on the

reactions G'r values but they still remain in the exothermic zone, even at partial pressures up to 5

bar. Among the 3 reactions investigated, glucose conversion to propionate (R2) was

thermodynamically the most favorable reaction. Because no carbon dioxide is involved in R2, its

partial pressure evidently has no impact on the G'r value.

110

Figure 4.1. Gibbs free energy change of reactions from glucose to acetate, propionate, n-butyrate under 1 bar and 5 bar hydrogen partial pressure as a function of carbon dioxide partial pressure. Note: The hollow dots indicate 1 bar hydrogen partial pressure and the filled dots indicate 5 bar hydrogen partial pressure for each reaction.

glucose + 2H2O 2 acetate- + 2H+ + 4H2 + 2CO2 G°'r = -216 kJ/mol (R1)

glucose + 2H2 2 propionate- + 2H+ + 2H2O G°'r = -364 kJ/mol (R2)

glucose n-butyrate- + H+ + 2H2 + 2CO2 G°'r = -266 kJ/mol (R3)

Energy calculations showed that all carboxylates production reactions from glucose are

feasible and that tuning hydrogen and carbon dioxide levels in the headspace is not sufficient to

direct fermentation reactions towards a single product. This makes it more difficult to know how to

drive fermentation reactions from real organic waste.

Propionate

Acetate

-100

-200

-300

-400

n-Butyrate

111

4.3.2. Carboxylates were the main fermentation products

As a substrate, a real organic waste stream from a potato processing industry was used. As

can be seen from Table 4.2, the characteristics of the waste material batches varied although they

were collected from the same point of the industrial process. This caused extra variations in

product spectrum apart from the controlled parameters. However the reason of using real waste

streams instead of pure substrates such as glucose was to obtain scalable results and increase the

practical relevance of the study. Although waste composition differed among the batches, the

control reactors always produced a mixture of carboxylates and changes in product spectrum due

to the modified headspace could be evaluated for each case.

The incoming waste often contained lactate concentrations above 1 g COD/L and lactate

concentration in the incoming feed changed with the waste stream collected at different times.

Typically, lactate was consumed.

In all experiments, total carboxylates constituted over 85% of the total products and the

main carboxylates were acetate, propionate and n-butyrate. Alcohols and n-caproate (C6) were

lower than 10%, except for a few cases. Ethanol was the main alcohol product. Methane was not

detected in any of the experiments.

112

Table 4.2. Characteristics of waste streams collected from potato processing industry*

Analysis

1st collection

Batch exp. 1

& 2

2nd collection

Batch exp. 3 &

Continuous exp.

days 0 - 16

3rd collection

Continuous

exp.

days 16 - 58

4th collection

Continuous exp.

days 58 - 64

TS (mg/g) 36.7 (1.2) 64.5 (0.5) 47.1 (0.4) 36.5 (1.4)

VS (mg/g) 30.8 (4.5) 56.1 (1) 870 (1.2) 852 (9.9)

Carbohydrate (g/kg) 22 (16) 13.2 (0.7) 20.6 (3.0) 9.9 (0.8)

Protein (g/kg) 5.9 (0.2) 7.3 (0.2) 6.2 (0.3) 4.1 (0.2)

Lipid (g/kg) 0.2 (0) Not measured Not measured Not measured

Total COD (g/kg) 26.4 (0.7) 67.5 (2.6) 48.6 (2.4) 32.9 (2.3)

*Concentrations are mean values and standard deviations are shown in brackets. Results are presented on wet weight basis.

4.3.3. External hydrogen addition increased total carboxylate concentrations

Figure 4.2 illustrates that applying a 2 bar hydrogen partial pressure into the headspace

increases total carboxylate production after 4 weeks of incubation with approximately 1.5 g COD

compared to the control with initial substrate concentrations of 8 and 13.5 g COD/L. For these two

experiments, hydrogen consumption from the headspace was 0.8 g COD and 0.5 g COD. This

indicates that the increase in carboxylate concentrations was well above the COD consumption as

hydrogen from the headspace. The difference is due to the higher hydrolysis degrees obtained

under 2 bar of hydrogen compared to the control reactors (Table 4.3). At the highest initial

substrate concentration, the total carboxylate increase was 0.7 g COD higher than in the control

reactor and hydrogen consumption was 1 g COD.

113

Total carboxylate production was elevated not only by applying external hydrogen. By

increasing the substrate concentration from 8 g COD/L to 23 g COD/L, under hydrogen headspace,

the increase in the concentrations of acetate and n-butyrate was 2 times higher than the increase

in the control reactor except for the control reactor in the 23 g COD/L experiment. However

fractions in total remained the same in all experiments after 4 weeks of incubation. n-Butyrate

fraction remained at around 42% and acetate at 33% (Figure 4.2 and Table 4.3). This indicates

that the product distribution was mainly determined by the hydrogen in the headspace. Because a

major single carboxylate was not found by applying 2 bar of hydrogen, it seems that hydrogen

supply to the headspace is not promising for selective production of a single compound.

n-Butyrate and acetate were always the main products with 2 bar hydrogen headspace

while propionate was never observed at fractions higher than 10%. The change in the control

reactor spectrum was sometimes quite big at different substrate concentrations. At the lowest

substrate concentration test, for instance, acetate and propionate were the major products. The

propionate fraction reached up to 33%. When substrate concentration increased, the fermentation

type shifted from acetate–propionate to acetate–n-butyrate. The final n-butyrate concentration in

the highest substrate concentration experiment reached up to 10 g COD/L with a 68% fraction of

the total products. In a mixed culture environment, a shift in populations can occur due to a

change in environmental conditions or chosen operational parameters. Such a shift can be a result

of a different pH and/or hydrogen partial pressure of the system (Vavilin et al. 1995). For example,

n-butyrate producing organisms dominated a microbial population at a pH lower than 5, whereas

acetate and propionate producing organisms were predominant at a pH higher than 5.5 (Fang &

Liu, 2002).

114

Figure 4.2. Concentrations and fractions of acetate, propionate and n-butyrate on COD basis after 1 week and 4 weeks at 2 bar hydrogen, 2 bar carbon dioxide headspace and (c) control reactors at initial substrate concentrations (a) 8 g COD/L, (b) 13.5 g COD/L and (c) 23 g COD/L. Note: Fractions higher than 10% are mentioned on the columns.

115

In our experiments, the pH dropped to 4.2 for both reactors which resulted both in high

propionate and high n-butyrate concentrations. So it is assumed that the pH is had a minor effect

on the product type (Table 4.3). Hydrogen partial pressure levels, however, differed. With

increasing substrate concentration the produced hydrogen in the headspace was increased.

Thermodynamically, propionate production was expected to be favorable at elevated levels of

hydrogen since it is accompanied with hydrogen consumption (R2) (Agler et al., 2010). However,

we never observed propionate in any of the hydrogen applied headspace cases. Our results thus

show that although thermodynamically it is favorable, propionate production is not always higher

at a high hydrogen partial pressure. This is confirmed by the control experiment. There we

observed also that fermentation reactions shifted from propionate to n-butyrate when the

produced amount of hydrogen increased.

The variations in the waste composition might also cause changes in product distribution or

concentrations as observed for the product spectrum differences in the control reactor (Figure

4.2). For instance, in the experiment with the highest substrate concentration, the hydrolysis

degrees were equal to 100% for both the 2 bar hydrogen and the control reactors whereas this

was not the case for the other experiments (Table 4.3). This shows that the organic waste which

was collected the 2nd time was more easily hydrolyzed and degradable comparing to the first batch

of waste material. This might also have reflected in the shift from propionate to n-butyrate.

In the 2 bar hydrogen reactor an increase in longer chain carboxylates was observed. n-

Caproate was the 3rd main product after acetate and n-butyrate produced in the hydrogen

headspace systems. The highest n-caproate production, 0.8 g COD/L reaching up to 11%, was

observed with the lowest substrate concentration within 4 weeks of fermentation whereas it did

not occur in the control reactor.

116

Microorganisms can use VFA as electron acceptor and hydrogen or ethanol as electron

donor to generate medium chain fatty acids such as n-caproate (Steinbusch et al., 2011). These

reactions are described as:

acetate- + 2 ethanol 2 n-caproate- + 2H2O G°'r = -79 kJ/mol (R4)

acetate- + H+ + 2H2 ethanol + H2O G°'r = -9 kJ/mol (R5)

3 acetate- + 2H+ + 4H2 n-caproate- + 4H2O G°'r = -177 kJ/mol (R6)

acetate- + n-butyrate- + H+ + 2H2 n-caproate- + 2H2O G°'r = -48 kJ/mol (R7)

The presence of hydrogen is thermodynamically favorable for formation of more reduced

products such as ethanol and caproic acid (with a higher degree of reduction). In our experiment,

ethanol was not consumed; on the contrary, ethanol production for each hydrogen reactor was at

least 2 times higher than in the control reactors (Table 4.3). Since hydrogen was consumed and

ethanol and n-caproate were formed, the likely explanation is R4 might have coupled with R5. n-

Caproate production could also have occurred via R6 and/or R7 by using hydrogen, acetate and n-

butyrate. In the end, elevated ethanol production together with n-caproate production under

hydrogen pressurized headspace supports the fact that chain elongation of short chain fatty acids

like acetate and n-butyrate in the presence of hydrogen is possible as was mentioned by Ding et

al. (2010) and Steinbusch et al. (2011).

117

4.3.4. External carbon dioxide addition favored n-butyrate production at low substrate

concentration

Applying 2 bar of carbon dioxide to the headspace resulted in a lower total carboxylate

production after 1 week of incubation when compared to the control reactor. This can be due to

inhibitory effect of carbon dioxide as also discussed in our previous study (Arslan et al., 2012).

However, after 4 weeks of incubation, such effects were no longer visible. The total

carboxylate production in the reactors under 2 bar carbon dioxide was either slightly higher than in

the controls for the 8 and 13.5 g COD/L substrate concentration experiments. Total carboxylates in

the 2 bar carbon dioxide reactor were even similar to the control for the 23 g COD/L substrate

concentration condition (Figure 4.2). This might be due to the higher levels of hydrogen

production and subsequent consumption from the headspace in the control reactor of the last

experiment. Indeed, in the 8 and 13.5 g COD/L substrate concentrations experiments, hydrogen

productions were negligible in both control reactors. In the experiment at the highest substrate

concentration however, the hydrogen fraction in the headspace of the control reactor reached

27% and dropped to 4% by the end of the experiment due to consumption. Another reason might

be that in the highest substrate concentration experiment, as mentioned earlier, the composition

of the waste was different and it was observed that this waste batch was more easily degradable

as shown from hydrolysis degree calculations. This indicates that for this experiment kinetics were

different. Carbohydrate conversion to carboxylates, hydrogen and carbon dioxide was faster in the

last experiment than for the experiments run with the previously collected waste material

Applying 2 bar carbon dioxide to the headspace reactors changed product fractions and

concentrations compared to the control reactors (Figure 4.2). The main product was always n-

butyrate. Especially in the 1st week of the experiments, there was a significant selective n-butyrate

118

production with fractions reaching up to 78% with the 8 g COD/L and 13.5 g COD/L initial

substrate concentrations. After the 1st week, the acetate production rate increased, so n-butyrate

fraction in the total carboxylate production dropped to approximately 50% for these 2

experiments, while it was only 17% in the control reactors. Concentrations of n-butyrate were also

3 to 4 times higher than in the control reactors. Carbon dioxide consumption in the 8 and 13.5 g

COD/L substrate concentration experiments were 5.6 and 12.3 mM which was rather low

compared to the total product concentrations. Even when carbon dioxide consumption in the

reactions was not significant, the presence of such a high carbon dioxide partial pressure can shift

the equilibria of reactions and impact the product spectrum if it is present at sufficiently high

pressure (Arslan et al., 2012).

At the initial substrate concentration of 23 g COD/L, the n-butyrate fraction and

concentration in the 2 bar carbon dioxide reactor was lower than in the control reactors. In the

former, the n-butyrate fraction was only 48% while it was 73% in the control and its concentration

remained 3 g COD/L lower than in the control reactor. This shows that the effect of carbon dioxide

was less outspoken at the elevated initial substrate concentrations. High levels of substrate

concentrations apparently already drove fermentation reactions to production of n-butyrate even

without carbon dioxide addition.

4.3.5. There are two ways to steer organic waste fermentation towards n-butyrate

n-Butyrate is a valuable carboxylate compound which can be used directly in the chemical,

plastic, and textile industry or can be used as substrate for the production of fuels such as butanol

(Zigova & Sturdik, 2000).

In this study, it was thus found that n-butyrate production could be selectively increased

under two conditions: (1) by applying carbon dioxide in the headspace if substrate concentration is

119

below 13.5 g COD/L, (2) by sufficiently increasing substrate concentration without any headspace

modification. Under the 1st condition, the n-butyrate fraction could be selectively increased as high

as 75% but the n-butyrate concentration remained low. Under the 2nd condition, both n-butyrate

concentration and fraction could be increased. In both ways, it seems that a 1 week incubation

period is enough to achieve selective n-butyrate production higher than 70%.

Common knowledge of the acetate-n-butyrate type fermentation is that acetate production

from glucose yields 4 ATP and n-butyrate yields 3 ATP (Kleerebezeem & Stams 2000; Zhang et.

al., 2009). Because acetate production provides more energy it is more likely that acetate is mainly

produced in the growth phase (R1). In the mean time, hydrogen and carbon dioxide accumulate in

the system which may lead to a shift from acetate production to n-butyrate since n-butyrate

production generates less hydrogen (R2). However, in this study, the opposite occurred.

A reason for the high n-butyrate concentrations and fraction at the beginning of the

experiments might be that acid forming bacteria fermenting glucose to propionate and n-butyrate

are faster growing organisms than acetate producers (Mosey, 1983, Mohan et al., 2007).

Therefore acetate production started later with assimilation by slow-growing acetogens.

Another reason might be the high lactate concentration in the waste material. Lactate

concentrations were 0.9 g COD/L in the first, 1.5 g COD/L in the second experiment and 2.2 g

COD/L in the last experiment. In the experiments where the influent had the lowest lactate

concentrations, the n-butyrate concentrations were indeed lower comparing to the experiment

with high lactate concentration in the feed. A study by Kim et al. (2012) showed that addition of

lactic acid into a feed together with glucose, increased n-butyrate production almost 30% in a

batch-wise fermentation process. Conversion of lactate to n-butyrate was previously observed with

single anaerobic bacteria by Duncan et al. (2004) and with a mixed consortium by Brummeler &

120

Koster (1989). With the mixed consortium, first 1 mole of glucose is converted to 2 moles of

lactate then to 1 mole of n-butyrate (R8). This reaction occurs at pH values higher than the pKa

value of n-butyrate/butyric acid (pH>4.8). n-Butyrate formation from lactate for the low pH range

was described according to R9 in Agler et al. (2011). However, our results do not match with the

stoichiometry of R9. Besides, R9 is only favorable when it is coupled with acetate reduction (R10)

(Agler et al., 2011). As such, hydrogen partial pressures can be kept low enough by R10 or other

hydrogen consuming reactions to keep the R9 thermodynamically feasible. In our study, during the

n-butyrate production period, acetate and hydrogen were not consumed however, so this route is

less plausible. We do realize that the production of hydrogen could have been equal to the

consumption of hydrogen for the production of n-butyrate. This makes it rather difficult to

determine the exact mechanism for its production. The observed hydrogen and carbon dioxide

production levels were also too low if you assume that n-butyrate was produced only via R3.

Therefore, n-butyrate might have been formed through a combination of R3 and R8 within the first

week of incubation. The produced hydrogen and carbon dioxide from R3 and R8 might then have

been consumed within other intermediate fermentation reactions. Finally, butyric acid production

might have stopped when its concentration reached inhibitory levels. These amounted to 50 mM

for mixed culture fermentation (van den Heuvel et al., 1992).

glucose 2 lactate- n-butyrate- + H+ + 2H2 + 2CO2 (R8)

2 lactate- + H+ n-butyrate- + 2H2 + 2CO2 G°'r = -209 kJ/mol (R9)

2 acetate- + H++ 2H2 n-butyrate- + 2H2O G°'r = -48 kJ/mol (R10)

The other way to increase the n-butyrate fractions and concentrations is to apply high

substrate concentration without adding external headspace. Here, our observations are in line with

literature. Increase in n-butyrate concentration with increasing substrate concentration was for

121

instance also observed by Fang et al. (2006) in a batch-wise reactor. The highest n-butyrate

concentration of 61 mM was obtained at high substrate concentration.

4.3.6. Batch results were confirmed in continuous mode

A continuous reactor was run to validate the results obtained in the batch experiments.

Such validation is crucial as for practical applications a continuous reactor is preferred because it

(i) is easier to operate, (ii) allows steady state operation under optimal conditions, and (iii) is more

robust to perturbations. According to the batch experiments, n-butyrate production could be

maximized to 10 g COD/L with almost 73% fraction without applying external hydrogen or carbon

dioxide (Table 4.3). This result was obtained within 1 week of batch fermentation. Therefore, 2

days HRT was considered to be enough for the continuous test. The pH of the system was

regulated at 4.0, since this was the pH of the batch reactor where n-butyrate reached its

maximum levels. Besides, from an economical point of view, regulating the pH around 4.0 would

reduce the base consumption compared to higher pH values.

122

Tabl

e 4.

3.Co

ncen

trat

ion

chan

ge o

f pr

oduc

ts a

nd f

ract

ions

(in

bra

cket

s) f

or e

ach

head

spac

e co

nditi

on w

ith in

crea

sing

initi

al s

ubst

rate

con

cent

ratio

n

Initi

al s

ubst

rate

conc

entr

atio

n

8 g

COD

/l13

.5 g

CO

D/l

23 g

CO

D/l

Hea

dspa

ce c

ondi

tions

H2

CO2

N2

H2

CO2

N2

H2

CO2

N2

Prod

ucts

(m

g CO

D/l)

Afte

r 1

wee

k

C21.

5 (3

1%)

-0.0

4 a

1.4

(38%

)-0

.3 a

0.3

(7%

)1.

6 (2

9%)

3.1

(26%

)3.

2 (2

7%)

2.2

(17%

)

C30.

1 (1

%)

0.1

(4%

)1.

3 (3

4%)

0.2

(6%

)0.

1 (4

%)

0.06

(1%

)0.

3 (3

%)

0.4

(3%

)0.

2 (2

%)

n-C4

2.4(

51%

)1.

5 (7

5%)

0.7

(17%

)2.

6 (8

3%)

2.9

(78%

)3.

3 (5

9%)

5.9

(49%

)5.

8 (4

8%)

9.6

(73%

)

n-C6

0.4

(8%

)<

0.1

(<1%

)0

<0.

1 (<

1%)

0.1

(3%

)0.

5 (8

%)

0.5

(4%

)0.

3 (3

%)

<-0

.1 a

Tota

l VFA

4.4

(93%

)1.

6 (8

1%)

3.4

(90%

)2.

9 (9

0%)

3.5

(94%

)5.

6 (9

9%)

10 (

84%

)9.

8 (8

2%)

12 (

92%

)

Lact

ate

cons

umed

-0.9

a-0

.9 a

-0.9

a-1

.5 a

-1.5

a-1

.5 a

-2.2

a-2

.2 a

-2.2

a

Etha

nol

0.2

(4%

)0.

2 (1

0%)

0.4

(10%

)0.

3 (9

%)

0.2

(5%

)<

0.1

(1%

)1.

0 (9

%)

1.0

(8%

)0.

9 (7

%)

Hyd

roly

sis

degr

ee (

%)

4310

389

432

6668

79

pH4.

55

4.2

5.6

4.6

4.7

4.3

44.

1

Afte

r 4

wee

ks

C22.

6 (3

3%)

1.9

(34%

)2.

2 (4

3%)

2.1

(28%

)1.

4 (2

0%)

3.3

(63%

)6.

6 (3

9%)

4.8

(31%

)3.

4 (2

3%)

C30.

2 (3

%)

0.4

(7%

)

1.7

(33

%)

0.7

(9%

)0.

7 (9

%)

0.2

(3%

)0.

3 (2

%)

0.7

(4%

)0.

6 (4

%)

n-C4

3.1

(41%

)2.

5 (4

7%)

0.9

(17%

)3.

2 (4

4%)

3.6

(52%

)0.

8 (1

7%)

7.5

(43%

)7.

2 (4

6%)

10 (

68%

)

n-C6

0.8

(11%

)0.

3 (5

%)

0.04

(1%

)0.

6 (7

%)

0.7

(10%

)0.

5 (1

0%)

0.7

(4%

)0.

7 (4

%)

0.05

(<

1%)

123

Tota

l VFA

7.2

(93%

)5.

4 (9

8%)

4.9

(97%

)6.

8 (9

3%)

6.8

(98%

)5.

0 (9

8%)

15 (

88%

)14

(86

%)

14 (

96%

)

Lact

ate

cons

umed

-0.9

a-1

.0a

-1.0

a-1

.5a

-1.5

a-1

.5a

-2.2

a-2

.2a

-2.4

a

Etha

nol

0.4

(5%

)<

0.1

(<1%

)0.

1 (3

%)

0.5

(6%

)0.

1 (1

%)

0.1

(2%

)1.

0 (6

%)

1.0

(7%

)0.

5 (3

%)

Hyd

roly

sis

degr

ee (

%)

6149

5166

6035

101

103

101

pH4.

44.

54.

24.

64.

84.

74.

14.

04.

1

COD

rec

over

y (%

)92

9710

678

7576

9695

102

Prod

ucts

alw

ays

occu

rrin

g in

fra

ctio

ns lo

wer

tha

n 10

% a

re n

ot s

how

n.

aN

egat

ive

valu

es in

dica

te c

onsu

mpt

ion

of c

ompo

unds

124

Figure 4.3 represents the product distribution and the concentrations for 64 days of

operation. Carboxylate concentrations continuously increased until 25 days after the start-up.

Due to technical problems with the grinder pump of the feeding tank, the amount of total fresh

solids entering to the reactor was less in between the days 25 and 40 than before. In this

period, carboxylate production declined somehow. Once the problems were solved, the

production increased further and leveled off in the last 10 days of the experiment. The average

COD recovery in this final period was 95% (-/+25%) and the average carboxylate concentration

increase was 12.3 g COD/L.

Most important results were that selective n-butyrate production was achieved in the

first 40 days of the run (Figure 4.3). The average n-butyrate concentration increase was 5.7 g

COD/L. This constituted 92% of the total carboxylate production which is in fact, comparable

with single culture conversion results under sterilized conditions. However, in this stage, the n-

butyrate concentration was lower than in the batch experiment with the same condition (Figure

4.3). The mechanism of n-butyrate production might have been R3 together with R8, since

both glucose and lactate were consumed. However, in that case more hydrogen and carbon

dioxide should have been formed. Maybe the produced hydrogen and carbon dioxide from R3

and R8 was consumed within other intermediate fermentation reactions. Since acetate

consumption was also observed in this stage of the experiment, R9 and R10 could take place as

intermediate reactions. In these two reactions lactate and acetate are consumed together with

hydrogen to produce n-butyrate.

125

Figure 4.3. Concentration change of carboxylates, lactate and headspace products as a function of time in the continuous experiment. Negative COD concentration is the amount of the consumed constitutes which was available in the incoming waste. Note: Arrows indicates that the system was run with freshly collected waste material.

The highest n-butyrate concentration was generated after 40 days of the continuous

experiment. When lactate and acetate concentration in the feed dropped to 0, the average n-

butyrate concentration increased from 5.7 g COD/L to 9 g COD/L. This was the highest and also

the potentially inhibitory level of n-butyrate obtained in the batch experiment. In the mean

time, acetate production started and the fractions of n-butyrate dropped. As a result, the

selectivity of n-butyrate production decreased from 92% to 74% (data not shown). In this

126

period of the experiment, the H2 yield was 0.45 mol/g sugars converted. Continuous production

of n-butyrate at the highest concentration under the selected conditions was thus validated for

60 days. However, when aiming at the production of valuable biochemicals, rather than the

highest concentration, obtaining the highest fraction of one compound could be more desirable

since product separation is easier. Hence, it is more interesting to maintain the conditions of the

first 40 days of the continuous experiment. For that, acetate producers could be washed out by

reducing the HRT.

The increasing n-butyrate fraction and concentration with the elevated substrate

concentrations can be explained by the difference of available electrons levels. Under the

condition of high fresh substrate concentration, the amount of available electrons/reactor

increases. The vast amount of electrons/reactor leads to a shift in reactions towards more

reduced compounds such as n-butyrate. This finding was also observed in the study of Badiei et

al. (2011). They found n-butyrate with fractions higher than 70% at increasing substrate

concentration daily fed into the reactor. The amount of electrons readily available in the reactor

also depends on the HRT and type of substrate material. At longer HRTs or even at shorter

HRTs but with increasing organic load and easily hydrolyzing waste materials, reactions can

shift towards even more reduced compounds. For instance, Lim et al. (2008) observed n-

caproate instead of n-butyrate at increasing organic loadings of food type waste at higher HRT

values than in our study. The amount of available electrons in the reactor might also change

the dominant microbial population in the reactor and cause a shift in product spectrum towards

n-butyrate formation.

Acetate production remained negligible compared to n-butyrate production and only

started in the last 25 days of the experiment (Figure 4.3). Simultaneously, a slight decrease in

127

hydrogen partial pressure was observed. It could be that homoacetogens produced acetate

from hydrogen and carbon dioxide which were produced from the waste material. However, the

amount of consumed hydrogen was not enough to generate the acetate produced and no

decrease in the carbon dioxide was observed. In fact, also in the batch experiments n-butyrate

concentrations were higher than acetate at the beginning of the experiments. As also explained

earlier, this might be due to slow-growing acetogens comparing to n-butyrate and propionate

forming bacteria.

Another remarkable thing is the correlation between propionate production and lactate

consumption. At the point where lactate concentration in the incoming waste decreased,

propionate concentration increased from 0.4 g COD/L to 1 g COD/L (Figure 4.3). In the

meantime, 4 mM of hydrogen was consumed from the headspace. When lactate concentration

in the waste elevated again propionate disappeared in the reactor and the hydrogen

consumption stopped. Propionate was thus assumed to be produced from hydrogen and

glucose because the amounts of consumed hydrogen and produced propionate fit the

stoichiometry of R2.

Overall, this experiment achieved continuous n-butyrate production with high selectivity

(92%) for over a month with organic waste substrate. Still it is difficult to determine the exact

conversion mechanism of each product since the analyzed concentrations only reveal a net

production or consumption and the overall conversion rates, but not the individual production

and consumption of each compound.

4.4. Conclusions

Thermodynamics indicated that carboxylate production reactions from glucose are all

spontaneous under standard conditions and increasing hydrogen or carbon dioxide up to 5 bar

128

is not enough to direct mixed culture fermentation towards a single product. Our experiments

showed that hydrogen or carbon dioxide in headspace helps selective carboxylate formation. n-

Butyrate fraction increased under 2 bar carbon dioxide in headspace. At elevated substrate

concentrations, effect of carbon dioxide on n-butyrate formation was limited and n-butyrate

increased without headspace manipulation. This research demonstrated two methods for

selective n-butyrate production in batch mixed culture processes and validated the results in

continuous operation.

129

References

Agler, M.T., Wrenn, B.A., Zinder, S.H., Angenent, L.T., (2011). Waste to bioproduct conversion with undefined mixed cultures: the carboxylate platform. Trends in Biotechnology, 29, 70 – 78.

Angenent, L.T., Wrenn, B.A., (2008). Optimizing mixed culture processing to convert waste into bioenergy. Bioenergy, ASM Press, Washington DC.

Amend, J.P., Shock, E.L., (2001). Energetics of overall metabolic reactions of thermophilic and hyperthermophilic archaea and bacteria. FEMS Microbiology Reviews, 25 (2), 175 – 243.

APHA, 1995. Standard Methods for the Examination of Water and Wastewater, 19th ed. APHA, Washington, DC.

Arslan, D., Steinbusch, K.J.J., Diels, L., De Wever H., Buisman, C.J.N., Hamelers, H.V.M., (2012). Effect of hydrogen and carbon dioxide on carboxylic acids patterns in mixed culture fermentation. Bioresource Technology, 118,227–234.

Badiei, M., Jahim, J.Md., Anuar, N., Abdullah, S.R.S., (2011). Effect of hydraulic retention time on biohydrogen production from palm oil mill effluent in anaerobic sequencing batch reactor. International journal of hydrogen energy, 36, 5912 – 5919.

Brummeler, E.T., Koster, I.W. (1989). The Effect of Several pH Control Chemicals on the Dry Batch Digestion of the Organic Fraction of Municipal Solid Waste. Resources, Conservation and Recycling, 3, 19 – 32.

Coats, E.R., Gregg, M.C., Ronald L., (2011). Effect of organic loading and retention time on dairy manure fermentation. Bioresource Technology, 102, 2572 – 2577.

Ding, H.B. Tan G.-Y. A., Wang, J.-Y., (2010). Caproate formation in mixed-culture fermentative hydrogen production. Bioresource. Technology, 101, 9550 – 9559.

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F., (1956). Colorimetric method for determination of sugars and related substances. Analytical Chemistry 28, 350 – 356.

Duncan, S.H., Louis, P., Flint, H.J., (2004). Lactate-utilizing bacteria, isolated from human feces, that produce butyrate as a major fermentation product. Applied and Environmental Microbiology, 70 (10), 5810 – 5817.

Fang, H.H.P., Li, C.L., Zhang, T., (2006). Acidophilic biohydrogen production from rice slurry. International Journal of Hydrogen Energy, 31, 683 – 692.

Fang, H.H.P., Liu, H., (2002). Effect of pH on hydrogen production from glucose by a mixed culture. Bioresource Technology, 82, 87 – 93.

Gajdos, R., 1998. Bioconversion of organic waste by the year (2010): to recycle elements and save energy. Resources, Conservation and Recycling 23, 67 – 86.

Gomez, X., Cuetos, M.J., Prieto, J.I., Moran A., (2009). Bio-hydrogen production from waste fermentation: Mixing and static conditions. Renewable Energy, 34, 970 – 975.

Holtzapple, M.T., Davison, R.R., Ross, M.K., Aldrett-lee, S., Nagwani, M., Lee, C.H., Lee, C., Adelson, S., Kaar, W., Gaskin, D., Shirage, H., Chang, N.S., Chang, V.S., Loescher, M.S., (1999). Biomass conversion to mixed alcohol fuels using the MixAlco process. Applied Biochemistry and Biotechnology, 77–79, 609 – 631.

Kleerebezeem R., Stams, A.J.M., (2000). Kinetics of syntrophic cultures: a theoretical treatise on butyrate fermentation. Biotechnology and Bioengineering, 67 (5), 529 – 543.

130

Kim, T.H., Lee, Y., Chang, K.H., Hwang, S.J., (2012). Effects of initial lactic acid concentration, HRTs, and OLRs on bio-hydrogen production from lactate-type fermentation. Bioresource Technology, 103, 136 – 141.

Kyazze,G., Martinez-Perez,N., Dinsdale, R., Premier,G.C., Hawkes,F.R. Guwy,A.J., Hawkes D.L., (2005). Influence of Substrate Concentration on the Stability and Yield of Continuous Biohydrogen Production. Biotechnology and Bioengineering, 93, (5), 971 – 979.

Li, W.W., Yu, H.Q., (2011). From wastewater to bioenergy and biochemicals via two-stage bioconversion processes: A future paradigm. Biotechnology Advances, 29, 972 – 982.

Lee, H.S., Salerno, M.B., Ritmann, B., (2008). Thermodynamic Evaluation on H2 Production in Glucose Fermentation. Environmental Science Technology, 42, 2401 – 2407.

Lim, S.J., Kim, B.J., Jeong, C. M., Choi, J., Ahn, Y.H., Chang, H.N., (2008). Anaerobic organic acid production of food waste in once-a-day feeding and drawing-off bioreactor. Bioresource Technology 99, 7866 – 7874.

Mosey, F.E. (1983). Mathematical modelling of the anaerobic digestion process: regulatory mechanisms for the formation of short-chain volatile acids from glucose. Water Science and Technology, 15, 209 - 232.

Mohan, S.V., Babu, V.L., Sarma, P.N., (2007). Anaerobic biohydrogen production from dairy wastewater treatment in sequencing batch reactor (AnSBR): Effect of organic loading rate. Enzyme and Microbial Technology, 41, 506 – 515.

Okamoto, M., Miyahara, T., Mizuno, O., Noike, T., (2000).Biological hydrogen potential of materials characteristic of the organic fraction of municipal solid wastes. Water Science and Technology, 41 (3), 25 – 32.

Phillips, J.R., Klasson, K.T., Clausen, E.C., Gaddy, J.L., (1993). Biological production of ethanol from coal synthesis gasmedium development studies. Applied Biochemistry and Biotechnology 39, 559 – 571.

Pierce, W.C., Haenisch, E.L., (1948). Quantitative Analysis, third ed. J. Wiley, New York.

Serafim, L.S., Lemos, P.C., Albuquerque, M.G.E., Reis, M.A.M., (2008). Strategies for PHA production by mixed cultures and renewable waste materials. Applied Microbiology Biotechnology, 81, 615 – 628.

Steinbusch, K.J.J., Hamelers, H.V.M., Plugge, C.M., Buisman, C.J.N., (2011). Biological formation of caproate and caprylate from acetate: fuel and chemical production from low grade biomass. Energy and Environmental Science, 4, 216 – 224.

Ueno, Y., Kawai, T., Sato, S., Otsuka, S., Morimoto, M., (1995). Biological production of hydrogen from cellulose by natural anaerobic microflora. Journal of fermentation and bioengineering, 79 (4), 395 – 397.

Vavilin, V.A., Rytow, S.V., Lokshina, L.Ya, (1985). Modelling hydrogen partial pressure change as a result of competition between the butyric and propionic groups of acidogenic bacteria. Bioresource Technology, 54,171 – 174.

van den Heuvel, J.C., Verschuren, P. G., Beeftink, H. H., de Beer, D., (1992). Determination of the critical concentration of inhibitory products in a repeated fed-batch culture. Biotechnology Techniques, 6 (1), 33 – 38.

Zhang , C., Yang, H., Yang, F., Ma, Y., (2009). Current Progress on Butyric Acid Production by Fermentation. Current Microbiology, 59, 656 – 663.

Zigova, J., Sturdik E., (2000). Advances in biotechnological production of butyric acid. Journal of Industrial Microbiology & Biotechnology, 24, 153–160.

131

132

I don’t make plans. I just let life come and find me.

Morrissey

133

Chapter 5

IIn-situ carboxylate recovery and simultaneous pH control with tailor-configured bipolar membrane electrodialysis during continuous mixed culture fermentation

134

Abstract

Anaerobic fermentation of organic waste streams by mixed culture generates a mixture of short

chain carboxylic acids. To avoid inhibitory effects of the acids or their consumption in internal

conversion reactions in the mixed culture environment, in-situ recovery of acids can be

beneficial. In this study, electrodialysis with bipolar membranes (EDBM) was applied to a mixed

culture fermentation on organic waste streams using a novel EDBM stack with “direct contact”

operation mode. We could demonstrate simultaneous recovery of carboxylates from the

fermenter by the EDBM stack and in-situ generation and transport of hydroxyl ions to the

fermenter allowing direct pH control. Experiments showed productivity increase after EDBM

coupling to the fermenter, and complete elimination of external base consumption. It was also

observed that EDBM was able to drive the mixed culture fermentation towards acetate and

propionate type of carboxylate production.

135

5.1. Introduction

Organic waste streams can be converted to valuable carboxylic acids such as acetic,

propionic and butyric acid for industrial use using a non-sterile mixed culture fermentation.

Compared to pure culture fermentation processes, carboxylate production from organic waste

with mixed cultures is very attractive because substrate costs are reduced and no expensive

sterilization process is needed. The outcome of the mixed culture fermentation, however, is

mostly a mixture of carboxylates rather than a single product in high concentration.

Furthermore, the carboxylate production capacity is generally limited in biological fermentation

processes. This might be caused by the various metabolic conversions occurring at the same

time by different organisms or by lower tolerance to inhibition or toxicity. Such limitations can

be overcome by continuously removing the carboxylates once they are formed in the reactor.

This approach is called in-situ product recovery (ISPR). Integrating the fermentation process

with a proper separation technology leads to process intensification and may steer product

formation. As a result, continuous removal of products may create a driving force for higher

production rates and higher selectivity of target compounds (Woodley et al., 2008).

Separation of organic acids in-situ can be achieved using several techniques such as

sorption (Chen & Ju, 2002; Gao et al., 2011) and ion-exchange (Anasthas & Gaikar, 2001),

liquid-liquid extraction (Alkaya et al., 2009), membrane extraction (Wodzki & Nowaczyk, 2002)

and electrodialysis (ED) (Wang et al., 2006).

The major advantage of ED technology over other ISPR techniques is that it can supply

protons or alkalinity by configuring bipolar membranes (BM) in the ED unit. The principle of

EDBM is based on separation of charged ions in aqueous medium by a series of BM, anion- and

cation- exchange membranes (AEM and CEM) arranged in alternating pattern in the ED stack.

136

Cations in the feed solution migrate towards the cathode through CEMs and anions migrate

towards the anode through AEMs. Eventually, ions are concentrated in alternating

compartments while the feed solution becomes ion-free (Strathmann, 2004). Cations combine

with OH- and anions with H+ generated by the BM in different product compartments. A

schematic diagram of a conventional EDBM stack is shown in Figure 5.1a.

Since the BM generate H+ and OH-, the base required to regulate the pH of a fermenter

can be reduced or in a best case completely eliminated (Huang et al., 2007). EDBM is thus

considered as a very suitable technology for ISPR coupling to fermentation as it does not

introduce any additional salt or toxic chemical for the organism like in crystallization, adsorption

or extraction (Urbanus et al., 2010).

The state-of-the-art on the ED and EDBM for organic acids and carboxylates production

is reviewed in Huang et al. (2007). Principal process parameters to control EDBM are current

density, stack potential, cross-flow rate, pressure drop, the ion concentration and the pH of the

incoming solution. Overall efficiency can be evaluated by the process outputs such as energy

efficiency, ion fluxes, recovery rate and concentrating factor (Zhang, 2011).

137

(a)

(b)

Figure 5.1. Schematic representation of conventional EDBM system (a) and of EDBM membrane configuration for 1 one cell pair as used in this study (b). CEM: Cation exchange membrane; AEM: Anion exchange membrane; BM: Bipolar membrane. ERS: Electrode rinse solution; F: Feed compartment; P: Product compartment; B: Base compartment.

138

The practical disadvantage of the conventional EDBM system is that all charged

molecules are transported in relative ratios depending on charge, molecular size and

concentration, thus the selectivity for the target compound over other ions may be low. Also, an

intermediate ultrafiltration step might be required to retain the microorganisms in the fermenter

and prevent clogging and fouling of the EDBM stack. Major limitations of EDBM application in

industry are the need of expensive membrane maintenance and replacement and electrical

energy cost (Wang et al., 2011; Iizuka et al., 2012).

Although some major economic problems needs to be overcome for EDBM applications,

some promising results were obtained with small scale laboratory set-ups for production of

selected carboxylates in pure culture fermentations and under sterilized conditions. Wong et al.

(2010) proved up to 75% acetate removal in an E. coli fermentation with coupled EDBM. The

total volume of base required to regulate the pH was reduced to 50%. Redwood et al. (2012)

recovered about 50 to 70% of a carboxylate mixture containing 40% n-butyrate, 26% acetate

and 2% propionate from EDBM coupled to an E. coli continuous fermentation and maintained

pH control in a 30 day experiment. A reactor not coupled with EDBM yielded only 1 mol of total

carboxylates whereas up to 5 mol was extracted from a coupled reactor. The majority of papers

on fermentation set-ups with integrated EDBM are based on single culture fermentations for

acids like lactic, citric or succinic rather than short chain carboxylates such as acetate,

propionate and n-butyrate (Lee et al., 1998; Bailly et al., 2001; Wee et al., 2005; Yi et al.,

2008; Wang et al., 2012; Wang 2013). In some studies, base generated in a separate unit and

is dosed as salt to the reactor or two sequential ED-EDBM systems are used. The first ED for

product recovery and the second EDBM for base generation by BM (Novalic et al., 2000;

Habova et al., 2004).

139

There are only few examples in literature from ISPR technology coupled to mixed culture

fermentation. Agler et al. (2012) demonstrated the in-line recovery of medium-chain mixed

carboxylates using liquid-liquid extraction. Cheng et al. (2008), Wang et al. (2006) and Vertova

et al. (2009) studied separation of short chain carboxylates mixtures from waste effluent by

batch-wise ED operation. Separation of carboxylates in one step EDBM in a mixed product

spectrum from complex waste feedstock has hardly been studied up to now.

Hence, the first aim of this work consists of expanding the EDBM technique to a new

situation, i.e. a mixed product spectrum from a waste fermentation, and evaluating it by

assessing the most important process parameters current density, stack potential, removal

capacity, and productivity. The diversity and sophisticated functions of EDBM allow it to be

implemented in a very creative way for ISPR. In this study, a novel membrane stack

configuration was proposed. Two different types of AEMs were used for selective separation of

negatively charged carboxylate ions from the fermentation broth while OH- ions were generated

by BM and transported into the fermentation broth. This “direct contact” operation mode,

allowed simultaneous carboxylate recovery from and pH control through OH- supply to the

recirculated fermentation broth. In conventional methods, instead of OH- ions, a base solution is

generated by BM and collected in an external tank from which it is then dosed to the fermenter

for pH control. CEM were excluded from the configuration to avoid scaling of divalent ions e.g.

Ca2+, Mg2+ coming from the fermentation medium on the membrane surface. An associated

benefit is that this avoided cation removal from the broth. Hence, the proposed EDBM

configuration mainly affects the anion composition of the recirculated fermentation broth.

In this study, we coupled the novel EDBM stack for the first time in-situ to a carboxylate

producing mixed culture reactor operated in continuous mode. The targets are to increase the

140

total carboxylate productivity and selectivity by in-situ recovery and simultaneously produce

sufficient alkalinity to eliminate the external base consumption, and costs of external base

dosage pump.


5.2.1. Continuous fermentation

Anaerobic granular sludge collected from an UASB reactor treating wastewater from a

potato processing company was used as inoculum. As substrate, waste from different potato

processing stream was used. Waste stream used and operation of the continuous reactor (3L

working volume) and analytical methods were as described in Arslan et al. (2013). The reactor

was operated at 2 days hydraulic retention time (HRT) with 23 g COD/L d organic loading at 30

(1993) without sulphate. The initial pH of the medium was set to 5 by using mixtures of 4 mM

monopotassium phosphate and 4 mM dipotassium phosphate buffers and was not further

controlled during the start-up phase of fermentation. When the pH spontaneously dropped to

4.7, it was kept at this level by dosing 2 M NaOH. Only when the feed pH was higher than 4.8,

2 M HCl was added to the reactor.

5.2.2. EDBM configuration

The laboratory scale ED stack (10 cm x 10 cm) was assembled with 5 cell pairs. Each

cell pair consisted of a BM (FumaTech, Germany) and 2 AEMs which were a standard AMX

membrane (Neosepta, Japan) and a PC-200D membrane (PCA, Germany) selective for short

chain carboxylates. The membrane configuration is presented in Figure 5.1b. In this novel

EBDM stack, the base compartment (B) was restricted by the AMX and BM membranes.

Because direct contact of the fermentation broth with BM might cause severe fouling issues, the

141

AMX membrane was placed before the BM for fouling protection. The membrane configuration

was designed to produce OH- ions by BM and transport OH- through the AMX membrane to the

fermentation broth in the feed compartment (F) resulting in pH control. From compartment F

configured by the AMX and PC-200D membranes, carboxylate and inorganic anions migrated to

the product compartment (P) through the PC-200D membrane while all the cations were

retained. Compartment P, formed by PC-200D and BM, produced protons and received the

mixture of carboxylates and inorganic anions coming from compartment F. Hence, it contained

carboxylic acids and inorganics.

The effective surface area of each membrane was 100 cm2. The spacer thickness of the

rubber gasket and silicone net was 0.2 cm. The EDBM stack consisted of 4 vessels each having

a submersible pump (Hendor N 11 PP). Electrodes were connected to the constant current

supply (Delta Electronika SM 60 – 100). The electrode vessel (E) was filled with 5 L 0.5 M

Na2SO4 solution.

5.2.3. Off-line (batch mode) EDBM experiments

To obtain sufficient fermentation broth for off-line tests, reactor effluents were collected

daily between days 20 and 70 of continuous operation. After each collection, effluent was

centrifuged at 1750 G for 15 minutes and stored at - -line tests were performed under

3 different constant currents: 0.23, 1.0 and 1.5 A. Vessel F were filled with 5 L reactor effluent

and vessel B with 5 L 0.1 M NaOH solution. Vessel P was filled with 5 L 0.1 M HCl in the off-line

test of 0.23 A, and 3 L 0.1 M HCl in the off-line tests 1.0 and 1.5 A. Each off-line test was run

for at least 2 days. Conductivity measurement was recorded on-line and pH was measured off-

line each day. The flow over vessel F was 30 L/h while it was 50 L/h in vessel P and B. The

operational conditions of the off-line tests can be found in Table 5.1.

142

5.2.4. In-situ coupled experiments

Coupled experiments were performed with the same membrane configuration as the off-

line tests (Figure 5.1b). An ultrafiltration (UF) unit was placed before the EDBM stack to

separate the cells and large particles from the fermentation broth to avoid blocking the

membranes. The UF consisted of one module equipped with a tubular PVDF membrane (30 nm

pore size, 0.025 m2 surface area, Pentair, The Netherlands). The flux was set to 19 L.m2/h,

corresponding to a filtrate flow of 0.47 L/h. A peristaltic pump (Watson Marlow, 520U)

circulated the fermentation liquor through the UF unit at a cross-flow velocity of 1 m/s. In

between the coupling experiments, the recirculation was kept running to avoid volume

disturbances in the reactor, and the ED stack was circulated with deionized water.

The UF permeate was pumped to the EDBM feed loop with a peristaltic pump

(Masterflex 7518-10). As opposed to the off-line ED tests, compartment F was replaced by a

direct feed line of the permeate line from the UF unit to the EDBM stack and back to the

fermenter. In the EDBM stack, an internal recirculation was provided on the feed line by a

peristaltic pump (Watson Marlow, 520U) at a flow of 24 L/h in the 1st coupled experiment. In

the 2nd and 3rd experiments, the flow was decreased to 12 L/h. The total liquid volume of

reactor, UF unit, feed loop of the EDBM stack and tubing was 4 L. A schematic diagram of the

reactor coupled with the UF and EDBM units is shown in Figure 5.2.

In each coupled experiment, vessel P was filled with 4 L 0.1 M HCl and vessel B with 4 L

0.1 M NaOH. In the last experiment, to avoid water transfer from compartment B to P, the

concentration of HCl in vessel P was dropped to 0.05 M HCl in a final volume of 3 L. The liquids

in the vessels P and B were both recirculated with a flow of 50 L/h. Operational conditions of

143

the coupled tests are presented in Table 5.1. Three coupled tests were executed between day

78 and 119 of continuous fermenter operation.

Table 5.1. Operational conditions of off-line and in-situ coupled experiments

Current Vessel

Off-line experiments

Base Feed Product Electrolyte

1 0.23 0.1 M NaOH (5 L) Centrifuged

fermentation

broth (5 L)

0.1 M HCl (5 L)0.1 M HCl (3 L)0.1 M HCl (3 L)

0.5 M Na2SO4 (5 L)2 1.03 1.5

Coupled experiments

1 0.6 0.1 M NaOH (4 L) Fermentation broth after UF (4 L)

0.1 M HCl (4 L) 0.5 M Na2SO4 (4 L)2 0.4 -> 0.253 0.3 0.05 M HCl (3 L)

Figure 5.2. A schematic diagram of reactor coupled with UF and EDBM units.

144

Ideally, the pH of the reactor was regulated during the coupled tests by the OH- ions

generated by the BM. In case the EDBM failed to do so, the separate pH controller on the

fermenter dosed either 2 M NaOH or 2 M HCl to keep the pH of the reactor at 4.7.

5.2.5. Calculations

Current efficiency represents the fraction of current used to transfer the ions through

the membrane as a function of time. Current efficiency can be calculated as:

= z. n. FN.I.t

(5.1)

where z is the charge number (e.g. 1 for acetate and chloride), n is the moles of ions

transferred (mol), F is the Faraday constant (96485 C/mol), N is the number of cells, I the

applied current (A), and t is the time of applied current (s).

Water flux through the product compartment P from the compartment B was calculated

as:

Jw= dVAtot

m . dt(5.2)

where dV is the volume change in compartment P (mL), Atotm is the total membrane

surface area (m2), and dt is the time of operation of the EDBM stack. Flux for ions was

calculated with the same formula by replacing the nominator with the molar change of the

specific ion taking into account the volume change.

Productivity of the in-situ coupled EDBM-fermenter system was calculated as:

Productivity=((C

ave,reactor-Cave, influent).Ffeed.t)+mpro

Ffeed.t . D (5.3)

145

where Cave, reactor is the average concentration of total carboxylates measured in the

reactor (mmol/L), Cave,influent is the average concentration of total carboxylates in the influent

tank Ffeed is the feed flow to the reactor (L/d), t is the total duration of the coupling period (d),

mpro is the total amount of carboxylates accumulated in the product vessel P until the end of

the coupling (mmol) and D is the dilution rate of the reactor (d-1).

Mass balance of carboxylates was calculated as:

Recovery (%)=Carboxylatesrecovered by EDBM

Carboxylatesrecovered by EDBM + Carboxylatesdiscarded(5.4)

where Carboxylatesrecovered is the total carboxylates recovered from the reactor by EDBM

over the period that the EDBM was connected to the reactor (mol),

and Carboxylatesdiscarded by EDBM is the total carboxylates discarded from reactor (mol).

5.3. Results and Discussion

5.3.1. Acetate was the major carboxylate in continuous mixed culture fermentation

Figure 5.3 shows the product spectrum and concentrations generated in the continuous

fermenter throughout the operation period with and without EDBM integration. Acetate was the

major compound with concomitant n-butyrate production from fermentation of the waste

stream under the selected conditions before in-situ recovery was implemented. After a start-up

period for the fermentation, a peak in carboxylate production could be observed concomitant

with complete consumption of lactate present in the feed. The average total carboxylate

concentration was 166 mM in the continuous reactor until day 26. In this period, lactate

concentration in the incoming waste was around 25 mM. Figure 5.3 shows that lactate in the

146

feed was consumed completely. After day 26, a new batch of waste was fed to the reactor and,

the average total carboxylate concentration level dropped from 166 mM to 120 mM.

147

Figu

re 5

.3.

Conc

entr

atio

n ch

ange

of

carb

oxyl

ates

, la

ctat

e an

d he

adsp

ace

prod

ucts

as

a fu

nctio

n of

tim

e in

the

con

tinuo

us r

eact

or.

A ne

gativ

e co

ncen

trat

ion

refe

rs t

o co

nsum

ptio

n of

con

stitu

ents

pre

sent

in t

he in

com

ing

was

te. (

NW

: N

ew b

atch

of

was

te;

ED1:

Ele

ctro

dial

ysis

tes

t 1,

etc

.).

148

5.3.2. Carboxylate flux was comparable at different currents applied in off-line batch EDBM

experiments

From day 20 to 70, effluent samples were collected for off-line ED tests. In that period,

acetate and n-butyrate were the 2 major products at average concentrations of 53 mM and 39

mM respectively. While acetate and n-butyrate were initially produced at comparable

concentrations, gradually a shift to acetate was observed with lower but often similar

propionate and n-butyrate levels. These results differ from our previous study (Arslan et al.,

2013), where n-butyrate was the main carboxylate compound over 64 days. The reason is that

the pH was then kept at 4.0 to promote n-butyrate production, while it was now increased to

4.7 to provide a higher fraction of charged ions, required for removal by EDBM.

For the off-line tests, 3 current levels were selected. As the limiting current density for

the stack, and operational conditions had been determined experimentally to be around 150

A/m2 (1.5 A), these levels were fixed at 0.23, 1 and 1.5 A. Based on carboxylate productivities

observed in previous continuous fermentations and the use of 5 cell pairs in the ED stack, this

range was expected to be in the right order to reach an equilibrium between carboxylate

produced in fermentation and removed through ISPR by EDBM, as aimed for the coupled tests.

Since the 3 off-line tests had a different duration with the shortest one running for 21 h,

the results will be compared after a similar duration of operation, and not at the end of the

tests. Figure 5.4 presents the total carboxylate amount as a function of time in vessel P under

three different constant currents. It was expected to observe a trend with increasing current

applied. However, except for one aberrant analysis measured for the 1 A case around 25h of

operation, the amount of carboxylates transferred was quite similar in all cases. Average

carboxylate fluxes from compartment F to P during the first 21 h of operation were quite

149

comparable and amounted to 0.41 mol/m2.h at 0.23 A; 0.36 mol/m2.h at 1.0 A and 0.49

mol/m2.h at 1.5 A (Table 5.2).

Figure 5.4. Accumulating carboxylate amounts in product vessels during the offline tests.

According to flux values it would presume low and similar final carboxylate

concentrations in the feed vessel but this is not observed. For each experiment, the final

carboxylate concentration was different. Thus, it was expected to observe the highest fluxes

would occur at the highest current. The reason might be the total amount of carboxylates and

the relative composition initially present in the feed differed substantially (Table 5.2). As the

type of carboxylate and its concentration in the fermenter varied between the 3 mixed samples

(each collected over 1 week) (Figure 5.3), the test results were compared after normalization.

When the amount of carboxylates transported was normalized for the initial amount present in

the feed, the normalized amount increased in the order 1 A – 0.23 A – 1.5 A. Importantly, this

coincides with a decrease in relative ratio of chloride versus total carboxylates present in the

0

300

600

900

0 20 40 60 80

Tota

l car

boxy

late

s (m

mol

)

time (h)

0.23 A 1.0 A 1.5 A

150

feed. It is unavoidable to transport chloride since chloride is smaller than carboxylates and the

second major negatively charged ion in the reactor. Although the PC200D membrane is

somehow selective for carboxylates, smaller negatively charged molecules will pass though

membrane too. This explains why a relatively lower level of chloride resulted in relatively higher

carboxylate transport.

Table 5.2. Initial feed conditions, current efficiencies and fluxes in offline tests under variable applied currents

Applied Current (A) 0.23 1 1.5

Initial carboxylate concentration in feed vessel (mg/L) 12 121 16 486 8 650

Carboxylate/chloride ratio in feed (mol/mol) 3.5 2.5 4.5

Average Carboxylate Flux (mol/m2.h)* 0.41 0.36 0.49

Acetate 0.24 0.22 0.27

Propionate 0.03 0.03 0.05

Butyrate 0.15 0.11 0.17

Average Current Efficiency (%)* 56 38 22

Average Carboxylate Current Efficiency (%)* 48 12 15

Final carboxylate concentration in product vessel (mg/L) 11 538 15 813 9 925

*After approximately 21 h of operation

Because acetate is the smallest molecule and present at the highest concentration

among the carboxylates in the reactor, acetate flux was the highest comparing to other

carboxylates (Table 5.2). After acetate, n-butyrate flux was the second highest for the off-line

experiments. Because propionate concentration was a lot lower comparing to the acetate and

n-butyrate levels in the effluent, its flux was very low and reached a maximum of only 0.05

mol/m2.h under 1.5 A.

151

The average carboxylate current efficiencies were 48% at 0.23 A, 12% at 1.0 A and

15% at 1.5 A. They were higher initially and showed a tendency to drop due to ion depletion in

the compartment F. When the limiting current density for carboxylates was exceeded, instead

of carboxylate transport, current was mainly spent on transporting of other anions, mainly OH-

and chloride and water splitting in BM. Also, back diffusion of carboxylates from product

compartment to feed due to the concentration gradient might occur and these carboxylates

might be subsequently transported back to product side once more. It should be noted that the

calculated current efficiency for the 0.23 A test run may be rather inaccurate since the

measured carboxylate levels were oscillating to some extent, probably due to analytical errors.

The carboxylates had a major contribution to the overall current efficiencies, which were in the

range 22-56% and dropped with increasing current applied (Table 5.2). In literature, batch-wise

EDBM studies with model solution (Habova et al., 2004) and real fermentation broth (Cheng et

al., 2008) resulted in much higher current efficiencies of 70% - 80% for one major product

lactate. However, Wang et al. (2006) and Vertova et al. (2009) detected current efficiencies

ranging between 20% and 50% for separation of organic carboxylates mixtures from waste

effluent by batch-wise ED operation. In our experiments, we observed a slight biofilm formation

on the PC200D membrane surface. Thus, membrane fouling can be a reason for the lower

current densities/efficiencies. Another reason can be co-transport of OH-, some organic and

inorganic anions from the waste material, which were not analysed. Finally, ‘membrane

poisoning’ can occur, a type of fouling on AEM due to humic acids in water (Kobus & Heertjes,

1973). When larger molecules like humic acids cannot pass through the AEM, their

concentration at the membrane can increase and this may in turn increase the membranes

electrical resistance.

152

As the organic ions were replaced by hydroxyl ions, the pH of the compartment F

increased to around 12 in each off-line test. Because the off-line tests were run for only 1 or 2

days, no water transport could be observed.

5.3.3. Simultaneous in-situ carboxylates removal and pH control was achieved when EDBM was

integrated into continuous fermenter

After performance of the off-line ED tests and when a rather stable carboxylate

production was achieved in the fermenter, the EDBM set-up was coupled to the reactor. As a

result, the average total carboxylate level in the fermenter dropped from 120 mM to 70 mM.

This is in line with the fact that at pH 4.7, half of the acids are present in dissociated form.

Table 5.3 presents the recovered amount of carboxylates by EDBM from the reactor during the

operation period. For the experiment at 0.6 A, it was observed that recovered carboxylates

were 62%. However in this experiment, pH control was not optimal yet and OH- produced by

BM was higher than demand. This resulted in more dissociated ions and higher recovery by

EDBM. Later at 0.4 A and 0.3 A, the recoveries were around 50%. Interestingly, the acetate

and n-butyrate concentrations in the fermenter dropped and became comparable, while

propionate concentration remained similar and its relative ratio to the other carboxylates

increased.

In the proposed novel ‘direct contact’ EDBM concept, the amount of anions removed

from the EDBM compartment F is balanced by the hydroxyl ions entering it. Hence, recycling

the compartment F solution to the fermenter allows pH control in the fermenter without

external base addition. The critical point in such a ‘direct contact’ approach is evidently the very

close linkage between the organic anion recovery and the generation of hydroxyl ions in the

BM. As other than the target carboxylate ions will also be removed from the compartment F,

153

the risk exists that hydroxyl ion addition exceeds the level required for pH control. On the other

hand, the organic waste substrate and fermentation broth with variable composition represent a

certain buffering capacity also determining the base requirements. The major goal of the

coupled tests was thus to fine-tune the applied current, such that sufficient carboxylate removal

was obtained simultaneously with adequate pH control, completely eliminating external base

addition.

Table 5.3. Process comparison of test period with and without EDBM coupling and under variable applied currents

Applied Current (A) No ED coupled 0.6 0.4 0.25 0.3

Volt over stack 13.2 8.9 6.7 6.7

Average current efficiency for

carboxylates %

20 10 13

Average current efficiency for

carboxylates and chloride %

31 18 19

Flux (mol/m2.h)

Acetate 0.20 5x10-2 6x10-2

Propionate 8x10-2 2x10-2 2x10-2

Butyrate 6x10-2 2x10-2 1x10-2

Chloride 0.23 0.17 0.13

Water flux (L/m2.h) 0.19 0.20 0.15

NaOH consumed in reactor (mmol/day) 80 0 0 36 0

NaOH production by EDBM (mmol/day) 273 94 114

HCl consumed in reactor (mmol/day) 0 108 264 0 0

Yield

(mmol carboxylate/L.d/g CODfed/L.d)

2.23 3.89 2.35 1.85

Recovery of carboxylates (%) 62 56 44

Energy kJ / g carboxylate separated 60 110 25 18

154

For the integrated tests, the EDBM set-up was coupled to the fermenter with an

intermediate UF unit for biomass removal. To avoid any excess base dosing, a preliminary

coupled experiment was performed at the lowest current (0.23 A), that was evaluated in the

off-line tests and that was calculated to be in the right range to remove the produced quantity

of carboxylates. However, under those conservative conditions, no significant effect on external

base dosing was observed (not shown). As base generation was obviously too low, the

following coupling test was performed at a near doubled constant current of 0.6 A. Under this

condition, carboxylates were depleted from the feed loop of the EDBM and also drastically

dropped in the fermenter (Figure 5.3). The amount of OH- produced by the BM was higher than

needed (Table 5.3). As a result of this, pH of the reactor started to raise and the external pH

control system started to add HCl to keep the reactor pH at 4.7 (Table 5.3). This was not how

the EDBM should be operating. Therefore, the current was reduced to 0.4 A in a second test. As

external HCl was still added to the reactor, the current was decreased from 0.4 A down to 0.25

A after 50 h. Under 0.25 A, a slight external NaOH addition was detected, indicating that OH-

production from BM was not sufficient. When the EDBM was temporarily uncoupled, the

carboxylate level in the fermenter quickly returned to its previous steady state (Figure 5.3). This

is a good indication that a fairly stable carboxylate production must have been maintained

during the first two coupled tests, in spite of temporarily high HCl addition. In a final third

coupled experiment, one of the most important targets of this study was achieved. At a slightly

increased current of 0.3 A, external base addition was completely eliminated (Figure 5.5).

155

Figu

re 5

.5.

On-

line

pH a

nd N

aOH

con

sum

ptio

n da

ta f

rom

con

tinuo

us r

eact

or.

(

) in

dica

tes

pH,

(

) in

dica

tes

the

NaO

H a

dded

fr

om e

xter

nal p

H c

ontr

ol t

o th

e re

acto

r

Coup

ling

1Co

uplin

g 2

Coup

ling

3

156

The fastest total carboxylate recovery was obtained at the highest current applied

(Figure 5.6a). Particularly in the first hours, the ion transport was faster comparing to the tests

at lower currents. This is also apparent from the current efficiency trends (Figure 5.7). At a

current of 0.6 A, the current efficiency was 30% for total carboxylate ions in the first 10 hours

when carboxylates were depleted fast. Later it dropped to lower than 20%. In later experiments

where current was 0.4 and 0.3 A, the highest current efficiencies both started from much lower

initial values than 30%. It is important to note though that this corresponded to the use of

virgin membranes in the first experiment. In later cases, the membranes were saturated with

ions, which most probably explains why the current efficiency never reached initial values as

high as 30% any more. For the tests at 0.4 A and 0.3 A the total amount of carboxylates

transferred was similar and current efficiencies for carboxylate ions were around 12% (Table

5.3). A current of 0.3 A was thus the optimal condition for in-situ coupling and pH control in this

set-up, but the current efficiency was low. The current efficiencies for the coupled tests were

lower than for the off-line batch experiments and low compared to related literature.

157

(a)

(b)

Figure 5.6. Accumulated (a) carboxylates and (b) chloride amounts as a function of time in the product vessel for the applied currents. The arrow indicates that the current was decreased from 0.4 A to 0.25 A in the second experiment

158

Figure 5.7. Current efficiency for carboxylates as a function of time in three coupled EDBM experiments.

Du et al. (2012) worked with in-situ coupling of ED to single culture fermentation and

Redwood et al. (2012) with EDBM. Both studies reported current efficiencies around 55%.

Constant current of 0.05 A was applied in study of Du et al. (2012), while current varied

between 0.01 A and 0.1 A in Redwood et al. (2012). Some reasons for the low current

efficiencies were already discussed before, but their effects may be more pronounced in

coupled tests, because the coupled tests lasted longer than the off-line ones. Another reason

for the low current efficiency in the coupled experiment might be depletion of available

carboxylates in the reactor (Wang et al., 2013). At pH 4.7, only half of the carboxylic acids were

in the dissociated form. In all experiments, half of the total carboxylate products in the reactor

were transported within the first 15 h. In a higher pH zone, there would be more charged

carboxylic ions and recovery rate would be higher too.

0,0

0,1

0,2

0,3

0,4

0 50 100 150 200 250

Curr

ent e

ffici

ency

time (h)

0.6 A 0.4 -> 0.25 A 0.3 A

159

Chloride fluxes were as high as carboxylate fluxes and increased as the applied current

increased. At current 0.6 A and 0.4 A, the chloride amount transported from the reactor was

higher than the total carboxylates (Figure 5.6b), due to external HCl addition. To which extent

the anions are separated selectively from each other depends on the membrane selectivity and

the feed solution itself (Strathmann, 2010). As chloride is the smallest anion, its transport via

AEM is inevitable and unfortunately no membrane exists to our knowledge that allows only

carboxylate passage.

5.3.4. In-situ EDBM coupling to fermenter steer the fermentation for selective acetate and

propionate production

An interesting aspect of coupling EDBM to the fermenter, lies in the potential avoidance

of side-reactions and alleviation of toxicity. When looking at the yields (Table 5.3) and

productivities (Figure 5.8) in the various integrated tests, this effect was most pronounced in

the first coupled test. Carboxylate productivity increased 1.4 times when EDBM was coupled

and operated at 0.6 A. The lower the applied current, the lower the carboxylate productivity.

However, currents lower than 0.6 A are preferable, because a better pH control by EDBM was

realized.

The low productivity in the last coupling test may have been due to operational issues. A

problem with influent supply caused lower carboxylate productivities and gas generation (Figure

5.3). When EDBM was uncoupled, carboxylate productivity was only half of that in the previous

stand-alone period (Figure 5.8). Hence, the productivity in the third EDBM test probably should

be compared with that last period of operation without EDBM. Waste fermentation processes

run with mixed culture are efficient but are sometimes difficult to handle at lab-scale due to

160

clogging issues. Together with waste variability, this is extra challenging for fully integrated

ISPR testing, as aimed for in this work.

Figure 5.8. Average carboxylate productivities before, during and after each ED coupling

Interestingly, the productivity for each carboxylate showed different trends. In the first

ED test, acetate and propionate productivities selectively doubled while n-butyrate productivity

remained stable. In the second test, acetate and propionate productivities were only slightly

higher than in the stand-alone test period, but n-butyrate productivity was only half as high.

Because acetate and propionate are smaller molecules compared to n-butyrate, they are

expected to be transported faster. As a consequence, it seems that the removed product was

produced more in the reactor. This finding definitely needs more investigation, since it is a big

research question how to drive mixed culture fermentation towards a certain product or

161

combination of products. If this can be achieved by EDBM systems, for sure it will bring another

dimension to the research field of in-situ coupling to mixed culture fermentation.

5.3.5. Significant water transport towards product side limits the concentration degree of

carboxylates

As much as separating carboxylates from the fermentation broth, concentrating them is

crucial for industrial application. In our tests, the maximal concentration obtained in the vessel

P was 4.5 times the concentrations found in the reactor. Water fluxes in the experiments with

0.6 A and 0.4 A were equal to each other around 0.20 L/m2.h although current was different. In

the last experiment at 0.3 A, water flux was 0.15 L/m2.h. Water fluxes are presented for

different test conditions in Table 5.3. Because of an increasing concentration gradient in

compartment P against compartments F and B, significant water transport (125 ml/day)

towards compartment P was realized, proportional with carboxylate ion passage. Around 80%

of the total water transported was from vessel B. The rest of water increase might be due to

the passage of H+ and OH- ions produced in the BM and recombine in the product

compartment. This might also explain why the current efficiency was partially spent for OH- ion

passage and resulted lower values for carboxylates.

To reduce the water transport impact, the concentration of HCl solution was reduced

from 1 M to 0.05 M and the vessel P volume was decreased from 4 L to 3 L in the last

experiment. Indeed, water flux in the last experiment was less, however not sufficient to

improve carboxylate levels. Similar studies conducted with EDBM by Redwood et al. (2012) and

Wang et al. (2013) observed water transport of around 70 ml/day and up to 700 ml/day

respectively, though Wang et al. (2013) used a smaller membrane area. Huang et al. (2006)

indicated that electro-osmosis is a main problem of EDBM processes. Apparently, water

162

transport capacity depends on many parameters such as membrane specific resistance, current

efficiency, type of the ion, concentration gradient between the compartments, etc. (Robbins et

al., 1996; Okada et al., 1998).

5.4. Challenges

Carboxylate production via mixed culture processes may reduce the product cost by

eliminating expensive sterilization processes and avoiding the need of pure substrates.

However, extra downstream processing efforts are required since carboxylate concentrations in

the final broth are lower and the target product has to be separated from a more complex

mixture. Pure culture continuous fermentation systems could reach carboxylate productivities of

1 mol/L.d with 90% purity (Rogers et al., 2006). Such levels of total carboxylates are not found

in mixed culture systems, let alone of one specific carboxylate. In this study, average total

carboxylate of 70 mmol/L.d were obtained on a waste substrate, compared to 44 mmol/L.d in a

non-coupled reactor. Acetate was the major product at 47% fractions. Based on the test

results, separation of carboxylates from such fermentation product spectrum will need 18 kJ/g

carboxylate separated (Table 5.3). The measured purity and separation rates of the coupled

system are still low and the process is costly compared to today’s conventional carboxylate

production. As already indicated in Huang et al. (2006), it is not economically and practically

feasible to apply EDBM to dilute feed solutions. Further research focus should be on reducing

the adverse impacts of membrane fouling to increase the current efficiency e.g. by applying

pre-treatment methods to feed solution. Water transport issue needs to be minimized to

increase intensification of target product. To increase the productivities and purities, the

integration of other separation techniques to EDBM such as using ion exchange resins (Zhang

et al., 2011; Du et al., 2012) or using ionic liquids (Lopez & Hestekin, 2013) can be investigated

163

further for in-situ integrated sustainable mixed culture carboxylate production. However,

combining EDBM with other techniques will increase the production cost further making the

process less compatible in industry.

5.5. Conclusions

A novel EDBM concept with “direct contact” operation mode was applied for the first

time to a mixed culture fermentation on organic waste. In-situ separation of carboxylates could

be demonstrated with simultaneous pH control in the fermentation. Through stepwise

optimization, the supply of external base could be completely eliminated. In-situ coupling of

EDBM to a continuous reactor showed increased total carboxylate productivities. Especially,

productivities of acetate and propionate were selectively doubled, possibly because they were

selectively removed from the reactor. In future work, competition with chloride and most

importantly water transport issues need to be eliminated to achieve higher degrees of up-

concentration.

164

Rerefences

Agler, M.T., Spirito, C.M., Usack, J.G., Werner, J.J., & Angenent, L.T. (2012). Chain elongation with reactor microbiomes: upgrading dilute ethanol to medium-chain carboxylates. Energy & Environmental Science, 5(8), 8189.

Alkaya, E., Kaptan, S., Ozkan, L., Uludag-Demirer, S., & Demirer, G.N. (2009). Recovery of acids from anaerobic acidification broth by liquid-liquid extraction. Chemosphere, 77(8), 1137–1142.

Anasthas, H.M., & Gaikar, V.G. (2001). Adsorption of acetic acid on ion-exchange resins in non-aqueous conditions. Reactive and Functional Polymers, 47(1), 23–35.

Arslan, D., Steinbusch, K.J.J., Diels, L., De Wever, H., Hamelers, H.V.M., & Buisman, C.J.N. (2013). Selective carboxylate production by controlling hydrogen, carbon dioxide and substrate concentrations in mixed culture fermentation. Bioresource Technology, 136(0), 452–460.

Bailly, M., Roux-de Balmann, H., Aimar, P., Lutin, F., & Cheryan, M. (2001). Production processes of fermented organic acids targeted around membrane operations: design of the concentration step by conventional electrodialysis. Journal of Membrane Science, 191(1-2), 129–142.

Chen, C., & Ju, L.K.K. (2002). Coupled lactic acid fermentation and adsorption. Applied Microbiology and Biotechnology, 59(2-3), 170–4.

Cheng, G., Wang, Q., Sun, X., Meng, H., & Li, J. (2008). Experimental study on concentration of ammonium lactate solution from kitchen garbage fermentation broth by two-compartment electrodialysis. Separation and Purification Technology, 62(1), 205–211.

Du, J., Lorenz, N., Beitle, R. R., & Hestekin, J. A. (2012). Application of Wafer-Enhanced Electrodeionization in a Continuous Fermentation Process to Produce Butyric Acid with Clostridium tyrobutyricum. Separation Science and Technology, 47(1), 43–51.

Gao, M.T.T., Shimamura, T., Ishida, N., & Takahashi, H. (2011). pH-Uncontrolled lactic acid fermentation with activated carbon as an adsorbent. Enzyme and Microbial Technology, 48(6-7), 526–30.

Habova, V., Melzoch, K., Rychtera, M. M., SekavovÃ¡, B., Hábová, V., & Sekavová, B. (2004). Electrodialysis as a useful technique for lactic acid separation from a model solution and a fermentation broth. Desalination, 162, 361–372.

Huang, C., & Xu, T. (2006). Electrodialysis with Bipolar Membranes for Sustainable Development. Environmental Science & Technology, 40(17), 5233–5243.

Huang,T.,Zhang, Y., Xue, Y., Chen, G.,C.X., & Huang, C. (2007). Application of electrodialysis to the production of organic acids: State-of-the-art and recent developments. Journal of Membrane Science, 288, 1–12.

Iizuka, A., Hashimoto, K., Nagasawa, H., Kumagai, K., Yanagisawa, Y., & Yamasaki, A. (2012). Carbon dioxide recovery from carbonate solutions using bipolar membrane electrodialysis. Separation and Purification Technology, 101, 49–59.

Kobus E.J.M & Heertjes, P.M. (1973). The poisoning substances of anion-selective membranes by humic substances.Desalination, 12, 333–342.

Lee, G.E., Moon, S.H.H., Keun Chang, Y., Yoo, I.K.K., & Nam Chang, H. (1998). Lactic acid recovery using two-stage electrodialysis and its modelling. Journal of Membrane Science, 145(1), 53–66.

165

Lopez, A.M., & Hestekin, J.(2013). Separation of organic acids from water using ionic liquid assisted electrodialysis. Separation and Purification Technology, 116, 162–169.

Novalic, S., Kongbangkerd, T., & Kulbe, K.D. (2000). Recovery of organic acids with high molecular weight using a combined electrodialytic process. Journal of Membrane Science, 166(1), 99–104.

Okada, T., Xie, G., Gorseth, O., & Kjelstrup, S. (1998). Ion and water transport characteristics of Nafion membranes as electrolytes. Electrochimica Acta,43(24), 3741–3747.

Phillips, J.R., Klasson, K.T., Clausen, E.C., & Gaddy, J.L. (1993). Biological Production of Ethanol from Coal Synthesis Gas. Applied Biochemistry and Biotechnology, 39, 559–571.

Redwood, M. D., Orozco, R. L., Majewski, A. J., & Macaskie, L. E. (2012). Electro-extractive fermentation for efficient biohydrogen production. Bioresource Technology, 107, 166–174.

Robbins, B.J., Field, R.W., Kolaczkowski, S.T., & Lockett, A.D. (1996). Water flux through anion exchange membranes, 118, 101–110.

Rogers, P., Chen, J.S.,Zidwick, M.J. (2006). The prokaryotes: Organic Acid and Solvent Production: Acetic, Lactic, Gluconic, Succinic, Polyhydroxyalkanoic Acids Propionic Butyric acid and aLcohols. Prokaryotes, 1, 511–755.

Strathmann H., (2004). Ion-exchange membrane Separation Processes, Membrane Science and Technology Series 9, Elsevier B.V., Amsterdam.

Strathmann H., (2010). NanoMemCourse Handout, University of Twente, Enschede, the Netherlands.

Urbanus, J., Bisselink, R.J.M., Nijkamp, K., ter Horst, J.H., Verdoes, D., & Roelands, C.P.M. (2010). Integrated product removal of slightly water-soluble carboxylates from fermentation by electrochemically induced crystallization. Journal of Membrane Science, 363, 36–47.

Vertova, A., Aricci, G., Rondinini, S., Miglio, R., Carnelli, L., & D’Olimpio, P. (2009). Electrodialytic recovery of light carboxylic acids from industrial aqueous wastes. Journal of Applied Electrochemistry, 39(11), 2051–2059.

Woodley, J.M., Bisschops, M., Straathof, A.J.J., & Ottens, M. (2008). Future directions for in-situ product removal (ISPR). Journal of Chemical Technology and Biotechnology, 83(2), 121–123.

Wodzki, R.& Nowaczyk, J. (2002). Propionic and acetic acid pertraction through a multimembrane hybrid system containing TOPO or TBP. Separation and Prurification Technology,26, 207–220.

Wang,Y., Luo, Y., & Yu, P. (2006). Rocovery of organic acids from waste salt solutions derived from the manufacture of cyclohexanone by electrodialysis. Journal of Membrane Science, 280, 134–137.

Wang, Y., Zhang, N., Huang, C., & Xu, T. (2011). Production of monoprotic, diprotic, and triprotic organic acids by using electrodialysis with bipolar membranes: Effect of cell configurations. Journal of Membrane Science, 385-386, 226–233.

Wang, X., Wang, Y., Zhang, X., & Xu, T. (2012). In situ combination of fermentation and electrodialysis with bipolar membranes for the production of lactic acid: Operational compatibility and uniformity. Bioresource Technology, 125(0), 165–171.

Wang, X., Wang, Y., Zhang, X., Feng, H., & Xu, T. (2013). In-situ combination of fermentation and electrodialysis with bipolar membranes for the production of lactic acid: Continuous operation. Bioresource Technology, 147, 442–448.

166

Wee, Y.J., Yun, J.S., Lee, Y., Zeng, A.P., & Ryu, H.-W. W. (2005). Recovery of lactic acid by repeated batch electrodialysis and lactic acid production using electrodialysis wastewater. Journal of Bioscience and Bioengineering, 99(2), 104–108.

Wong, M., Woodley, J., & Lye, G. (2010). Application of bipolar electrodialysis to E. coli fermentation for simultaneous acetate removal and pH control. Biotechnology Letters, 32(8), 1053–1057.

Yi, S.S., Lu, Y.C.; Luo, G.S. (2008). Separation and concentration of lactic acid by electrodialysis. Separation and Purification Technology, 60, 308–314.

Zhang, Y. (2011). Wastewater Reclamation and Ion Fractionation by Electrodialysis. Ph.D. thesis, Katholieke Universiteit Leuven. Leuven.

Zhang, X., Li, C., Wang, Y., Luo, J., & Xu, T. (2011). Recovery of acetic acid from simulated acetaldehyde wastewaters: Bipolar membrane electrodialysis processes and membrane selection. Journal of Membrane Science, 379(1-2), 184–190.

167

168

You have to do it by yourself and you cannot do it alone.

Martin Rutte

169

Chapter 6

General Results and Discussion

170

6.1. Introduction

This thesis investigates how to steer the mixed culture fermentation processes towards

desired end products at high concentration. The focus was set on the impact of three major

parameters on carboxylate concentration and spectrum: (1) the substrate type and

concentration in liquid phase (2) the composition and the concentrations of gas compounds in

headspace and (3) integration of in-situ product recovery to mixed culture fermentation. In this

chapter, the outcomes of this research, the limitations that we encountered and

recommendations for applicability of the process will be addressed.

6.2. The impact of substrate type and concentration on carboxylate production

Food wastes generally contain a mixture of carbohydrate, protein and lipid polymers. In

batch experiments, as a carbon source, 3 industrial organic waste streams different in

carbohydrate, lipid and protein composition were selected to observe the individual effect of

polymer compositions on carboxylate type and concentration (Chapter 3).

Up to now, many studies have been applying mixed culture fermentation with pure

substrates such as glucose, starch (Hussy et al., 2003; Khanal et al., 2004) and different

organic waste streams, such as artificial food waste (Okamoto et al., 2000; Komemoto et al.,

2009, Dong et al., 2009), kitchen food waste (Min et al., 2005), manure (Coats et al., 2011),

dairy wastewater (Bengtsson et al., 2008), paper waste (Ueno et al.,2007), organic fraction of

municipal solid waste (Cavdar et al., 2011). Literature studies reported the maximum

carboxylate production in batch studies to be around 20 g carboxylate/L for food or

carbohydrate rich waste substrates. For single protein- and lipid-based streams, the total

carboxylate concentrations are on average below 3 g/L. When pure substrates (mostly glucose)

171

are fermented in fed batch and fed-batch systems, single products like acetate or n-butyrate

can reach up to 200 g/L (Straathof, 2014).

Table 6.1 represents the results of the experiments for each substrate in batch tests. In

our study, the highest total carboxylate production was obtained with the carbohydrate waste

and amounted to 3.7 g/L. The lower carboxylate production compared to literature was due to

lower initial substrate concentration in our experiments. For the protein waste, total carboxylate

concentration was 1.5 g/L and for the lipid waste lower than 0.1 g/L. Relatively low carboxylate

concentration in the protein and lipid waste compared to carbohydrate waste is caused by the

low hydrolysis rate (Christ et al., 2000). Increasing the hydrolysis rate will also increase

carboxylates concentrations.

It is difficult to make a direct link between carbohydrate type of substrate and product

selectivity from the literature mixed culture fermentation studies. However, the highest product

selectivity reported in literature was 80% acetate from mixed culture fermentation of

carbohydrate rich waste stream (Nachiappan et al., 2011). Protein-rich substrates often yield

odd-numbered carboxylates such as propionate and i-valerate with maximum 35% fraction in

mixed culture fermentation (Chen et al., 2007; Ucisik & Henze, 2008). Product selectivity can

reach higher fractions up to 90% in pure culture processes. This is not commonly realized by

open fermentation of organic wastes.

172

Tabl

e 6.

1.Co

ncen

trat

ion

chan

ge o

f pr

oduc

ts a

nd fr

actio

ns (

in b

rack

ets)

for

each

hea

dspa

ce c

ondi

tion

with

incr

easi

ng in

itial

sub

stra

te c

once

ntra

tion

for

carb

ohyd

rate

and

pro

tein

ric

h w

aste

str

eam

sin

bac

th t

ests

Initi

al s

ubst

rate

co

ncen

trat

ion

&

type

8 g

COD

car

bohy

drat

e w

aste

/L8

g CO

D p

rote

in w

aste

/L13

.5 g

CO

D c

arbo

hydr

ate

/L23

CO

D c

arbo

hydr

ate

/L

Hea

dspa

ce

H2

CO2

N2

H2

CO2

N2

H2

CO2

N2

H2

CO2

N2

Prod

ucts

(g/

l)Af

ter

1 w

eek

Acet

ate

1.3

(45%

)-0

.1a

1.3

(51%

)0.

2 (4

6%)

0.4

(62%

)0.

6 (6

5%)

-0.3

a0.

3 (1

2%)

1.5

(41%

)2.

8 (4

3%)

2.9

(45%

)2.

0 (2

7%)

Prop

iona

te<

0.1

(<1%

)0.

1(3

%)

0.9

(34%

)<

0.1

(4%

)<

0.1

(4%

)<

0.1

(4%

)0.

1 (4

%)

0.1

(5%

)0

0.2

(3%

)0.

3 (4

%)

0.1

(2%

)

n-Bu

tyra

te1.

4 (4

6%)

0.8

(95%

)0.

4 (1

4%)

0.1

(23%

)0.

1 (1

9%)

0.2

(17%

)1.

5 (9

5%)

1.6

(79%

)1.

8 (5

1%)

3.3

(49%

)3.

2 (4

9%)

5.3

(71%

)

n-Ca

proa

te0.

2(6

%)

00

<0.

1 (8

%)

<0.

1 (2

%)

00

0.1

(3%

)0.

2 (6

%)

0.2

(4%

)0.

1 (2

%)

0

Tota

l car

boxy

late

s2.

90.

92.

50.

50.

61

1.5

23.

66.

66.

67.

5

Afte

r 4

wee

ks

Acet

ate

2.3

(48%

)1.

7(4

6%)

2.0

(54%

)0.

8 (5

2%)

1.5

(74%

)1.

2 (7

8%)

1.9

(41%

)1.

3 (2

9%)

3 (77%

)6 (5

6%)

4.4

(47%

)3.

1 (3

4%)

Prop

iona

te0.

1 (3

%)

0.2

(7%

)1.

1 (3

0%)

<0.

1 (3

%)

0.1

(4%

)<

0.1

(2%

)0.

5 (1

0%)

0.4

(10%

)0.

1 (3

%)

0.2

(2%

)0.

5 (5

%)

0.4

(4%

)

n-Bu

tyra

te1.

7 (3

6%)

1.4

(39%

)0.

5 (1

3%)

0.4

(26%

)0.

3 (1

3%)

0.1

(7%

)1.

8 (4

0%)

2.0

(47%

)0.

4 (1

1%)

4.1

(39%

)4.

0 (4

3%)

5.6

(61%

)

n-Ca

proa

te0.

4 (8

%)

0.1

(3%

)0 (0

%)

0.1

(6%

)<

0.1

(1%

)0

0.2

(6%

)0.

3 (8

%)

0.2

(6%

)0.

3 (3%

)0.

3 (3

%)

0

Tota

l car

boxy

late

s4.

83.

63.

71.

52.

01.

54.

54.

34.

010

.79.

39.

2

a Neg

ativ

e va

lues

indi

cate

con

sum

ptio

n of

com

poun

ds

NO

TE:T

est

with

lipi

d ric

h w

aste

is o

mitt

ed b

ecau

se t

he t

otal

car

boxy

late

s w

as lo

wer

tha

n 0.

5 g/

L fo

r al

l hea

dspa

ce c

ondi

tions

. Als

o, e

xper

imen

ts w

ith H

2an

d CO

2he

adsp

ace

cond

ition

ar

e ex

clud

ed f

rom

the

tab

le b

ecau

se t

hese

yie

lded

abu

ndan

ce o

f ace

tate

by

cons

umin

g he

adsp

ace

gase

s in

depe

nden

t of

the

sub

stra

te t

ype

and

conc

entr

atio

n. T

he r

esul

ts a

re p

rese

nted

af

ter

1 w

eek

and

4 w

eeks

sin

ce t

he p

rodu

ct d

istr

ibut

ion

diff

ers

sign

ifica

ntly

as

a fu

nctio

n of

tim

e.

173

Acetate was the major product for all substrate types 4 weeks after incubation. However

the acetate concentrations and fractions varied with substrate type. The maximum acetate

concentration was 2 g/L at 54% of the total carboxylates on carbohydrate waste. The

concentration of acetate was only 1.2 g/L on protein waste. This was lower than on

carbohydrate waste, but its relative fraction was 78%, much higher than on carbohydrate

waste. The selectivity obtained in this experiment was one of the highest reported in literature.

In literature it was indicated that with protein rich substrates yield propionate as a dominant

product. This is different than our observation. The reason might be that propionate rich

experiments were performed at neutral pH (Ucisik & Henze, 2008; Feng et al., 2009) while in

the experiment with protein waste, pH was kept at 5.0. This might cause dominant acetate

formation rather than propionate.

Initial substrate concentration

In principle, short chain carboxylate production increases with increasing carbohydrate

content of the substrate in the reactor. Because our target was to direct mixed culture towards

maximum achievable concentrations, a set of experiments was designed to increase product

concentrations and observe the impact of increasing substrate concentration on carboxylate

formation. (Chapter 4)

In literature, the studies indicate that for mixed culture process, although substrate

concentration increased up to 100 g COD/L or 15 g VS/d, total carboxylate concentration

stabilizes around 20 g/L. This might be related with the inhibitory effect of acids on mixed

culture. When substrate concentrations become too high in short retention times in batch

fermentation, or under conditions of feed overloading in continuous, product inhibition may

occur. The inhibition phenomena in mixed culture system was exceptionally overcome by Fu &

174

Holtzapple (2011). They reported 40 g carboxylate/L through reducing the contact time of acid

and organism by applying “counter-current” fed-batch fermentation technique. By this method

the inhibitory effect of carboxylates on organism was eliminated. Literature reports that the

product shifts with increasing substrate concentration from acetate to n-butyrate or even n-

caproate (if enough incubation time provided) (Fang et al., 2006; Hafez et al., 2010).

In our study, batch reactors were operated at 3 different initial substrate concentrations,

8 g COD waste/L, 13.5 g COD/L and 23 g COD/L with only carbohydrate rich substrate for a

month. The result of increasing substrate concentration in batch reactors was that the total

carboxylate concentration raised from 2.5 g/L to 7.5 g/L (Table 6.1). For the selectivity, it was

observed that the product type shifted from acetate towards n-butyrate as the substrate

concentration increased. n-Butyrate selectivity increased from 17% to 70% when substrate

concentration was increased from 8 g COD waste/L to 23 g COD/L (~ 4 g VS/d). This is in line

with reported values in literature.

Limited carboxylate production in mixed culture fermentation

Carboxylate concentration in mixed culture fermentation is often reported much lower

than pure culture. Pure culture, in optimal conditions for that specific culture, can convert 98%

of the glucose to the desired product. In mixed culture fermentation, even rich in sugar type of

waste substrates, the yield rarely reaches higher than 55% in literature. The reason for this is

that each organism has different optimal conditions and it is difficult to find the best conditions

for several cultures at the same time. Also, the organisms show competition for substrate and

electron sources and several ions and impurities in the waste streams decrease the metabolic

performance of the organisms. Because hydrolysis is a limiting factor, conversion of protein and

lipid type of wastes usually remains even lower than 20%.

175

In our experiments, the carboxylate concentrations were even lower than what was

obtained usually in literature. We were not able to reach the high carboxylate concentration

partially due to the operational parameters such as pH and the low hydrolysis rate of starch. It

is often reported in literature that the maximum carboxylate concentrations are obtained at

neutral pH values while in our experiments the pH was lower than 5.0. For the hydrolysis of

starch in fermenter, many studies apply heat or enzymatic pre-treatment of starch. Substrate

pre-treatment was not considered in our study to prevent higher costs. Increasing the inoculum

concentration from 1% up (weight inocula/volume reactor) might improve the conversion rate

of substrates, but this might lead to changes in product spectrum as well. Aslo, increasing

substrate concentration higher than 23 g COD/L might help to increase the total carboxylate

concentration.

6.3. The impact of headspace on carboxylate production

Hydrogen and carbon dioxide are two gas compounds involved in fermentation reactions

either by being produced or consumed. Certain fermentation pathways can be energetically

suppressed or favored by hydrogen and carbon dioxide levels in the headspace (Thauer et al.,

1977). Accordingly, it was anticipated that by changing the levels of hydrogen and carbon

dioxide in headspace, the fermentation reactions can be steered towards a certain direction.

In our headspace study, the experiments were designed by adding 2 bar of hydrogen, 2

bar carbon dioxide and 2 bar of hydrogen and carbon dioxide mixed by 50:50 (v:v) to the

headspace of batch reactors. A control reactor was prepared by flushing the headspace with

nitrogen at the beginning of the experiment, without any further adjustment during the

experiment. Each headspace condition was tested with carbohydrate, protein and lipid rich

176

substrates and initial substrate concentrations of 8, 13.5 and 23 g COD/L with only

carbohydrate waste (Chapter 4).

Thermodynamic approach

Table 6.2 presents the reaction equations from glucose to acetate, propionate and n-

butyrate and how headspace gases contribute these reactions. According to the

thermodynamics, because acetate and n-butyrate production from glucose produces hydrogen

and carbon dioxide, their presence in headspace is not favorable for acetate and n-butyrate

production (equations 6.1 and 6.3 in Table 6.2). Propionate production becomes

thermodynamically more spontaneous under hydrogen presence while carbon dioxide does not

have impact since it is not involved the reaction (equation 6.2). All free enthalpies are much

lower than the thermodynamical limit for spontaneous reactions of -20 kJ/mol. So, even under 5

bar pressurized headspace conditions, all reactions can still occur, making it difficult to predict

the product spectrum in advance.

Table 6.2. Equations for different fermentation reactions

Reaction Reaction Equation r°' (kJ/mol)a

Eqn.

Acetate from glucose C6H12O6 + 2H2O 2 acetate- + 2H+ + 4H2 + 2CO2 -292.3 6.1

Propionate from glucose C6H12O6 + 2H2 2 propionate- + 2H+ + 2H2O -308.5 6.2

n-Butyrate from glucose C6H12O6 n-butyrate- + H+ + 2H2 + 2CO2 -309.4 6.3

Acetate by homoacetogens

2CO2 + 4H2 acetate- + H+ + 2H2O -138.2 6.4

ar°') values is calculated at 30°C and a pH of 7 under 1 atm pressure.

Concentrations of reactants and products in aqueous solution are 1 M.

177

The results of the experiments for each headspace condition are represented in Table

6.1. The experiments showed that the results differed from the thermodynamic predictions. In

case of hydrogen-filled headspace, it could be expected to have propionate since this reaction

consumes hydrogen (equation 6.2). On the contrary, in our experiments, under 2 bar hydrogen

partial pressure, propionate formation was lower than 10% in any substrate condition and

acetate and n-butyrate were the two main compounds. According to the equations 6.1 and 6.3,

carbon dioxide is produced in all reactions. So, it was unclear which product would be favorable

when head space was filled with 2 bar carbon dioxide. At substrate concentrations lower than

23 g COD/L, n-butyrate was the major product after a week, after that the product distribution

shifted to equal fraction of acetate and n-butyrate. Hydrogen can promote the chain elongation

by involving in the reactions as electron donor. Thus, another expectation was that at hydrogen

pressurized reactor, n-caproate would be the dominant product according to thermodynamic

data (not shown). Indeed, n-caproate production reached 8% fraction in 2 bar hydrogen head

space after 4 weeks and at the lowest substrate concentration. Because n-caproate formation

from acetate and n-butyrate needs long residence time, higher concentration and fraction could

be observed with experiments longer than 4 weeks in a hydrogen pressurized reactor.

Carboxylate formation under modified headspace reactors

There are only few studies available in literature about the effect of hydrogen and

carbon dioxide on selective production of one carboxylate. These studies mainly worked with

the removal of fermentation gases by sparging or flushing with nitrogen or carbon dioxide. The

effect of sparging on total carboxylate production is found negligible (Hussy et al., 2003; Kim et

al., 2006) or even negative (Kyazze et al., 2005; Massanet-Nicolau et al., 2010). Addition of

extra carbon dioxide to the headspace increased the n-butyrate fraction relative to the other

178

fermentation products (Kim et al., 2006). No obvious trends can be derived from the impact of

elevated hydrogen partial pressure on the short chain carboxylate levels and composition. In

some studies, a high partial pressure of hydrogen favored the production of longer chain

carboxylates and more reduced compounds (Smith & McCarty, 1989; Yu & Mu, 2006). Neither

sparging nor flushing with nitrogen was found to affect the carboxylate composition.

The experimental results showed that headspace gases have a direct impact on the

concentration of products. The carboxylate concentration increase was almost 17 g/L in

hydrogen and carbon dioxide mixed headspace with protein rich substrate. With carbohydrate

rich substrates, the increase was around 10 g/L in all conditions. The drastic increase in this

headspace condition was mainly due to homoacetogenic conversion of headspace gases to

acetate (equation 6.4). For the reactors solely having 2 bar hydrogen, the total carboxylate

production increase was aproximately 1.5 g/L more than the control reactors. For the reactors

solely having 2 bar carbon dioxide, the carboxylate concentrations were either equal or slightly

higher than the control reactors. The headspaces in these reactors were not consumed.

Acetate as a major product

Addition of a mixture of hydrogen and carbon dioxide favored acetate formation. The

carboxylate produced in this case was almost 85% acetate, independent of the substrate type.

With increasing substrate concentration the fraction was dropped to 66%. The acetate highest

concentration was 16 g/l with protein rich substrate and mixed hydrogen and carbon dioxide

headspace. Almost all acetate produced in this conditions was not produced by conversion of

organic waste but mainly because of homoacetogenic conversion of headspace, which makes

this condtion not interesting for the carboxylate platform concept. Acetate was also abundant

under any headspace condition where protein rich substrate was used.

179

Propionate as a major product

From the findings it was observed that under any headspace manipulated condition,

propionate was not formed at all. In the condition where headspace was not controlled and the

lowest carbohydrate rich substrate concentration, propionate fraction reached to 31% with only

1 g/L concentration. This was the highest selectivity obtained overall. Once substrate

concentration was elevated, propionate production disappeared.

n-Butyrate as a major product

n-Butyrate production was affected by headspace manipulation depending on the

substrate concentration and experiment period. Carbon dioxide partial pressure at 2 bar in the

headspace promoted the selective production of n-butyrate within a week of incubation at

carbohydrate substrate 8 g COD/L substrate concentration. Here, n-butyrate concentration was

low (~1 g/L) and selectivity reached 95%. In fact, 95% n-butyrate production is promising

result for mixed culture process, though the concentration is quite low. Our observation is in

line with Kim et al., (2006) who observed 80% n-butyrate with 4.5 g/L concentration when they

sparged the reactor with carbon dioxide. After 4 weeks in our experiments, n-butyrate fraction

was only 40%.

At substrate concentration of 13.5 g COD/L, n-butyrate was selectively produced in

similar fractions under only hydrogen and only carbon dioxide pressurized headspace after 1

week. Similar to the previous case, its fraction decreased to around 45%. The concentration of

n-butyrate was not higher than 2 g/L in these conditions. In the experiment with the highest

substrate concentration 23 g COD/L, n-butyrate was dominant in the reactor when there was no

headspace manipulation. n-butyrate selectivity was 71% with 5.3 g/L in 1 week and 61% with

5.6 g/L in 4 week fermentation.

180

Headspace addition in industrial applications

Although headspace manipulation elevated the carboxylate concentration, the increase is

either not enough to meet industrial standards for application (>50g/L) or it is achieved by

transformation of headspace gases rather than organic waste. Selectivity of carboxylates

increased for some reactors with manipulated headspace but the trends were aberrant and

depending on the substrate concentrations. High carboxylate selectivity together with high

concentration could be obtained without manipulating headspace conditions and by high

substrate concentration.

Continuous experiment for stable carboxylate production

A continuous experiment was done to validate the results obtained in the batch

experiments. Such validation is crucial for practical applications as continuous mode is preferred

because it (i) is easier to operate, (ii) allows steady state operation under optimal conditions,

and (iii) is more robust to perturbations. According to the batch experiments, n-butyrate

production could be maximized to 5.3 g/L with almost 71% fraction without applying external

hydrogen or carbon dioxide, within 1 week of batch fermentation. Therefore, 2 days HRT was

considered to be enough for the continuous test. The pH of the system was regulated at 4.0,

being pH of the batch reactor where n-butyrate reached its maximum levels.

From the continuous mixed culture studies in literature, it was observed that carboxylate

concentration was not reaching higher than 20 g/L with average productivity of 1.5 g/L.h

independent of operational condition. In continuous fermentation with pure substrates, product

concentrations are higher than in mixed culture. For instance, n-butyrate concentration can be

40 g/L with very high production rate of 9 g/L.h which is 3 to 4 times higher compared to batch

or fed-batch pure substrate fermentation (Rogers et al., 2006). Table 6.3 shows the normally

181

achievable concentrations and productivities obtained in pure and mixed culture continuous

processes.

Table 6.3 Normally achievable concentrations and productivities obtained in pure and mixed

culture continuous processes

Fermentation Substrate Concentration

(g/L)

Productivity

(g/L.h)

End product

Pure culture Glucose 40 9 n-Butyratea

Mixed culture Glucose 17 1.6 Mixed carboxylatesb

Mixed culture Organic waste 20 1.5 Mixed carboxylatesc

Mixed culture Organic waste 10 0.2 Mixed carboxylatesd

aMichel-Savin et al., 1990b70% n-butyrate 30% acetate (Kyazze et al., 2005)cUsually obtained in efficient mixed culture of organic waste substrate process in literature d67% n-butyrate 25% acetate (maximum and stable concentration obtained in chapter 4 of this study)

In the continuous experiment conducted in this study, total carboxylate concentration

was around half of the highest value mainly reported for mixed cultures. Maximum and stable

concentration was 10 g/L with productivity only 0.20 g carboxylate/L.h. n-Butyrate with 70%

fraction was selectively produced over a month of continuous reactor. The reasons for low

carboxylate concentration compared to literature studies might be due to similar reasons

explained for batch experiments, such as low pH, low substrate concentration and limited

degradation of substrate. The reason for low productivity in continuous experiment is not very

clear.

6.4. The impact of iin-situ product recovery on carboxylate production

One of the most important targets of this study was to apply in-situ product recovery

technology to increase the carboxylate production. By continuously removing the products from

182

the fermenter, reactions thermodynamically shift towards the product side. Another purpose of

in-situ coupling is to steer the fermentation towards selective production of a desired

carboxylate. It was also expected that the in-situ recovery would selectively enrich a single

compound due to differences in concentration between the different carboxylates. Finally in-situ

product recovery could be beneficial since production and separation are achieved in one step,

reducing production costs. In this thesis, the focus was on the integration of ED as in-situ

separation technology. By incorporating bipolar membranes, the pH of the fermentation can be

regulated internally instead of adding external base addition. Considering the fact that high

carboxylates production is achieved near neutral pH values, using bipolar membrane for large

scale production applications can reduce the chemicals cost.

Recovery and purification of products from the fermenter is quite energy-intensive and

costly. In mixed culture processes downstream processes require more energy compared to

pure systems due to the low product concentrations. In literature, there is a research gap for

ISPR in mixed culture processes. In conventional pure culture acetate production, the product is

already concentrated in the final broth making downstream process more economically viable

than for mixed culture. For some high yield acetic acid process no recovery process other than

cell mass removal is needed (Lopez-Garzon & Straathof, 2014). However, when very high purity

is desired, distillation can be used.

In pure culture processes, ED is mainly used as a concentration step after a separation

unit so it is difficult to make a comparison with our study. However, fermentation with

simultaneous recovery (e.g. extractive fermentation) can promote the carboxylate (e.g. butyric

acid) productivity 8 to 10 times (Yang, 2007).

183

In our experiments, we coupled the continuous reactor together with an ED bipolar

membrane (EDBM) (Chapter 5). A novel membrane stack configuration was proposed,

consisting of two different types of AEMs for selective separation of negatively charged

carboxylate ions from the fermentation broth while OH- ions are generated by bipolar

membrane and pumped into the fermentation broth. By this way, OH- ions were directly

transported from the EDBM stack to the reactor in so called “direct contact” operation mode. An

ultrafiltration (UF) unit was placed before the EDBM stack to separate the cells and large

particles from the fermentation broth to avoid membranes blocking.

Preliminary experiments indicated that in-situ recovery of carboxylates from mixed

culture fermenter did not improve the productivities in the desired direction, and concentration

effect remained in limited range. Carboxylate productivity was only 1.5 times higher, after

approximately 15 days of operation.

The low efficiency of ED in our experiments was mainly due to membrane fouling

occurring within couple of hours. We observed biofilm growth on the anion exchange

membrane surface. Another drawback of our set-up was that the fluctuating carboxylate

production due to instabilities in waste material and difficulties like clogging of tubing of

fermenter due to solid particles in the waste stream. For sure, large scales fermenters can be

more robust than small lab scale equipment. However organic waste material should be

screened carefully and in case of big fluctuations in feed composition, waste material should be

adjusted accordingly. More research on integration of membrane technology especially to avoid

fouling and how to keep the stability on the overall process is required to show the

effectiveness and potential for mixed culture fermentation.

184

Oher targets of this study was to control pH of the fermenter by OH--production at the

bipolar membrane. The production of OH- ion by the bipolar membrane balanced the acid

production in the reactor at a current of 0.3 A. The pH was controlled at 4.7. Because at this

pH, half of the carboxylates are dissociated form, the recovery was stable around 50%.

There is a strong link between pH and carboxylate productivity. For mixed culture

processes, total carboxylate concentration increases when pH reaches to neutrality. The highest

carboxylate concentration reported in literature was obtained at pH 6.5 and 7 (Fu & Holtzapple,

2011; Nachiappan et al., 2011). In our experiments, the pH was kept lower than 5.0 to avoid

high alkali consumption. From our experiences, increasing the pH from 4.0 to 4.7 in continuous

mixed culture process, at a carboxylate productivity around 0.2 g/L.h, additionally 0.25

kg/month more was dosed. Ideally, to remove the base addition completely, the pH can be

controlled by the coupled bipolar membrane ED in the fermenter. This mechanism separates

the acids in the ion forms from the reactor and simultaneously produce sufficient alkalinity to

eliminate the external base needed for fermentation. This would remove the chemical

consumption completely while regulating the pH in desired zones.

Relative concentration of acetate and n-butyrate in total carboxylates were higher in

equal concentrations. Consequently, it was foreseen that acetate and n-butyrate would separate

from the fermenter most and their productivities would selectively increase. However, it was

observed that acetate together with propionate were selectively separated and their

productivities increased more compared to n-butyrate. The reason is that acetate and

propionate are shortest carbon chain carboxylates and they pass through membranes faster

under electrical field. The selectivity and purity specifically to certain products could increase by

185

additional techniques such as ion exchange resins. This would bring, however, much extra costs

to the process.

Mixed culture fermentation as an industrial application

Currently, industries producing organic waste streams use anaerobic digestion (AD) to

convert their wet streams into methane and electricity or heat energy. To become a preferable

process in industry over AD, there are some standards that mixed culture process still needs to

meet. Table 6.4 compares the process performances of two competing technologies; AD and

mixed culture fermentation for a carbohydrate rich waste stream studied in this thesis.

Table 6.4. Process performance comparison of anaerobic digestion (AD) and mixed culture fermentation (MCF)

Vol.loading rate

(kg COD/m3.d)a

Reactor volume (m3)

HRT (d)

Yield

(%)b

CH4

(Nm3/d)cAcid

(kg/d)d

Energy

(kWh/d)e

€/d €/m3

AD 20 350 3.5 70 1206 2534 238f 2.4

MCF 60 3818 1909g 19

aBase case applied in continuous reactor in Chapter 4. Daily amount of waste needs to be treated is assumed 7000 kg COD. Flow rate is 100 m3/d.bYields are maximum reported in literature.cBiogas composition is assumed to be 65% CH4 and 35% CO2. The amount of CH4 was back calculated from the COD value (4 kg O2/kg CH4).dAll produced acids are assumed in the form of acetate.e2.1 kWh/m3 biogas fElectricity cost is 0.094€/kWhgAcetate price is 500€/tons

Table 6.4 shows that the turnover in case of methane production is lower than for mixed

culture fermentation. The feedstock cost, capital and operational costs for methane digester

and acid fermenter are mainly comparable, although different reactor designs will be chosen.

The most important advantage of methane production is that it separates naturally making it

easy to collect from the headspace avoiding the need for special downstream process. The

186

investment costs after digestion comprise mainly equipment to convert the biogas to electricity,

e.g. gas engines or more efficient combined heat and power plants (CHP). Transport of biogas

is difficult and expensive, hence it should be avoided.

The costs for the production of the acids are substantially higher than the biogas

separation and utilisation, especially separating and concentrating the acids. From Table 6.4, it

is clear that to meet the standards of a feasible application, the total expenses of downstream

process should be lower than 20 €/m3. First and foremost, it is necessary to produce acid

concentrations higher than 50 g/L in the fermentation broth (Lopez Garzon & Straathof, 2014).

The ideal separation technology for mixed culture fermentation

It is difficult to draw a conclusion which separation technology would be the most ideal

for mixed culture fermentation. Currently, crystallization is commonly used in industry,

adsorption and solvent extraction techniques are extensively discussed in literature. Among

these studies, ED technology is still the most suitable technology for ISPR due to its

compatibility. However, selectivity and mass transport are still low and make the process

combined with high maintenance and membrane costs economically unfeasible. Crystallization

and ion exchange resins techniques provide higher recovery yields and selectivity than ED. But,

for ion exchange resins, the carboxylates are usually recovered as salts while industrial

applications mainly use the acid form. For crystallization method, water remaoval (usually by

evaporation) and cooling is required so that crystallization can occur. Also, crystallization and

adsorption techniques cannot be considered sustainable processes, due to the chemical

consumption to regenerate the acids and the resulting wastewater that needs to be treated.

Table 6.5 compares the most ideal ISPR techniques. Solvent extraction though widely studied in

187

literature, is not included in Table 6.5. Because its toxic effect, it is least suited as ISPR

technology in biological processes.

Table 6.5. The most ideal and highest potential ISPR applications for mixed culture fermentation

ED Adsorption Crystallization

Robustness + + +

Performance +/- + +

Product selectivity +/- + +

ISPR compatibility + +/- -

Cost - + +/-

Sustainability + - -

Main challenges -Membrane costs &

performance

-Mass transport

limitations

-Wastewater generation

-Chemical use

-Product regeneration

-Waste generation

(gypsum)

-Chemical use

6.5. Potential applications of mixed culture carboxylate production

It seems that carboxylate production by mixed culture fermentation, the concentration

and rates are difficult to reach as high and pure as with single culture fermentation. Depending

on the economics of the production process, mixed culture carboxylate production can be used

for different applications. Unfortunately, elaborated cost analysis of downstream of mixed

culture process is not available in literature. The paper of Granda et al. (2009) gives insight

about the fermentation and dewatering costs of the carboxylate production mixed culture

(Table 6.6). The results indicate that prices would be even higher for diluted fermentation broth

since more energy intensive separation techniques would be required. This would make the

carboxylate platform process less suitable for direct industrial applications. However, the

products from the carboxylate platform can still be used as building blocks in secondary

188

processes such as medium chain fatty acids, alcohols or polymers production, avoiding

intermediate and costly purification steps.

Table 6.6. Capital and operational costs of fermentation and dewatering unit in mixed culture fermentation of municipal solid wastea

Fermentationb,c Dewatering

Fixed capital cost (M€) 34 22.8Operational cost (€/ton biomass)d 96.9 34

aThe capacity of carboxylate production plant is 40 tons/h. bMunicipial solid waste was pre-treated with lime before fermentation. The pre-treatment is included to the fermentation costs.cTotal carboxylate concentration is 50 g/L.dOperational cost covers feedstock, utility, labor costs and fixed charges defined in Granda et al., 2009

Fermentative production of acetate is mainly used in vinegar manufacturing. Propionate

and n-butyrate are mainly used in food industry as preservative or flavouring agents, antibiotic

drug preparation and cosmetic sector and also in cellulose based plastic production for textile,

filter and membrane production. However, the carboxylates for these applications should have

high purities and prices ideally lower than petro-chemical based carboxylates. Considering the

low yields and large expenses of concentrating carboxylates from mixed culture fermentation,

direct conversion of mixed carboxylates into plastics or mixed alcohols by chemical post-

processing might be economically more feasible. In such case, a second process which would

receive the effluent from fermentation and would directly upgrade the short chain carboxylates

either in second fermentation step, electrochemical or chemical conversion. The products of

secondary process can be alcohols, medium and longer chain fatty acids due to their higher

carbon to oxygen ratio, have higher energy content compared to short chain carboxylates or

polymers, surfactants and flocculants which are valuable chemicals used as plastics, cleaning

189

agents, in water treatment systems, etc. Such secondary products cannot be produced directly

from biomass with mixed culture due to complex structure of biomass. Only single organisms

can provide the conversion in specific, controlled conditions. Thus short chain carboxylates are

excellent precursors for production of the secondary products by mixed culture. Upgrading of

short chain carboxylates at low concentrations is that the maximum solubility drops and the

hydrophobic effect increases. This results in improved separation capability of the desired

products, like caproate or bioplastics.

Several approaches for secondary applications exist but still not realized in plant

application yet. MixAlco process is one of the pilot plant examples of biological conversion of

carboxylates to higher grade products. In MixAlco, mixed carboxylates are firstly produced from

mixed substrates by mixed culture fermentation. Later they are converted through thermal

conversion into first ketones then to mixed alcohols which can be used as gasoline. Some other

examples of this approach exist already, through mainly at laboratory scale by Holtzapple’s

group. An acetate rich stream was upgraded to ethanol by Steinbusch et al. (2008), an acetate-

ethanol mixture to n-caproate and caprylate (Grootscholten et al., 2012; van Eerten-Jansen et

al., 2013) and n-butyrate-glucose mixture to butanol (Richter et al., 2011). Short chain

carboxylates can also be converted to polyhydroxyalkanoates (PHA), or polymers (Lemos et al.,

2006; Srikanth et al., 2012), surfactants and flocculants (Li & Yu, 2011). For all these

applications, except MixAlco concept, steering the fermentation process towards certain product

or combination of products is still quite important. Because the composition of acids generated

from the fermentation has significant impact on the structure and physical properties of the

polymer, surfactant, single alcohol, etc. produced.

190

It might be necessary to include an additional cell separation or a pre-treatment unit

between carboxylate production and second application processed. For example, for PHA

production nitrogen deficiency in the feeding regime can be required, so carboxylate rich

effluents need to be pre-treated to remove nitrogen for efficient PHA production. Optimization

of in-situ recovery of alcohols is still being developed. For alcohol production from short chain

carboxylates, like the conversion of n-butyrate to butanol, sugar addition at certain

concentrations is necessary though it is unwanted from an economical perspective (Richter et

al., 2011). Also, alcohol should be constantly removed from the continuous fermentation to

avoid inhibition (Perez et al., 2013). Bio-electrochemical application of short chain carboxylates

to medium chain carboxylates requires pre-treatment to remove organics which can interfere

with the reduction of acetate, propionate and n-butyrate. Apart from pre-treatment, bio-

electrochemical process is still costly due to not yet optimized electrodes and membranes.

6.6. Concluding remarks

Carboxylate Platform is a promising technology for utilization of large amounts of

organic waste into valuable chemicals as in the form of short chain fatty acids. The most

important advantage of carboxylate platform is that the conversion is achieved by mixed

culture. In a mixed culture fermentation, the costs are drastically reduced since no sterilization

step is required and cheap feedstock is abundantly and locally available. Yet, there are still

major issues limit the mixed culture process in industrial applications. Low and instable

productivities challenge downstream processes and increase the product costs drastically. Using

waste as substrate is potential to generate high fractions up to 70-80% but usually in low

concentrations (<20 g/L).

191

High carboxylate concentrations can be achieved by altering the process

design basically. Holtzapple’s group used 4 stage semi-continuous reactor in sequence to allow

enough retention time for substrate to hydrolyse while avoiding the contacts of acids with

organism too long. This way of reactor design can be investigated further with alternative

operational parameters e.g. pH lower than 6 and substrate types e.g. carbohydrate rich food

waste. Changes in operational parameters in each stage can be investigated to optimize the

process and to observe the shift between product type.

ED technology can be a promising solution because of the possibility to in-situ integrate

into the fermentation process. Besides, the chemical use in ED is significantly lower than in

other separation techniques such as crystallization, solvent extraction or adsorption. However,

high membrane and electric energy costs limits ED application to high-added value products.

Overcoming the membrane fouling problems would increase the efficiency of coupling.

Adsorption is another separation technique which is promising for carboxylate separation. A

comparative ISPR study between ED and adsorption techniques is necessary to observe the

potentials for downstream applications. .

Potential promising applications of short chain carboxylates generated by open

fermentation lies in the conversion of these chemicals via a secondary post-processing step to

medium chain fatty acids or polymers.

192

References

Bengtsson, S., Hallquist J., Werker, A., Walender, T. (2008). Acidogenic fermentation of industrial wastewaters: Effects of chemostat retention time and pH on volatile fatty acids production. Biochemical Engineering Journal, 40, 492–499.

Cavdar, P., Yilmaz, E., Tugtas, A.E., Calli, B., (2011). Acidogenic fermentation of municipal solid waste and its application to bio-electricity production via microbial fuel cells (MFCs). Water Science and Technology, 64, 789 – 795.

Chen, Y., Jiang, S., Yuan, H., Zhou, Q., & Gu, G. (2007). Hydrolysis and acidification of waste activated sludge at different pHs. Water Research, 41(3), 683–689.

Christ,O., Wilderer, P.A., Angerhöfer R.R., Faulstich, M., (2000). Mathematical modeling of the hydrolysis of anaerobic processes. Water Science and Technology, 41, 61 – 65.

Coats, E.R., Gregg, M., & Crawford, R.L. (2011). Effect of organic loading and retention time on dairy manure fermentation. Bioresource Technology, 102(3), 2572–2577.

Dong, L., Zhenhong, Y., Yongming, S., Xiaoying, K., & Yu, Z. (2009). Hydrogen production characteristics of the organic fraction of municipal solid wastes by anaerobic mixed culture fermentation. International Journal of Hydrogen Energy, 34(2), 812–820.

Fang, H.H.P., Li, C., & Zhang, T. (2006). Acidophilic biohydrogen production from rice slurry. International Journal of Hydrogen Energy, 31(6), 683–692.

Feng, L., Chen, Y., & Zheng, X. (2009). Enhancement of waste activated sludge protein conversion and volatile fatty acids accumulation during waste activated sludge anaerobic fermentation by carbohydrate substrate addition: The effect of pH. Environmental Science and Technology, 43(12), 4373–4380.

Fu, Z., & Holtzapple, M. (2011). Anaerobic thermophilic fermentation for carboxylic acid production from in-storage air-lime-treated sugarcane bagasse. Applied Microbiology and Biotechnology, 1–11.

Granda, C., Holtzapple, M., Luce, G., Searcy, K., & Mamrosh, D. (2009). Carboxylate Platform: The MixAlco Process Part 2: Process Economics. Applied Biochemistry and Biotechnology, 156(1), 107–124.

Grootscholten, T.I.M., Steinbusch, K.J.J., Hamelers, H.V.M., & Buisman, C.J.N. (2012). Chain elongation of acetate and ethanol in an upflow anaerobic filter for high rate MCFA production. Bioresource Technology, 135, 440 –445.

Hafez, H., Nakhla, G., El. Naggar, M.H., Elbeshbishy, E., Baghchehsaraee, B., & Naggar, M.H. (2010). Effect of organic loading on a novel hydrogen bioreactor. International Journal of Hydrogen Energy, 35(1), 81–92.

Hussy, I., Hawkes, F.R., Dinsdale, R., & Hawkes, D.L. (2003). Continuous fermentative hydrogen production from a wheat starch co-product by mixed microflora. Biotechnology and Bioengineering, 84(6), 619–626.

Khanal, S.K., Chen, W.H.H., Li, L., & Sung, S. (2004). Biological hydrogen production: effects of pH and intermediate products. International Journal of Hydrogen Energy, 29(11), 1123–1131.

Kim, D.H.H., Han, S.K.K., Kim, S.H.H., & Shin, H.S.S. (2006). Effect of gas sparging on continuous fermentative hydrogen production. International Journal of Hydrogen Energy, 31(15), 2158–2169.

193

Komemoto, K., Lim, Y. G., Nagao, N., Onoue, Y., Niwa, C., Toda, T. (2009). Effect of temperature on VFA’s and biogas production in anaerobic solubilization of food waste. Waste Management, 29, 2950–2955.

Kyazze, G., Martinez-Perez, N., Dinsdale, R., Premier, G.C., Hawkes, F.R., Guwy, A.J., & Hawkes, D.L. (2005). Influence of substrate concentration on the stability and yield of continuous biohydrogen production. Biotechnology and Bioengineering, 93(5), 971–979.

Lemos, P.C., Serafim, L.S., & Reis, M.A. (2006). Synthesis of polyhydroxyalkanoates from different short-chain fatty acids by mixed cultures submitted to aerobic dynamic feeding. Journal of Biotechnology, 122(2), 226–238.

Li, H.W., & Yu, H.Q.Q. (2011). From wastewater to bioenergy and biochemicals via two-stage bioconversion processes: A future paradigm. Biotechnology Advances, 29(6), 972–982.

López-Garzón, C.S., & Straathof, A.J.J. (2014). Recovery of carboxylic acids produced by fermentation. Biotechnology Advances. In press, Corrected proof.

Massanet-Nicolau, J., Guwy, A., Dinsdale, R., Premier, G., & Esteves, S. (2010). Production of hydrogen from sewage biosolids in a continuously fed bioreactor: Effect of hydraulic retention time and sparging. International Journal of Hydrogen Energy, 35(2), 469–478.

Min, K.S., Khan, A.R., Kwon, M.K., Jung, Y.J., Yun, Z., & Kiso, Y. (2005). Acidogenic fermentation of blended food-waste in combination with primary sludge for the production of volatile fatty acids. Journal of Chemical Technology & Biotechnology, 80(8), 909–915.

Nachiappan, B., Fu, Z., & Holtzapple, M.T. (2011). Ammonium carboxylate production from sugarcane trash using long-term air-lime pretreatment followed by mixed-culture fermentation. Bioresource Technology, 102(5), 4210–4217.

Okamoto, M., Miyahara, T., Mizuno, O., Noike, T. (2000). Biological hydrogen potential of materials characteristic of the organic fraction of municipal solid wastes. Water Science and Technology, 41(3), 25–32.

Perez, J. M., Richter, H., Loftus, S.E., & Angenent, L.T. (2013). Biocatalytic reduction of short-chain carboxylic acids into their corresponding alcohols with syngas fermentation. Biotechnology and Bioengineering, 110(4), 1066–77.

Rogers, P., Chen, J.S.,Zidwick, M.J. (2006). The prokaryotes: Organic Acid and Solvent Production: Acetic, Lactic, Gluconic, Succinic, Polyhydroxyalkanoic Acids Propionic Butyric acid and aLcohols. Prokaryotes, 1, 511–755.

Smith, D.P., & McCarty, P.L. (1989). Reduced product formation following perturbation of ethanol- and propionate-fed methanogenic CSTRs. Biotechnology and Bioengineering, 34(7), 885–895.

Srikanth, S., Reddy, M.V., & Mohan, S.V. (2012). Microaerophilic microenvironment at biocathode enhances electrogenesis with simultaneous synthesis of polyhydroxyalkanoates (PHA) in bioelectrochemical system (BES). Bioresource Technology, 125, 291–9.

Steinbusch K.J.J., Hamelers, H.V.M., Buisman, C.J.N. (2008). Alcohol production through volatile fatty acids reduction with hydrogen as electron donor by mixed cultures. Water Research, 42, 4059–4066.

Straathof, A.J.J. (2014). Transformation of Biomass into Commodity Chemicals Using Enzymes or Cells. Chemical Reviews, 114, (3), 1871–1908.

Thauer, R.K., Jungermann, K., Decker, K., (1977). Energyconservation in chemotrophic anaerobic bacteria. Bacteriology Reviews 41, 100–180.

194

Ucisik, A.S., & Henze, M. (2008). Biological hydrolysis and acidification of sludge under anaerobic conditions: The effect of sludge type and origin on the production and composition of volatile fatty acids. Water Research,42(14), 3729–3738.

Ueno, Y., Fukui, H., & Goto, M. (2007). Operation of a Two-Stage Fermentation Process Producing Hydrogen and Methane from Organic Waste. Environmental Science & Technology, 41(4), 1413–1419.

van Eerten-Jansen, M.C.A.A., Ter Heijne, A., Grootscholten, T.I.M., Steinbusch, K.J.J., Sleutels, T.H.J.A., Hamelers, H.V.M., & Buisman, C.J.N. (2013). Bioelectrochemical Production of Caproate and Caprylate from Acetate by Mixed Cultures. ACS Sustainable Chemistry & Engineering, 1(5), 513–518.

Yang, S.T., (2007). Bioprocessing Value Added Products from Renewable Resources New Technologies and Applications. Extractive Fermentation for the Production of Carboxylic Acids, p 420-447. 1st edition, Elsevier, Amsterdam, The Netherlands.

Yu, H.Q.Q., & Mu, Y. (2006). Biological hydrogen production in a UASB reactor with granules. II: Reactor performance in 3-year operation. Biotechnology and Bioengineering, 94(5), 988–995.

195

196

Summary

The production of short chain carboxylates currently depends on petro-chemical sources.

Fossil fuel sources are however depleted fast by growing industry and population. Also, these

sources contribute negatively to global warming, climate change and modern human society.

Biomass, alternative to fossil fuel, is a sustainable resource and eliminates these negative

impacts. Production of organic chemicals from biomass via anaerobic fermentation is not a new

technology, e.g. ethanol generation from corn or lactic acid from sugar cane. However, short

chain carboxylates are still predominantly produced from oil and gas due to easy accessibility

and low oil prices compared to biomass feedstock. Among the biomass resources, organic waste

streams are abundant, locally available and have little or no competition with food. Bio-waste,

therefore, can be an excellent substitute source for the generation of short chain carboxylates.

The current biological processes often convert pure substrates and use single organisms

through sterile anaerobic fermentation. This yields high conversion of substrate to a single

product in high concentrations and fractions. Substrate cost and sterile operational conditions

are two parameters increasing the production cost of bio-based products. In this thesis, the

‘carboxylate platform’ is evaluated as a cost-effective biological technology for the production of

short chain carboxylates. The principle of the carboxylate platform is based on the conversion of

cheap organic feedstock to a mixture of short chain carboxylates by using non-sterile, mixed

culture fermentation.

Short chain carboxylates are organic chemicals that can be used in industry directly or as

building blocks for other valuable chemicals like alcohols and polymers. Fermentation of bio-

waste generates abundant short chain carboxylates, acetate, propionate and n-butyrate and a

small fraction of valerate, n-caproate and ethanol. The outcome of mixed culture fermentation

197

results in a combination of fermentation products in various amounts, rather than obtaining a

maximal total concentration of a single product. It is, thus, important to be able to direct the

fermentation process towards a desired major product or combination of products in certain

fractions.

This thesis describes how mixed culture fermentation processes can be steered towards

one specific carboxylate at sufficiently high concentration. Manipulation of environmental

conditions or operational parameters and integration of in-situ product recovery technology into

the fermentation are considered key parameters to drive the reactions towards a desired end

product.

In mixed culture fermentation, depending on the substrate type, inocula and headspace

composition, and also, process parameters like hydraulic and solid retention time, organic load

and pH, carboxylate concentrations and speciation can vary considerably. Thus it is crucial to

know how to use these key elements to obtain a desired product or composition of products in

the highest concentration(s) possible. Chapter 2 reviews mixed culture fermentations studies

dealing with conversion of organic substrates towards carboxylates in batch, fed-batch and

continuous reactor configurations. Literature studies report that the use of more concentrated

carbohydrate-rich substrates, at longer residence times and at neutral pH ranges stimulates

total acid production. The total carboxylate concentration reaches maximum for most of the

studies 20 g/L at production rate of 1.5 g/L.h. Exceptionally Fu & Holtzapple (2011) reported a

total carboxylate concentration of 40 g/L by eliminating the inhibitory effect of carboxylates on

the organisms. They developed a “counter-current” fed-batch operation system to reduce the

contact time of carboxylates and organism. Product inhibition was usually reported under too

high substrate concentrations at short retention times, or under conditions of feed overloading,

198

and at low pH. When increasing pH to the neutral range, not only total acid concentration, but

also acetate and propionate fractions are elevated, though high propionate concentrations and

fractions are only reported occasionally. A high n-butyrate fraction > 70% is usually found

when pH < 6, at somewhat longer incubation or retention times or high OLRs, under carbon

dioxide atmosphere or on high substrates concentrations. Increasing residence times creates a

shift from shorter to longer chain fatty acids such as valerate or n-caproate.

In this thesis, two major key operational parameters: headspace composition and

substrate type were further investigated experimentally in Chapter 3. The aim was to observe

to what extent the headspace composition at elevated partial pressures (2 bar) affects the

fermentation process, product type and productivity in batch-wise fermentation. Three reactors

were adjusted to a final headspace pressure of 2 bar by adding hydrogen, carbon dioxide or a

50:50 (v:v) mixture of these two gases. A control reactor was flushed with nitrogen only at the

beginning of the experiment and left it without headspace manipulation. The highest

carboxylate concentration was 17 g/L at hydrogen and carbon dioxide mixed pressurized

headspace. Here, the acetate fraction reached 85% due to the homoacetogenic activity.

However, this condition is not interesting for this study since acetate was formed from

hydrogen and carbon dioxide instead of degradation of waste. In the carbon dioxide reactor,

the carboxylate concentration was only 3.6 g/L however n-butyrate fraction was 70% at the

end of the first week. One of the important targets of this study, the production of one major

compound at fractions around 70% from waste streams, was successfully achieved. The

carboxylate production in the hydrogen pressurized reactor was 1.5 g/L higher than in the

control reactor but product distribution showed equal fractions of acetate and n-butyrate. For

the effect of substrate, three different organic waste streams were selected according to their

elemental composition and polymer constituents like carbohydrates, proteins and lipids.

199

Carbohydrate waste produced higher carboxylate concentrations compared to protein and lipid

wastes. In the protein rich waste, acetate was the main carboxylate and it was produced

independently of the headspace. According to literature propionate and i-valerate were the

major compounds to be found in protein rich fermentation at neutral pH. In our case the pH

was controlled at 5, resulting in mainly acetate production. Lipid rich waste yielded

concentrations below 0.5 g/L and was not considered further.

In Chapter 4, the study on headspace composition was further investigated at

increasing concentration of carbohydrate rich substrate in batch reactor configuration. The

concentrations were defined as 8, 13.5 and 23 g COD waste/L. The final target was to define

the operational conditions for a continuous process to produce high and stable total carboxylate

concentration of a single product with high fraction by mixed culture fermentation. The

experiments showed that the most optimal carboxylate production was achieved under the

conditions of the highest substrate concentration without any headspace manipulation.

Although, the highest carboxylate concentration, 11 g/L, was detected under the hydrogen

pressurized headspace, product distribution was not in the direction of a single product. Without

headspace control, the total carboxylate level was 9 g/L with 61% n-butyrate fraction. When a

continuous reactor was operated at 23 g COD waste/L at 2 days retention time without

headspace manipulation, 7 g/L and 71% of stable n-butyrate production over a month of

operation was successfully realized.

It should be considered that the biological processes can be limited if the final total

carboxylate concentration reaches inhibitory levels for mixed culture organisms. In Chapter 5,

the process parameters were overviewed for the improvement of carboxylate formation and

spectrum with integrated in-situ recovery technology. Here, a novel ‘direct contact’

200

electrodialysis configuration with bipolar membranes was integrated to continuous mixed

culture fermentation. The significant achievement of the in-situ integrated fermentation system

was that external base addition was completely eliminated; the pH of the reactor was regulated

by the bipolar membranes. Preliminary experiments indicated that acetate and propionate

productivities were selectively increased among the mixture of carboxylates. The carboxylate

productivity only increased 1.5 times, probably because carboxylate levels in the fermenter

were lower than expected and the inhibitory effect may thus have been low. The concentration

of the product compartment was maximum 3 times the carboxylate level in the reactor. These

results were preliminary data of a research on mixed culture set-ups which needs more

investigation and optimization.

Finally, carboxylate production from organic waste with mixed culture fermentation

process was evaluated in Chapter 6. The most important outcome of this research showed

that mixed culture fermentation can be steered towards a major short chain carboxylate in

fraction of 70% (n-butyrate), by carbon dioxide addition in headspace or high substrate

concentrations without headspace manipulation. However, in general total carboxylate

concentration doesn’t exceed 20 g/L. In current industrial practice, the concentrations are five

times higher. Acetate can be produced at 100 g/L up to 95% by pure cultures. Because

products are quite diluted in mixed culture systems, it is required to integrate costly, intensive

downstream processing after the fermentation. Mixed culture fermentation should still improve

product yields. Finally, mixed culture technologies can be promising if the mixture of

carboxylates is used as a precursor for valuable chemical production such as alcohols, polymers

or medium chain carboxylates rather than applying complex purifying technologies after the

fermentation process.

201

Samenvatting

De productie van kortketenige carbonzuren gebeurt op dit moment vooral via

petrochemische weg. Almaar groeiende wereldbevolking en industrie putten niet enkel de

voorraad aan fossiele brandstoffen uit, maar zorgen ook voor opwarming van de aarde,

klimaatsveranderingen en een nefaste impact op onze huidige samenleving. Het gebruik van

biomassa kan een alternatief voor deze fossiele grondstoffen vormen en op die manier

bijdragen tot een duurzame maatschappij.

De productie van organische stoffen uit biomassa via anaerobe fermentatie is geen

nieuwe technologie, bijvoorbeeld de productie van ethanol uit mais of productie van melkzuur

uit suikerriet. Kortketenige carbonzuren daarentegen worden momenteel in hoofdzaak nog uit

olie en gas gemaakt, dankzij de vrije beschikbaarheid en de lage kostprijs van fossiele bronnen

in vergelijking met biomassa. Organische afvalstromen zijn het meest overvloedig aanwezig als

biomassa bron, ze zijn lokaal beschikbaar en concurreren niet met de voedingsketen. Bioafval

kan daarom dan ook bij uitstek als vervanging voor fossiele grondstoffen dienen bij de

productie van kortketenige carbonzuren.

De huidige biologische processen gebruiken erg zuivere substraten en reinculturen in

zgn. steriele anaerobe fermentaties. Deze zorgen voor een hoge omzettingsgraad van het

substraat naar één enkel product in hoge concentraties. De kosten voor het substraat en de

steriele procesomstandigheden leiden tot een hoge productiekost van dergelijke bio-gebaseerde

producten. In deze thesis wordt het ‘carboxylaat-platform’ geëvalueerd als kosteneffectieve

biologische technologie voor de productie van kortketenige carbonzuren. Het

carboxylaatplatform is gebaseerd op de omzetting van organische stromen tot een mengsel van

kortketenige vetzuren door middel van een niet-steriele fermentatie met gemengde culturen.

202

Korteketenvetzuren worden in de industrie gebruikt, hetzij rechtstreeks, hetzij als

bouwstoffen voor andere, waardevolle chemicaliën zoals alcoholen en polymeren. Fermentatie

van bioafval levert vooral, azijnzuur, propionzuur, n-boterzuur en een kleinere fractie valeraat,

n-caproaat en ethanol. De gemengde fermentatie –zoals toegepast in dit onderzoek- resulteert

steeds in een waaier aan producten, veeleer dan één bepaalde stof. Daarom is het belangrijk

om dit type fermentatie te sturen naar één bepaald product of een combinatie van een aantal

producten in bepaalde fracties.

Deze thesis beschrijft hoe gemengde fermentatieprocessen gestuurd kunnen worden

naar één specifiek carbonzuur in een voldoende hoge concentratie. Manipulatie van de

procescondities en integratie van in-situ product winning in de fermentatie zijn de sleutel om de

fermentatiereacties te sturen naar één bepaald product.

In gemengde fermentatie hebben inoculum en samenstelling van de headspace, samen

met proces parameters zoals verblijftijd, organische belasting en pH, een belangrijke invloed op

het type en de concentratie van de carbonzuren. Het is dan ook van belang te weten hoe deze

parameters kunnen gebruikt worden om bepaalde producten te bekomen in zo hoog mogelijke

concentraties. Hoofdstuk 2 bespreekt de beschikbare literatuur met betrekking tot gemengde

fermentaties waarbij organische substraten worden omgezet in carbonzuren, in batch, fed-

batch en continue reactoren. De literatuur beschrijft dat het gebruik van geconcentreerde

substraten, rijk aan koolwaterstoffen, lange verblijftijden en neutrale pH de productie van

organische zuren stimuleert. De totale carbonzuurconcentratie bedraagt maximaal 20g/l bij een

productiesnelheid van 1.5g/L.h. Uitzondering hierop zijn de resultaten van Fu&Holtzapple

(2011) die spreken over zuurconcentraties tot 40 g/L wanneer het inhiberende effect van de

carbonzuren op de organismen geëlimineerd wordt. Ze ontwikkelden een tegenstroom

203

reactorconcept om de contacttijd van de gevormde zuren met de organismen te beperken.

Product inhibitie komt immers meestal voor bij erg hoge substraatconcentratie en korte

verblijftijden, bij te hoge organische belasting of bij lage pH. Wanneer gewerkt wordt bij

neutrale pH, verhoogt niet alleen de totale zuurconcentratie, maar ook de azijnzuur- en

propionzuurfracties, alhoewel hoge propionzuurconcentratie slechts af en toe gemeld worden.

Een hoge boterzuurfractie (>70%) wordt meestal ofwel bereikt bij pH<6 en iets langere

incubatie- of verblijftijden en een hoge organische belasting, ofwel onder CO2 atmosfeer ofwel

bij hoge substraatconcentratie. Verhogen van de verblijftijden levert vooral een verschuiving op

van korte naar langere vetzuurketens, zoals valeraat of n-caproaat.

In hoofstuk 3 van deze thesis worden twee belangrijke operationele parameters verder

onderzocht en beschreven: samenstelling van de headspace en substraattype. Het doel was om

na te gaan in welke mate een verhoging van de partiële druk in de headspace (2bar) bij een

batchfermentatie het proces, het product type en de productiviteit beïnvloedt. Drie reactoren

werden op 2 bar druk gebracht respectievelijk met waterstof, stikstof en een mengsel van

gelijke hoeveelheden waterstof en stikstof. Een controlereactor werd gespoeld met stikstof bij

het begin van het experiment en nadien ongemoeid gelaten. De hoogste

carbonzuurconcentratie werd gemeten bij de reactor met de waterstof en stikstof in de

headspace en bedroeg 17 g/L. In deze reactor bedroeg de azijnzuurfractie 85%, tengevolge

van homoacetogenese. Het azijnzuur werd dus in dit geval gevormd werd door omzetting van

waterstof en koolstofdioxide in plaats van het de organische componenten, aanwezig in het

substraat, wat niet gewenst was in het licht van deze studie. In de reactor onder CO2-

atmosfeer, was de totale carbonzuurconcentratie slechts 3.6 g/L; hoewel de butyraatfractie

70% bedroeg aan het einde van de eerste week. Bij deze was ook één van de belangrijke

doelen van deze studie –het bereiken van carbonzuurfracties rond 70% bij gemengde

204

fermentatie met afvalstromen- bereikt. De carbonzuurproductie in de reactor gevuld met

waterstof was 1.5 g/L hoger dan in de controlereactor, maar met gelijke fracties acetaat en n-

butyraat. Om het effect van verschillende substraten na te gaan werden drie verschillende

afvalstromen geselecteerd, volgens hun concentratie aan koolwaterstoffen, eiwitten en vetten.

Afval met een hoog koolwaterstofgehalte leidde tot hogere carbonzuurconcentraties in

vergelijking met proteïne- en vetrijke afvalstromen. Bij proteïne-rijke stromen werd vooral

azijnzuur gevormd, onafhankelijk van de condities en samenstelling van de headspace. Volgens

literatuur zou in dit geval vooral propionaat en i-valeraat gevormd worden bij neutrale pH. In

ons geval werd bij pH 5 gewerkt, wat vooral in azijnzuurproductie resulteerde. Vetrijke

afvalstromen leverden carbonzuurconcentraties lager dan 0.5 g/L op en werden dan ook verder

buiten beschouwing gelaten in het onderzoek.

Hoofdstuk 4 beschrijft het onderzoek naar de invloed van de samenstelling van de

headspace op de batch fermentatie bij stijgende concentratie van het substraat rijk aan

koolwaterstoffen. De substraatconcentraties bedroegen 8, 13.5 en 23 g/L. Het doel was te

bepalen welke operationele condities nodig zijn in een continu proces om een hoge, stabiele,

concentratie van één bepaald carbonzuur te bereiken. Uit de experimenten bleek dat dit het

geval was bij de hoogste substraatconcentratie, zonder toevoeging van gassen in de

headspace. Alhoewel de hoogste carbonzuurconcentratie (11 g/L) bereikt werd bij toevoeging

van waterstof in de headspace werd enkel een mengsel van verschillende organische zuren

gevormd, in plaats van één bepaald zuur. Zonder manipulatie van de headspace bedroeg de

zuurconcentratie 9 g/L met 61% n-butyraat. Wanneer een continue reactor gedurende een

maand werd bedreven zonder toevoeging van gassen in de headspace, gevoed met 23 g/L

substraat en een verblijftijd van 2 dagen, kon een concentratie van 7 g/L en 71% n-butyraat

bereikt worden in een stabiele procesvoering.

205

Bij hogere carbonzuurconcentraties kunnen de biologische processen bij gemengde

culturen gelimiteerd of geïnhibeerd worden. Hoofdstuk 5 beschrijft de proces parameters bij

zgn. ‘in-situ recovery’ van de organische zuren uit het fermentatiemedium met als doel het

verhogen van de productiviteit in de reactor. Een nieuwe technologie, ‘direct contact’

elektrodialyse met gebruik van bipolaire membranen werd geïntegreerd in de continue

fermentor. Door deze ingreep kon de toevoeging van base in de reactor volledig worden

geëlimineerd: de pH werd nu volledig gecontroleerd door productie van base aan het bipolaire

membraan. De eerste experimenten toonden dat de azijnzuur- en propionzuurproductie

selectief verhoogden ten opzichte van andere carbonzuren. De totale productie van carbonzuren

verhoogde slechts met 50%, waarschijnlijk omdat de carbonzuurconcentraties in de fermentor

en bijgevolg het inhiberend effect op de organismen lager waren dan verwacht. De

zuurconcentratie in het productcompartiment van de elektrodialyse was drie keer hoger dan

deze van de reactor. Deze resultaten waren de eerste gegevens in een onderzoek naar

koppeling van fermentatie en scheiding in fermentor met gemengde culturen. Verder onderzoek

en optimalisatie zijn hier duidelijk aan de orde.

In hoofdstuk 6 wordt de productie van organische zuren uit afvalstromen samengevat.

De belangrijkste bevinding uit dit onderzoek is dat de gemengde fermentatie gestuurd kan

worden naar productie van één enkel vetzuur, n-butyraat, bij toevoeging van koolstofdioxide in

de headspace óf bij toepassing van hoge substraatconcentratie. In deze gevallen wordt een n-

butyraat fractie bekomen van 70%. De totale zuurconcentratie blijft echter beperkt tot 20 g/L.

Bij de huidige industriële toepassingen liggen de concentraties zo’n 5 keer hoger. Via

reinculturen kan bijvoorbeeld 100 g/L azijnzuur geproduceerd worden met 95% zuiverheid.

Vermits de producten bij gemengde fermentatie steeds verdund aanwezig zijn is een

uitgebreide –en bijgevolg dure- nabehandeling nodig. Ook de opbrengst zou nog verhoogd

206

moeten worden om deze processen industrieel relevant te maken. Gemengde fermentatie kan

ontwikkelen tot een veel belovende technologie indien het mengsel van carbonzuren

rechtstreeks gebruikt kan worden als precursor voor de productie van waardevolle chemicaliën

zoals alcoholen, polymeren of middellange keten vetzuren.

207

208

209

Acknowledgments

Completing a Ph.D. is certainly a long and though journey. I would like to thank everyone who supported

me during my journey. I know I could never make it this far without you.

First of all I would like to thank my supervisors. Heleen, thank you very much for making it possible for me

to start, progress and finalize this study successfully. You contributed great in designing the experiments, evaluating

and publishing the results. I learned a lot from you! I would like to thank Cees for all his broad-minded but observant

approaches about my study, not least for being open to collaborate work with VITO for my Ph.D. Kirsten, I highly

value your knowledge and experience in the field. I appreciate a lot that you always make me realize the significant

points that I tend to overlook.

Special thanks to Bert and Ludo who supervised me the first two years of my doctorate. Bert, I am amazed

with your enthusiasm about our scientific conversations and guidance to my personal development. I always felt very

motivated after our meetings in Wageningen. Thank you Ludo, for introducing me to the `Bio’ group in VITO seven

years ago and keep supporting me during my master and doctorate.

Karolien, I highly appreciate your great efforts to overcome all the difficulties from simple lab problems to

complicated administrative issues. I really enjoyed your motivating attitude in every aspect during our short

discussions.

Many thanks to Hans, Rob, Jef, Helmut, Kristof, and Filip for your great help to my experiments. You all

constructed a fantastic lab for me, kept improving my reactors and bring practical solutions to each problem. Your

input on my work prompted to several publications and eventually this book. Christof, Guy, Silvia, Nicole, Queenie,

Lieve and Leen, thank you very much for your tremendous support to build suitable lab for my experiments. Miranda

and Carine, special thanks to you for valuable input to my analysis. As well as work, I enjoyed very much our long

discussions about movies, bands and England. Truly, I learned a lot from each one of you who helped me make this

study a success.

Yang, I am grateful for your valuable input to my work and also getting to know your warm personality.

Your knowledge and experience in the field helped me a lot to build the experimental setup and obtain promising

results to construct the last paper. I enjoyed working with you and having deep conversations.

210

Bachelor and master students, Bart, Cherity and Fang, thank you very much for choosing to work with me

and improving this study with your valuable input.

I would like to thank to my ‘bio’colleagues in VITO and Wageningen University. Bio-team in VITO; Linsey,

Wouter, former members Deepak, Melida and Suman thanks for valuable comments in the meetings and nice talks

during coffee breaks. Linsey, thanks for the special support towards the end where I needed it a lot. Bioenergy group

in Wageningen; Annemiek, Tim, Bruno, Mieke, Ralph, David, Tom, Nienke, Ruud, Adriaan, Marjolein, Alexandra,

thank you all for fruitful discussions.

My dear paranymphs, Annemiek and Wim, two precious people I met during my study. Annemiek, thank

you so much for opening your house every time (!) to this poor girl coming from Belgium. Together with Dirk, Tilia

and Wietse, I loved the times we spent in your place. Wim, I always liked your open and calming atmosphere at

work. Last years I found out you are full of interesting ideas, experiences, music, movies, books. Thank you both for

laughter and many pleasant moments you bring in my life and supporting me during my defense.

All my friends from work, Katrien and Sofie (girls, I miss our nice chats), John and Ben (I would kill to work

in another i-sup with you guys), Celine (thanks for nice talks), Eylem and Milica (phd victims, same time, same

place), Bert and Roel (thanks for the laughter in PRD corridors), Sofie Van Ermen, Winnie, Yamini, Nabin, Jan

Boeckx, Etienne, Inge, Sandra and Bas thank you all for creating very nice working environment. And Liesbeth

Kesaulya thank you very much for the unlimited help from Wageningen anytime I need.

My dearest friends, Meltem Abla, you encouraged me to start this Ph.D. Thank you very much for making

me feel at home in Belgium. Dearest Adoline, I am very happy to know you. We had fantastic friendship growing in

dormitories in Mol. You are still a true friend to me. Thanks Nele, I appreciate that going far lands couldn’t put

distance in our warm friendship. Artul, thanks for never-ending support for almost everything in my life. Uncle Eric,

thanks for the fatherly support. Thank you Alba, Tu çe, Kelly, Özlem, K vanç, Maciek, Taina, Jacob, Ludo, Sumru and

Birten for being with me here or on the other side of the phone whenever I needed.

My dear brother, thank you for the laughter and love that you give to my life.

My wonderful parents, sevgili anneci im ve babac m, beni sevgiyle büyüten ve her zaman destekleyen

ailem. Bu kitabi size adamak istiyorum. Her ey için size sonsuz te ekkürler. Sizi çok seviyorum.

And, thank you my soul mate, Joost. I love you.

211

About the author

Do a Arslan was born in Bayramic, Turkey in 1983. She

obtained her B.Sc. from Environmental Engineering

department, Marmara University in 2006. In the same

year, she started in the Master program of Bioengineering

Department, Marmara University. During her master

study, she was accepted to Erasmus scholarship program

to complete the second year of her thesis in the Flemish

Institute for Technological Research (VITO). Her thesis aimed at bioconversion of organic

substrates via pure and mixed culture anaerobic fermentation. After her graduation, she

continued as a researcher in SewagePlus project where she worked on optimization of bio-

hydrogen production from wastewater. She got an opportunity to start her Ph.D. study in

collaboration with Environmental Technology Department in Wageningen University and VITO.

the Chairman of the SENSE board the SENSE Director of Education

Prof. dr. Huub Rijnaarts Dr. Ad van Dommelen

The SENSE Research School has been accredited by the Royal Netherlands Academy of Arts and Sciences (KNAW)

Netherlands Research School for the

Socio Economic and Natural Sciences of the Environment

D I P L O M AFor specialised PhD training

The Netherlands Research School for theSocio Economic and Natural Sciences of the Environment

(SENSE) declares that

Doga Arslan

born on 11 April 1983 in Bayramic, Turkey

has successfully fulfilled all requirements of theEducational Programme of SENSE.

Wageningen, 21 October 2014

SENSE Coordinator PhD Education

Dr. ing. Monique Gulickx

The SENSE Research School declares thatMs Doga Arslan has successfully fulfilled allrequirements of the Educational PhD Programme of SENSE with a work load of 34.5 EC,

including the following activities:

SENSE PhD Courses

o Environmental Research in Context (2010)o Techniques for Writing and Presenting a Scientific Paper (2011)o Research Context Activity: Co organising the Bio based Economy session in ‘i SUP 2012

Conference: Focus on International Sustainability Issues’, 6 10 May, Bruges, Belgium(2012)

Other PhD and Advanced MSc Courses

o Advanced Course Downstream Processing (2011)o Soil chemical applicationso Techniques for Writing and Presenting a Scientific Papero Competence Assessmento Teaching and supervising Thesis studentso Reviewing a Scientific Paper

Management and Didactic Skills Training

o Supervision MSc thesis entitled ‘Production and Separation of Carboxylic Acids fromMixed Culture Fermentation’ (2012)

Oral Presentations

o Energy Recovery from Sewage Enriched Organic Kitchen Waste. IWA Symposium, 21September 2011, Leuven, Belgium

o Steering Mixed Culture by Headspace Manipulation. i SUP 2012: Focus on internationalsustainability issues, 6 10 May 2012, Brugge, Belgium

o Valorisation of Organic Waste Streams Through Fermentation Process. Orbit Conference,12 15 June 2012, Rennes, France

o Steering Mixed Culture Process by Headspace and Substrate Composition. SENSE Sciencemarket – Towards Biobased Economy, 25 October 2012, The Hague, The Netherlands

o Business Cases for Biobased Delta Region for the South West Netherlands and Flanders.Workshop – Industrial crops for feedstock towards a business case in Delta Region, 26March 2013, Terneuzen, The Netherlands

214

Cover design by Nejdet Arslan.

This thesis was printed on wood-free paper to respect our environment by CPI-

Wöhrmann (www.wohrmann.nl) and De Weijer Design (www.deweijerdesign.nl).

Selective Short Chain

Carboxylates Production

by Mixed Culture

Fermentation

Doğa Arslan

Doğ

a Arslan

Selective Sh

ort C

hain

Carb

oxy

lates Pro

du

ction

by

Mix

ed C

ultu

re Ferm

entatio

n

VK ST thesis omslag def1.indd 1 25-09-2014 23:48:44

Documents

Selective Short Chain Carboxylates Production by Mixed