Biochemical Engineering Journal 93 (2015) 250259

Contents lists available at ScienceDirect

Biochemical Engineering Journal

journal homepage: www.elsevier.com/locate/bej

Regular Article

Comparison of several methods for the separation of

poly(3-hydroxybutyrate) from Cupriavidus necator H16 cultures
M. Lpez-Abelairas a, , M. Garca-Torreiro a , T. L-Chau a , J.M. Lema b , A. Steinbchel c,d
a
Department of Chemical Engineering, Institute of Technology, University of Santiago de Compostela, 15782 Santiago de Compostela, Spain
b
Department of Chemical Engineering, School of Engineering, University of Santiago de Compostela, 15782 Santiago de Compostela, Spain
c
Institut fr Molekulare Mikrobiologie und Biotechnologie, Westflische Wilhelms-Universitt Mnster, D-48149 Mnster, Germany
d
Department of Environmental Sciences, King Abdulaiziz University, Jeddah, Saudi Arabia

a r t i c l e i n f o a b s t r a c t

Article history: Biopolymers, such as polyhydroxyalkanoates (PHA), are an environmentally friendly alternative to plas-
Received 28 January 2014 tics derived from fossil fuels. However, producing PHA in a cost-effective way requires the development
Received in revised form 21 October 2014 of highly efcient separation and purication treatments. In this study, one acid treatment (sulphuric acid
Accepted 23 October 2014
combined with a subsequent bleaching step) and three alkaline treatments (sodium hypochlorite, sodium
Available online 1 November 2014
hydroxide and a combination of the latter with an halogenated solvent) were evaluated for recovering
poly(3-hydroxybutyrate) (PHB), a type of PHA, from Cupriavidus necator H16 cells with high biopolymer
Keywords:
content (65%). Purity, percent of recovery and the properties of the PHB obtained after each treatment,
Polyhydroxyalkanoates
Cupriavidus necator together with the costs and environmental impacts associated with each treatment, were determined
Purication and compared. The lowest recovery costs were obtained with the sodium hydroxide and sulphuric acid
Bioseparation treatments (1.02 and 1.11 D kg1 , respectively). Estimated CO2 emissions of these two treatments were
Cell disruption 18% of those based on the use of sodium hypochlorite. However, the highest purity (98%) and lowest poly-
Downstream processing mer degradation were achieved with the acid treatment. Consequently, the acid treatment was selected
as the most effective choice for PHA recovery.
2014 Elsevier B.V. All rights reserved.

1. Introduction
A potentially interesting way to decrease the environmental
impact of conventional petroleum-derived plastics is to replace
them with biodegradable polymers. In this context, polymers
of biological origin, such as polyhydroxyalkanoates (PHA) and
among them polyhydroxybutyrate (PHB), play an important role.
These polymers are accumulated inside bacterial cells as energy
and carbon storage. Cupriavidus necator, the best-studied PHB-
accumulating bacterium, stores PHB when there is an excess of
carbon with respect to some other essential nutrient in the medium,
such as nitrogen, phosphorous or oxygen [1,2]. To compete with
conventional plastics, PHA production costs must be minimised.
It has been estimated that more than 50% of the cost of PHB pro-
duction are associated with the recovery and purication of the
polymer [3]. Due to the large impact of the recovery step on pro-
duction costs, efcient methods for separation and purication
Corresponding author. Tel.: +34 881 816014; fax: +34 881 816 702.
E-mail address: maria.lopez@usc.es (M. Lpez-Abelairas).
of PHA from PHA-containing cell mass are essential for produc-
ing bioplastics from renewable resources in a cost-effective and
environmentally friendly way. The ideal method would maintain
polymer properties and achieve high purity and recovery levels
with low production costs. On an industrial scale, the use of halo-
genated solvents is the most common method for PHA extraction.
However, this method has important drawbacks, such as its high
chemical costs and its associated hazards [4]. Among the alternative
recovery processes that have been proposed, digestion of the cells
with chemicals or enzymatic cocktails and extraction with non-
halogenated agents have received the greatest interest due to their
simplicity and more affordable cost [58]. More recently, the use of
chemicals that can selectively dissolve non-PHA biomass has also
been proposed [9].
In this study, four separation processes that constitute promis-
ing alternatives to the commonly applied processes on an industrial
scale were studied using a PHB-containing biomass. For each pro-
cess, the processing conditions that achieved the highest purity and
polymer recovery were determined. Then, the different treatments
were compared considering each treatments operational perfor-
mance, economical and environmental criteria and nal product
characteristics.

http://dx.doi.org/10.1016/j.bej.2014.10.018
1369-703X/ 2014 Elsevier B.V. All rights reserved.
 M. Lpez-Abelairas et al. / Biochemical Engineering Journal 93 (2015) 250259
2. Materials and methods
2.1. PHB production
A mineral salt medium composed by 2.0 g L1 (NH4 )2 HPO4 ,
2.1 g L1 KH2 PO4 , 0.2 g L1 MgSO4 7H2 O, 0.1 g L1 CaCl2 2H2 O,
0.006 g L1 FeCl3 6H2 O, 0.1 mL L1 of trace element solution SL6
[10] and 30 g L1 of sodium gluconate was used to produce the
C. necator H16 inoculum in a Biostat DL-30 reactor (with 25 L of
medium).
Polymer accumulation with this strain was performed in a 400-L
bioreactor (Biostat D650, Sartorius) for 68 h, using a fed-batch con-
guration into which gluconate was continuously fed to induce a
nitrogen limitation [11]. A medium composed of 400 g L1 of glu-
conate, 5 g L1 MgSO4 7H2 O, 57 g L1 (NH4 )2 HPO4 and 257 g L1
NH4 Cl was fed during the rst 40 h of operation at a rate that led
to a constant gluconate concentration of approximately 17 g L1 in
the reaction medium. Then, the cells were cultivated with an excess
of carbon to attain nitrogen-limitation conditions.
2.2. PHB separation
C. necator H16 cells were separated by centrifugation and then
freeze-dried. The particle size of the resulting solid was rst
homogenised by milling using a conventional blender (maximum
speed, 30 s). Then, the homogenised solid was subjected to one
of the following treatments to recover the polymer accumulated
inside the cells. All experiments were performed in triplicate in
closed 500 mL glass bottles with a volume of liquid of 100 mL,
except for chloroform extraction.
2.2.1. Chloroform extraction
Chloroform extraction is a well-established reference standard
procedure. 1 mg mL1 of lyophilised cells was suspended in chloro-
form in closed glass tubes at 60 C for 36 h [12]. Then, the polymer
was precipitated by adding ten volumes of methanol and dried at
55 C.
2.2.2. NaOH treatment (PHB-R1)
Alkaline digestion of biomass was performed using a solid con-
tent of 2.5% w/v with NaOH solutions at different concentrations
(0.25, 0.5, 1.0, 2.0 and 4.0 N) for 4 h at 37 C and 500 rpm. Then,
the procedure was repeated with the optimal NaOH concentration
selected in the previous step and various solids contents (2.5, 5, 7.5
and 10% w/v). Samples were centrifuged, and the solid phase was
washed twice with water and once with ethanol and freeze-dried.
In each washing step, approximately 0.8 mL of liquid per g of initial
biomass was used.
2.2.3. NaOCl treatment (PHB-R2)
Similarly, a commercial NaOCl solution (13% v/v) was used to
perform biomass digestion at different solids contents (within the
range 2.57.5%, w/v) at 37 C and 500 rpm for 4 h. The solid was
separated as described above.
2.2.4. NaOCl and dichloromethane treatment (PHB-R3)
At the same temperature and agitation conditions and with a
solids content of 2.5% (w/v), a combination of NaOCl solution (13%
v/v) with a halogenated solvent, dichloromethane, in a 1:1 (v/v)
ratio was used to digest the biomass and to extract the polymer.
The polymer in the organic phase precipitated when 10 volumes
of ethanol were added. The resulting solid was washed with water
and ethanol and freeze-dried.
251
2.2.5. Acid treatment (PHB-R4)
H2SO4 solutions at several acid concentrations were used to
digest the biomass (5%, w/v) at different temperatures and for dif-
ferent times. Three levels were dened for each of these process
conditions: (i) temperatures of 37, 65 and 100 C, (ii) treatment
times of 1, 15 and 30 h; and (iii) acid concentrations of 2.5, 5 and
10% (v/v). After the acid treatment, the pH value was set to 10 using
a 0.5 N NaOH solution, and the solid was washed with water. Finally,
a mild bleaching step with sodium hypochlorite at 3% was applied
for 1 h to remove residual protein. The mathematical relationships
of the response variables (purity and recovery percentages) to the
independent variables (temperature, acid concentration and time)
were represented by quadratic model equations. Finally, response
surfaces for purity and recovery percentages were plotted using the
software MATLABTM Version 7.3.0 (The Mathworks, Inc).
2.3. Reuse of chemical solutions
After the biomass was chemically digested with alkali or acid
and the resulting solution was centrifuged to recover the solids,
different fractions of the liquid phase (20%, 40%, 60% and 80%) were
mixed with fresh solution for reuse. This operation was repeated
four times with each digestion agent.
2.4. Polymer quantication
The polymer percentage in the samples was determined using
gas chromatography (GC). Samples were subjected to methanol-
ysis in a solution with 1 mL of chloroform, 0.85 mL of methanol
and 0.15 mL of H2 SO4 [13]. The resulting esters were determined
by gas chromatography in split injection mode with helium as the
carrier gas (5 cm min1 ) using GC equipment with an FID detec-
tor and a PEG Permaphase column (60 m, 0.32 mm of diameter,
0.5 m, Restek GmbH, Bad Soden, Germany) [14]. A commercial
P(3HB) (Ref. 363502, Sigma-Aldrich) was used to determine the
calibration curve.
Purity and recovery percentages of the polymer were calculated
using Eqs. (1) and (2), respectively:
polymer weight
Purity (%) = 100 (1)
total sample weight
Wf Pf
Recovery (%) = 100 (2)
Wi Pi
where Wi is the initial dry weight of the solid introduced into the
recovery step (g); Wf is the total dry weight of the solid recovered
after the recovery step (g); Pi , Pf represent the purity of the solid
before and after the recovery process, respectively (%).
2.5. Polymer characterisation
2.5.1. X-ray powder diffraction (XRD)
The crystalline structure of the samples was studied using an
X-ray diffractometer (Philips XPert), which provides Cu K radia-
tion (40 kV, 40 mA), employing the powder method. Every scan was
recorded in the range of 2 = 250 in step-by-step mode with step
size 0.02 .
2.5.2. Differential scanning calorimetry (DSC)
In DSC, samples were heated from room temperature to 650 C
at a heating rate of 20 C min1 in an N2 atmosphere. The melting
temperature and enthalpy of fusion (Hf) were calculated from the
maximum and the area of the rst endothermic peak, respectively,
whereas decomposition temperature (Td) was calculated from the
252 M. Lpez-Abelairas et al. / Biochemical Engineering Journal 93 (2015) 250259

second endothermic peak. The degree of crystallinity (Xc) was cal- 3. Results and discussion
culated using by the following equation:
C. necator H16 cultivations were performed in a pilot bioreactor
Hf
Xc = (3) at the Institut fr Molekulare Mikrobiologie und Biotechnolo-
Hf0 gie (IMMB) at the Westflische Wilhems-Universitt [11]. Carbon
source concentration was held constant during the entire pro-
where Hf0 corresponds to the enthalpy of fusion of 100% crys- cess; nitrogen supplementation was stopped when the desired cell
talline polymer (146 J g1 ) [15]. density was achieved. When ammonium concentration attained a
critical level, the accumulation phase began. After approximately
2.5.3. Gel permeation chromatography (GPC) 68 h of operation, a total cell mass of almost 11 kg was obtained
The polymer was analysed by GPC in order to determine the with a PHB content of 65%.
weight-average molar mass (Mw ) and number-average molar mass
(Mn ). This analysis was performed using a Phenogel 5 10E5A col- 3.1. Treatment PHB-R1
umn and an IR detector. Chloroform was used as the eluent with
a ow rate of 0.8 mL min1 . Calculations were based on calibration Alkaline treatment with NaOH, here termed PHB-R1, is com-
curves obtained from monodisperse polystyrene standards ranging monly used as a pretreatment step before the main separation
from 400 Da1 up to 2106 kDa1 (Fluka 81434). The polydisper- process. However, its utility as a unique separation treatment has
sion index (PI) and degree of polymerisation (DP) were calculated lately become an object of interest due to its attractive cost. In the
according to Eqs. (4) and (5), respectively. rst step of this study, the optimal NaOH concentration for alkaline
digestion of the biomass was determined using a biomass percent-
Mw
PI = (4) age of 2.5%. Solid purity after the treatment increased as the alkali
Mn
concentration increased, up to 0.5 N (p < 0.001). Above this con-
Mn centration, no improvement in purity could be detected (p = 0.010)
DP = (5)
m (Fig. 1a). This optimal value for alkali concentration agrees with that
where m is the molar mass of the repeat unit. found by Lo et al. [6] to treat Escherichia coli cells. Moreover, as the
concentration of NaOH solutions increased, it was more difcult to
separate the biomass from the solution by centrifugation. There-
2.6. Statistical analysis
fore, a 0.5 N NaOH solution was employed to digest the biomass
obtained from C. necator fermentation. Additionally, temperatures
Statistical analysis was conducted using the software SPSS
over 37 C or treatment times longer than 5 h did not result in higher
IncPASW Statistics 18 (IBM) to determine the standard devia-
purity or recovery efciency. Particularly extreme temperatures or
tion (SD) of results obtained at different conditions, determine the
treatment times actually had a negative effect on recovery (data
existence of signicant differences and conduct pair-wise multiple
not shown).
comparisons. First, a one-way analysis of variance, using a para-
Furthermore, purity and recovery efciency were reduced when
metric method (ANOVA, based on a linear model), was performed to
the initial solids concentration was increased (Fig. 1b). Specically,
determine whether values obtained using different treatment were
at a solids concentration over 5%, purity decreased to 90% and
signicantly different. A post hoc analysis (Tukey HSD) was used
recovery fell sharply (p < 0.001). The sharp drop in recovery ef-
to determine which pairs of treatments had signicantly different
ciency may be explained by the increased difculty in separating
values, at a signicance level of 0.05.
solid and liquid phases, which decreases the recovery ratio. This
decrease in purity and recovery efciency may be a drawback for
2.7. Economic and environmental analysis
the industrial application of this methodology, despite the advan-
tage of its low cost. The best values of purity and recovery were
Energy and mass balances were solved using the process mod-
obtained with a solids content of 2.5% in a 0.5 N NaOH solution
eling software Aspen PlusTM, version 7.3 (Aspen Technology, Inc,
(0.8 g NaOH g1 solid).
MA, USA). Some components for which the ASPEN library did not
contain information, such as the biomass or the polymer, were
3.2. Treatment PHB-R2
specied using the compound structure from the CheBI database
(http://www.ebi.ac.uk/chebi/). The simulated plant would have the
The best-known alternative among the four treatments tested
capacity to treat 1000 kg h1 of cellular biomass from the polymer
for PHB recovery is the sodium hypochlorite digestion method
accumulation step. Once the simulations were performed, mate-
developed by Williamson & Wilkinson [16]. It has been com-
rial and heat ows were introduced using Aspen Process Economic
monly used at the bench scale due to its high effectiveness and
Analyzer, Economic Evaluation V7.3 (ASPEN Tech, Cambridge MA).
its simplicity [1719]. Treatment conditions such as tempera-
The costs considered in the economic analysis were those asso-
ture and hypochlorite concentration have been widely studied
ciated with the use of reagents and utilities (steam and electricity
and optimised. High purity and recovery levels can be achieved
requirements). Costs for chemicals and utilities were calculated per
at moderate temperatures (approximately 40 C) using only a
kg of polymer.
commercial sodium hypochlorite solution [20]. Suspensions of C.
Greenhouse gas (GHG) emissions, expressed as equivalent kg of
necator cells with different biomass percentages were digested
CO2, were calculated taking into account the reagents and utilities
using this method. Similar percentages of the total solid (between
required for the different processes and the associated emissions
54 and 57%) were recovered from hypochlorite digestion of differ-
reported for each of these inputs by the following equation:
ent solids loadings. However, the percentage of recovered non-PHB
n
  F
i=1 i
ei biomass increased as the solids loading increased (for example, it
GHG emissions kgCO2 eq/kg PHA = (6) rose from 3% at 2.5% of solids to 31% at 7.5% of solids), while the per-
P
centage of recovered PHB decreased (at the same solid loadings, it
where Fi is the mass ow of reagent i(kg h1 ); ei is the emissions fell from 82 to 70%). These results show that the alkali present in
associated with the production of reagent i(kg CO2 -eq kg1 ); P is the solution was not enough to digest all of the biomass, which
the mass ow of product (PHA)(kg h1 ). remained as an insoluble solid. The drop in PHB recovery may be
 M. Lpez-Abelairas et al. / Biochemical Engineering Journal 93 (2015) 250259
Fig. 1. Purity (grey bars) and recovery (white bars) percentages after 4 h at (a) several NaOH concentrations with a solids percentage of 2.5% w/v and (b) several solids
percentages in a 0.5 N NaOH solution.
explained by the presence of some of the undigested biomass on
the surface of the basic solution, which could not be recovered
by centrifugation. This decrease in digestion performance led to
a signicant decrease in the purity of the nal solid. For example,
with a solids loading of 2.5%, the purity percentage achieved with
hypochlorite was 9798%; with a solids loading of 5%, the purity
was close to 90% and, with a solids loading of 7.5%, the purity fell
to 80% (data not shown).
3.3. Treatment PHB-R3
PHA extraction using solvents is the oldest separation method.
The solvent functions include modifying cell membrane per-
meability and then dissolving the polymer inside the cells [4].
Halogenated compounds such as dichloromethane, chloroform or
1,2-dichloroethane are commonly used in these processes because
of their high extraction efciency. The combination of halogenated
solvents with sodium hypochlorite for PHA separation has also been
tested [5,18]. In this method, cells are degraded by the alkaline
solution, and the polymer is released into the medium. Hahn et al.
[18] have suggested that extensive polymer degradation is avoided
by fast extraction of the released polymer into the organic solvent
phase during the process of cellular digestion. In this method, a
mixture of sodium hypochlorite and dichloromethane was applied
for PHB isolation. PHB precipitation was performed using ethanol
as a non-solvent. Precipitation with ethanol yielded a polymer with
very high purity (over 99%) and achieved a good recovery per-
centage (90%) (Fig. 2). Similar results for purity and recovery were
Fig. 2. Purity (grey bars) and recovery (white bars) percentages for extraction meth-
ods under optimised conditions (i.e., PHB-R1: 0.5 N NaOH solution and 2.5% w/v
solids content, PHB-R2 and PHB-R3: 2.5% w/v solids content and PHB-R4: 0.64 M
H2 SO4 solution, 5% w/v solids content and 6 h).
253
reported by Hahn et al. [18], who used chloroform as the solvent
under optimised conditions. However, the hazards associated with
these solvents, as well as their high cost, are important drawbacks
for the application of this method on an industrial scale [4].
3.4. Treatment PHB-R4
Acid treatment has been proposed as an alternative digestion
method that would allow selective separation of the polymer from
the cells. The acid solution degrades cells with a marginal hydroly-
sis of the polymer stored inside, in contrast to alkaline treatments
[9,21]. At moderate acid concentrations (0.14 N), Yu et al. [21]
found that there was a dynamic balance between cleavage and
repair of ester bonds, because protons from the acid are the catalyst
for both hydrolysis and esterication. In contrast, hydroxyl anions
from alkali are reagents in polymer hydrolysis because they can
remove protons from the acid produced during ester bond cleav-
age. Thus, alkaline treatments prevent re-esterication between
the formed acids and alcohols, while acid treatments allow re-
esterication. Because of this advantage over alkaline treatments,
as well the novelty of this method with respect to the treatments
previously described, an in-depth study of acid separation was
performed. The inuence of operational conditions on purity and
recovery efciency using sulphuric acid as a digestion agent was
determined.
Three variables were studied: temperature, acid concentra-
tion and treatment time. A face-centred central composite design
(FCCD) is applied to carry out optimisation. According to the FCCD,
a total of 15 combinations of the levels of the three variables stud-
ied were selected as experimental points. The purity and recovery
percentages obtained in this assay are expressed as functions of
the process variables identied as signicant by ANOVA analysis
by Eqs. (7) and (8), respectively.
Purity(%) = 63.03 + 0.55 T + 1.86 C + 1.08 t 0.0117
(7)
T t + 0.0006 T C t 0.0023 T 2 0.17 C 2
Recovery(%) = 49.29 + 0.71 T 0.0041 T 2
0.114 C 2 + 0.0046 t 2 (8)
T is the treatment temperature ( C), C is the acid concentration
(% v/v) and t is the treatment time (h). R2 values were 0.988 for
the purity equation and 0.997 for the recovery equation, indicating
that the equations accurately reproduce the experimental data. The
maximum recovery percentage calculated from these equations
254 M. Lpez-Abelairas et al. / Biochemical Engineering Journal 93 (2015) 250259

within the current working range was 79%. Likewise, maintain-

ing this recovery value, a high purity level (almost 98%) could
be achieved. This result is similar result to that obtained for the
hypochlorite treatments. The operational conditions at which these
purity and recovery percentages can be achieved correspond to a
temperature of 80 C, an acid concentration of 3.5% (v/v) (0.64 M)
and a treatment time of 6 h. Yu and Chen [9] suggested that the
high purity of the polymer is related to the selective degradation
of molecules such as peptidoglycan that form a protective mem-
brane that stabilises the granules. They found a high rate of protein
release from cells and a clear increase of crystalline fraction in the
granules. These authors connected these two observations because
the release of the proteins that participate in granule stabilisation
facilitates polymer crystallisation. Finally, because the obtained
structure has a higher degree of crystallisation than native PHB
granules, it has an increased resistance to the alkaline attack suf-
fered in the second step of purication with sodium hypochlorite
[9]. However, Eqs. (7) and (8) imply that, although more severe
conditions (i.e., higher temperatures and acid concentrations) lead
to higher purities, they also lead to lower polymer recovery levels,
perhaps due to partial degradation of the polymer.

3.5. Recovery processes overview

3.5.1. Product properties

The polymers extracted by the methods with the best results
under optimised conditions (PHB-R1, PHB-R2, PHB-R3, PHB-R4)
and those isolated using acid treatment under more severe con-
ditions (100 C, 10% v/v and 15 h), here termed PHB-R4*, were
characterised. GC analysis of these polymers indicated that they
were fully constituted by 3HB monomers, as expected for this
strain and the applied carbon source. Additionally, X-ray analy-
sis reected diffractograms consistent with those generated by
poly(3-hydroxybutyrate) and corresponding to an orthorhombic
structure with a space group of P212121 and dimensions a = 5.76 A,
b = 13.20 A and c = 5.96 A.
The mechanical properties of amorphous polymers such as PHB
are signicantly inuenced by molar mass. Both weight-average
molar mass (Mw ) and number-average molar mass (Mn ) were
determined for each extraction method (Table 1). Molar mass dis-
tribution was also determined (Fig. 3). These parameters indicate
the degree of degradation suffered by the polymer during the
separation process. A reduction in molecular weight was observed
for all treatments compared with the chloroform-extraction con-
trol (Table 1 and Fig. 3). In this case, the most aggressive method
resulting in the lowest molar masses was treatment PHB-R2
(Mn = 36 kDa, Mw = 249 kDa). Due to their amorphous structure,
polymer granules are hydrolysed more easily than crystalline
polymers by alkaline solutions [21]. This fact is reected in the
displacement of the molar mass distribution toward lower log M
values (Fig. 3a) and the increase in the range of molar masses
present. This latter outcome is quantied by the polydispersion
index (PI), whose value increased 4-fold over control. Dispersion of
molar mass was also high for treatment PHB-R4* (Fig. 3b) implying
a structure with short polymer chains among crystalline domains
consisting of chains of greater length. Another parameter calcu-
lated using the molar mass is the degree of polymerisation (DP),
which indicates the average number of repeat units that constitute
a polymer chain. The DP depended strongly on the separation
process: the DP values found after separation using treatments
PHB-R2 and PHB-R4* were less than half of those obtained using
the other two treatments. This result indicates the signicant poly-
mer degradation caused by treatments PHB-R2 and PHB-R4*. The
molar mass distribution was unimodal for all procedures except
PHB-R4*, for which it was bimodal. This difference is similar to
that reported by Penloglou et al. [22] when long sonication times
Fig. 3. Molar mass distributions (differential weight fraction versus molar mass)
for the different recovery procedures (control with chloroform is represented by a
grey dashed line): (a) basic treatments: PHB-R1 (black line), PHB-R2 (dotted black
line) and PHB-R3 (grey line) and (b) acid treatments: PHB-R4 (black line) and PHB-
R4*(grey line).
were applied to recover PHB from Azohydromonas lata cells. Severe
conditions lead to mechanical breakage of polymer chains as a
consequence of the high energy input [22].
Degradation of polymer structure when an alkaline digestion
method is used in the separation step has been reported by many
authors [4,23]. The decrease in molar mass observed after sepa-
ration using only hypochlorite (approximately 70%) is consistent
with that reported by Berger et al. [17]. In addition, Lo et al.
[6] reported that molar mass was reduced by 66% when NaOH
solutions with concentrations above 0.5 N were applied, which
is consistent with the results of the current work. These authors
proposed the use of a surfactant, SDS (sodium dodecylsulfate), to
avoid the polymer degradation caused by sodium hydroxide. How-
ever, a surfactant dosage over 5 wt% increases recovery costs and
causes problems in wastewater treatment and reuse [8]. By con-
trast, treatments with a dichloromethane/hypochlorite mixture or
acid under moderate conditions achieved better preservation of the
polymer structure (Fig. 3a and b, respectively) and less reduction
in the molecular mass of the polymer chains. For treatment with
dichloromethane/hypochlorite mixtures, the PI value is markedly
reduced in comparison with the value for the treatment with only
hypochlorite. Hahn et al. [18] suggest that the fast migration of
the polymer into the organic phase avoids prolonged contact times
with basic solutions, thereby decreasing the breakdown of polymer
chains. PHB-R3 and PHB-R4 treatments both resulted in a polymer
with high purity (greater than 98%), although recovery efciency
 M. Lpez-Abelairas et al. / Biochemical Engineering Journal 93 (2015) 250259 255

Table 1
Characteristics of biopolymer extracted by the different recovery methods: enthalpy of fusion (H), melting temperature (Tm ), decomposition temperature (Td ), degree of
crystallinity (Xc ), weight-average molar mass (Mw ) and number-average molar mass (Mn), polydispersion index (PI) and degree of polymerisation (DP).

Treatment Conditions Tm ( C) Td ( C) H (J g1 ) Xc Mn (kDa) Mw (kDa) PI DP

PHB-R1 0.5 N NaOH solution 2.5% w/v solids content 4 h, 37 C 159171 221247 76.45 0.52 98 5 283 1 2.90 0.15 1140
PHB-R2 Sodium hypochlorite solution (13% v/v) 2.5% w/v solids 151169 284294 87.55 0.60 36 2 249 2 6.82 0.38 420
content 4 h, 37 C
PHB-R3 Dichloromethane Sodium hypochlorite solution (13% 152167 281296 88.10 0.60 138 5 361 2 2.61 0.10 1605
v/v) 2.5% w/v solids content 4 h, 37 C
PHB-R4 H2 SO4 solution (3.5% v/v) 5% w/v solids content 6 h, 159171 286300 76.38 0.52 130 4 395 1 3.03 0.10 1510
80 C
PHB-R4* H2 SO4 solution (10% v/v) 5% w/v solids content 15 h, 152165 285303 123.1 0.84 44 3 321 1 7.29 0.50 510
100 C
Control Chloroform extraction 149165 283310 87.15 0.60 521 5 837 2 1.61 .0.02 6058

was slightly higher when the combination of chemical digestion
with solvent extraction was applied.
No signicant differences in melting temperature (Tm ) were
found in PHB recovered by the different treatments (Table 1). How-
ever, the decomposition temperature (Td ) of the polymer recovered
by treatment PHB-R1 was markedly lower (221247 C) than that of
the polymer recovered by the other treatments. This result implies a
drawback for treatment PHB-R1compared with the other methods
due to the reduction in the size of the processing window within
which the polymer could be melted without being degraded. This
result may be related to the lower purity of the polymer recovered
by this method. Hong et al. [24] have reported that the addition of a
small amount of impurities to PHB results in a signicant decrease
in thermal stability due to a catalytic degradation effect.
The high degree of crystallinity (almost 85%) measured for
the polymer isolated by the PHB-R4* method is noteworthy; this
parameter was in the range of approximately 50 to 60% for the
other extraction processes and the control. This signicant increase
in the degree of crystallinity for PHB-R4* was detected in both
DSC and XRD measurements. However, similar purities (9899%)
were obtained for the acid procedures under severe and moderate
conditions, while different polymer recovery levels (79% at mod-
erate conditions and 65% at severe conditions) were found. These
different recovery levels may be explained by increased polymer
degradation under more severe conditions, which mainly affects
the amorphous fraction. Taking into account recovery, purity and
crystallinity data, the weight of amorphous and crystalline polymer
after the application of each method can be calculated. Treatment
under moderate conditions yielded 0.37 g amorphous polymer g1
initial biomass and 0.40 g crystalline polymer g1 initial biomass;
under severe conditions, treatment yielded 0.10 g amorphous poly-
mer g1 initial biomass and 0.54 g crystalline polymer g1 initial
biomass. The reduction in amorphous polymer weight under severe
conditions is most likely due to the fact that amorphous polymer
is more easily degradable than crystalline material. There was also
a net increase in crystalline polymer under severe conditions com-
pared with moderate conditions. Therefore, it can be concluded that
the use of more severe treatment conditions causes an increase in
native polymer crystallinity.
Previous studies have reported the amorphous nature of PHB
granules inside cells. Therefore, it seems that all extraction methods
lead to a rearrangement of polymer molecules toward the forma-
tion of crystalline structures. This result agrees with Lauzier et al.
[25], who indicated that polymer molecules within the amorphous
structure are in a metastable state, which can change with modi-
cations in the conditions of the surrounding medium. These authors
found that small changes in the composition or an increase in the
temperature of the medium can induce surface crystallisation of
the PHB granules. This rearrangement of polymer molecules into
lamellar crystals may be explained by the diffusion out of the gran-
ules of small molecules that act as plasticising agents, which occurs
without the protection of the lipidprotein membrane that helps to
stabilise the granules. Taking this into account, it can be supposed
that the acid treatment studied in the current work leads to the
formation of a crystalline shell in the granules. In addition, it was
observed that the more severe the acid treatment conditions, the
higher the polymer crystallinity. This observation may be explained
by two phenomena affecting PHB granules:
(1) The partial degradation of the amorphous structure of the
granules, similar to the degradation that occurs in alkaline
treatments.
(2) The increase of the crystalline shell thickness, which depends
on the diffusion of plasticizing agents within the surface. The
increase in temperature contributes to increased diffusion and,
consequently, an increased crystalline fraction in the granules.
Concerning the presence of protein in recovered polymers, Yu
and Chen [9] have described the high vulnerability of proteins to
acid dissolution. The introduction of hypochlorite digestion after
the acid step provides further protein removal. However, Hahn et al.
[18] found a negligible protein content in the polymer extracted
using hypochlorite-organic solvent dispersion and hypochlorite
solution with a concentration higher than 10%. Therefore, a neg-
ligible protein concentration should be expected in the polymer
recovered by applying hypochlorite and acid digestions.
3.5.2. Economic and environmental analysis
Notable similarities in Mw , melting temperature and crys-
tallinity data were observed for the polymer commercialised
by Biomer (Mw = 230 kDa, crystallinity = 0.60, Tm = 175 C [26],
Td = 285 C [27]) and the polymer obtained after applying treat-
ments PHB-R2, -R3 and -R4. In addition, similar characteristics were
found for the PHB produced and commercialised by Copersucar
(Mw = 220 kDa, Tm = 173 C, crystallinity = 0.55 [26]). A signicantly
higher Mw (521 kDa) was found for the polymer produced by the
control chloroform treatment. However, it has been observed that
variation in molecular weight does not lead to changes in the
mechanical properties of this type of polymer at Mn above 3.2 kDa
[28]. These characteristics imply that the polymers obtained from
treatments PHB-R2, -R3 and -R4 can be processed and applied sim-
ilarly to commercial PHB, at a similar commercial price. The selec-
tion of the most suitable recovery method should be then based on
criteria such as operational costs and environmental concerns.
In this investigation, it was experimentally demonstrated that
acid dissolution could be partially reused by replacing approxi-
mately 20% of the volume by fresh solution without losing efciency
during the treatment (data not shown). Alkaline solutions could
also be partially reused; a replacement rate of 40% resulted in a min-
imal reduction in purity of the recovered polymer (approximately
2%). For hypochlorite solution, the pH was reset to its original
value with NaOH before reuse. According to these experimental
256 M. Lpez-Abelairas et al. / Biochemical Engineering Journal 93 (2015) 250259

Fig. 4. Schemes of extraction processes: (a) PHB-R1 and PHB-R2, (b) PHB-R3 and (c) PHB-R4.
 M. Lpez-Abelairas et al. / Biochemical Engineering Journal 93 (2015) 250259
Table 2
Operational costs associated to each isolation method.
Input Cost
Chemicalsa
NaOH 0.10 D kg1 b
H2 SO4 0.08 D kg1 b
NaOCl 0.11 D kg1 b
EtOH 0.76 D kg1 b
CH2 Cl2 2.36 D kg1 b
Electricity 0.017 D MJ1 c
Steam 13.80 D t1 c
Total costs D kg1 PHB
a
Bi-distilled water was used to carry out the experiments at the bench scale.
b
http://www.sunivo.com/ennew/Products/Products list.asp.
c
ASPEN database.
results, partial recirculation of these streams was considered dur-
ing the economic and environmental analysis of these processes.
Moreover, dichloromethane recirculation is a suitable alternative
process conguration, taking into account the hazards and high
cost of this halogenated solvent. With these criteria, the process
schemes represented in Fig. 4 were proposed for treatments with
better results. Basic treatments PHB-R1 and PHB-R2 follow the
same process conguration (Fig. 4a): after digestion, the polymer
is separated by centrifugation and then washed with water and
ethanol. In treatment PHB-R3 (Fig. 4b), separation of the polymer
from the solvent (dichloromethane) is performed by precipita-
tion with ethanol acting as a non-solvent. During centrifugation
for solid recovery, a cake with a moisture content of 30% was
obtained. This moisture content indicates that small quantities of
dichloromethane and ethanol are lost during the centrifugation
step, remaining in the separated solid. Therefore, these small quan-
tities of dichloromethane and ethanol must be replaced. Otherwise,
the polymer obtained after acid digestion (PHB-R4) was processed
as previously described (Fig. 4c). The costs of the chemicals asso-
ciated with each treatment were calculated from the process
simulation and were summarised in Table 2. The quality and quan-
tity of the chemicals used in this evaluation were included as
supplementary material (SM1). To compare processes that provide
similar polymer purity percentages and in view of the experimental
results, an initial solids percentage of 2.5% was considered for treat-
ments PHB-R1, PHB-R2 and PHB-R3 and an initial solids percentage
of 5% was considered for PHB-R4. PHB-R1 and PHB-R4 involved the
lowest chemical investment (1 D kg1 PHB), although utility costs
were lower for PHB-R1 and PHB-R2 (0.02 D kg1 PHB). By con-
trast, costs for treatment PHB-R3 were signicantly higher (above
6 D kg1 PHB), because utility costs were elevated by the higher
energetic requirements in the solvent recovery steps.
It should be noted that, in this study, the biomass was freeze-
dried to attain optimal preservation and homogeneity for use in
different experiments. However freeze-drying is not viable on an
industrial scale because of its high energy requirements. A centrifu-
gation stage followed by drying at mild temperature may be the
most feasible process. Chen et al. [29] found different PHB recovery
performances using a surfactantchelate combination when differ-
ent drying procedures were applied. The best results were obtained
with freeze-drying compared with drying at 60 C or no drying at all.
However, the authors also found that the performance loss could be
offset by modifying the recovery process conditions. Consequently,
after the best strategy to dry biomass on an industrial scale is iden-
tied, recovery conditions should be reevaluated and readjusted as
necessary.
257
Treatment
PHB-R1 PHB-R2 PHB-R3 PHB-R4
1
Dh
42 19 19 20
36
2288 2288 506
472 472 624
118
5.51 5.51 18 5.51
7.11 6.76 759 11.87
1.02 5.23 6.61 1.11
Another point that inuences the recovery performance is the
polymer content in the initial biomass as well as its crystallinity
and composition. Yang et al. [30] have reported a clear inuence
of initial PHA content on recovery yield after the treatment with
detergents. This effect can be explained by the digestion of non-PHA
biomass by these compounds, similarly to the acid and basic solu-
tions used in this study. From the side of crystallinity, a crystalline
polymer would show a high resistance to chemicals and it would
suffer a lower degradation during recovery process. According with
this fact, the copolymers of PHB/PHV would show a higher degrada-
tion due to its lower crystallinity in relation with the homopolymer
of PHB.
It should be remarked that distilled water has been used in the
current experiments. On an industrial scale, different qualities can
be found depending on water source and they can affect the process
performance. For example, a signicant hardness would lead to a
low thermal stability and to the formation of precipitates in pipes
and vessels. Water consumption has not been contemplated in the
economic and environmental analyses but it has been calculated
and included in the supplementary material (SM1). The lowest
water consumption was found in treatments PHB-R2 and PHB-R3
in which water uptake is only contemplated during the washing
step after sodium hypochlorite digestion. The other two treatments
(PHB-R1 and PHB-R4) present an additional water uptake for the
preparation of the digestion solutions. The pH value of the output
streams must be adjusted to a neutral value as a part of wastewater
treatment. After organic matter removal, the reuse of the treated
water in the process may be performed.
The greenhouse gas (GHG) emission is a common metric used
to assess relative differences in the environmental impact of indus-
trial processes. GHG release is associated with the environmental
impact of global warming. Although direct GHG emissions were not
considered in the current isolation processes, emissions derived
from the production of the supplied chemicals and the energy used
as electricity or steam were taken into account. In this case, the
highest emissions resulted from the use of NaOCl (Table 3, SM2).
Therefore, the extraction methods requiring the largest amounts of
this chemical (PHB-R2 and PHB-R3) exhibit the greatest environ-
mental impact. The lowest emissions resulted from the use of acid
and sodium hydroxide.
In summary, PHB-R4 has a low cost (1.11 D kg1 ) and achieves
similar purity and recovery percentages to PHB-R2 (p = 0.93 and
0.99, respectively). In addition, acid digestion leads to a lower
environmental impact, in terms of GHG emissions, than the other
separation alternatives. The lower polymer degradation observed
with the acid treatment is another advantage. PHB-R1 is another
258 M. Lpez-Abelairas et al. / Biochemical Engineering Journal 93 (2015) 250259

Table 3
Greenhouse gasses emissions assigned to each process input and total emissions calculated for each separation process.

Treatment

PHB-R1 PHB-R2 PHB-R3 PHB-R4

Input emissions kg CO2 -eq/h

Chemicalsa
NaOH 0.47 kg CO2 -eq/kgb 196 90 90 96
H2 SO4 0.21 kg CO2 -eq/kgb 95
NaOCl 0.66 kg CO2 -eq/kgc 13728 13728 3036
EtOH 3.00 kg CO2 -eq/kgd 1863 1863 2463
CH2 Cl2 3.00 kg CO2 -eq/kgd 150
Electricity 0.128 kg CO2 -eq/MJb 43 43 141 43
Steam 0.03 kg CO2 -eq/kgb 0.54 0.52 57.87 0.91

Total emissions kg CO2 -eq/kg PHB 4.08 29.46 28.71 6.27

a
Bi-distilled water was used to carry out the experiments at the bench scale.
b
http://biograce.net/content/ghgcalculationtools/standardvalues/Bio-Grace addittional standard values - version 1 - Public.pdf.
c
http://www.waterrf.org/PublicReportLibrary/4443.pdf.
d
lca.ncms.org/VOC/CradleToSpray.xls.

economical alternative, with an operational cost of approximately
1.02 D kg1 of polymer. However, it yields a polymer with more
impurities, which modify its initial properties, and most likely give
it less commercial value. The presence of impurities limits the
medical application if the polymer: medical applications require
a high-purity polymer, without the presence of bacterial proteins,
endotoxins and other contaminating chemicals and solvents.
4. Conclusions
Acid treatment is an effective choice for PHB recovery from
cells, with high recovery efciencies and polymer purities. Acid
treatment also results in less polymer degradation and lower GHG
emissions than hypochlorite treatments. Digestion with NaOH is
also a low-cost treatment but this treatment achieves lower poly-
mer purity than hypochlorite or acid treatments. Treatments that
use sodium hypochlorite exhibit the most negative environmental
impacts.
Acknowledgments
This work was economically supported by the CDTI (Project
CEN-20091040) and also supported by the Ministry of Economy and
Competitiveness of Spain through the Local Investment Fund for
Employment (Government of Spain) and was carried out in collab-
oration with Abengoa Bionerga Nuevas Tecnologas. The authors
(MLA, MGT, TALC and JMLR) belong to the Galician Competitive
Research Group GRC 2013-032, programme co-funded by FEDER.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bej.2014.10.018.
References
[1] A. Steinbchel, H.G. Schlegel, Physiology and molecular genetics of poly(-
hydroxyalkanoic acid) synthesis in Alcaligenes eutrophus, Mol. Microbiol. 5
(1991) 535542.
[2] A. Steinbchel, H.E. Valentin, Diversity of bacterial polyhydroxyalkanoic acids,
FEMS Microbiol. Lett. 128 (1995) 219228.
[3] Y. Ling, D.R.G. Williams, C.J. Thomas, A.P.J. Middelberg, Recovery of poly-3-
hydroxybutyrate from recombinant Escherichia coli by homogenization and
centrifugation, Biotechnol. Tech. 11 (1997) 409412.
[4] N. Jacquel, C.-W. Lo, Y.-H. Wei, H.-S. Wu, S.S. Wang, Isolation and purication
of bacterial poly(3-hydroxyalkanoates), Biochem. Eng. J. 39 (2008) 1527.
[5] H.W. Ryu, K.-S. Cho, E.G. Lee, Y.K. Chang, Recovery of poly(3-hydroxybutyrate)
from coagulated Ralstonia eutropha using a chemical digestion method,
Biotechnol. Prog. 16 (2000) 676679.
[6] C.-W. Lo, H.-S. Wu, Y.-H. Wei, High throughput study of separation of poly(3-
hydroxybutyrate) from recombinant Escherichia coli XL1 blue, J. Taiwan Inst.
Chem. Eng. 42 (2011) 240246.
[7] S.N.S. Anis, M.I. Nurhezreen, K. Sudesh, A.A. Amirul, Enhanced recovery
and purication of P(3HB-co-3HHx) from recombinant Cupriavidus neca-
tor using alkaline digestion method, Appl. Biochem. Biotechnol. 167 (2012)
524535.
[8] M.H. Madkour, D. Heinrich, M.A. Alghamdi, I.I. Shabbaj, A. Steinbchel,
PHA recovery from biomass, Biomacromolecules (2013), http://dx.doi.org/
10.1021/bm4010244.
[9] J. Yu, L.X.L. Chen, Cost-effective recovery and purication of polyhydrox-
yalkanoates by selective dissolution of cell mass, Biotechnol. Prog. 22 (2006)
547553.
[10] H.G. Schlegel, H. Kaltwasser, G. Gottschalk, Ein Submersverfahren zur Kultur
wasserstoffoxidierender Bakterien: wachstumsphysiologische Untersuchun-
gen, Arch. Mikrobiol. 38 (1961) 209222.
[11] D. Heinrich, M.H. Madkour, M.A. Al-467 Ghamdi, I.I. Shabbaj, A. Steinbchel,
Large scale extraction of poly(3-hydroxybutyrate) from Ralstonia eutropha H16
using sodium hypochlorite, AMB Express 2 (2012) 59.
[12] V. Yellore, A. Desai, Production of poly-3-hydroxybutyrate from lactose
and whey by Methylobacterium sp. ZP24, Lett. Appl. Microbiol. 26 (1998)
391394.
[13] H. Brandl, R.A. Gross, R.W. Lenz, R.C. Fuller, Pseudomonas oleovorans as a
source of poly(-hydroxyalkanoates) for potential applications as biodegrad-
able polyesters, Appl. Environ. Microbiol. 54 (1988) 19771982.
[14] A. Timm, D. Byrom, A. Steinbchel, Formation of blends of various poly(3-
hydroxyalkanoic acids) by a recombinant strain of Pseudomonas oleovorans,
Appl. Microbiol. Biotechnol. 33 (1990) 296301.
[15] P.J. Barham, A. Keller, E.L. Otun, Crystallization and morphology of a bacterial
thermoplastic: poly-3-hydroxybutyrate, J. Mater. Sci. 19 (1984) 27812794.
[16] D.H. Williamson, J.F. Wilkinson, The isolation and estimation of the poly--
hydroxy-butyrate inclusions of Bacillus species, J. Gen. Microbiol. 19 (1958)
198209.
[17] E. Berger, B.A. Ramsay, J.A. Ramsay, C. Chavarie, G. Braunegg, PHB recovery
by hypochlorite digestion of non-PHB biomass, Biotechnol. Tech. 3 (1989)
227232.
[18] S.K. Hahn, Y.K. Chang, B.S. Kim, H.N. Chang, Communication to the editor opti-
mization of microbial poly (3-hydroxybutyrate) recovery using dispersions of
sodium hypochlorite solution and chloroform, Biotechnol. Bioeng. 44 (1994)
256261.
[19] C.H. Lu, Purication and Separation of Polyhydroxyalkanoates from Bacteria,
Yuan Ze University, Taiwan, 2006 (M.S. Thesis).
[20] D. Jendrossek, B. Mller, H.G. Schlegel, Cloning and characterization of the
poly(hydroxyalkanoic acid)-depolymerase gene locus, phaZ1 of Pseudomonas
lemoignei and its gene product, Eur. J. Biochem. 218 (1993) 701710.
[21] J. Yu, D. Plackett, L.X.L. Chen, Kinetics and mechanism of the monomeric prod-
ucts from abiotic hydrolysis of poly[(R)-3-hydroxybutyrate] under acidic and
alkaline conditions, Polym. Degrad. Stab. 89 (2005) 289299.
[22] G. Penloglou, E. Kretza, C. Chatzidoukas, S. Parouti, C. Kiparissides, On the con-
trol of molecular weight distribution of polyhydroxybutyrate in Azohydromonas
lata cultures, Biochem. Eng. J. 62 (2012) 3947.
[23] J. Choi, S.Y. Lee, Factors affecting the economics of polyhydroxyalkanoate
production by bacterial fermentation, Appl. Microbiol. Biotechnol. 51 (1999)
1321.
[24] S.-G. Hong, Y.-C. Lin, C.-H. Lin, Crystallization and degradation behav-
iors of treated polyhydroxybutyrates, React. Funct. Polym. 68 (11) (2008)
15161523.
[25] C. Lauzier, R.H. Marchessault, P. Smith, H. Chanzy, Structural study of isolated
poly(-hydroxybutyrate) granules, Polymer 33 (4) (1992) 823827.
[26] A.M. Abdel Ghaffar, Development of a Biodegradable Material Based on Poly(3-
Hydroxybutyrate) (PHB), Martin Luther Universitt Halle, 2002 (Ph.D. Thesis).
 M. Lpez-Abelairas et al. / Biochemical Engineering Journal 93 (2015) 250259 259

[27] C.R. Arza, P. Jannasch, F.H.D. Maurer, Network formation of graphene oxide in
poly(3-hydroxybutyrate) nanocomposites, Eur. Polym. J. 59 (2014) 262269.
[28] J.M. Torres, C.M. Stafford, B.D. Vogt, Impact of molecular mass on the elastic
modulus of thin polystyrene lms, Polymer 51 (18) (2010) 42114217.
[29] Y. Chen, Q. Xu, H. Yang, G. Gu, Effects of cell fermentation time and
biomass drying strategies on the recovery of poly-3-hydroxyalkanoates from
Alcaligenes eutrophus using a surfactantchelate aqueous system, Process
Biochem. 36 (2001) 773779.
[30] Y.-H. Yang, C. Brigham, L. Willis, C. Rha, A. Sinskey, Improved detergent-
based recovery of polyhydroxyalkanoates (PHAs), Biotechnol. Lett. 33 (2011)
937942.