pubs.acs.org/NanoLett
Size Dependence in Hexagonal Mesoporous Germanium: Pore Wall Thickness versus Energy Gap and Photoluminescence Gerasimos S. Armatas,†,§ and Mercouri G. Kanatzidis*,†,‡ †
Department of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208-3113, Materials Science Division, Argonne National Laboratory, Chicago, Illinois 60439, and § Department of Materials Science and Technology, University of Crete, 71003 Heraklion, Crete, Greece ‡
ABSTRACT A series of hexagonal mesoporous germanium semiconductors with tunable wall thickness is reported. These nanostructures possess uniform pores of 3.1-3.2 nm, wall thicknesses from 1.3 to 2.2 nm, and large internal BET surface area in the range of 404-451 m2/g. The porous Ge framework of these materials is assembled from the templated oxidative self-polymerization of (Ge9)4- Zintl clusters. Total X-ray scattering analysis supports a model of interconnected deltahedral (Ge9)-cluster forming the framework and X-ray photoelectron spectroscopy indicates nearly zero-valence Ge atoms. We show the controllable tuning of the pore wall thickness and its impact on the energy band gap which increases systematically with diminishing wall thickness. Furthermore, there is room temperature photoluminescence emission which shifts correspondingly from 672 to 640 nm. The emission signal can be quenched via energy transfer with organic molecules such as pyridine diffusing into the pores. KEYWORDS Mesoporous, zintl compounds, nanoporous, quantum confinement
S
zation of linear polymer (1/∞){(Ge9)2-} chains around of surfactant micelles. By adopting a different synthetic procedure, we demonstrated mesoporous germanium networks through the surfactant-templated self-polymerization of (Ge9)4- clusters.8 That involves the slow oxidative coupling of the clusters to form the framework. Initially, these materials came with a fixed wall thickness of ∼2 nm, but now we succeeded in creating systems with variable wall thickness. It would be interesting to study a series of such porous semiconductors where the tuning parameter is only the size of wall thickness. This could allow the investigation of whether quantum confinement effects can exist in regular nanoporous structures in the same fashion they exist in semiconductor nanocrystals. The nanocrystals possess a size-dependent optical band gap which is larger than that of the bulk system. By analogy the nanostructured frameworks could show a wall-thickness-dependent optical band gap as shown schematically in Scheme 1 that is larger than that of the bulk systems. In this context a systematic blue shift would be expected as the pore wall thickness decreases. Herein, we report a series of hexagonally ordered germanium mesostructures, which to our knowledge are the first examples of semiconductors with tunable pore wall thickness. The synthetic procedure involves surfactant templated self-polymerization of (Ge9)n2n- oligomers and permits control over the wall thickness of the mesoporous structure that can be effected with the reaction and annealing time. The resulting nanostructured frameworks exhibit strong sizedependent optical absorption and photoemission properties. As the wall thickness decreases, a continuous blue shift in optical absorption is observed.
emiconductors having one or more dimensions in the nanometer range (typically 1-100 nm) are commonly referred to as quantum wells, wires, or quantum dots and are of considerable scientific and technological interest. The ability to tune the physical characteristics in a controlled manner by adjusting only size and shape is a core feature of nanoscience, and it makes nanomaterials attractive for applications in light-emitting nanodevices, photocatalysis, energy conversion, and biological sensing.1 The sizedependent energy band gap and luminescence are fascinating features of nanostructures which can be explained by sizeinduced quantum confinement effects.2 For these reasons group IV nanostructures, such as Si and Ge, have been studied extensively in recent years with procedures that affect their chemical composition, size, and shape.3,4 Recently, we described nanostructured semiconducting frameworks perforated with regular arrays of open mesopores.5 Such nanostructures represent a substantial dimensional reduction of the semiconducting framework from the bulk to the nanoscale regime. In principle, this could induce new functionalities such as quantum-confined optical absorption and luminescence, analogous to those observed in discrete nanoparticles. Surfactant-directed assembly is the key to stabilizing mesoporous elemental germanium and binary intermetallic germanium-based mesostructures.6,7 The synthetic methods involve metal-mediated coupling of discrete Ge4- and (Ge9)4- Zintl ions or oxidative polymeri-
* To whom correspondence should be addressed,
[email protected]. Received for review: 03/22/2010 Published on Web: 08/10/2010 © 2010 American Chemical Society
3330
DOI: 10.1021/nl101004q | Nano Lett. 2010, 10, 3330–3336
SCHEME 1. Band Gap Evolution with Wall Thickness in a Nanostructured Semiconductor (VB ) valence band, CB ) conduction band)
EDMHEAB + {(Ge9)n}2n- + 2nNH2(CH2)2NH2 f (EDMHEA){(3/∞)(Ge9)} + 2nNH2(CH2)2NH- + nH2 NU-Ge-1(t)
(2)
where, N-eicosane-N,N-dimethyl-N-(2-hydroxyethyl)ammonium bromide (EDMHEAB) is the surfactant. The mesoporous NU-Ge-1(t) semiconductors were obtained after removal of the surfactant molecules from the framework through ion-exchange with ammonium nitrate in ethanol followed by mild heating at 80 °C under vacuum to remove NH3. This process produces a stable all-Ge mesoporous framework with H-passivated surface.8 Energy dispersive X-ray spectroscopy (EDS) acquired on transmission electron microscopy (TEM) confirmed that the framework was composed only of germanium. The EDS analysis showed the presence of Ge element without detection of any traces of phosphorus, potassium, or halides (Figure S1 in Supporting Information). The absence of phosphorus suggests that species such as diphenyl phosphide (Ph2P-) anions are not incorporated into the products. The chemical composition of “as-prepared” (containing surfactant) and template-free mesoporous NU-Ge-1(t) materials was determined by elemental C, H, and N and thermogravimetric (TGA) analysis. The analytical results of as-prepared materials reveal ∼29-37% of surfactant content incorporated between the framework. After templateremoval only ∼6-7% of surfactant was estimated to remain (Table S1 in Supporting Information). The mesoscopic order of the NU-Ge-1(t) materials was probed by small-angle X-ray scattering analysis and TEM images. The SAXS patterns of mesoporous samples show a strong peak at scattering wave vector q ()4π sin θ/λ, where
We utilized surfactant-templated polymerization of partially coupled deltahedral (Ge9)4- cluster oligomers to construct hexagonal mesoporous Ge with adjustable size of pore wall width. Previous studies of (Ge9)4- anions have shown that they are susceptible to oligomerization/polymerization, producing {(Ge9)n}2n- oligomers, linear polymer (1/∞){(Ge9)2-} chains9 and even three-dimensional crystalline clathrate superstructures.10 The oxidative coupling of (Ge9)4- clusters includes the formation of dimer {Ge9-Ge9}4-,11 trimer {Ge9-Ge9-Ge9}6,12 or tetramer {Ge9-Ge9-Ge9-Ge9}8-13 anions. Herein, taking advance of this chemistry we prepared different oligomeric {(Ge9)n}2n- species as secondary building blocks to assemble a series of mesoporous all-germanium frameworks. To exercise kinetic control on the oligomerization reaction, we used triphenylphosphine (Ph3P) as an electron acceptor in ethylenediamine solution because it is a mild oxidant. The strongly reducing character of (Ge9)4clusters causes it to oxidize slowly to oligomers by producing diphenyl phosphide and phenyl ions as byproducts (eq 1).13
n(Ge9)4- + nPh3P f {(Ge9)n}2n- + n(Ph2P)- + nPh(1)
Adjusting the reaction time t between the (Ge9)4- precursor and Ph3P allows control over the mesoporous wall thickness. The gradual formation of {(Ge9)n}2n- oligomeric species was indicated by the color change of the solution with time from deep-red, to brown, and to dark-green.13 The templated polymerization of {(Ge9)n}2n- building units proceeds via a surfactant-assisted route and seems to be accompanied with a two-electron reduction of the ethylenediamine solvent according to eq 2.12a © 2010 American Chemical Society
FIGURE 1. Small-angle X-ray scattering patterns of mesoporous (a) NU-Ge-1(1), (b) NU-Ge-1(3), (c) NU-Ge-1(6), and (d) NU-Ge-1(12) materials (surfactant molecules removed). 3331
DOI: 10.1021/nl101004q | Nano Lett. 2010, 10, 3330-–3336
FIGURE 3. Nitrogen adsorption-desorption isotherms at 77 K of mesoporous (a) NU-Ge-1(1), (b) NU-Ge-1(3), (c) NU-Ge-1(6), and (d) NU-Ge-1(12). The NLDFT pore size distributions calculated from the adsorption branches are shown in the insets.
exhibit typical type IV curves accompanied with a distinct condensation step at relative pressure (P/P0) ∼ 0.16, suggesting mesoporous structures with uniform pores. The lack of hysteresis loop between the adsorption and desorption branches indicates negligible pore blocking (percolation) effects and corroborates to the tubular pore structure. The sharp increase in the adsorption branch at high relative pressures (P/P0 > 0.9) is ascribed to the capillary condensation of nitrogen in large voids between the particles. The mesoporous NU-Ge-1(t) samples exhibited BrunauerEmmett-Teller (BET) surface area as high as 404-451 m2 g-1 and total pore volume of 0.27-0.35 cm3 g-1. Such surface areas are very large given the heavy elements that compose the framework and correspond to 624-697 m2/g silica and 1057-1180 m2/g COF equivalent surface area.14 We also conducted small-angle X-ray scattering measurements to estimate the total surface area and pore size of mesoporous NU-Ge-1(t) materials.9,15 The X-ray scattering from mesoporous structures is sensitive to electron density fluctuations that primarily exist at the interface between the inorganic framework and the pores. Therefore, SAXS analysis on NU-Ge-1(t) samples can probe the total surface area, regardless of whether the pores are empty or filled with surfactant. The good agreement between the BET surface areas obtained from nitrogen physisorption measurements and those from SAXS analysis indicates that the porous
FIGURE 2. Typical TEM images of mesoporous (a) and (b) NU-Ge1(1) and (c) and (d) NU-Ge-1(12) materials viewed along the pore channel axis (a and c) and perpendicular to the pore channel axis (b and d). The corresponding fast Fourier transform images are shown in the insets.
2θ is the scattering angle) range of 1.44-1.60 nm-1, which is associated with interplanar distance (d ) 2π/q) from 4.37 to 3.92 nm, respectively (Figure 1). In the wide-angle range (>10°), the absence of Bragg reflections indicates the nonperiodic nature of the germanium frameworks. Figure 2 depicts typical TEM images and the corresponding fast Fourier transform (FFT) patterns obtained from mesoporous NU-Ge-1(1) and NU-Ge-1(12) samples. Direct observation of the structure with TEM reveals parallel mesoporous channels that are hexagonally arranged according to 2-D hexagonal p6mm symmetry. On the basis of this symmetry and the SAXS interplanar (d100) distances, we obtain a lattice parameter (a0) from 4.53 to 5.05 nm in a series of mesoporous NUGe-1(t) samples (Table 1). Since the mesoporous structures are obtained as inorganic replicas of organized assemblies from the same surfactant molecules, the systematic increase of the lattice parameters as a function of reaction time directly reflects the corresponding increase in the thickness of the framework walls. Figure 3, shows the nitrogen adsorption-desorption isotherms of the various NU-Ge-1(t) samples. The isotherms
TABLE 1. Surface Area, Pore Volume, Mesopore Size, Unit Cell Size, and Framework Wall Thickness of Mesoporous NU-Ge-1(t) material
SBET(SSAXS) (m2/g)
pore volumea (cm3/g)
DNLDFT (DSAXS) (nm)
a0 (a0)b (nm)
WTN2 (WTSAXS)c (nm)
NU-Ge-1(1) NU-Ge-1(3) NU-Ge-1(6) NU-Ge-1(12)
436 (439) 406 (442) 451 (453) 404 (460)
0.31 0.27 0.35 0.31
3.2 (3.13) 3.2 (3.10) 3.1 (3.03) 3.1 (2.98)
4.53 (4.84) 4.71 (5.15) 4.97 (5.07) 5.05 (5.11)
1.33 (1.40) 1.51 (1.61) 1.87 (1.94) 1.95 (2.07)
a Cumulative pore volume at relative pressure (P/P0) equal to 0.95. b Unit cell size a0 ) 2/3 d100, where d100 is the (100) interplanar distance. In parentheses the unit cell size of as-prepared containing surfactant materials is given. c Framework wall thickness WT ) a0 - D, where D is the diameter of the mesopores.
© 2010 American Chemical Society
3332
DOI: 10.1021/nl101004q | Nano Lett. 2010, 10, 3330-–3336
FIGURE 5. (a) Pair distribution function plots of (a) templateremoved mesoporous NU-Ge-1(1) (black), NU-Ge-1(3) (red), NU-Ge1(6) (green), and NU-Ge-1(12) (blue line) materials. (b) Calculated PDF plots from {-(Ge9)-}n models (black, n ) 1; red, n ) 2; green, n ) 3; blue line, n ) 4) and (c) experimental PDF of polycrystalline Ge.
atomic vectors at ∼2.5, ∼3.9, and ∼6.0 Å, which correspond to Ge-Ge first, second, and third distances, respectively, within the (Ge9) clusters. The 6.0 Å vector arises from intercluster correlations. It can be observed that the PDF plots for NU-Ge-1(t) are similar to the theoretical PDFs obtained from {-(Ge9)-}n (n ) 1, 2, 3, and 4) oligomer models and very different from that of polycrystalline bulk Ge. Notably, the PDF plots of NU-Ge-1(t) reveal a local structure that is defined essentially by two interconnected (Ge9) clusters since they lack interatomic vectors that correspond to Ge-to-Ge distances of three or more interconnected (Ge9) clusters. This could be due to the small distortions in the deltahedral (Ge9) structure caused by polymerization which degrades the long-range atomic pair correlation interactions. A detailed look at the pair correlation peak of Ge···Ge second neighbors of mesoporous samples from NU-Ge-1(1) to NU-Ge-1(12) reveals a continuous shift from 3.95 to 3.88 Å. Such a shift is in line with the observation that the Ge···Ge second neighbor correlation peak of discrete (Ge9) clusters appears at ∼4.14 Å, while peaks of tetramer {-(Ge9)-}4 units occur slightly lower at ∼3.92 Å. The NU-Ge-1(t) series of materials exhibit sharp optical absorption onsets in the visible region which is considerably higher than the band gap absorption of bulk Ge (0.66 eV) and amorphous Ge (0.88 eV)18 (Figure 6a). This large blue shift by over 1 eV is attributed to the nanosize wall thickness of the Ge frameworks which does not allow for broad electronic bands to develop. This nanoscale reduction and inability to develop broad bulklike electronic bands is the reason for the well-known blue shifts observed in conventional nanocrystals. The energy band gap of NU-Ge-1(t) varies systematically with the wall thickness from 1.97 eV for the ∼1.3 nm wall to 1.86 eV for the ∼2 nm wall sample (Table 2). Figure 6b, shows the energy gap of NU-Ge-1(t) as a function of wall thickness. It appears that the energy band gap systematically increases with decreasing the size of framework wall thickness. The wall thickness, in Figure 6b, was determined from nitrogen physisorption measurements
FIGURE 4. High-resolution Ge 3d core-level photoelectron spectra of mesoporous (a and b) NU-Ge-1(6) and (c and d) NU-Ge-1(12) samples. In (b) and (d), the corresponding Ge 3d photoelectron spectra after surface sputtering with low-energy Ar+ ion beam.
structures are fully accessible on adsorption and that the mesoporous frameworks possess limited microporosity. Analysis of the adsorption branches by using the nonlocal density function theory (NLDFT)16 reveals quite narrow pore size distributions with maximum of the peaks centering at ∼3.1-3.2 nm. These pore sizes are in very good agreement with those obtained by the independent SAXS analysis (see Table 1). The pore-to-pore distance (ao) calculated from SAXS interplanar (d100) distances together with the diameter of mesopores from NLDFT and SAXS analysis gives an estimate for the framework wall thickness from ∼1.3 to ∼2 nm, which is strongly correlated with reaction time t (see Table 1). The oxidation state of germanium in the mesoporous frameworks was investigated with X-ray photoelectron spectroscopy (XPS). Because of the large number of surface Ge-Ge bonds and their high sensitivity to oxidation (i.e., by formation of Ge-O-Ge bonds), sample exposure to air was minimized during handling. The XPS spectra of mesoporous NU-Ge-1(6) and NU-Ge-1(12) showed only one peak for the Ge 3d core-level at 29.4 and 29.3 eV, respectively (Figure 4). These binding energies imply a framework structure predominantly made with nearly zerovalent Ge atoms. The XPS peaks shifted very slightly to lower binding energy (29.1 eV) after sputtering the samples with a low-energy Ar+ ions beam, suggesting only minimal surface oxidation of the particles. Clear evidence for the presence of interconnected deltahedral (Ge9) clusters into the framework is obtained upon investigation of the local structure of NU-Ge-1(t) with X-ray diffuse scattering and pair distribution function (PDF) analysis.17 Figure 5 depicts the PDF plots as a function of the interatomic distance for NU-Ge-1(t). The PDF plots demonstrate a well-defined local structure reflected in the inter© 2010 American Chemical Society
3333
DOI: 10.1021/nl101004q | Nano Lett. 2010, 10, 3330-–3336
while the other directions (along the pore axis and the framework) are much longer (bulklike), the observed blue shift in these mesostructures is consistent with that in 3.3 nm size Ge nanoparticles. The possible connection to quantum confinement comes from the fact that semiconducting porous structures, with uniform wall thickness and pore size such as the materials described here, are the “negative image” (an inside out version) of an array of regularly spaced quantum dots; i.e., they represent a system of “antidots”.22 These types of materials have been envisioned and have been termed “exosemiconductors” or “quantum antidots”.23 The relationship between these two types of nanostructures (i.e., semiconductors with holes and arrays of nanocrystals) is fascinating and has yet to be explored. The NU-Ge-1(t) materials are photoluminescence (PL) at room temperature under the excitation wavelength of 370 nm, Figure 7a. Since the surfactant molecules are not photoluminescent, the PL emission peak in NU-Ge-1(t) apparently originated from the semiconducting germanide framework. The peak of the PL spectra is located in the visible region slightly below the band-edge absorption, suggesting a predominantly band-edge emission process, although emission from surface defects or trapping states cannot be ruled out. The fact that the PL spectra exhibit a definitive red shift of the peak maxima from 640 to 672 nm with increasing wall thickness from ∼1.3 to ∼2 nm is consistent with band-edge emission (see Table 2). Figure 7b shows the systematic variation of the emission as well as the energy band gap with wall thickness. The systematic red shift in the PL emission is reminiscent of the size-dependent PL emission observed from Ge nanocrystals.24 Is the observed emission from the Ge framework or from the surface? This question is perhaps moot since in porous materials with wall thicknesses of
