Ethanol 2G: Development of a methodology to evaluate the morphology of lignocellulosic substrates Ana Sofia Brazão Borrego Thesis to obtain the Master of Science Degree in Chemical Engineering Supervisors: Dr. Damien Hudebine Dr. Nadège Charon Prof. João Carlos Moura Bordado Examination Committee Chairperson: Supervisor: Member of the Committee: Prof. Sebastião Manuel Tavares Silva Alves Prof. João Carlos Moura Bordado Dr. Maria Margarida Pires dos Santos Mateus September 2015
105
Embed
Ethanol 2G: Development of a methodology to evaluate the ... · Ethanol 2G: Development of a methodology to evaluate the morphology of lignocellulosic substrates Ana Sofia Brazão
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Ethanol 2G: Development of a methodology to evaluate
the morphology of lignocellulosic substrates
Ana Sofia Brazão Borrego
Thesis to obtain the Master of Science Degree in
Chemical Engineering
Supervisors: Dr. Damien Hudebine
Dr. Nadège Charon
Prof. João Carlos Moura Bordado
Examination Committee
Chairperson:
Supervisor:
Member of the Committee:
Prof. Sebastião Manuel Tavares Silva Alves
Prof. João Carlos Moura Bordado
Dr. Maria Margarida Pires dos Santos Mateus
September 2015
This page was intentionally left blank.
You have to learn the rules of the game.
And then you have to play better than anyone else.
Albert Einstein (1879-1955)
This page was intentionally left blank.
I
ACKNOWLEDGMENTS
I appreciate this awesome chance that IFP Energies nouvelles provided me. Damien and Nadège,
the achievement of this work will not be possible without your significant collaboration. I’m grateful for
the continuous learning that you gave me, the challenges, the confidence and all suggestions. You
made this internship an enriching experience. Also a special thanks to Marie-Olive that supported me
in my first days and always cared about my good being. I extend my gratitude to the excellent people
from Elbaite (Serge, Amandine, Karine). A particular thanks to Michel for his availability and good
humor. Generally, the support from both R12 and R05 was indispensable.
I want to thank deeply to the person that made possible that this opportunity took part of my life.
Prof. Filipa Ribeiro: thank for your teaching during last years and dedication to allow this extraordinary
experience to me and to my colleagues. Also an word to Joana Fernandes and Vitor Costa that
welcomed us in IFP in the best way. To my supervisor from Instituto Superior Técnico, Prof. João
Bordado, I thanks for the given suggestions and the final revision of this work.
This period of time made me grow up professionally but also personally. I’m grateful to the
portuguese community that integrated us (Ruben, Sonia, Leonor, Max, Leonel, Mafalda,…). I also had
the opportunity to meet people that encouraged me to learn French language and costumes (Fabien,
Mathieu, Swetan,…). The coffees and lunches together were also important, a word to Larissa, Alexis
and Raido. To my office partners, Rami, Romain, and Laure, that taught me the first words in french.
To all my friends from the university (particularly, Mariana, Filipe and Bernardo) and my friends from
ever (Ana Marta, Ana Luísa, Sara, …). People, you are incredible! Thanks for sharing extraordinary
moments with me.
A huge thank to my family in Lyon: Loios, Solange, Casinhas, David, Catarina, Joana, Diogo. For
the sharing of experiences during these six months, the adventures in the trips, the everyday dinners
together, every moments, always together. Thank you all for the friendship.
I reserve this last paragraph to express my biggest acknowledgments. To whom that told me (and
still remember me every time) to work and put all my best qualities in everything that I do. The people
who deserve all my respect: Mãe, Pai, Mano. Pedro, I thank you for believing me and for your
unconditional support.
II
This page was intentionally left blank.
III
RESUMO
Este trabalho teve como objectivo o desenvolvimento de um método que permite caracterizar a
área superficial disponível de substratos linhocelulósicos e relacioná-la com a reactividade desses
substratos na hidrólise enzimática.
Numa primeira fase, foi efectuado um estudo bibliográfico extensivo que permitiu identificar os
métodos disponíveis para o propósito. Uma técnica baseada na exclusão molecular de solutos foi
proposta com o objectivo de determinar o volume acessível nos materiais linhocelulósicos a
moléculas sonda de diferentes tamanhos. Foram exploradas duas abordagens distintas baseadas na
utilização de um substrato saturado ou seco. A segunda abordagem não existe na literatura e foi
adaptada com sucesso a partir da primeira.
A metodologia utilizando um substrato seco foi testada com uma celulose comercial (Alphacel
C40) e palha de trigo (nativa e pré-tratada a 160 oC, lavada e não lavada). Também foi proposta uma
equação modelo que descreve a distribuição de poros por tamanhos. Foi feito um estudo completo
com a Alphacel, no entanto, é necessário mais estudo sobre os restantes substratos. A técnica
caracteriza-se pelo longo tempo de espera (1 dia por molécula sonda, 10 dias por substrato). Para
solucionar este problema, foram sugeridas diversas optimizações neste trabalho.
A metodologia proposta é reprodutível e foi validada para a Alphacel. Este trabalho deverá ser
completado com a aplicação do método na caracterização de outros substratos pré-tratados, com o
objectivo de obter uma base de comparação da eficiência dos pré-tratamentos.
PALAVRAS-CHAVE:
Biocombustíveis; Etanol 2G; Linhocelulose; Área acessível; Exclusão de solutos.
IV
This page was intentionally left blank.
V
ABSTRACT
This work focused on the development of a methodology that allows to characterize the available
surface area of lignocellulosic substrates and to relate it with their reactivity on enzymatic hydrolysis.
Firstly, an extended literature review was done on the methods used for this purpose. A method
based on solute exclusion was proposed and aimed to measure the accessible volume of
lignocellulosic materials by using chemical probes of different sizes. Two approaches were explored
based on a saturated or a dried substrate. The second method does not exist in literature and was
adapted with success from the first one.
The methodology using a dried substrate was tested using a commercial cellulose (Alphacel C40)
and wheat straw (native and pretreated at 160 oC, non-washed and washed). A model equation was
also proposed in order to describe pore size distribution. A complete study was done on Alphacel but
more studies are still required for the other substrates. The main drawback of this technique is its long
experimental standby time (1 day per probe, 10 days per substrate). To solve this issue, several
optimizations were suggested in this work.
The methodology proposed is reproducible and was validated for Alphacel. The present work shall
be completed with the characterization of other pretreated substrates, in order to provide a basis to
compare pretreatment’s effectiveness.
KEYWORDS:
Biofuels; Ethanol 2G; Lignocellulose; Available area; Solute-exclusion.
VI
This page was intentionally left blank.
VII
Table of contents
Acknowledgments .................................................................................................................................... I
Resumo .................................................................................................................................................. III
Abstract .................................................................................................................................................... V
Table of contents ................................................................................................................................... VII
List of Symbols and Abbreviations ......................................................................................................... IX
List of Figures ......................................................................................................................................... XI
List of Tables ........................................................................................................................................ XIII
As can be noted on Table 5-5, an error of 0.00001% on refractive index will reproduce an error of
0.1% on the value of final concentration. Furthermore, this will represent an error of almost 20% in the
total porous volume. The last value is tremendous and certainly will affect the final distribution.
At this point, the first point was obtained. In order to obtain the other points, progressively smaller
molecules were used. By this, the accessible volume to the probes was calculated using Equation 5-6,
once the external volume is constant for the substrate.
Finally, the model equation proposed can be employed and the pore size distribution was obtained.
The data for this substrate are clearly repeatable (Figure 5-8) and Equation 5-11 reflects these results.
Figure 5-8: Pore volume distribution for Alphacel (Table 8-21, Appendix 8.10).
Vi = (0.77 ± 0.04)(1 − e−0.05D) Equation 5-11
0.0
0.2
0.4
0.6
0.8
1.0
0 50 100 150 200
Inaccessib
le p
ore
vo
lum
e
(mL
/g m
ds)
diameter (Å)
47
As can be noticed, the data reveal that the pores are almost impermeable to probes with a
diameter above 50 Å. So, it can be concluded that exists micro and mesoporosity in this substrate, in
the studied range of diameters.
Another standard cellulose was tried by Gama and his co-workers, named Avicel, and the same
evidences were verified – Figure 5-10. To clearly compare the data obtained in this work with these
results, the accessible volume as function of diameter is represented on Figure 5-9:
Figure 5-9: Pore volume distribution for Alphacel (Table 8-21, Appendix 8.10).
Figure 5-10: Pore volume distribution of celluloses (Table 5-2) [19].
For the five substrates studied by Gama, it was observed a fiber saturation point that correspond to
probe molecules with a 50 Å diameter or higher.
In particular, regarding Avicel, it was found a behavior of distribution similar to Alphacel. However,
it is important to retain that Gama and his co-workers worked with a saturated substrate. In the present
work, the dried substrate methodology is used.
To conclude, the dried substrate methodology was validated for Alphacel.
0.0
0.2
0.4
0.6
0.8
1.0
0 50 100 150 200
Ac
ce
ss
ible
po
re v
olu
me
(m
L/g
md
s)
diameter (Å)
48
Non-washed native wheat straw
As can be seen in Appendix 8.7, the accessible volume is negative for the majority of the
experimental points. This does not make sense. So, the first approach was to define that points as
zero, meaning that all porous volume is inaccessible to the molecules in question. Additionally, the
point correspondent to Dextran 75000 was eliminated, since the behavior of this molecule in solution is
unknown, and possibly can be different from the PEG’s. This resulted in:
Figure 5-11: Pore volume distribution for non-washed native wheat straw (Table 8-22, Appendix 8.10).
Vi = (5.01 ± 0.02)(1 − e−0.33D) Equation 5-12
In the figure below, it is possible to observe a zoom of the Figure 5-11, that allows to regard clearly
the error bars associated to the points. The results are not surprising due to the hypothesis applied to
the data. In this way, the points are apparently repeatable.
About porosity, the data show an impermeability above approximately 20 Å. This defines the
substrate as a microporous solid.
Figure 5-12: Pore volume distribution for non-washed native wheat straw (Table 8-22, Appendix 8.10).
0.0
1.0
2.0
3.0
4.0
5.0
6.0
0 50 100 150 200
Ina
ccessib
le p
ore
vo
lum
e
(mL
/g m
ds)
diameter (Å)
4.6
4.8
5.0
0 50 100 150 200
Inaccessib
le p
ore
vo
lum
e
(mL
/g m
ds)
diameter (Å)
49
As said earlier, a blank sample was prepared and analyzed at each trial. From the experimental
point of view, this sample containing only water instead of probe solution was prepared as control and
is advised by some authors [24, 42]. The reason for this is to allow deducting the contributions of water
soluble extractives, or other contaminants, in the refractive index measurements.
Thus, for this substrate, final concentrations were corrected for the possible existence of
contaminants (directly deducted on refractive index measurements).
The correction was done by subtracting out the specific reading from blank to wheat straw sample.
For this, the average of all measurements for the blanks was calculated. Then, the refractive index
correspondent to the contribution of the soluble can be determined through:
∆soluble = nDblank,AV − nDwater Equation 5-13
where nDblank,AV and nDwater are 1.33420 and 1.33286, respectively. As can be noticed by the
value obtained, there is a strong contribution of the soluble components.
Likewise, it is possible to determine the variation associated to the presence of probe:
∆probe = nDsample − nDblank,AV Equation 5-14
where nDsample depends on the sample and probe solution.
To finish, the corrected refractive index can be calculated by:
nD′ = ∆probe − nDwater Equation 5-15
This equations were then applied and a completely different set of results was obtained, as can be
seen in the next figure:
Figure 5-13: Pore volume distribution for non-washed native wheat straw, with refractive index correction – all points included (Table 8-23, Appendix 8.10).
The new data obtained look like as expected. In a calculation of this type, there is a subtraction of
two big numbers, with a small deviation associated to each one. The result of this deduction is a small
number, with a deviation associated of its order of magnitude.
To model these results, the point correspondent to PEG 2000 was eliminated. Also, it was used the
same hypothesis than the one applied for Alphacel: once the accessible or inaccessible volume is
negative, then it turns zero. Then, for the three smallest diameters, there are not accessible volume. In
-1.5
-1.0
-0.5
0.0
0.5
1.0
0 50 100 150 200
Inaccessib
le p
ore
vo
lum
e (
mL
/g m
ds)
diameter (Å)
50
addition, a reasoning was done concerning the measured concentrations: the points with higher
deviation between them were eliminated. These data can be extensively consulted in Appendix 8.7.
Figure 5-14: Pore volume distribution for non-washed native wheat straw, with refractive index correction. (Table 8-23, Appendix 8.10)
Vi = (0.18 ± 0.05)(1 − e−0.01D) Equation 5-16
Therefore, the new result is traduced by Figure 5-14. As explained, the points with no error bars
represent an apparent repeatability, since they result from the elimination of points in order to optimize
them. This was done with a logical reasoning due to the knowledge acquired with Alphacel
experiments.
To conclude, the results with non-washed native wheat straw are not exploitable. Too much
hypotheses and approximations were applied. This seriously affected the results. Moreover, the final
result is not similar to literature data (Table 5-3).
At this point of the work, the question of the possible existence of soluble components was taken in
account. Once the mathematically correction did not result in exploitable data, another solution was
established, acting directly on the substrate preparation. This introduces the next section, where the
native wheat straw was previously washed.
Washed native wheat straw
To confirm the hypothesis that the soluble compounds affect tremendously the refractive index
measurements, the native wheat straw was previously washed with water. Due to the time available
for the trials, only three points were obtained.
The quantity of soluble was then determined using Equation 5-13. Concerning Table 5-6 it is
perfectly visible that the quantity of soluble decreases drastically with the washing. Furthermore, the
results are measured in the refractive index apparatus. As can be seen, the order of magnitude of the
contribution of these compounds is significant for non-washed wheat straw when compared with
washed. By this, the results for this substrate were extremely affected. Subsequently, the contribution
of soluble was also deducted to the results obtained for washed native wheat straw.
0.00
0.05
0.10
0.15
0.20
0.25
0 50 100 150 200
Inaccessib
le p
ore
vo
lum
e
(mL
/g m
ds)
diameter (Å)
51
Table 5-6: Contribution of soluble components for refractive index measurements.
Substrate ∆𝐬𝐨𝐥𝐮𝐛𝐥𝐞
native non-washed 0.00134
native washed 0.00014
Figure 5-15 : Pore volume distribution for washed native wheat straw (Table 8-24, Appendix 8.10).
Vi = (1.51 ± 0.03)(1 − e−0.13D) Equation 5-17
Figure 5-16: Pore volume distribution for washed native wheat straw, with refractive index correction (Table 8-25, Appendix 8.10).
Vi = (0.72 ± 0.04)(1 − e−0.04D) Equation 5-18
Concerning these figures (Figure 5-15 and Figure 5-16), the differences are evident. The first thing
that can be noted is the difference in the fiber saturation point value. By subtracting the contribution of
soluble, the plateau decreases from 1.5 to 0.7 mL/g mds, half of the first value obtained. It can be firstly
0.0
0.5
1.0
1.5
0 50 100 150 200
Ina
cc
es
sib
le p
ore
vo
lum
e
(mL
/g m
ds)
diameter (Å)
0.0
0.2
0.4
0.6
0.8
0 50 100 150 200
Inaccessib
le p
ore
vo
lum
e
(mL
/g m
ds)
diameter (Å)
52
conclude that this contribution should be subtracted. Afterwards, if the soluble are not deducted, the
distribution obtained will correspond to a pore size value higher than the real one.
Regarding porosity, a conclusion can be made, based on the behavior of model equation
proposed. A fiber saturation point was encountered at 0.7 mL/g mds, since a size diameter of
approximately 50 Å. This indicates the presence of micro and mesopores.
Comparing the corrected data of washed native wheat straw with non-washed native wheat straw.
The fiber saturation point is completely different. Actually, the total pore volume for the washed sample
is approximately four times higher than the non-washed.
Due to this evidence, the non-washed data were not considered to make conclusions. Excepting
the evidence about the soluble compounds. With this, the existence of soluble can be corroborated
and it should be taking in account.
Washed and pretreated wheat straw
The other product tried was a sample pretreated with acid, at 160 oC. After pretreatment, the
sample was subjected to a washing in order to eliminate some contaminants.
Due to limited time, only three experiments were performed using as probes PEG 35000, PEG
20000 and glucose. Among these, the results obtained for glucose were not considered in
calculations.
Figure 5-17: Pore volume distribution for washed and pretreated wheat straw (Table 8-26, Appendix 8.10).
Vi = (0.75 ± 0.00)(1 − e−0.02D) Equation 5-19
As can be seen on Figure 5-17, two points did not allow to make conclusions. Additionally, the
model equation cannot describe this distribution due to this absent of data. A value for the fiber
saturation was obtained, nevertheless, it is not clear if the total porous volume corresponds to that
value. Regarding the same figure, it is not observed the plateau. Can be supposed that the plateau will
be at high value, as result of the pretreatment. Also, no one conclusion should be made about the type
of porosity. Accordingly, at this point, more study is required on pretreated sample.
0.0
0.2
0.4
0.6
0.8
0 50 100 150 200
Inaccessib
le p
ore
vo
lum
e
(mL
/g m
ds)
diameter (Å)
53
Conclusion
At this moment is time to make a summary of all data obtained until now. Once the data for non-
washed native wheat straw was found to be not exploitable, it is not represented in this section. Then,
the results obtained were put all in the same figure and represented using a logarithmic scale:
Figure 5-18: Pore volume distribution for Alphacel and different wheat straw products.
Once there are not published studies using this new approach of solute exclusion, as well, there
are few information on wheat straw, only a preliminary comparison can be made. Recalling Stone and
Scalan work, can be noted on Figure 5-19 that the behavior of data from the present work is
comparable with their study on pulp fibers dried and reswollen.
Figure 5-19: Pore volume distribution for pulp fibers – zoom of Figure 5-6 [40].
0.0
0.2
0.4
0.6
0.8
1 10 100 1000
Inaccessib
le p
ore
vo
lum
e (
mL
/g m
ds)
diameter (Å)
alphacel
washed native
pretreated 160oC
54
Similarly to the same authors, in the current study, the plateau was observed between 0.6 and 0.8
mL/g mds. The behavior of the curve (defined by a constant) is also analogous, however, about this
none conclusion should be made, since that this behavior depends on the substrate used.
Determination of specific surface area
As previously referred, surface area available is an important parameter on enzymatic hydrolysis,
regarding the pretreatment method (section 2.3.2). Consequently, it is interesting to study the five
different samples pretreated in this work in order to compare them with native sample and to notice the
differences between them.
Though, the limitation of the applicability of this method should be retained: nitrogen adsorption
measurements provide specific surface area for a molecule that is 3200 times smaller than the
average cellulase [60]. Concerning the figure above, this affirmation turns completely clear. There, the
influence of probe size on the determination of available surface area is perceptible.
Figure 5-20: Schematic representation of the structural features of the cellulose particle surface [19].
Afterwards, this analysis was performed on the different substrates and the results can be seen on
Table 5-7. The specific surface area, SSABET, increases from the sample pretreated at 100 °C until the
sample pretreated at 180 °C. These results were predictable since it was expected an increment on
SSA with the severity of the pretreatment. Furthermore, the native sample has a SSA lower than the
other ones. Likewise, this relation was expected and translates the effectiveness of the pretreatment
on the increasing of the available surface area.
Table 5-7: Results of specific surface area from N2 gas adsorption, in this work.
Substrate ID Pretreatment SSABET (m2/g)
wheat straw
C0033 – 0.77
CR_1082L 100 °C, 20 min, 1 %w/w H2SO4 0.97
CR_1081L 120 °C, 20 min, 1 %w/w H2SO4 1.27
CR_1083L 140 °C, 20 min, 1 %w/w H2SO4 1.71
CR_1080L 160 °C, 20 min, 1 %w/w H2SO4 2.79
CR_1084L 180 °C, 20 min, 1 %w/w H2SO4 6.27
55
Some similar results on pretreated wheat straw were found in literature. In fact, the values reported
have the same order of magnitude than the ones achieved in this work even if the operating conditions
are not the same: different temperatures (between 120 and 190 °C) with different quantities of acid
(0.5 to 2 %w/w), as well, variable residence time (7 to 240 min). By this, the results are only coherent
for the higher temperatures.
Table 5-8: Specific surface area from literature, for wheat straw, by N2 gas adsorption.
Substrate Pretreatment SSABET (m2/g) Reference
wheat straw
– 3.3 [25] – 4.0
[62]
120 °C, 240 min, 1 %w/w H2SO4 5.7 140 °C, 25 min, 2 %w/w H2SO4 5.4 160 °C, 7 min, 1.5 %w/w H2SO4 5.5 170 °C, 10 min, 1 %w/w H2SO4 6.2 180 °C, 7 min, 0.5 %w/w H2SO4 7.1 190 °C, 10 min, 0.5 %w/w H2SO4 6.6
Summary and discussion
At this point, some considerations can be done. Using Alphacel, the results of size exclusion
analysis were reproducible and dried substrate methodology was validated. Regarding non-washed
and washed native wheat straw, it can be settled that the washing is significant in order to avoid
deviation on the results. For washed native wheat straw, as well, pretreated at 160 oC and washed, a
more complete study is required.
This section intends to make some general conclusions to summarize the data obtained and to
explain details that can considerably affect the experiments.
In the figure below is represented the accessible volume of untreated and pretreated corn stover,
as function of probe diameter proposed by Ishizawa et al. [42]. In parentheses are indicated the
cellulose digestibilities after seven days. The error bars represent the standard deviation of three
replicates.
Figure 5-21: Accessible pore volume of corn stover, measured by solute exclusion [42].
56
Ishizawa and his co-workers obtained the measurements of concentration of probe using an HPLC
apparatus equipped with a refractive index detector. Considering these results can be noted that the
error bars present a random and important deviation from sample to sample. This bad reproducibility
was found in the present study and is also encountered by these authors. This reinforces the
imprecision associated to a measurement of this type.
Due to this evidence, it is important to solve the problem of the measurements. So, in order to
improve the acquisition of data, a brainstorming was performed. The issue that seems to affect more
these results is the small gap between the initial and the final concentration of probe. Consequently, it
is needed to find a solution for this.
Using Equation 5-5 and Equation 5-6, the ratio between final and initial concentration of probe can
be obtained:
Cf
Ci
=Vsol,i
Va + Ve
Equation 5-20
As it is known, the volume of initial solution is already defined (Equation 5-7) and the porous
volume as well (Equation 5-8). By this, rearranging the equation, it results in:
Cf
Ci
=1
1 −Vi
Vsol,i⁄
Equation 5-21
where Vi is characteristic of the solid in study and can be expressed by:
Vi = εsolid × Vsolid Equation 5-22
where Vsolid is the volume of biomass and εsolid its porosity. Substituting this variable in Equation
5-21, the ratio of concentrations will be defined by:
Cf
Ci
=1
1 − εsolid ∙msolid
dsolid∙
1Vsol,i
Equation 5-23
The last equation represents the relation between the ratio of concentrations at the begging and at
the end of the experiment, with the variables of the trial. As can be noted on Table 5-9, it can be seen
that the final concentration can be very different regarding the substrate used. This concentration will
allow to determine the porous volume. Therefore, this measurement should be obtained accurately.
Table 5-9: Example of concentrations of probe (Appendix 8.6; Appendix 8.7).
Substrate Probe Ci (g/100mL) Cf (g/100mL)
Alphacel PEG 20000 1.00
1.10
Wheat straw 1.02
Regarding the same expression (Equation 5-23) it is evident that two parameters can be modified
from the experimental point of view in order to maximize the ratio of concentrations: the volume of the
probe solution, Vsol,i, and the mass of substrate used, msolid. Porosity and density will depend on the
substrate and cannot be modified.
57
Influence of the mass of substrate and the volume of probe
To evaluate the impact of the mass of substrate used, Equation 5-23 can be simplified:
Cf
Ci
=1
1 − K ∙ msolid
; K =εsolid
dsolid
∙1
Vsol,i
Equation 5-24
In this case, the influence of the quantity of solid is been evaluated, maintaining constant the
volume of probe solution. Supposing that the ratio between porosity and density it is one, than the
constant K will take the value of 0.1 g-1 (once the initial volume of solution is 10 mL).
Concerning this example and the results on Table 5-10, it is clear that an increase in solid mass will
allows the final concentration to increase. In this way, with the possibility of increasing the
concentration in the end, more precise results would be obtained.
Table 5-10: Influence of substrate quantity in final concentration of probe.
Increment 𝐊 ∙ 𝐦𝐬𝐨𝐥𝐢𝐝 𝟏 − 𝐊 ∙ 𝐦𝐬𝐨𝐥𝐢𝐝 𝟏
𝟏 − 𝐊 ∙ 𝐦𝐬𝐨𝐥𝐢𝐝
𝐂𝐟
2 ∙ msolid 2 0.8 1.25 ↑
6 ∙ msolid 6 0.4 2.50 ↑↑
8 ∙ msolid 8 0.2 5.00 ↑↑↑
10 ∙ msolid 10 0 - -
Nevertheless, it exists an asymptote in the increment of mass. As can be noted on Figure 5-22, at
certain point, that increasing does not make sense: thus, there is a limit to this parameter. Observing
Equation 5-23 it is perceptible that this asymptote corresponds to the point where the denominator of
the equation turns zero.
Figure 5-22: Influence of substrate quantity in final concentration of probe.
This behavior was expected. Once the mass of substrate is increased, the porous volume is also
increasing. By this, the final concentration of probe will be equal to the initial concentration or higher
than the initial concentration if not all the probe can enters the pores. Additionally, the quantity of
solution is fixed in 10 mL. In this case, if the quantity of substrate is being increased, the porous
volume is increasing as well. Consequently, this limit of concentration can happen before, since the
0
2
4
6
8
10
0 2 4 6 8 10 12
Cf(g
/100m
L)
msolid (g)
Ci
58
porous volume can equalize the volume of solution. In this situation, the final concentration will depend
totally on the probe size diameter: if the probe size is higher than all pores, than only water will
penetrate into the substrate; on the other hand if it is smaller than the porous diameter, the final
concentration will be zero. The last case is a limit case and should not be achieved because the
volume of solution will be higher than the porous volume available and the measurement will not be
correspondent to the reality.
Then, regarding Equation 5-23, it can be noted that reducing the volume in order to increase the
final concentration, is the same thing that increase the mass of substrate. By this, can be concluded
that the true parameter will be the ratio instead of the parameters by themselves. From a mathematical
point of view, if the mass of substrate and the volume of solution are considered as independent
parameters, it will results in one equation and two parameters (over-parameterization). In order to
reduce this problem, the ratio is considered as the only parameter and then there are one equation
and one parameter.
Following this, the Equation 5-23 will be simplified as:
Cf
Ci
=1
1 − K′ ∙ Z; K′ =
εsolid
dsolid
; Z =msolid
Vsol,i
Equation 5-25
Concerning the equation below, to increase final concentration, it is required an increment on the
ratio of mass of substrate and volume of probe solution. This can mean an increasing of mass of
substrate or a decreasing on the volume of probe solution.
Figure 5-23: Influence of the ratio mass of substrate by volume of solution in final concentration of probe.
In the figure above, is represented the variation of the ratio used in this work (0.1 g/mL). The graph
was obtained by multiplying the original ratio by values higher than one, in order to make the
increment. The asymptote determined corresponds to the value obtained by multiplying the ratio by
ten. As expected it was obtained the same result generated for mass influence.
Influence of the concentration of the initial solution
Hereupon, the impact of change the initial concentration was studied. For this, the mass of solid
and the initial volume of probe solution were maintained constants. The same assumption was done
with the ratio porosity by density and this resulted in:
0
2
4
6
8
10
0.1 0.3 0.5 0.7 0.9 1.1
Cf(g
/100
mL
)
msolid/Vsol,i (g/mL)
59
Cf = Y ∙ Ci ; Y =1
1 − K′′; K′′ =
εsolid
dsolid
∙msolid
Vsol,i
Equation 5-26
As can be noted by the equation above, the influence on final concentration will be proportional
with the increment in the initial concentration, with a constant that will depend on the substrate.
Accordingly, change only the initial concentration does not make sense, except for limiting the impact
of soluble in the refractive index compared to the impact of the probe.
Experimental issues
As said, the mathematic analysis performed to study the influence of the variables does not
evaluate the issues associated to the experimental part. The parameters can be modified, still the
experience with the materials in cause is determinant to make decisions.
Regarding the protocol described on section 4.3 that was the one applied in this work, some
suggestions can be done in order to optimize the experiments. These recommendations result from
the observations performed during the laboratory work.
Considering the elementary steps, the first issue is related to the mixing of substrate with probe
solution. First of all, the probe solution is added to the dried substrate (case A). Then, with the help of
a spatula, the substrate is pushed to the bottom of the flask in order to allow the contact between
probe and substrate (case B). After the stirring begins.
Figure 5-24: Experimental issue on stirring.
Figure 5-25: Comparison between a native and a pretreated wheat straw samples, after stirring.
60
As can be noted in Figure 5-24, after stirring and decantation, there is straw that ascend the flask.
This could be affect the results, since a part of substrate cannot being in contact with probe solution.
This evidence was noted for native wheat straw but it is not so visible with Alphacel and pretreated
wheat straw (Figure 5-25). To solve this problem the type of agitation should be changed.
Another important point is the washing of the substrate. In Figure 5-26, there is a visible difference
in the color of the samples if the substrate is washed or not. It was verified that there is a contribution
of soluble compounds that affects the results. A solution is to increase the initial concentration of
probe, to limit the impact of these soluble compounds. Another solution passes by the washing of the
substrate. If these solutions are not possible, the blank sample should be maintained in order to
deduce these contributions.
Figure 5-26: Comparison between a washed (1’) and a non-washed (1) native wheat straw supernatants.
Methodology optimization
Since this technique is very labor-intensive, it took some time to learn how to do it and get
consistent and reproducible data. By this, this section intends to propose a way to reduce the time of
the experiment and obtain the high number of data possible in a reduced time space.
As said before, instead of what is described in literature, an approach to dry substrate was
performed in this study. The final protocol resulted in the performing of one experiment by day, as can
be seen in next table:
Table 5-11: Day work plan to performed one experiment with the dried substrate methodology.
08H 09H 10H 11H 12H 13H 14H 15H 16H 17H
Step b)
Step c)
Step d)
Step e), Step f)
Step g)
Step h)
Results treatment
Preparation of next assay
As can be seen in the timetable above, each experiment (one experiment corresponds to one
probe) takes one entire day to be performed, concerning the experimental part, as well the treatment
of the results and the preparation of the next one. So, subsequently to the first experiments, with the
intention to obtain a high number of results in a short period of time, an adjustment was done, to try
perform two experiments in one day:
61
Table 5-12: Day work plan to performed two experiments by dried substrate methodology.
08H 09H 10H 11H 12H 13H 14H 15H 16H 17H 18H
Step b)
Step c)
Step d)
Step e), Step f)
Step g)
Step h)
Results treatment
Preparation of next assay
Making this upgrading, it was possible to obtain more data during this work. Still, this planning can
be improved. After to known how to perform the experiments, the operator will be able to perform three
different trials simultaneously. For this, step d) (settling down) should be reduced or eliminated, for
example. Another way to reduce the obtain data quickly is to use a different analyzer instead of the
refractometry apparatus. Actually, this step comprises more than 25% of the total time of the
methodology (due to the calibration).
The number of probes used can also be a point to explore. At this moment the expected behavior
of data for this method is known. Therefore, the probes can be selected more accurately and not by
trial and error experiments. Recovering the example of Alphacel:
Figure 5-27: Pore volume distribution for Alphacel (Table 8-21, Appendix 8.10).
If this trial was being repeat now, some changes should be done in order to adjust better the
experimental points to the proposed equation. First, more trials should be performed using the
smallest probe molecules, since this data revealed to be not feasible. The points referred to the probes
between 30 and 70 Å should be maintained since they define the form of the curve. Last, but not the
0.0
0.2
0.4
0.6
0.8
1.0
0 50 100 150 200
Inaccessib
le p
ore
vo
lum
e
(mL
/g m
ds)
diameter (Å)
62
least, the high size diameter molecules should be kept and maybe a bigger ones should be tried to
verify the possible existence of other type of macroporosity. By this, concerning Figure 5-27 can be
suggested a reduction of the number of probes used to define the distribution of pores. Five probes
can be enough: two to define the fiber saturation point, two to define the curve (and the constant as
result) and one smaller to better adjust the model equation.
Regarding the type of analyzer to the concentration measurements an HPLC apparatus coupled to
a refractometric detector can be used. By this, a solution containing of several probe molecules can be
used and the time of measurement of final concentration will considerably decrease.
63
6 CONCLUSIONS AND FUTURE PROSPECTS
The literature review on the available methods allowed us to explore the techniques used for the
structural characterization of lignocellulosic substrates. From this study, one method named solute
exclusion technique was selected and applied on wet substrates. Rapidly, the first experiments have
shown poor results in terms of reproducibility and a new approach was established using dry
substrates.
The present work focused on the solute exclusion technique, which was firstly performed by Stone
and Scalan, in 1968, and is still employed. Regarding the method selected, and, particularly, the
preparation of the substrate, all methods have in common the use of a water saturated substrate. This
methodology was performed using commercial products (Avivel and Alphacel) and it was found to be
not interesting due to the high uncertainties on the final results, being coherent with literature. By this,
it was discarded.
Subsequently, the new approach of the method was performed. The main difference was the use
of a dry substrate. Nevertheless, it is referred in literature an issue that involves the possible collapse
of pores during drying [60].
In despite of this risk, the dried substrate method has been tested during this training period. It was
found reproducible and was validated for a commercial product, Alphacel C40. Similar results were
found in literature [19] for other standard celluloses, such as Avicel. The technique was also applied to
wheat straw (native non-washed and washed, and pretreated at 160 oC). For these substrates a more
complete study would be required due to the low number of points obtained.
For determining the concentration of the probes after mixing, a refractive index apparatus was
used. It was found the need to eliminate contaminants that affected the measurements, especially in
the case of the wheat straw wherein the amount of soluble compounds is high. This evidence was
found by comparing native wheat straw samples after washing step or not. Following this, it is strongly
recommended the preparation of a blank sample which allows to deduct the contribution of these
compounds on the measurements. Additionally, the washing of the substrate is recommended to
minimize their contribution in the refractive index of the final solution.
To describe the pore volume distribution, a model equation was also proposed. This equation
describes the distribution of the porosity of a substrate as function of the pore diameter using two
variables: the fiber saturation point, or maximal porous volume, and a constant that depends on the
substrate. Using this equation it was possible to known the pore size distribution of various substrates.
It is proposed the use of only five or six probes for a substrate, and, applying the equation proposed it
will be possible to have a complete description of the distribution of pores.
In summary, this work proposes a new approach of the solute exclusion technique, as well, a
model equation that can describe the pore size distribution. Optimizations shall be done and the
technique must be validated for other type of substrates. It was also suggested the use of combined
characterization techniques in order to obtain complete information about accessibility. This work was
significant since it deal with the relation between substrate accessibility and enzymatic hydrolysis.
64
65
7 REFERENCES
[1] P. Halling and P. Simms-Borre, "Overview of lignocellulosic feedstock conversion into ethanol -
focus on sugarcane bagasse," 2008. [Online]. Available: www.internationalsugarjournal.com.
[Accessed 4 March 2015].
[2] European Biofuels - Technology Platform, "Cellulosic Ethanol," [Online]. Available:
http://www.biofuelstp.eu/cellulosic-ethanol.html#ce1. [Accessed 10 August 2015].
[3] "Dioxyde de Carbone," [Online]. Available: http://www.respire-asso.org/dioxyde-de-carbone-co2/.
[Accessed 15 June 2015].
[4] M. Guo, W. Song and J. Buhain, "Bioenergy and biofuels: History, status, and perspective,"
Renewable and Sustainable Energy Reviews, no. 42, pp. 712-725, 2015.
[5] International Transport Forum, "Reducing Transport Greenhouse Gas Emissions - Trends&Data,"
2010.
[6] "2030 Energy Strategy," [Online]. Available: http://ec.europa.eu/energy/node/163. [Accessed 15
June 2015].
[7] IFP Energies nouvelles, "Biofuels 2G, Biocarburants - Production/consommation par zone
The average molecular weight and diameter for each probe were obtained from the literature:
Table 8-2: Solution molecular diameters of probes from literature [40, 53].
Probe Molecular weight
(g/mol)
Diameter
(Ǻ)
Glucose 180 8
Cellobiose 342 10
PEG 200 190-210 13
PEG 400 285-315 18
PEG 600 570-630 21
PEG 1000 950-1050 27
PEG 1500 1300-1600 33
PEG 3500 3000-3700 50
PEG 8000 7000-9000 84
Dextran 75000 72000 120
PEG 20000 15000-20000 130
72
For the probes used, some diameters were not found, and for that one’s an equation obtained by
power curve was employed:
Figure 8-1: Correlation obtained for PEG probes by power curve.
Results from water retention value method
Table 8-3: Drying of substrate WRV1, for Avicel PH101.
ID Set A Set B
t (h) m1 m2 m3 m4 m5 m6
0 1.1984 1.2150 1.3864 0.9793 1.0289 1.1076
17.5 0.7695 0.7976 0.7568 0.6138 0.7826 0.8392
Table 8-4: Drying of substrate WRV2, for Avicel PH101.
ID Set A Set B
time (h) m1 m2 m3 m4 m5 m6
0 1.1726 1.0822 0.8832 0.9990 1.0546 1.0236
0.5 0.7352 0.6836 0.5657 0.6681 0.7020 0.6783
1 0.6612 0.5578 0.4941 0.4887 0.5452 0.5222
1.5 0.6604 0.5571 0.4932 - - -
2 - - - 0.4635 0.4896 0.4654
2.5 0.6599 0.5566 0.4930 - - -
3 - - - 0.4632 0.4882 0.4651
3.5 0.6594 0.5565 0.4931 - - -
4 - - - 0.4630 0.4880 0.4657
4.5 0.6598 0.5564 0.4931 - - -
5 - - - 0.4620 0.4873 0.4644
5.5 0.6586 0.5554 0.4918 - - -
73
Table 8-5: Drying of substrate WRV3, for Avicel PH101.
ID Set A Set B
time (h) m1 m2 m3 m4 m5 m6
0 1.0954 0.9998 1.0690 1.0555 1.0773 1.1101
17 0.5949 0.5260 0.5445 0.5960 0.6163 0.6284
Table 8-6: Drying of substrate WRV4, for Alphacel C40.
ID Set A Set B
time (h) m1 m2 m3 m4 m5 m6
0 1.1606 1.2046 1.2871 1.1842 1.3343 1.3151
0.5 0.7254 0.7481 0.8217 0.8181 0.9491 0.9941
1 0.6698 0.6894 0.7341 - - -
1.5 0.6676 0.6869 0.7302 0.6935 0.7784 0.7589
2 0.6664 0.6863 0.7290 - - -
2.5 - - - 0.6928 0.7774 0.753
3 0.6665 0.6866 0.7296
3.5 - - - 0.6916 0.7772 0.7533
4 0.6653 0.6863 0.7290 - - -
4.5 - - - 0.6933 0.7778 0.7531
5 0.6661 0.6870 0.7309 - - -
Table 8-7: Results of WRV for Avicel PH101.
ID m1 m2 m3 m4 m5 m6
mss (g) 1.1606 1.2046 1.2871 1.1842 1.3343 1.3151
mds (g) 0.6677 0.6890 0.7330 0.6949 0.7795 0.7549
WRV (g/g) 0.78 0.95 0.80 1.16 1.16 1.20
AV±SD (g/g) 1.01±0.19
Table 8-8: Results of WRV for Alphacel C40.
ID m1 m2 m3 m4 m5 m6
mss (g) 1.1726 1.0822 0.8832 0.9990 1.0546 1.0236
mds (g) 0.6586 0.5554 0.4918 0.4620 0.4873 0.4644
WRV (g/g) 0.74 0.75 0.76 0.70 0.71 0.74
AV±SD (g/g) 0.73±0.02
74
Calibration curves (refractometry)
For all the assays a new set of solutions was prepared and the calibration of the refractometer was
performed. In the next table are presented the linear correlations correspondent to the considered
assays:
Table 8-9: Linear correlations between refractive index and concentration of probe.
ID Probe Date Linear correlation R2
SE01 PEG 20000 29/04/2015 y = 0.00137x + 1.33283 0.9988
SE02 PEG 20000 29/04/2015 y = 0.00137x + 1.33283 0.9988
SE03 PEG 8000 06/05/2015 y = 0.00138x + 1.33280 0.9995
SE04 PEG 8000 06/05/2015 y = 0.00137x + 1.33281 0.9995
SE06 PEG 20000 19/05/2015 y = 0.00143x + 1.33278 0.9974
SE10 PEG 200 29/05/2015 y = 0.00126x + 1.33276 0.9975
SE12 PEG 8000 01/06/2015 y = 0.00139x + 1.33281 0.9993
SE13 PEG 10000 02/06/2015 y = 0.00143x + 1.33277 0.9977
SE14 PEG 1500 05/06/2015 y = 0.00142x + 1.33278 0.9904
SE15 PEG 600 08/06/2015 y = 0.00134x + 1.33280 0.9993
SE16 PEG 2000 08/06/2015 y = 0.00140x + 1.33280 0.9936
SE18 Cellobiose 09/06/2015 y = 0.00141x + 1.33295 0.9890
SE19 PEG 200 10/06/2015 y = 0.00133x + 1.33272 0.9985
SE21 PEG 20000 17/06/2015 y = 0.00132x + 1.33291 0.9999
SE22 PEG 20000 18/06/2015 y = 0.00138x + 1.33286 0.9995
SE23 Dextran 75000 19/06/2015 y = 0.00142x + 1.33288 0.9999
SE24 PEG 2000 22/06/2015 y = 0.00115x + 1.33314 0.9767
SE25 PEG 8000 22/06/2015 y = 0.00133x + 1.33289 0.9999
SE26 PEG 1500 23/06/2015 y = 0.00132x + 1.33289 0.9999
SE27 PEG 4000 23/06/2015 y = 0.00132x + 1.33290 0.9999
SE28 PEG 200 24/06/2015 y = 0.00121x + 1.33285 0.9999
SE29 Glucose 30/06/2015 y = 0.00134x + 1.33298 0.9993
SE30 Glucose 30/06/2015 y = 0.00141x + 1.33289 0.9999
SE31 PEG 35000 02/07/2015 y = 0.00135x + 1.33289 0.9999
SE32 PEG 35000 02/07/2015 y = 0.00135x + 1.33289 0.9999
SE33 Glucose 09/07/2015 y = 0.00142x + 1.33289 0.9986
SE34 Glucose 10/07/2015 y = 0.00144x + 1.33287 0.9973
SE35 PEG 35000 10/07/2015 y = 0.00135x + 1.33289 0.9998
SE36 PEG 200 15/07/2015 y = 0.00132x + 1.33274 0.9996
SE37 Cellobiose 15/07/2015 y = 0.00147x + 1.33288 0.9998
SE38 PEG 600 16/07/2015 y = 0.00131x + 1.33285 0.9999
75
Saturated substrate methodology – Alphacel
Table 8-10: Results from saturated substrate methodology, for Alphacel.
PE
G 2
00
00 ID SE01 mi (g) mss (g) WRV Ci (g/100mL) nD Cf (g/100mL) Vp (mL)
m1 1.0 1.0014 0.52
1.02
1.33427 1.05 -0.29
m2 1.0 1.0000 0.64 1.33429 1.07 -0.46
m3 1.0 1.0045 0.65 1.33426 1.04 -0.24
m4 1.0 1.0082 0.63 1.33428 1.05 -0.34
PE
G 2
00
00
ID SE02 mi (g) mss (g) WRV Ci (g/100mL) nD Cf (g/100mL) Vp (mL)
m1 1.0 1.0006 0.63
1.02
1.33424 1.03 -0.09
m2 1.0 1.0103 0.62 1.33424 1.03 -0.10
m3 1.0 1.0017 0.67 1.33423 1.02 -0.05
m4 1.0 1.0056 0.63 1.33428 1.05 -0.34
PE
G 8
00
0
ID SE03 mi (g) mss (g) WRV Ci (g/100mL) nD Cf (g/100mL) Vp (mL)
m1 1.0 1.0052 0.61
1.00
1.33418 1.00 0.03
m2 1.0 1.0050 0.65 1.33415 0.98 0.25
m3 1.0 1.0011 0.65 1.33416 0.99 0.17
m4 1.0 1.0010 0.61 1.33417 0.99 0.10
PE
G 8
00
0
ID SE04 mi (g) mss (g) WRV Ci (g/100mL) nD Cf (g/100mL) Vp (mL)
m1 1.0 1.0069 0.70
1.00
1.33420 1.01 -0.16
m2 1.0 1.0035 0.63 1.33419 1.01 -0.08
m3 1.0 1.0080 0.61 1.33419 1.01 -0.08
m4 1.0 1.0002 0.65 1.33419 1.01 -0.08
PE
G 2
00
ID SE10 mi (g) mss (g) WRV Ci (g/100mL) nD Cf (g/100mL) Vp (mL)
m1 1.0 1.0180 0.67
1.00
1.33401 0.99 0.10
m2 1.0 1.0190 0.67 1.33402 1.00 0.02
m3 1.0 1.0120 0.65 1.33400 0.98 0.18
m4 1.0 1.0250 0.59 1.33399 0.98 0.26
76
Dried substrate methodology – Alphacel
Results that present a significant deviation were not considered in calculations (signed with an asterisk – *). For the molecules with high diameter, when the result of determination of accessible volume is negative, is considered zero for the treatment of the data (signed with double asterisk – **).
Table 8-11: Results from dried substrate methodology, for Alphacel (part 1).
PE
G 2
00
00
ID SE06 mi (g) mf (g) H (%) mds (g) Ci (g/100mL) nD Cf (g/100mL) Ve (mL) Va (mL) Vi (mL) Ve (mL/g mds) Va (mL/g mds) Vi (mL/g mds)
Results that present a significant deviation were not considered in calculations (signed with an asterisk – *). When the result of determination of accessible
volume or inaccessible is negative, is considered zero for the treatment of the data (signed with double or triple asterisk, respectively).
Table 8-13: Results from dried substrate methodology, for non-washed native wheat straw (part 1).
PE
G 3
50
00
ID SE31 mi (g) mf (g) H (%) mds (g) Ci (g/100mL) nD Cf (g/100mL) Ve (mL) Va (mL) Vi (mL) Ve (mL/g mds) Va (mL/g mds) Vi (mL/g mds)
Results that present a significant deviation were not considered in calculations (signed with an asterisk – *). For the molecules with high diameter, when
the result of determination of accessible volume is negative, is considered zero for the treatment of the data (signed with double asterisk – **).
Table 8-18: Results from dried substrate methodology, for washed native wheat straw.
PE
G 3
50
00
ID SE32 mi (g) mf (g) H (%) mds (g) Ci (g/100mL) nD Cf (g/100mL) Ve (mL) Va (mL) Vi (mL) Ve (mL/g mds) Va (mL/g mds) Vi (mL/g mds)
Table 8-19: Results from dried substrate methodology, for washed native wheat straw – nD correction.
PE
G 3
50
00
ID SE32 nD nD’ Cf (g/100mL) Ve (mL) Va (mL) Vi (mL) Ve (mL/g mds) Va (mL/g mds) Vi (mL/g mds)
m1 1.33449 1.33435 1.08 9.26 0 0.74 9.23 0 0.74
m2 1.33448 1.33434 1.08 9.32 0 0.68 9.27 0 0.68
m3 1.33449 1.33435 1.08 9.26 0 0.74 9.19 0 0.74
m4 1.33451 1.33437 1.10 9.13 0 0.87 9.10 0 0.87 *
PE
G 2
00
ID SE36 nD nD’ Cf (g/100mL) Va (mL) Vi (mL) Va (mL/g mds) Vi (mL/g mds)
m1 1.33426 1.33412 1.05 0.37 0.35 0.37 0.35
m2 1.33424 1.33410 1.03 0.51 0.21 0.51 0.21 *
m3 1.33426 1.33412 1.05 0.37 0.35 0.37 0.35
m4 1.33430 1.33416 1.08 0.10 0.62 0.10 0.62 *
Glu
co
se
ID SE34 nD nD’ Cf (g/100mL) Va (mL) Vi (mL) Va (mL/g mds) Vi (mL/g mds)
m1 1.33446 1.33432 1.01 0.65 0.07 0.65 0.07
m2 1.33447 1.33433 1.02 0.58 0.14 0.58 0.14
m3 1.33447 1.33433 1.02 0.58 0.14 0.58 0.14
m4 1.33450 1.33436 1.04 0.38 0.34 0.38 0.34 *
85
Dried substrate methodology – Wheat straw pretreated at 160 °C and washed
Results that present a significant deviation were not considered in calculations (signed with an asterisk – *). For the molecules with high diameter, when
the result of determination of accessible volume is negative, is considered zero for the treatment of the data (signed with double asterisk – **).
Table 8-20: Results from dried substrate methodology, for wheat straw pretreated at 160 °C and washed.
PE
G 3
50
00
ID SE35 mi (g) mf (g) H (%) mds (g) Ci (g/100mL) nD Cf (g/100mL) Ve (mL) Va (mL) Vi (mL) Ve (mL/g mds) Va (mL/g mds) Vi (mL/g mds)