Journal of Environmental Management 157 (2015) 194e204
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
CO2 adsorption on diatomaceous earth modified with cetyltrimethylammonium bromide and functionalized with tetraethylenepentamine: Optimization and kinetics Phuwadej Pornaroonthama a, Nutthavich Thouchprasitchai a, Sangobtip Pongstabodee a, b, * a b
Department of Chemical Technology, Faculty of Science, Chulalongkorn University, 254 Phayathai Road, Pathumwan, Bangkok 10330, Thailand Center of Excellence on Petrochemical and Materials Technology, Chulalongkorn University, 254 Phayathai Road, Pathumwan, Bangkok 10330, Thailand
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 February 2015 Received in revised form 7 April 2015 Accepted 9 April 2015 Available online
The carbon dioxide (CO2) adsorbent diatomaceous earth (DE) was modified with cetyltrimethylammonium bromide (CTAB) and functionalized with varying levels of tetraethylenepentamine (TEPA). The CO2 absorption at atmospheric pressure was optimized by varying the TEPA-loading level (0 e40% (w/w)), operating temperature (40e80 C) and water vapor concentration (0e16% (v/v)) in a 10% (v/v) CO2 feed stream in helium balance using a full 23 factorial design. The TEPA/CTAB-DE adsorbents were characterized by X-ray diffractometry, Fourier transform infrared spectrometry and thermogravimetric analyses. The CO2 adsorption capacity increased as each of these three factors increased. The TEPA loading level-water concentration interaction had a positive influence on the CO2 adsorption while the operating temperatureewater concentration interaction was antagonistic. The optimal condition for CO2 adsorption on 40%TEPA/CTAB-DE, evaluated via a factorial design response surface method (RSM), was a temperature of 58e68 C and a water vapor concentration of 9.5e14% (v/v), with a maximum CO2 adsorption capacity of 149.4 mg g1 at 63.5 C and 12% (v/v) water vapor concentration in the feed. Validation and sensitivity tests revealed that the estimated CO2 adsorption capacity was within ±4% of the experimental values, suggesting that the RSM model was satisfied and acceptable. From three kinetic models (pseudo-first-order, pseudo-second-order model and Avrami's equation), assessed using an error function (Err) and the coefficient of determination (R2), Avrami's equation was the most appropriate to describe the kinetics of CO2 adsorption on the 40%TEPA/CTAB-DE adsorbent and suggested that more than one reaction pathway occurred in the CO2 adsorption. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Carbon dioxide adsorption Diatomaceous earth-CTABTetraethylenepentamine Optimization Sesnsitivity
1. Introduction A large net annual emission of greenhouse gaseous carbon dioxide (CO2) from human activity, especially from fossil-fuels combustion, has raised serious environmental concerns on global climate change and unsustainable perturbance of ecosystems. An increasing amount of research on the reduction of CO2 emission levels to the environment has been reported. Conventional scrubbing with an aqueous amine solution was first used to capture the
* Corresponding author. Department of Chemical Technology, Faculty of Science, Chulalongkorn University, 254 Phayathai Road, Pathumwan, Bangkok 10330, Thailand. E-mail address: [email protected] (S. Pongstabodee). http://dx.doi.org/10.1016/j.jenvman.2015.04.013 0301-4797/© 2015 Elsevier Ltd. All rights reserved.
CO2 from the exhaust of coal- and gas-fired power plants and industrial plants. However, it has drawbacks in terms of solvent evaporation, equipment corrosion, high energy consumption for the solvent regeneration and high maintenance costs (Rochelle, 2009; Yu et al., 2012b; Wang et al., 2011). To overcome these drawbacks the use of solid absorbents like amines anchored on porous materials has gained in interest. The porous solid adsorbents most widely used to date for CO2 adsorption are activated carbon (Guo et al., 2006; Siriwardane et al., 2001; Songolzadeh et al., 2012), zeolites (Siriwardane et al., 2001; Songolzadeh et al., 2012), porous silicas (Belmabkhout and Sayari, 2009; Belmabkhout et al., 2009), carbon fibers (Thiruvenkatachari et al., 2009), silica gel (Goodman, 2009; Yu et al., 2012b), ion-exchange resins (Meng and Park, 2012), metal-organic frameworks (MOFs) (Li et al., 2011), organic-inorganic hybrid sorbents (Chaffee et al.,
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
2007) and mineral clays (Maroto-Valer et al., 2005; Martunus et al., 2012). The efficiency of CO2 adsorption on amine functionalized porous solid adsorbents depends on the porous support, the type and number of active amine sorption sites and the operating mode for sorption. Generally, there are three main operating modes for CO2 adsorption on amine functionalized porous solid adsorbents. The first is pressure swing adsorption (PSA), where pressurization is used for the adsorption stage and desorption takes place by depressurization. The second is temperature swing adsorption (TSA), where the adsorption/desorption stages are mediated via increasing and decreasing the operating temperature, respectively. Finally, the third is a combination of both PSA and TSA. Diatomaceous earth (DE), also known as diatomite or diatoms, is a natural siliceous sedimentary rock that is extremely hydrophilic with a very light density, high porosity and excellent thermal resistance (Yuan et al., 2004). Due to these properties, DE was selected to be the porous support in this work. The surface hydrophobicity of DE has previously been modified by the use of some surfactants (Hu et al., 2013), and in this study a surfactant was employed as a compatibilizer between the hydrophilic DE and the hydrophobic amine phases. Tetraethylenepentamine (TEPA) is an interesting hydrophobic amine since it contains primary and secondary amino groups. The modification of siliceous mesocellular foam with TEPA has been shown to result in the formation of a high efficiency and stable solvent for CO2 capture (Feng et al., 2013). Nevertheless, many studies have been performed as sequential univariate analyses, which are based on the assumption of nointeraction between the factors. In contrast, studies of the CO2 adsorption when the effects of each factor on the adsorption capacity were investigated according to the levels of other factors are scarce. In practice, it is necessary to identify the critical process factors and to adjust these factors to optimize the adsorption capacity. In this study the CO2 adsorption capacity of DE modified with cetyltrimethylammonium bromide (CTAB) and functionalized with TEPA at atmospheric pressure was evaluated for different TEPA loading levels, operating temperatures and water vapor concentrations in a constant 10% (v/v) CO2 feed stream. Moreover, a more optimal condition for CO2 adsorption capacity (in terms of absorbent capacity) was evaluated via a face-centered central composite design response surface model (FCCCD-RSM) based on the derived important factors and their interactions using a full 23 factorial design with three central points. The CO2 capture capability of TEPA/CTAB-DE was reported in terms of mg CO2 adsorbed per g of adsorbent. The adsorbents were characterized by means of X-ray diffraction (XRD), Fourier transform infrared spectrometry (FT-IR) and thermal gravimetric analysis (TGA). For the kinetic studies, two separate two-parameter models and one three-parameter model, which were associated with the surface-reaction kinetic steps, were used to gain further insight into the CO2 adsorption process. 2. Experimental 2.1. Preparation of the TEPA/CTAB-DE adsorbents Three g of DE, obtained from the T.K. Dinkao factory, amphur Mae Tha, Lampang, Thailand, was stirred vigorously in absolute ethanol at a 1:2 mass by volume (w/v) solid: liquid ratio for 12 h. The washed DE was then harvested by vacuum filtration and dried in a static air oven at 90 C for 5 h. The dried DE (3 g) was suspended in 150 ml of 0.1 M CTAB (aqueous) and stirred with a magnetic stirrer at 200 rpm for 4 h. The resultant CTAB-DE was harvested from the suspension by vacuum filtration and washed several times with de-ionized water to eliminate excess CTAB prior to drying in a static air oven at 90 C for 5 h. The dried CTAB-DE (3 g) was then
195
added to the TEPA solution (10, 25 or 40 percent by mass (% (w/w)) in ethanol at a constant 1:5 (w/v) ratio of TEPA: ethanol) and heated to 90 C with stirring at 400 rpm for 4 h. The solid particles were then harvested as above and dried at 90 C for 8 h to obtain the TEPA/CTAB-DE absorbents with the desired TEPA loading mass (10, 25 or 40% (w/w), noted as 10%TEPA/CTAB-DE, 25%TEPA/CTAB-DE and 40%TEPA/CTAB-DE, respectively). For the preparation of 40% TEPA/DE, the procedure was performed as mentioned above except without modification of the DE with CTAB. 2.2. Characterization of the adsorbents The crystal structures of the components in the different synthesized adsorbents were examined by XRD on a Bruker D8 Advance X-Ray diffractometer. The detector was operated at 40 kV and 30 mA using monochromatic CuKa radiation (l ¼ 0.15406 nm). The scan speed rate was about 5 min1 and the XRD profiles were recorded over a 2Theta range from 5 to 90 . The functional groups of the adsorbents were analyzed by FT-IR on a Perkin-Elmer (Spectrum one) spectrometer equipped with a mercury-cadmium-telluride detector. The wavenumber range of the functional groups was determined by comparing the obtained spectra to the literature. For the TGA, around 10 mg of the sample was placed in a chamber of thermal gravimetric analyzer (PerkinElmer Pyrisdiamond) and heated at a rate of 10 min1 from room temperature to 800 C under a nitrogen atmosphere at a flow rate of 10 ml min1 in order to analyze their thermal stability. 2.3. Adsorption of CO2 The CO2 capture experiments were investigated at atmospheric pressure. About 3 g of the adsorbent was packed between two layers of quartz wool inside a 0.25 inch inner-diameter quartz-tube reactor. The reactor temperature was controlled via a K-type thermocouple placed in the center of the reactor and equipped with a temperature controller. Prior to measuring the CO2 adsorption capacity, the reactor was heated up to 90 C for 1 h under a helium (He) flow at 30 ml min1 to remove any residual gas in the adsorbent sample. The reactor was then cooled down under the He flow to the desired adsorption temperature and then the He flow was switched to the 10 percent by volume (% (v/v)) CO2 concentration in a He balance at a flow rate of 10 ml min1 until the CO2 concentration in the outflow gas was the same as that in the feed flow, defined as the point of CO2 saturation of the adsorbent. The operating temperature was investigated over a range of 50e80 C, whilst the water vapor in the feed stream was studied over the 0e16% (v/v) range. The steam was generated via an evaporator and routed to the reactor. The effluent gases from the reactor were passed through a water-trap unit before being analyzed by gas chromatography (GC). The influent and effluent gases were analyzed every minute by on-line GC (GC-2014; Shimadzu Scientific Instruments) equipped with a thermal conductivity detector and using He as the carrier gas. The equilibrium CO2 adsorption capacity of the synthesized adsorbent was determined via analysis of the breakthrough curve in terms of mg CO2/g of adsorbent. 2.4. Experimental design for optimization of the CO2 adsorption capacity In order to evaluate the optimum condition for a maximal CO2 adsorption capacity on the adsorbent via a FCCCD-RSM, a full 23 factorial design with three central points was first conducted (random order of runs) to evaluate the influence of each factor on the adsorption capacity when varying one factor within the level of
196
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
2.5. Data analysis
the other factors (see Table 1). The three independent factors (adsorption temperature ( C), TEPA loading level in the adsorbent (% (w/w)) and water concentration (% (v/v)) in the influent) in original measurement units were encoded in dimensionless variables as a low level (1), central point (0) and a high level (þ1), and the CO2 adsorption capacity on the adsorbent was used as the response. Other factors, that might affect the response, were held constant. The adsorbent mass/total gas flow rate (W/F) ratio was maintained at 1.8 g s ml1 and the CO2 concentration in the feed was constant at 10% (v/v) in a He balance. All tests were performed in three replications and the data represented as a mean of the three. Based on the factorial design results, a FCCCD-RSM was subsequently adopted to optimize the conditions for the maximal CO2 sorption capacity (see Table 2). To investigate the validity of the three dimensional (3D)-RSM, six more trials were randomly performed. The magnitude of the difference in the experimental result from the evaluated one was used to reflect the validity of the RSM and was reported in terms of the percentage error. In addition, two more trials were performed in which the level of each of the factors was outside the given range in order to test the sensitivity of the model.
All tests were performed in triplicate and the mean data were analyzed at a 95% confidence interval with The Design-Expert 7.0 software package (Stat Ease Inc. Minneapolis, USA), including the analysis of variance (ANOVA), percentage of contribution, the Pareto chart of absolute standardized effects, normal probability plot of the effects and the residual plot. Statistical significance of the differences in means was accepted at the p < 0.05 level. 3. Results and discussion 3.1. Characterization of the adsorbents The crystalline structure of the different adsorbents (DE, CTABDE and CTAB-DE with TEPA loading levels of 10, 25 and 40 wt%) was characterized by XRD analysis, with representative results shown in Fig. 1. For the crystalline structure of DE, the XRD diffractions of quartz were found at a 2Theta of 19.87, 26.65 , 39.49 , 50.17, 54.90 and 59.98 . These patterns corresponded to JCPDS card no. 76e0823 and so a hexagonal crystal with lattice parameters of a ¼ b ¼ 4.9135 Å and c ¼ 5.4045 Å. The diffraction peaks of
Table 1 Experimental variables in coded values for the 23 factorial design with three central points. Standard order
Run order
Aa Operating temperature (Degree celsius)
Bb TEPA loading (Percent by mass)
Cc Water concentration (Percent by volume)
CO2 adsorption capacityd (mg g1)
1 2 3 4 5 6 7 8 9 10 11
4 6 5 9 1 8 2 7 11 10 3
1 þ1 1 þ1 1 þ1 1 þ1 0 0 0
1 1 þ1 þ1 1 1 þ1 þ1 0 0 0
1 1 1 1 þ1 þ1 þ1 þ1 0 0 0
43.9 74.1 81.2 109.2 55.1 57.6 125.3 129.9 92.0 92.5 95.1
Coded values of (1), (0), and (þ1) refer to the actual values of 40, 60, and 80 C, respectively. Coded values of (1), (0), and (þ1) refer to the actual values of 10, 25, and 40 percent by mass, respectively. c Coded values of (1), (0), and (þ1) refer to the actual values of 0, 8, and 16 percent by volume, respectively. d The adsorbent mass/total gas flow rate ratio and carbon dioxide concentration in the feed were constant at 1.8 g s ml1 and 10 percent by volume carbon dioxide in a helium balance, respectively. a
b
Table 2 Experimental variables in coded values for the FCCCD-RSM with three central points. Standard order
Run order
Aa Operating temperature (Degree celsius)
Cb Water concentration (Percent by volume)
CO2 adsorption capacityc (mg g1)
1 2 3 4 5 6 7 8 9 10 11
4 2 11 6 8 10 1 5 3 7 9
1 þ1 1 þ1 1 þ1 0 0 0 0 0
1 1 þ1 þ1 0 0 1 þ1 0 0 0
74.1 109.2 57.6 129.7 53.2 124.8 131.6 147.3 135.0 134.4 137.4
Coded values of (1), (0), and (þ1) refer to the actual values of 40, 60, and 80 C, respectively. Coded values of (1), (0), and (þ1) refer to the actual values of 0, 8, and 16 percent by volume, respectively. c The tetraethylenepentamine loading level on diatomaceous earth modified with cetyltrimethylammonium bromide, adsorbent mass/total gas flow rate ratio and carbon dioxide concentration in the feed were constant at 40 percent by mass, 1.8 g s ml1 and 10 percent by volume carbon dioxide in a helium balance, respectively. a
b
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
Fig. 1. Representative XRD patterns of DE, CTAB-DE and the 10%/25%/40%TEPA/CTABDE absorbents.
cristobalite were observed at a 2Theta of 20.96 , 36.30 and 42.89 , which corresponded to JCPDS card no. 39e1425 and so a tetragonal structure with lattice parameters of a ¼ b ¼ 4.9709 Å and c ¼ 6.9180 Å. In addition, the diffraction peak of kyanite was seen at a 2Theta ¼ 27.96 , which is an acceptable match to that in JCPDS card no. 11e46 and so a triclinic structure with lattice parameters of a ¼ 7.1150 Å, b ¼ 7.8410 Å and c ¼ 5.5724 Å. The peaks at 2Theta of 12.41, 21.41, 24.97 and 34.97 corresponded to the diffraction peaks of kaolinite reported in JCPDS card no. 89e6538, and so was be ascribed as a triclinic crystal structure with lattice parameters of a ¼ 5.1535 Å, b ¼ 8.9419 Å and c ¼ 7.3906 Å. The diffraction peaks of hematite at a 2Theta of 33.15 and 40.86 were very similar to those in JCPDS card no. 33-0664, and so was defined as a rhombohedral structure with lattice parameters of a ¼ b ¼ 5.0376 Å and c ¼ 13.7489 Å. There was no shift in these diffraction peaks in the different adsorbents when DE was modified with CTAB (CTAB-DE) or when the CTAB-DE was loaded with TEPA (TEPA/CTAB-DE), showing that CTAB and TEPA had no effect on the crystal lattice of the DE, but rather were only modified at the surface of the DE. The FT-IR spectra of the different absorbents were evaluated in order to analyze the surface functional groups of the samples (Fig. 2). In addition, the spent 40%TEPA/CTAB-DE absorbent used at 80 C with or without 16% (v/v) water vapor in the feed was analyzed. The vibration bands of pure TEPA (Fig. 2Ia) showed a broad peak at 3348 cm1, assigned to the NeH antisymmetric stretching (NH2) mode of TEPA, and a peak shoulder at 3274 cm1 corresponding to the NeH symmetric stretching (NH and NH2) mode (Yang et al., 2012). Two overlapped peaks appeared at 2930 and 2825 cm1, ascribed to the CeH antisymmetric stretching (CH2) and stretching (NeCH2) vibrations of the TEPA chains, respectively (Yang et al., 2012; Yu et al., 2012a,b). The shoulder at 1647 cm1 was due to the ring skeleton stretching mode of cyclization among TEPA. The peak located at 1581 cm1 represented the NeH scissoring (NH2) of TEPA. The CeH deformation (CH2) of TEPA was observed at 1473 cm1 with a shoulder at 1433 cm1 (Hiyoshi et al., 2005). Additionally, the peak at 1310 cm1 with overlapped peaks at 1384 and 1350 cm1 was assigned to the CeH wagging (CH2) vibration of TEPA (Rusu et al., 2008; Yang et al., 2012; Yu et al., 2012a), and that at 1119 cm1 was ascribed to the CeN stretching (ReNeR) mode (Yang et al., 2012; Yu et al., 2012a). The vibration of CTAB in the wavenumber range of 4000e400 cm1 is shown in Fig. 2Ib. The wavenumber at 3022 and 3018 cm1 were assigned to the antisymmetric CeH stretching mode of the Nþ(CH3)3 group. The symmetric CeH stretching vibration was seen as a shoulder at 2956 cm1. The antisymmetric
197
and symmetric stretching of CH3-vibrational bands appeared at 2934 and 2861 cm1, respectively. The antisymmetric CeH stretching (CH2) mode at 2917 cm1 and symmetric CeH stretching (CH2) at 2851 cm1 were observed in the CeH stretching region (Venkataraman and Vasudevan, 2001). The band at 1487 cm1 was assigned to the antisymmetric NeCH3 scissoring, while those at 1472 and 1462 cm1 were the vibrational modes of the CH2 scissoring. Two peaks for CeC skeletal stretching mode were observed (962 and 937 cm1) with the CeN stretching vibration at 912 cm1. The band at 720 cm1 was assigned to the (CH2)n rocking mode, and the broad bands at 3418 and 1622 cm1 to the stretching and bending modes of adsorbed water molecules, respectively. The FT-IR spectra of DE (Fig. 2Ic) exhibited a broad band at 3418 cm1, attributed to the symmetric OeH stretching mode of H et al., 2002). The bonded adsorbed water molecules (Madejova band centered at 1622 cm1 is the HeOeH bending vibration of water molecules sorbed on the surface of DE (Xu et al., 2000). The bands at 3632 and 3704 cm1 were assigned to the OeH stretching vibrations of the structural OeH (or inner hydroxyl) groups , 2003; Xue et al., 2007) whilst the (Johnston et al., 1992; Madejova SieOeSi, SieO, AleOeAl and AleO metal oxides bond vibrated in the wavenumber range of 400e1200 cm1. The bands at 1110, 788 and 476 cm1 corresponded to antisymmetric stretching, symmetric stretching and bending vibrations of SieOeSi siloxane (Huang et al., 2007; Sheng et al., 2009; Sun et al., 2013), respectively. The SieO stretching vibration was observed at 1090 and 1030 cm1 (Xue et al., 2007). The AleOeSi stretching vibration was indicated at 800 cm1 for the DE sample and that at 700 cm1 was assigned to the AleO stretching vibration (Xue et al., 2007). The band at 540 cm1 was assigned to the SieOeAl (octahedral Al) , 2003). From the FT-IR spectra of the CTAB-DE (Fig. 2Id), (Madejova the spectra showed the vibration of both CTAB and DE, which indicated that DE was modified with CTAB. The 10%TEPA/CTAB-DE (Fig. 2Ie) expressed the characteristic vibration bands of TEPA, CTAB and DE, supporting that TEPA was loaded on the CTAB-DE via the method used in this work. When loading a higher (25 or 40% (w/w)) amount of TEPA on the CTAB-DE the intensity of the peaks that corresponded to TEPA increased (Fig. 2II). The FT-IR spectra of the spent 40%TEPA/CTAB-DE, used with or without the addition of 16% (v/v) water vapor in the feed at 80 C, are shown in Fig. 2III. Without the addition of water vapor in the feed, the vibration band of the carbonyl stretching of carbamate species was observed at 1658 cm1 (Hasib-ur-Rahmana et al., 2012) and bicarbonate species was observed at a wavenumber of 1410 cm1 due to the reaction among the amine groups of TEPA with the sorbed water molecules on DE and CO2. In the presence of added water vapor in the feed, the vibration bands of bicarbonate bands were found at 3395, 1623, 1410 and 1224 cm1, which were the vibrations of OH stretching, CO2 antisymmetric stretching, CO2 symmetric stretching, and COH bending, respectively (Garand et al., 2010). Additionally, the CeO stretching mode of imino-carbonate occurred at 1473 cm1 (Jackson et al., 2009). The change in the FT-IR absorption bands for CO2 sorption in the absence/presence of water was in agreement with Yang and He (Yang and He, 2014). The thermal stability of the different absorbents was evaluated by TGA, where Fig. 3 shows the TGA profile in terms of the mass loss as a function of the temperature. For pure TEPA, the mass loss was around 4.1 wt% at below 150 C, attributed to the moisture vaporization in TEPA. Decomposition of TEPA started at 180 C and was complete at 260 C. For DE the mass loss of 4 wt% from 30 to 200 C and a further 3.3 wt% from 250 to 497 C were assigned to the evaporation of adsorbed water and dehydroxylation of some associated SieOH groups on the external surface of the DE, respectively (Chaisena and Rangsriwatananon, 2004). For CTAB-DE, the slight mass loss (2.3 wt%) from 30 to 200 C corresponded to the
198
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
dehydration of physically adsorbed water, and the mass loss at 200e290 C was attributed to the partial decomposition of CTAB (Wang et al., 2007). Further increasing the temperature to 500 C caused the complete decomposition of CTAB. For the TEPA/CTAB-DE adsorbents with different TEPA loading levels, the TGA profiles showed two steps of mass loss. The first one, from 60 to 130 C with a total mass loss of 4.8 wt%, was attributed to the vaporization of moisture on the adsorbent. The second step, from 130 to 290 C with a total mass loss of 13.9, 27.5 and 41.4 wt% for the 10, 25 and 40%TEPA/CTAB-DE, respectively, was due to the decomposition of CTAB and TEPA. In addition, two mass loss steps were found in the
TGA curve of 40%TEPA/DE, but each step was shifted to a lower temperature range. This indicated that a higher thermal stability was achieved when modifying DE with CTAB before loading the TEPA. Regardless, the thermal stability behavior of the adsorbents was in an acceptable temperature range for this application. 3.2. Measurement of CO2 adsorption The CO2 adsorption capability of the TEPA/CTAB-DE adsorbents was investigated with different levels of TEPA loading, operating temperature and amount of water vapor in the feed stream.
Fig. 2. Representative FT-IR spectra of the synthesized absorbents. [I] The fresh (a) pure TEPA, (b) CTAB, (c) DE, (d) CTAB-DE and (e) 10%TEPA/CTAB-DE. [II] The 10%/25%/40%TEPA/ CTAB-DE. [III] The fresh (solid line) and spent (dashed line) 40%TEPA/CTAB-DE with and without the addition of 16% (v/v) water vapor in the feed at an operating temperature of 80 C.
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
Fig. 3. Representative TGA profiles of DE, CTAB-DE, 10%/25%/40%TEPA/CTAB-DE, 40% TEPA/DE and pure TEPA.
The influence of the TEPA loading level on the CO2 adsorption capacity of the TEPA/CTAB-DE adsorbent at an operating temperature of 70 C without any added water vapor in the feed stream is shown in Fig. 4a. The adsorption capacity of CTAB-DE was 58.1 mg g1 and this increased 1.17-, 1.32- and 1.59-fold after loading 10, 25 and 40% (w/w) TEPA, respectively, onto the CTAB-DE, reaching a CO2 adsorption capacity of 92.2 mg g1 at a 40% (w/w) TEPA loading. That increasing the TEPA loading level increased the CO2 adsorption capacity is due to the higher amount of available amine groups (from the TEPA molecules) on the surface of the CTAB-DE support to react with CO2 (Xu et al., 2002; Veneman et al., 2012) to form carbamates, as seen in the FT-IR analysis. Thus, the TEPA loading level can affect the capacity performance of the adsorbents. However, loading over 40% (w/w) TEPA on the CTAB-DE resulted in the formation of moist adsorbents during the preparation and so the CO2 adsorption capacity could not be measured. Accordingly, a TEPA loading level of 40% (w/w) was selected for further study. The CO2 adsorption capacity of the 40%TEPA/CTAB-DE adsorbent without any added water vapor in the feed stream increased 1.31-,
199
1.85- and 2.2-fold with increasing adsorption temperature at 60, 70 and 80 C, respectively, from 49.6 mg g1 at 50 C up to 109.2 mg g1 at 80 C (Fig. 4b). Increasing the temperature increased the kinetic energy of the CO2 molecules and so (in this temperature range) allows them to penetrate more easily into the TEPA particles in the adsorbent pores and accordingly make contact with more CO2 affinity sites (Xu et al., 2002; Yue et al., 2006). Therefore, an operating temperature of 80 C was selected for further studies. Generally, water vapor in the range of 8e20% (v/v) is one of the components in flue gas and other industrial gases. The CO2 adsorption capacity of the 40%TEPA/CTAB-DE at 80 C increased from 109.2 mg g1 with no added water vapor absence in the feed stream to 116.5 mg g1 and 129.7 mg g1 when adding 8 and 16% (v/v) water vapor, respectively (Fig. 4c). That the addition of water vapor in the feed promotes the CO2 adsorption is because without added water vapor one mole of CO2 molecules reacts with two moles of the primary (RNH2) and secondary (R2NH) amino groups in TEPA, leading to the formation of carbamate, as shown in Eqs. (1) and (2):
CO2 þ 2RNH2 4RNHCOO þ RNH3 þ
(1)
CO2 þ 2R2 NH4R2 NCOO þ RNH3 þ
(2)
The formation of carbamate species on the adsorbent was evident in the FT-IR results (Fig. 2III). However, when water vapor was added into the feed the OH groups of water were involved in the CO2 adsorption and bicarbonate formation, as supported by the increased bicarbonate levels seen in the FT-IR analysis. One mole of CO2 molecules reacts with one mole of the amine in the presence of H2O or hydroxyl groups to form bicarbonate species (Yue et al., 2006). When no water vapor was added some bicarbonate species were still formed, as shown in the FT-IR spectra, due to the presence of residual water in the adsorbent, as supported by the TGA results. However, the adsorption capacity could not be evaluated at levels of water vapor above 16% (v/v) due to slurry formation and blockage of the gas flow in the reactor. Accordingly, the addition of an appropriate amount of water vapor in the feed, in this case 16% (v/v), has a positive effect on the CO2 adsorption.
Fig. 4. Representative CO2 adsorption capacity as a function of (a) the TEPA loading level without any added water vapor in the feed and at an operating temperature of 70 C; (b) the operating temperature for 40%TEPA/CTAB-DE without any added water vapor in the feed and (c) the H2O concentration in the feed for a 40%TEPA/CTAB-DE adsorbent at 80 C. In all cases the W/F ratio was constant at 1.8 g s mL1 and the CO2 concentration in the feed was constant at 10% (v/v) in a He balance.
200
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
3.3. Experimental design 3.3.1. Screening factors using a 23 factorial design With respect to the CO2 adsorption capacity of the adsorbent, which factors and their interactions had a significant effect was not determined, and thus the optimal condition for CO2 adsorption was not ascertained in Section 2. Accordingly, a full 23 factorial design with three central points was employed to determine the important factor(s) and their interaction(s) on the adsorption capacity (Table 1). Statistical analysis at a 95% confidence interval was used to evaluate the data. The normal probability plot of the effects (Fig. 5a) revealed that the operating temperature, TEPA loading level and the water concentration in the feed, plus the TEPA loading level-water concentration interaction had a normal probability of more than 50%, inferring that they had a synergistic effect on the adsorption capacity. That the adsorption capacity increased as these factors increased from a low level to a high level is in accordance with the CO2 adsorption measurement. Only the interaction between the operating temperature and water concentration had a significantly less than 50% probability, indicating that this interaction is antagonistic on the CO2 adsorption capacity.
The Pareto chart for the T-value of jEffectj compared to the Bonferroni limit (5.068 in this case) revealed that only the interaction between the operating temperature and water concentration had a significant negative effect on the CO2 adsorption capacity (Fig. 5b), whilst the three main factors and the TEPA loading levelwater vapor interaction were all positive. Thus, the direction of effect of each factor on the CO2 capacity was similar to that expressed by the normal probability plot. Moreover, the Pareto chart revealed that there is a curvature in the relationship among the adsorption capacity, the important factors and their interaction. In agreement with the above analysis, the ANOVA revealed that the three main factors, and the TEPA loading-water concentration and operating temperatureewater concentration interactions all had a significant effect on the CO2 adsorption capacity (Table 3). Moreover the percentage contribution (related to the total sum of squares) suggested their relative importance was ranked (highest to lowest) as the TEPA loading level >>> TEPA loading level-H2O concentration interaction > temperature > H2O concentration > temperatureeH2O concentration interaction. This ranked result and the relative importance of each factor and their interactions is similar to that obtained in the normal probability plot and the Pareto chart. The R2 of 0.9990 indicates a good agreement between the experimental and estimated values for the CO2 sorption capacity. The randomized error variable in the experimental results was rather small, and the Adj. R2 (0.9977) was very close to that of R2, indicating that a requisite term was included in the model. The adequate precision (76.160) was much greater than 4, and so the selected model had a moderate performance ability in evaluating the CO2 sorption capacity and was effective at navigating the design space. In particular, only 1.61% of the data points were dispersed around the mean, as indicated by the coefficient of variation (C.V.), with no significant terms likely to be missing from the model. The regression equation (in terms of coded factors) of the CO2 adsorption capacity (Qe) could be expressed by Eq. (3),
Qe ¼ þ84:50 þ 8:13A þ 26:83B þ 7:41C 6:42AC þ 8:73BC (3) The positive sign of the coefficient estimate of the operating temperature (A), TEPA loading level (B), water concentration in the feed (C), and the TEPA loading level-H2O concentration interaction (BC), and the negative sign of the operating temperatureewater concentration interaction (AC), in the regression equation are in agreement with the adsorption capacity results (Fig. 4aec). From the ANOVA, the indicated curvature in the model suggested that the average response value did not correspond to the response value at the central point for the factors studied, in accord with the curvature shown in the results of the Pareto chart. Accordingly, a RSM analysis was employed to achieve a better perception.
Fig. 5. Representative (a) normal probability plot of the effects for the full 23 factorial design and (b) Pareto chart of the factors for the 23 factorial design.
3.3.2. Optimization by FCCCDeRSM analysis Due to the extreme contribution of the TEPA loading effect on the response, which was from 9.4- to 17.5-fold more than the contribution of the other effects (as obtained from the factorial design in Section 3.3.1) the other effects would be hidden by the TEPA loading effect in a RSM analysis. Accordingly, the value of the TEPA loading at 40% (w/w) obtained from the results of the CO2 adsorption measurement and the factorial design was set as a constant in the FCCCD-RSM. Based on the factorial design results, the operating temperature and water vapor concentration in the feed were then varied to determine the optimum conditions for the adsorption capacity (see Table 2). The CO2 adsorption capacity response surface equation in terms of coded factors could be expressed by Eq. (4),
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
201
Table 3 Analysis of variance (ANOVA) of the significant terms for the 23 factorial design with three central points. Source
Sum of squares
DF
Model A-Temperaturea B-TEPA loadinga CeH2O concentrationa ACb BCb Curvature Residual Lack of Fit Pure Error Total R-Squared Adj. R-Squared
7667.97 528.79 5759.66 439.51 329.98 610.04 163.58 7.78 2.32 5.46 7839.33 0.9990 0.9977
5 1533.59 1 528.79 1 5759.66 1 439.51 1 329.98 1 610.04 1 163.58 4 1.95 2 1.16 2 2.73 10 e C.V. percent Adeq. Precision
a b c
Mean square
F-value
P-valuec
Percent contribution
788.14 271.75 2959.98 225.87 169.58 313.51 84.06 e 0.43 e e 1.606 76.10
Contents lists available at ScienceDirect
Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman
CO2 adsorption on diatomaceous earth modified with cetyltrimethylammonium bromide and functionalized with tetraethylenepentamine: Optimization and kinetics Phuwadej Pornaroonthama a, Nutthavich Thouchprasitchai a, Sangobtip Pongstabodee a, b, * a b
Department of Chemical Technology, Faculty of Science, Chulalongkorn University, 254 Phayathai Road, Pathumwan, Bangkok 10330, Thailand Center of Excellence on Petrochemical and Materials Technology, Chulalongkorn University, 254 Phayathai Road, Pathumwan, Bangkok 10330, Thailand
a r t i c l e i n f o
a b s t r a c t
Article history: Received 25 February 2015 Received in revised form 7 April 2015 Accepted 9 April 2015 Available online
The carbon dioxide (CO2) adsorbent diatomaceous earth (DE) was modified with cetyltrimethylammonium bromide (CTAB) and functionalized with varying levels of tetraethylenepentamine (TEPA). The CO2 absorption at atmospheric pressure was optimized by varying the TEPA-loading level (0 e40% (w/w)), operating temperature (40e80 C) and water vapor concentration (0e16% (v/v)) in a 10% (v/v) CO2 feed stream in helium balance using a full 23 factorial design. The TEPA/CTAB-DE adsorbents were characterized by X-ray diffractometry, Fourier transform infrared spectrometry and thermogravimetric analyses. The CO2 adsorption capacity increased as each of these three factors increased. The TEPA loading level-water concentration interaction had a positive influence on the CO2 adsorption while the operating temperatureewater concentration interaction was antagonistic. The optimal condition for CO2 adsorption on 40%TEPA/CTAB-DE, evaluated via a factorial design response surface method (RSM), was a temperature of 58e68 C and a water vapor concentration of 9.5e14% (v/v), with a maximum CO2 adsorption capacity of 149.4 mg g1 at 63.5 C and 12% (v/v) water vapor concentration in the feed. Validation and sensitivity tests revealed that the estimated CO2 adsorption capacity was within ±4% of the experimental values, suggesting that the RSM model was satisfied and acceptable. From three kinetic models (pseudo-first-order, pseudo-second-order model and Avrami's equation), assessed using an error function (Err) and the coefficient of determination (R2), Avrami's equation was the most appropriate to describe the kinetics of CO2 adsorption on the 40%TEPA/CTAB-DE adsorbent and suggested that more than one reaction pathway occurred in the CO2 adsorption. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Carbon dioxide adsorption Diatomaceous earth-CTABTetraethylenepentamine Optimization Sesnsitivity
1. Introduction A large net annual emission of greenhouse gaseous carbon dioxide (CO2) from human activity, especially from fossil-fuels combustion, has raised serious environmental concerns on global climate change and unsustainable perturbance of ecosystems. An increasing amount of research on the reduction of CO2 emission levels to the environment has been reported. Conventional scrubbing with an aqueous amine solution was first used to capture the
* Corresponding author. Department of Chemical Technology, Faculty of Science, Chulalongkorn University, 254 Phayathai Road, Pathumwan, Bangkok 10330, Thailand. E-mail address: [email protected] (S. Pongstabodee). http://dx.doi.org/10.1016/j.jenvman.2015.04.013 0301-4797/© 2015 Elsevier Ltd. All rights reserved.
CO2 from the exhaust of coal- and gas-fired power plants and industrial plants. However, it has drawbacks in terms of solvent evaporation, equipment corrosion, high energy consumption for the solvent regeneration and high maintenance costs (Rochelle, 2009; Yu et al., 2012b; Wang et al., 2011). To overcome these drawbacks the use of solid absorbents like amines anchored on porous materials has gained in interest. The porous solid adsorbents most widely used to date for CO2 adsorption are activated carbon (Guo et al., 2006; Siriwardane et al., 2001; Songolzadeh et al., 2012), zeolites (Siriwardane et al., 2001; Songolzadeh et al., 2012), porous silicas (Belmabkhout and Sayari, 2009; Belmabkhout et al., 2009), carbon fibers (Thiruvenkatachari et al., 2009), silica gel (Goodman, 2009; Yu et al., 2012b), ion-exchange resins (Meng and Park, 2012), metal-organic frameworks (MOFs) (Li et al., 2011), organic-inorganic hybrid sorbents (Chaffee et al.,
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
2007) and mineral clays (Maroto-Valer et al., 2005; Martunus et al., 2012). The efficiency of CO2 adsorption on amine functionalized porous solid adsorbents depends on the porous support, the type and number of active amine sorption sites and the operating mode for sorption. Generally, there are three main operating modes for CO2 adsorption on amine functionalized porous solid adsorbents. The first is pressure swing adsorption (PSA), where pressurization is used for the adsorption stage and desorption takes place by depressurization. The second is temperature swing adsorption (TSA), where the adsorption/desorption stages are mediated via increasing and decreasing the operating temperature, respectively. Finally, the third is a combination of both PSA and TSA. Diatomaceous earth (DE), also known as diatomite or diatoms, is a natural siliceous sedimentary rock that is extremely hydrophilic with a very light density, high porosity and excellent thermal resistance (Yuan et al., 2004). Due to these properties, DE was selected to be the porous support in this work. The surface hydrophobicity of DE has previously been modified by the use of some surfactants (Hu et al., 2013), and in this study a surfactant was employed as a compatibilizer between the hydrophilic DE and the hydrophobic amine phases. Tetraethylenepentamine (TEPA) is an interesting hydrophobic amine since it contains primary and secondary amino groups. The modification of siliceous mesocellular foam with TEPA has been shown to result in the formation of a high efficiency and stable solvent for CO2 capture (Feng et al., 2013). Nevertheless, many studies have been performed as sequential univariate analyses, which are based on the assumption of nointeraction between the factors. In contrast, studies of the CO2 adsorption when the effects of each factor on the adsorption capacity were investigated according to the levels of other factors are scarce. In practice, it is necessary to identify the critical process factors and to adjust these factors to optimize the adsorption capacity. In this study the CO2 adsorption capacity of DE modified with cetyltrimethylammonium bromide (CTAB) and functionalized with TEPA at atmospheric pressure was evaluated for different TEPA loading levels, operating temperatures and water vapor concentrations in a constant 10% (v/v) CO2 feed stream. Moreover, a more optimal condition for CO2 adsorption capacity (in terms of absorbent capacity) was evaluated via a face-centered central composite design response surface model (FCCCD-RSM) based on the derived important factors and their interactions using a full 23 factorial design with three central points. The CO2 capture capability of TEPA/CTAB-DE was reported in terms of mg CO2 adsorbed per g of adsorbent. The adsorbents were characterized by means of X-ray diffraction (XRD), Fourier transform infrared spectrometry (FT-IR) and thermal gravimetric analysis (TGA). For the kinetic studies, two separate two-parameter models and one three-parameter model, which were associated with the surface-reaction kinetic steps, were used to gain further insight into the CO2 adsorption process. 2. Experimental 2.1. Preparation of the TEPA/CTAB-DE adsorbents Three g of DE, obtained from the T.K. Dinkao factory, amphur Mae Tha, Lampang, Thailand, was stirred vigorously in absolute ethanol at a 1:2 mass by volume (w/v) solid: liquid ratio for 12 h. The washed DE was then harvested by vacuum filtration and dried in a static air oven at 90 C for 5 h. The dried DE (3 g) was suspended in 150 ml of 0.1 M CTAB (aqueous) and stirred with a magnetic stirrer at 200 rpm for 4 h. The resultant CTAB-DE was harvested from the suspension by vacuum filtration and washed several times with de-ionized water to eliminate excess CTAB prior to drying in a static air oven at 90 C for 5 h. The dried CTAB-DE (3 g) was then
195
added to the TEPA solution (10, 25 or 40 percent by mass (% (w/w)) in ethanol at a constant 1:5 (w/v) ratio of TEPA: ethanol) and heated to 90 C with stirring at 400 rpm for 4 h. The solid particles were then harvested as above and dried at 90 C for 8 h to obtain the TEPA/CTAB-DE absorbents with the desired TEPA loading mass (10, 25 or 40% (w/w), noted as 10%TEPA/CTAB-DE, 25%TEPA/CTAB-DE and 40%TEPA/CTAB-DE, respectively). For the preparation of 40% TEPA/DE, the procedure was performed as mentioned above except without modification of the DE with CTAB. 2.2. Characterization of the adsorbents The crystal structures of the components in the different synthesized adsorbents were examined by XRD on a Bruker D8 Advance X-Ray diffractometer. The detector was operated at 40 kV and 30 mA using monochromatic CuKa radiation (l ¼ 0.15406 nm). The scan speed rate was about 5 min1 and the XRD profiles were recorded over a 2Theta range from 5 to 90 . The functional groups of the adsorbents were analyzed by FT-IR on a Perkin-Elmer (Spectrum one) spectrometer equipped with a mercury-cadmium-telluride detector. The wavenumber range of the functional groups was determined by comparing the obtained spectra to the literature. For the TGA, around 10 mg of the sample was placed in a chamber of thermal gravimetric analyzer (PerkinElmer Pyrisdiamond) and heated at a rate of 10 min1 from room temperature to 800 C under a nitrogen atmosphere at a flow rate of 10 ml min1 in order to analyze their thermal stability. 2.3. Adsorption of CO2 The CO2 capture experiments were investigated at atmospheric pressure. About 3 g of the adsorbent was packed between two layers of quartz wool inside a 0.25 inch inner-diameter quartz-tube reactor. The reactor temperature was controlled via a K-type thermocouple placed in the center of the reactor and equipped with a temperature controller. Prior to measuring the CO2 adsorption capacity, the reactor was heated up to 90 C for 1 h under a helium (He) flow at 30 ml min1 to remove any residual gas in the adsorbent sample. The reactor was then cooled down under the He flow to the desired adsorption temperature and then the He flow was switched to the 10 percent by volume (% (v/v)) CO2 concentration in a He balance at a flow rate of 10 ml min1 until the CO2 concentration in the outflow gas was the same as that in the feed flow, defined as the point of CO2 saturation of the adsorbent. The operating temperature was investigated over a range of 50e80 C, whilst the water vapor in the feed stream was studied over the 0e16% (v/v) range. The steam was generated via an evaporator and routed to the reactor. The effluent gases from the reactor were passed through a water-trap unit before being analyzed by gas chromatography (GC). The influent and effluent gases were analyzed every minute by on-line GC (GC-2014; Shimadzu Scientific Instruments) equipped with a thermal conductivity detector and using He as the carrier gas. The equilibrium CO2 adsorption capacity of the synthesized adsorbent was determined via analysis of the breakthrough curve in terms of mg CO2/g of adsorbent. 2.4. Experimental design for optimization of the CO2 adsorption capacity In order to evaluate the optimum condition for a maximal CO2 adsorption capacity on the adsorbent via a FCCCD-RSM, a full 23 factorial design with three central points was first conducted (random order of runs) to evaluate the influence of each factor on the adsorption capacity when varying one factor within the level of
196
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
2.5. Data analysis
the other factors (see Table 1). The three independent factors (adsorption temperature ( C), TEPA loading level in the adsorbent (% (w/w)) and water concentration (% (v/v)) in the influent) in original measurement units were encoded in dimensionless variables as a low level (1), central point (0) and a high level (þ1), and the CO2 adsorption capacity on the adsorbent was used as the response. Other factors, that might affect the response, were held constant. The adsorbent mass/total gas flow rate (W/F) ratio was maintained at 1.8 g s ml1 and the CO2 concentration in the feed was constant at 10% (v/v) in a He balance. All tests were performed in three replications and the data represented as a mean of the three. Based on the factorial design results, a FCCCD-RSM was subsequently adopted to optimize the conditions for the maximal CO2 sorption capacity (see Table 2). To investigate the validity of the three dimensional (3D)-RSM, six more trials were randomly performed. The magnitude of the difference in the experimental result from the evaluated one was used to reflect the validity of the RSM and was reported in terms of the percentage error. In addition, two more trials were performed in which the level of each of the factors was outside the given range in order to test the sensitivity of the model.
All tests were performed in triplicate and the mean data were analyzed at a 95% confidence interval with The Design-Expert 7.0 software package (Stat Ease Inc. Minneapolis, USA), including the analysis of variance (ANOVA), percentage of contribution, the Pareto chart of absolute standardized effects, normal probability plot of the effects and the residual plot. Statistical significance of the differences in means was accepted at the p < 0.05 level. 3. Results and discussion 3.1. Characterization of the adsorbents The crystalline structure of the different adsorbents (DE, CTABDE and CTAB-DE with TEPA loading levels of 10, 25 and 40 wt%) was characterized by XRD analysis, with representative results shown in Fig. 1. For the crystalline structure of DE, the XRD diffractions of quartz were found at a 2Theta of 19.87, 26.65 , 39.49 , 50.17, 54.90 and 59.98 . These patterns corresponded to JCPDS card no. 76e0823 and so a hexagonal crystal with lattice parameters of a ¼ b ¼ 4.9135 Å and c ¼ 5.4045 Å. The diffraction peaks of
Table 1 Experimental variables in coded values for the 23 factorial design with three central points. Standard order
Run order
Aa Operating temperature (Degree celsius)
Bb TEPA loading (Percent by mass)
Cc Water concentration (Percent by volume)
CO2 adsorption capacityd (mg g1)
1 2 3 4 5 6 7 8 9 10 11
4 6 5 9 1 8 2 7 11 10 3
1 þ1 1 þ1 1 þ1 1 þ1 0 0 0
1 1 þ1 þ1 1 1 þ1 þ1 0 0 0
1 1 1 1 þ1 þ1 þ1 þ1 0 0 0
43.9 74.1 81.2 109.2 55.1 57.6 125.3 129.9 92.0 92.5 95.1
Coded values of (1), (0), and (þ1) refer to the actual values of 40, 60, and 80 C, respectively. Coded values of (1), (0), and (þ1) refer to the actual values of 10, 25, and 40 percent by mass, respectively. c Coded values of (1), (0), and (þ1) refer to the actual values of 0, 8, and 16 percent by volume, respectively. d The adsorbent mass/total gas flow rate ratio and carbon dioxide concentration in the feed were constant at 1.8 g s ml1 and 10 percent by volume carbon dioxide in a helium balance, respectively. a
b
Table 2 Experimental variables in coded values for the FCCCD-RSM with three central points. Standard order
Run order
Aa Operating temperature (Degree celsius)
Cb Water concentration (Percent by volume)
CO2 adsorption capacityc (mg g1)
1 2 3 4 5 6 7 8 9 10 11
4 2 11 6 8 10 1 5 3 7 9
1 þ1 1 þ1 1 þ1 0 0 0 0 0
1 1 þ1 þ1 0 0 1 þ1 0 0 0
74.1 109.2 57.6 129.7 53.2 124.8 131.6 147.3 135.0 134.4 137.4
Coded values of (1), (0), and (þ1) refer to the actual values of 40, 60, and 80 C, respectively. Coded values of (1), (0), and (þ1) refer to the actual values of 0, 8, and 16 percent by volume, respectively. c The tetraethylenepentamine loading level on diatomaceous earth modified with cetyltrimethylammonium bromide, adsorbent mass/total gas flow rate ratio and carbon dioxide concentration in the feed were constant at 40 percent by mass, 1.8 g s ml1 and 10 percent by volume carbon dioxide in a helium balance, respectively. a
b
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
Fig. 1. Representative XRD patterns of DE, CTAB-DE and the 10%/25%/40%TEPA/CTABDE absorbents.
cristobalite were observed at a 2Theta of 20.96 , 36.30 and 42.89 , which corresponded to JCPDS card no. 39e1425 and so a tetragonal structure with lattice parameters of a ¼ b ¼ 4.9709 Å and c ¼ 6.9180 Å. In addition, the diffraction peak of kyanite was seen at a 2Theta ¼ 27.96 , which is an acceptable match to that in JCPDS card no. 11e46 and so a triclinic structure with lattice parameters of a ¼ 7.1150 Å, b ¼ 7.8410 Å and c ¼ 5.5724 Å. The peaks at 2Theta of 12.41, 21.41, 24.97 and 34.97 corresponded to the diffraction peaks of kaolinite reported in JCPDS card no. 89e6538, and so was be ascribed as a triclinic crystal structure with lattice parameters of a ¼ 5.1535 Å, b ¼ 8.9419 Å and c ¼ 7.3906 Å. The diffraction peaks of hematite at a 2Theta of 33.15 and 40.86 were very similar to those in JCPDS card no. 33-0664, and so was defined as a rhombohedral structure with lattice parameters of a ¼ b ¼ 5.0376 Å and c ¼ 13.7489 Å. There was no shift in these diffraction peaks in the different adsorbents when DE was modified with CTAB (CTAB-DE) or when the CTAB-DE was loaded with TEPA (TEPA/CTAB-DE), showing that CTAB and TEPA had no effect on the crystal lattice of the DE, but rather were only modified at the surface of the DE. The FT-IR spectra of the different absorbents were evaluated in order to analyze the surface functional groups of the samples (Fig. 2). In addition, the spent 40%TEPA/CTAB-DE absorbent used at 80 C with or without 16% (v/v) water vapor in the feed was analyzed. The vibration bands of pure TEPA (Fig. 2Ia) showed a broad peak at 3348 cm1, assigned to the NeH antisymmetric stretching (NH2) mode of TEPA, and a peak shoulder at 3274 cm1 corresponding to the NeH symmetric stretching (NH and NH2) mode (Yang et al., 2012). Two overlapped peaks appeared at 2930 and 2825 cm1, ascribed to the CeH antisymmetric stretching (CH2) and stretching (NeCH2) vibrations of the TEPA chains, respectively (Yang et al., 2012; Yu et al., 2012a,b). The shoulder at 1647 cm1 was due to the ring skeleton stretching mode of cyclization among TEPA. The peak located at 1581 cm1 represented the NeH scissoring (NH2) of TEPA. The CeH deformation (CH2) of TEPA was observed at 1473 cm1 with a shoulder at 1433 cm1 (Hiyoshi et al., 2005). Additionally, the peak at 1310 cm1 with overlapped peaks at 1384 and 1350 cm1 was assigned to the CeH wagging (CH2) vibration of TEPA (Rusu et al., 2008; Yang et al., 2012; Yu et al., 2012a), and that at 1119 cm1 was ascribed to the CeN stretching (ReNeR) mode (Yang et al., 2012; Yu et al., 2012a). The vibration of CTAB in the wavenumber range of 4000e400 cm1 is shown in Fig. 2Ib. The wavenumber at 3022 and 3018 cm1 were assigned to the antisymmetric CeH stretching mode of the Nþ(CH3)3 group. The symmetric CeH stretching vibration was seen as a shoulder at 2956 cm1. The antisymmetric
197
and symmetric stretching of CH3-vibrational bands appeared at 2934 and 2861 cm1, respectively. The antisymmetric CeH stretching (CH2) mode at 2917 cm1 and symmetric CeH stretching (CH2) at 2851 cm1 were observed in the CeH stretching region (Venkataraman and Vasudevan, 2001). The band at 1487 cm1 was assigned to the antisymmetric NeCH3 scissoring, while those at 1472 and 1462 cm1 were the vibrational modes of the CH2 scissoring. Two peaks for CeC skeletal stretching mode were observed (962 and 937 cm1) with the CeN stretching vibration at 912 cm1. The band at 720 cm1 was assigned to the (CH2)n rocking mode, and the broad bands at 3418 and 1622 cm1 to the stretching and bending modes of adsorbed water molecules, respectively. The FT-IR spectra of DE (Fig. 2Ic) exhibited a broad band at 3418 cm1, attributed to the symmetric OeH stretching mode of H et al., 2002). The bonded adsorbed water molecules (Madejova band centered at 1622 cm1 is the HeOeH bending vibration of water molecules sorbed on the surface of DE (Xu et al., 2000). The bands at 3632 and 3704 cm1 were assigned to the OeH stretching vibrations of the structural OeH (or inner hydroxyl) groups , 2003; Xue et al., 2007) whilst the (Johnston et al., 1992; Madejova SieOeSi, SieO, AleOeAl and AleO metal oxides bond vibrated in the wavenumber range of 400e1200 cm1. The bands at 1110, 788 and 476 cm1 corresponded to antisymmetric stretching, symmetric stretching and bending vibrations of SieOeSi siloxane (Huang et al., 2007; Sheng et al., 2009; Sun et al., 2013), respectively. The SieO stretching vibration was observed at 1090 and 1030 cm1 (Xue et al., 2007). The AleOeSi stretching vibration was indicated at 800 cm1 for the DE sample and that at 700 cm1 was assigned to the AleO stretching vibration (Xue et al., 2007). The band at 540 cm1 was assigned to the SieOeAl (octahedral Al) , 2003). From the FT-IR spectra of the CTAB-DE (Fig. 2Id), (Madejova the spectra showed the vibration of both CTAB and DE, which indicated that DE was modified with CTAB. The 10%TEPA/CTAB-DE (Fig. 2Ie) expressed the characteristic vibration bands of TEPA, CTAB and DE, supporting that TEPA was loaded on the CTAB-DE via the method used in this work. When loading a higher (25 or 40% (w/w)) amount of TEPA on the CTAB-DE the intensity of the peaks that corresponded to TEPA increased (Fig. 2II). The FT-IR spectra of the spent 40%TEPA/CTAB-DE, used with or without the addition of 16% (v/v) water vapor in the feed at 80 C, are shown in Fig. 2III. Without the addition of water vapor in the feed, the vibration band of the carbonyl stretching of carbamate species was observed at 1658 cm1 (Hasib-ur-Rahmana et al., 2012) and bicarbonate species was observed at a wavenumber of 1410 cm1 due to the reaction among the amine groups of TEPA with the sorbed water molecules on DE and CO2. In the presence of added water vapor in the feed, the vibration bands of bicarbonate bands were found at 3395, 1623, 1410 and 1224 cm1, which were the vibrations of OH stretching, CO2 antisymmetric stretching, CO2 symmetric stretching, and COH bending, respectively (Garand et al., 2010). Additionally, the CeO stretching mode of imino-carbonate occurred at 1473 cm1 (Jackson et al., 2009). The change in the FT-IR absorption bands for CO2 sorption in the absence/presence of water was in agreement with Yang and He (Yang and He, 2014). The thermal stability of the different absorbents was evaluated by TGA, where Fig. 3 shows the TGA profile in terms of the mass loss as a function of the temperature. For pure TEPA, the mass loss was around 4.1 wt% at below 150 C, attributed to the moisture vaporization in TEPA. Decomposition of TEPA started at 180 C and was complete at 260 C. For DE the mass loss of 4 wt% from 30 to 200 C and a further 3.3 wt% from 250 to 497 C were assigned to the evaporation of adsorbed water and dehydroxylation of some associated SieOH groups on the external surface of the DE, respectively (Chaisena and Rangsriwatananon, 2004). For CTAB-DE, the slight mass loss (2.3 wt%) from 30 to 200 C corresponded to the
198
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
dehydration of physically adsorbed water, and the mass loss at 200e290 C was attributed to the partial decomposition of CTAB (Wang et al., 2007). Further increasing the temperature to 500 C caused the complete decomposition of CTAB. For the TEPA/CTAB-DE adsorbents with different TEPA loading levels, the TGA profiles showed two steps of mass loss. The first one, from 60 to 130 C with a total mass loss of 4.8 wt%, was attributed to the vaporization of moisture on the adsorbent. The second step, from 130 to 290 C with a total mass loss of 13.9, 27.5 and 41.4 wt% for the 10, 25 and 40%TEPA/CTAB-DE, respectively, was due to the decomposition of CTAB and TEPA. In addition, two mass loss steps were found in the
TGA curve of 40%TEPA/DE, but each step was shifted to a lower temperature range. This indicated that a higher thermal stability was achieved when modifying DE with CTAB before loading the TEPA. Regardless, the thermal stability behavior of the adsorbents was in an acceptable temperature range for this application. 3.2. Measurement of CO2 adsorption The CO2 adsorption capability of the TEPA/CTAB-DE adsorbents was investigated with different levels of TEPA loading, operating temperature and amount of water vapor in the feed stream.
Fig. 2. Representative FT-IR spectra of the synthesized absorbents. [I] The fresh (a) pure TEPA, (b) CTAB, (c) DE, (d) CTAB-DE and (e) 10%TEPA/CTAB-DE. [II] The 10%/25%/40%TEPA/ CTAB-DE. [III] The fresh (solid line) and spent (dashed line) 40%TEPA/CTAB-DE with and without the addition of 16% (v/v) water vapor in the feed at an operating temperature of 80 C.
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
Fig. 3. Representative TGA profiles of DE, CTAB-DE, 10%/25%/40%TEPA/CTAB-DE, 40% TEPA/DE and pure TEPA.
The influence of the TEPA loading level on the CO2 adsorption capacity of the TEPA/CTAB-DE adsorbent at an operating temperature of 70 C without any added water vapor in the feed stream is shown in Fig. 4a. The adsorption capacity of CTAB-DE was 58.1 mg g1 and this increased 1.17-, 1.32- and 1.59-fold after loading 10, 25 and 40% (w/w) TEPA, respectively, onto the CTAB-DE, reaching a CO2 adsorption capacity of 92.2 mg g1 at a 40% (w/w) TEPA loading. That increasing the TEPA loading level increased the CO2 adsorption capacity is due to the higher amount of available amine groups (from the TEPA molecules) on the surface of the CTAB-DE support to react with CO2 (Xu et al., 2002; Veneman et al., 2012) to form carbamates, as seen in the FT-IR analysis. Thus, the TEPA loading level can affect the capacity performance of the adsorbents. However, loading over 40% (w/w) TEPA on the CTAB-DE resulted in the formation of moist adsorbents during the preparation and so the CO2 adsorption capacity could not be measured. Accordingly, a TEPA loading level of 40% (w/w) was selected for further study. The CO2 adsorption capacity of the 40%TEPA/CTAB-DE adsorbent without any added water vapor in the feed stream increased 1.31-,
199
1.85- and 2.2-fold with increasing adsorption temperature at 60, 70 and 80 C, respectively, from 49.6 mg g1 at 50 C up to 109.2 mg g1 at 80 C (Fig. 4b). Increasing the temperature increased the kinetic energy of the CO2 molecules and so (in this temperature range) allows them to penetrate more easily into the TEPA particles in the adsorbent pores and accordingly make contact with more CO2 affinity sites (Xu et al., 2002; Yue et al., 2006). Therefore, an operating temperature of 80 C was selected for further studies. Generally, water vapor in the range of 8e20% (v/v) is one of the components in flue gas and other industrial gases. The CO2 adsorption capacity of the 40%TEPA/CTAB-DE at 80 C increased from 109.2 mg g1 with no added water vapor absence in the feed stream to 116.5 mg g1 and 129.7 mg g1 when adding 8 and 16% (v/v) water vapor, respectively (Fig. 4c). That the addition of water vapor in the feed promotes the CO2 adsorption is because without added water vapor one mole of CO2 molecules reacts with two moles of the primary (RNH2) and secondary (R2NH) amino groups in TEPA, leading to the formation of carbamate, as shown in Eqs. (1) and (2):
CO2 þ 2RNH2 4RNHCOO þ RNH3 þ
(1)
CO2 þ 2R2 NH4R2 NCOO þ RNH3 þ
(2)
The formation of carbamate species on the adsorbent was evident in the FT-IR results (Fig. 2III). However, when water vapor was added into the feed the OH groups of water were involved in the CO2 adsorption and bicarbonate formation, as supported by the increased bicarbonate levels seen in the FT-IR analysis. One mole of CO2 molecules reacts with one mole of the amine in the presence of H2O or hydroxyl groups to form bicarbonate species (Yue et al., 2006). When no water vapor was added some bicarbonate species were still formed, as shown in the FT-IR spectra, due to the presence of residual water in the adsorbent, as supported by the TGA results. However, the adsorption capacity could not be evaluated at levels of water vapor above 16% (v/v) due to slurry formation and blockage of the gas flow in the reactor. Accordingly, the addition of an appropriate amount of water vapor in the feed, in this case 16% (v/v), has a positive effect on the CO2 adsorption.
Fig. 4. Representative CO2 adsorption capacity as a function of (a) the TEPA loading level without any added water vapor in the feed and at an operating temperature of 70 C; (b) the operating temperature for 40%TEPA/CTAB-DE without any added water vapor in the feed and (c) the H2O concentration in the feed for a 40%TEPA/CTAB-DE adsorbent at 80 C. In all cases the W/F ratio was constant at 1.8 g s mL1 and the CO2 concentration in the feed was constant at 10% (v/v) in a He balance.
200
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
3.3. Experimental design 3.3.1. Screening factors using a 23 factorial design With respect to the CO2 adsorption capacity of the adsorbent, which factors and their interactions had a significant effect was not determined, and thus the optimal condition for CO2 adsorption was not ascertained in Section 2. Accordingly, a full 23 factorial design with three central points was employed to determine the important factor(s) and their interaction(s) on the adsorption capacity (Table 1). Statistical analysis at a 95% confidence interval was used to evaluate the data. The normal probability plot of the effects (Fig. 5a) revealed that the operating temperature, TEPA loading level and the water concentration in the feed, plus the TEPA loading level-water concentration interaction had a normal probability of more than 50%, inferring that they had a synergistic effect on the adsorption capacity. That the adsorption capacity increased as these factors increased from a low level to a high level is in accordance with the CO2 adsorption measurement. Only the interaction between the operating temperature and water concentration had a significantly less than 50% probability, indicating that this interaction is antagonistic on the CO2 adsorption capacity.
The Pareto chart for the T-value of jEffectj compared to the Bonferroni limit (5.068 in this case) revealed that only the interaction between the operating temperature and water concentration had a significant negative effect on the CO2 adsorption capacity (Fig. 5b), whilst the three main factors and the TEPA loading levelwater vapor interaction were all positive. Thus, the direction of effect of each factor on the CO2 capacity was similar to that expressed by the normal probability plot. Moreover, the Pareto chart revealed that there is a curvature in the relationship among the adsorption capacity, the important factors and their interaction. In agreement with the above analysis, the ANOVA revealed that the three main factors, and the TEPA loading-water concentration and operating temperatureewater concentration interactions all had a significant effect on the CO2 adsorption capacity (Table 3). Moreover the percentage contribution (related to the total sum of squares) suggested their relative importance was ranked (highest to lowest) as the TEPA loading level >>> TEPA loading level-H2O concentration interaction > temperature > H2O concentration > temperatureeH2O concentration interaction. This ranked result and the relative importance of each factor and their interactions is similar to that obtained in the normal probability plot and the Pareto chart. The R2 of 0.9990 indicates a good agreement between the experimental and estimated values for the CO2 sorption capacity. The randomized error variable in the experimental results was rather small, and the Adj. R2 (0.9977) was very close to that of R2, indicating that a requisite term was included in the model. The adequate precision (76.160) was much greater than 4, and so the selected model had a moderate performance ability in evaluating the CO2 sorption capacity and was effective at navigating the design space. In particular, only 1.61% of the data points were dispersed around the mean, as indicated by the coefficient of variation (C.V.), with no significant terms likely to be missing from the model. The regression equation (in terms of coded factors) of the CO2 adsorption capacity (Qe) could be expressed by Eq. (3),
Qe ¼ þ84:50 þ 8:13A þ 26:83B þ 7:41C 6:42AC þ 8:73BC (3) The positive sign of the coefficient estimate of the operating temperature (A), TEPA loading level (B), water concentration in the feed (C), and the TEPA loading level-H2O concentration interaction (BC), and the negative sign of the operating temperatureewater concentration interaction (AC), in the regression equation are in agreement with the adsorption capacity results (Fig. 4aec). From the ANOVA, the indicated curvature in the model suggested that the average response value did not correspond to the response value at the central point for the factors studied, in accord with the curvature shown in the results of the Pareto chart. Accordingly, a RSM analysis was employed to achieve a better perception.
Fig. 5. Representative (a) normal probability plot of the effects for the full 23 factorial design and (b) Pareto chart of the factors for the 23 factorial design.
3.3.2. Optimization by FCCCDeRSM analysis Due to the extreme contribution of the TEPA loading effect on the response, which was from 9.4- to 17.5-fold more than the contribution of the other effects (as obtained from the factorial design in Section 3.3.1) the other effects would be hidden by the TEPA loading effect in a RSM analysis. Accordingly, the value of the TEPA loading at 40% (w/w) obtained from the results of the CO2 adsorption measurement and the factorial design was set as a constant in the FCCCD-RSM. Based on the factorial design results, the operating temperature and water vapor concentration in the feed were then varied to determine the optimum conditions for the adsorption capacity (see Table 2). The CO2 adsorption capacity response surface equation in terms of coded factors could be expressed by Eq. (4),
P. Pornaroonthama et al. / Journal of Environmental Management 157 (2015) 194e204
201
Table 3 Analysis of variance (ANOVA) of the significant terms for the 23 factorial design with three central points. Source
Sum of squares
DF
Model A-Temperaturea B-TEPA loadinga CeH2O concentrationa ACb BCb Curvature Residual Lack of Fit Pure Error Total R-Squared Adj. R-Squared
7667.97 528.79 5759.66 439.51 329.98 610.04 163.58 7.78 2.32 5.46 7839.33 0.9990 0.9977
5 1533.59 1 528.79 1 5759.66 1 439.51 1 329.98 1 610.04 1 163.58 4 1.95 2 1.16 2 2.73 10 e C.V. percent Adeq. Precision
a b c
Mean square
F-value
P-valuec
Percent contribution
788.14 271.75 2959.98 225.87 169.58 313.51 84.06 e 0.43 e e 1.606 76.10
