Cooperative Formation of Inorganic-Organic interfaces in the synthesis of silicate mesostructures

Cooperative Formation of Inorganic-Organic interfaces in the synthesis of silicate...

(Parte 1 de 2)

Cooperative Formation of Inorganic-Organic Interfaces in the Synthesis of

Silicate Mesostructures

A. Monnier, F. Schuth, Q. Huo, D. Kumar, D. Margolese, R. S. Maxwell, G. D. Stucky,* M. Krishnamurty, P. Petroff, A. Firouzi, M. Janicke, B. F. Chmelka

A model is presented to explain the formation and morphologies of surfactant-silicate mesostructures. Three processes are identified: multidentate binding of silicate oligomers to the cationic surfactant, preferential silicate polymerization in the interface region, and charge density matching between the surfactant and the silicate. The model explains present experimental data, including the transformation between lamellar and hexagonal mesophases, and provides a guide for predicting conditions that favor the formation of lamellar, hexagonal, or cubic mesostructures. Model Q230 proposed by Mariani and his co-workers satisfactorily fits the x-ray data collected on the cubic mesostructure material. This model suggests that the silicate polymer forms a unique infinite silicate sheet sitting on the gyroid minimal surface and separating the surfactant molecules into two disconnected volumes.

The invention ofa new family ofmesoporous silica materials, designated M41S, by scientists at Mobil Oil Corporation (1), has dramatically expanded the range of crystallographically defined pore sizes from the micropore (< 13 A) to the mesopore (20 to 100 A) regime. The synthesis uses ordered arrays of surfactant molecules as a "template" for the three-dimensional polymerization of silicates. The mesoporous materials obtained by this route exhibit several remarkable features: (i) well-defined pore sizes and shape, as compared to other mesoporous materials; (i) fine adjustability of the pore size within the limits stated above; (i) high thermal and hydrolytic sta-

bility ifproperly prepared; and (iv) a very high degree of pore ordering over micrometer length scales. These unusual properties are a direct result of the interplay between organized arrays of the surfactant molecules and silicate species in the aqueous phase. Beck et al. (2) outlined two general pathways for the formation of the mesoporous silicates. The first model assumes that the primary structure-directing element is the water-surfactant liquid crystal phase. The second model suggests that the addi-

A. Monnier, Department of Chemistry, University of

California, Santa Barbara, CA 93106, and Departement de Chimie Physique Sciences I, 1211 Geneva, Switzerland. F. Schuth, Department of Chemistry, University of

California, Santa Barbara, CA 93106, and Institut fur Anorganische Chemie, Johannes-Gutenberg Universitat, 6500 Mainz, Germany. Q. Huo, D. Kumar, D. Margolese, R. S. Maxwell, G. D. Stucky, Department of Chemistry, University of California, Santa Barbara, CA 93106. M. Krishnamurty and P. Petroff, Materials Department,

University of California, Santa Barbara, CA 93106. A. Firouzi, M. Janicke, B. F. Chmelka, Department of Chemical and Nuclear Engineering, University of California, Santa Barbara, CA 93106.

*To whom correspondence should be addressed.

tion of the silicate orders the subsequent silicate-encased surfactant micelles. These general models, however, are insufficient for establishing the mechanistic under- standing needed for better control of the synthesis process, which is key to efforts aimed at improving or adding to this exciting new class of materials. On the basis of experimental results, we present here a more detailed model of the mesophase formation process, which explains presently known experimental data and successfully predicts conditions needed for the synthesis of desired structures. We believe that this model can be generalized to the synthesis of other nonsiliceous materials as well.

From considerations in surfactant and silicate chemistry, three closely coupled phenomena are identified as crucial to the formation of surfactant-silicate mesophases. These include: (i) multidentate binding of silicate oligomers, (i) preferred polymerization of silicates at the surfactant-silicate interface, and (i) charge density matching across the interface.

Mesostructure syntheses can be carried out under conditions in which the silicate alone would not condense (at pHs from 12 to 14 and silicate concentrations of 0.5 to 5%) and the surfactant cetyltrimethylam- monium (CTA+) alone would not form a liquid crystal phase. In fact, surfactantsilicate mesophases can form at surfactant concentrations as low as 1%, a regime in the CTABr-water phase diagram in which only micelles are present. For a CTABrwater solution at typical surfactant-silicate synthesis temperatures in the absence of silicates, a hexagonal phase is favored at surfactant concentrations from -25 to 70% by weight whereas a lamellar phase forms at concentrations above 70% (3, 4). Never- theless, a solid mesophase precipitate is formed, the structure of which will be discussed below, as soon as surfactant (chain length of 8 to 20 carbon atoms) and silicate solutions are combined. The rapidity of this precipitation indicates that there is a strong interaction between the cationic surfactant and anionic silicate species in the formation of surfactant-silicate mesophases.

We performed syntheses aimed at identifying conditions important for the formation of mesoporous materials over a wide range of reactant compositions and temper- atures (5). For the purpose of investigation, we found that we could slow the evolution of the surfactant-silicate systems by undertaking the syntheses at moderate temperatures (between 300 and 100'C) (6). During freeze-dry kinetic experiments with CTACI used as the surfactant, a layered (lamellar) material with a primary d spacing (repeat distance) of 31 (+ 1) A was produced, together with amorphous silica, after reaction times on the order of 1 min. For the synthesis conditions given in Fig. 1, the lamellar mesophase disappears after approximately 20 min, at which point the diffrac- tion pattern of the hexagonal mesostructure is simultaneously detected. This hexagonal material has a primary d spacing of 40(± 1) A and attains its final degree of ordering after --10 hours (7). A layered material with a primary d spacing of 31 (± 1) A (Fig. 2, pattern A) can be isolated in pure form (8); a transmission electron microscopy (TEM) micrograph of this mesostructure is depicted in Fig. 3. The variation of the d spacing as a function of the chain length of a cationic surfactant CH2, 1[NCCH3)31+ (for 14 ' n < 2) is 1.0 to 1.2 A per carbon, which a. Layered phase c l ; Hexagonal phase

Fig. 1. Time evolution of the intensity of x-ray diffraction features associated with layered and hexagonal (M41S) mesostructures at 348 K. The layered material is precipitated rapidly, whereas the hexagonal material appears later, as a result of a higher degree of silica polymerization. The composition of the reaction mixture was as follows: 1 M SiO2:0.025 M A1203:0.115

M Na2O:0.233 M CTACI:0.089 M TMAOH:125 M H2O.

m'REPORTS on February 18, 2010 Downloaded from corresponds to a monolayer assembly. Ifthis new layered material is hydrothermally treated at 373 K (pH = 7), it is converted to the hexagonal mesostructure over 10 days, with intermediate and final x-ray patterns shown in Fig. 2, patterns B and C, respectively. During this transformation the degree of silica polymerization increases, as measured by the relative number of incompletely condensed (Q3) and fully condensed (Q4) silicon atoms determined by 29Si magic-angle spinning nuclear magnetic resonance spectroscopy. The ratio between Q3 and Q4 silicon decreases from typical values of 1.0 for the layered material to 0.4 to 0.5 for the hexagonal mesostructure, reflecting a significant increase in the number of silicon atoms fully coordinated to other silicate nearest neighbors.

Mesophase formation and associated silica polymerization are intimately tied to Coulombic interactions between surfactant and silicate species at the micelle interfaces. Silicates present in the form of monovalent monomers, Si(OH)30-, however, are expected to have little energetic advantage over other monovalent anions competing for access to the cationic surfactant head groups. At high pH, the reaction mixture also contains small silica oligomers

(three to seven silicon atoms) of varying degrees of polymerization and charge (9). These oligomers are appreciably more acid-

ic (pKa 6.5) than the monomer or dimer species [PKa 9.8 and 10.7, respectively

(10)], although all such silicates will be highly dissociated under the high pH conditions used here (1 1).

The oligomeric silica polyanions, how- ever, can easily act as multidentate ligands for the cationic head groups of the surfac- tant, leading to a strongly interacting surfactant-silicate interface. Indeed, the interaction of ionic surfactants with polyions of opposite charge encourages strong cooperative binding, manifested by increases in the binding constants of up to two orders of

Fig. 2. Powder x-ray diffraction patterns of surfactant-silicate mesostructures precipitated from the same reaction mixture (1 M SiO2: 0.034 M A1203:0.07 M Na2O:0.27 M CTABr:0.14 M TMAOH:0.28 M TMB:100 M H20), and then treat-

ed hydrothermally at 373 K for different times. X-ray patterns are shown for (curve A) the initially precipitated layered material, (curve B) an intermediate material, and (curve C) the M41 S hexagonal mesostructure acquired 0, 1, and 10 days, respectively, after initiation of the hydrothermal treatment.

p c0 magnitude in similar systems (12). Preferential multidentate binding of the silicate polyanions causes the interface to quickly become populated by tightly held silicate oligomers, which can subsequently polymerize further. Silicate polymerization within the surfactant-silicate interface region is favorable for two related reasons: (i) the concentration of silicate species near the interface is high and (i) their negative charges are partially screened by the surfac- tant. Furthermore, as polymerization proceeds, the formation of highly connected silicate polyanions, which act as very large multidentate ligands, further enhances the cooperative binding between the surfactant and silicate species.

Multidentate ionic binding in surfactant-silicate systems has an important consequence; namely, it leads to precipitation of a given mesophase from solution. Through the interactions driving the precipitation process, the appearance of a given mesostructure is established, although this process is expected to operate on a different time scale from polymerization of the silica, which accounts ultimately for the thermal, mechanical, and hydrolitic stability of the final material. If small silica oligomers are present in sufficient quantity, precipitation of the surfactant-silicate system is primarily the result of electrostatic interactions, combined with packing constraints associated with the hydrophobic surfactant chains. Whereas precipitation is fast and essentially thermodynamically controlled, silica polymerization into a strong and extended framework is slow and reaction rate-limited. This two-stage process is in agreement with experimental findings that contrast the mesostructures obtained at room temperature after short reaction times with those obtained at high temperature after long reaction times: very similar x-ray patterns are obtained for both sets of conditions, indicating identical precipitated mesostructures; however, the materials syn- thesized by the low-temperature route are thermally and mechanically much less stable than the high-temperature analogs. The coupling between the precipitation and polymerization processes in surfactant-silicate systems provides the basis for the lamellar-to-hexagonal mesophase transformation in a way that we now describe. The resemblance, in shape and size, of the surfactant-silicate mesostructures with the corresponding water-surfactant liquid crystal phases indicates that the interactions responsible for these morphologies are of a similar nature. The governing role of the head-group area (A) in the selection of a particular mesophase has already been recognized in water-surfactant systems: the favored mesophase is that which permits A to be closest to its optimal value AO, while maintaining favorable packing of the hy- drophobic surfactant chains (13). In surfac- tant-silicate systems, the value of Ao is strongly affected by electrostatic and steric interactions between the silicate and surfactant micelle species. More quantitatively, its value is obtained by minimizing the Gibbs free energy G as a function of A:

Ao- (aG/aA) = 0

G(A,p) = Gintra(A) + Gwall(P) where Gintra reflects the van der Waals forces and conformational energy of the hydrocarbon chains and the van der Waals and electrostatic interactions of the head group within a single micelle; Gwaii accounts for the polysilicate structural free energy, including the solvent, counterion, and silicate van der Waals and electrostatic interactions within the inorganic silicate framework or "wall"; Ginter reflects the van der Waals and electrostatic effects associated with wall-micelle and micelle-micelle interactions; G..1 describes the solution phase; and p specifies the compositions of the various species within the wall. The chemical potentials of these species are set by the concentration of the corresponding species in the aqueous solution, as account- ed for by G501.

I Fig. 3. Transmission electron micrograph of the layered surfactant-silicate mesostructure whose x-ray data are shown in Figs. 1 and 2 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 (curve A). The d spacing of this material is

20(d) 31(+1) A.

SCIENCE * VOL. 261 * 3 SEPTEMBER 1993M'W-5,0

on February 18, 2010 Downloaded from

Physically, Gintra governs the formation of a particular micelle shape for a given value of A and is also responsible for the observed swelling of the micelles when hydrocarbon "expanders," such as trimethylbenzene (TMB) are added to the solution

(14). The term Gwaii drives the chemistry within the wall, including the polymeriza- tion process, and contains the structural constraints responsible for the multidentate binding. The term Ginter establishes the relation between A and the state of the wall described by p. This coupling across the interface can be understood in terms of electrostatic interactions (which most likely predominate), whereby the silicate charge density within the wall, p,, is mutually screened by the charges on the sur- factant head groups, which have an average surface charge density of 1/A. Thus, the electrostatic interactions link AO, as defined by Eq. 1, with p6, a relation we refer to as

"charge density matching." Such interdependent electrostatic effects control the d spacings of surfactant intercalates in different mica-type silicates (15) and have been invoked to explain the "self-replication" process of silica layers in purely inorganic systems (16).

In surfactant-silicate systems, polymer- ization driven by Gwall will profoundly affect

Pe, providing a mechanism to explain the transition between the lamellar and hexag- onal mesophases. In the early stage of the synthesis, the presence of highly charged silica oligomers favors a small value of A0, which can be achieved with a lamellar surfactant configuration. As rearrangement and polymerization of the silicate species proceed, the density of anionic silanol groups diminishes, so that Ao increases, while the number of compensating cations decreases. At the same time, the wall thickness can decrease from its initial value without energy cost, because the most stable ionized silanol groups are confined to the wall surface, thus reducing repulsive dipole-dipole interactions between the two opposite-facing wall surfaces. The silicate wall is still poorly condensed during early stages of the synthesis, allowing the system to increase its A toward Ao by adopting the hexagonal structure according to charge- density matching criteria. Under these circumstances, the wall thickness simultaneously decreases to keep the volume ratio CTA/Si02 constant. The actual wall thickness has been estimated to be 10 to 1 A (17) for the lamellar mesophase and 8 to 9 A (18) for the hexagonal mesophase. Simple geometrical arguments can be used to show that these values are consistent with a con- stant CTA/Si02 volume ratio throughout the phase transition.

The regularity of the product mesostructures supports mediation of the silicate wall

Fig. 4. Schematic diagramD..

of the mechanism pro- posed for the transformation of a surfactant-silicate system from the lamellar to the hexagonal mesophase. On the left, small silica oligomers (not shown ex- plicitly in the gray SiO2 region) act as multidentate

ligands, which have sufficiently high charge density to permit a lamellar surfac- tant configuration. As polymerization of the silica pro- ceeds, diminished charge density of larger silica polyanions increases the *SiO2 1 XJeReactionocoordinate average head-group area of the surfactant assembly, driving the transformation into the hexagonal mesophase.

Fig. 5. Chart showing the approximate domains of formation of the lamellar and hexagonal surfactantsilicate mesophases, as functions of pH and silica source. Cab-O-Sil is composed of -100 A oligomeric silica particles, whereas Na-silicate is a solution of hydrolyzed and essentially monomeric silicates.


AITTTITT|U3P'Clear _------

IsNa-silicate Cab-O-Sil

pH of reactionAmorphous g Hexagonal 1 Lamellar thickness during the assembly process. The high efficiency ofthis mediation is reflected by the experimental observation that the wall thickness ofthe hexagonal phase is essentially constant (8 to 9 A) over a wide range of reaction conditions, independent of the sur- factant chain length, and by the clearly hexagonal, as opposed to circular, pore shape established by both high-resolution TEM and modeling of the powder x-ray diffraction pat-terns (19). Control of the silicate wall thick- ness is undoubtedly related to the double layer potential: silicate species are only accumulated at the surfactant interface to the extent necessary for charge compensation. Polymerization normal to the interface, which would thicken the wall or produce amorphous bulk SiC2, does not occur because of the strong electrostatic repulsion produced by the high negative charge on the silicate species at pH

12 and above (10). Figure 4 shows a mechanism consistent with current experimental investigations by which the lamellar-to-hexagonal meso- phase transformation may occur. Silica polymerization leads to an increase in interfacial area that is achieved through corrugation of the lamellar surfactant-silicate sheets. As implied in the final step, this corrugation progresses until connection between the sheets is made at the cusps, resulting ulti- mately in the formation of the hexagonal mesophase. Another way to accommodate the change in A would be to maintain a planar structure while tilting the hydrocarbon chains. Such a transition, however, is entropically disfavored by the restrictive chain configuration this suggests.

Yanagisawa et al. (20) recently reported a hexagonal mesostructure, with pore di- mensions similar to that ofM41S, produced by the inclusion of CTAt cations into the sheet silicate kanemite. During their syn- thesis these researchers observed a layered intermediate that subsequently transformed into a hexagonal phase material. This process is probably driven by the same forces as the transformation we report, although it is not yet clear to what extent the kanemite structure is preserved during the conversion. If the pH is sufficiently basic, for example, the sheets can be partially or fully destroyed during the process.

We propose that the surfactant-silicate mesophase structure is governed primarily by the terms Gintra and Ginter of Eq. 2. In this respect, the main effect of the silicate wall and of the reaction conditions are to determine A0. This provides predictive capability for establishing the reaction condi- tions that favor the lamellar or the hexagonal mesophases. We have tested this mod- el experimentally by monitoring the effects of pH and the degree of polymerization of the silica source on the mesostructure syntheses, with the results summarized in Fig.

on February 18, 2010 Downloaded from

Fig. 6 (left). Transmission electron micrograph of the cubic surfactantsilicate mesostructure showing an ordered -2000 A aggregate viewed along its [1 1 1] axis. Fig. 7 (right). X-ray powder diffraction pattern of the cubic mesostructure, with la3d symmetry, synthesized from a reac- tion mixture with a molar composition of 1 M TEOS:0.25 M Na2O:0.65 M CTACI:62 M H20 for 3 days at 373 K (curve A). Calculated diffraction pattern using the Q230 model proposed by Mariani et al. (21) with a lattice parameter a = 97.3 A (curve B).

5. These results lead to the following conclusions in accordance with our predictions: 1) The lamellar phase is favored at high pH and for a low degree of polymerization of the silica source. 2) The hexagonal phase is favored at low pH and for a highly polymerized silica source.

(Parte 1 de 2)