Monash University Synthesis of Zeolite Nanocrystals and Their Application for Mixed Matrix Membranes by Dan Li July 2009 A dissertation submitted for the degree of Doctor of Philosophy in the Department of Chemical Engineering at Monash University Supervisor: Dr Huanting Wang Department of Chemical Engineering
180
Embed
Synthesis of Zeolite Nanocrystals and Their Application for Mixed … › ... › monash24299.pdf · 2017-02-03 · Synthesis of Zeolite Nanocrystals and Their Application for Mixed
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Monash University
Synthesis of Zeolite Nanocrystals and
Their Application for Mixed Matrix
Membranes by
Dan Li
July 2009
A dissertation submitted for the degree of Doctor of Philosophy in the Department of
Chemical Engineering at Monash University
Supervisor: Dr Huanting Wang
Department of Chemical Engineering
Declaration
i
Declaration
I hereby declare that this thesis contains no material which has been accepted for the
award of any other degree or diploma at any university or equivalent institution and that,
or the best of my knowledge and belief, this thesis contains no materials previously
published or written by another person, excepted where due reference is made in the
text of the thesis.
Signed: _________________
Date: ___________________
Acknowledgements
ii
Acknowledgements I would like to take this opportunity to thank my supervisor Dr Huanting Wang for his
willingness to accept me into his research team and his unwavering support throughout
my PhD study. His insight and suggestions have helped me progress so far.
I extend my appreciation to all of the people working in our lab, including Dr Xinyi
Zhang, Dr Jianfeng Yao, Dr Wei Zhu, Dr Jingping Wei, Ms Zhanli Chai, and Dr Dehua
Dong for their nice help during my experiments. I have enjoyed so much in working
with all of them.
Special thanks go to Dr Yunxia Yang and Ms Na Hao for their assistance on gas
adsorption-desorption test and 29Si MAS NMR analysis. Also thanks Yuan Fang, Nicky
Eshtiaghi, and Hue-chen Au Yong for their encouragement and kind care.
Many thanks go to the Australian research council discovery for funding this research
project.
Lastly, I would like to thank my family members, especially my parents and my
boyfriend Weihan Wang, for all of their love, support and encouragement in all that I
set out to achieve.
在我博士毕业在即,我希望借此机会感谢我的父母对我一直以来的教育和关怀。
回首过往的生活和学习中,每当我遇到艰辛和困惑的时候,他们总是给予我无限
的支持,信任和鼓励。没有他们多年来对我付出的心血,也不会有今天的我。亲
爱的爸爸妈妈,在我的心中,我永远爱你们。
Thank you all.
Dan Li
July 2009
Table of Contents
iii
Table of Contents
Declaration.................................................................................................. i
Acknowledgements.................................................................................... ii
Table of Contents ..................................................................................... iii
Abstract..................................................................................................... vi
List of Publications................................................................................... ix
List of Figures........................................................................................... xi
List of Tables ........................................................................................... xv
List of Schemes ....................................................................................... xvi
Abbreviations........................................................................................ xviii
Zeolite crystallization in crosslinked chitosan hydrogels: Crystal size control andchitosan removal
Dan Li a, Yi Huang a, Kyle R. Ratinac b, Simon P. Ringer b, Huanting Wang a,*
a Department of Chemical Engineering, Monash University, Clayton, Vic. 3800, Australiab Australian Key Center for Microscopy and Microanalysis, The University of Sydney, Sydney, NSW 2006, Australia
a r t i c l e i n f o
Article history:Received 15 January 2008Received in revised form 7 April 2008Accepted 30 April 2008Available online 24 May 2008
For the purpose of controlling zeolite crystal size, crystallization of zeolite NaA and NaY in glutaraldehydecrosslinked chitosan (GA-CS) hydrogels was studied in this paper. The zeolite crystals were produced bypenetration of Na2O–Al2O3–H2O alkaline solution into GA-CS hydrogels filled with colloidal silica, fol-lowed by hydrothermal treatment and removal of GA-CS hydrogels. We systematically investigatedthe effects of the synthesis parameters – including the amounts of silica, chitosan, and glutaraldehyde,and the aging and heating times – on the size, size distribution and crystallinity of the particles. A hydro-gen peroxide treatment method was shown to be an effective way for removing GA-CS hydrogels, therebyavoiding the conventional calcination step. X-ray diffraction (XRD), light scattering, scanning electronmicroscopy (SEM), thermogravimetric analysis (TG), and N2 sorption were used to characterize the zeo-lite samples. This work showed that GA-CS is a promising space-confinement medium for the synthesis ofzeolite nanocrystals with tunable crystal sizes and excellent dispersibility.
� 2008 Elsevier Inc. All rights reserved.
1. Introduction
There has been considerable interest in confined-space synthe-sis of zeolite nanocrystals since the principle was first reported byMadsen and Jacobsen in the late 1990s [1]. To date, a variety ofadditives such as carbon blacks and polymer hydrogels have beenused to confine zeolite crystallization. As a class of soft space-con-finement additives, polymer hydrogels comprise three-dimen-sional networks that are created via physical or chemicalcrosslinking [2, 3], which can be readily introduced into zeolitesynthesis due to good compatibility between zeolite precursorsand polymer gels [4]. We have recently demonstrated the con-trolled synthesis of zeolite crystals in chemically crosslinked poly-acrylamide hydrogel [4] and thermoreversible methyl cellulosehydrogels [5]. The crystal sizes of SAPO-34 molecular sieves weresubstantially reduced by forming crosslinked polyacrylamidehydrogel from the water-soluble organic monomers acrylamide(AM) and N,N0-methylenebisacrylamide (MBAM), followed by avapor phase transport process [4]. However, the synthesizedSAPO-34 nanocrystals exhibited a very poor dispersibility in sol-vents. Similarly, NaA (20–180 nm in size) and NaX (10–100 nmin size) nanocrystals were synthesized by employing thermore-versible methylcellulose hydrogels to confine crystal growth [5].The zeolite nanocrystals were easily collected by washing awaythe water-soluble methylcellulose at room temperature, and they
ll rights reserved.
: +61 3 9905 5686.u (H. Wang).
were highly dispersible in water and ethanol. Given the successof these techniques, it would be considerable interest to further ex-plore the feasibility for the synthesis of zeolite nanocrystals withcontrollable sizes and size distributions in other polymerhydrogels.
In this paper, therefore we report attempts to control the syn-thesis of zeolite nanocrystals in the system of Na2O–SiO2–Al2O3–H2O by using crosslinked chitosan hydrogels. Chitosan is a partiallydeacetylated polymer of chitin, which is found in a wide range ofnatural sources, such as crab, lobster and shrimp shells. Its ali-phatic primary amino groups are regularly distributed along thepolymer backbones, and can be crosslinked to form more rigidpolymer networks [6–9]. The crosslinked chitosan hydrogels havebeen studied for various applications such as in pervaporation sep-aration through chitosan [10] or chitosan-zeolite membranes [11],enzyme immobilization [12] and cationic specimen transportation[13], controlled ingredient-release [14, 15], environmental applica-tions [16] and fuel cells [17]. In the present work, zeolite crystalli-zation in glutaraldehyde crosslinked chitosan hydrogels wassystematically investigated to determine the effects of differentcompositions of the synthesis mixture (ratios of chitosan to silicato glutaraldehyde), and the duration of aging and heating. Unlikeother polymer hydrogels, chitosan is only soluble in an acidic solu-tion, and does not dissolve in an alkaline zeolite synthesis gel.Therefore, a two-step method involving the formation of silica-filled crosslinked chitosan hydrogel and the subsequent penetra-tion of Na2O–Al2O3–H2O alkaline solution was developed to forma sodium aluminosilicate gel inside the crosslinked chitosan hydro-
D. Li et al. / Microporous and Mesoporous Materials 116 (2008) 416–423 417
gel. After zeolite synthesis, hydrogen peroxide was employed todegrade chitosan to retrieve zeolite crystals, and a comparison be-tween hydrogen peroxide degradation and high-temperature calci-nation was made.
Scheme 1. Synthesis of zeolite crystals within crosslinked chitosan hydrogels (GA-CS).
2. Experimental section
2.1. Synthesis of zeolite LTA (NaA)
Firstly, 0.6–1.5 g of chitosan (average molecular weight120,000 g/mol, �80% deacetylation, Sigma-Aldrich, denoted CS)was dissolved in 21 g of 1 M acetic acid (Sigma-Aldrich). Theresulting solution was stirred at room temperature for 1 h and thenleft overnight without stirring, after which 2.0–4.0 g of colloidalsilica (HS-30, 30%, Sigma-Aldrich) were added. A given amount(0.6–5.0 g) of glutaraldehyde (50%, Sigma-Aldrich, denoted GA)was added into the CS-silica solution, and left undisturbed at roomtemperature for 2 h, resulting in crosslinked chitosan hydrogel (de-noted GA-CS). Therefore, the silica-filled crosslinked chitosan (GA-CS) hydrogels were synthesized from molar compositions in therange 0.005–0.0125CS:10–20SiO2:1.88–25 glutaraldehyde(GA):21acetic acid (HAc):1243–2499H2O, corresponding to masscompositions of 0.6–1.5CS:0.6–1.2SiO2:0.4–5.0GA:1.3HAc:22.4–24.7H2O.
Secondly, an alkaline solution was prepared by dissolving 5.56 gof NaOH (99%, Merck) in 20.00 g of deionized water, with subse-quent addition of 2.45 g of NaAlO2 (anhydrous, Sigma-Aldrich) dur-ing stirring. The molar composition of the alkaline solution was7.7Na2O:1.0Al2O3:111.0H2O. This alkaline solution was introducedinto the silica-filled crosslinked chitosan hydrogel with a finalmolar composition of 0.005–0.0125CS:10–20SiO2:1.88–25GA:21-HAc:80Na2O:10Al2O3:2396–2455H2O, and aged for 12–72 h atroom temperature. After aging, the gel was removed from the alka-line solution, transferred to a sealed polypropylene bottle and thenheated at 90 �C for 1, 3 or 6 h to allow zeolite crystallization. Tomake a comparison, another sample prepared without aging washeated at 90 �C for 3 h in the presence of the alkaline solution.
2.2. Synthesis of zeolite FAU (NaY)
In this case, the synthesis of silica-filled hydrogels was per-formed from a system with a molar composition of0.01CS:17.5SiO2:12.5GA:21HAc:1302H2O. The same chemicals forthe synthesis of NaA crystals described earlier were used. Typically,1.2 g of CS was dissolved into 21 g of 1 M HAc. As with NaA, thesolution was stirred at room temperature for 1 h, and then leftovernight, after which 3.5 g of colloidal silica was added, and2.5 g of GA was added to form GA-CS. The alkaline solution wasprepared as follows: 4.14 g of NaOH was dissolved in 25.83 g ofdeionized water, with subsequent addition of 0.75 g of NaAlO2 dur-ing stirring. The molar composition of alkaline solution was17.3Na2O:1Al2O3:455.6H2O. The solution was stirred for 0.5–1 huntil it became clear and then it was introduced into the cross-linked chitosan gel system with a molar composition of0.01CS:17.5SiO2:12.5GA:21HAc:55Na2O:3.18Al2O3:2769H2O, andallowed to age at room temperature for 36 h. After aging, the gelwas removed from the alkaline solution, transferred to a sealedpolypropylene bottle, and then heated at 90 �C for 5 h.
2.3. Removal of crosslinked chitosan hydrogels
The heat-treated gels, which contained zeolites, were repeat-edly washed with deionized water until a pH of less than 8 was at-tained. Approximately 3 g of hydrogel was stirred into 150 mL of10% H2O2 solution and then heated at 80–90 �C for 1–2 h. The zeo-
lite crystals were retrieved by high-speed centrifugation and re-peated washing with deionized water; these were dried at 60 �C.For comparison, gels also were calcined to remove the crosslinkedCS. After washing, the zeolite-containing gels were dried at 80 �Covernight, ground by hand using a mortar and pestle, and calcinedat 550 �C under air for 2 h at an initial heating rate of 2 �C min�1.
2.4. Characterization
Scanning electron microscopy (SEM) images were taken with aJSM-6300 F microscope (JEOL). The particle size distributions forzeolite crystals were determined by manual measurement of 300crystals for each sample from the SEM images with Adobe Photo-shop software. Elemental Si/Al ratios of samples were determinedby energy dispersive X-ray spectroscopy (EDXS) on the JSM-6300Fmicroscope. X-ray diffraction (XRD) patterns were recorded on aPhilips PW1140/90 diffractometer with CuKa radiation (25 mAand 40 kV) at a scan rate of 2�/min and a step size of 0.02�. Thermo-gravimetric analysis (TGA, Perkin Elmer, Pyris 1 analyzer) was per-formed at a heating rate of 5 �C/min to 700 �C in oxygen with aflow rate of 15 cm3 min�1. Nitrogen adsorption–desorption exper-iments were performed at 77 K with a Micrometritics ASAP2020MC analyzer. The NaA sample and NaY sample were degassedat 673 K for 24 h, and 623 K for 4 h, respectively, prior to analysis,and the specific surface areas were calculated according to the Bru-nauer–Emmett–Teller (BET) method. To study the dispersibility ofzeolite nanocrystals, the particle size distributions of colloidal zeo-lite suspension were analyzed by light scattering with a MalvernMastersizer 2000 analyzer. Approximately 12–15 mL samples ofcolloidal zeolite suspension were prepared for this purpose by dis-persing 50 mg of each sample into 50 mL of deionized water duringultrasonication.
3. Results and discussion
A schematic diagram for the formation of zeolite nanocrystals inGA-CS hydrogels is shown in Scheme 1. A colloidal silica solution isdispersed in the solution of CS and acetic acid. When the cross-linker (GA) is added, the amino groups from the backbones ofchitosan are crosslinked [9], which causes chitosan solution tosolidify into yellow gels that contain colloidal silica. To form alumi-nosilicate zeolite gels, the alkaline solution is added into the yellow
418 D. Li et al. / Microporous and Mesoporous Materials 116 (2008) 416–423
gel. During the aging, alkaline solution penetrates into the GA-CSgel and reacts with the entrapped silica. Aluminosilicate gels crys-tallize during the subsequent heating, producing zeolite crystalswithin the crosslinked chitosan hydrogel networks. To removethe chitosan hydrogels, hydrogen peroxide is added to solubilizecrosslinked chitosan by degradation of the network structure. Thenthe zeolite crystals are readily collected through high-speed centri-fugation and repeated washing (Scheme 1).
3.1. Effect of the amount of SiO2
Fig. 1 shows the XRD patterns of the samples prepared with mo-lar compositions of 0.01CS:ySiO2:12.5GA:21HAc:(1166 + 8y)H2O(y = 10-20). They are denoted A-10SiO2, A-12.5SiO2, A-15SiO2, A-17.5SiO2, and A-20SiO2, respectively. From Fig. 1, A-10SiO2 appearsto be amorphous, and A-12.5SiO2 exhibits very low crystallinity,whereas A-15SiO2, A-17.5SiO2, and A-20SiO2 are pure zeolite A.Interestingly, A-17.5SiO2 exhibits the smallest average particle size(148 nm), which is much smaller than the mean particle sizes of325 nm and 239 nm of A-20SiO2 and A-15SiO2, respectively. It iswell known that organic additives play an important role in zeolitenucleation and growth [18, 19]. Previous studies have found thatthe addition of water-soluble surfactants (e.g. sodium dodecyl sul-fate, sodium dioctylsulfosuccinate and cetyltrimethylammoniumbromide) and organic polymers (e.g. poly(ethylene glycol)) in thezeolite gel dramatically shortened prenucleation and nucleationperiods and accelerated crystal growth. [18] Some recent mathe-matical [20, 21] modeling and experimental results [22] haveshown that the zeolite nucleation takes place at the interface be-tween the solution and the gel solid by adsorption and rearrange-ment of soluble precursor. Our experimental results could possiblybe explained by the effect of the ratio of SiO2 to CS on the rates ofnucleation and growth. When the ratio of SiO2 to CS is high at2000:1, both initial nucleation rate and subsequent growth rateare presumably high due to the high concentration of aluminosili-cate gel, ultimately leading to the larger crystal sizes. As the ratio ofSiO2 to CS is lowered to 1750:1, the initial nucleation rate mightdrop slightly, accounting for far smaller crystals. When the ratioof SiO2 to CS decreases further to 1250:1, both nucleation rateand growth rate would significantly decrease, resulting in decreasein the number of nuclei formed in the system, but an overall in-crease in final particle size. As expected, if the ratio of SiO2 to CSbecomes too low (e.g. 1000), the amount of silica is insufficientfor zeolite crystallization.
10 20 30 40 50 60
(e)
(d)
(c)
(b)
(a)
Inte
nsity
(a.
u.)
2θ (degrees)
Fig. 1. XRD patterns of the samples prepared with molar compositions of0.01CS:ySiO2:12.5GA:21HAc:(1166 + 8y)H2O (y = 10 � 20) under the same aging(36 h) and heating (90 �C for 3 h) conditions: (a) A-10SiO2; (b) A-12.5SiO2; (c) A-15SiO2; (d) A-17.5SiO2 and (e) A-20SiO2. All of the samples were collected afterH2O2 treatment.
The SEM image, particle size distributions, and nitrogen sorp-tion isotherm of A-17.5SiO2 are shown in Fig. 2. The SEM image(Fig. 2a) reveals that the crystals exhibit irregular shapes. The par-ticle sizes determined by SEM range from 100 nm to 200 nm, aver-aging 148 nm (Fig. 2b), which is slightly smaller than their meanparticle size 174 nm measured by light scattering (Fig. 2c). Thesimilarity of these size distributions and mean sizes suggests thatthe produced particles were well dispersed in water. Furthermore,the suspension formed by dispersing the zeolite particles in wateris stable under lab condition for at least one week. In the nitrogenadsorption–desorption isotherm (Fig. 2d), the amount of nitrogenadsorbed in the sample is very low at low relative pressures, andsubstantially increase at high relative pressures (e.g. P/Po > 0.8).It is clear that the nitrogen adsorption arises from the external sur-faces of nanocrystals because the micropores of well-crystallizedzeolite LTA crystals are inaccessible to nitrogen molecules at the li-quid nitrogen temperature (77 K) [23]. The BET surface area is cal-culated to be 26.2 m2/g, and this further supports a highcrystallinity of the sample.
3.2. Effect of the amount of chitosan
Fig. 3 shows the XRD patterns of the samples produced withmolar compositions of xCS:17.5SiO2:1250xGA:21HAc:(1233 +6944x)H2O (x = 0.005-0.0125), and clearly the crystallinity of sam-ples declines as the amount of chitosan increases. For instance,when x = 0.0125 (Fig. 3a), the sample (A-0.0125CS) exhibits verylow crystallinity. This is probably due to an increase in the densityof the polymer networks with the CS concentration, resulting insignificant reduction of diffusion of zeolite precursors, and henceslow crystallization. The same argument can be applied to the dif-ference in particle size. The SEM results indicate that the crystalsize of zeolite A decreases with increasing amounts of CS; averagesizes are 399 nm, 341 nm and 148 nm for samples prepared withx = 0.0050, 0.0075, and 0.01, respectively. Again this can be ex-plained by the decrease in local concentration of aluminosilicateavailable within the chitosan hydrogels as the amount of chitosanincreases.
3.3. Effect of the amount of glutaraldehyde (GA)
The amount of GA was varied in the CS hydrogels to study theeffect of crosslinking. The samples were prepared from crosslinkedCS hydrogels with a molar composition of 0.01CS:17.5SiO2:zGA:21-HAc:(1233 + 6z)H2O (z = 1.88-25) under the same aging (36 h) andheating (90 �C for 3 h) conditions, and H2O2 treatment. The GA con-centration in the crosslinking gel was expressed as the molar ratio‘‘GA molecules:amino groups from chitosan” (GA:NH2), which wasvaried at 0.3GA:1.0NH2 (z = 1.88), 1.0GA:1.0NH2 (z = 6.25),2.0GA:1.0NH2 (z = 12.5), 4.0GA:1.0NH2. (z = 25). The samples ob-tained were denoted A-0.3GA, A-1.0GA, A-2.0GA and A-4.0GA,respectively. Fig. 4 displays the SEM images for A-0.3GA, A-1.0GA, A-2.0GA and A-4.0GA. It is clear that A-0.3GA has a widesize distribution ranging from 150 nm to over 500 nm (Fig. 4a),whereas A-1.0GA exhibits a narrower size distribution rangingfrom 100 nm to approximately 340 nm (Fig. 4b). A-2.0GA and A-4.0GA exhibit still smaller sizes and narrower size distributions(Fig. 4c and d).
Fig. 5 compares the particle size distributions measured by SEMand light scattering for A-2.0GA and A-4.0GA. In terms of SEM-de-rived particle size distributions, both A-2.0GA and A-4.0GA havesimilar particle size distributions, with average sizes of 148 nmand 147 nm, respectively (Fig. 5a and c). Light scattering measure-ments indicate that both A-2.0GA and A-4.0GA are well dispersedin deionized water, and that their mean particle size is 174 nmfor A-2.0GA (Fig. 5b) and 167 nm for A-4.0GA (Fig. 5d).
Fig. 2. (a) SEM image, (b) particle size distribution determined by SEM, (c) particle size distribution measured by light scattering, and (d) N2 sorption isotherm of the sampleA-17.5 SiO2.
2θ (degrees)10 20 30 40 50 60
(d)
(c)
(b)
Inte
nsity
(a.
u.)
(a)
Fig. 3. XRD patterns of the samples prepared with molar compositions ofxCS:17.5SiO2:1250xGA:21HAc:(1233 + 6944x)H2O (x = 0.005-0.0125) under thesame aging (36 h) and heating (90 �C for 3 h) conditions: (a) A-0.0125CS; (b) A-0.01CS; (c) A-0.0075CS; (d) A-0.005CS. All of the samples were collected after H2O2
treatment.
Fig. 4. SEM images of the particles produced with different amounts of added GA:(a) A-0.3GA; (b) A-1.0GA; (c) A-2.0GA; (d) A-4.0GA.
D. Li et al. / Microporous and Mesoporous Materials 116 (2008) 416–423 419
These results suggest that the degree of crosslinking of chitosanhydrogel has a strong effect on the crystal size. As the amount ofGA increases, the three-dimensional 3 D networks of the hydrogelsbecome denser and more uniform [24]. The higher degree of cross-linking leads to lower degree of swelling, thereby decreasing thediffusion of zeolite precursors in solution [25]. Therefore,crosslinked chitosan gels with more GA provide more rigid, con-fined-spaces for zeolite crystallization and less material for crystal-lization, resulting in smaller and more uniform zeolitenanocrystals. A-2GA and A-4GA exhibit similar morphologies andparticle size distributions because, as has been pointed out in theliterature [25], there is little change in the degree of crosslinkingat high concentrations of GA and, therefore, little change in thecrystallization environment.
3.4. Effect of aging time
The crosslinked chitosan gels with a molar ratio of0.01CS:17.5SiO2:12.5GA:21HAc:1302H2O were aged for differentperiods of 12, 36 or 72 h, and then heated at 90 �C for 3 h. Theresulting samples were denoted A-12h, A-36h and A-72h, respec-tively. Fig. 6 shows the XRD patterns of A-12h, A-36h and A-72h.A-12h possesses a LTA crystal phase with a very low crystallinity
Fig. 5. Particle size distributions determined by SEM (a and c) and by light scattering (b and d) for A-2.0GA (a and b) and A-4.0GA (c and d).
Fig. 7. SEM images of the particles obtained from the chitosan gels after (a) 36 haging (A-36h), and (b) 72 h aging (A-72h).
420 D. Li et al. / Microporous and Mesoporous Materials 116 (2008) 416–423
(Fig. 6a). As the aging time increases, the crystallinity of sample in-creases (Fig. 6b and c). This is because the penetration of the alka-line solution through the crosslinked chitosan hydrogels isessential for producing aluminosilicate gels via reaction with silicaand longer aging times allow greater penetration. In addition, long-er aging allows more uniform nucleation to occur through the gelmatrix, which assists the formation of zeolite crystals with smallsizes and narrow size distributions. The SEM images of A-36hand A-72h are shown in Fig. 7.
Fig. 7a exhibits that the sample A-36h has particle sizes from100 to 200 nm, averaging 148 nm as mentioned above. When theaging period is extended to 72 h (sample A-72h), the resultantNaA particles have larger sizes ranging from approximately 200to 350 nm (Fig. 7b). This is probably due to the swelling behaviorof crosslinked chitosan, which is enhanced during the long expo-sure time to alkaline solution [26]. Moreover, it is possible thatthe crosslinked chitosan may partly degrade after days of exposurethese highly alkaline conditions [25]. As a result, the crosslinkedchitosan can adsorb more precursor Na and Al species and provideslarger spaces for the further growth of zeolite particles.
2θ (degrees)10 20 30 40 50 60
(a)
(b)
Inte
nsity
(a.
u.) (c)
Fig. 6. XRD patterns of the samples obtained from the crosslinked chitosan gelsafter (a) 12 h aging (A-12h), (b) 36 h aging (A-36h), and (c) 72 h aging (A-72h).
For further comparison, sample A-0h was made by directlyheating a gel after the addition of alkaline solution without anyaging period. Fig. 8 and Fig. 9 show the XRD patterns and SEMimages, respectively, for samples A-0h and A-36h. A-0h appearsto have a higher degree of crystallinity (as seen in greater peak
2θ (degrees)10 20 30 40 50 60
(a)
Inte
nsity
(a.
u.)
(b)
Fig. 8. XRD patterns of the crystals in samples (a) A-0h (with alkaline solution) and(b) A-36h (without alkaline solution).
Fig. 9. SEM images of the crystals in samples (a) A-0h and (b) A-36h.
2θ (degrees)10 20 30 40 50 60
(a)
(b)
Inte
nsity
(a.
u.) (c)
Fig. 10. XRD patterns of the samples (a) A-1h, (b) A-3h, and (c) A-6h.
Fig. 11. SEM images of the particles produced after heating for (a) 3 h (A-3h) and(b) 6 h (A-6h).
2θ (degrees)
0 10 20 30 40 50 60
(a)Inte
nsity
(a.
u.) (b)
Fig. 12. XRD patterns of (a) A-H2O2 and (b) A-cal.
D. Li et al. / Microporous and Mesoporous Materials 116 (2008) 416–423 421
heights) than A-36h (Fig. 8b), as well as the presence of someamorphous material (Fig. 8a). The SEM images of A-0h (Fig. 9a) ex-hibit generally coarser particles with a broad size distribution –some particles even exceed 500 nm – and a mean particle size of380 nm. In contrast, A-36h possesses a far more uniform particlesize distribution (Fig. 9b) with a smaller average size of 148 nm.
The reasons for the difference in particle sizes can be explainedby the extent and uniformity of penetration of alkaline solutions.After the crosslinking reaction with GA, the CS hydrogels entrapcolloidal silica particles within their networks. To produce zeolitecrystals, an aluminosilicate gel of the desired composition needsto be formed by diffusion of the alkaline solution throughout theGA-crosslinked CS hydrogel network, and subsequent reactionwith colloidal silica. Without aging (i.e., sample A-0h), the largecompositional gradients of aluminosilicate gel exist during heating,such that the zeolite nucleation and growth can only start from theoutside of the polymer hydrogel surfaces, resulting in non-uniformgrowth. Thus, during the limited heating time of 3 h, the poorlydistributed precursor material deep within the polymer gel cannot fully crystallize into zeolite, and this unconverted is the amor-phous phase found in the XRD pattern (Fig. 8a). However, aging atroom temperature for 36 h (sample A-36h) allows the alkalinesolution to be evenly distributed throughout the crosslinkedhydrogels, leading to an aluminosilicate gels with a uniform com-position. After aging, the polymer hydrogels that incorporate thealuminosilicate gel are removed from the solutions for heating,which helps prevent the polymer hydrogels from over-swelling.On the other hand, during hydrothermal reaction, the wet gelsmay also slightly shrink due to water evaporation; presumably thismakes confined-space growth more effectively [27].
3.5. Effect of heating time
To study the effect of heating time, the molar composition ofthe gels was fixed at 0.01CS:17.5SiO2:12.5GA:21HAc:1302H2Oand all the synthesis gels were aged for 36 h and then heated at90 �C. The heating time was varied from 1, 3 or 6 h, and the corre-sponding as-synthesized samples were denoted A-1h, A-3h, and A-6h, respectively. The XRD patterns (Fig. 10) indicate that there is nocrystalline material formed after 1 h heating, whereas zeolite Acrystals are produced once the heating period is extended to 3 h(A-3h) or 6 h (A-6h). Fig. 11 shows the SEM images of A-3h andA-6h. It can be seen that the zeolite A which are produced under6 h hydrothermal treatment has larger particle sizes than thoseproduced during 3 h of heating. This difference might be attributedto a combination of the flexibility, interconnected pore channels,and large pore sizes (e.g. a few microns) of polymer hydrogel net-works. Therefore, optimized hydrothermal conditions are requiredfor controlled synthesis of zeolite nanocrystals with a narrow sizedistribution.
3.6. Comparison between the treatment of H2O2 and conventionalcalcination
In this study, we applied a novel method to remove the confin-ing polymer network by degradation of crosslinked chitosanhydrogels in a hydrogen peroxide solution. H2O2 easily decom-poses to form the highly reactive hydroxyl radical (HO�), especiallyunder heating. The hydroxyl radical attacks polymer hydrogels,degrading the crosslinked structure and chitosan molecules [28–32]. For comparison, high-temperature calcination, which is a con-ventional method for removing organic agents, was also used. Thesample for this comparison was crosslinked chitosan with zeoliteA, which was produced from gels with a molar ratio of0.01CS:17.5SiO2:12.5GA:21HAc:1302H2O, which were aged for
100 200 300 400 500 600 700
0
20
40
60
80
100
(a)
(b)
Temperature (ºC)
Wei
ght (
%)
(c)
Fig. 13. TG curves of the samples after the treatment of hydrogen peroxide: (a)plain crosslinked chitosan (GA-CS), (b) A-cal, and (c) A-H2O2.
422 D. Li et al. / Microporous and Mesoporous Materials 116 (2008) 416–423
36 h in alkaline solution and heated for 3 h at 90 �C. The sampletreated by hydrogen peroxide is denoted as A-H2O2 and that trea-
Fig. 14. (a) XRD pattern, (b) SEM image, and (c) particle size distribution by SEM and
Fig. 15. (a)TG curve and (b) nitrogen adsorption–desorption isotherm of
ted by calcination as A-cal. Fig. 12 compares the XRD patterns ofpure particles treated with H2O2 (Fig. 12a) and calcination at550 �C for 2 h (Fig. 12b). Clearly, the zeolite A crystals retain great-er crystallinity under the hydrogen peroxide treatment than thoseobtained from high-temperature calcination.
Fig. 13 shows the thermogravimetric (TG) curves of plain cross-linked chitosan, A-cal and A-H2O2. Under flowing pure oxygen,there is a continuous mass loss from 100 �C to 540 �C for the plaincrosslinked chitosan (Fig. 13a). Its total mass loss reaches 100%after 540 �C. A-H2O2 (Fig. 13c) has a total mass loss of approxi-mately 11% occurring, which is mainly attributed to loss of thestructural water from the zeolites as well as physically adsorbedwater. This confirms that the polymer hydrogels were completelyremoved by the hydrogen peroxide treatment method. For A-cal,however, there is another mass loss after 400 �C (Fig. 13b) in addi-tion to the loss of adsorbed and structural water at around 100 �C.This suggests that the crosslinked chitosan was not completelyburned off during calcination. Given the greater crystallinity re-tained and cleaner removal of the hydrogel, the hydrogen peroxidetreatment is clearly the preferred method for removal of GA-CSafter hydrothermal synthesis.
(d) particle size distribution by light scattering of zeolite FAU (NaY) nanocrystals.
zeolite Y samples obtained after treatment with hydrogen peroxide.
D. Li et al. / Microporous and Mesoporous Materials 116 (2008) 416–423 423
3.7. Synthesis of FAU nanocrystals
FAU nanocrystals were synthesized in GA-crosslinked CS hydro-gels by simply varying the compositions of the alkaline solutionand heating times. Fig. 14 displays the XRD pattern, SEM image,and particle size distributions (from SEM and light scattering) ofthe FAU nanocrystals. The Si/Al ratio of the synthesized crystalswas determined to be 2.07:1.00 by the EDXS analysis, suggestingthe nanocrystals are zeolite Y. The nanocrystals have a high crys-tallinity and a narrow size distribution. Their average crystal sizemeasured by SEM is 192 nm; the mean particle size from showsexcellent agreement at 193 nm. This confirms that the zeoliteNaY produced in this way can be well dispersed in deionized water.
Fig. 15 shows TG and nitrogen adsorption–desorption isothermof the zeolite Y nanocrystals. A mass loss of approximately 15% oc-curs, which is due mainly to loss of structural water from the zeo-lites. This confirms that no polymer molecules remain in the voidsof zeolites, a conclusion that is supported by the nitrogenadsorption–desorption isotherm in Fig. 15b. The sample exhibitsa much higher nitrogen adsorption capacity than the NaA samplesbecause NaY has larger, nitrogen accessible pores [33]. The BETspecific surface area of zeolite Y nanocrystals is calculated to be602.2 m2/g.
4. Conclusion
We have shown that glutaraldehyde crosslinked chitosan (GA-CS) hydrogels with three-dimensional network structures wereeffective for controlling the growth of zeolite NaA and NaY. Thezeolite crystal sizes were significantly affected by the formulationof silica-containing GA-CS hydrogels and alkaline solution, and bythe aging and heating conditions. The zeolite NaA nanocrystalswith an average size of 148 nm and NaY with an average size of192 nm were synthesized in this study. A novel method of usinghydrogen peroxide solution was developed to remove GA-CShydrogels after zeolite synthesis. TGA results confirmed that poly-mer hydrogels were completely removed by this hydrogen perox-ide treatment method. The NaA samples obtained via this methodexhibited much higher crystallinity than those obtained via con-ventional calcination. This suggested that the hydrogen peroxidetreatment method be preferred for removal of GA-CS hydrogels.In addition, the zeolite NaA and NaY nanocrystals produced hereare readily dispersed in common solvents, and therefore theymay be useful for applications such as in the fabrication of zeo-lite-polymer composite membranes and hierarchical porous zeo-litic structures.
Acknowledgments
This work was supported by the Australian Research Council(DP0452829) and by Monash University. The technical assistancefrom staff at the Monash Center for electron microscopy is grate-fully acknowledged. H.W. thanks the Australian Research Councilfor the QEII Fellowship.
References
[1] C. Madsen, J.H. Jacobsen, Chem. Commun. (1999) 673.[2] J. Kopeek, Nature 417 (2002) 388.[3] E.H. Schacht, J. Phys.: Conf. Series 3 (2004) 22.[4] J.F. Yao, H. Wang, S.P. Ringer, K.-Y. Chan, L. Zhang, N. Xu, Microporous
Mesoporous Mater. 85 (2005) 267.[5] H.T. Wang, B.A. Holmberg, Y.S. Yan, J. Am. Chem. Soc. 125 (2003) 9928.[6] K. Ogawa, K. Nakata, A. Yamamoto, Y. Nitta, T. Yui, Chem. Commun. 8 (1996)
2349.[7] K. Ogawa, Chem. Mater. 5 (1993) 726.[8] P. He, S.S. Davis, L. Illum, Int. J. Pharm. 187 (1999) 53.[9] T. Wang, M. Turhan, S. Gunasekaran, Polym. Int. 53 (2004) 911.
[10] D. Anjali Devi, B. Smitha, S. Sridhar, T.M. Aminabhavi, J. Membr. Sci. 262 (2005)91.
Soc. 78 (1956) 5963.[24] O.A.C. Monteiro Jr., C. Airoldi, Int. J. Biol. Macromol. 26 (1999) 119.[25] R.M. Silva, G.A. Silva, O.P. Coutinho, J.F. Mano, R.L. Reis, J. Mater. Sci. Mater.
Med. 15 (2004) 1105.[26] V.H. Kulkarni, P.V. Kulkarni, J. Keshavayya, J. Appl. Polym. Sci. 103 (2007) 211.[27] Z. Qian, Z. Zhang, H. Li, H. Liu, Z. Hu, J. Polym. Sci. Part A: Polym. Chem. 46
(2008) 228.[28] Q. Gao, A. Wan, Y. Zhang, J. Appl. Polym. Sci. 104 (2007) 2720.[29] C.Q. Qin, Y.M. Du, L. Xiao, Polym. Degrad. Stab. 76 (2002) 211.[30] S. Tanioka, Y. Matsui, T. Irie, T. Tanigawa, Y. Tanaka, H. Shibata, Y. Sawa, Y.
Kono, Biosci. Biotechnol. Biochem. 60 (1996) 2001.[31] J.M. Fang, R.C. Sun, D. Salisbury, P. Fowler, J. Tomkinson, Polym. Degrad. Stab.
66 (1999) 423.[32] B. Kang, Y. Dai, H. Zhang, D. Chen, Polym. Degrad. Stab. 92 (2007) 359.[33] H. Yang, Q.T. Nguyen, Z. Ping, Y. Long, Y. Hirata, Mater. Res. Innovations 5
Cubes of Zeolite A with an Amorphous Core**Jianfeng Yao, Dan Li, Xinyi Zhang, Chun-Hua Kong, Wenbo Yue, Wuzong Zhou, andHuanting Wang*
The syntheses of zeolites involve very complex nucleation andgrowth processes. During the past decade, significant progresshas been made towards understanding zeolite crystallizationmechanisms. This progress has been made possible byadvanced analytical techniques, such as high-resolution trans-mission electron microscopy (HRTEM), small-angle X-rayscattering, and atomic force microscopy.[1–5] A number ofzeolite growth mechanisms were proposed based on therespective synthesis of the zeolites. For instance, by monitor-ing the crystallization of silicalite-1 from silica sols intetrapropylammonium ion (TPA) at room temperature, anoriented aggregation mechanism was proposed.[4] In thegrowth mechanism of zeolite A evolving from the nucleiinside the amorphous gel, the particles gradually grow intolarger crystals by consuming the surrounding amorphousgels.[2] The gel was formed by using aluminosilicate solutionsand tetramethylammonium hydroxide as the structure-direct-ing agent (SDA).[2] For zeolite A formation, evidences ofnucleation at the solid–liquid interface of the gel cavities werealso found in sodium aluminosilicate gels without organicSDA.[3] In addition, a reversed crystal growth process fromthe surface to the core of nanocrystallite aggregates wasobserved in the crystal growth of zeolite analcime icosite-trahedra.[6] These studies have undoubtedly provided newinsights into zeolite crystallization processes.
As non-structure-directing agents, organic polymers havesignificant effects on zeolite nucleation and growth. Theconfinement of sodium aluminosilicate zeolite gels in ther-moreversible methylcellulose hydrogels resulted in zeolite A
and X nanocrystals under hydrothermal treatment.[7] Cross-linked polyacrylamide hydrogels was used to reduce SAPO-34 crystal sizes in vapor-phase transport synthesis.[8] In allthese cases, the small crystal sizes is due to space confinementof the polymer hydrogel networks. Hollow sodalite spheresand zeolite A crystals were also synthesized hydrothermallyin the presence of crosslinked polyacrylamide hydrogels. Itwas suggested that the scaffolds of polyacrylamide hydrogelswere the preferential sites for zeolite nucleation, andpromoted the direction of nanoparticle aggregation subse-quent to the surface-to-core growth.[9] These results suggestthat the roles of polymer hydrogels in zeolite synthesis arecomplex, and syntheses of the zeolites depend on themicrostructure of the polymer hydrogels and the interactionbetween the polymer chains and the zeolite gels.
Herein we report the formation of cubes of zeolite A witha single crystalline shell and an amorphous core by in-situcrystallization of sodium aluminosilicate gel inside thepolymer networks of uncrosslinked chitosan hydrogel. Thiswork provides further direct evidence for the surface-to-corereversed-growth mechanism. Chitosan is a biopolymerderived from chitin that is found in a wide range of naturalsources, such as crab, lobster, and shrimp shells. Chitosan,containing abundant amino and hydroxy groups, was used asthe orientation-directing matrix for the synthesis of b-oriented TS-1 films.[10] Glutaraldehyde-crosslinked chitosan(GA-CS) hydrogels were recently used to control zeolitecrystallization, and thus zeolite A and Y nanocrystals weresynthesized.[11] It is noted that chitosan is only soluble in anacidic aqueous solution, and the resulting chitosan solutionturns into a polymer hydrogel when an alkaline solutionpenetrates through the gel. Therefore, for the synthesis ofcore-shell cubes of zeolite A, a two-step process, involving thedispersion of silica in a chitosan acidic solution and subse-quent penetration of Na2O/Al2O3/H2O alkaline solution wasemployed to form a sodium aluminosilicate gel inside theuncrosslinked chitosan hydrogel.
XRD pattern (Figure 1a) indicates the as-synthesizedsample has the structure of zeolite A. SEM image (Figure 1b)shows cube-like crystals with a particle size of 0.5–1.5 mm.This morphology with six {100} facets is typical for zeolite A,which has a cubic structure with the unit cell parameter a =
2.461 nm, and space group Fm3c. The characteristic poly-hedron normally indicates a single crystal property ofzeolite A. According to the classic crystal growth theory,crystals normally develop from nuclei and the appearance ofthe facets is due to the differences in their growth rate.[12–15]
TEM confirms the cube-like or rectangular morphologyof the samples. Figure 2a shows a TEM image of a typicalrectangular particle of zeolite A with the corresponding
[*] Dr. J. F. Yao,[+] D. Li, Dr. X. Y. Zhang, Dr. H. T. WangDepartment of Chemical EngineeringMonash University, Clayton, Victoria 3800 (Australia)Fax: (+ 61)3-9905-5686E-mail: [email protected]
W. B. Yue, Dr. W. Z. ZhouSchool of Chemistry, University of St. Andrews,St. Andrews, Fife KY16 9ST (United Kingdom)
Dr. C. H. (Charlie) KongElectron Microscope UnitUniversity of New South Wales, Sydney, NSW 2052 (Australia)
[+] Present address:State Key Laboratory of Materials-Oriented Chemical Engineeringand College of Chemistry and Chemical EngineeringNanjing University of Technology, Nanjing 210009 (P.R. China)
[**] This work was supported by the Australian Research Council (GrantNo.: DP0452829). H.W. thanks the Australian Research Council forthe QEII Fellowship. W.Z. thanks University of St Andrews for anEaStChem studentship to W.Y.
Supporting information for this article is available on the WWWunder http://dx.doi.org/10.1002/anie.200802823.
selected area electron diffraction (SAED) pattern, which is astandard single crystal diffraction pattern viewed down the[100] zone axis. Many particles were examined, and the singlecrystalline nature of zeolite A was observed in each casewithout evidence of any polycrystallinity and twin defects.However, the image contrast implies a core–shell structure, inwhich the core appears to be disordered. Under the electronbeam of the microscope, the disordered core of the zeolite Areduced in volume and separated itself from the shell in amatter of few minutes. The shell remained intact, whichclearly appeared as a rectangle with a thickness of about 7 nm(Figure 2b). The SAED pattern from the particle in Figure 2bis almost identical to the pattern in Figure 2a, which indicatesthat the shell structure was maintained after the irradiationand the separation of the core. No other diffraction spots wereobserved, indicating that the core is amorphous, rather thanpolycrystalline as in the case of zeolite analcime.[6] As thematerial is very sensitive to the beam, HRTEM images of thecrystalline shell were not acquired. The low-magnificationTEM images and the SAED patterns allows us to describe thezeolite A as a monocrystalline cube-like or rectangular boxwith an amorphous core.
The core-shell structure of the as-synthesized zeolite Awas further supported by dark field TEM images of the cross-sections of the cubes prepared by focused ion beam milling(see Supporting Information, Figure SI1) and by dissolutionof the core component in an acidic aqueous solution. Duringthe latter process, 30 mL of 0.35m acetic acid solution wasadded to 1 g of the zeolite A under stirring for 4 h. Most of thecubes lost their inner filling and micrometer-sized hollow
cube-like structures were produced (Supporting Information,Figure SI2a). According to the XRD pattern these hollowstructures were amorphous (Supporting Information, Fig-ure SI2b). As the rate of dissolution of the amorphous core ismuch faster than the crystalline shell, the cube-like orrectangular outer shape was retained, although the crystal-linity of the shell was lost during the acidic treatment. Forcomparison, zeolite A crystals were prepared without chito-san; however, the resulting particles were amorphous (Fig-re SI3).
To investigate the crystallization of zeolite A during thehydrothermal reaction, the products obtained during 1 to 6 hof reaction time were examined. The sample was amorphousafter 1 h, whereas those obtained after 2 and 3 h had zeolite Astructure. After 4 h and 6 h of hydrothermal reactions, amixture of zeolite A and sodalite (Supporting Information,Figure SI4) was obtained. After the 0.35m acetic acid treat-ment, all the crystalline samples obtained from hydrothermalsynthesis of 2–6 h had a rectangular or cube-like morphology(Supporting Information, Figure SI5). These results indicatethat extending hydrothermal reaction time did not lead to thecrystallization of the cores of zeolite A, but resulted in thetransformation of the crystal structure of the shell.
Figure 3 shows the formation of cubes of zeolite A with anamorphous core. Initially, the silica sol is dispersed in anacidified chitosan solution (Figure 3a). After addition of thealkaline solution, the chitosan molecules are deprotonated,resulting in a hydrogel with microsized three-dimensionalpores (Figure 3b). The sodium aluminosilicate gel is producedinside the chitosan hydrogel network by the reaction betweensilica and the alkaline solution. During the hydrothermaltreatment, zeolite nucleation takes place mainly on thesurface of the aluminosilicate aggregates (Figure 3c). Similarto the case of zeolite analcime,[6] some crystalline islandsmight initially form on the surface of the aluminosilicate.These islands then join together, leading to a monocrystallinecube-like shapes by self-alignment of their crystallographicorientations (Figure 3d). Thus, it is an interesting observationthat a very thin-walled crystalline cube-like or rectangular
Figure 2. TEM images of a typical zeolite A particle with a cube-likemorphology and the corresponding SAED patterns obtained from theentire particle. a) Original particle and b) the same particle after beamirradiation for a few minutes.
Figure 3. Representation of the formation of cubes of zeolite A, withan amorphous core (e). The rounded boxes in a)–d) are about4 mm � 4 mm in dimension.
Figure 1. a) XRD pattern and b) SEM image of the as-synthesizedsample.
morphology can be developed on the surface of an amor-phous cluster without any specific relationship to the crystalgrowth rate of the crystal planes from a nucleus in theamorphous center (Figure 3d). The driving force behind theformation of such polyhedral shells is the one that minimizesthe surface energy.
It has also been observed that organic polymers havesignificant effects on zeolite crystallization.[7–9,11, 16] The addi-tion of water-soluble polymers in the zeolite gel coulddramatically shorten the prenucleation and nucleation peri-ods and thus accelerate the crystal growth.[16] Mathematicalmodeling[17, 18] and experimental results[3] indicate that zeolitenucleation takes place at the interface between the solutionand the gel by adsorption and rearrangement of the solubleprecursors. In the synthesis described herein, the uncros-slinked chitosan hydrogel networks are highly swollen by thesolution, and the interfaces between the chitosan polymernetworks and zeolite aluminosilicate gel can serve as idealnucleation sites. Such unique interfaces facilitate zeolitecrystallization from the surface of the aliminosilicate gelaggregates. On the other hand, the chitosan hydrogel may alsoplay a role in confining the aluminosilicate aggregates andthus controlling the sizes of the zeolite A cubes. During thehydrothermal treatment, the crystalline shell limits thediffusion of the solution and thus the crystallization of thecores is not able to proceed. It is worth mentioning that fullycrystallized zeolite A cubes were obtained when the gel wasaged overnight at room temperature before the hydrothermaltreatment. The aging process most likely makes the systemmore uniform, in which the chitosan-facilitated zeolitenucleation becomes kinetically less pronounced. In addition,the desired network structure of chitosan hydrogels isessential for the formation of cubes of zeolite A with anamorphous core. Owing to the presence of crosslinkedchitosan hydrogels, the small hydrogel pores greatly confinedzeolite growth leading to zeolite nanocrystals.[11]
In summary, we have shown that cubes of zeolite Aconsisting of a thin crystalline shell and an amorphous corecan be grown within uncrosslinked chitosan hydrogels. It isindicative that the formation of cube-like or rectangular core-shell structures involves particle aggregation and surface-to-core crystallization induced by chitosan networks. This workmay provide a new model system for studying complex zeolitenucleation and growth mechanisms.
Experimental SectionAcetic acid (99%, Sigma–Aldrich; 7 g of 1m) was dissolved indeionized water (14 g) in a polypropylene bottle. Chitosan (averagemolecular weight 120000 gmol�1, ca. 80% deacetylated, Sigma–Aldrich; 1.2 g) was dissolved in the prepared acetic acid solutionunder magnetic stirring for 1 h, followed by addition of the silica sol(HS-30 30 wt %, Sigma–Aldrich; 3.38 g) to the chitosan/acetic acidsolution. The alkaline solution was prepared by mixing NaOH (99%,Merck; (5 g), and NaAlO2 (anhydrous, Sigma–Aldrich; 2.45 g) withdeionized water (20 g). The solution was stirred for 0.5–1 h until itbecame clear. The Na2O/Al2O3/H2O alkaline solution was added tothe chitosan/acetic acid solution without stirring, resulting in a sodium
aluminosilicate gel entrapped inside the chitosan hydrogel. The finalmolar composition of chitosan/SiO2 was 1.18:1. After hydrothermaltreatment at 90 8C for 3 h, the samples were washed with sufficientwater and dried at 80–1008C overnight, followed by calcining thedried sample at 500 8C in oxygen, or treating them with 10%hydrogen peroxide to remove chitosan.[11] In addition, samples werealso synthesized at 90 8C with different hydrothermal reaction times(1, 2, 4, and 6 h).
Scanning electron microscopy (SEM) images were taken with aJSM-6300F microscope (JEOL). Transmission electron microscopy(TEM) images and selected-area electron diffraction (SAED) weretaken with a JEOL JEM-2011 electron microscope operated at200 kV. X-ray diffraction (XRD) patterns were recorded on a PhilipsPW1140/90 diffractometer with Cu Ka radiation at a scan rate of28min�1 and a step size of 0.028.
Received: June 14, 2008Published online: October 2, 2008
[1] C. S. Cundy, P. A. Cox, Microporous Mesoporous Mater. 2005,82, 1 – 78.
[2] S. Mintova, N. H. Olson, V. Valtchev, T. Bein, Science 1999, 283,958 – 960.
[3] V. P. Valtchev, K. N. Bozhilov, J. Am. Chem. Soc. 2005, 127,16171 – 16177.
[4] a) T. M. Davis, T. O. Drews, H. Ramanan, C. He, J. S. Dong, H.Schnablegger, M. A. Katsoulakis, E. Kokkoli, A. V. McCormick,R. L. Penn, M. Tsapatsis, Nat. Mater. 2006, 5, 400 – 408; b) M. A.Snyder, M. Tsapatsis, Angew. Chem. 2007, 119, 7704 – 7717;Angew. Chem. Int. Ed. 2007, 46, 7560 – 7573.
[5] J. R. Agger, N. Pervaiz, A. K. Cheetham, M. W. Anderson, J.Am. Chem. Soc. 1998, 120, 10754 – 10759.
[6] X. Y. Chen, M. H. Qiao, S. H. Xie, K. N. Fan, W. Z. Zhou, H. Y.He, J. Am. Chem. Soc. 2007, 129, 13305 – 13312.
[7] H. T. Wang, B. A. Holmberg, Y. S. Yan, J. Am. Chem. Soc. 2003,125, 9928 – 9929.
[8] J. F. Yao, H. T. Wang, S. P. Ringer, K. Y. Chan, L. X. Zhang, N. P.Xu, Microporous Mesoporous Mater. 2005, 85, 267 – 272.
[9] L. Han, J. F. Yao, D. Li, J. Ho, X. Y. Zhang, C. H. Kong, Z. M.Zong, X. Y. Wei, and H. T. Wang, J. Mater. Chem. 2008, 18,3337 – 3341.
[10] X. D. Wang, B. Q. Zhang, X. F. Liu, Y. S. Lin, Adv. Mater. 2006,18, 3261 – 3265.
[11] D. Li, Y Huang, K. R. Ratinac, S. P. Ringer, H. T. Wang,Microporous Mesoporous Mater. 2008, DOI:10.1016/j.micro-meso.2008.04.032.
[12] A. Bravais, e�tudes Cristallographic, Gauthier-Villars, Paris,1866.
[13] M. G. Friedel, Bull. Soc. Fr. Mineral. Cristallogr. 1907, 30, 326 –455.
[14] J. D. H Donnay, D. Harker, Am. Mineral. 1937, 22, 446 – 467.[15] P. Hartman, W. G. Perdok, Acta Crystallogr. 1955, 8, 49 – 52; P.
Hartman, W. G. Perdok, Acta Crystallogr. 1955, 8, 521 – 524; P.Hartman, W. G. Perdok, Acta Crystallogr. 1955, 8, 525 – 529.
[16] G. J. Myatt, P. M. Budd, C. Price, F. Hollway, S. W. Carr, Zeolites1994, 14, 190 – 197.
[17] V. Nikolakis, D. G. Vlacho, M. Tsapatsis, Microporous Meso-porous Mater. 1998, 21, 337 – 346.
[18] T. O. Drews, M. A. Katsoulakis, M. Tsapatsis, J. Phys. Chem. B2005, 109, 23879 – 23887.
Microporous and Mesoporous Materials 106 (2007) 262–267
Organic-functionalized sodalite nanocrystals and their dispersionin solvents
Dan Li a, Jianfeng Yao a, Huanting Wang a,*, Na Hao a, Dongyuan Zhao a,Kyle R. Ratinac b, Simon P. Ringer b
a Department of Chemical Engineering, Monash University, Clayton, VIC 3800, Australiab Australian Key Center for Microscopy and Microanalysis, The University of Sydney, Sydney, NSW 2006, Australia
Received 30 January 2007; received in revised form 2 March 2007; accepted 5 March 2007Available online 13 March 2007
Abstract
Hydroxy-sodalite nanocrystals with organic functional groups (i.e., @Si–(CH3)(CH2)3NH2, denoted Sod-N, or „Si–CH3, denotedSod-C) were synthesized by the direct transformation of organic-functionalized silicalite nanocrystals. The chemical structure oforganic-functionalized sodalite nanocrystals was confirmed by 29Si MAS NMR spectroscopy. Gas sorption results showed that the soda-lite nanocrystals contained uniform pore channels that were accessible to hydrogen, but inaccessible to nitrogen, as expected. The BETsurface areas are calculated to be 22.8, 19.6 and 19.1 m2/g for plain sodalite nanocrystals (Sod), Sod-N, and Sod-C, respectively; simi-larly, Sod-N and Sod-C exhibited slightly lower hydrogen adsorption than Sod. The dispersion of Sod-N and Sod-C in organic solventswas favored by the presence of organic functional groups. Therefore, the organic-functionalized sodalite nanocrystals prepared in thiswork may be very useful for fabricating zeolite nanostructures and sodalite-polymer nanocomposite membranes.� 2007 Elsevier Inc. All rights reserved.
Sodalite is a small-pore zeolite whose framework con-sists of a six-membered ring aperture with a pore size of2.8 A. Because of its small pore size and high ion exchangecapacity, sodalite has been considered as a good candidatematerial for a wide range of applications such as opticalmaterials, waste management, hydrogen storage, andhydrogen separation [1]. The active research into efficientstorage and separation of hydrogen has been driven byits potential as an essential component of future energyeconomies. Consequently, we are interested in developinghigh-selectivity, high-flux membranes for the separationand purification of hydrogen gas. Among the various
1387-1811/$ - see front matter � 2007 Elsevier Inc. All rights reserved.
possible membranes, polymeric ones have been extensivelystudied for hydrogen separation because they are of low-cost and can be easily fabricated into compact hollow fibersand flat sheets with a high separation-area-to-volume ratio[2–4]. Although some polymer membranes exhibit goodhydrogen selectivity and permeability, there is still plentyof room for development of membranes with improvedperformance [2]. Previous studies by a number of groupshave suggested that the incorporation of zeolites into thepolymer matrix can significantly increase gas separationselectivity by enhancing selective gas adsorption and diffu-sion through the membranes [3–6]. Therefore, the additionof sodalite into polymers promises to yield sodalite-poly-mer composite membranes with superior selectivity forhydrogen separation. It has been suggested that template-free sodalite nanocrystals with good interfacial compatibil-ity with the chosen polymer are needed to effectivelyfabricate sodalite-polymer composite membranes [3]. As
D. Li et al. / Microporous and Mesoporous Materials 106 (2007) 262–267 263
part of our project aiming to design such membranes, thefocus of this paper is the synthesis of template-free sodalitenanocrystals with suitably tailored surface properties.
Our newly developed method for synthesizing colloidalhydroxy-sodalite nanocrystals by the transformation of sil-icalite nanocrystals is used in this study [1], since there is noother method for the synthesis of colloidal structure-direct-ing agent free hydroxy-sodalite nanocrystals available. Thehydroxy-sodalite nanocrystals obtained have a sodalitestructure whose framework charges are balanced byhydroxide ions, and they do not enclose template moleculeswithin their pore channels. The strategy of attachingorganic functional groups to zeolites is adopted to achievesuitable surface properties of the hydroxy-sodalite nano-crystals because it is one of the most effective ways formodifying surface properties or adding surface reactivityto zeolite crystals [7–12]. Two kinds of organic groupsincluding methyl and amino moieties are introduced intothe sodalite nanocrystals by adding controlled amountsof 3-aminopropyl(diethoxy) methylsilane and methyltri-methoxysilane during the growth of silicalite nanocrystals.The sodalite nanocrystals are thus expected to be mademore hydrophobic (–CH3) or reactive (–NH2). The prepa-ration and characterization of organic-functionalized soda-lite nanocrystals and their dispersion in solvents aredetailed in this paper.
2. Experimental section
2.1. Synthesis of organic-functionalized silicalite
nanocrystals
Clear synthesis solutions were prepared by dropwiseaddition of 20 g of 1 M tetrapropylammonium hydroxide(TPAOH, Sigma–Aldrich) solution into the mixture of17.8 g of tetraethyl orthosilicate (TEOS, 99%, Sigma–Aldrich) and 1.8 g of 3-aminopropyl(diethoxy) methylsilane(ADMS, 97%, Sigma–Aldrich) or 1.3 g of methyltrimethox-ysilane (MTMS, 98%, Sigma–Aldrich) under vigorous stir-ring, followed by continued stirring at room temperaturefor 3 h. The molar composition of final solution was 1TPAOH:4.32 SiO2:0.48 ADMS (or MTMS): 44 H2O. Crys-tallization was carried out at 80 �C for 12–15 days. Themilky silicalite suspensions obtained were dried at 90–100 �C leading to solid silicalites (denoted Sil-N and Sil-Cfor silicalites prepared with ADMS and MTMS, respec-tively). To observe their morphologies by scanning electronmicroscopy, the samples were prepared by repeated cyclesof washing with deionized water and centrifuging, followedby drying at 90–100 �C overnight.
2.2. Synthesis of organic-functionalized sodalite nanocrystals
An alkaline solution with a molar composition of 6.07Na2O:1 Al2O3:66 H2O was prepared by mixing 20 g ofsodium hydroxide (99%, Merck), 9.2 g of sodium alumi-nates (anhydrous, Sigma–Aldrich), and 60 g of deionized
water at room temperature for 1–2 h. 1 g of the dried silica-lite sample (i.e. Sil-N and Sil-C) was added to 11 g of thealkaline solution during 2–3 min of stirring, and then agedat room temperature for 4 h without further stirring. Thetransformation was carried out at 80 �C for 0–4 h. Thesamples obtained were cooled to room temperature andcollected by repeated cycles of washing with deionizedwater and centrifuging, followed by drying at 90–100 �Covernight. The samples were denoted Sod-N and Sod-C,respectively, when Sil-N and Sil-C were used as silicasource, respectively. For comparison, hydroxy-sodalitenanocrystals (denoted Sod) were also prepared from silica-lite nanocrystals according to our previous paper [1].
2.3. Characterization
Scanning electron microscopy (SEM) images were takenwith a JSM-6300F microscope (JEOL). The particle sizedistributions for Sil-N, Sil-C, Sod-N and Sod-C were deter-mined by manual measurement of 300 nanocrystals each inSEM images with a Photoshop software. X-ray diffraction(XRD) patterns were measured on a Philips PW1140/90 dif-fractometer with Cu Ka radiation (25 mA and 40 kV) at ascan rate of 1�/min with a step size of 0.02�. Thermogravi-metric analysis (TGA, Perkin Elmer, Pyris 1 analyzer) wasperformed in air at a heating rate of 5 �C/min to 600 �C.29Si solid-state nuclear magnetic resonance (NMR) wasconducted on a Bruker DSX300 spectrometer (Germany)under conditions of cross polarization (CP) and magic anglespinning (MAS). 29Si solid-state MAS NMR spectra werecollected at room temperature with a frequency of59.6 MHz, a recycling delay of 600 s, a radiation frequencyintensity of 62.5 kHz, and a reference sampleof Q8M8([(CH3)3SiO]8Si8O12]). Nitrogen and hydrogenadsorption–desorption experiments were performed at77 K with a Micrometritics ASAP 2020MC analyzer anda Micrometritics ASAP 2010MC analyzer, respectively.The samples were degassed at 473 K before analysis. Thesurface areas were determined by the Brunauer–Emmett–Teller (BET) method. Suspended particle size distributionswere quantified by light scattering with a Malvern Master-sizer 2000 analyzer. Different solvents-deionized water, iso-propanol (97%, Sigma–Aldrich), dichloromethane (DCM,Sigma–Aldrich) and dimethylformamide (DMF, Sigma–Aldrich) – were used for sample dispersion. Approximately12–15 ml of suspension was prepared by dispersing 50 mgof sample into 50 ml of solvent under ultrasonication beforeinjection into the Mastersizer for size distribution analysis.
3. Results and discussion
3.1. Transformation of silicalite
The XRD patterns (Fig. 1) show the transformationof organic-functionalized silicalites (Sil-N and Sil-C)under hydrothermal treatment at 80 �C. The organic-functionalized silicalites (Sil-N and Sil-C) became amorphous
10 20 30 40 50 60
3h
2h
1h
0h
Inte
nsity
(a.u
.)
2θ (degrees)10 20 30 40 50 60
3h2h1h
0h
4h
Inte
nsity
(a.u
.)
2θ (degrees)
Fig. 1. XRD patterns of samples prepared with dried organic-functionalized silicalites by hydrothermal treatment at 80 �C for different times. (a)Sil-N toSod-N and (b) Sil-C to Sod-C.
264 D. Li et al. / Microporous and Mesoporous Materials 106 (2007) 262–267
after 1 h hydrothermal treatment. However, in our previ-ous study [1], plain silicalite (without organic groups) waslargely transformed into zeolite A after only 1 h hydrother-mal treatment. This is because the presence of @Si–(CH3)(CH2)3NH2 and „Si–CH3 in silicalite structures(Sil-N and Sil-C, respectively) does not favor aluminosili-cate structure rearrangement during the incorporation ofAl and Na. After 2 h treatment, both samples were a mix-ture of zeolite A and sodalite. The pure organic-functional-ized sodalite, Sod-N, was obtained after 3 h. However, thetransformation of Sil-C into Sod-C took a longer time (4 h)to complete.
Fig. 2 shows the SEM images and particle size distribu-tions of organic-functionalized silicalite nanocrystals (Sil-Nand Sil-C) and organic-functionalized sodalite nanocrystals(Sod-N and Sod-C). All samples exhibit similar morpholo-gies. Sil-N exhibits smaller particle sizes as compared withSil-C, though the synthesis conditions were identical. Thismay be explained by the presence of the –NH2 groups
Fig. 2. SEM images (c and f) and particle size distributions a, b, d, and e of oimages: (a) dried silicalite Sil-N, (b) sodalite Sod-N, (d) dried silicalite Sil-C, andSil-C and Sod-C.
accelerating nucleation in the silicalite synthesis solution,leading to smaller particles on average [13]. This is alsoconsistent with the XRD results above showing that thetransformation of Sil-N into Sod-N took a shorter time.The particle sizes of the organic-functionalized sodalitenanocrystals are larger than those of their precursor silica-lite nanocrystals. This is related to the recrystallization inthe transformation as indicated by XRD. The mean parti-cle sizes are 95 nm, 105 nm, 105 nm and 140 nm for Sil-N,Sod-N, Sil-C and Sod-C, respectively (Fig. 2c and f).
3.2. Evidence of organic functionalization of sodalite
To prove that the organic functional groups have beenincorporated into the sodalite nanoparticles, Sod-N andSod-C samples were characterized by solid-state NMRspectroscopy. The 29Si MAS NMR spectra shown inFig. 3a display a strong resonance peak at around�85 ppm, which arises from Si (4Al) in Sod-N and Sod-C
rganic-functionalized silicalites and organic-functionalized sodalites. SEM(e) sodalite Sod-C. Particle size distributions: (c) Sil-N and Sod-N, and (f)
100 50 0 -50 -100 -150 -200 -250
Si-C (*)
Sod-NSod-C
ppm
Si (4Al)
SiO
OAl
OSi
O AlO
SiO
Si CH3
SiHO O
AlO
Si
(CH2)3NH2
SiO
OAl
OSi
O AlO
SiO
Si CH3
SiHO O
AlO
Si
Sod-N Sod-C
Fig. 3. (a) 29Si-NMR of organic-functionalized sodalite nanocrystals and (b) the bonding scheme for organic-functionalized sodalite nanocrystals.
0 100 200 300 400 500 600
85
90
95
100
cb
Temperature (°C)
Mas
s (%
)
a
Fig. 4. TGA curves of organic-functionalized sodalite nanocrystals andhydroxy-sodalite nanocrystals. (a) Sod-N, (b) Sod-C and (c) Sod.
D. Li et al. / Microporous and Mesoporous Materials 106 (2007) 262–267 265
[14,15]. The NMR spectra also exhibit a resonance peak ataround �55 ppm, which is ascribed to Si–C bonds [14]. Theresults confirm the existence of organic functional groupsin the Sod-N and Sod-C, and thus the organic groups havebeen incorporated into the sodalites. The integrated area ofthe functionalized silicon peak represents 7.4 mole% and7.2 mole% of the total silicon in Sod-N and Sod-C, respec-tively. The amounts of organic functional groups incorpo-rated into Sod-N and Sod-C are less than those added insilicalite synthesis solutions (10 mole% was added for bothSil-N and Sil-C), but this is reasonable given that a propor-tion of the hydrolyzed ADMS and MTMS would haveremained in the synthetic solutions. The Si–C bondslabeled with asterisk in organic-functionalized sodalitesare illustrated in Fig. 3b.
The organic functionalization of the sodalite nanocrys-tals receives further support from the TGA results, whichare shown in Fig. 4. The mass loss of the pure hydroxy-sodalite was about 11 wt% owing to the loss of the struc-tural water (Fig. 4c) [1]. The mass losses for Sod-N andSod-C were 13.6 wt% and 11.5 wt%, respectively. As com-pared with the pure sodalite nanocrystals, the additionalmass loss of 2.6 wt% for Sod-N and of 0.5 wt% for Sod-C was due to decomposition of organic functional groups(i.e., –(CH3)(CH2)3NH2 or –CH3) at high temperatures[7]. These figures are quite consistent with the expectedmass losses of 3.18 wt% for Sod-N and 0.65 wt% for Sod-C that can be calculated from the proportion of Si–Cbonds measured by 29Si MAS NMR.
3.3. Gas adsorption and pore structures
To further compare the organic-functionalized sodalitenanocrystals (Sod-N and Sod-C) and plain hydroxy-soda-
lite nanocrystals (Sod), nitrogen and hydrogen adsorp-tion–desorption analyses were conducted. The isothermsof Sod-N, Sod-C and Sod are shown in Fig. 5. Theamounts of nitrogen adsorbed in all three samples are verylow at low relative pressures, and substantially increase athigh relative pressures (e.g., P/P0 > 0.8). This is becausewell-grown sodalite pores are inaccessible to nitrogen (N2
kinetic diameter 3.6 A is larger than sodalite pore size2.8 A), and the main nitrogen adsorption arises from theexternal surfaces of nanocrystals. The BET surface areasare calculated to be 22.8, 19.6 and 19.1 m2/g for Sod,Sod-N, and Sod-C, respectively, which is consistent withthe particle size distributions observed by SEM. By con-trast, all samples exhibit much higher H2 adsorption atlow relative pressures as compared with N2 adsorption(Fig. 5a and b), implying that the sodalite channels in thesethree samples are readily accessed by H2 molecules.
0.0 0.2 0.4 0.6 0.8 1.00
10
20
30
40
50
60
Vol
ume
Ads
orbe
d (c
m3 g
-1)
P / Po
Sod-N Sod-C Sod
0.0 0.2 0.4 0.6 0.8 1.0
0
5
10
15
20
25
30
35
Vol
umed
ads
orbe
d (c
m3 g
-1)
P / Po
Sod-N Sod-C Sod
Fig. 5. (a) Nitrogen and (b) hydrogen adsorption–desorption isothermals of plain sodalites (Sod) and organic-functionalized sodalites (Sod-N and Sod-C).
266 D. Li et al. / Microporous and Mesoporous Materials 106 (2007) 262–267
Furthermore, the organic-functionalized sodalites (Sod-Nand Sod-C) possess slightly lower H2 adsorption than puresodalite (Sod). At P/P0 = 0.99, the volume of hydrogenabsorbed is around 33.0 cm3/g for Sod, 26.5 cm3/g forSod-N, and 28.0 cm3/g Sod-C. Therefore, the organicgroups do not substantially change the hydrogen adsorp-tion of the sodalite nanocrystals. Clearly, this finding isessential if the functionalized nanoparticles are to be usedsuccessfully in H2 separation membranes.
3.4. Surface modification: dispersion in solvents
To study the effect of organic functionalization on thedispersibility of sodalite nanocrystals, a series of solventsof different polarities was selected: deionized water, isopro-panol, dichlormethane (DCM), and dimethlformamide
100 1000
0
5
10
15
20
25
Num
ber
(%)
Particle size (nm)
Sod-N Sod-C Sod
100 1000
0
5
10
15
20
25
Num
ber
(%)
Particle size (nm)
Sod-N Sod-C Sod
a b
c d
Fig. 6. Particle size distributions of organic-functionalized sodalite nanocrysta(b) dimethylformamide (DMF), (c) isopropanol and (d) dichloromethane (DC
(DMF). The solvent polarity of this series, in descendingorder, is water (100) > DMF (42.88) > isopropanol(36.72) > DCM (23.04) [16]. The particle size distributionsof Sod-N, Sod-C, and Sod shown in Fig. 6 are used asan indicator of their relative dispersibility. When deionizedwater is used as a dispersion medium, both Sod-N and Sodhave a similar particle size distribution and their mean par-ticle sizes are approximately 160 nm, which is slightlygreater than that observed by SEM due to the surface sol-vation effect (e.g., surface ionization and adsorption)[17,18]. In contrast, Sod-C exhibits a wider particle size dis-tribution and its mean particle size is approximately270 nm (Fig. 6a). The different dispersibility betweenSod-N/Sod and Sod-C arises from their different surfaceenergy components: Sod-N with –(CH3)(CH2)3NH2 groupsand Sod with –OH groups have similar hydrogen-bonding
100 1000
0
5
10
15
20
25
30
Num
ber
(%)
Particle size (nm)
Sod-N Sod-C Sod
100 1000
0
5
10
15
20
25
30
Num
ber
(%)
Particle size (nm)
Sod-N Sod-C Sod
ls and plain sodalite nanocrystals in different solvents: (a) deionized water,M).
D. Li et al. / Microporous and Mesoporous Materials 106 (2007) 262–267 267
forces, whereas Sod-C with –CH3 groups is more hydro-phobic. Sod-N, Sod-C, and Sod show similar dispersibilityin DMF (Fig. 6b) because DMF combines a high polarityand high hydrogen-bonding force with hydrophobicgroups. In isopropanol, both Sod-N and Sod-C exhibitslightly better dispersion than Sod (Fig. 6c). Sod-N andSod-C exhibit similar degrees of dispersion in DCM, butthe Sod nanocrystals severely aggregate, leading to a meanparticle size of 880 nm (Fig. 6d). These are because isopro-panol and DCM, with relatively low polarity and poorhydrogen-bonding force, preferentially interact withorganic-functionalized surfaces [19]. These results clearlyshow that the surface properties of sodalite nanocrystalscan be tailored by organic functionalization, which isessential for preparing zeolite-polymer nanocomposites[10,20].
4. Conclusion
We have successfully incorporated organic functionalgroups into hydroxy-sodalite nanocrystals through thedirect transformation of organic-functionalized silicalitenanocrystals. The organic-functionalized sodalite nanocrys-tals showed high crystallinity and well-grown pore struc-tures based on XRD and nitrogen sorption measurements.The micropores of the organic-functionalized sodalite nano-crystals were highly accessible to hydrogen molecules,though there was a slight reduction of hydrogen adsorptioncompared with sodalite nanocrystals without organicgroups. Sodalite nanocrystals with –(CH3)(CH2)3NH2 moi-eties showed good dispersibility in all four solvents (i.e.,water, isopropanol, dichloromethane, and dimethylform-amide) tested whereas sodalite nanocrystals with –CH3
groups were dispersible in isopropanol, dichloromethaneand dimethylformamide, but were agglomerated in water.Without organic functionalization, sodalite nanocrystalsshowed very poor dispersibility in dichloromethane. There-fore, we expect that the organic-functionalized sodalitenanocrystals synthesized in this work will be highly suitedfor fabricating sodalite-polymer nanocomposite mem-branes and other zeolite nanostructures.
Acknowledgments
This work was supported by the Australian ResearchCouncil (Discovery Project No. DP0559724) and MonashUniversity. The facilities and technical assistance from staffat the Electron Microscopy and Microanalysis Facility,Monash University, are gratefully appreciated. H.W.thanks the Australian Research Council for the QEIIFellowship.
Synthesis and characterization of sodalite–polyimide nanocomposite membranes
Dan Li a, Huai Yong Zhu b, Kyle R. Ratinac c, Simon P. Ringer c, Huanting Wang a,*
a Department of Chemical Engineering, Monash University, Clayton, VIC 3800, Australiab School of Physical and Chemical Sciences, Queensland University of Technology, Brisbane, QLD 4001, Australiac Australian Key Center for Microscopy and Microanalysis, The University of Sydney, Sydney, NSW 2006, Australia
a r t i c l e i n f o a b s t r a c t
Article history:Received 28 March 2009Received in revised form 10 May 2009Accepted 12 May 2009Available online xxxx
Please cite this article in press as: D. Li et al., M
Nanocomposite membranes are fabricated from sodalite nanocrystals (Sod-N) dispersed in BTDA-MDApolyimide matrices and then characterized structurally and for gas separation. No voids are found uponinvestigation of the interfacial contact between the inorganic and organic phases, even at a Sod-N loadingof up to 35 wt.%. This is due to the functionalization of the zeolite nanocrystals with amino groups(@SiA(CH3)(CH2)3NH2), which covalently link the particles to the polyimide chains in the matrices. Theaddition of Sod-N increases the hydrogen-gas permeability of the membranes, while nitrogen permeabil-ity decreases. Overall, these nanocomposite membranes display substantial selectivity improvements.The sodalite–polyimide membrane containing 35 wt.% Sod-N has a hydrogen permeability of 8.0 Barrersand a H2/N2 ideal selectivity of 281 at 25 �C whereas the plain polyimide membrane exhibits a hydrogenpermeability of 7.0 Barrers and a H2/N2 ideal selectivity of 198 at the same testing temperature.
� 2009 Elsevier Inc. All rights reserved.
1. Introduction
During the past two decades, mixed matrix membranes(MMMs) have attracted much attention due to their potential forsuperior gas separation performance [1–4]. A variety of inorganicfillers such as zeolite, porous carbon, and nonporous silica havebeen used to fabricate inorganic–polymer composite membranes.Of the many possible separations, hydrogen purification is ofindustrial importance because of its applications in the chemicalindustry, and its use as a fuel in fuel cells. Until now, the many at-tempts to develop zeolite–polymer composite membranes withimproved hydrogen permeability and selectivity have met withlimited success. For instance, S�en et al. developed polycarbonate-matrix membranes filled with highly crystalline zeolite-4A withparticles sizes of 3 lm [5]. At a zeolite loading of 30 wt.%, the com-posite membranes had an improved H2/N2 selectivity of 73.2 com-pared with 56.7 for plain polycarbonate. However, they also founda decrease in hydrogen permeability, which they attributed to in-creased rigidity of the polymer chains in the presence of the zeoliteparticles [6,7], the partial blockage of the zeolite pore by the poly-mer chains [7] and/or the extended diffusion pathways of thehydrogen molecules through the membrane [8,9]. A similar trendwas reported by Li et al. [10] and Huang et al. [7]. Li et al.demonstrated that membranes of polyethersulfone and zeolite5A (1–5 lm) exhibited about 25% higher H2/N2 selectivity than aplain polyethersulfone membrane, but had a decrease in gas
ll rights reserved.
u (H. Wang).
icropor. Mesopor. Mater. (2009
permeability of at least 25% [7,9]. Huang et al. prepared their com-posite membranes by incorporating 20 wt.% of micrometer-sized(1–5 lm) or nanometer-sized (50–140 nm) zeolite A in polyether-sulfone (PES) [7]; the hydrogen permeability of the PES membranedropped from 8.96 Barrers to 8.3 Barrers when filled with thenano-zeolite and down to 4.94 Barrers for the micro-zeolite. Inter-estingly, the gas permselectivity enhancement was much morepronounced when zeolite-4A nanocrystals were incorporated in aPES membrane. Indeed, nano-sized zeolites are required for fabri-cating composite membranes because the polymeric membranesare usually shaped into asymmetric hollow fibers or flat sheetswith a thin selective layer (e.g., <1 lm) for practical applications[10]. Up to 40.2 wt.% silicalite-1 (MFI) nanocrystals (80 nm) werecombined with Telfon AF 1600 polymers by Golemme et al. [11].The composite membranes had a hydrogen permeability of 3580Barrers, a 15-fold increase relative to the pure polymer mem-branes. However, the H2/N2 selectivity of the composite mem-branes, at just 4.6, was 50% less than the pure Telfon film. Thisresult was probably due to interfacial voids between the zeoliteand polymer, which were formed because of low adhesion be-tween the polymer matrix and the zeolite crystals [5,6,12–14].
Several approaches have been proposed to fabricate the mixed-matrix membranes that are free of voids and have enhanced selec-tivity [15–17]. One of the most effective ways is the surface mod-ification of the zeolite particles with silane-coupling agents[3,16,18]. For instance, Duval et al. promoted the adhesion betweenzeolite particles and polymer matrices by modifying the zeolitesurfaces with silane-coupling agents (e.g., y-aminopropyltriethoxysilane, N-p-(aminoethy1)-y-aminopropyltrimethoxy silane and
2 D. Li et al. / Microporous and Mesoporous Materials xxx (2009) xxx–xxx
ARTICLE IN PRESS
styryl amino functional silane). Unfortunately, the measured per-meabilities were slightly lower compared with polymer matrixwhile the ideal selectivities were largely unchanged [16]. Pecharet al . developed mixed-matrix membranes from polyimide andzeolite L or ZSM-2 zeolites, which were functionalized by APTES(aminopropyl-triethoxysilane) coupling agents. The gas selectivityof the composite membranes was enhanced, but the gas perme-ability was unexpectedly lowered relative to the pure polyimidemembranes [18,19]. Similarly, Li et al. found that the increase inselectivity of membranes made with zeolite A, which had beenmodified with APDEMS (3-aminopropyl)-diethoxymethyl silane),was offset by a decrease in permeability [3,9].
Nonporous fillers such as silica nanoparticles were also incorpo-rated into polymer to yield inorganic–organic polymer compositemembranes. Merkel et al. found that silica-poly(4-methyl-2-pen-tyne) nanocomposite membranes exhibited significantly enhancedmembrane permeability and selectivity for large organic moleculesover small permanent gases. This was because physical dispersionof nanoporous nanoparticles yielded polymer-particle interfaces,disrupted polymer chain packing and thus affected moleculartransport [20].
The objective of our work is to study the feasibility of fabricatingnanocomposite membranes with improved separation propertiesby incorporating organic-functionalized sodalite nanocrystals intopolymer. Sodalite is a type of small-pore zeolite, which has asix-membered-ring aperture with a 2.8 Å pore size. Sodalite wasreported to exhibit good hydrogen adsorption property at high tem-peratures (e.g., >573 K) [21]. Our recent study shows that sodalitenanocrystals exhibit attractive hydrogen adsorption–desorptionbehavior at very low temperatures (e.g., 77 K) [22]. This interestingtemperature-dependent hydrogen sorption property is related tothe change in the size of sodalite cage at different temperatures.It would be of fundamental interest to investigate how sodalitenanocrystals affect the microstructure and separation propertiesof polymer membranes. In this study, we used modified sodalitenanocrystals, functionalized with = SiA(CH3)(CH2)3NH2 groups aswe previously reported [22,23], as the inorganic phase in compositemembranes. We chose polyimide as the continuous polymer matrixfor this study; polyimides have attracted considerable interest forhydrogen separation, because of their good gas transport proper-ties, their thermal and chemical stability, and their mechanicalproperties [24–28]. Previous research has reported excellent selec-tivity, varying from 64.8 to 365, for separating hydrogen from nitro-gen by using polyimide prepared from different kinds of monomers[26,27]. Fluorinated polyimides usually possess higher H2 perme-ability and lower selectivity over other gases as compared withnon-fluorinated polyimides. For instance, 6FDA-DDBT polyimideexhibits a H2 permeability of 156 Barrers and a H2/CH4 selectivityof 78.8 [29] whereas BPDA-ODA polyimide has a H2 permeabilityof 1.33 Barrers and a H2/N2 selectivity of 365 [27]. Here, we havechosen a polyimide with a moderate H2/N2 selectivity for thefabrication of sodalite-polyimide membranes. Two monomers(benzophenone-3,30,4,40-tetracarboxylic dianhydride and 4,40-diaminodiphenylmethane) are thus used to synthesize polyimidethat is bonded directly to nanoparticles of organic-functionalizedsodalite, resulting in membranes free from interfacial defects. Thefabrication, characterization and separation performance of thesecomposite membranes are detailed in this paper.
2. Experimental
2.1. Sodalite synthesis and membrane fabrication
The amino-functionalized sodalite nanocrystals (denoted Sod-N) with a mean size of 105 nm were synthesized by transforming
Please cite this article in press as: D. Li et al., Micropor. Mesopor. Mater. (2009
silicalite nanocrystals according to our reported method [22].Briefly, a clear synthesis solution was prepared by dropwise addi-tion of 20 g of 1 M tetrapropylammonium hydroxide (TPAOH, Sig-ma–Aldrich) solution into a mixture of 17.8 g of tetraethylorthosilicate (TEOS, 99%, Sigma–Aldrich) and 1.8 g of 3-aminopro-pyl(diethoxy) methylsilane (ADMS, 97%, Sigma–Aldrich) withvigorous stirring, followed by continued stirring at room tempera-ture for 3 h and then crystallization at 80 �C for 12–15 days. Themilky silicalite suspensions so obtained were dried at 90–100 �Cto obtain solid silicalites. An alkaline solution was prepared bymixing 20 g of sodium hydroxide (99%, Merck), 9.2 g of sodium alu-minates (anhydrous, Sigma–Aldrich), and 60 g of deionized waterat room temperature for 1–2 h. We added 1 g of the dried silicalitesample (denoted Sil-N) to 11 g of the alkaline solution during 2–3 min of stirring, and then allowed it to age at room temperaturefor 4 h without further stirring. The transformation was carriedout at 80 �C for 4 h. The resulting amino-functionalized sodalitenanocrystals were cooled to room temperature and collected by re-peated cycles of washing with deionized water and centrifuging,followed by drying overnight at 90–100 �C.
Monomers benzophenone-3,30,4,40-tetracarboxylic dianhydride(BTDA; 96%, Sigma–Aldrich) and 4,40-diaminodiphenylmethane(MDA; 97%, Sigma–Aldrich) were dried at �150 �C for at least12 h under vacuum. Dimethylformamide (DMF) (GR, Merck) wasdried and stored with 4-ÅA
0
molecular sieves prior to use. To fabri-cate each composite membrane, a given quantity of Sod-N nano-crystals was dispersed in 10 g of DMF under ultrasonication atroom temperature for 30 min. Then 1.5 g of BTDA and 1.92 g ofMDA were dissolved in the Sod-N suspension. The resulting mix-ture was stirred for 5 h in an ice-water bath at approximately0 �C under N2 gas to obtain a Sod-N/PAA (polyamic acid) precursor,which was a cloudy yellow, viscous solution. The Sod-N/PAA solu-tion was cast directly onto a glass plate and placed into a vacuumoven and heat treated for 2 h each at 50 �C, at 100 �C and at 150 �C,before it was held at 200 �C overnight. The resulting sodalite–poly-imide nanocomposite membrane (denoted Sod-N/PI) was slowlycooled to room temperature. All of the yellow Sod-N/PI films wereimmersed in hot water at 90 �C for 1 h to allow removal from theglass plates, after which they were dried under vacuum at 150 �Covernight before analysis. In this paper, the sodalite–polyimidenanocomposite membranes were made with sodalite loadings of15, 25 and 35 wt.% (based on the mass of polyimide) and theseare denoted PI-15, PI-25, and PI-35, respectively. For comparison,pure polyimide membranes were prepared by applying the aboveprocedures without any Sod-N additions and these are referredto as PI-0.
2.2. Characterization
Scanning electron microscopy (SEM) images of cross sections ofmembranes were taken with a JSM-6300F microscope (JEOL). X-raydiffraction (XRD) patterns were measured on a Philips PW1140/90diffractometer with Cu Ka radiation (25 mA and 40 kV) at a scanrate of 1�/min with a step size of 0.01�. Fourier-transform infraredspectra (FT-IR) were recorded for the samples embedded in KBrpellets with a GX Spectrometer (Perkin–Elmer). Thermogravimet-ric analysis (TGA, Perkin–Elmer, Pyris 1 analyzer) was performedat a heating rate of 5 �C/min to 700 �C in oxygen with a flow rateof 15 cm3 min�1. Hydrogen adsorption–desorption experimentswere performed at 77 K and room temperature, and a pressure ofup to 900 mm Hg with a Micrometritics ASAP 2010MC analyzer.The samples were degassed at 473 K before analysis. To test gasseparation properties, the composite membrane or pure polyimidemembrane samples were firstly attached to a porous stainless-steel stand (pore size � 200 nm), which was then fixed in asample holder by using Torr Seal epoxy resin (Varian). Before
), doi:10.1016/j.micromeso.2009.05.014
Fig. 2. XRD patterns of samples Sod-N, PI-0, PI-15, PI-25, and PI-35. The peakslabeled with asterisks arise from Sod-N.
Fig. 3. IR spectra of samples Sod-N, PI-0 and PI-35.
D. Li et al. / Microporous and Mesoporous Materials xxx (2009) xxx–xxx 3
ARTICLE IN PRESS
measurements, the samples were evacuated and dried in a vacuumoven at 200 �C overnight to remove any residual solvent and ad-sorbed water. The gas permeation tests were performed at 25, 60and 100 �C on pure H2 and pure N2. The pressure rise of the perme-ate stream was measured with a Series 901 Transducer (MKS),which was connected to computer. Membrane permeability, Pi,was defined as [10,30],
Pi ¼dNi
DPiA
where d is the membrane thickness (cm), Ni the permeation rate ofcomponent i (cm3 s�1), DPi the transmembrane pressure differenceof i (cm Hg), and A the membrane area (cm2). 1 Barrer = 10�10
cm3(STP) cm cm�2 s�1 cm Hg�1. The ideal selectivity, aij, betweentwo gases, i and j, was defined as, [31,32]
aij ¼Pi
Pj
The apparent activation energy Ep was analyzed according tothe Arrhenius equation [31,33–35],
P ¼ P0 exp�Ep
RT
� �
where P is the permeability, P0 the pre-exponential factor, R theideal gas constant (8.3143 J mol�1 K�1) and T is the temperaturein Kelvin (K).
3. Results and discussion
3.1. Membrane characterization
Fig. 1 shows photographs of the series of polyimide compositemembranes with a thickness of 50 lm, which were all intact andhomogeneous, laid over the word ‘‘Monash”. Pure polyimides areclear, flexible and have good tear strength. All of the compositemembranes have a yellow appearance, but their transparency de-creases with increasing content of Sod-N nanocrystals (Fig. 1), asis evident from the gradual obscuration of the word from PI-0 toPI-35. Fig. 2 shows the XRD patterns of pure Sod-N and for PI-0,PI-15, PI-25 and PI-35. The Sod-N nanocrystals exhibit good crys-tallinity, giving sharp peaks in XRD pattern, which have been in-dexed in Fig. 2. In contrast, the pure polyimide membrane (PI-0)appears to be amorphous, as expected. With increasing contentsof Sod-N nanocrystals in the polyimide membranes, the peaks inFig. 2 increase in intensity from PI-15 to PI-35.
Fig. 1. Photos of PI-0, PI-15, PI-25 and PI-35 showing the change in transparencywith increasing Sod-N content.
Please cite this article in press as: D. Li et al., Micropor. Mesopor. Mater. (2009
Fig. 3 shows the IR spectra of Sod-N, PI-0 and PI-35. For the lasttwo samples, absorption bands, which correspond to the polyimidestructure, are observed at 1780 cm�1 (C@O asymmetric stretch-ing), 1720 cm�1 (C@O symmetric stretching), 1380 cm�1 (CANstretching), and 720 cm�1 (imide ring deformation); these indicatethe successful chemical imidization of the membranes [25,36–38].For the pure Sod-N sample, the broad band at approximately990 cm�1 is assigned to the asymmetric stretch (T-O-T, T = Si, Al),and the adsorption at 661 cm�1 is ascribed to the symmetricstretch (T-O-T) [22,39]. The presence of Sod-N in sample PI-35causes the peaks at around 1000 cm�1 to broaden in comparisonwith pure PI-0 film. Furthermore, there is a new small peak ap-pears for PI-35 at 661 cm�1, which is due to asymmetric stretchT-O-T (T = Si, Al) arising from added Sod-N.
Fig. 4 shows the SEM images of cross sections of PI-0, PI-15, PI-25 and PI-35. These micrographs confirm that Sod-N nanocrystalsare well dispersed throughout the polyimide matrix at all loadingsof Sod-N. No voids are apparent between the nanocrystals andpolyimide, even at 35-wt.% Sod-N where some large-scale surfaceroughness is evident, which suggests good bonding and compati-bility between the zeolite and polymer. Other studies also havefound that improving the interaction between zeolites and poly-mer tends to inhibit formation of interfacial voids [3,18,19,40].
The thermogravimetric (TG) curves of pure polyimide and thecomposite membranes with different loadings of Sod-N are shownin Fig. 5; Table 1 summarizes the corresponding thermogravimet-ric (TG) and differential thermogravimetric (DTG) results. Underflowing oxygen, the pure polyimide membrane, PI-0, lost 1.6% ofits mass in the temperature range from 30–400 �C. This is due tothe loss of residual organic solvent (DMF has a boiling point of153 �C) and/or adsorbed water. In the temperature range from400 to 700 �C, the remaining 98.4% of mass was lost, leaving no
), doi:10.1016/j.micromeso.2009.05.014
Fig. 5. TGA curves of samples PI-0, PI-15, PI-25, and PI-35.
Table 1DTG and TGA results of PI-0, PI-15, PI-25 and PI-35.
Fig. 6. H2 adsorption–desorption isotherms of amino-functionalized sodalitenanocrystals at 77 and 298 K.
Fig. 4. SEM images of cross sections of PI-0, PI-15, PI-25 and PI-35.
4 D. Li et al. / Microporous and Mesoporous Materials xxx (2009) xxx–xxx
ARTICLE IN PRESS
residue after the TGA run, which we ascribed to the completedecomposition and combustion of the polyimide at high tempera-ture [25,36]. The DTG peak (Td) for the corresponding mass loss liesat 573 �C.
The mass losses varied for the composite membranes duringheating between 30 and 400 �C – 3.0%, 2.7% and 3.4% for PI-15,PI-25 and PI-35, respectively – but all the composites lost moremass than PI-0. This might be due to increased adsorption of waterand/or DMF caused by the hydrophilic Sod-N particles and/or bythe presence of inorganic–organic cross-linked networks after
Please cite this article in press as: D. Li et al., Micropor. Mesopor. Mater. (2009
polymerization [36]. However, most of the mass loss occurs inthe temperature range from 400 to 700 �C and is 84.5% for PI-15,78.0% for PI-25 and 71.4% for PI-35. Interestingly, the Td valuesfor the composite materials are all higher than that of pure polyim-ide, and increase with Sod-N content: 580 �C for PI-15, 595 �C forPI-25 and 600 �C for PI-35. Some previous research attributed thiskind of trend to the interaction between the amino moieties frominorganic nanoparticles (Sod-N) and the polymer matrix, whichcan reduce the movement (increase the rigidity) of the polymerchains and, thus, increase the decomposition temperature of com-posite membranes [25,37,41].
The residual masses after TG analysis are 12.5%, 19.3% and25.2% for PI-15, PI-25 and PI-35, which would correspond to plainsodalite nanocrystals, given that organic functional groups (i.e.,A(CH3)(CH2)3NH2) would have been completely decomposed andremoved by the high temperatures [42]. The use of 29Si-NMR inour previous work showed that 3.18 wt.% of Sod-N comprises or-ganic functional groups [22]. This allows recalculation of the actualSod-N loading of PI-15, PI-25 and PI-35 as 14.8%, 24.7% and 34.8%,respectively, based on the mass of polyimide, which are close tothe theoretical values.
3.2. H2 sorption of sodalite nanocrystals and gas permeation ofmembranes
H2 sorption isotherms of amino-functionalized sodalite nano-crystals are shown in Fig. 6. It is clear that the temperature hassubstantial influence on H2 adsorption capacity of sodalite nano-crystals. At 77 K, H2 adsorptive volume significantly increases withincreasing the adsorption pressure, and it reaches a maximum vol-ume of 26.9 cm3/g. However, at room temperature (298 K), amino-functionalized sodalite nanocrystals exhibit almost no H2 adsorp-tion as P/Po is raised to 1 (Po = 900 mm Hg). This is due to sodalitecage contraction when the sorption temperature increases from77 K to 298 K. XRD analysis confirms that the crystallinity in ami-no-functionalized sodalite nanocrystals remains unchanged afterH2 sorption analysis. According to Ref. [21], sodalite cage expandsand starts to uptake hydrogen at 573 K or above. These indicatethat amino-functionalized sodalite nanocrystals may function asnonporous nanoparticles in nanocomposite membranes in ourgas permeation temperatures.
Table 2 summarizes the permeability values of two pure gases(H2 and N2) and the ideal selectivity aðH2=N2Þ for pure polyimidefilms and composite membranes at three different temperatures(25, 60 and 100 �C). Our permeability and ideal selectivity datafor pure polyimide membranes fabricated from BTDA and MDA iscomparable to similar polyimide membranes in the literature
), doi:10.1016/j.micromeso.2009.05.014
Table 2Gas permeation results of the PI-0, PI-15, PI-25, and PI-35 membranes.
Fig. 8. Ideal selectivity aðH2=N2 Þ of PI-0, PI-15, PI-25, and PI-35 at differenttemperatures (25, 60 and 100 �C).
D. Li et al. / Microporous and Mesoporous Materials xxx (2009) xxx–xxx 5
ARTICLE IN PRESS
[29]. The addition of Sod-N causes the composite membranes to re-duce the permeation of N2, leading to a substantial improvement inH2/N2 selectivity. This should be attributed to the interfacial effectsand disrupted polyimide chain packing caused by the covalentbonding between Sod-N and polyimide. The structure of sodalite-polyimide interface is illustrated in Fig. 7a. Sodalite nanocrystalsare composed of a crystalline sodalite core and a thin amorphousaluminosilicate shell with amino-groups (@SiA(CH3)(CH2)3NH2).The thickness of the amorphous aluminosilicate shell is roughlyestimated to be around 2 nm assuming that all amino-groups arecontained in the shell [22]. The high-quality bonding betweenthe sodalite nanocrystals and the polymer matrix is realized byforming covalent linkers via the imidization reaction of the ami-no-groups with the polyimide monomers (Fig. 7b). The additionof Sod-N also affects the chain length of polyimide molecules sur-rounding Sod-N nanocrystals because the polyimide chains react-ing with amino-groups are terminated. This would increase therigidity of the polymer chains in the interfaces and polyimide ma-trix [6,7]. These unique structures allow H2 to diffuse throughwhile reducing the passage of N2 molecules. This explains thatthe H2 permeability of all composite membranes at 25 �C is slightlyhigher than that of the pure polyimide membrane. On the otherhand, these data provide strong evidence that there are no voidspresent at the polyimide and sodalite interface in any of the com-posite membranes, because such voids would have resulted in alarge increase in permeability of H2 or even N2 [18].
When the testing temperature is elevated, there is a subsequentincrease in the permeability of H2 or N2 for the pure-polyimide andthe composite membranes. There was a more significant increasein permeability for the pure polymer with temperature than wasfound for the composite membranes, especially for N2 gas. PI-0has a N2 permeability of 0.036 Barrer at 25 �C, compared with0.13 Barrer at 100 �C, a 3.6-fold increase. However, PI-35 showed
Fig. 7. Schematic representation of sodalite–polyimide interfacial structure (a) andcovalent linker between Sod-N and polyamide (b).
Please cite this article in press as: D. Li et al., Micropor. Mesopor. Mater. (2009
an increase of only 1.1 times for PN2 between room temperatureand 100 �C. In addition, at 60 and 100 �C the permeabilities of H2
and N2 for the composite membranes are lower than those forthe PI membrane (Table 2). As the temperature increases, the per-meabilities of both H2 and N2 increase because of the increase ofthe diffusivity and the decrease of the solubility in polyimides[43]. We attribute this result to the increase in polymer chain rigid-ity in the composite membranes with increasing Sod-N loading,and the increase in the permeabilities of both H2 and N2 for thePI membrane is greater than those for the composite membranes.
The H2/N2 selectivity for PI-0, PI-15, PI-25 as a function of tem-perature, which are included in Table 2, are shown in Fig. 8. At25 �C, the nanocomposite membranes demonstrate perselectivitiesof 223, 238, and 281 for PI-15, PI-25, and PI-35, respectively. Thesevalues represent 13%, 20%, and 46% greater ideal selectivity,respectively, than PI-0.
It is also plain from Fig. 8 that elevating the temperature lowersthe ideal selectivities of all the membranes, but that increasing thesodalite content considerably retards the falling-off of gas selectiv-ity from 25 to 100 �C. For instance, PI-0’s selectivity drops from 198at 25 �C to 110 at 100 �C, which is a fall of 44%. In contrast, PI-35sees a decrease in aðH2=N2Þ of only 25%. As Sod-N loading increases,the number of Sod-N terminated increases substantially, affectingthe chain configuration beyond the interfaces; in other words,the interfacial area may be extended. Therefore, the increased inor-ganic content in composite materials restricts the thermal motionof the polymer segments, and thus reduces the decrease in the gasselectivity [31].
Fig. 9 shows the apparent activation energy, Ep, of PI-0, PI-15, PI-25 and PI-35 for the pure H2 and pure N2. It is apparent that allsamples have higher values Ep for N2 than H2, confirming that N2
molecules need more energy to penetrate the membranes thanH2 molecules. Compared with composite membranes, pure poly-imide polymer (PI-0) has the highest activation energies – 8.5 kJ/mol for H2 and 15.9 kJ/mol for N2. In the composite membranes,the presence of Sod-N lowers Ep below that of the pure polymermembranes. For example, PI-15 and PI-25 have Ep values of 6.7and 4.1 kJ/mol, respectively, for H2 and 14.9 and 9.5 kJ/mol, respec-tively, for N2. Interestingly, PI-35 shows an increase in activationenergy relative to PI-25 for H2, but not for N2. Similarly, the H2
permeability for PI-35 increases largely from 9.9 Barrers to 13.1Barrers as the temperature is increased from 25 to 100 �C. In thecomposite membranes, gas diffusion requires relatively smallsegmental motions of polymer matrix in the packing-disruptedpolyimide chains and sodalite–polyimide interfaces, because theypossess relatively more unoccupied free space. When Sod-Nloading is increased to a certain point (e.g., 35%), the overlap of
), doi:10.1016/j.micromeso.2009.05.014
Fig. 9. Apparent activation energy (Ep) for PI-0, PI-15, PI-25 and PI-35.
6 D. Li et al. / Microporous and Mesoporous Materials xxx (2009) xxx–xxx
ARTICLE IN PRESS
interfacial layers becomes significant [44]. Such overlapped inter-faces favor H2 diffusion, and are more temperature-dependent inthe permeation of small hydrogen molecules. It is clear that theseparation performance of polyimide membrane has been signifi-cantly enhanced. The strategy of forming nanocomposite mem-branes demonstrated in this work could be applied to fabricatepractical H2 separation membranes by incorporating functionalsodalite nanocrystals into a more permeable polyimide skin layer.It would be interesting to study water transport property of soda-lite-polyimide nanocomposite thin membranes for potential appli-cations, such as in water/organic solvent separation and waterpurification, given that sodalite membranes have been reportedto exhibit good water permeation property [45].
4. Conclusions
We have used organic-functionalized sodalite nanocrystals(Sod-N) and polyimide to fabricate nanocomposite membranes.Characterization by SEM showed that Sod-N can be well distrib-uted with polyimide phase, even at a loading of 35 wt.%, as is con-firmed by the FTIR spectroscopy and XRD results. From TG and DTGanalysis, the DTG peaks for corresponding major mass loss increasewith the increasing Sod-N content of the composite, which isattributed to restricted movement of the main chains arising fromthe interaction between the amino moieties from inorganic nano-particles (Sod-N) and polymer matrix. The gas permeation experi-ments were performed with two pure gases, H2 and N2, and theresults revealed that H2 permeability was improved, while N2 per-meability decreased. In particular, the PI-35 composite membraneshad the highest ideal selectivity (aðH2=N2Þ = 281) and a good perme-ability (8.0 Barrers) at room temperature.
Acknowledgments
This work was supported by the Australian Research Council(ARC) and the CSIRO Flagships – Advanced Membrane Technology
Please cite this article in press as: D. Li et al., Micropor. Mesopor. Mater. (2009
for Water Treatment Cluster. H.W. thanks the ARC for the QEII Fel-lowship. D.L. gratefully acknowledges Monash University for thepostgraduate scholarships.
References
[1] H. Cong, M. Radosz, B.F. Towler, Y. Shen, Sep. Purif. Technol. 55 (2007) 281.[2] C.V. Funk, D.R. Lloyd, J. Membrane Sci. 313 (2008) 224.[3] Y. Li, H.-M. Guan, T.-S. Chung, S. Kulprathipanja, J. Membrane Sci. 275 (2006) 7.[4] C.F. Zeng, L.X. Zhang, X.H. Cheng, H.T. Wang, N.P. Xu, Sep. Purif. Technol. 63
(2008) 628.[5] D. S�en, H. KalIpçIlar, L. Yilmaz, J. Membrane Sci. 303 (2007) 194.[6] T.T. Moore, W.J. Koros, J. Mol. Struct. 739 (2005) 87.[7] Z. Huang, Y. Li, R. Wen, M.M. Teoh, S. Kulprathipanja, J. Appl. Polym. Sci. 101
(2006) 3800.[8] M.G. Süer, N. Baç, L. Yilmaz, J. Membrane Sci. 91 (1994) 77.[9] Y. Li, T.-S. Chung, C. Cao, S. Kulprathipanja, J. Membrane Sci. 260 (2005) 45.
[10] H.T. Wang, B.A. Holmberg, Y.S. Yan, J. Mater. Chem. 12 (2002) 3640.[11] G. Golemme, A. Bruno, R. Manes, D. Muoio, Desalination 200 (2006) 440.[12] R. Mahajan, W.J. Koros, Ind. Eng. Chem. Res. 39 (2000) 2692.[13] R. Mahajan, W.J. Koros, Polym. Eng. Sci. 42 (2002) 1420.[14] R. Mahajan, W.J. Koros, Polym. Eng. Sci. 42 (2002) 1432.[15] R. Mahajan, R. Burns, M. Schaeffer, W.J. Koros, J. Appl. Polym. Sci. 86 (2002)
881.[16] J.-M. Duval, A.J.B. Kemperman, B. Folkers, M.H.V. Mulder, G. Desgrandchamps,
C.A. Smolders, J. Appl. Polym. Sci. 54 (1994) 409.[17] H.H. Yong, H.C. Park, Y.S. Kang, J. Won, W.N. Kim, J. Membrane Sci. 188 (2001)
151.[18] T.W. Pechar, S. Kim, B. Vaughan, E. Marand, M. Tsapatsis, H.K. Jeong, C.J.
Cornelius, J. Membrane Sci. 277 (2006) 195.[19] T.W. Pechar, M. Tsapatsis, E. Marand, R. Davis, Desalination 146 (2002) 3.[20] T.C. Merkel, B.D. Freeman, R.J. Spontak, Z. He, I. Pinnau, P. Meakin, A.J. Hill,
Science 296 (2002) 519.[21] J. Weitkamp, M. Fritz, S. Ernst, Int. J. Hydrogen Energy 20 (1995) 967.[22] D. Li, J.F. Yao, H.T. Wang, N. Hao, D.Y. Zhao, K.R. Ratinac, S.P. Ringer, Micropor.
Mesopor. Mater. 106 (2007) 262.[23] J.F. Yao, H.T. Wang, K.R. Ratinac, S.P. Ringer, Chem. Mater. 18 (2006) 1394.[24] R.W. Baker, Membrane Technology and Applications, John Wiley, New York,
2004.[25] M.S. Boroglu, I. Boz, M.A. Gurkaynak, Polym. Adv. Technol. 17 (2006) 6.[26] K. Tanaka, H. Kita, M. Okano, K.-I. Okamoto, Polymer 33 (1992) 585.[27] Y. Li, X. Wang, M. Ding, J. Xu, J. Appl. Polym. Sci. 61 (1996) 741.[28] M.K. Ghosh, K.L. Mittal, Polyimides: Fundamentals and Applications, Marcel
Dekker, New York, 1996.[29] L.M. Robeson, J. Membrane Sci. 320 (2008) 390.[30] L.Y. Jiang, T.S. Chung, S. Kulprathipanja, AIChE J. 52 (2006) 2898.[31] H.B. Park, J.K. Kim, S.Y. Nam, Y.M. Lee, J. Membrane Sci. 220 (2003) 59.[32] H. Lin, E. Van Wagner, B.D. Freeman, L.G. Toy, R.P. Gupta, Science 311 (2006)
639.[33] J.P.G. Villaluenga, B. Seoane, J. Hradil, P. Sysel, J. Membrane Sci. 305 (2007) 160.[34] S. Pauly, Permeability and diffusion data, in: J. Brandrup, E.H. Immergut, E.A.
Grulke (Eds.), Polymer Handbook, John Wiley, New York, 1999. Chapter 6.[35] L.Y. Jiang, T.S. Chung, S. Kulprathipanja, J. Membrane Sci. 276 (2006) 113.[36] S.-H. Zhong, C.F. Li, X.F. Xiao, J. Membrane Sci. 199 (2002) 53.[37] S. Al-kandary, A.A.M. Ali, Z. Ahmad, J. Mater. Sci 41 (2006) 2907.[38] J.S. Im, J.H. Lee, S.K. An, K.W. Song, N.J. Jo, J.O. Lee, K. Yoshinaga, J. Appl. Polym.
Sci. 100 (2006) 2053.[39] M. Alkan, C. Hopa, Z. Yilmaz, H. Guler, Micropor. Mesopor. Mater. 86 (2005)
176.[40] C.C. Hu, T.C. Liu, K.R. Lee, R.C. Ruaan, J.Y. Lai, Desalination 193 (2006) 14.[41] L. Liu, B. Liang, W. Wang, Q. Lei, J. Compos. Mater. (2006) 2175.[42] S. Li, Z.J. Li, D. Medina, C. Lew, Y.S. Yan, Chem. Mater. 17 (2005) 1851.[43] T.H. Kim, W.J. Koros, J. Membrane Sci. 46 (1989) 43.[44] L. Khounlavong, V. Ganesana, J. Chem. Phys. 130 (2009) 104901.[45] M. Kazemimoghadam, A. Pak, T. Mohammadi, Micropor. Mesopor. Mater. 70