Phase Transition Behavior and Oriented Aggregation During

Article pubs.acs.org/JPCC

Phase Transition Behavior and Oriented Aggregation During Precipitation of In(OH)3 and InOOH Nanocrystals Martin Klaumünzer,† Mirza Mačković,‡ Pascal Ferstl,§ Michael Voigt,*,† Erdmann Spiecker,‡ Bernd Meyer,§ and Wolfgang Peukert† †

Institute of Particle Technology, Friedrich-Alexander-University Erlangen-Nürnberg, Cauerstrasse 4, 91058 Erlangen, Germany Center for Nanoanalysis and Electron Microscopy (CENEM), Department of Materials Science and Engineering (WW7), Friedrich-Alexander-University Erlangen-Nürnberg, Cauerstrasse 6, 91058 Erlangen, Germany § Interdisciplinary Center for Molecular Materials (ICMM) and Computer-Chemistry-Center (CCC), Friedrich-Alexander-University Erlangen-Nürnberg, Nägelsbachstrasse 25, 91052 Erlangen, Germany ‡

S Supporting Information *

ABSTRACT: The phase transition behavior and oriented aggregation (OA) during colloidal synthesis of In(OH)3 nanocrystals in water are investigated by TEM, SEM, Xray diffraction, and density functional theory (DFT) calculations. Besides the cubic In(OH)3 phase, also orthorhombic InOOH is formed in a precipitation route using indium acetate as the In3+ source. Well-developed nano- and microcuboids are observed that consist solely of In(OH)3. In contrast, the InOOH phase remains semicrystalline even for long reaction (refluxing) times. The irregular growth of the InOOH phase is explained by proton transfers from hydroxyl groups to oxygen ions within the InOOH lattice that lead to OH disorder and lattice strain. DFT calculations of the surface energies of ideal and water-saturated low-index InOOH and In(OH)3 surfaces predict that the In(OH)3 phase becomes energetically more favorable than InOOH above a critical crystallite size. This explains why InOOH is formed before the In(OH)3 phase, which is an unusual pathway for a hydrothermal process. Once InOOH has transformed to In(OH)3 by incorporation of water, the crystallites can grow without restriction due to the disappearance of OH-disorder-induced strain. Finally, for the In(OH)3 cuboids a three-step formation process is suggested: In the first step, one-dimensional OA under formation of nanorods occurs. In the second step, parallel bundles are formed from the nanorods. In the third step these bundles merge into cuboids by three-dimensional OA.

■

INTRODUCTION In the past few years several attempts were made to synthesize indium hydroxide In(OH)3 nanocrystals with tailored size and shape. Nanocubes, nanospheres, nanorods, nanotubes, multipods, nanosheets, and nanoflowers of this material were obtained by liquid phase colloidal synthesis.1−14 In(OH)3 can be transformed to In2O3 by thermal decomposition at temperatures above 200 °C.8,14−17 In(OH)3 nanocubes were observed to form by oriented aggregation (OA) of small In(OH)3 nanocrystals.3,14 OA is defined as aggregation of nanocrystals along specific crystal directions and was formerly investigated by several research groups.18−20 A detailed overview about the mechanisms of OA is reviewed by Niederberger et al.19 OA can result in the formation of structural defects such as twin boundaries or edge dislocations.18,21−23 Despite previous studies on the synthesis of In(OH)3 nanoand microcrystals their exact formation mechanism is still under debate. Besides the cubic In(OH)3 phase, the formation of a second phase is reported for different precipitation routes, which was identified as orthorhombic indium oxide hydroxide InOOH.11,12 As reported by Schlicker et al.,24 orthorhombic InOOH is metastable, can be only stabilized in nanoscaled systems, and does not form during the transformation from © 2012 American Chemical Society

cubic In(OH)3 to bixbyite-type In2O3. Surprisingly, for precipitation in water the InOOH phase appears to be formed before the In(OH)3 phase as reported by Lin et al.12 As also stated by Lin et al.,12 the transition from InOOH to In(OH)3 is not the common phase transition pathway for a hydrothermal process. Lin et al.12 speculate that the presence of urea as OH− source in their reaction is responsible for this unusual phase behavior. However, the classical pathway from In(OH)3 to InOOH is also discussed in the literature.25 Within the present work we give an explanation for the unusual phase transition behavior (see above) during colloidal synthesis of In(OH)3 nanocrystals in water. It is also explained why the InOOH phase remains semicrystalline even for long refluxing times. In addition, the role of OA in the formation of the In(OH)3 nanocrystals is discussed by a three-step growth model. Our results on the formation mechanism of In(OH)3 cuboids are compared to those published recently by Jean et al.14 (see the Discussion). In addition to Jean et al.,14 we observe that InOOH is formed as a second phase besides In(OH)3 and we give an explanation for the unusual phase Received: June 14, 2012 Revised: October 8, 2012 Published: November 12, 2012 24529

dx.doi.org/10.1021/jp305858n | J. Phys. Chem. C 2012, 116, 24529−24537

The Journal of Physical Chemistry C

■

transition pathway from InOOH to In(OH)3. For this we performed density functional theory (DFT) calculations for the ideal and water-saturated low-index surfaces of InOOH and In(OH)3.

EXPERIMENT In(OH)3 samples were prepared by precipitation from In(Ac)3 in 99% purity with KOH (≥85% purity) in water (milli pore Qwater). All chemicals were purchased from Sigma Aldrich and used as received without further purification. For a typical synthesis the initial concentration of In(Ac)3 was 0.13 mol/L and KOH was 2.3 mol/L. Ten milliliters of the KOH solution was dropped into 40 mL of the indium acetate solution via a dropping funnel within less than 1 min followed by up to 24 h of refluxing. The refluxing time is further denoted as t. To remove remaining salts, after the synthesis all the samples were treated two times by the following procedure: centrifuging 5 min at 3000 rpm and exchanging the supernatant with the same amount of water. As characterization methods we used X-ray diffraction in Bragg−Brentano geometry (XRD), scanning electron microscopy (SEM), and transmission electron microscopy (TEM). SEM images were recorded with a field emission scanning electron microscope (ULTRA 55, Carl Zeiss NTS GmbH, Germany) at an operating voltage of 20 kV using an in-lens secondary electron detector. Conventional TEM imaging was performed using a Philips CM30 TWIN/STEM microscope. For high-resolution TEM (HRTEM) imaging a Philips CM300 UltraTWIN having a nominal point resolution of 1.7 Å and a Titan3 80−300 equipped with an image-side aberration corrector were used. All TEMs were operated at 300 kV acceleration voltage. TEM images were recorded with slow scan charge coupled device (CCD) cameras having an image size of 1024 × 1024 (CM30) and 2048 × 2048 pixels (CM300, Titan3). Electron diffraction patterns were evaluated using the software JEMS (version 3.5505U2010), incorporating the crystal data from the inorganic crystal structure database (ICSD). Possible beam damage during illumination with electrons was prevented by short exposure times of the samples to the electron beam, working at higher spot sizes (less beam intensity) and by spreading the beam. XRD measurements were performed with a Bruker AXS Advance D8 (Karlsruhe, Germany) X-ray diffractometer using Cu Kα radiation (λKα = 1.54 Å) at an acceleration voltage of 30 kV. The diffractometer was equipped with a onedimensional Vantec-1 high-speed detector and a rotating sample stage. The XRD coherence lengths Lhkl were calculated from the full width at half-maximum (FWHM) of the diffraction peaks by using the Scherrer equation assuming Gaussian shaped peaks:26

In(OH)3 → InOOH + H 2O

B − b2 cos Θhkl

(2)

we calculated a reaction energy of Er = +0.734 eV; i.e., at zero temperature and pressure In(OH)3 is thermodynamically more stable than InOOH plus gas phase water. All surface structures were represented by periodically repeated slabs with a thickness of six or seven indium layers (Figure 6), separated by a vacuum region of at least 12 Å. For the lateral extension of the slabs we used the theoretical bulk lattice parameters. Only surface structures with a periodicity of (1 × 1) primitive surface unit cells were considered in the present study. A k-point mesh with the same density as for the respective bulk calculation was used throughout. Water molecules were adsorbed symmetrically on both sides of the slabs. In the structural optimization all atoms were allowed to relax. Convergence was assumed when the largest residual force component acting on the atoms was less than 0.005 eV/Å. The energy required to cleave a crystal along a (hkl) lattice plane is given by

KλKα 2

THEORY

DFT calculations for ideal and water-saturated InOOH and In(OH)3 surface structures were carried out with the PWscf code of the Quantum Espresso software package.27 The PBE exchange−correlation functional of Perdew, Burke, and Ernzerhof28 was used throughout all calculations together with Vanderbilt-type ultrasoft pseudopotentials29 and a planewave representation of the wave functions with kinetic energy cutoff of 25 Ry. The idealized crystal structure of In(OH)3 is described by a body-centered cubic Bravais lattice and a primitive unit cell containing four formula units (space group symmetry Im3̅, ICSD 35636). However, in real crystals the OH groups show a high degree of orientational disorder.30 In the present calculations we used a conventional cubic unit cell with 56 atoms, which allows us to probe a distribution of OH orientations while maintaining the cubic symmetry of the Bravais lattice. k-point sampling was done with a (3,3,3) Monkhorst−Pack mesh. The energy differences between structures with different OH orientations turned out to be marginal. For the lowest-energy configuration we obtained a lattice constant of a = 8.052 Å with our calculational setup, which is +1.0% larger than the experimental value30 of a = 7.974 Å. InOOH has an orthorhombic crystal structure with space group symmetry Pmn21 containing two formula units per unit cell (ICSD 15081).31 Using a (4,4,6) Monkhorst−Pack k-point mesh, we obtained lattice parameters of a = 5.366 Å (+2.0%), b = 4.639 Å (+1.7%), and c = 3.330 Å (+1.8%). The deviations with respect to the experimental values31 of a = 5.26 Å, b = 4.56 Å, and c = 3.27 Å are given in parentheses. For the transformation of In(OH)3 into InOOH according to

■

Lhkl =

Article

(1)

γcleav =

B is the FWHM on the 2Θ scale in radiant, b is the instrumental broadening, K is the Scherrer constant, and Θhkl is the position of the intensity maximum of the corresponding diffraction peak. Elemental analysis of the samples for carbon was done using a LECO CS-200 by measuring CO. Elemental analysis of the samples for indium was done using ICP-OES. Before the elemental analysis the samples were dried carefully under vacuum at room temperature.

1 hkl (Eslab (0) − E bulk ) Ahkl

(3)

Ahkl is the geometric area of a (hkl) surface unit cell, Ehkl slab(0) is the total energy of a slab with (hkl) surface structure and no additional adsorbed water, and Ebulk is the bulk energy of the same number of formula units used in the slab calculation. As a measure for the interaction strength of water molecules with a (hkl) surface, we define the incremental water adsorption energy 24530



Article

Figure 1. SEM images recorded for samples taken after refluxing times of (a) 25 min, (b) + (c) 180 min, and (d) 1440 min. hkl hkl w Eadw, Nw = Eslab (Nw ) − Eslab (Nw − 1) − Emol

water layer. Only water molecules close to the surface, which experience some interaction with the surface atoms and have a binding energy significantly different from the average value in liquid water, contribute to the interface energy in eq 7. In the present work we make the approximation that only the water molecules in the very first adsorbate layer, which saturate the vacant coordination sites of the In surface atoms, have to be considered to evaluate the interface energy γint. Water molecules in the second and subsequent adsorbate layer are only bound via hydrogen bonds to OH groups of the surfaces and other water molecules beneath. Therefore, it can be assumed that their binding energy has already reached the average cohesion energy within liquid bulk water. The chemical potential Δμw of liquid water at room temperature can be deduced from thermochemical reference tables.33 With our choice of zero energy we obtain Δμw = −0.57 eV. This value is close to our calculated PBE binding energy of water molecules in ice: −0.61 eV (no zero point vibration correction included).

(4)

Ehkl slab(Nw) is the total energy of a slab with (hkl) surface structure and Nw adsorbed water molecules per (1 × 1) surface unit cell, and Ewmol is the energy of a gas phase water molecule. The interaction of water with the (hkl) surfaces is described by the equilibrium (surface)hkl + Nw H 2O ↔ (surf/water)hkl

(5)

The Gibbs free energy variation per surface unit area associated with this reaction at a given temperature T and pressure p is Δγ =

1 hkl hkl (Gslab (Nw ) − Gslab (0) − Nwμw (T ,p)) Ahkl

(6)

where μw(T,p) is the chemical potential of water.32 In the following we use the total energy of a gas phase water molecule as energy reference for the water chemical potential by introducing Δμw = μw + Ewmol. For the condensed phases we neglect the entropic and pV terms and replace the Gibbs free energies G by their respective total energies E from the DFT calculations. Within this approximation we introduce the water/solid interface energy γint and rewrite eq 6: γint = γcleav + Δγ = γcleav +

1 Ahkl

■

RESULTS Experiment. SEM images for samples taken from the aqueous dispersions after different refluxing times are shown in Figure 1a−d. As demonstrated in Figure 1a, for an early sample (t = 25 min) small nanocrystals with sizes below 30 nm are present where upon some of the nanocrystals exhibit a rod-like shape. For the later samples (t = 180 and 1440 min) nano- and microcuboids with a size of up to 250 nm occur besides a fraction of a smaller semicrystalline phase, as shown in Figure 1b−d. Figure 1c shows exemplarily a magnification of a cuboid for t = 180 min. The cuboid exhibits a surface with several linear defects (see arrows A) oriented parallel to its edges and vacancy clusters (see arrows B). Single rods are attached to the surface and are aligned parallel to the edges of the cuboid (see arrows C). This shows that OA of rods to the surface of the

Nw

∑ (Eadw,i − Δμw (T ,p)) i=1

(7)

In the limit of a thick water film approaches the average binding energy of a water molecule in liquid water. Thus, if we evaluate eq 7 for a chemical potenial Δμw that corresponds to liquid water, which is our reservoir with which we exchange molecules that are adsorbed/desorbed from the surface, the result for γint becomes independent of the thickness of the Ew,i ad

24531



Article

cuboids occurs. Further refluxing leads to cuboids with smooth and defect-less surfaces, as shown exemplarily in Figure 1d for t = 1440 min. This effect is related to surface ripening by a growth/dissolution process. TEM images of an early sample taken at t = 1 min show a mixture of rod-like nanocrystals and a small fraction of a semicrystalline phase (Figure 2). The diameter of the rod-like

Figure 2. TEM images and electron diffraction pattern of the sample taken at t = 1 min refluxing time, confirming the presence of rod-like structures and a semicrystalline phase. (a) Overview of the sample. (b) Electron diffraction pattern observed for a large sample region shown in (a). The asterisk indicates that the first diffraction ring can be indexed to the (220) reflection of the cubic In(OH)3 phase or the (101) and (011) reflections of the orthorhombic InOOH phase. (c) Higher magnification TEM image, showing rod-like structures and structures with irregular morphology. (d) HRTEM image, showing crystalline and semicrystalline regions present in the sample.

Figure 3. Bright-field TEM images and electron diffraction pattern of the sample taken at t = 4.5 min refluxing time, confirming the presence of In(OH)3 rod bundles behind a semicrystalline phase with irregular morphology. (a) Sample overview. (c) TEM image showing bundles of In(OH)3 rods. (b) and (d) SAED pattern for a region shown in (a) and (c), respectively. The asterisk indicates that the corresponding diffraction ring can be indexed to the (220) reflection of the cubic In(OH)3 phase or the (101) and (011) reflections of the orthorhombic InOOH phase. (e) Higher magnification TEM image showing rod bundles and structures with irregular morphology (see upper right section). (f) SAED pattern of the rod bundle marked with the circle in (e). The SAED pattern in (f) is depicted in correct relative orientation to the TEM image and contains a row of individual reflections of type (200) and (400) of In(OH)3. This confirms that the long axis of the In(OH)3 nanorods is parallel to the [100] direction.

nanocrystals is measured to be ≈5 nm. The selected area electron diffraction (SAED) pattern shown in Figure 2b is observed from a larger region of the sample as indicated in Figure 2a. Two intense diffraction rings can be identified in the SAED pattern in Figure 2b. The first diffraction ring is marked with an asterisk and can be referred to three different lattice planes: (a) (220) lattice planes of the cubic In(OH)3 phase, having a spacing of 2.82 Å, (b) (101) lattice planes of the orthorhombic InOOH phase, having a spacing of 2.78 Å, and (c) (011) lattice planes of the orthorhombic InOOH phase, having a spacing of 2.66 Å. The second diffraction ring can be referred to the (422) lattice planes of the cubic In(OH)3 phase solely and gives evidence for the presence of the cubic In(OH)3 phase. The presence of the orthorhombic InOOH phase, however, can be neither confirmed nor excluded from the SAED pattern because InOOH as well as In(OH)3 lattice planes may contribute to the first diffraction ring (see above). Therefore, additional XRD measurements are performed (Figure 5), which confirm the assignment of the SAED pattern to orthorhombic InOOH and cubic In(OH)3. Within the HRTEM image shown in Figure 2d crystalline and amorphous regions can be identified, showing the complexity of this system already in the early formation stages. TEM images for a later sample taken at t = 4.5 min are shown in Figure 3. Three-dimensional parallel bundles of In(OH)3 nanorods are observed, which consist of several tens

of nanorods and are up to ≈50 nm in size. A magnified region of the bundles is shown in Figure 3c. From Figure 3c it is obvious that the rod bundles partially merge into cuboids. The corresponding SAED pattern is shown in Figure 3d. The (200), (400), and (422) reflections clearly correspond to the cubic In(OH)3 phase and are attributed to the rod bundles. Merging of rod bundles into cuboids is indicated by the occurrence of diffraction dots for the (200) and (400) lattice planes. For the (220) and (422) lattice planes of In(OH)3 diffraction rings are observed, which is attributed to the random lateral orientation of the rod bundles. The first intense diffraction ring is again marked with an asterisk, with the meaning that, besides the (220) lattice planes of the cubic In(OH)3, also the (101) and (011) lattice planes of the orthorhombic InOOH phase are possible, as discussed above. Besides the bundles, also areas with a semicrystalline phase appear (Figure 3a,c,e). As confirmed by the corresponding SAED patterns and also by XRD (see below), the orthorhombic InOOH phase is still present at t = 4.5 min. An electron diffraction analysis (Figure 3f) was performed on a rod bundle marked with the circle in Figure 3e. It revealed the [100] direction as the main growth direction of the In(OH)3 nanorods. This agrees with results of 24532



Article

Zhuang et al.,34 who also found the [100] direction as the preferential growth direction of their In(OH)3 nanorods. Furthermore, the (200) and (400) reflections of In(OH)3 in Figure 3f show streaking, which is attributed to a slight misorientation and/or separation of the single rods inside the bundles. It is noteworthy that the dispersions appeared in a gellike state for t < 20 min and thereafter converted into a fluid state. This indicates that during the early stages of growth a 3D particle network is formed by agglomeration. TEM images for a late sample taken at t = 180 min are shown in Figure 4. A mixture of In(OH)3 nano- and microcuboids and

indicate the oriented agglomeration of single In(OH) 3 nanorods/bundles into a larger nano/microcuboid. A HRTEM image acquired in the edge region of the In(OH)3 cuboid (Figure 4a) is shown in Figure 4b. It reveals (220) lattice planes of cubic In(OH)3 with a spacing of 2.82 Å. The orientation of the lattice planes with respect to the edge of the cuboid agrees with the result obtained by electron diffraction and conventional TEM (Figure 4a). In some regions of the HRTEM image the crystal lattice appears disturbed (dotted circles in Figure 4b). These defects are not “real” but result from beam damage under HRTEM imaging conditions (strongly condensed beam), which leads to a gradual amorphization of In(OH)3. No such effects were observed under normal imaging conditions used for electron diffraction and conventional imaging (Figure 4a). In Figure 5 a series of X-ray diffraction patterns from samples taken at refluxing times from t = 1 min to t = 180 min is shown.

Figure 5. XRD patterns for samples taken after refluxing times of 1, 4.5, and 180 min. The reflections can be assigned to cubic In(OH)3 (α) and orthorhombic InOOH (β).

Figure 4. TEM images and electron diffraction pattern of a sample taken at t = 180 min. In (a) a In(OH)3 cuboid with (200) facets is observed together with a small fraction of the semicrystalline phase: (i) [001] zone axis SAED pattern of the In(OH)3 cuboid; (ii) SAED pattern observed in the region marked with the dotted circle in (a). The asterisk in (ii) indicates that the corresponding diffraction ring can be indexed to the (220) reflection of the cubic In(OH)3 phase and the (101) and (011) reflections of the orthorhombic InOOH phase. The dotted arrows indicate planar defects present in the In(OH)3 cuboid. In (b) a HRTEM image of the In(OH)3 cuboid is shown. The dotted circles indicate amorphized regions that form during illumination as a result of the strongly condensed beam in HRTEM mode.

For all the refluxing times the orthorhombic InOOH phase and the cubic In(OH)3 phase are identified by XRD, consistent with the TEM results. For the reflections associated to In(OH)3 the intensity increases and the FWHM decreases with refluxing time. This shows that the crystallinity of the In(OH)3 phase increases with refluxing time. In contrast, the reflections related to the InOOH phase remain broad and weak even at t = 180 min. The XRD coherence lengths for the lattice planes of InOOH remain below 5 nm for all refluxing times. This shows that the InOOH phase remains its semicrystalline nature even for long refluxing times. For the (200) reflection of In(OH)3 a coherence lengths of 37 nm is obtained. As expected for a high defect density, the coherence length is small as compared to the size of the cuboids as observed in the SEM and TEM images. The reflections associated to the InOOH phase (Figure 5) are shifted to low angles in comparison to the expected peak positions for bulk InOOH.31 This might be due to lattice expansion either by the insertion of water or by the semicrystalline nature of the InOOH phase in our samples. Furthermore, the proton transfer, as described below, might cause an expansion of the a-axis and the b-axis of the InOOH lattice. Besides the InOOH and In(OH)3 reflections, an additional not identified reflection is observed, as marked by # in Figure 5. This reflection appears broad and weak as in the

a small fraction of the semicrystalline phase containing In(OH)3 and InOOH is present at t = 180 min. The presence of cubic In(OH)3 and orthorhombic InOOH at this growth stage is confirmed by the SAED patterns i and ii in Figure 4 and also by XRD (see below). The SAED pattern shown in Figure 4i was observed for the big cuboid shown in Figure 4a and clearly corresponds to the cubic In(OH)3 phase, with the corresponding reflections valid for the [001] zone axis of cubic In(OH)3. The SAED pattern shown in Figure 4ii was observed in the region marked with the dotted circle shown in Figure 4a. This SAED pattern can be indexed for both phases, as already discussed above. Figure 4a shows a In(OH)3 cuboid with (200) facets, which contains several planar defects perpendicular to its edges, as marked by the dotted arrows. These planar defects 24533



Article

case of the InOOH reflections but can be attributed to neither the InOOH nor the In(OH)3 phase. The reflections associated to InOOH and the not identified reflection also appear in the X-ray diffraction patterns published in Lin et al.,12 for example, at exactly the same positions. In contrast to our study, Lin et al.12 comment neither on the shift of the InOOH reflections nor on the presence of the unidentified reflection nor on the semicrystalline nature of the InOOH phase. Elemental analysis for a sample taken at t = 180 min revealed an indium content of 58.8 mass % and a carbon content of 0.22 mass %. The indium content is below the expected value of larger than 69.2 mass % in the case of pure In(OH)3. This might be due to the insertion of water into the lattice of both phases, In(OH)3 and InOOH, leading to a nonstoichiometric composition. The small carbon content can be explained by acetate adsorbed on the surface of our particles. Because the carbon content is small, the presence of a significant amount of In(Ac)3 in our samples can be excluded. Dynamic light scattering on a sample taken at t = 180 min (not shown here) shows a fraction of clusters with a size of ≈3 nm besides a second fraction of larger particles. The cluster containing fraction is attributed to the semicrystalline InOOH phase. The second fraction consists of particles with sizes of ≈150 nm and is attributed to crystalline In(OH)3. To obtain In2O3, we performed temper experiments on our dried samples at moderate temperatures. As can be seen from Figure S1 (Supporting Information), exclusively reflections of cubic In2O3 appear in XRD of the calcinated sample. Because cubic In(OH)3 is known to convert to cubic In2O3 under calcination and because our samples mainly consist of cubic In(OH)3, our observations are in accord with refs 24 and 35. Theory. The bulk structure of InOOH can be described as a sequence of alternating ac-planes of O, In, and OH units stacked in the b-direction (Figure 6b). The OH groups form hydrogen bonds to the O atoms in the next layer. In our calculations we observed that the protons of the OH groups can be transferred easily with low activation barrier and small energy cost to the H-bond accepting O atom in the adjacent atomic plane. Therefore, significant disorder in the distribution of OH groups and O ions in the InOOH bulk phase has to be expected. Actually, if the proton of every second OH group along the a-axis is transferred to its neighboring O atom, the ideal InOOH bulk crystal structure is obtained again, but with the role of the a- and b-axis being reversed (Figure 6b). However, the a- and b-axis have slightly different lengths; i.e., the transfer of protons to the adjacent O ions will create a stress field within the bulk structure. So if a nanocrystal starts to grow with a random distribution and orientation of OH groups, stress will build up with increasing size of the nanoparticle. To reduce the stress, the nanocrystal has to transform, so that one direction becomes the a- and another direction becomes the baxis. Thus, an initial random distribution of OH groups represents a frustrated configuration, which might be one of the reasons why InOOH does not grow to larger crystalline nanoparticles, in contrast to In(OH)3, where the disorder of OH groups does not lead to significant geometrical distortions in the In lattice and to the formation of stress. Because InOOH is orthorhombic, the (100), (010), and (001) surfaces of InOOH crystals are not equivalent. The (100) termination is polar, whereas the (010) and (001) are nonpolar due to a mirror plane and a glide mirror plane symmetry of the crystal. The cleavage of the InOOH crystal reduces the 6-fold bulk coordination of the In atoms to 5-fold

Figure 6. Side view of the ideal (top) and water saturated (bottom) structures of the InOOH and In(OH)3 surfaces. In, O, and H atoms are depicted in gray, red, and blue, respectively. Golden ellipses mark positions of proton transfers in response of surface polarity.

at the (100) and (010) surfaces and to 4-fold at the (001) surface. These vacant coordination sites turn out to be the preferred adsorption sites for additional water molecules (see below). The (100) surface is polar, because in the ideal InOOH structure all OH groups are oriented in the same direction (Figure 6a). Polar surfaces are intrinsically unstable and have to reconstruct to suppress the macroscopic electrostatic dipole moment perpendicular to the surface. In the present case we found that the polarity of this surface is quenched by an appropriate number of proton transfers from OH groups in the uppermost surface layers to adjacent O atoms in the neighboring atomic plane (see golden circles in Figure 6a). This proton transfer happened spontaneously in the geometry optimization calculation, resulting in a nonmetallic surface structure with a band gap comparable to the bulk. This is an interesting new stabilization mechanism for polar surfaces that, to our knowledge, has not been reported previously for other polar surfaces. Elimination of an electrostatic dipole moment by transfer of protons might be a quite universal stabilization mechanism for polar surface terminations of oxyhydroxides. In InOOH, such polarity-driven proton transfers are an important intrinsic source for OH disorder that cannot be avoided. This mechanism is less likely to occur in hydroxides because two OH groups would have to decompose into an oxygen anion and a water molecule. However, a mechanism that only requires reorientation of OH groups is also conceivable. InOOH crystals can be cut at four different positions perpendicular to the [010]-direction (b-axis). The different cuts 24534



Article

(1 × 1) surface unit cell contains two 5-fold coordinated In. The two water molecules adsorbing at these sites become embedded in an H-bond network with the neighboring OH groups, which leads to a rather large adsorption energy. For all four surfaces the water adsorption energy is larger than the average water cohesion energy within liquid water. Thus, the interaction with water lowers the energy of the InOOH and In(OH)3 surfaces. In Table 1 we have listed the water/solid interface energies γint calculated according to eq 7 for a water chemical potential of Δμw = −0.57 eV, which is representative for liquid water at standard conditions. Interestingly, the water/solid interface energies for InOOH are slightly lower than for In(OH)3. Though the water adsorption energy is rather large for the In(OH)3(100) surface, the reduction in surface energy is slightly less pronounced than for the InOOH surfaces because of the much larger area of the surface unit cell and the associated lower density of undercoordinated In atoms in the surface layer that act as water adsorption sites. The total energy of a crystallite is given by the sum of energy gain by the formation of the bulk crystal and the energy cost to form the water/solid interface. The gain in volume energy scales with the number N of formula units, in contrast to the water/solid interface energy, which scales with N2/3. With our DFT calculations we found that the bulk phase of In(OH)3 is energetically more favorable than the bulk phase of InOOH, whereas for the water/solid interface energy it is vice versa. Thus, whether In(OH)3 or InOOH nanoparticles are thermodynamically more stable in aqueous solution depends on the bulk to surface ratio. In the present case we have the interesting situation that below a critical size InOOH nanoparticles are more stable because of the lower water/ solid interface energy at small crystal volume. Above the critical size, however, In(OH)3 particles become more favorable due to the larger gain in volume energy.

are either terminated by 1- or 2-fold coordinated O atoms or OH groups, respectively. The surface structure with 2-fold coordinated OH groups is found to be the energetically most favorable one (Figure 6b). Finally, the (001) surface, which has the simplest structure of all InOOH surfaces, is displayed in Figure 6c. The In(OH)3(100) surface is shown in Figure 6d. A (1 × 1) surface unit cell contains four In atoms in the surface layer. Two of them remain 6-fold coordinated with an OH group adsorbed on top, whereas two In surface atoms have a reduced coordination of 5. We probed several different possibilities for the distribution and orientation of the top two OH groups in the (1 × 1) surface unit cell. The lowest energy was found for a configuration in which the OH groups form pairs that are connected by an hydrogen bond (Figure 6d). The calculated cleavage energies γcleav for the four surfaces are summarized in Table 1. It can be seen that γcleav is mainly Table 1. Calculated Cleavage Energy γcleav (J/m2), Water Adsorption Energies Ew,i ad (eV), and Water/Solid Interface Energy γint (J/m2) for Different InOOH and In(OH)3 Surface Structures surface

γcleav [J/m2]

Ahkl [Å2]

Ew,1 ad [eV]

InOOH(100) InOOH(010) InOOH(001) In(OH)3(100)

1.26 1.20 2.25 1.31

15.5 17.9 24.9 64.8

−0.76 −0.81 −1.35 −1.28

Ew,2 ad [eV]

γint [J/m2]

−1.09 −1.28

0.45 0.33 0.57 0.61

determined by the reduction of coordination of the In surface atoms: although the cleavage energies are rather similar for all three surfaces that only contain 5-fold coordinated In atoms, it is almost doubled for the surface with 4-fold coordinated In in the topmost surface layer. Thus, in an aqueous environment, where the vacant In coordinations will be saturated by water molecules, the relative stability of the four surfaces might be very different. Therefore, in the next step, we determined the energetically most favorable structure of all surfaces after adsorption of water on the vacant In coordination sites. For all four surfaces we considered molecular and dissociative adsorption as well as several different orientations of the adsorbed water molecules and a rearrangement of the OH groups that are already present on the surfaces. For the structures with lowest energy the water adsorption energies for the first, and if applicable, the second water molecule per (1 × 1) surface unit cell are given in Table 1 and the atomic configurations are shown in Figure 6. For the InOOH(100) and (010) surfaces, both containing one 5-fold coordinated In per (1 × 1) surface unit cell, we found that molecular adsorption is most favorable with a moderate water adsorption energy. At the InOOH(001) surface we need two water molecules to saturate the 4-fold coordinated In atom in the surface unit cell. The first water molecule dissociates, thereby protonating an O atom in the subsurface layer, and the second water molecule remains intact. The adsorption energy for the first water molecule is rather high, as to be expected from the high cleavage energy. But also the adsorption energy of the second water molecule is larger than in the case of the other two surfaces, which is due to the formation of additional hydrogen bonds. On In(OH)3(100) only molecular adsorption is possible because there are no O atoms available that can accept the dissociated proton. The

■

DISCUSSION I. Why Does the InOOH Phase Remain Semicrystalline? As evidenced by our results from XRD (see above), the cubic In(OH)3 phase and the orthorhombic InOOH phase appear within the whole range of refluxing time. As indicated by the SAED patterns, the nano- and microcuboids solely consist of cubic In(OH)3, which appears highly crystalline. This results in sharp and intense reflections in the SAED patterns and XRD, as observed in Figures 2−5. Because the XRD reflections for the orthorhombic InOOH phase appear broad and weak, we conclude that the semicrystalline fraction within our samples contains the orthorhombic InOOH phase. In contrast to In(OH)3 the InOOH phase maintains its semicrystalline nature even for long refluxing times, as is also observed by Lin et al.12 So far no comment and no explanation for the semicrystalline nature of the InOOH phase are given in the literature. The InOOH phase also forms semicrystallites if other In3+ sources like indium chloride or other OH− sources like urea (Avivi et al., 11 Lin et al.12) are used. Thus, the formation of semicrystalline InOOH is not restricted to our system and surface adsorption of urea or acetate do not seem to play a dominant role. As discussed in the Theory section, the growth of InOOH crystallites will be irregular, because of disorder in the distribution of OH groups and O ions in the InOOH lattice and the associated lattice strain. This is attenuated by the fact that some OH disorder must be maintained to quench the polarity of the (001) surface orientation of InOOH. The 24535



Article

surface of the cuboids occurs. However, this plays only a minor role in the cuboid formation. A growth mechanism for In(OH)3 nanocubes similar to that observed by us has been reported very recently by Jean et al.14 Consistent with our results Jean et al.14 observed that In(OH)3 nanocubes predominantly form by a multistep OA mechanism. However, in contrast to our findings, Jean et al.14 do not observe InOOH as a second phase behind In(OH)3. Because Jean et al.14 use urea as the OH− source, they expect the release of OH− and the formation rate of In(OH)3 particles to be slow. In contrast to this, we use KOH as the direct OH− source and thus expect a formation rate of In(OH)3, which is fast compared to that in the experiment of Jean et al. Because a similar growth mechanism is observed in both cases, we conclude that the release rate of OH− and also the presence of urea as surface blocking agent as well as the In3+ source do not play a dominant role in the formation of the In(OH)3 cuboids. This shows that the formation of anisotropic cuboids is not predominantly effected by the presence of urea, as supposed by Jean et al.14

quenching of the surface polarity might be the reason why no growth by OA is observed for the InOOH phase. II. What Is the Reason for the Unusual Phase Transition Pathway? As reported by Lin et al.,12 the InOOH phase appears to be formed before the In(OH)3 phase. As already mentioned above, Lin et al.12 speculate the presence of urea as OH− source in their reaction to be responsible for this unusual phase behavior. Because we do not use urea as the OH− source, the presence of urea cannot be the reason for this unusual phase transition behavior. From our experiments it becomes evident that the InOOH phase is already present at low refluxing times, which is in accord with the results obtained by Lin et al.12 The formation of InOOH and the phase transition to In(OH)3 can be described as follows: In the alkaline solution the initial nucleus has to consist of In(OH)3 units. However, already at an very early stage, the In(OH)3 nuclei should eliminate water molecules from their core while maintaining a hydroxide shell because of the lower water/solid interface energy of InOOH. The same mechanism has been suggested for the growth of ZnO nanocrystals in alkaline solution.36 From a critical size onward, In(OH)3 particles become energetically more favorable than InOOH. Once InOOH particles have transformed to In(OH)3 by incorporation of water, the particles can grow without restriction and without frustration because of the regular cubic crystal structure. As reported by Schlicker at al.,24 no intermediate phase forms during transition from cubic In(OH)3 to bixbyite-type In2O3. This means that orthorhombic InOOH does not form due to dehydroxylation of cubic In(OH)3.24 III. What Is the Role of OA during the Formation of the In(OH)3 Cuboids? Based on our experimental results above, a three-step growth model can be supposed for the In(OH)3 nano- and microcuboids as illustrated in Scheme 1. In

■

CONCLUSIONS Precipitation of In3+ with OH− under the presence of acetate ions in water results in the formation of two different material fractions. The dominant fraction contains crystalline nano- and microcuboids, which solely consist of cubic In(OH)3. The second fraction contains orthorhombic InOOH and is semicrystalline. On the basis of DFT calculations we find that for small crystallites the InOOH phase is more stable than the In(OH)3 phase, leading to a size-dependent transition from the InOOH phase to In(OH)3 above a critical crystallite size. This explains the unusual phase behavior, as observed for this material system. Former studies only speculate about the reasons for this unusual phase behavior. In addition, we explain the semicrystalline nature of the InOOH phase by a previously unknown proton transfer mechanism within the crystal lattice of InOOH. The growth of In(OH)3 nano- and microcuboids follows a three-step OA process: (1) 1D OA of In(OH)3 nanocubes to In(OH)3 nanorods occurs. (2) 2D OA of rods to parallel bundles, and (3) merging of the rod bundles into cuboids. Due to imperfect OA planar defects and vacancy clusters are formed reducing the XRD coherence lengths.

Scheme 1. Growth Model for the In(OH)3 Nano- and Microcuboidsa

■

ASSOCIATED CONTENT

* Supporting Information S

XRD patterns for a calcinated sample. This material is available free of charge via the Internet at http://pubs.acs.org/.

■

a

The cuboids form by three-step OA from nanocubes: (i) 1D OA of nanocubes under formation of nanorods with an aspect ratio of two, (ii) 3D OA to parallel bundles of nanorods, (iii) merging of the rod bundles into cuboids.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

the first step small nanorods form by one-dimensional (1D) oriented agglomeration. Growth of rods is likely due to oriented agglomeration because anisotropic growth is not expected for a conventional growth mechanism for a cubic phase. In the second step, three-dimensional (3D) OA to parallel bundles of nanorods occurs. In the third step the rod bundles merge into cuboids. The second and the third step are evidenced by the TEM results (see above). Besides the threestep OA, as discussed above, also OA of single rods to the

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) via Research Training Group 1161 “Disperse Systems for Electronic applications” and Cluster of Excellence “Engineering of Advanced Materials”. We also thank Evonik Industries AG. 24536



■

Article

REFERENCES

(1) Zhu, H.; Wang, Y.; Wang, N.; Li, Y.; Yang, J. Mater. Lett. 2004, 58, 2631−2634. (2) Abe, Y.; Hasegawa, T.; Takahashi, Y.; Yamada, T.; Tokura, Y. Appl. Phys. Lett. 2005, 87, 153506. (3) Chen, S.-Y.; Wu, M.-C.; Lee, C.-S.; Lin, M. C. J. Mater. Sci. 2009, 44, 794−798. (4) Tao, X.; Sun, L.; Li, Z.; Zhao, Y. Nanoscale Res. Lett. 2009, 5, 383−388. (5) Xu, J. Q.; Chen, Y. P.; Pan, Q. Y.; Xiang, Q.; Cheng, Z. X.; Dong, X. W. Nanotechnology 2007, 18, 115615. (6) Motta, F. V.; Lima, R. C.; Marques, A. P. A.; Li, M. S.; Leite, E. R.; Varela, J. A.; Longo, E. J. Alloys Compd. 2010, 497, L25−L28. (7) Yang, J.; Frost, R. L.; Martens, W. N. J. Therm. Anal. Calorim. 2010, 100, 109−116. (8) Liu, X.; Zhou, L.; Yi, R.; Zhang, N.; Shi, R.; Gao, G.; Qiu, G. J. Phys. Chem. C 2008, 112, 18426−18430. (9) Huang, J.; Gao, L. J. Cryst. Growth Des. 2006, 6, 1528−1532. (10) Du, J.; Yang, M.; Cha, S. N.; Rhen, D.; Kang, M.; Kang, D. J. Cryst. Growth Des. 2008, 8, 2312−2317. (11) Avivi, S.; Mastai, Y.; Gedanken, A. Chem. Mater. 2000, 12, 1229−1233. (12) Lin, S.-E.; Wei, W.-C. J. Am. Ceram. Soc. 2006, 89, 527−533. (13) Yan, T.; Wang, X.; Long, J.; Lin, H.; Yuan, R.; Dai, W.; Li, Z.; Fu, X. New J. Chem. 2008, 32, 1843−1846. (14) Jean, S.-T.; Her, Y.-C. CrystEngComm 2011, 13, 2324−2330. (15) Koga, N.; Kimizu, T. J. Am. Ceram. Soc. 2008, 91, 4052−4058. (16) Gurlo, A.; Kroll, P.; Riedel, R. Chem.Eur. J. 2008, 14, 3306− 3310. (17) Zhu, H.; Wang, X.; Wang, Z.; Yang, C.; Yang, F.; Yang, X. J. Phys. Chem. C 2008, 112, 15285−15292. (18) Burrows, N. D.; Yuwono, V. M.; Penn, R. L. MRS Bull. 2010, 35, 133−137. (19) Niederberger, M.; Cölfen, H. Phys. Chem. Chem. Phys. 2006, 8, 3271−3287. (20) Lee, E. J. H.; Ribeiro, C.; Longo, E.; Leite, E. R. J. Phys. Chem. B 2005, 109, 20842−20846. (21) Penn, R. L.; Banfield, J. F. Am. Mineral. 1998, 83, 1077−1082. (22) Penn, R. L.; Banfield, J. F. Science 2008, 281, 969−971. (23) Ribeiro, C.; Vila, C.; Matos, J. M. E. d.; Bettini, J.; Longo, E.; Leite, E. R. Chem.Eur. J. 2007, 13, 5798−5803. (24) Schlicker, L.; Riedel, L.; Gurlo, A. Nanotechnology 2009, 20, 495702. (25) Ashok, V. D.; De, S. K. J. Phys. Chem. C 2011, 115, 9382−9392. (26) Krischner, H.; Koppelhuber-Bitschnau, B. Röntgenstrukturanalyse und Rietveldmethode, 5th ed.; Vieweg: Braunschweig/Wiesbaden, 1994. (27) Giannozzi, P.; et al. J. Phys.: Condens. Matter 2009, 21, 395502. (28) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865−3868; Erratum. Phys. Rev. Lett. 1997, 78, 1396. (29) Vanderbilt, D. Phys. Rev. B 1990, 41, 7892−7895. (30) Mullica, D. F.; Beall, G. W.; Milligan, W. O.; Korp, J. D.; Bernal, I. J. Inorg. Nucl. Chem. 1979, 41, 277−282. (31) Lehmann, M. S.; Larsen, F. K.; Christensen, A. N.; Poulsen, F. R.; Christensen, A. N.; Rasmussen, S. E. Acta Chem. Scand. 1970, 24, 1662−1670. (32) Qian, G.-X.; Martin, R. M.; Chadi, D. J. Phys. Rev. B 1988, 38, 7649−7663. (33) Stull, D. R.; Prophet, H. JANAF Thermochemical Tables, 2nd ed.; U.S. National Bureau of Standards, U.S. GPO: Washington, DC, 1971. (34) Zhuang, Z.; Peng, Q.; Liu, J.; Wang, X.; Li, Y. Inorg. Chem. 2007, 46, 5179−5187. (35) Chen, C.; Chen, D.; Jiao, X.; Wang, C. Chem. Commun. 2006, 4632−4634. (36) Kawska, A.; Duchstein, P.; Hochrein, O.; Zahn, D. Nano Lett. 2008, 8, 2336−2340.

24537


Phase Transition Behavior and Oriented Aggregation During

Recommend Documents