Report

Manganese Oxide Mesoporous Structures: Mixed-Valent Semiconducting Catalysts

See allHide authors and affiliations

Science  09 May 1997:
Vol. 276, Issue 5314, pp. 926-930
DOI: 10.1126/science.276.5314.926

Abstract

Hexagonal and cubic phases of manganese oxide mesoporous structures (MOMS) have been prepared by means of the oxidation of Mn(OH)2. The hexagonal MOMS materials form a hexagonal array of pores with an open porous structure, thick walls (1.7 nanometers), and exceptional thermal stability (1000°C). The walls of the mesopores are composed of microcrystallites of dense phases of Mn2O3 and Mn3O4, with MnO6 octahedra as the primary building blocks. The calcined hexagonal MOMS have an electrical conductivity of 8.13 × 10−6 per ohm·centimeter, an average manganese oxidation state of 3.55, and a band gap of 2.46 electron volts. Catalytic oxidations of cyclohexane and n-hexane in aqueous solutions in a batch reactor show conversions of ∼10 and ∼8 percent, respectively. Characterization and catalytic data suggest that MOMS systems show significant enhancement in thermal stability with respect to octahedral molecular sieve materials.

Since their discovery at Mobil Corporation in 1992, mesoporous aluminosilicate materials (designated M41S) (1) have attracted considerable attention. Significant advances resulting from research in this area include novel properties of these materials (2), new synthetic chemistry (1-4), unique structures (5), and synthesis of related materials (5).

Oxides of transition metals have some advantages over aluminosilicate materials for use in electromagnetics, photoelectronics, and catalysis because transition metal atoms can exist in various oxidation states. However, syntheses and structures of transition metal oxides can be much more complicated than oxides of main group metals because of the multitude of different coordination numbers and oxidation states. Mesoporous structures of transition metals doped into aluminosilicates [Cr (6)] or transition metal oxides such as Ti (7), V (8), W (3, 5), Zn (9), Nb (10), and Ta (11) have been reported. Most of these transition metal oxide mesoporous materials are insulators with transition metals in isolated oxidation states. The generation of mixed-valent transition metal oxide mesoporous materials might lead to versatile systems for redox catalysis and battery applications (12).

The wall materials of the mesoporous aluminosilicate and transition metal oxide materials discussed above are believed to be composed of noncrystalline amorphous phases (1, 3, 13), that is, the materials can be viewed as glasses with ordered pores. There is considerable interest in determining the coordination numbers of the metal atoms (primary structure) and local structure of the wall materials (secondary structure) because the amorphous (local structure) nature of the walls has been largely unexplored.

We report here the synthesis of semiconducting mixed-valent manganese oxide mesoporous structures (MOMS), which are believed to consist of crystalline wall material and are related to octahedral molecular sieve (OMS) materials (12, 14, 15). Mixed-valent MOMS systems may represent a distinct family of mesoporous materials.

Cetyltrimethylammonium bromide (CTAB) cationic surfactants were used as micellar templates in water. Air and inexpensive chemicals (MnCl2 and NaOH) were used as starting materials (16). Precursors were prepared by two distinct steps: (i) the formation of a layered phase of Mn(OH)2 and the generation of an aqueous solution of surfactant in a separate system (16) and (ii) the mixing of the two systems such that the crystallites of layered Mn(OH)2 reacted with the surfactants, leading to the formation of a mesoporous phase (16). The crystalline layered phase of Mn(OH)2 was then mildly oxidized in air to form a mixed-valent manganese oxide shell (16). Finally, surfactant templates were removed by further oxidation during calcination, leading to the formation of a semiconducting mesoporous material (16). The control of the primary building block units and oxidation processes are the basis for developing our syntheses.

If the mixing step is done in air, some Mn2+ present in Mn(OH)2 can be readily oxidized to Mn3+ and Mn4+ (12, 14, 15). This type of oxidation process has been observed in many of the syntheses of manganese oxide OMS phases, where mixtures are white [Mn(OH)2] under N2 atmosphere but brown (MnO2, Mn2O3, or Mn3O4) under oxidizing atmospheres. The cylindrical micelles then react with two different building blocks [Mn(OH)6 and MnO6]. The oxidative atmosphere used during synthesis also provides a pathway to mixed valency and the possibility of forming edge-shared MnO6 octahedra.

The Mn(OH)2 is a white, layered crystalline material built from the edge-sharing of [Mn(OH)6]4−octahedral clusters (17). Oxidation of Mn(OH)2 can readily produce brown to dark-gray phases of manganese oxides (17). The negatively charged building blocks of [Mn(OH)6]4− of the Mn(OH)2 microcrystallites have a higher probability of binding to positively charged surfactant head groups than of binding to one another. This bias may limit the size of the Mn(OH)2crystallites. The Mn(OH)2 crystallites oxidize in air in the presence of surfactant much more quickly than they do in the absence of surfactant because the smaller Mn(OH)2 particles react with O2 faster and more completely. Microcrystallites of Mn(OH)2 react with surfactants to eventually form mesoporous phases by means of self-assembly, with the help of large, cylindrical micelle templates.

Our x-ray diffraction (XRD) patterns (18) of the resultant solids (Fig. 1) suggest that the primary building blocks are octahedral MnO6 units that link together to produce crystalline wall material. Similar XRD patterns were observed for these materials after heating to 1000°C.

Figure 1

(Curve A) XRD pattern of the uncalcined hexagonal MOMS-1. Note the strong (100) peak (d = 4.7 nm) and small (110) peak (d = 2.7 nm). (Curve B) XRD pattern of calcined hexagonal MOMS-1. The positions of the (100) and (110) peaks shifted to lower d spacings, and the peak intensities decreased. (Inset) The XRD patterns of calcined samples show two additional broad peaks (d = 0.50 and 0.30 nm), which are Mn2O3 (gamma phase) and Mn3O4 (hausmannite) microcrystallite phases.

Calcined samples prepared with a CTAB concentration of 28 weight % have arrays of pores of hexagonal shape (Fig. 2). This hexagonal symmetry is believed to be due to the organization of micelles, not the geometry of the primary building blocks of manganese oxide. The high degree of ordering is in direct contrast to reports of M41S mesopore systems (1, 3) and OMS systems (12, 14,15). Electron paramagnetic resonance (EPR) data (18) suggest that the primary building blocks of the wall materials are octahedral MnO6 units.

Figure 2

(A) Lattice morphology of the calcined hexagonal MOMS-1 (CTAB concentration = 28%) shown by HRTEM. Crystallites with an average particle size of 200 Å are observed with no evidence of any other phase. (B) Convergent beam electron diffraction (CBED) pattern of the calcined hexagonal MOMS-1 showing hexagonal symmetry. The nonideal 120° angles for the MnO6 octahedral building blocks suggest some stacking stress in these systems. Ordering of MOMS-1 in two dimensions shows nine orders of observed reflections. Stacking faults and twinning, observed in some OMS systems, were not observed in MOMS-1 or MOMS-2.

Argon sorption data (19) (Fig. 3) from Brunauer, Emmett, and Teller (BET) measurements and pore-size distribution data show that mesopores exist in these materials, and XRD, transmission electron microscopy (TEM), and EPR data show that these are hexagonal MOMS materials, hereafter called MOMS-1. Cubic-phase XRD patterns were observed for materials prepared with a CTAB concentration of 10 weight %, with two peaks for the as-prepared material but seven peaks after calcination at 600°C (18). The surface area for this calcined cubic sample is 46 m2g−1. This cubic phase will hereafter be referred to as MOMS-2.

Figure 3

(A) Low–relative pressure Ar sorption isotherm for the calcined hexagonal MOMS-1;V is the volume adsorbed per gram of MOMS-1. Saturation is reached at low relative pressure (P/P 0 ∼ 10−5). Relatively low surface areas (170 m2 g−1) are related to the dense, thick wall structure and are similar to microporous crystalline OMS phases (12, 14, 15) and mesoporous manganese nodules (27). Classical surface areas may not be truly representative for materials with transition metals like MOMS: A density-normalized surface area may be more appropriate. The density-normalized surface area of MOMS-1 is 886 m2cm−3 and that of MOMS-2 is 250 m2cm−3. (B) N2 sorption hysteresis for the calcined hexagonal MOMS-1. A clear hysteresis loop for adsorption-desorption is observed. (C) Pore size (radiusr) distribution for the calcined hexagonal MOMS-1. The broad peak at about 30 Å is indicative of mesopores.

The wall thickness of the mesopores, about 1.7 nm, was deduced from the interlattice d spacing of the (100) reflection (4.7 nm in Fig. 1A) and the pore diameter (3.0 nm) determined for the hexagonal MOMS-1 material. These data suggest that MOMS-1 consists of thick walls. The walls of MOMS-1 are also more dense than the walls of M41S materials, because Mn atoms are heavier than Si and Al atoms and the edge-shared MnO6 octahedra are more tightly packed than the more flexible vertex-shared SiO4 and AlO4tetrahedra. Subsequent oxidation of the manganese oxide wall materials is necessary for formation of edge-shared structures (17). Too much reduction to Mn2+ or to Mn3+ leads to reduced nonporous materials. Corner-shared tetrahedra are expected to be more flexible than edge-shared octahedra. A large number of relatively rigid edge-shared MnO6 octahedra may be needed to form a hexagonal MOMS-1 array. Similar thick, dense walls are proposed for the cubic MOMS-2 materials on the basis of corresponding XRD, high-resolution TEM (HRTEM), and adsorption data.

The shapes of the hysteresis loop in the adsorption-desorption isotherms of the calcined samples (Fig. 3B) are indicative of mesopores that are poorly defined (20) and may further support the thick and dense wall structure for MOMS-1. The Ar sorption data at low relative pressure (P/P 0 ∼ 10−5) (Fig. 3A) suggest that no micropores are present in these systems. The adsorption data further support data from XRD and HRTEM experiments that suggest that neither crystalline nor amorphous (15) microporous impurity phases are present in MOMS-1 or MOMS-2.

The average oxidation state (AOS) of MOMS-1 (Table 1) changes from a majority of 3+ before oxidation to a majority of 4+ after calcination. Changes in AOS are related to the ability of MnO6 units to form mixed-valent microporous (12, 14, 15) and mesoporous systems by means of edge and vertex sharing.

Table 1

Ratios of Mn2+:Mn3+:Mn4+, average oxidation states, conductivities, and band gaps for uncalcined and calcined hexagonal MOMS-1.

View this table:

The conductivities (22) of MOMS-1 before and after calcination are 5.0 × 10−8 and 8.1 × 10−6 (ohm·cm)−1, respectively. Diffuse-reflectance ultraviolet-visible (DRUV-Vis) data (21) of the as-prepared MOMS-1 sample show an absorption at 390 nm, corresponding to a band gap (23) of 3.18 eV. The calcined MOMS-1 sample shows a DRUV-Vis peak at 505 nm, corresponding to a band gap of 2.46 eV. The conductivity of the calcined MOMS-2 cubic phase is 2.28 × 10−6 (ohm·cm)−1. Conductivities of the calcined MOMS-1 and MOMS-2 samples show temperature-dependent semiconducting behavior. Fermi level shifts (23) due to quantum size effects for Mn2O3 and Mn3O4microcrystallites and the mesopore size of ∼3.0 nm may have some effect on the conductivity. The conductivity of microporous todorokite-type OMS-1 crystalline phases (14) with similar AOS values of 3.5 (MOMS-1 = 3.55) is about 2 × 10−7 (ohm·cm)−1 [MOMS-1 = 8.13 × 10−6 (ohm·cm)−1], and pure dense-phase Mn2O3 has a conductivity of 5.0 × 10−8 (ohm·cm)−1. Calcined MOMS-1 thus has a higher conductivity than synthetic todorokite, although both have similar AOS values, suggesting that special structural features of MOMS-1 may play an important role in the electrical conduction.

Oxidation with air converts the Mn2+ in the surfactant Mn(OH)2 material into Mn3+ and Mn4+as it does in microporous OMS materials. The Mn(OH)2 layers closest to the micelle may not be reached by O2 because of diffusion limitations of O2 through unusually thick and dense walls in the initial mild oxidation step. In the following oxidation step, the structure may be stabilized by strong bonding of the negatively charged innermost Mn(OH)2 layers to the positively charged surfaces of surfactant micelles. The structure may be further stabilized during calcination by the formation of hexagonal arrays of thermally stable outer MnO2layers. The color change from brown (Mn3+) before calcination to almost black (Mn4+) after calcination suggests that the removal of surfactant during calcination was accompanied by further oxidation of wall material. Such color changes are in agreement with our proposal that the primary structural units are MnO6 units and that crystalline dense-phase microcrystallites of manganese oxides in the walls are the secondary structural units of the MOMS systems. This two-step oxidation mechanism is supported by several observations, including changes in the color, AOS, conductivity, and band gap of the samples after calcination.

Calcined samples heated to 1000°C show a 4% weight loss, as shown in thermogravimetric analysis (TGA) data (19). This small weight loss may be due to the evolution of leftover surfactant or hydroxyl groups. Differential scanning calorimetry (DSC) data (19) do not show any clear phase changes for the heated calcined sample. The DSC and TGA data suggest that surfactant molecules are not present in the calcined samples because melting of the surfactant phase was not observed. These thermal data suggest that the calcined hexagonal MOMS-1 structure is thermally stable to 1000°C, which is also in agreement with XRD studies. The cubic MOMS-2 materials show similar TGA and DSC properties with thermally stable crystalline phases being observed up to 1000°C.

In microporous OMSs consisting of primary MnO6building blocks, the thermal stability decreases as pore size increases. The high thermal stability of these MOMS phases is unusual for manganese oxide systems. Fourier transform infrared (FTIR) data (19) for pyridine adsorbed on calcined MOMS-1 (Fig.4) suggest that there are two types of acid sites that co-exist on the surface of calcined MOMS-1: The surface is dominated by Lewis acid sites, which is usual for manganese oxide systems, as most crystalline microporous OMS materials have an overwhelming majority of Lewis acid sites. There is little information on the acidity of M41S materials (24).

Figure 4

FTIR spectrum for the pyridine on calcined hexagonal MOMS-1 showing an intense absorption at 1460 cm−1 and a considerably weaker absorption at 1540 cm−1. The peak at 1460 cm−1 is due to Lewis acid sites and that at 1540 cm−1 is due to a much smaller number of Brönsted acid sites.

However, temperature-programmed desorption (TPD) data for NH3 adsorbed on calcined MCM-41 systems (24) and similar data for MOMS-1 show significant differences in the acidity of these two classes of materials. In the MCM-41 systems, the NH3 TPD data mimicked behavior for amorphous silica aluminas with a relatively broad desorption. The NH3 TPD data for MOMS-1 show three distinct peaks: at 215°C, between 300° and 310°C, and at 490°C. Such behavior is more typical for crystalline systems like zeolites and is in line with XRD data and the dense, thick structure of the crystalline walls of MOMS.

Data for the catalytic oxidation of cyclohexane (25) were obtained for calcined MOMS-1 (Table 2). The catalytic oxidation of stable alkanes to more valuable products like alkyl alcohols and ketones under mild reaction conditions is known to be catalyzed by mixed-valent microporous manganese oxide OMSs (26). Reactions of hydrogen peroxide with alkanes in the presence of manganese oxide OMS materials have been studied (26), and the role of lattice oxygen in catalytic oxidations has also been discussed (14, 15). These catalytic data suggest that MOMS-1 materials can function as active and selective oxidation catalysts and that total oxidation to CO2 can be avoided.

Table 2

Yields for the oxidation of cyclohexane and hexane over calcined hexagonal MOMS-1. The conversions for cyclohexane andn-hexane were 10.2 and 7.65%, respectively. No products were observed under similar reaction conditions when MOMS-1 catalyst was absent. Dashes indicate that the product was not observed.

View this table:

Results of NH3 TPD suggest that these reactions are driven by redox reactions, as previously suggested (12, 14, 15, 26) for microporous OMS systems. The catalytic data of Table 2 are also in line with suggestions of shape-selective redox catalysis at internal sites, because the smallest pore material, OMS-2, which has square pores (4.6 Å on a side, 6.5 Å in diameter), is most active and selective in these oxidations (26). The activities and selectivities of large-pore OMS-1 (6.9 Å on an edge, 9.7 Å in diameter) are similar to those for the hexagonal MOMS-1 materials.

The total conversion and yields for n-hexane over MOMS-1 (Table 2) are promising, indicating that the total conversion of hexane and the less-stable cyclohexane are similar. The yields of both 2- and 3-hexanol are about one order of magnitude higher than the yield of 1-hexanol, probably because of thermodynamic constraints.

The catalytic oxidation of alkanes shown here suggests that the mesoporous hexagonal phase is active because of a redox mechanism that is a direct result of the mixed valency and unusually high thermal stability of MOMS materials. Future work should be aimed at understanding the interactions between surfactant molecules and Mn(OH)2 materials, the nature of bonding in such systems, the effects of doping the wall phases, the local electronic environments, and other kinds of applications.

REFERENCES AND NOTES

View Abstract

Stay Connected to Science

Navigate This Article