Partially Calcined Gismondine Type Silicoaluminophosphate SAPO

Partially Calcined Gismondine Type. Silicoaluminophosphate SAPO-43: Isopropylamine. Elimination and Separation of Carbon Dioxide, Hydrogen. Sulfide, a...
0 downloads 0 Views 190KB Size
Langmuir 2003, 19, 2193-2200

2193

Partially Calcined Gismondine Type Silicoaluminophosphate SAPO-43: Isopropylamine Elimination and Separation of Carbon Dioxide, Hydrogen Sulfide, and Water Arturo J. Herna´ndez-Maldonado and Ralph T. Yang* Department of Chemical Engineering, University of Michigan, Ann Arbor, Michigan 48109-2136

Daniel Chinn and Curtis L. Munson Separation Technology, Chevron-Texaco, Richmond, California 94802-0627 Received August 17, 2002. In Final Form: January 6, 2003 Selective removal of carbon dioxide and hydrogen sulfide from other gases is usually difficult to achieve because most of the available sorbents will also adsorb other unwanted molecules. SAPO-43 is a microporous sorbent with appropriate dimensions and characteristics for these types of separations, but previous studies have shown that, because of low thermal stability, it is not suitable. This study discusses a procedure developed to partially remove the organic moiety used for synthesis of SAPO-43 while preserving the framework. The thermal process to remove the template (isopropylamine) results in decomposition of the species, which is believed to be a result of a Hofmann elimination process. Afterward, the partially calcined silicoaluminophosphate can be used as a selective adsorbent to separate molecules like CO2, H2O, and H2S while excluding molecules such as N2, CH4, and O2. In general, it was found that SAPO-43 has adsorption capacities of 1.1, 4.93, and 2.52 mmol/g for CO2, H2O, and H2S, respectively, at 25 °C and atmospheric conditions.

Introduction Concern about removal of gases like carbon dioxide and hydrogen sulfide by more efficient ways is growing every day. There are many applications that require high amounts of energy content that is usually hindered by the presence of gases such as carbon dioxide and hydrogen sulfide as in the case of raw natural gas streams. Also, there are some possible global warming implications. Processes like gas scrubbing, separations by membranes,1,2 absorption, and distillation are useful somehow in achieving removal but have drawbacks that include low selectivity and/or energy intensive problems. There are approximately 40 000 columns in operation in the United States, of which distillation is used to make 90% or more of all chemical separations. These consume an equivalent of 1.2 million barrels per day of crude oil.3 For gas separations, the process gets more energy intensive, requiring liquefaction of the mixture before entering the distillation column (cryogenic separation). Synthetic zeolites, for instance, have shown some attractive properties that can be exploited in such separations. Some of these molecular sieves are used in applications that have been around for decades, such as preferential adsorption of nitrogen over oxygen4-6 and other interesting separations. But even molecular sieves have limitations in their * Corresponding author: Tel (734) 936-0771; Fax (734) 763-0459; e-mail [email protected]. (1) Watanabe, H. J. Membr. Sci. 1999, 154, 121. (2) Poshusta, J. C.; Tuan, V. A.; Pape, E. A.; Noble, R. D.; Falconer, J. L. AIChE J. 2000, 46, 779. (3) Humphrey, J. L.; Keller, G. E. Separation Process Technology; McGraw-Hill: New York, 1997. (4) Barrer, R. M. Proc. R. Soc. London 1938, A167, 392. (5) Yang, R. T. Gas Separations by Adsorption Processes; Butterworth: Boston, 1987; reprinted by Imperial College Press: London and World Scientific Publishing Co.: River Edge, NJ, 1997.

applicability to various separations because, for example, selectivity is not greatly demarcated. This is particularly true for separations that involve molecules like carbon dioxide, nitrogen, and methane.7-10 This study focuses on the use of highly selective synthetic microporous materials, specifically, a silicoaluminophosphate of the Gismondine family called SAPO-43. Silicoaluminophosphates (SAPOs) are crystalline microporous materials formed by silicon, aluminum, phosphorus, and oxygen atoms in tetrahedral coordination.11-13 This arrangement results in formation of uniform pore channels with molecular dimensions. SAPOs have a framework with a net charge that varies depending upon how the silicon substitution into an aluminophosphate group is achieved. That is, if silicon substitutes for aluminum, phosphorus, or both, the resulting net charge will be +1, -1, or 0, respectively.14 Studies have shown that usually the second and third substitution mechanisms are present during the crystallization process.15-17 The (6) Talu, O.; Jianmin, Li.; Kumar, R.; Mathias, P. M.; Moyer, J. D.; Schork, J. M. Gas Sep. Purif. 1996, 10, 149. (7) Habgood, H. W. Can. J. Chem. 1964, 42, 2340. (8) Valenzuela, D. P.; Myers, A. L. Adsorption Equilibrium Data Handbook; Prentice-Hall: Englewood Cliffs, NJ, 1989. (9) Hernandez-Huesca, R.; Diaz, L.; Aguilar-Armenta, G. Sep. Purif. Technol. 1999, 15, 163. (10) Van Der Vaart, R.; Huiskes, C.; Bosch, H.; Reith, T. Adsorption 2000, 6, 311. (11) Lok, B. M.; Messina, C. A.; Patton, R. L.; Gajek, R. T.; Cannan, T. R.; Flanigen, E. M. US Patent 4,440,871, 1984. (12) Lok, B. M.; Vail, L. D.; Flanigen, E. M. US Patent 4,758,419, 1988. (13) Hartmann, M.; Kevan, L. Chem. Rev. 1999, 99, 635. (14) Djieugoue, M. A.; Prakash, A. M.; Kevan, L. J. Phys. Chem. B 1999, 103, 804. (15) Hartmann, M.; Azuma, N.; Kevan, L. J. Phys. Chem. 1995, 99, 10988.

10.1021/la026424j CCC: $25.00 © 2003 American Chemical Society Published on Web 02/07/2003

2194

Langmuir, Vol. 19, No. 6, 2003

Figure 1. Gismondine framework viewed along [100] and 8-ring viewed along [100]. Source: Structure Commission of the International Zeolite Association (IZA).

former requires presence of counterbalance species such as cations and/or protons. SAPO-43, as well as other Gismondine members, shows a pore opening formed by eight oxygen rings with dimensions depicted in Figure 1. This crystalline material was first synthesized as a pure phase by Akporiaye et al.18 after modifying and optimizing a procedure followed by Helliwell et al.,19 in which a chromium-substituted version was obtained. This material has a three-dimensional framework with an orthorhombic unit cell (a ) 14.158 Å, b ) 14.441 Å, and c ) 10.062 Å) as determined by Helliwell et al. It was believed that the use of this material for separations was impossible because of lower thermal stability, resulting in structure collapsing.18 The following discussion shows a procedure to remove the template partially and preserve some adsorption capacity. Experimental Section Sorbent Synthesis. SAPO-43 was synthesized following procedures available elsewhere.18 Reactants used were aluminum isopropoxide (Acros Chemicals), o-phosphoric acid (85 wt %, Fisher), colloidal silica (40 wt % LUDOX-AS, Aldrich), isopropylamine (Aldrich), and deionized water. The initial gel oxide composition was 1.0Al2O3:1.0P2O5: 0.7SiO2:10iPrNH2:50-70H2O. This gel was placed inside a Teflon-protected 450 mL autoclave and sealed. The mixture was then heated to 160 °C under autogenous pressure for 120 h. The final product was recovered by decantation, filtered, and washed with copious amounts of deionized water. The solid was dried at 100 °C for a minimum of 24 h. Some other possible impurities were removed by heating the samples in air at 20 °C/min up to 250 °C for 20 h. All phases were verified using X-ray powder diffraction. (16) Hartmann, M.; Azuma, N.; Kevan, L. In Zeolites: A Refined Tool for Designing Catalytic Sites; Bonneviot, L., Kaliaguine, S., Eds.; Elsevier: New York, 1995; p 335. (17) Prakash, A. M.; Hartmann, M.; Kevan, L. J. Chem. Soc., Faraday Trans. 1997, 93, 1233. (18) Akporiaye, D. E.; Dahl, I. M.; Mostad, H. B.; Wendelbo, R. Zeolites 1996, 17, 517. (19) Helliwell, M.; Kaucic, V.; Cheetham, G. M. T.; Harding, M. M.; Kariuki, B. M.; Rizkallah, P. J. Acta Crystallogr. B 1993, 49, 413.

Herna´ ndez-Maldonado et al. X-ray Powder Diffraction. X-ray powder diffraction patterns were obtained using a Rigaku rotating anode unit equipped with a Cu KR target operating at 40 kV and 100 mA. For hightemperature X-ray analysis a Rigaku 2 kW standard X-ray generator with a θ-θ diffractometer was used. This unit has a high-temperature (1400 °C) stage for use with a controlled atmosphere. All samples were scanned at 2 deg/min and using steps of 0.02 deg. SEM-EDAX Analysis. Electron microscopy data were obtained using a Philips XL30 FEG SEM at The University of Michigan Electron Microbeam Analysis Lab (EMAL). The unit has a UTW Si-Li solid state X-ray detector with an integrated energy dispersive analysis by X-rays (EDAX) system, which can be used for bulk elemental analysis. For the EDAX analysis the samples were not coated with a layer of carbon since the tests included this element. Images were obtained following standard procedures for low conductive samples. Adsorption Equilibrium Isotherms, Weight Loss Profiles, and Uptake Rates. Adsorption tests were done using a Micromeritics ASAP 2010 static volumetric unit and a Cahn TGA microbalance. For the gravimetric method, isotherms and uptake rates were obtained following the procedure described elsewhere.20,21 Adsorbate gases used were CO2 (Coleman Grade, Metro Welding, 99.99%), CH4 (Matheson Grade, 99.99%), N2 (Metro Welding, 99.99%), O2 (Extra-Dry Grade, Metro Welding, 99.99%), and He (Ultra High Purity Grade, Metro Welding, 99.995%). For water vapor adsorption tests helium was saturated with deionized water using gas bubblers. All samples were treated in either vacuum or helium at 325 °C before adsorption experiments. Adsorption/desorption runs were performed at temperatures ranging from 25 to 75 °C. Desorption Activation Energies. Activation energies for desorption were estimated using gravimetrical thermal analysis (TG/DTG) data and Redhead’s equation for first-order kinetics

( )

ln

( )

Ed Tm2 Ed ) + ln β RTm AR

(1)

where Tm is the temperature at peak maximum, β is the heating rate, Ed is the activation energy, and A is a preexponential factor. The procedure on how to use DTG and Redhead’s model can be found elsewhere.22 Isosteric Heat of Adsorption and Diffusion Rate. Pure component equilibrium adsorption data were fitted using Langmuir-Freundlich (L-F) and Dubinin-Astakhov (D-A) models.23 For Langmuir-Freundlich isotherms the equation has three adjustable parameters:

θ)

q BP1/n ) q0 1 + BP1/n

(2)

where θ is fractional coverage, q0 is the saturated adsorbed amount, B is the Langmuir constant, P is pressure, and n is a constant. For Dubinin-Astakhov isotherms the equation has three adjustable parameters:

[(

)]

Ps V ) exp - C ln V0 P

n

(3)

where

C)

RT βE0

(4)

V is the volumetric adsorbed amount, V0 is the volumetric saturated adsorbed amount, Ps is the saturated vapor pressure (20) Ackley, M. W.; Yang, R. T. AIChE J. 1991, 37, 1645. (21) Yeh, Y. T. Diffusion and Adsorption of Gases in Molecular Sieves. Ph.D. Dissertation, University of New York at Buffalo, Buffalo, NY, 1989. (22) Cvetanovic, R. J.; Amenomiy, Y. Catal. Rev.sSci. Eng. 1972, 6, 21. (23) Do, D. D. Adsorption Analysis: Equilibria and Kinetics; Imperial College Press: London, 1998.

Partially Calcined Gismondine Type SAPO-43

Langmuir, Vol. 19, No. 6, 2003 2195

Figure 2. X-ray powder diffraction (XRD) patterns of asynthesized SAPO-43.

Figure 3. Scanning electron micrograph of SAPO-43. of the adsorbate, T is temperature, β is the affinity coefficient, n is a constant, and E0 is the characteristic energy of the sorbent. The conversion of molar to volumetric adsorbed amount was performed using liquid molar volumes at the corresponding temperature. Isosteric heats of adsorption were calculated using the Clausius-Clapeyron equation and pure component isotherm models obtained at different temperatures. Thus, the isosteric heat of adsorption is given by

∆Hiso ) -R

d ln P | d(1/T) θ)const

(5)

For rate of uptake measurements, diffusion time constants (D/ R2) were estimated by fitting data with a phenomenological model for spherical particles24

mt m∞

)1-

6





1

π2n)1n2

(

exp -

)

n2π2Dt r2

(6)

where mt is the uptake at any time t and m∞ is the uptake as t approaches infinity. This model assumes that the sorbate pressure remains constant following the initial step change.

Results and Discussion X-ray Diffraction and SEM Results. Figure 2 shows the XRD pattern for our SAPO-43. The pattern matches well those in the literature and corresponds to a structure similar to that of Gismondine, whose framework was shown before. The sample’s crystal size is about 5 µm (see Figure 3), which is smaller than the one previously reported by Akporiaye et al. (about 20-40 µm). Steps in the synthesis that could influence the crystal size include addition of inorganic cations, especially if this is done after (24) Ruthven, D. M. Chem. Eng. Sci. 1992, 47, 4305.

Figure 4. Transient XRD patterns of SAPO-43 calcined in situ at 325 °C (10 °C/min) under helium.

seeding the reacting gel.25 Our synthesis procedure included some trace amounts of sodium hydroxide (used to stabilized the colloidal silica), and some in situ seeding because the gel was aged for about 30 min at room temperature. For this study, however, a small crystal size is advantageous because intraparticle mass transfer limitation is diminished. Before proceeding with adsorption tests, it was necessary to remove the organic moiety (isopropylamine) from within the channels of SAPO-43. Initially, the sample was calcined in a quartz reactor under air atmosphere at a temperature of 400 °C for 2-20 h at a heating rate of 10 °C/min. The calcination temperature corresponds to the sorbent’s thermal stability limit as reported by Akporiaye et al.18 Afterward, adsorption tests showed little or no carbon dioxide uptake capacity, indicating either collapsing of the structure or no template removal. CO2 was chosen because of its small collision diameter, which is about 3.3 Å and should be allowed to enter the sorbent’s framework. Figure 4 shows XRD patterns obtained while heating the sample in situ at 325 °C under a helium atmosphere. The temperature was risen from room to a target value at a rate of 10 °C/min. Other heating rates were not used for this analysis because of limitations encountered with the X-ray diffraction unit. The patterns show peak shifting occurring after 1.25 h of sample exposure to 325 °C, indicating that the pores or channels are shrinking after removal/desorption of the template. Eventually, the sorbent structure collapses as indicated by the absence of almost all peaks in the XRD pattern at 2 h. Thus, even temperatures below 400 °C will result in destruction of the sample, although this could change at different heating rates and/or atmospheres. Akporiaye et al. did a similar test in their study but with air as the atmosphere. It should be mentioned that the sample has a brownish color at the end of the treatment, which is a typical observation found in some zinc-gallium phosphates with the Gismondine structure,26 but could also indicate coke formation. In general, Figure 4 shows clearly why CO2 was not adsorbed during the preliminary adsorption tests. Thermal (TG/DTG) and Redhead Analyses. Further analysis of template removal using gravimetrical thermal analysis (Figure 5) shows two major weight losses, as reported previously for MAPO-43 (GIS) and SAPO-43 (GIS).18,27 Also, as the heating rate is increased, desorption (25) Szostak, R. Molecular Sieves: Principles of Synthesis and Identification, 2nd ed.;Blackie Academic and Professional, Thomson Science: New York, 1998. (26) Chippindale, A. M.; Cowley, A. R.; Peacock, K. J. Microporous Mesoporous Mater. 1998, 24, 133. (27) Akolekar, D. B.; Kaliaguine, S. Microporous Mater. 1994, 2, 137.

2196

Langmuir, Vol. 19, No. 6, 2003

Figure 5. SAPO-43 TG profiles at different heating rates. Calcination done under helium atmosphere.

regions are shifted to higher temperature areas. A similar behavior was observed for Amicite, parent of Gismondine, under similar heating rates conditions.28-30 An important observation made in these previous studies is that, upon heating treatment, Amicite’s framework also goes through distortion after dehydration, which could be the case also for SAPO-43 (Figure 4). It is also known that calcium containing Gismondine goes through distortion upon dehydration, which, as a result, diminishes its rehydration capacity from 21 to about 15.6 wt %. Apparently, the usage of higher heating rates leaves some of the “template product” inside the channels, which may lead to think that complete removal is achieved by not only physical desorption but also decomposition as proposed by Akolekar and Kaliaguine in their MAPO-43 work.27 They stated that template elimination in MAPO43 occurs via a two-stage process: elimination of the template via desorption followed immediately by decomposition of the remaining occluded species and, then, slow desorption of the decomposition products. The latter occurs at temperatures greater than 620 °C. Most zeolitic templates are eliminated following a similar process, which is known as Hofmann-type degradation. Therefore, since SAPO-43 is parent to MAPO-43 and based on Figure 5, desorption activation energies for isopropylamine elimination should indicate whether a reaction is occurring during the elimination process. Apparent activation energies were estimated using Redhead’s equation for first-order desorption. Figure 6 shows a differential weight change plot from TG analysis depicting maximum peak temperatures at which apparent desorption of species occurs. Table 1 shows activation energies obtained from the Redhead equation (Figure 7). Fesenko et al. studied the desorption of isopropylamine from HY (FAU) zeolites using TPD and calculated apparent activation energies for the process.31 Their analysis showed similar weight losses regions, and they reported apparent desorption energies of 60-75 and 104-140 kJ/ mol for the desorption and Hofmann elimination of isopropylamine, respectively. Activation energies for isopropylamine desorption/elimination in H-ZSM-532 were also in the same order of magnitude as in the case of HY. (28) Alberti, A.; Hentschel, G.; Vezzalini, G. Neues Jahrb. Mineral., Monatsh. 1979, 481. (29) Gottardi, G.; Galli, E. Natural Zeolites; Springer: Berlin, 1985. (30) Vezzalini, G.; Alberti, A.; Sani, A.; Triscari, M. Microporous Mesoporous Mater. 1999, 31, 253. (31) Fesenko, E. A.; Barnes, P. A.; Parkes, G. M. B.; Brown, D. R.; Naderi, M. J. Phys. Chem. B 2001, 105, 6178. (32) Parrillo, D. J.; Gorte, R. J. J. Phys. Chem. 1993, 97, 8786.

Herna´ ndez-Maldonado et al.

Figure 6. DTG profile of SAPO-43 as a function of heating rate. Refer to Figure 5 for TG profiles.

Figure 7. Plot of ln(Tm2/β) vs TmSAPO-43.

1

for template removal of

Table 1. Redhead Analysis Data; Refer to Figure 6 for DTG Data peaks region

β (K/min)

Tm (K)

Ed (kJ/mol)

A (s-1)

1

5 10 15 20 5 10 15 20

373.15 390.15 400.15 414.15 633.15 650.15 664.15 683.15

59.54

8.58 × 105

140.68

1.41 × 109

2

Despite the fact that Fesenko et al. and others did the desorption analysis for zeolites structures different than GIS, our results agree extremely well with their findings, which allows to think that the residual weight observed in our TG analyses at different heating rates belong to products from decomposition of isopropylamine that are trapped within the solid’s channels. Perhaps these products and unreacted isopropylamine are responsible for keeping the structure from collapsing, at least partially. Akolekar and Kaliaguine’s thermal analysis for MAPO43 showed that template decomposition products are probably released at a temperature higher than 620 °C, which is already past the point of our sorbent’s thermal stability. At 325 °C those byproducts surely still remain inside the sorbent voids. As an additional comment, it should be mentioned that Fesenko et al. did not observe any isopropylamine reaction in NaY, indicating this that the process occurs in the presence of acid sites (i.e., HY). This may lead to assume that only a small portion of the sodium (traces) present in the reacting gel during the hydrothermal synthesis were

Partially Calcined Gismondine Type SAPO-43

Langmuir, Vol. 19, No. 6, 2003 2197

Table 2. Elemental Analysis Obtained by EDAX for SAPO-43 after Treatment at Different Heating Rates under a Helium Atmospherea heating rate (°C/min)

C (wt %)

N (wt %)

O (wt %)

Na (wt %)

Al (wt %)

Si (wt %)

P (wt %)

Si/Al

P/Al

C/Al

N/Al

5 10 20 40 av lit.b

6.09 3.99 4.08 5.95 5.03 17.99

5.01 4.83 6.09 4.84 5.19 6.99

41.46 43.57 38.53 42.47 41.51 8.65

0.39 0.40 0.43 0.37 0.40 N/A

24.26 22.17 19.49 22.77 22.18 29.18

8.60 9.12 8.42 8.46 8.65 12.48

14.18 15.90 14.00 14.76 14.71 19.68

0.35 0.41 0.43 0.37 0.39 0.43

0.58 0.71 0.72 0.65 0.66 0.67

0.25 0.18 0.21 0.26 0.22 0.62

0.21 0.22 0.31 0.21 0.23 0.240

a

The calcination temperature was 325 °C. b Akporiaye, D. E.; Dahl, I. M.; Mostad, H. B.; Wendelbo, R. Zeolites 1996, 17, 517. Table 3. Single-Component Adsorption Ratios for Different Sorbents at 1 atm adsorbed amounts (mmol/g) sorbent

temp (°C)

CO2

SAPO-43 BPL (activated carbon)a ASC (activated carbon)a fiber carbon (activated)a NUXIT-AL (activated carbon)a 4A zeolite SAPO-5b

25 30 30 25 20 25 32

1.07 1.50 1.70 2.50 2.30 3.55 ∼0.75

CH4 0.16 1.10 0.80 0.80 ∼0.25

single-component ratios (mol/mol)

N2

CO2/CH4

CO2/N2

0.07 0.35 0.25

6.68

15.28 4.29 6.80

2.27 2.88 4.44 3.00

a Valenzuela, D. P. and Myers, A. L. Adsorption Equilibrium Data Handbook, Prentice Hall, New Jersey, 1989. b Approximated from isotherms calculated by Choudhary, V. R.; Mayadevi, S. Langmuir 1996, 12, 980.

included in the framework, and thus our sorbent should be initially charge balanced by isopropylammonium cations. After the template goes through the elimination process, proton sites should emerge eventually. EDAX Analysis. Table 2 shows elemental analysis obtained by SEM-EDAX. The samples were heated to 325 °C for 1 h at different heating rates. The results show there are some traces of carbon left after treating the samples at different heating rates under helium. This carbon residue has to be due to the presence of isopropylamine and/or the products from its decomposition, such as propylene. Furthermore, since there is detection of elemental nitrogen, this leads to think that the products include also ammonia or other amines produced from the oligomerization of isopropylamine. The detected amount of nitrogen as a function of calcination heating rate seems unchanged, thus indicating this that the final temperatures were not high enough to release byproducts. Also shown in Table 2 are weight ratios of the elements present in SAPO-43. These ratios agree very well with those in the literature, and major differences can be found in the carbon-to-aluminum ratio, but this is due to the calcination process under discussion. Adsorption Isotherms, Uptake Rates, and Heat of Adsorption. After treating the sample at a heating rate of 20 °C/min up to 325 °C (in either vacuum or helium atmosphere), CO2 was allowed to contact the sorbent. This time the uptake amount was considerable, slightly higher than 1 mmol/g, and the isotherm shows a very steep slope at low pressures (see Figure 8). Hence, partial removal of the template and/or its products allows the sorbent framework to be preserved to some extent. Also shown in Figure 8 are pure component isotherms for methane, nitrogen, and oxygen. Methane (kinetic diameter σ ) 3.8 Å) does adsorb more than oxygen (σ ) 3.46 Å) and nitrogen (σ ) 3.64 Å), but this could be due to its high polarizability value and the low concentration of sodium cations measured by EDAX (Table 2), which hinders the possibility of sorbent interaction with a nitrogen molecule quadrupole. Nonetheless, methane uptake is negligible when compared to carbon dioxide adsorption. In general, these gases are almost excluded from the voids of the crystals, which agrees well with the Gismondine free apertures

Figure 8. Pure component adsorption isotherms of selected gases at 25 °C in SAPO-43.

dimensions reported elsewhere.33 Free apertures could be obstructed by cations upon dehydration, but this behavior is not expected in our sorbent. CO2 and CH4 adsorption studies on SAPO-5,34 which has larger pore size, have shown that both species are allowed to enter the framework, but with a much lower selectivity. This also evidences the presence of a steric effect in SAPO-43 for the molecules mentioned previously. Table 3 shows a comparison of adsorption selectivities for different adsorbents. Further adsorption experiments with other gases were done to follow the extent of possible separation applications for SAPO-43, such as the purification of natural gas. Pure component isotherms show that both hydrogen sulfide and water vapor are also allowed to adsorb on SAPO-43 (see Figures 9-11). Results show that the maximum adsorbed amounts under the conditions tested here follow the order H2O > H2S > CO2. According to collision/kinetic data, H2S should show lower adsorption capacity than that of CO2 as the former should face resistance when passing through the channel windows. But the H2S molecule has also a dipole moment that could enhance sorbent-adsorbate interactions. This behavior will be (33) Breck, D. W. Zeolite Molecular Sieves; Wiley: New York, 1973; reprinted by Krieger, Malabar, FL, 1984. (34) Choudhary, V. R.; Mayadevi, S. Langmuir 1996, 12, 980.

2198

Langmuir, Vol. 19, No. 6, 2003

Herna´ ndez-Maldonado et al.

Table 4. Parameters for the L-F and D-A Isotherms Models at Different Temperatures L-F equation molecule

temp (°C)

q0 (mmol/g)

CO2

25 50 75 25 75 25 50

2.12 1.51 1.23 9.97 6.33 3.97 2.25

H2O H2S

B

(atm-n) 0.99 1.11 1.01 3.02 5.16 1.54 4.61

D-A equation n 2.26 1.57 1.35 3.06 2.23 3.13 2.38

V0

(cm3/g)

0.079 0.082 0.104 0.089

C

n

Vm (cm3/mol)

Ps (atm)

0.240 0.201 0.125 0.108

1.42 1.63 2.50 2.58

16.03 17.36 41.11 42.78

0.03 0.38 20.57 58.13

Table 5. Isosteric Heats of Adsorption;a Refer to Figures 10, 11, and 13 for Adsorption Isotherms molecule

Hiso (kJ/mol)

CO2 H2O H2Sb

60.7-35.59 (0.1-0.55) 78.12-35.59 (0.1-3.0) 59.47-18.78 (0.1-1.0) 18.78-56.50 (1.0-1.85)

a Values in parentheses indicate adsorbed amounts (mmol/g) used for calculation. L-F isotherm models were used for calculations. b Hydrogen sulfide shows a minimum at about 0.1 mmol/g.

Figure 9. Pure component isotherms for hydrogen sulfide and carbon dioxide on SAPO-43 at 25 °C.

Figure 10. Hydrogen sulfide adsorption isotherms on SAPO43 at different temperatures.

Figure 11. Water vapor adsorption isotherms on SAPO-43 at different temperatures.

discussed later in light of heat of adsorption data, too. For the case of water vapor adsorption at 25 °C the uptake capacity is lower than expected, being only 8 wt % of dehydrated sorbent. As seen in the TG/DTG patterns, the capacity should be close to 18% instead. Again, this indicates that there is some void volume occupied by the template decomposition products, or part of the structure

was damaged, as in the case of calcium containing Gismondine.33 Tables 4 and 5 show isotherm models parameters and isosteric heats of adsorption, respectively. The adsorption data were fitted using the L-F and D-A models for all cases, except carbon dioxide. Use of these isotherm models indicates that the adsorption mechanism for the aforementioned gases in partially calcined SAPO-43 follows a pore-filling mechanism. For H2O and H2S it was found that the D-A equation was more representative of the adsorption behavior, in particular at higher and lower pressures, but for CO2 the L-F model fit was excellent. However, the isosteric heats of adsorption calculated here were obtained using the L-F model because it is more appropriate for the pressure range under study.23 The isosteric heats of adsorption values calculated here are higher than expected, indicating this the presence of additional interactions. For instance, CO2 adsorption on SAPO-5 produces an isosteric heat of ca. 20 kJ/mol at 0.1 mmol/g,34 which is more than 50% lower than the value reported here for SAPO-43. The difference could be due to interactions of the carbon dioxide molecules with the byproducts of the template elimination or the template itself. Satyapal et al. calculated the heat of adsorption of CO2 on solid-based amine sorbents,35 and they reported an average value of ca. 95 kJ/mol, which is higher than most zeolites cases. This could lead one to think CO2 is interacting with some SAPO-43 entrapped isopropylamine residue or just other amines produced during the Hofmann elimination process. Also, despite the fact that H2O and H2S both have dipole moments and CO2 does not, the heat of adsorption values at low coverage for this molecule should be high as a result of its quadrupole. The large isosteric heats of adsorption values for water were expected since most silicoaluminophosphates are hydrophilic. In terms of observed trends, it was found that for both CO2 and H2O there is an increase in heat of adsorption at low loadings followed by a uniform decrease at higher loadings. But for H2S, the heat of adsorption has a minimum (see Table 5). The latter is typical of adsorbateadsorbate interactions, which also explains why the adsorption capacity for this gas is larger than expected. Table 6 shows CO2, H2O, and H2S diffusion time constants obtained at different pressure steps. Carbon (35) Satyapal, S.; Filburn, T.; Trela, J.; Strange, J. Energy Fuels 2001, 15, 250.

Partially Calcined Gismondine Type SAPO-43

Figure 12. Fractional uptakes of various gases in SAPO-43 at 25 °C.

Figure 13. Adsorption and desorption isotherms for carbon dioxide on SAPO-43 at different temperatures. Table 6. Diffusion Time Constants of Carbon Dioxide, Hydrogen Sulfide, and Water Vapor in SAPO-43 at 25 °C adsorbate carbon dioxide hydrogen sulfide water vapor

pressure change (atm)

diffusion time constant D/R2 (s-1)

0 f 0.15 0.15 f 0.4 0.4 f 0.6 0 f 0.2 0.2 f 0.4 0.4 f 0.6 0 f 0.03

2.68 × 10-4 4.72 × 10-4 7.34 × 10-4 6.06 × 10-5 7.03 × 10-5 15.9 × 10-5 2.03 × 10-4

dioxide diffuses faster than water vapor and hydrogen sulfide, which could be suitable for time-dependent operations such as PSA (pressure swing adsorption.) These data also support the fact that H2S is quite large to pass freely through the pores of SAPO-43; thus, the large uptake observed in the isotherms is due to other interactions that will eventually prevail over kinetic effects. This kinetic behavior is depicted in Figure 12. Sorbent Regeneration. Carbon dioxide adsorption is reversible as shown in Figure 13, which is suitable for multiple PSA cycles. Nevertheless, water vapor and hydrogen sulfide are not, requiring the sorbent to be regenerated for further use. The regeneration experiments were done also in a TGA unit using helium saturated with water vapor (about 3%) and 1% H2S also in helium, respectively, and about 10 mg of sorbent. For the water vapor case (Figure 14) a temperature of about 180 °C is sufficient for complete sorbent regeneration, but for

Langmuir, Vol. 19, No. 6, 2003 2199

Figure 14. SAPO-43 regeneration cycle after water vapor adsorption. Regeneration was done under pure helium atmosphere.

Figure 15. SAPO-43 regeneration cycle after hydrogen sulfide adsorption. Regeneration was done under a pure helium atmosphere.

hydrogen sulfide postadsorption this is not the case. Figure 15 shows that the working capacity of SAPO-43 is reduced by about 60% after H2S uptake. This behavior can be due to possible strong interactions of the gas with the sorbent, which is evident from the slope of the single component isotherm at low coverage or because of some reaction of the gas with some of the sorbent’s oxygen atoms. Reshetenko et al.36 studied the reaction and decomposition of H2S on Al2O3 and found that, as the temperature is increased, hydrogen sulfide interacts with the oxide to form oxygen-sulfur compounds. However, this was for temperatures between 400 and 600 °C, which is beyond the thermal stability of SAPO-43. Obviously, a better regeneration scheme must be designed to fully restore the sorbent’s working capacity. Perhaps the use of hydrogen peroxide, which has been used in activated carbon regeneration,37,38 could serve as a promoter to remove the hydrogen sulfide. Conclusion TG/DTG, elemental, and adsorption analyses when combined indicate that the framework can be preserved under partial calcination conditionssat least at high heating rates. Elemental analysis results and carbon (36) Reshetenko, T. V.; Khairulin, S. R.; Ismagilov, Z. R.; Kuznetsov, V. V. Int. J. Hydrogen Energy 2002, 27, 387. (37) Cal, M. P.; Strickler, B. W.; Lizzio, A. A. Carbon 2000, 38, 1757. (38) Cal, M. P.; Strickler, B. W.; Lizzio, A. A.; Gangwal, S. K. Carbon 2000, 38, 1767.

2200

Langmuir, Vol. 19, No. 6, 2003

dioxide heats of adsorption point to the presence of aminelike compounds entrapped in the surface of SAPO-43. These amines should come from unreacted isopropylamine or the products of its oligomerization. It is believed that these compounds keep the framework from collapsing. Adsorption tests have shown that partially calcined SAPO-43 can be used to separate molecules like CO2 and H2S from natural gas streams in a selective way, but these tests showed also that the sorbent is highly hydrophilic. For regeneration, a temperature of ca. 180 °C is sufficient to remove any residual water after its uptake, but any

Herna´ ndez-Maldonado et al.

remaining hydrogen sulfide cannot be desorbed within the thermal stability range of SAPO-43. Acknowledgment. SEM-EDAX studies were conducted at the University of Michigan’s Electron Microscopy Analysis Laboratory (EMAL). We thank Dr. Corinna Wauchope in EMAL for teaching how to use SEM equipment and EDAX package. Support by NSF CTS0138190 is acknowledged. LA026424J