preparation of volatile fatty acid (vfa) calcium salts by anaerobic digestion of glucose
TRANSCRIPT
This article has been accepted for publication and undergone full peer review but has not been through the copyediting,
typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of
Record. Please cite this article as doi: 10.1002/bab.1301.
This article is protected by copyright. All rights reserved. 1
Preparation of volatile fatty acid (VFA) calcium salts by anaerobic digestion of glucose
Xiaofen Li*, Janis E. Swan, Giridhar R. Nair and Alan G. Langdon
School of Engineering, The University of Waikato, Private Bag 3105, Hamilton, New Zealand 3240
Abstract
Many potentially useful intermediates such as hydrogen and volatile fatty acids (VFAs) are formed during the
complex anaerobic digestion processes that produce methane from biomass. This study recovers VFAs from an
anaerobic digester by a combination of gas stripping and absorption with calcium carbonate slurry. Glucose was
used as the model substrate because it is readily available, inexpensive and easily digested. Sludge from a meat
works anaerobic digester produced methane and carbon dioxide (and sometimes a small amount of hydrogen)
when batch-fed with glucose. Conditioning the neutral anaerobic sludge to an acidic pH (below 4.8) was
achieved using repeated 1-g L-1
doses of glucose. After conditioning, mainly VFAs and hydrogen were
produced. The intermediate VFAs could be stripped using headspace gas. In subsequent fed-batch
digestion/stripping cycles, the pH decreased when glucose was added and then rose when the VFA was gas
stripped. The predominant acids formed at low pH values were lactic, butyric and acetic acids. Lactic acid was
converted to VFAs during stripping. The VFA calcium salts recovered were 80% butyrate, and 20% acetate with
minor quantities of propionate and valerate.
Keywords:
Anaerobic digestion, bioreactors, volatile fatty acid, stripping, absorption, recovery
1. Introduction
The search for sustainable and environmentally-compatible alternatives to fossil fuels for producing organic
commodity chemicals has led to renewed interest in converting biomass using microbial processes. Anaerobic
digestion is a multi-step series of hydrolytic, acidogenic, acetogenic and methanogenic biological reactions that
converts complex polymeric substances to simpler molecules [1]. This process is routinely used to process
waste biomass from food processing, agriculture and municipal wastewater treatment into stabilized sludge, a
nutrient-rich effluent, methane and carbon dioxide. Hydrogen and volatile fatty acids (VFAs) are formed as
intermediates. These are converted to methane in the final methanogenic step. Hydrogen and VFAs have
higher values as industrial chemicals than methane [2], which provides an incentive to find economic methods
for their recovery.
Although there have been many reports on controlling anaerobic digestion to produce hydrogen [3-9], there are
fewer reports on recovering VFAs. A process for removing VFAs could have the added benefit of preventing
the “souring” problem when trying to maintain continuous anaerobic fermentations [10]. The purpose of this
research was to condition an anaerobic sludge to an acidic pH to inhibit methanogenesis and ensure that VFAs
were predominantly in their volatile undissociated forms, and then to apply a recently reported procedure [2, 10]
to strip and recover VFAs as VFA calcium salts.
This article is protected by copyright. All rights reserved. 2
1.1 Volatile fatty acid stripping and recovery
Continuous operation of an anaerobic digester to produce acetic acid and other VFAs involves inhibiting
methanogenesis and simultaneously removing and recovering the VFAs. Literature indicates that liquid/liquid
extraction and reactive extraction using tertiary amines [11] can be used to recover acetic acid from dilute
solutions. These methods are not compatible with maintaining an active fermentation. Recently, a more
suitable gas stripping method, based on volatility of the VFAs, has been developed [2, 10]. While gas stripping
is routinely used to remove volatile contaminant components from liquid process streams, there is little literature
on using this process to remove and/or recover VFAs from dilute solutions of processing waste streams or
anaerobic digesters. This may due to VFAs having much higher Henry’s law constant values compared with
strippable components such as hydrogen sulphide, ammonia, carbon dioxide, etc [12]. Conventional stripping is
successful when the volatile solutes have Henry’s Law constants less than 10 mol L-1
atm-1
[13]. Values for
VFAs are more than two orders of magnitude higher than this and the concentrations of undissociated volatile
forms of VFAs are pH dependent (the higher the pH, the lower the concentration).
Gas stripping used in industrial processes such as wastewater treatment needs to be rapid and as near complete
as possible. This is favored if Henry’s law constants are small. However, because anaerobic digestion is a slow
process, even slow gas stripping of VFAs, can be effective. An optimized continuous process only requires that
stripping remove the acids as fast as they are produced. The stripping should maintain the pH at a level that
avoids excessive VFA dissociation and inhibition of the digestion.
1.2 Inhibiting methanogenesis
To obtain high VFA concentration, methanogenesis, which converts VFAs to methane, must be inhibited. This
can be achieved by acidification [14, 15, 16], exposure to air [6, 17], thermal treatment [4, 6, 7, 18] or chemical
inhibition with, for example, halogenated hydrocarbons [8]. Lowering the pH will also satisfy the requirement
that the VFAs are in their volatile, undissociated forms and therefore strippable. In conventional anaerobic
digestion, the aim is to maintain a neutral pH. Methane production falls markedly when pH is below 6.3 or
above 7.8 [15, 19]. Lowering bioreactor pH directly by adding acid or indirectly by overloading with easily-
digested substrate are possible ways of inhibiting activity of methanogens and favoring the acidogens and
acetogens that produce hydrogen and VFAs. Panamas et al. reported that consumption of large amounts of
glucose lowered pH and produced high VFAs concentrations, however, glucose concentrations of 10 g L-1
and
greater are toxic for the anaerobic microbes [20]. When methanogenesis has been suppressed, acid
accumulation during batch fermentation will eventually lower pH to levels that inhibit further fermentation and
the reactor “sours”. Hydrogen (and lack of methane) in the gaseous products can be correlated with
accumulation of VFAs in the bioreactor fluids [9], allowing gas composition to indicate VFAs accumulation, at
least while the reactor is producing gas.
2. Materials and methods
2.1 Fed-batch bioreactor/stripping/recovery system
The recently reported technology for stripping VFAs from aqueous systems [10, 12] was applied to a bench-
scale reactor (Figure 1) made of a 5-L Pyrex bottle bioreactor containing 1.7 L sludge. The 1-L conical flask
recovery system contained 200 mL 2M CaCO3 slurry. The bioreactor contents, containing dilute VFAs, were
stripped using water-saturated nitrogen gas entering via a 7-mm (i.d.) open tube. The gas containing the VFAs
stripped from the bioreactor was then passed through a 19-mm (i.d.) open tube immersed in the CaCO3 recovery
This article is protected by copyright. All rights reserved. 3
system. The peristaltic pump was fitted with 166 cm silicon tube (6.4 mm i.d., Masterflex 6404-17) and set to
deliver 1.06 L min-1
(stripping rate of 0.62 L-1
L-1
min-1
). The system was maintained at 35oC in a modified
incubator.
The VFAs of interest (acetic, propionic, butyric, and valeric) all have pKa values close to 4.8. All have similar
Henry’s law constants [12] (1.1 -5.7) x 103. Henry’s Law describes the partial pressure of a volatile component
(e.g. VFA) in a gas phase (PVFA) in equilibrium with a dilute solution of that component at a concentration,
CVFA(aq):
VFA(aq) H VFAC = K P (1)
where, KH is the Henry’s Law constant (mol L-1
atm-1
).
The maximum transfer rate (TR) of a specific VFA that can achieved at a stripping gas volumetric flow rate, Q,
is:
VFA(aq)
VFA(g)
H
CTR= Q C = Q x
K RT
(2)
Existence of this relationship between TR and Q will demonstrate VFA equilibration with the gas stream during
stripping and will also indicate that VFA is quantitatively removed from the gas stream in the recovery system.
Once the VFAs in the gas stream have been transferred into the CaCO3 recovery system, the gas can absorb
more acid when it is recirculated through the bioreactor. The stripping gas remains water saturated and at
constant temperature.
Previous research [10] showed that Equation 2 was valid for nitrogen flow rates up to 2.0 L min-1
when
stripping acetic acid from dilute solution and then recovering the acetic acid in CaCO3 slurry, indicating that
liquid-vapor equilibrium was established during stripping and quantitative removal of the stripped acid was
achieved during the recovery step.
2.2 Anaerobic sludge
Anaerobic sludge samples were obtained from the ambient temperature plug flow reactor at a local meat
processing company. The total solids (TS) of the sludge was 63 g L-1
. Of this, 59% (37 g L-1
) was volatile
solids (VS). As-collected sludge alkalinity, 5 day carbonaceous biochemical oxygen demand (BOD5), chemical
oxygen demand (COD), ammonia (NH4) nitrogen as N, total suspended solids (TSS ), total phosphorus as P
were 2008 mg CaCO3 L-1
; 758 mg L-1
, 944 mg L-1
, 457 mg L-1
, 390 mg L-1
and 67 mg L-1
respectively. The
high ammonium and phosphorous content were considered to be sufficient to maintain microbial activity
without further additions. Anecdotal evidence indicated this sludge often produced hydrogen and hence might
be readily conditioned to produce VFAs. When operated at 20oC and batch-fed with a single 1 g L
-1 dose of
glucose, the sludge produced a biogas containing 2.7% hydrogen. The sludge stored at 20oC and fed
periodically with glucose but no other nutrients retained good activity during the trials.
Insert Fig. 1 here
2.3 Fed-batch conditioning experiment (Cycle 1)
Anaerobic sludge (1.7 L at 6.3% volatile solids) was transferred to the 5-L bioreactor and conditioned to acid
pH by adding 1-g L-1
doses of glucose at the loading rate of 0.05 g glucose (g VS)-1
. Each successive addition
This article is protected by copyright. All rights reserved. 4
was made after gas production and pH lowering from the preceding dose had ceased (usually after 1 to 3 days).
After each glucose dose, the system was flushed with oxygen-free nitrogen to remove any introduced oxygen.
Ambient temperature (20oC) was maintained during the conditioning process because the sludge had been
collected and stored at ambient temperature.
Gas composition during digestion was monitored regularly by in-line gas chromatography (GC). The gas
produced was collected over water (pH 7.32 ± 0.05) in inverted 250, 1000 or 2000-mL measuring cylinders.
The pH in the measuring cylinder water did not change, indicating negligible adsorption of carbon dioxide. The
batch feeding was continued until the reactor pH was lowered to pH 4.65
2.4 Fed-batch digestion /stripping/recovery cycles
After conditioning of the bioreactor sludge at 20oC (with 15 1-g L
-1 additions of glucose), the reactor contents
were gas stripped with head space gas (mainly nitrogen) at 35oC until pH increased to approximately pH 5.0
(Cycle 1). Three further successive fed-batch digestion /stripping/recovery cycles were performed, all at 35oC
(Cycles 2, 3 and 4). Stripping was started after the pH in the reactor had dropped below pH 4.8. The glucose
digestion step of Cycle 1 was the 15 successive 1-g L-1
doses at a loading rate of 0.046 g glucose (g VS)-1
needed to condition the sludge to acid pH. In Cycle 2, six successive glucose additions were batch-fed as 1-g L-
1 doses. In Cycle 3, glucose was added as nine 1-g L
-1 doses at a loading rate of 0.046 g glucose (g VS)
-1,
followed by one 5-g L-1
dose at loading rate of 0.23 g glucose (g VS)-1
and then one 10-g L-1
glucose dose at
loading rate of 0.46 g glucose (g VS)-1
. In Cycle 4, in which there was no stripping step, one 10-g L-1
dose was
added at a loading rate of 0.46 g glucose (g VS)-1
. After the initial conditioning at 20oC, the reactor was run at
35oC. Total glucose added during each cycle was 15 g L
-1, 6 g L
-1, 24 g L
-1 and 10 g L
-1 for cycles 1, 2, 3 and 4
respectively. Individual additions during each cycle were kept at 10 g L-1
or less to avoid the possibility of
glucose toxicity. Generally, digestion of the successive additions was complete within a few days. In cycle one,
the 15 glucose additions were added over a period of 79 days allowing approximately 5 days for digestion after
each addition. For cycles 2 and 3, the average digestion times were 1.7 and 4.4 days. Stripping was continued
for 9, 10 and 67 days for cycles 1, 2 and 3 respectively. For cycle 2 the stripping time was kept approximately
the same as the total digestion time to test the possibility of operating the reaction under conditions of
continuous simultaneous digestion and stripping.
2.5 Gas analysis
Head space gas samples were analyzed using a Perkin-ElmerTM GC with a thermal conductivity detector
(TCD) and a column sequence of HaySep Q (80/100), molecular sieve 13X (pore size 13 Å with sodium as the
primary cation) and HaySep D (100/120) columns. The molecular sieve separated hydrogen, oxygen, nitrogen,
methane, and carbon monoxide. Because carbon dioxide is irreversibly adsorbed by the zeolite, the GC-TCD
program was set to reverse gas flow before carbon dioxide had passed through the HaySep Q columns. The
following program was used: oven temperature 50oC, detector temperature 250
oC, inject temperature 120
oC,
carrier gas argon (90 psi.), flow rate 20 mL min-1
, program time 10 min with reversed flow at 2.65 min. Gas
flow was reversed using pneumatic switches operated with dry air at 65 psi.
The GC system was calibrated with a standard gas mixture (Matheson Tri. Gas, Grace Davison Discovery
Science, USA) at regular intervals during the trials. Peak areas had coefficients of variation (CVs) of 2.5%,
2.7%, 2.3%, and 5.0% for H2, CH4, N2 and CO2 respectively. Head space gas samples from the bioreactors were
injected at room temperature to the columns via a 1-mL sample loop operated pneumatically by a dry air supply.
This article is protected by copyright. All rights reserved. 5
2.6 Analysis of digester contents
Digester contents (glucose, lactic acid, acetic acid, propionic acid, iso-butyric acid, butyric acid and valeric acid)
were analyzed by HPLC [21]. Samples of reactor fluid (1 mL) taken at regular intervals during the runs were
stored at -20oC until analyzed. When required, samples were centrifuged at ambient temperature and 12,000
rpm for 10 min during which the samples thawed and produced a clear supernatant. A 1-mL sample of the
supernatant was withdrawn using a 1-mL syringe fitted with a 22 G x 1 ½ (0.71 x 38 mm) sterile single use
needle and filtered through a 0.20-µm minisart®
-plus syringe filter. Forty µL of 7% sulfuric acid was added to
the filtered supernatant to convert the weak acids to their protonated forms and help maintain functionality of
the column.
The samples were shaken in a vortex (MS1 Minishaker IKA® Works, Malaysia) mixer for about 10 seconds and
then analyzed by HPLC (Waters 515 Pump, Waters Column Heater Module, Waters 996 Photodiode Array
Detector (PDA), Waters 2414 Refractive Index (RI) Detector and a Rheodyne 7725i manual sample injector
(20-µL sample loop)). The system was controlled using Waters EmpowerTM
2 Chromatography Software. The
analyses were performed isocratically at 0.6 mL min-1
and 60oC with a 300 x 7.8-mm i.d. cation exchange
column (Aminex HPX-87H) equipped with a cation H+ microguard cartridge (Bio-Rad Laboratories). Mobile
phase was 5 mmol L-1
sulfuric acid prepared with milliQ water, filtered through a 0.2-µm minisart®-plus
syringe filter and degassed under vacuum in an ultrasonic bath for 20 min. Peak areas were highly reproducible
with CVs of 0.54%, 0.23%, 1.26%, 1.39%, 3.45%, 3.17% and 1.05% for the acetic acid, glucose, propionic acid,
iso-butyric acid, butyric acid, valeric acid and L-lactic acid standards respectively.
3. Results and discussion
3.1 Conditioning the sludge with glucose substrate
Gas production at 20oC (collected over water) for 1 g L
-1 glucose doses leveled off to approximately 190 mL (g
glucose)-1
. After each successive glucose dose had been digested, pH decreased. Initially the pH drops were
about 0.1 pH units but increased to about 0.3-0.4 pH units below pH 6.0 and then by 0.1 pH units at pH below
5.0. After 15 doses, pH had decreased from 7.5 to 4.65. The sludge that had been producing predominantly
methane and carbon dioxide was conditioned to one that produced hydrogen and carbon dioxide and no
methane.
3.2 Fed-batch digestion and gas stripping cycles
Compositions of the reactor fluid were determined before digestion (BD), after digestion (AD) and after
stripping (AS) for three complete fed-batch/stripping /recovery cycles (Cycles 1, 2 and 3). The BD data for
Cycle 1 is for the unconditioned sludge. The AS and BD data for the successive batch fed/stripping cycles are
similar. The results (Table 1) show that glucose is converted to predominantly butyric acid, acetic acid and
lactic acid, which is consistent with the pathways for anaerobic digestion of glucose. At neutral pH, glucose is
converted predominantly to methane and carbon dioxide. As pH decreases, methane production decreases,
hydrogen production increases and VFAs (mainly acetic acid and butyric acid with smaller amounts of lactic
acid) accumulate. These transformations, which depend on the conditions, can be described by the overall
stoichiometries [23, 24, 25].
6 12 6 3C H O 3CH COOH
6 12 6 3 2 2 2 2C H O CH CH CH COOH + 2CO + 2H
This article is protected by copyright. All rights reserved. 6
6 12 6 3C H O 2CH CHOHCOOH
The predominance of butyric acid rather than acetic acid in the supernatant is consistent with the presence of
significant volumes of hydrogen in the head space gas. The stoichiometries and Gibbs free energy changes for
forming acetic acid from the higher carbon VFAs [26, 27] mean there will little tendency for any butyric acid
formed to be converted to acetic acid when hydrogen is present.
- - +
3 3 2 2 3 2CH CH CH COO + 2H O 2CH COO +2H +2H , ΔGo' = 48.3 kJ mol
-1
- - + -
3 2 2 3 3 2CH CH COO + 3H O CH COO +H +HCO +3H , ΔGo' = 76.1 kJ mol
-1
- -
3 2 2 3 2 2CH CH COO + 2H O CH COO +3H +CO , ΔGo' = 71.7 kJ mol
-1
During stripping, pH always increased and lactic and butyric acid concentrations always decreased but there was
no consistent trend for acetic acid (Table 2). No VFA was completely stripped. The disappearance of the non-
volatile lactic acid during acidic anaerobic stripping indicates its conversion to VFAs, as reported previously
[22]. Acetic and propionic acid contents in the supernatant did not appear to be significantly affected by
stripping. The acid stripped at low pH values appeared to be replaced by acids formed in their unstrippable
anionic forms at the higher pH values towards the end of stripping. It is known that acetate-producing microbes
become active as pH rises, allowing for head space H2 and CO2 to be converted to acetate. Alternatively, if
lactic acid is converted to acetate during the higher pH stages of stripping, this acetate would also remain largely
unstripped. The decrease in HAc after stripping during cycle 3 but its increase during cycles 1 and 2 can be
explained by the fact that the pH during the stripping of cycle 3 remains below pH 4.95. The lower pH would
allow HAc to be more exhaustively stripped while inhibiting the methanogenic processes that would allow un-
strippable acetate to be formed from higher C acids and CO2 and H2.
Data for pH changes during the digestion and stripping steps are summarized in Table 2. Changes in pH should
relate to changes in concentrations of the acid and salt species present. In the case of a single weak acid and its
salt, the relationship between acid and salt concentrations is the well-known Henderson equation [26]:
a 10
[A ]pH = pK + log
[HA]
Insert Table 1 here
While it is possible to devise individual equations for systems of mixed acids and their salts, because the pKa
values of the VFAs are all close to 4.8, an approximate estimate of pH can be made by summing the
concentrations of all the undissociated acids and their salts and using a pKa of 4.80 (the mean pKa value of the
two main VFAs present, butyric acid and acetic acid). The total VFA salt concentration in the system will
depend on the initial alkalinity of the sludge and will remain essentially constant. It was estimated from half the
total VFA content (salt and undissociated acid) of Cycle 2 after digestion where the pH of 4.78 (Table 1), was
very close to the mean pKa value so that salt and undissociated acid concentrations would be approximately the
same. The observed and calculated pH values for the various systems investigated (Table 2) were usually
within ± 0.1 pH units for all systems except those with significant concentrations of lactic acid (Cycle 1 AD,
Cycle 4 AD (1) and AD (2)). The presence of lactic acid (pKa 3.86) or other acid species such as polylactic acid
(pKa = 3.1) [29], which have pKa values lower than 4.80, will produce a lower pH for the given carboxyl
This article is protected by copyright. All rights reserved. 7
content. For example, the observed pH of 3.85 for Cycle 4 AD1 is close to the value calculated by considering
only the lactic acid equilibrium. The lower than calculated pH of 3.58, recorded after further standing for Cycle
4 AD2, indicates the presence of a species with a pKa even lower than that of lactic acid.
Insert Table 2 here
To determine carboxylic acid yields and carbon conversion efficiency during digestion and stripping, the
compositions of the supernatant and headspace gas were used to calculate moles carboxylic acid formed and
moles of carbon associated with each product species. Comparing the total moles of carbon identified as acid
and carbon dioxide products with moles of carbon added as glucose (Table 3) indicates that carbon recovery was
65%, 58% and 77% for cycles 1, 2 and 3 respectively. The glucose carbon not accounted for in these products
is likely to have been used in other digestion metabolites produced, including polymeric material detected (but
not quantified) as a significant peak at the column void volume and cellular biomass that would have been
removed from the samples during sample preparation. The fact that conversion efficiency to acids was greatest
for cycle 3 where pH was generally much lower than for the other cycles indicates that the formation of the
other metabolic products was inhibited by low pH. In general, yields for cycles 1, 2 and 3 were similar (cycles 1
and 3) or lower (cycle 1) than those reported in the literature for similar conditions [30] consistent with the
absence of contributions from breakdown of biomass during these experiments. However the anomalous
behaviour during cycle 4 is of interest. This system was not stripped because the lack of gas evolution after
adding glucose was initially interpreted as loss of bioactivity. After 23 months at room temperature, pH had
dropped to 3.58 but no gas had been produced. Total dry solids content decreased from 6.3 wt% to 4.3 wt% and
a carbon balanced indicated that supernatant lactic acid contained more carbon than could have been produced
from the added glucose (Table 3). It appears that some of this lactic acid had been produced from the dry solids
that had been consumed.
Insert Table 3 here
3.3 Recovery of crystalline product
In early work recovery of pure product was compromised by carry-over to the CaCO3 recovery system of foam
produced in the bioreactor. Foaming was eliminated by adding SE-15 anti-foaming agent (Sigma, USA). A
pure salt solution was then obtained as filtrate from the CaCO3 slurry. The XRD profile of the 3.46 g of white
crystalline product obtained from the evaporated filtrate collected from pH 4.8 to 4.95 for stripping cycle 3
indicated the presence of VFA salts but no calcium carbonate. However, the individual components could not
be identified further from XRD. The HPLC analysis of the white powder confirmed it consisted of calcium salts
of VFAs, predominantly butyric acid (Table 4).
Insert Table 4 here
4. Conclusions
Neutral methane-producing anaerobic sludge can be conditioned to a hydrogen-producing acidic sludge by
repeated dosing with glucose. At pH values below 4.8, VFAs are formed in their undissociated volatile form
This article is protected by copyright. All rights reserved. 8
and can be gas stripped from the reactor and recovered as calcium salts, predominantly calcium butyrate, by
passing the stripping gas through calcium carbonate slurry. Prolonged gas stripping can raise the pH of an
acidic digester to pH 5. The operation of an anaerobic digester under acidic conditions, using appropriate
substrate addition and gas stripping rates to remove VFAs as fast as they are formed, provides a possible new
method for continuous conversion and recovery of VFAs (particularly butyric acid) from suitable biomass
substrates. These might include biomass wastes of high sugar or starch content such as fruit and food
processing wastes.
5. Acknowledgment
The authors gratefully acknowledge the financial support of WaikatoLink Ltd and The University of Waikato,
Hamilton, New Zealand.
6. References
1. Evans, G, ed. (2001) Anaerobic digestion, in: Biowaste and Biological Waste Treatment, James &
James, London, 94-95
2. Langdon, A and L Li (2010) Anaerobic digestion: Harvesting high value by-products. NZ553786
3. Baghchehsaraee, B, Nakhla, G, Karamanev, D and Margaritis A (2010) Fermentative hydrogen
production by diverse microflora. Int J Hydrogen Energy 35:5021-5027
4. Fan, Y, Li C, Lay JJ, Hou H, and Zhang G (2004) Optimization of initial substrate and pH levels for
germination of sporing hydrogen-producing anaerobes in cow dung compost. Bioresour Technol
91:189-193
5. Lay, JJ (2001) Biohydrogen generation by mesophilic anaerobic fermentation of microcrystalline
cellulose. Biotechnol Bioeng 74: 280-287
6. Li, Z, Wang, H, Tang, ZX, Wang, XF and Bai JB (2008) Effects of pH value and substrate
concentration on hydrogen production from the anaerobic fermentation of glucose. Int J Hydrogen
Energy 33:7413-7418
7. Liang, DW, Shayegan SS, Ng, WJ and He, J (2010) Development and characteristics of rapidly formed
hydrogen-producing granules in an acidic anaerobic sequencing batch reactor (AnSBR). Biochem Eng
J 49:119-125
8. Wang, CC, Chang, CW, Chu, CP, Lee, DJ, Chang, BV, Liao, CS and Tay, JH (2003) Using filtrate of
waste biosolids to effectively produce bio-hydrogen by anaerobic fermentation. Water Res 37:2789-
2793
9. Zhu, H, Parker, W, Basnar, R, Proracki, A, Falletta, P, Béland, M and Seto, P (2009) Buffer
requirements for enhanced hydrogen production in acidogenic digestion of food wastes. Bioresour
Technol 100:5097-5102
10. Li, L, Langdon, AG, Nair, GR & Swan, JE (2010) Acetic acid recovery and pH control by gas stripping
and acid anaerobic sludge. Chemeca 2010, Adelaide, Australia 992-1001
11. Z Li, W Qin and Y Dai (2002) Liquid-liquid equilibria of acetic, propionic, butyric, and valeric acids
with trioctylamine as extractant. J Chem Eng Data 47:843-848
12. Li L (2013) Recovering Volatile Fatty Acids from an Acidic Anaerobic Digester, PhD Thesis,
University of Waikato, Hamilton, New Zealand
This article is protected by copyright. All rights reserved. 9
13. Nazaroff, WW and Alvarez-Cohen, L (2001) Environmental Engineering Science, John Wiley & Sons,
NY
14. Chang, JS, KS Lee, and PJ Lin (2002) Biohydrogen production with fixed-bed bioreactors. Int J
Hydrogen Energy 27:1167-1174
15. Chen, CC, CY Lin, and MC Lin (2002) Acid-base enrichment enhances anaerobic hydrogen production
process. Appl Microbiol Biotechnol 58:224-228
16. Fang, HHP, C Li, and T Zhang (2006) Acidophilic biohydrogen production from rice slurry. Int J
Hydrogen Energy 31:683-692
17. Morimoto, M, M Atsuko, AAY Atif, MA Ngan, A Fakhru'l-Razi, SE Iyuke, and AM Bakir (2004)
Biological production of hydrogen from glucose by natural anaerobic microflora. Int J Hydrogen
Energy 29: 709-713
18. Lay, JJ, KS Fan, J Chang, and CH Ku (2003) Influence of chemical nature of organic wastes on their
conversion to hydrogen by heat-shock digested sludge. Int J Hydrogen Energy 28:1361-1367
19. van Haandel, AC, and G Lettinga (1994) Anaerobic Sewage Treatment - A Practical Guide for Regions
with a Hot Climate. John Wiley & Sons, New York
20. Pantamas1, P, Chaiprasert P, and Tanticharoen M (2003) Anaerobic Digestion of glucose by Bacillus
licheniformis and Bacillus coagulans at low and high alkalinity, Asian J Energy Environ, 4:(1-2), 1-17
21. Niven, SJ, JD (2004) Beal, and PH Brooks, The simultaneous determination of short chain fatty acid,
monosaccharides and ethanol in fermented liquid pig diets. Animal Feed Sci Tech 117:339-345
22. Oude Elferink, SJWH, Krooneman, J, Gottschal, JC., Spoelstra, Faber, F and Driehuis, F (2001)
Anaerobic Conversion of Lactic Acid to Acetic Acid and 1,2-Propanediol by Lactobacillus buchneri.
Appl Environ Microbiol, 67: 125-132
23. Bilitewski, B, Härdtle, G and Marek, K (1997) Waste Management, Berlin, Springer
24. Dwidar, M, Park, JY, Mitchell, RJ and Sang, B (2012) The Future of Butyric Acid in Industry. The
Scientific World Journal
25. Campbell, NA and Reece, JB, Biology, San Francisco Pearson, Benjamin Cummings (2005).
26. McCarty, PL and Smith, DP (1986), Anaerobic wastewater treatment, Environ Sci Technol, 20: (12),
1200-1206
27. Ostrem, K, (2004), Greening Waste: Anaerobic Digestion for Treating the Organic Fraction of
Municipal Solid Wastes, MSc Thesis, Columbia University, New York.
28. Beynon, RJ and Easterby, JS (1996) Acids and Bases Buffer Solutions. Oxford Tokyo: Oxford
University Press Inc, New York. p 20
29. Siparsky, G, Voorhees, K and Miao, F (1998) Hydrolysis of polylactic Acid (PLA) and
polycaprolactone (PCL) in aqueous acetonitrile solutions: Autocatalysis. J Environ Polym Degr, 6: 31-
41
30. Zoetemeyer, RJ, Heuvel VD and Cohen A (1982) pH influence on acidogenic dissimilation of glucose
in an anaerobic digestor, Water Res, 16:303-311
This article is protected by copyright. All rights reserved. 10
Bioreactor Recovery system
Volatile fatty acids
Pump
N2
Vent
Sampling port
StirrerStirrer
Return line
Calcium carbonate slurrySludge
Fig 1. Schematic of the fed-batch/stripping/recovery system
This article is protected by copyright. All rights reserved. 11
Table 1 Products formed during the acid anaerobic digestion of glucose. The glucose added (GA) and
remaining (GR) and acid content before digestion (BD) and after digestion (AD) for repeated fed batch
digestion/stripping cycles are expressed as the mmols in the 1.7 L reactor fluid. For Cycles 2 and 3 the BD
values are those for the supernatant after stripping (AS)
Cy
cle
GA
(m
mo
l)
CH
4 (
L)
H2 (
L)
CO
2 (
L)
Sta
ge
GR
(m
mo
l)
LA
(m
mo
l )
HA
c (m
mo
l)
PA
(m
mo
l)
IBA
(m
mo
l)
BA
(m
mo
l )
IVA
(m
mo
l)
VA
(m
mo
l)
1 142 0.16 0.83 3.9 BD nd nd nd nd nd nd nd nd
AD 8.8 36.2 54.4 6.7 nd 35.4 nd nd
2 56.7 nd 0.69 2.1 AS/BD nd nd 72 6.4 1.9 22.2 nd nd
AD 3.7 5.7 71.7 4.9 2.1 46.9 0.6 nd
3 227 nd 3.0 8.9 AS/BD nd nd 84.9 3.9 1.1 13.9 nd nd
AD nd 11.9 62.8 3.3 1.0 184 0.8 4.0
4 94.4 nd nd nd
AS/BD nd nd 42.6 1.6 nd 66.6 nd 4.0
AD1* nd 251 52.9 1.8 nd 76.3 nd nd
AD2* nd 259 52.5 1.9 1.0 82.5 nd nd
nd - not detected, AD1* - sampled 19 months after glucose addition, AD2* - sampled 23 months after glucose addition, GA - Glucose added,
GR - glucose remaining, LA - lactic acid, HAc - acetic acid, PA - propionic acid, IBA - iso-butyric acid, BA - butyric acid, IVA - iso-valeric
acid, VA - valeric acid
This article is protected by copyright. All rights reserved. 12
Table 2 Comparison of observed and calculated pH values for pH before and after stripping
Cycle Stage Observed pH Calculated pH
1
AD 4.65 5.09
AS 5.12 5.01
2
AD 4.78 4.81
AS 5.01 5.01
3
AD 4.32 4.32
AS 4.95 4.92
4
AD1 3.85 3.86*
AD2 3.58 3.83*
AD1 Digestion Cycle 4 after 19 months, AD2 digestion Cycle 4 after 23 months, *Calculate using pKa of lactic acid (3.86)
This article is protected by copyright. All rights reserved. 13
Table 3 Carboxylic acid production and carbon conversion for successive digestion/stripping cycles Amounts
are expressed as either mmol (for carboxyl) or mmol C (for glucose, CO2 and acids) added to, stripped from or
present in the 1.7 L of reactor contents)
Cycle
GA
(mmol
C)
CO2
(mmol
C)a
Total carboxylb
(mmol)
Carboxyl strippedc
(mmol)
Total acid
(mmol C)
Total products
(mmol C)
AD AS
1 852 164 133 103 30 379 543
2 340 82 132 103 29 114 196
3 1362 351 268 112 156 696 1047
4 566 nd
382AD1
nm nm 809AD1
809AD1
397AD2
nm nm 861AD2
861AD2
nm – not measured, nd – not detected aCollected over water, bDetermined from the moles of all the carboxylic acids measured, cCalculated from change in COOH before and after stripping
AD – after digestion, AS – after stripping
This article is protected by copyright. All rights reserved. 14
Table 4 VFA composition of product recovered from a glucose-fed, acidic anaerobic digester
Volatile fatty acid HAc PA BA VA TVFA
Amount (mmol) 5.49 0.92 29.5 1.11 37.0
Composition (% (w/w) ) 11.4 2.3 82.9 3.5 100
HAc - acetic acid, PA - propionic acid, BA - butyric acid, VA - valeric acid, TVFA – Total volatile fatty acid