Improving the synthesis of Zn-Ta-TUD-1 for the Lebedev process using the Design of Experiments methodology

Article abstract of DOI:10.1016/j.apcata.2019.117386

The synthesis method of a zinc-tantalum catalyst supported on three-dimensional mesoporous silica with high specific surface area was studied. Its activity in the conversion of ethanol to butadiene was optimized using the Design of Experiment approach. A Plackett-Burman screening design identified the important preparation parameters, notably the ratio of Zn to Ta. It was subsequently optimized using the Response Surface Methodology, affording a highly active catalyst.

Full text of DOI:10.1016/j.apcata.2019.117386

Journal Pre-proof
Improving the synthesis of Zn-Ta-TUD-1 for the Lebedev process using
the Design of Experiments methodology
Guillaume Pomalaza (Investigation) (Methodology) (Formal
¨
analysis) (Writing - original draft), Mickael Capron (Supervision)
(Funding acquisition), Franck Dumeignil (Supervision) (Writing -
review and editing) (Funding acquisition) (Validation)
PII:
S0926-860X(19)30541-1
DOI:
https://doi.org/10.1016/j.apcata.2019.117386
APCATA 117386
Reference:
To appear in:
Applied Catalysis A, General
Received Date:
Revised Date:
Accepted Date:
12 October 2019
22 December 2019
24 December 2019
Please cite this article as: Pomalaza G, Capron M, Dumeignil F, Improving the synthesis of
Zn-Ta-TUD-1 for the Lebedev process using the Design of Experiments methodology, Applied
Catalysis A, General (2019), doi: https://doi.org/10.1016/j.apcata.2019.117386
This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.
© 2019 Published by Elsevier.

Improving the synthesis of Zn-Ta-TUD-1 for the Lebedev process using the Design of
Experiments methodology
Guillaume Pomalaza,a Mickaël Capron,a Franck Dumeignila,*
aUniv. Lille, CNRS, Centrale Lille, ENSCL, Univ. Artois, UMR 8181 – UCCS – Unité de Catalyse
et Chimie du Solide, F-59000 Lille, France
*corresponding author, franck.dumeignil@univ-lille.fr
Graphical abstract
Highlights
1

●
Zinc-tantalum on mesoporous silica TUD-1 catalyst showed remarkable activity in the conversion
of ethanol to butadiene
●
●
●
Plackett-Burman design identified key synthesis parameters for high productivity
TUD-1 synthesis parameters compromised between surface area and pore diameter
Response Surface Methodology identified high performing catalyst composition
Abstract:
The synthesis method of a zinc-tantalum catalyst supported on three-dimensional mesoporous
silica with high specific surface area was studied. Its activity in the conversion of ethanol to
butadiene was optimized using the Design of Experiment approach. A Plackett-Burman
screening design identified the important preparation parameters, notably the ratio of Zn to Ta.
It was subsequently optimized using the Response Surface Methodology, affording a highly
active catalyst.
Keywords: Ethanol; 1,3-Butadiene; TUD-1; Design of Experiment; Supported catalyst
1. Introduction
The Lebedev process, the conversion of ethanol to 1,3-butadiene (BD), is being considered as
a sustainable alternative to hydrocarbon steam cracking. The latter which currently produces 95%
of BD—the world’s most consumed diolefin [1–6]. Not only does the Lebedev process use a widely
available feedstock derivable from biomass, it is also much more selective towards BD than steam
cracking [1]. Selectivity comes into play when considering the purity needed by polymerization
catalytic processes used to synthesize rubber from BD [5–7]. However, to financially compete with
fossil-based routes, the Lebedev process requires—amongst other things—better performing
catalysts [8,9].
Silica-supported metal and metal oxide mixtures have demonstrated high catalytic activity in
the Lebedev process [1,4]. Their performances are owed to the multi-functionality provided by the
2

combination of different metals or metal oxides, each possessing complementary chemical
properties required to catalyze the multi-step ethanol-to-butadiene reaction (Fig. 1). A balance
between these properties has been cited as the key to maximizing BD production [10]. We found
through a preliminary rough screening study (not published) that silica-supported Zn and Ta
yielded the largest amount of BD compared with the other transition metals tested, e.g., Al, V, Cu,
Ga, Zr, Nb,Hf, La, and Ce. In addition, catalyst structural properties have been linked to superior
catalytic activity in the Lebedev process: high active phase dispersion [11–15], large specific
surface areas [15,16], three-dimensional mesoporous morphology [8,11,17] were all found to
improve catalytic performances in metrics such as BD productivity, BD selectivity and resistance
to coke deactivation.
In our previous work, we used the procedure developed at the Delft University of Technology
to synthesize mesoporous silica (TUD-1) to prepare a Zn-Ta-TUD-1 catalyst showing remarkable
performances in the conversion of ethanol to BD compared to commercial silica and dealuminated
BEA materials [18,19]. The Zn-Ta-TUD-1 material proved to be more productive and stable than
other highly active catalysts under comparable reaction conditions, notably hierarchical MgO-SiO2
reported by Men et al. and Zn-Y/SiBEA reported by Li et al. [20,21]
We seek to improve the performances of the Lebedev process by tuning the synthesis of
catalysts possessing the important physical and chemical properties mentioned above. TUD-1
materials have a three-dimensional sponge-like mesoporous morphology and many advantages
over conventional mesoporous catalyst carriers [22,23]. They boast a simple, yet cost-effective
one-pot synthesis based on the sol-gel process, with tunable pore size and specific surface area,
ranging from 2 – 50 nm and 400 – 1000 m2/g, respectively. TUD-1 materials are also reported to
have a high hydrothermal stability, which suits them well for processes involving the dehydration
of alcohol at high temperature. Furthermore, metals are easily introduced and dispersed within the
3

silica framework with minor adaptation of the preparation procedure [23]. A key component of the
TUD-1 synthesis is the addition of an organic chelating agent during the sol-gel process: it forms
complexes of the metal and silica precursors, insuring their homogeneous dispersion throughout
the preparation by preventing cluster formation; it also acts as a structure-directing agent to produce
the sponge-like morphology when the silica precursor condenses during thermal treatment of the
gel [23,24]. Catalysts with a highly dispersed active phase, a large specific surface area and a
mesoporous morphology for the Lebedev process can thus be obtained. However, despite its
simplicity, the TUD-1 preparation procedure needs to be treated carefully: the effects of several
synthesis parameters are unclear in the literature, which may cause unexpected result when
scientists attempt to adapt the method for their own purposes. Furthermore, authors working with
TUD-1 sometimes omit to justify their preferences when adapting the synthesis method. One
instance we encountered was the use of tetraethyl ammonium hydroxide (TEAOH) as an alkalizing
agent during the sol-gel process. Although descripted as optional in the original paper by Jansen et
al. [18] most scholars resort to it, undoubtedly due to its role as gelation catalyst. However, the
quantity in relation to silica precursor amount appears to arbitrarily change from one publication
to another [22,25,26]. Parameters we found to change depending on the publication were the
calcination method [25,27] and solvent used [22,25,28–30], amongst other.
The objective of our work was thus two-fold: to prepare a Zn-Ta-TUD-1 catalyst active in the
Lebedev process based on the observations of our previous work, as well as sorting and
understanding the effect of certain synthesis variables on the morphology of bimetallic TUD-1.
These goals were achieved using a Design Of Experiment (DOE) methodology combined with
mathematical and statistical techniques which allow the modeling of dependent responses to the
independent variables of a process. Such models can be used for process optimization, but also for
statistical interpretation in order to study the influence exerted by each independent variable on the
4

selected response. First, a Plackett-Burman (PB) experimental design was used to identify
important variables of the Zn-Ta-TUD-1 synthesis and their impact on BD productivity, specific
surface area and pore size. It is a two-level factorial design of experiment that allows the screening
of n – 1 factors in a maximum of n experiments, where n is the number of runs and a multiple of
four [31,32]. This highly economical design is ideal for studying processes that are expensive or
time-consuming, but comes at the cost of screening resolution, meaning only the main effects of
each variables can be calculated. With a better understanding of the Zn-Ta-TUD-1 synthesis, the
catalyst was further optimized for BD productivity using a three-level factorial design of
experiment combined with the response surface methodology (RSM), a mathematical-statistical
technique used in engineering for experiment design and process optimization [31,33–35]. In this
case, only two independent variables were selected—Zn and Ta concentration in the catalyst—
enabling a more descriptive study of their effect on BD productivity. Catalytic testing and
characterization of the morphological properties were performed to gather the experimental data
needed for empirical modelling.
2. Material and methods
2.1. Reagents & materials
For the synthesis of Zn-Ta-TUD-1, two sources of each metal were used alternatively: tantalum
chloride (Alfa Aeser, 99.8%) or optical grade tantalum ethoxide (Alfa Aeser, 99.95%), and zinc
chloride (Acros Organics, 97+%) or zinc acetate dehydrate (Acros Organics, 98+%). Two different
chelating agents were used: triethanol amine (or TEAH3, Acros Organic, 99+%) or tetraethylene
glycol (or TEG, Agros Organics, 99.5%). Tetratethyl orthosilicate, (or TEOS, Agros Organics,
98%) was the silica precursor. Tetraethyl ammonium hydroxide (or TEAOH, Aldrich, 35 wt. % in
water) was used as the alkalizing agent to catalyze gelation. Ethanol (Aldrich, 99.8%) was used as
the solvent for the synthesis and as reactant during catalytic testing.
5

Thermal treatment of the dried gel was performed in a 35 mL PTFE-lined autoclave from the
Parr Instrument Company. Calcination under air flow was done in a quartz tubular reactor and
under static air in a muffled oven.
2.2. Characterization
Catalyst structures were characterized with nitrogen physisorption experiments at -196 °C using
a Micromeritics Tristar II instrument. Prior to analysis, 50–200 mg of catalyst were outgassed
under vacuum at 150 °C for 6 hours. Specific surface area (SBET) was calculated with the Brunauer–
Emmett–Teller (BET) method. The Barret-Joyner-Halenda model was used to calculate the pore
diameter (Dp) distribution using the desorption isotherm.
Transmission electron microscopy (TEM) was used to characterize the microstructure of
selected Zn-Ta-TUD-1 samples with a FEI Tecnai G2 transmission electron microscope operated
at 200 kV.
2.3. Catalytic testing
Ethanol conversion was performed with a Multi-R® apparatus from Teamcat Solutions SAS [36],
which is a high-throughput equipment for heterogeneous catalyst screening. Four glass reactors
can be used simultaneously, with the gaseous feed being calibrated to ensure an equal inlet flow
using a splitter; the reactor outputs were analyzed with an online Agilent 7890 A equipped with an
FID detector. An independently controlled valve enables selecting the output of each reactor for
analysis.
Catalyst testing was performed at 350 °C and a pressure of 1 atm. Each catalyst was ground and
sieved to 120 mesh granules, 30 mg of which were loaded in glass reactors and kept in place with
SiC. To feed the reactors with ethanol, He was used as a carrier gas. It was passed through a bubbler
containing ≥99.8 % ethanol, set at pressure and temperature to afford vapor concentration of 4.5%
6

according to the Antoine’s law. Weighted hourly space velocity of ethanol (WHSVEtOH) was set
to5.3 h-1 by adjusting the inlet flow and catalyst mass.
Ethanol conversion (X, %), the selectivity towards each product (Si, %), the molar yield of each
product (Yi, %) and the productivity in butadiene (PBD, gBD·gcat-1·h-1) were used to describe catalytic
activity—equation 1, 2, 3,and 4 respectively, where ci represents the number of carbon moles
measured for a given compound i. These values were recorded after 1 hour on stream, after initial
stabilization of the reactor output. The carbon balance (CB) for each test was calculated by dividing
the sum of carbon moles detected with the molar amount of carbon introduced as ethanol and found
to range between 95 – 105 %.
− c
X= c
EtOH,out ∙100
(1)
(2)
EtOH,in
c
EtOH,in
c
i,out
Si= c
∙100
EtOH,out
− c
EtOH,in
Yi=푋∙Si
(3)
(4)
PBD=X∙SBD∙푊퐻푆푉퐸푡푂퐻∙0.587/100
2.4. General Zn-Ta-TUD-1 synthesis
The default TUD-1 preparation method was inspired by the work of Pescarmona et al.
[29,37,38]. However, we substituted 2-propanol—the original solvent—by ethanol as the former
failed to adequately dissolve some metal precursors. In a typical synthesis (see Error! Reference
source not found.), 1.741 g of TEOS and the metal precursors, i.e. 0.067 g of TaCl5 and 0.257 g
of Zn Zn(NO₃)₂·6H₂O, were added to 30 mL of ethanol under vigorous stirring at room
temperature. After obtaining a clear solution, the chelating agent was added dropwise while
stirring; if TEAH3 was used, it was first dissolved in water with 1:11 molar ratio; a typical synthesis
used 1.741 g of TEAH3. The mixture was left to stir for 1 hour, resulting in a clear solution. TEAOH,
7

35 wt.% in water (i.e. 1.767 g of it) was added dropwise to the clear solution under vigorous stirring.
During this step, the solution quickly became white and opaque, before returning to a clear,
colorless solution, which was further stirred for 2 hours. This sol was left to age for 24 hours,
resulting in gelation. The obtained gel was dried overnight at 100 °C, resulting in a solid,
transparent xerogel with varying shades of dark orange. It was gently ground to a fine powder and
placed in a Teflon-lined autoclave for a thermal treatment at 180 °C during 6 to 48 hours. The
ensuing solid—a sticky power reminiscent of brown sugar—was calcined at 600 °C for 10 hours.
2.5. Plackett-Burman screening study
XLstat, an add-on for the Microsoft Excel® software, was used to generate the Plackett-Burman
design used for studying the effects of the synthesis parameters on the properties and activity of
Zn-Ta-TUD-1 and analyze the responses obtained experimentally (Table 1). XLstat can model the
effect of each parameter (also known as variable or factor) of a given response by fitting a first-
order polynomial function of the studied parameters (equation 5) with the experimental response.
푘
∑
푌 = 훽0 + 푖=1 훽푖 ∙ 푋푖 + 휀
(5)
푗
where Yj is the fitted response, β0 the model intercept, βi is the linear coefficient of independent
variable i with Xi its level, k the number of involved variables, and ε the residual error. Equation 5
was solved using the least square method, which is a multiple regression technique that fits
mathematical models to experimental data by minimizing the value of residuals between
experimental and fitted responses. Quality of fit and model significance were established by the
coefficient of determination (R2) and Fischer’s F-test, respectively. The obtained statistical results
showed the medialization to be statistically acceptable for further study.
The effect of each variable was judged according to their statistical significance, which was
assessed with an analysis of variance (ANOVA) performed with XLstat. For each response, this
8

required transforming three of the eleven variables into ‘dummy’ variables to reach the minimum
variable-to-observation ratio required for statistical analysis. Variables were considered ‘dummy’
when the contribution of their coefficient to the response model was less than 1%. The ANOVA
afforded standardized main effects of variables, which are t-statistics that test the null hypothesis,
e.g., that the effect of a variable on the response is 0.
BD productivity (YBD) was chosen as the first response to model due to its industrial importance
[2,39]. BET specific surface (YSBET) and average pore diameter (YDp) were selected as responses
due to their importance as morphological properties of catalyst carriers. The choice of synthesis
variables was based on the literature concerning both the Lebedev process and TUD-1 catalysts, as
well as preliminary experiments (not shown). The Zn-to-Ta (Zn:Ta) and total Si-to-metal molar
ratios (Si:M) in the precursor gel were selected due to the reported importance of balanced active
phases in catalysts for the Lebedev process [10,13,16,40–42]. The nature of the metal precursors,
was reported as influential on TUD-1 morphology [22], but also on activity in the Lebedev
process.[43] In this case, zinc chloride and zinc acetate hydrate were selected as levels for the zinc
precursor parameter (ZnPr). Tantalum chloride and tantalum ethoxide were chosen as tantalum
precursors (TaPr). Thermal treatment (ThTr) duration is reported as an important TUD-1 synthesis
parameter because of its influence on morphology [23–25]. The TEAOH-to-Si mole ratio in the
precursor gel (Alk:Si), the type of chelating agent (ChAg) and the choice of calcination method
(CalcM) were selected due to the ambiguity in the literature regarding their influence. For instance,
TEAOH is described as optional [18], yet is used in most publications, without an optimal ratio
being reported [22,25,26]. The calcination temperature ramp (CalcR) and the need for a drop-wise
addition of TEAOH under vigorous stirring (StiDW) were investigated as potential time-saving
measures. The order by which the chelating agent was added to the precursor solution (ChOrd)
with regards to the metal precursor was also investigated out of curiosity. Fig. 2 illustrates the Zn-
9

Ta-TUD-1 synthesis methods used in the PB experiment, as well as the different levels of all
parameters with the exception Zn:Ta and Alk:Si.
Choosing the two levels of each factor, represented by + and – in Table 1, was largely a matter
of preliminary experimentation with the TUD-1 and the result of our unpublished screening study
previously mentioned. Table 2 lists the levels of each variable investigated in the Zn-Ta-TUD-1
synthesis; Fig. 2 illustrates these levels in relation to the Zn-Ta-TUD-1 preparation procedure.
Experiments were performed in a random order generated by the XLStat software to minimize
errors and biases.
2.6. Response surface methodology
RSM is a technique that encompasses multi-variant experimental design, statistical modelling
and process optimization. It is generally performed in three steps: (1) DOE, (2) response surface
modelling through regression and (3) optimization of the response [44]. The XLstat software was
used for all three steps. RSM was used to optimize the productivity in BD (YPDB) by establishing
its relationship to two independent variables: Zn and Ta molar content in Zn-Ta-TUD-1, Zn mol.%
and Ta mol.% respectively. The variables were selected after the PB screening study showed that
the Zn:Ta molar ratio had a significant effect on the activity of the catalyst. In addition, the Zn-Ta-
TUD-1 preparation method used corresponded to the best performing procedure identified by
screening, which was equivalent to that used for sample PB12 in Table 1.
For a single-response, two-variable experiment, a three-level full factorial design was found
suitable, as it did not require many experiments, yet provided a reasonable amount of information
[35]. The three levels used were symbolized by -1, 0, 1. Table 3 lists the experimental design, the
corresponding experimental values for each level and the YPBD response obtained via catalytic
testing. Like above, experiments were performed in random order to minimize errors and biases
via the XLstat software.
10

Response surface modelling was performed by an empirical quadratic model of the response
(Equation 6) to the experimental data using the least square root method.
푘
푘
2
푘
∑
∑
∑
푌 = 훽0 + 푖=1 훽푖 ∙ 푥푖 + 푖=1 훽푖푖 ∙ 푥푖 + 푖<푗 훽푖푗 ∙ 푥푖 ∙ 푥푗 + 휀
(6)
푗
where Yj is the fitted response, β0 the model intercept, βi is the linear coefficient of independent
variable i with xi its input factor, and k the number of involved variables, βi is the quadratic
coefficient of variable i, βij is the linear interaction coefficient between variable i and j, and ε the
residual error. Goodness of fit of the model was evaluated with R2 and its significance with
Fischer’s F-test. Contrarily to the modelling used in the PB experiment, the introduction of second-
order terms allows the study of variable interaction effects. The relevance of each coefficient was
judged according to the t-statistics resulting from an ANOVA.
Optimization, e.g., finding the variable level providing the theoretical maximum response, was
performed using the method of steepest ascent, which is available due to the model being limited
to a single response [31].
3. Results and discussion
3.1. Plackett-Burman screening
3.1.1. Statistical interpretation
The coded value of experimental points representing the variables of the Zn-Ta-TUD-1
synthesis and the corresponding responses are listed in Table 1. For each response—BD
productivity, BET surface area and average pore diameter—a first-order polynomial equation was
generated and fitted to the experimental data. Accuracy of fit and F-test results (Error! Reference
source not found.) indicate that all three models explain >94% of the response variation and are
overall significant at 95% confidence level. An association test of the studied responses with
11

Pearson-type correlation was performed at 95% confidence level. The correlation matrix can be
found in Table 4.
The calculated linear coefficient βi of every independent variable i can be used to estimate their
influence on each response. A more rigorous interpretation considers the standardized main effects,
which are the t-values of variable effects computed with the ANOVA of the models [32]. Two
criteria were used to judge the importance of each variable: the t-value limit at confidence level of
95% (α = 0.05) and the Bonferroni limit, which tests the null hypothesis at more conservative
confidence level [32]. Factors with standardized effects above the t-value limit were interpreted as
likely to be significant; above the Bonferroni limit, variables were considered significant [32,45].
Below the t-value limit, variables were deemed unlikely to be significant.
Pareto charts of standardized effects are simple bar charts, but a useful visualization tool to
quickly interpret the results of factorial screening studies through. By plotting the t-value and
Bonferroni limits, the significant of each variable of the Zn-Ta-TUD-1 synthesis can be easily
assessed. The length of each bar also indicates the relative weight of each variable. The effects
each experimental level of the synthesis parameters had on the responses were also considered. For
a given response, dashed bars indicate that the low level (–) of the parameter afforded the greater
response value comparatively. Contrarily, dash-less bars indicate that the high level (+) gave a
higher response. Pareto charts of standardized effects of Zn-Ta-TUD-1 synthesis on BD
productivity, SBET and Dp are illustrated in Fig. 3.
3.1.2. Zn-Ta-TUD-1 morphology
The results of N2 porosimetry with the catalysts prepared according to the PB design are listed
in Table 1. These confirm the formation of mesoporous materials with high surface area. As Table
1 indicates, BET specific surface area (YBET) ranged between 228 and 747 m2g-1 and average pore
12

diameter (YDp) varied between 3.0 and 25.1 nm. This degree of irregularity in terms of
morphological properties is consistent with the high tunability of TUD-1 materials.
In the original paper introducing the TUD-1 synthesis procedure, Jansen et al. explained how
the mesoporous morphology could be tuned [18]. By adjusting the thermal treatment duration of
the silica xerogel, pore diameter and specific surface area could be modified, with the value of each
characteristic being inversely proportional to one another as a function of time—lengthening
treatment time reducing specific surface area and increasing mesopore size. Similar observations
were made with metal-containing TUD-1 materials when time was the only synthesis variable [24].
The xerogel is an organic-inorganic hybrid in which the chelating agent and its metal complexes
are homogeneously dispersed [24]. Upon heating, silica particles grow and organic species
agglomerate, shaping the mesoporous framework by steric hindrance. In theory, lengthening the
heating period promotes the organic agglomeration [46], resulting in larger, but fewer agglomerates
for silica to condense around. The morphological consequence of this phenomenon is larger pores,
but a reduced specific surface area. This trade-off between the two morphological properties as a
result of thermal treatment time is well established [18,23,46].
Surprisingly, the statistical analysis of the effects exerted by the 11 variables of the Zn-Ta-
TUD-1 synthesis under study (Fig. 3) found thermal treatment time not to influence BET specific
surface area or the average mesopore size. Nevertheless, the association test (Table 4) indicated a
strong inverse correlation between the two morphological properties. In other words, the trade-off
between surface area and pore size typical of TUD-1 still took place, but was the subject of
variables other than thermal treatment time. Fig. 5 illustrates this relationship. The type of chelating
agent and TEAOH:Si ratio in the precursor gel were identified as statistically significant variables
influencing both morphological properties. According to the literature, TEAH3 and TEG play the
identical dual role of precursor chelating and structure directing agents, with the former being the
13

predominant choice in TUD-1 synthesis.[23] However, no study could be found that directly
compared both molecules. Interestingly, TEG led to larger specific surface area and TEAH3 to
larger pores. This is consistent with the fact the latter has a larger molar volume than TEG, both
when determined empirically at 25 °C and using Connolly’s molecular surface package [47], since
equimolar amounts were used in the synthesis of Zn-Ta-TUD-1. Incidentally, greater quantities of
TEAOH—used for catalyzing the gelation process and introduce micropores within the
framework—was correlated with bigger pore diameter at the expense of surface area, although no
micropores could be detected by N2 porosimetry. The additional organic matter within the
precursor gel likely increases the size of structure-shaping agglomerates during the thermal
treatment. Consequently, the use of TEOH should be limited to gelation catalysis, as its structure-
directing properties could be fulfilled by the less expensive, safer chelating agents.
Other synthesis variable studied showed significant effect on the morphological properties of
Zn-Ta-TUD-1 (Fig. 3). However, these were not reciprocal between both responses studied.
Considering the inverse correlation observed, two possibilities main explain this discrepancy: these
variables exclusively affected one of the morphological properties independently of the other;
interaction effects between variables also influencing TUD-1 morphology could not be estimated
due to the low degree of freedom of PB designs [31]. The most important synthesis parameter
identified to only affect pore size was the TEAOH addition procedure during the sol-gel process.
Most authors indicate TUD-1 should be prepared by adding TEAOH drop-wise under vigorous
stirring and left stirring for up to two hours until a clear gel is obtained. Surprisingly, directly
pouring TEOH consistently afforded a clear colorless gel, whereas the traditional method
occasionally resulted in milky mixtures, which have been observed elsewhere.[26,46] In the sol-
gel methodology, the basic catalyst feed rate controls the silica precursor hydrolysis and
condensation kinetics; higher feed rates have been associated to faster particle growth.[48]
14

Consequently, the influence of the TEAOH addition method on the morphology of Zn-Ta-TUD-1
may be owed to the change in gelation kinetics it induces. Why this effect is statistically significant
only for the average pore diameter remains to be answered.
3.1.3. BD productivity
The catalytic performances of Zn-Ta-TUD-1 catalysts prepared according to PB design varied
significantly in terms of productivity (Table 3). In a typical test, ethanol was converted to
predominantly three products: BD, acetaldehyde and ethylene; 1–3 % yield consisted of diethyl
ether, propylene, 1-butanol and butenes. Selectivity towards the three main products depended on
the catalyst used. The best performances were achieved with PB12: its activity is depicted in Fig. 4.
As illustrated, BD selectivity reached 70%, a value comparable to many of the best catalysts found
in the literature [4]. Although BD selectivity remained stable, deactivation took place, as evidenced
by the decreasing ethanol conversion. Nevertheless, high BD productivity was achieved.
As previously indicated, several authors have associated the morphology of studied materials
and their performances in the Lebedev process [8,11,16,17]. Association tests of BD productivity
with specific surface area and average pore diameter were performed. Accordingly, a Pearson-type
correlation between BD productivity and specific surface area at 95% confidence interval was
found (Table 4); although statistically significant, it is unlikely that the correlation is linear, as a
better fit was found with a quadratic equation (Fig. 5 (b)). Similar correlations have been reported
by other scholars for this reaction [16]. In fact, it is well-established that greater surface area allows
for a better accessibility to active sites and is often considered a desirable feature of catalysts.
Contrarily to Jones et al. [8] and Palkovits et al. [17], who reported improvements in BD yield with
increasing pore size, no correlation could be found between the average pore diameter and BD
productivity on Zn-Ta-TUD-1. This can be explained by the trade-off between SBET and Dp
mention in section 3.1.2.: the benefits of greater pore size may be cancelled due to the loss in
15

specific surface area, suggesting the latter to be the most important morphological property of the
two for maximization BD formation. Consequently, it is unsurprising that two of the most
influential factors on SBET, the nature of the chelating agent and the calcination method, were also
statistically significant on BD productivity, as depicted by Fig. 3 (c).
The only factor with no impact on TUD-1 morphology, but significantly influential on BD
productivity was the Zn-to-Ta molar ratio. Zinc oxide is well-established for catalyzing the
dehydrogenation of ethanol and tantalum oxide can perform the conversion of ethanol-
acetaldehyde mixtures of BD. However, many authors have reported that a subtle balance must be
struck between the dehydrogenating and condensation promoters, as the active sites are also known
to catalyze undesirable side-reaction. This theory is given statistical evidence through the results
of our PB screening. The fact that the Zn-to-Ta molar ratio was statistically insignificant on
synthesis procedure with regards to the resulting morphological properties further indicates that it
is solely attributable to the chemical properties of Zn-Ta-TUD-1.
Further increase in BD productivity proceeded by tuning the synthesis of Zn-Ta-TUD-1. Of all
significant preparation parameters identified by the PB design experiment, the Zn-to-Ta ratio was
selected for the RSM experiment. To accommodate a two-variable design, Zn-to-Ta was split into
the molar amount of each element, thereby providing information of the effect of low metal content.
All other variables were set to their low-level setting, as Fig. 3 shows them to improve BD
productivity. Incidentally, this corresponds to the procedure used to synthesize PB12, except for
the ratio and amount of Zn and Ta.
3.2. Response surface methodology
3.2.1 Statistical interpretation
16

The obtained response listed in Table 3 was correlated with independent variables Zn mol.%
and Ta mol.% using the quadratic equation, Eq. (6). The least square regression method was used
to fit the experimental data to Eq. (6), resulting in the model below:
푌
푃퐵퐷
= 1.191 + 0.146 ∙ 푋1 + 0.156 ∙ 푋2 + 0.059 ∙ 푋12 − 0.366 ∙ 푋2 − 0.031 ∙ 푋1 ∙ 푋2 (7)
2
where X1 is Zn mol.% and X2 is Ta mol.%. Validity of the model was tested through statistical
means (Error! Reference source not found.Error! Reference source not found.). The
coefficient of determination and its adjusted form, 0.971 and 0.924, respectively, showed that the
experimental results were well represented by the model. ANOVA of the model indicated an F-
value of 20.347 and a p-value below 0.05; theses statistical results demonstrated the significance
and adequacy of the model.
The importance of each factor on the response (BD productivity) was assessed by comparing
their standardized effect to the minimum t-value at 95% confidence interval. The Pareto chart
depicted in Fig. 6 reveal the most important factors. The main effect of Zn and Ta content were
found to be important, naturally suggesting both elements contribute to the catalytic activity of Zn-
Ta-TUD-1. However, no interaction effect could be discerned between the two variables; this
implies Zn and Ta—although both required for forming BD—do not have a synergy effect that can
be discerned using our quadratic model. Only the squared effect of Ta loading was significant, but
also negative. This can be interpreted as a non-linear detrimental effect of Ta mol.% on BD
productivity.
The two-dimensional contour plot of BD productivity corroborated with Zn and Ta loadings is
shown in Fig. 7; it is the visual representation of the quadratic response model, Eq. 7. A noticeable
plateau effect with regards to the Ta loading can be deduced from its shape [35], reflecting the
squared negative effect noted above. A linear relation between BD productivity with Zn content
17

within the experimental region can also be observed. The method of steeped ascent indicated BD
productivity can be maximized with a catalyst containing 3 mol.% of Zn and 2.2 mol.% of Ta.
However, the elliptical shape of the response maxima suggests the true optimal value to be outside
the experimental region with regards to Zn content. Incidentally, PB12—synthesized for the
screening experiment with a loading of 4 mol.% Zn and 2.1 mol.% Ta—showed a BD productivity
of 1.60 gBD·gcat-1·h-1. A Zn-Ta-TUD-1 catalyst with 6 mol.% and 2.2 mol.% of Zn and Ta was
synthesized using the same methodology to further test the influence of Zn. BD productivity
dropped to 0.86 gBD·gcat-1·h-1, with a noticeable gain in acetaldehyde selectivity (not shown). The
resulting curve of PB selectivity versus Zn mol.% at fixed Ta content the RSM indicated the
optimal Zn-Ta-TUD-1 catalyst should have a Zn content between 3 and 4 mol.%. Ta content
between 2 and 2.2 mol.% was found optimal with the method of steepest ascent. This amounts to
a Zn-to-Ta ratio between 1.5 and 2.
3.2.2. RSM series characterization
The TUD-1 preparation has been described as an easy way to homogeneously disperse metals
within a mesoporous silica framework [23]. Optimization of Zn-Ta-TUD-1 synthesis to maximize
its activity in the Lebedev process afforded highly active materials. To verify that the materials
prepared were comparable to those found in the literature, thereby confirming the success of the
synthesis method used, characterization was performed.
N2 porosimetry results (Table S3) indicated the final Zn-Ta-TUD-1 method afforded materials
with an average BET surface area of 661±41 m2∙g-1, indicative of its repeatability. Average pore
size diameter of 9.8±1.5 nm was obtained, with an outlier at 7.0 nm. Interestingly, no correlation
between BET surface area and activity could be observed. This suggested the metal content
becomes the predominant factor once specific area is large enough, e.g., ≥600 m2∙g-1at which point
this morphological property appears to no longer be an issue.
18

SEM images typical of samples prepared during the RSM experiments are shown in Fig. 8. The
results are similar to those reported in the literature for M-TUD-1 at low metal loading
[25,30,38,49–51]. Zn-Ta-TUD-1 consisted of <100 μm particles apparently without a well-defined
morphology. At high magnification, the catalyst surface is shown to be rough and irregular, typical
of the sponge-like morphology resulting from the agglomeration of silica particles formed during
the synthesis procedure [23,24,52].
Fig. 9 illustrates HR-TEM images of RSM series Zn-Ta-TUD-1 catalysts. Inspection of various
samples confirmed the sponge-like 3D structure with “worm-like” pores characteristic of TUD-1
materials [23,26]. The absence of discernable metal oxide nanoparticles suggests Zn and Ta were
completely isolated within the carrier framework. Their presence was confirmed by energy-
dispersive X-ray spectroscopy. Furthermore, nanoparticles could be detected upon electron
irradiation of the samples, which provoked the degradation of silica and metal oxide agglomeration
(Fig. 9, right, and Error! Reference source not found.) [53,54].
3 Conclusion
The effect of various parameters in the synthesis of Zn-Ta-TUD-1 materials on their
morphology and ability to convert ethanol to BD was studied using designs of experiments. A
Placket-Burman screening design coupled with mathematical modelling and statistical tools
identified the most important preparation variables for attaining high BD productivity and
understanding their effect on surface area and pore size. Response surface methodology was used
to optimize BD productivity by tuning the Zn and Ta content of catalysts prepared according to the
most suitable procedure resulting from the screening study.
We found the nature of the chelating agent to play a statistically significant role on the
morphology of Zn-Ta-TUD-1. Use of TEG—a sterically smaller molecule—resulted in larger
surface area and smaller average pore diameter than TEAH3. There existed a trade-off situation
19

between the two structural properties depending on the agent used and the total amount of organic
species present in the precursor gel. Ostensibly, the difference manifests itself during the structure
shaping process taking place under thermal treatment. Choosing a favorable chelating agent may
be an alternative to tuning the thermal treatment duration for obtaining desirable morphologies, the
common practice with TUD-1 material. New chelating agents and their effect should also be
investigated.
Substituting the drop-wise addition under stirring of TEAOH for rapid pouring influenced pore
size, likely due to changes in the gelation kinetics. It showed great reproducibility in obtaining
materials with large surface area (≥600 m2·g-1) and mesopores diameters averaging 10.5 nm. In
practical terms, this finding enables time saving during the synthesis. However, a more thorough
study of the gelation kinetics with better controlled alkalizing agents addition rates is advised.
Besides the chelating agent, high BD productivity required a balanced Zn:Ta ratio and
calcination of the samples under air. RSM optimization of Zn and Ta loadings further indicated the
optimal content of Ta was between 2 and 2.2 mol.%. Maximum BD productivity required a Zn
content between 3 and 4 mol.%. Despite finding no mathematical evidence of interaction between
the amount of Zn and Ta, the results highlight the need for a balanced quantity of each element for
maximizing BD production. This observation coincides with other findings of the literature which
concluded that the multi-step reaction of the Lebedev process requires catalysts with balanced
properties, often obtaining by tuning their different components [10,16,40].
The butadiene productivity of catalyst RSM-8 prepared via the improved synthesis method can
be compared with that Zn-Ta-TUD-1 from our previous work—synthesized by conventional
method with unoptimized metal content [19]. At 400 °C and WHSVEtOH of 5.3 h-1, RMS-8 achieves
a comparable BD productivity of 2.18 gBD·gcat-1·h-1 after 3 hours on stream, despite possessing 53%
of the total metal content (3 and 2 mol.% vs. 6.1 and 3.4 mol% of Zn and Ta, respectively).
20

Furthermore, RSM-8 displayed a better resistance to deactivation over a 20 h period, decreasing
by 8 percentage points, compared to 16 percentage points for the catalyst of our previous work (Fig.
S3). Consequently, we conclude that the design of experiment approach successfully improved the
synthesis method for preparing Zn-Ta-TUD-1 materials highly active in the Lebedev process.
Guillaume Pomalaza: Investigation, Methodology, Formal analysis, Writing – Original Draft. Mickaël
Capron: Supervision, Funding Acquisition. Franck Dumeignil: Supervision, Writing – Review & Editing,
Funding Acquisition, Validation.
4 Acknowledgements
Authors acknowledge the support from the French National Research Agency (ANR-15-CE07-
0018-01). Chevreul Institute (FR 2638), Ministère de l’Enseignement Supérieur, de la Recherche
et de l’Innovation, Région Hauts-de-France and FEDER are acknowledged for supporting and
funding partially this work.
21

References
[1] E. V Makshina, M. Dusselier, W. Janssens, J. Degrève, P.A. Jacobs, B.F. Sels, Chem. Soc.
Rev. 43 (2014) 7917–7953.
[2] D. Cespi, F. Passarini, I. Vassura, F. Cavani, Green Chem. 18 (2016) 1625–1638.
[3] S. Farzad, M.A. Mandegari, J.F. Görgens, Bioresour. Technol. 239 (2017) 37–48.
[4] G. Pomalaza, M. Capron, V. Ordomsky, F. Dumeignil, Catalysts 6 (2016) 203.
[5] H.N. Sun, J.P. Wristers, in: Kirk-Othmer Encycl. Chem. Technol., John Wiley & Sons, Inc.,
Hoboken, NJ, USA, 2002.
[6] M. Dahlmann, J. Grub, E. Löser, in: Ullmann’s Encycl. Ind. Chem., Wiley-VCH Verlag
GmbH & Co. KGaA, Weinheim, Germany, 2011, pp. 1–24.
[7] C. Angelici, M.E.Z. Velthoen, B.M. Weckhuysen, P.C.A. Bruijnincx, ChemSusChem 7
(2014) 2505–2515.
[8] M. Jones, C. Keir, C. Iulio, R. Robertson, C. Williams, D. Apperley, Catal. Sci. Technol. 1
(2011) 267.
[9] G.O. Ezinkwo, V.P. Tretyakov, A. Aliyu, A.M. Ilolov, ChemBioEng Rev. 1 (2014) 194–
203.
[10] C. Angelici, M.E.Z. Velthoen, B.M. Weckhuysen, P.C.A. Bruijnincx, Catal. Sci. Technol. 5
(2015) 2869–2879.
[11] T.W. Kim, J.W. Kim, S.Y. Kim, H.J. Chae, J.R. Kim, S.Y. Jeong, C.U. Kim, Chem. Eng. J.
278 (2014) 217–223.
[12] J.L. Cheong, Y. Shao, S.J.R. Tan, X. Li, Y. Zhang, S.S. Lee, ACS Sustain. Chem. Eng. 4
(2016) 4887–4894.
[13] V.L. Sushkevich, I.I. Ivanova, V. V. Ordomsky, E. Taarning, ChemSusChem (2014) 2527–
22

2536.
[14] A. Tripathi, K. Faungnawakij, A. Laobuthee, S. Assabumrungrat, N. Laosiripojna, Int. J.
Chem. React. Eng. 14 (2016) 945–954.
[15] V.L. Dagle, M.D. Flake, T.L. Lemmon, J.S. Lopez, L. Kovarik, R.A. Dagle, Appl. Catal. B
Environ. 236 (2018) 576–587.
[16] S. Da Ros, M.D. Jones, D. Mattia, J.C. Pinto, M. Schwaab, F.B. Noronha, S.A. Kondrat,
T.C. Clarke, S.H. Taylor, ChemCatChem 8 (2016) 2376–2386.
[17] A. Klein, R. Palkovits, Catal. Commun. 91 (2016) 72–75.
[18] J.C. Jansen, Z. Shan, L. Marchese, W. Zhou, N. v d Puil, T. Maschmeyer, Chem. Commun.
(2001) 713–714.
[19] G. Pomalaza, G. Vofo, M. Capron, F. Dumeignil, Green Chem. 20 (2018) 3203–3209.
[20] X. Huang, Y. Men, J. Wang, W. An, Y. Wang, Catal. Sci. Technol. 7 (2017) 168–180.
[21] W. Dai, S. Zhang, Z. Yu, T. Yan, G. Wu, N. Guan, L. Li, ACS Catal. 7 (2017) 3703–3706.
[22] A. Ramanathan, M. Carmen Castro Villalobos, C. Kwakernaak, S. Telalovic, U. Hanefeld,
Chem. - A Eur. J. 14 (2008) 961–972.
[23] S. Telalović, A. Ramanathan, G. Mul, U. Hanefeld, J. Mater. Chem. 20 (2010) 642.
[24] M.S.H.M. Saad, Functionalized TUD-1ꢀ: Synthesis , Characterization and (Photo-) Catalytic
Performance, Universiteit Van Hewan, 2005.
[25] S. Lima, M.M. Antunes, A. Fernandes, M. Pillinger, M.F. Ribeiro, A.A. Valente, Molecules
15 (2010) 3863–3877.
[26] A. Ranoux, K. Djanashvili, I.W.C.E. Arends, U. Hanefeld, RSC Adv. 3 (2013) 21524–21534.
[27] G. Imran, M.P. Pachamuthu, R. Maheswari, A. Ramanathan, S.J. Sardhar Basha, J. Porous
Mater. 19 (2012) 677–682.
[28] B. Karmakar, A. Sinhamahapatra, A.B. Panda, J. Banerji, B. Chowdhury, Appl. Catal. A
23

Gen. 392 (2011) 111–117.
[29] L. Li, D. Cani, P.P. Pescarmona, Inorganica Chim. Acta 431 (2015) 289–296.
[30] M.S. Hamdy, O. Berg, J.C. Jansen, T. Maschmeyer, J.A. Moulijn, G. Mul, Chem. - A Eur.
J. 12 (2006) 620–628.
[31] M. Natrella, NIST/SEMATECH e-Handbook of Statistical Methods, NIST/SEMATECH,
2010.
[32] J. Madinger, P.J. Whitcomb, DOE Simplified, Productivity Press, 2017.
[33] M. Zougagh, P.C. Rudner, A.G. De Torres, J.M. Cano Pavôn, J. Anal. At. Spectrom. 15
(2000) 1589–1594.
[34] T. Lundstedt, E. Seifert, L. Abramo, B. Thelin, Å. Nyström, J. Pettersen, R. Bergman,
Chemom. Intell. Lab. Syst. 42 (1998) 3–40.
[35] M.A. Bezerra, R.E. Santelli, E.P. Oliveira, L.S. Villar, L.A. Escaleira, Talanta 76 (2008)
965–977.
[36] F. Dumeignil, S. PAUL, L. Duhamel, J. FAYE, P. Miquel, M. CAPRON, J.L. Dubois,
Dispositif d’évaluation d’au Moins Un Critère de Performance de Catalyseurs Hétérogènes,
WO2015118263 A1, 2015.
[37] L. Li, T.I. Korányi, B.F. Sels, P.P. Pescarmona, Green Chem. 14 (2012) 1611.
[38] I.C. Neves, G. Botelho, A. V. Machado, P. Rebelo, S. Ramôa, M.F.R. Pereira, A.
Ramanathan, P. Pescarmona, Polym. Degrad. Stab. 92 (2007) 1513–1519.
[39] E. V. Makshina, W. Janssens, B.F. Sels, P.A. Jacobs, Catal. Today 198 (2012) 338–344.
[40] S.-H. Chung, C. Angelici, S.O.M. Hinterding, M. Weingarth, M. Baldus, K. Houben, B.M.
Weckhuysen, P.C.A. Bruijnincx, ACS Catal. 6 (2016) 4034–4045.
[41] O. V. Larina, P.I. Kyriienko, S.O. Soloviev, Theor. Exp. Chem. 52 (2016) 51–56.
[42] P.I. Kyriienko, O. V. Larina, S.O. Soloviev, S.M. Orlyk, C. Calers, S. Dzwigaj, ACS Sustain.
24

Chem. Eng. 5 (2017) 2075–2083.
[43] T. De Baerdemaeker, M. Feyen, U. Müller, B. Yilmaz, F.S. Xiao, W. Zhang, T. Yokoi, X.
Bao, H. Gies, D.E. De Vos, ACS Catal. 5 (2015) 3393–3397.
[44] P.N. Panahi, D. Salari, A. Niaei, S.M. Mousavi, J. Ind. Eng. Chem. 19 (2013) 1793–1799.
[45] A. Hesari, P. Mohammad, A.H. Fererdoon, A. Mehdi, Iran. J. Chem. Chem. Eng. 35 (2016)
51–62.
[46] Z. Shan, J.C. Jansen, W. Zhou, T. Maschmeyer, Appl. Catal. A Gen. 254 (2003) 339–343.
[47] M.L. Connolly, M.P.W. Shell, 11 (1993) 139–141.
[48] I.A. Rahman, P. Vejayakumaran, C.S. Sipaut, J. Ismail, M.A. Bakar, R. Adnan, C.K. Chee,
Colloids Surfaces A Physicochem. Eng. Asp. 294 (2007) 102–110.
[49] M.P. Pachamuthu, V.V. Srinivasan, R. Maheswari, K. Shanthi, A. Ramanathan, Appl. Catal.
A Gen. 462–463 (2013) 143–149.
[50] W. Yan, A. Ramanathan, P.D. Patel, S.K. Maiti, B.B. Laird, W.H. Thompson, B.
Subramaniam, J. Catal. 336 (2016) 75–84.
[51] M.P. Pachamuthu, K. Shanthi, R. Luque, A. Ramanathan, Green Chem. 15 (2013) 2158.
[52] A. Ranoux, K. Djanashvili, I.W.C.E. Arends, U. Hanefeld, RSC Adv. 3 (2013) 21524.
[53] N. Jiang, Reports Prog. Phys. 79 (2016) 016501.
[54] B. Martin, O.W. Flörke, E. Kainka, R. Wirth, Phys. Chem. Miner. 23 (1996) 409–417.
25

5 ANNEX - FIGURES
Fig. 1 Generally accepted mechanism for the conversion of ethanol to 1,3-butadiene. Reaction
steps: (a) ethanol dehydrogenation; (b) aldol condensation of acetaldehyde; (c) acetaldol
Table 1 Plackett-Burman experimental design used for studying the main effects of Zn-Ta-TUD-1
synthesis variables.
Variable
Responses
Run no. Cat. name
Zn:Ta Si:M ThTr TaPr ZnPr ChAg Alk:Si ChOrd StiDW CalcM CalcR YPBD YSBET YDp
1
PB1
PB2
PB3
PB4
PB5
PB6
PB7
PB8
PB9
PB10
PB11
PB12
+
-
+
-
-
-
+
+
+
-
+
-
+
+
-
+
-
-
-
+
+
+
-
-
+
-
+
+
-
+
-
-
+
+
-
+
+
-
+
-
-
+
+
+
-
+
+
-
+
-
-
-
-
-
-
-
-
+
+
+
-
+
+
-
+
-
-
-
+
-
-
-
+
+
+
-
+
+
-
0.583 341 18.7
1.089 336 25.3
0.644 424 28.3
1.139 747 3
10
11
2
6
4
5
3
7
8
+
+
-
+
-
-
-
+
+
+
-
+
+
+
-
+
+
-
+
-
-
+
+
+
-
+
+
-
+
-
-
-
+
+
+
-
+
+
-
0.748 516 7
0.753 228 25.11
0.669 401 26
0.745 505 15.9
0.640 486 12.6
0.912 601 11.9
1.181 740 6.6
1.602 739 10.7
-
+
+
-
-
+
-
9
12
-
-
+
-
-
-
-
Variable signification: Zn:Ta, the zinc-to-tantalum molar ratio; Si:M, the silica-to-total-metal ratio; ThTr, the thermal
treatment time; TaPr, nature of the Ta precursor; ZnPr, nature of the Zn precursor; ChAg, nature of the chelating
agent; Alk:Si, the TEAOH-to-Si ratio; ChOrd, the order of chelation; StiDW, Dropwise addition of TEAOH with stirring;
CalcM, calcination method; CalcR, calcination ramp.
26

Fig. 2 Preparation scheme of Zn-Ta-TUD-1; (+) and (-) signs represent the levels of the Plackett-
Burman experimental design.
Table 2 Level of variables in the Plackett-Burman experiment of Zn-Ta-TUD-1 synthesis
Variable
Zn-to-Ta ratio
Silica-to-metal ratio
Unit
n/a
n/a
Symbol
Zn:Ta
Si:M
-
2
16
+
5
8
Thermal treatment duration
Nature of Ta precursor
Nature of Zn precursor
Nature of chelating agent
TEAOH-to-Si ratio
Order of chelation addition step
Dropwise addition of TEAOH with stirring
Calcination method
Hour
n/a
n/a
n/a
n/a
n/a
n/a
n/a
°C/min
ThTr
TaPr
ZnPr
ChAg
Alk:Si
ChOrd
StiDW
CalcM
CalcR
6
24
TaCl5
ZnCl5
TEAH3
1.0
After metal
Yes
Under static air
3
Ta(EtO)5
Zn(CH3CO2)2
TEG
0.5
Before metal
No
Under air flow
1
Calcination temperature ramp
Table 3 Three-level factorial design and corresponding levels of variable for the optimization of
Zn-Ta-TUD-1 preparation to maximize butadiene productivity and the corresponding experimental
responses.
Variable
Ta mol.%
Run no.
Catalyst name
YPBD (gBDgcath-1)
Zn mol.%
27

2
5
4
9
7
8
1
6
3
RSM1
RSM2
RSM3
RSM4
RSM5
RSM6
RSM7
RSM8
RSM9
-1
0
1
-1
0
1
-1
0
1
1 %
2 %
3 %
1 %
2 %
3 %
1 %
2 %
3 %
-1
-1
-1
0
0
0
1
1
1
1 %
1 %
1 %
2 %
2 %
2 %
3 %
3 %
3 %
0.555
1.048
0.978
0.715
1.188
0.939
0.849
1.456
1.150
Table 4 Correlation matrix between the responses selected for the PB experiment.
Response
YPBD
YSBET
YPBD
1
0.689**
-0.455
YSBET
0.689**
1
YDp
-0.45
-0.828**
1
YDp
-0.828**
** indicates correlations that are statistically significant. YBD: BD productivity; YSBET: BET specific
surface area; YDp: Average pore diameter.
Fig. 3 Pareto chart of standardized effects each TUD-1 synthesis has on the selected responses: (a)
butadiene productivity; (b) BET specific surface area; (c) Average pore diameter. In comparing
28

both levels, dashed bars indicate that the low level of the parameter gave the highest response;
dash-less bars indicate the high level resulted in the highest response.
Fig. 4 Conversion and selectivity towards major products of ethanol conversion on (a) PB12 and
over time. T = 350 C, P = 1 atm, WHSVEtOH = 5.3 h-1. EtOH: ethanol. AcH: acetaldehyde. C2=:
ethylene.
Fig. 5 Relationship between BET specific surface area and (a) the average pore diameter, (b) BD
productivity of PB series of catalysts.
29