Synthesis Optimization of MCM-41 for CO2 Adsorption Using Simplex-centroid Design

According to Scripps of Oceanography Institute, from San Diego University, where the Mauna Loa station is monitored, the average concentration of CO2 in Hawaii in April of 2014 was 401.33 ppm. For the first time in human history, the concentration of CO2 in the atmosphere remained above 400 ppm during a whole month. The concentration of CO2 at the start of the industrialization period was around 280 ppm, according to the Intergovernamental Panel on Climate Change (IPCC). The new report from IPCC, from April 2014 shows that the emissions of gases from the greenhouse effect increased to unprecedented levels, despite of a series of policies to mitigate them. The emissions between 2000 and 2010 grew faster than during the previous three decades1. To decrease the environmental impact from intensified emission of CO2, it is necessary to developmore efficient removal technologies. A promising approach to reduce CO2 emission is the capture and geological storage of Carbon (CCS)2. CO2 can be captured by pre-combustion, post-combustion or oxyfuel techniques3 by capture processes including absorption, adsorption, hybrid processes such as adsorption/membrane system or cryogenic distillation. Nowadays, among a diversity of technologies to capture CO2, adsorption has been vastly applied in many industrial processes including the production of synthesis gas and hydrogen with high contents of CO2. Physical adsorption is a compelling technology applied to CO2 due to the use of low-cost adsorbents that exhibits low heat capacity, fast kinetics, high CO2 adsorption capacity in addition to thermal, Synthesis Optimization of MCM-41 for CO2 Adsorption Using Simplex-centroid Design


Introduction
According to Scripps of Oceanography Institute, from San Diego University, where the Mauna Loa station is monitored, the average concentration of CO 2 in Hawaii in April of 2014 was 401.33 ppm.For the first time in human history, the concentration of CO 2 in the atmosphere remained above 400 ppm during a whole month.The concentration of CO 2 at the start of the industrialization period was around 280 ppm, according to the Intergovernamental Panel on Climate Change (IPCC).The new report from IPCC, from April 2014 shows that the emissions of gases from the greenhouse effect increased to unprecedented levels, despite of a series of policies to mitigate them.The emissions between 2000 and 2010 grew faster than during the previous three decades 1 .
To decrease the environmental impact from intensified emission of CO 2, it is necessary to developmore efficient removal technologies.A promising approach to reduce CO 2 emission is the capture and geological storage of Carbon (CCS) 2 .CO 2 can be captured by pre-combustion, post-combustion or oxyfuel techniques 3 by capture processes including absorption, adsorption, hybrid processes such as adsorption/membrane system or cryogenic distillation.Nowadays, among a diversity of technologies to capture CO 2, adsorption has been vastly applied in many industrial processes including the production of synthesis gas and hydrogen with high contents of CO 2 .Physical adsorption is a compelling technology applied to CO 2 due to the use of low-cost adsorbents that exhibits low heat capacity, fast kinetics, high CO 2 adsorption capacity in addition to thermal, Synthesis Optimization of MCM-41 for CO 2 Adsorption Using Simplex-centroid Design chemical and mechanical stability under extensive cycling.In recent years, considerable efforts in developing new adsorbents have been reported by the scientific community aiming at optimizing the adsorption technology 4 .
Since 1990s, the use of mesoporous materials synthesized from M41S has been extensively investigated in areas including environment 5,6 catalysis 7-9 , energy 10 , and biomedicine 11 .A wide variety of reports has addressed the versatility of MCM-41 [12][13][14][15][16] which became the most popular example of the M41S family due to a combination of superior properties, such as high surface area, thermal stability and porous volume that can vary according to the surfactant used.In addition, average pore sizes can vary between 2 and 50 nm 17,18 .These features yield fast mass transport 19 , making this material appealing to adsorption processes 9 .The high adsorption capacity of mesoporous silica for CO 2 , CH 4 , N 2 , H 2 and O 2 5 has been reported along with the possibility of adjusting the mesoporous 9 improving CO 2 adsorption.
The experimental design of component mixtures is an essential tool to optimize production at reduced costs.Both the time and number of experimentsmay be reduced to obtain reliable results.To study the effect of mixture compositions on a variable outcome, one may use composition modelling.This method basically consists in solving an equation by the least square method to find results which show the behavior of a dependent variable with respect to each component of a mixture 20 .The simplex-centroid mixture design was chosen in this study since it is specially used for three-component mixtures.In this case, due to the peculiarity of ternary mixtures represented by only one mixture, planning is known as "simplex-centroid design" 21,22

Synthesis of MCM-41-type absorbents
Mesoporous materials were synthesized by the hydrothermal method using tetraethylorthosilicate (TEOS) as silica source, sodium silicate, water and three surfactants with different chain sizes as structural conductors 12  was added in a teflon autoclave and incubated for a period of 120 h at 373 K. pH was daily adjusted between 9 and 10 using acetic acid 30%.The resulting materials were filtered, washed and dried for 3 h at 373 K, followed by 2 h of calcination at 823 K initially under inert nitrogen atmosphere and subsequently air.After calcination, the materials were characterized by XRD, FTIR, BET,TG and SEM.

Characterization
The calcined samples were characterized by X-ray diffraction (XRD) in a Shimadzu XRD-6000 equipment set to 30 kV and 30 mA, with monochromatic CuKα radiation (λ = 1.5406Å).The diffraction data were scanned in the 2θ range of 1°-10° at a step size of 0.02°.Nitrogen adsorption/desorption isotherms were measured using a Micromeritics ASAP 2020 equipment.The specific surface area and porous distribution were obtained using BET and BJH methods applied to the desorption isotherm, respectively 23,24 .Thermogravimetric analyses were performed using a TA balance model Q500, under N 2 atmosphere, under a flow of 100 mL/min, between 30 and 800°C.The mass of the samples was 3.0 mg.The heating rate was 10°C min -1 .FTIR spectra were obtained in a Shimadzu IR Prestige-21 spectrophotometer, using KBr as dispersing agent.The pellets were prepared using approximately 0.7 mg of material with enough KBr to reach the concentration of 1 wt.%.The material was homogenized and hydraulically pressed under 8 ton.The absorption spectra of MCM-41 were obtained in the medium infrared range of 400-4000 cm -1 with resolution of 4 cm -1 .The morphology of the materials was observed by scanning electron microscopy (SSX-550 SHIMADZU).

Adsorption experiments
The adsorption equilibrium of CO 2 for different samples was gravimetrically measured on a Rubotherm Magnetic Suspension balance at 298 K.The methodology employed consisted of three steps: (a) activation of the sample, (b) determinationof the effect of the thrust to correct the values recorded by the balance during adsorption and (c) adsorption of CO 2 .
a) Activation of the sample The samples were activated before performing the measurements.The method consisted of flushing the sample with 50 mL/min of Helium gas and heating it up to 393 K.After the sample was sufficiently flushed and heated, it was cooled down under vacuum to measure the temperature.The activation process was continuously monitored by a magnetic suspension balance (MSB).The weight of the sample container and sample itself was measured by the MSB during activation.At the end, the mass was measured in vacuum.From this, the activated sample was calculated by subtracting the mass of the empty sample container 25 .
b) Determining the effect of the thrust After activation, the thrust effect was corrected to determine the excess amount of adsorbent gas using Equation 1.
where, m bal is the mass recorded by the MSB, ρ is the density of the atmosphere surrounding the sample and V is the volume of the weighed sample [26][27][28] .

c) CO 2 Adsorption
The samples were submitted to a flow rate of 100 mL min -1 of CO 2 until they reached equilibrium at 1 bar.This procedure was repeated for different pressures, i.e., 1, 3, 5, 10, 15, 20, 25, 30 and 40 bar, always waiting for system equilibrium at constant temperature of 25°C.The purity of He and CO 2 used in this study was above 99.9%.The absorbed mass of CO 2 in balance for each pressure was determined by subtracting the mass from the sample holder.Thus, the capacity of CO 2 adsorption was demonstrated by the relationship of the absorbed mass of absorbent with respect to the range of analyzed pressure 25,[29][30][31] .

Use of statistical tools to evaluate CO 2 adsorption
The adsorption of CO 2 using different mesoporous materials was evaluated using the simplex-centroid method (Statistic Software.v. 7.0, Tulsa, OK, USA, 2004).Contrary to the factorial experiment planning, where the variables are independent, using mixture planning, components represent ratios instead of independence.These ratios cannot be negative and, if they are presented as fractions of mixture, they should sum up to unit.Such conditions reduce the spatial representation of mixtures, denominated 'simplex', to the dimension of (q -1), where q is the number of components.Therefore, q = 2 corresponds to a straight line, q = 3 an equilateral triangle, q -1 a tetrahedron and values of q > 4 are impossible to represent in the three-dimensional space.This limitation is a consequence of the interdependence of components (x i ) from the mixture given by Equation 2.
Therefore, the geometrical representation of the experimental space of "q" components consists of all sides, e.g., vertex, edges, and faces of a regular picture of dimension (q -1).For a mixture of three components, q = 3, the simplex is an equilateral triangle (Figure 1), formed by a diagonal plain contained in a cube where the addition of proportions x 1 , x 2 and x 3 in any side of the triangle is equal to 1.
The sides located above the triangle vertex correspond to pure components: C 17 (  32,33 .The model was adjusted using combinations of component ratios chosen to set the experimental planning.The "simplex-centroid" model developed by Scheffé 34 yields responses of all subsets of components in the mixture where they appear in equal proportions.A number of (2 4 -1) observations are used to reduce the number of experiments and still adjust an especial cubic model.In order to obtain regression equations, the model must be adjusted to experimental data.Experiments involving response models with independent variables have shown that Y can be estimated using polynomials derived from Taylor series 35 .Mixture modelling is common in simple computations.Linear, quadratic and especial cubic models are described by the following Equations 3-5, respectively: Where βi are interaction coefficients of first, second, and third orders, calculated for x factors x, and y is the experimental response 36,37 .In a mixture of 3 components to be optimized, those models should be gradually expanded, taking into consideration the effect of three pure components of the linear model, binary interactions, and an interaction of the three factors of the quadratic model and special cubic one.
The planning which allows to adjust the special cubic model without additional experiments of varied components is the simplex-centroid 34,38 and, for this reason, it is the most used one.According to this planning for 3 components, e.  2).The response of the model was CO 2 adsorption (Table 1).The necessary calculations to build and evaluate the models were performed using the least square solution of the Statistic Software.v. 7.0 computer program.110) and (200) of the hexagonal mesoporous phase.It can also be observed that the material which contains a mixture of surfactants in the pores showed higher intensity peaks, especially C 17 C 19.It is interesting to notice that the surfactant of longest hydrophobic chain, DTMA + , used in the synthesis of sample C 21 was the one which presented the lowest peak intensity.Selvam and co-authors reported that surfactants with high molecular weight (≥C 18 ) are difficult to solubilize, which can explain the behavior reported for sample C 21 13 .

BET and BJH analyses
The physico-chemical parameters of samples C  2. The specific surface area was determined according to the BET method whereas the distribution of pores was estimated by BJH algorithm.It can be observed that sample C 17 C 19 presented the highest crystallinity observed (DRX, Figure 3) along with highest surface area and pore volume.
The adsorption/desorption isotherms for N 2 and the pore size distribution of calcined samples can be seen in Figure 4.According to IUPAC classification, the sample depicted in Figure 4a showed type IV isotherm and type H1 hysteresis, typical of mesoporous material with capillary condensation, suggesting the presence of particles of even size regularly ordered 15,23  The main vibrational frequencies and their respective attributions are shown in Figure 5.For calcined samples, the stretch between C-H from groups, CH 2 and CH 3 corresponding to TTMA + , CTMA + and DTMA + used on sample formation of MCM-41 was absent.This confirmed the efficiency of the calcination process 39 .
Figure 6 shows TG/DTG profiles of C samples.Three mass loss events can be observed and were all characteristic of the MCM-41 mesoporous material.The first one corresponded to loss of adsorbed water, the second one to the decomposition of surfactants present in the pores of the material and, finally, the third event corresponded to condensation of silanol groups from internal pore surfaces 40 .TG/DTG plots (Figure 6) showed faster deterioration of samples with mixtures of surfactants in the pores of the MCM-41 structure.The temperature range of the deterioration of surfactants and the percentage of mass loss are shown in Table 3.
SEM images of calcined samples are shown in Figures 7 and 8.The images reveal that the morphology of the material is similar for all samples and depicted clusters of hollow vermiform tubes which tend to curl over the surface.There are reports of MCM-41 morphologies that include spherical, rodlike, discoids and gyroids, in addition to millimeter-to-micrometer sized particles and hollow spheres, hollow tubes, wormlike, monolithic gels, thin films, among   others.Images similar to those reported here in can also be seen elsewhere [41][42][43] .
The results of the adsorption tests of C The experimental design revealed that trial 4 depicted the best result, with CO 2 adsorption of 0.62g/g at 40 bar.These results can be explained by the synergetic effect of physisorption and chemisorption on MCM-41 associated with the high surface area and the uniform mesoporous channels of MCM-41.shape and curvature of pores were claimed to be important for the diffusion of molecules through the structure and the ultimate adsorption capacity 16 .Thus, it was expected that the best CO 2 adsorption performance would be obtained by the highest surface area material, the highest pore volume, and the high degree of uniformity 25 .
Table 5 shows the coefficients to build the quadratic model obtained by statistical analysis.The limit of 95% reliabilitywas adopted.It can be observed that the quadratic model was best fitted to experimental data once the value of p was below 0.05 to the limit of reliability of 95%.
The analysis of the statistically significant terms from the adjusted model suggested that in the adjusted equation the components interact synergistically, increasing the capacity of CO 2 adsorption , except for C 19 C 21 .From the Equation 6, it can be seen that the effect of trimethyloctadecylammonium bromide -C 21 (C 21 H 46 NBr) associated with cetyltrimethylammonium bromide -C 19 (C 19 H 42 NBr) used in the formation of MCM-41 C 19 C 21 interacted decreasing the capacity of CO 2 adsorption, as it can be established by the negative value of the coefficient.This may be attributed to the relatively lower surface area of this material and its low degree of uniformity as inferred from BJH (Figure 4b) and XRD data (Figure 3).Thus, there exists a clear correlation between high pressure CO 2 adsorption data and both surface area and pore ordering 25 .
Table 6 shows the results from analysis of the adsorption variance of CO 2. It can be observed that the quadratic model best adjusted the experimental data, once the value of p<0.05 and the coefficient of regression R-Sqrwas 0.99.
The quality of the adjustment shown in Table 6 is given by the coefficient of determination, R 2 , defined by Equation 7.
However, high values of R 2 do not necessarily imply that the model is adequate.Therefore, it is very important the parallel use of the adjusted value of 2 A R , defined by Equation 8( ) ( ) Models with high 2 A R values are generally preferable 44 .Figure 9 shows the response models of C g. in this study A = C 17 H 38 NBr; B = C 19 H 42 NBr and C = C 21 H 46 NBr, experiments must include 7 different mixtures: 3 experiments with pure components (1.00 A, B, and C), corresponding to the vertex of the diagram), 3 experiments with binary mixtures (0.50 A and B, 0.50 A and C and 0.50 B and C), corresponding to the mid-points of the edges, and a central point (0.33 A, B and C), corresponding to the centroid of the diagram (Figure

Figure 1 .
Figure 1.Representation of the experimental space "simplex" of three independent variables.

X
-ray diffracton patterns of samples C 17 , C 19 , C 21 , C 17 C 19 , C 17 C 21 , C 19 C 21 and C 17 C 19 C 21 are shown in Figure 3.All the samples presented profiles typical of MCM-41, with the presence of major diffraction peaks (100), (

Figure 4 .
Figure 4. (a) Isotherms and (b) Pore distribution curves, based on the BET and BJH method, respectively.
, C 19, C 21, C 17 C 19, C 17 C 21, C 19 C 21 andC 17 C 19 C 21 materials showed well-defined hexagonal structure after calcination.By statistical planning of the mixtures, it was possible to determine the best combination rate among the surfactants tested to maximize CO 2 adsorption .The response models showed that the best combination of the surfactants resulted from C 17 C 19 sample, which presented synergistic interactions reaching the highest value of CO 2 adsorption (0.62 g CO 2 /g adsorbent), compared to all other samples.Moreover, that sample showed highest crystallinity, surface area and pore volume.By statistical planning, it was clear that there was a synergistic interaction of the surfactants from different hydrophobic chains in the formation of MCM-41 and CO 2 adsorption.
tetradecyltrimetylammonium bromide-C 17 H 38 NBr ), C 19 (cetyltrimethylammonium bromide -C 19 H 42 NBr) and C 21 (trimethyloctadecylammonium bromide C 21 H 46 NBr).Along the sides the binary mixtures can be found and, finally, the spot inside the triangle corresponds to the ternary mixture C 17 C 19 C 21 (C 17 H 38 NBr + C 19 H 42 NBr + C 21 H 46 NBr).The mixture response is represented by the contour line

Proportion of the surfactants at MCM-41(%) C 17 H 38 NBr C 19 H 42 NBr C 21 H 46 NBr
Synthesis Optimization of MCM-41 for CO 2 Adsorption Using Simplex-centroid Design 17 , C 19 , C 21 , C 17 C 19 , C 17 C 21 , C 19 C and C 17 C 19 C 21 are listed in Table . In Figure 4b, it can be observed pore size distributions in the range of 26 and 35 A. It can also be observed that samples C 17 C 19 C 21 and C 17 C 19 depicted uniform mesoporous distributions.Conversely, samples C 21 and C 17 C 21 , which used DTMA + surfactant in their synthesis, presented poor mesoporous uniformity.

Table 2 .
Physico-chemical parameter of materials.

Table 3 .
Temperature range of deterioration of the surfactants occluded in pores from samples.
17, C 19, C 21, C 17 C 19, C 17 C 21, C 19 C 21 and C 17 C 19 C 21 components.The result demonstrates that the region between vertex C 17 and C 19 is the one which showed the best CO 2 adsorption rate.Comparing with the results shown in Table 4, it can be observed that sample C 17 C 19 presented high synergistic interaction reaching

Table 5 .
Regression analysis for CO 2 adsorption from MCM-41 samples.Factors in which p-value is less than the level chosen of significance (0.05) are considered to have a statistically significant effect over CO 2 adsorption .. *

Table 6 .
Analysis of variance (ANOVA) from quadratic model.