Article pubs.acs.org/Langmuir
Selective Adsorption to Particular Crystal Faces of ZnO Nathan Johann Nicholas,†,‡ George V. Franks,*,‡ and William A. Ducker*,† †
Virginia Polytechnic Institute and State University, Blacksburg, Virginia, United States Chemical and Biomolecular Engineering, University of Melbourne, Parkville, Victoria, Australia
‡
ABSTRACT: We examine the hypothesis that selective adsorption to a particular face of ZnO is responsible for the ability of small organic molecules to control the aspect ratio of ZnO crystals during hydrothermal synthesis. Large, single crystals of ZnO were prepared such that the vast majority of a surface consisted of a single crystal plane, as shown by atomic force microscopy, and the adsorption to a single crystal plane was determined by attenuated total reflectance spectroscopy. The results show that citrate strongly and selectively adsorbs to the (0001) face. Similarly, results show that ethylenediamine selectively adsorbs to the (1010̅ ) face. Each of these results separately shows a correlation between selective adsorption to and growth of large areas of a particular face, and thus, each result is consistent with the proposed hypothesis.
■
influence of amines (primary,16,17 secondary,18,19 tertiary,20,21 and quaternary14,22,23), carboxylic acids,14,24−27 or combinations of the two.18 The effects of sulfonates14,24 and metal ions28,29 have also been studied. From these studies, the overall trend observed was that synthesis in the presence of amines tends to create crystals with a high aspect ratio (i.e., crystals with a large amount of (101̅0) planes) while carboxylates tended to form a lower aspect-ratio crystal, with significantly more (0001) plane present. For molecules containing both amines and carboxylates (such as ethylenediaminetetraacetic acid (EDTA)), the crystal shape could be tuned depending on pH.18 Researchers hypothesize that the action of the small molecules is to adsorb selectively to particular crystal faces20,26 and thereby to decrease the energy of the face or block access of incoming Zn-containing solution species (i.e., growth units) to that face. If a molecule is adsorbed onto a particular face, it would prevent growth in the direction normal to that face, while growth in other directions is relatively unhindered. For example, adsorption to the (0001) face is expected to hinder growth in the [0001] direction while allowing growth in the (1010̅ ) directions to continue, thereby producing larger areas of the (0001) face and a crystal with a low aspect ratio. The experimental aim of this paper is to determine whether particular molecules adsorb selectively to particular faces of ZnO. The scientific aim is to combine adsorption results with the aspect ratios during hydrothermal synthesis (from the literature) to confirm or refute the hypothesis that selective adsorption is responsible for the observed aspect ratios. Before this goal could be achieved, it was necessary to (a) develop a
INTRODUCTION The potential use of ZnO crystals in catalysis,1,2 piezoelectric devices,3 solar cells,4 or LEDs5−8 has stimulated renewed interest in the control of the shape of ZnO particles during crystal growth.9 The normal growth habit of ZnO is a hexagonal prism, shown in Figure 1a, with an aspect ratio
Figure 1. ZnO crystals with two different aspect ratios (L/D). (a) High aspect ratio crystals with large (101̅0) faces are formed when the primary growth direction is [0001]. (b) Low aspect ratio crystals with large (0001)/(0001̅) faces are formed by growth in the ⟨101̅0⟩ directions. Note: the (0001) plane in the 4 index system is equivalent to the (001) plane in the 3 index system, and the (101̅0) plane is equivalent to the (100) plane.
(L/D) that is greater than 1. This type of crystal shape is suited for use in either piezoelectric devices,3 solar cells,4 or syngas catalysis.10,11 However, crystals with larger areas of the (0001) (basal) face (L/D < 1) are more suitable for use in LEDs.12 One promising way to achieve selective growth of crystals faces is via the introduction of small organic molecules during hydrothermal growth of ZnO.13−15 Studies have focused on the © 2012 American Chemical Society
Received: December 22, 2011 Revised: April 9, 2012 Published: April 11, 2012 7189
dx.doi.org/10.1021/la2050674 | Langmuir 2012, 28, 7189−7196
Langmuir
Article
Table 1. Structures and Concentrations of Adsorbates Investigated
Another route to understanding the mechanism by which small molecules affect the crystal geometry is to examine correlations between the ZnO aspect ratio and the structure of the organic modifier. Few studies have done this. One notable exception is by Yahiro et al,24 who systematically studied the influence of dicarboxylates on (0001) growth. They found that succinate (butanedioate) does not cause an increase in the growth of the (0001) plane but maleate (the cis isomer of butenedioate) does. Furthermore, the trans isomer of butenediote, fumarate, did not. From this, it is evident that the 3-D structure of a molecule is key to its ability to act as a growth modifier.
method to selectively measure the adsorption to a single face and (b) produce crystals with large areas of a particular face. Attenuated total internal reflectance Fourier transform infrared spectroscopy (ATR-FTIR) was chosen to measure the selective adsorption because ZnO is transparent in the frequency ranges of interest, whereas both the small organic molecules and water have adsorption bands. However, only small ZnO crystals (10 mm × 10 mm × 0.5 mm) are available and considerable effort was required to adapt an FTIR spectrometer to establish a light path into and out of the crystal to produce evanescent waves. A second issue is that ATR-IR probes large areas (several mm2) of crystal, so it was necessary to produce crystal surfaces that were predominantly a single crystal face over this area. Here we study the adsorption of a shape-directing molecule, sodium citrate. Citrate is a tricarboxylate molecule that has been reported14,25−27 to encourage the growth of ZnO with large areas of (0001) faces. The adsorption of three additional carboxylates (sodium citrate, sodium succinate, and sodium glutarate; see Table 1) was studied to better understand which structural features of citrate are important for selective adsorption. In this regard, the IR spectrum is useful for determining whether the adsorbing molecule interacts directly with a surface Zn atom (an “inner sphere complex”) or indirectly via water or hydroxide groups (an “outer sphere complex”). We also study an amine, ethylendiamine (en), which has been present in the reaction mixture when crystals are produced with a high aspect ratio (i.e., a large % of the area is covered by (101̅0) faces).16,17 FTIR has previously been used to probe the interactions between small molecules and mineral surfaces. Kubicki et al.30 studied the adsorption of various carboxylic acids on slurries of different alumina-silicates, correlating the observed IR spectra with theoretically calculated spectra for different bonding arrangements of the functional groups. They observed that small differences in mineral structure (such as that between kaolinite, illite, and montmorillonite) affect the adsorption of different carboxylic acids. They also found that chemical impurities within the samples (such as the presence of iron ions) encourage adsorption. A study by Biber and Stumm31 demonstrated salicyclic acid adsorption on different aluminates and iron(III) oxide suspensions. By comparing the peak positions of the adsorption spectra of salicyclic acid to minerals with different elemental compositions, but similar structures (corundum (α-Al2O3) and hematite (Fe2O3)), they found that the elemental composition was more crucial than surface geometry in determining how carboxyl groups bond to mineral surfaces. Both of these studies demonstrate how changes in mineral structure and composition can influence molecular adsorption, but both examined powders so that spectral results are the sum from different crystal faces. Therefore, it is impossible to determine face-selectivity in adsorption.
■
EXPERIMENTAL SECTION
Solution Preparation. The chemical species and concentrations are shown in Table 1. A constant concentration of carboxylate groups was used in the experiments (except in en solutions). All solutions were prepared using AR grade (purity 99+% Aldrich) chemicals dissolved in 99% D2O (Cambridge Isotopes). All chemicals, with the exception of the glutarate, were used without further purification. Na2glutarate was prepared by reaction of glutaric acid with 2 equiv of NaOH (e.g., 15 mM glutaric acid with 30 mM NaOH). The solvent was then evaporated at ca. 70 °C, and the residual powder resuspended in an equivalent volume of 1 mM NaOH in D2O. All other solutions were made directly using 1 mM NaOH in 99% D2O, which will be referred to as the “background solution”. The synthesis of ZnO is sensitive to pH, but the experiments here all use D2O as the solvent, so the relevant property is pD. To obtain an approximate estimate of pD, we measured the pH of solutions that were nominally identical to the D2O solutions, but instead used H2O. Throughout the paper, when we refer to pH, we mean the pH of the equivalent H2O solution. For example, the pH of the background solution was 10.7, but this is the pH of 1 mM NaOH in H2O. Shifts in pH of the solutions when the sodium salts were added were negligible (±0.1), with the exception of Na2glutarate, where the pH did vary by about ±0.5. Sample Preparation. Square prisms (10 mm × 10 mm × 0.5 mm) of ZnO single crystals were obtained from MTI Corporation. The large (10 mm × 10 mm) faces were either (0001) or (101̅0). Prisms were etched for 5 min in a 1 M NaOH solution, rinsed with ethanol/water, and then annealed in an open air furnace (Barnstead Thermolyne 48000) for 5 h at 1000 °C (ramp rate 5 °C/min). After annealing, the samples were etched for 1 min (1 M NaOH) and then rinsed with ethanol/water to remove any contaminants. ATR-FTIR. ATR-FTIR adsorption studies were performed using a customized ATR setup shown in Figure 2. The spectrometer (Varian 670 FTIR) produces a convergent hollow cone IR beam, that is, with very little intensity in the center of the cross section. Such a beam did not have sufficient intensity on the 10 mm × 0.5 mm face of the ZnO crystal for our ATR measurement. A ZnSe biconvex lens was used to further focus the beam, and it was necessary to divert the beam laterally to produce a focused beam far enough away from the detector to leave space for the prism. The beam impinged approximately normal to the 10 mm × 0.5 mm face, but because of the spread in angles of the incoming beam, much of this beam resulted in total internal reflection at the interface between the solution and the large 10 mm × 10 mm face. Inside the fluid cell, gold gaskets were placed between the cell walls and the ZnO prism to aid in sealing the cell, 7190
dx.doi.org/10.1021/la2050674 | Langmuir 2012, 28, 7189−7196
Langmuir
Article
in 99% D2O) for at least 90 min to allow the system to come to equilibrium. For a particular chemical species (listed in Table 1), the spectrum was measured at two concentrations. The lower concentration was injected into the fluid cell using a disposable syringe, and periodic scans were taken until equilibrium was reached (i.e., after several consecutive IR scans showed no change). The second, higher concentration solution was then injected and allowed to come to equilibrium. After no further change in adsorption was observed at the higher concentration, the fluid cell was flushed with background solution to examine desorption. If negligible desorption occurred, the cell was flushed a second time. All FTIR scans were taken as an average of 64 scans taken at a resolution of 2 cm−1. As both sides of a (101̅0) crystal are the same, both sides were exposed to solution (in series) and the spectrum is the sum of spectra from both sides. A key feature of a dipolar crystal like ZnO is that the (0001̅) and the (0001) surfaces of the crystal are different, so the adsorption studies on the (0001) surface were done differently. Initially, both sides were exposed to background solution. All subsequent fluid exchanges were then done on the (0001) side only. This was done so that the IR spectrum of the molecules adsorbed onto the (0001) surface could be determined unambiguously without interference from adsorption to the (0001̅) face. For comparison, the transmission spectrum of the different compounds was taken at the higher concentrations listed in Table 1 using a PIKE transmission cell under the same conditions as the absorbance spectra (64 scans at 2 cm−1 resolution). Atomic Force Microscopy (AFM). An MFP3D atomic force microscope (Asylum Research) using Veeco NP cantilevers was used to determine the topography of the large faces in air. Images were
Figure 2. Schematic of ATR-FTIR fluid cell and light path. (a) Beam path. A convergent IR beam was focused into ZnO ATR prism in contact with a solution containing the adsorbing molecule. The exit beam was focused onto an MCT detector using an off axis parabolic mirror. (b) Cross section of the fluid cell perpendicular to the incoming beam. while preventing access of the IR beam to the PTFE. Without these gaskets PTFE absorption bands appeared in the spectra. The beam exiting the prism was refocused on to a 1 mm × 1 mm liquid nitrogen cooled MCT detector using an aluminum off-axis parabolic mirror. Adsorption Studies. Single crystal ZnO with either large (0001) or (101̅0) faces were exposed to background solutions (1 mM NaOH
Figure 3. (a−d) 10 μm × 10 μm AFM images of the (0001) and (101̅0) ZnO prism surfaces before (a,c) and after (b,d) annealing. (e−h) 1 μm × 1 μm enlargements of (a−d). Height scale is 10 nm for images (a) and (c), 40 nm for (b) and (d), and 5 nm for (e−h). (i, j) O1s XPS spectra for the (0001) surface before (i) and after (j) annealing, showing a reduction in hydroxide content (532.3 eV) from 67% before annealing to 48% after annealing. 7191
dx.doi.org/10.1021/la2050674 | Langmuir 2012, 28, 7189−7196
Langmuir
Article
obtained at a scan rate of 1 Hz with 512 scan lines and 512 scan points, regardless of the area of the scan. X-ray Photoelectron Spectrometry (XPS). The surface composition of the ZnO prisms before and after annealing was determined using a Perkin-Elmer PHI 5400 spectrometer with a Mg Kα photon energy of 1253.6 eV at 45° angle of incidence. The source was run at 250 W (12.5 kV and 20 mA).
■
RESULTS AND DISCUSSION Preparation of Planar ZnO Surfaces. The (0001) and (101̅0) ZnO prisms were received precut and polished to roughness of 1200 cm−1), so it is difficult to make a definite conclusion.
spectra of 30 and 300 mM sodium acetate to (0001) plane ZnO. The spectrum of acetate in solution was taken using a transmission cell and has been included for comparison. First, there are no shifts in peaks, showing that the (0001) ZnO plane does not have a strong influence on the acetate carboxylate. Second, there is a large difference in absorption between the spectra obtained at 30 and 300 mM, showing that the adsorption of acetate to (0001) is either very weak or nonexistent (the adsorption is not saturated at 30 mM, as it was for citrate at 10 mM). Finally, almost all the acetate ion easily rinses off the (0001) surface. Clearly, a single carboxylate group does not have a strong affinity for the surface. Is the presence of two carboxylates sufficient to cause adsorption? Succinate has two carboxylate groups that have the same spacing (two carbons) as the central and terminal carboxylates on citrate (see Table 1). The IR spectra of succinate are shown in Figure 7 at concentrations of 15 and 150 mM on (101̅0) and (0001) plane ZnO. The transmission spectrum of succinate in solution has been included for comparison. The most obvious difference to citrate is that there is a large difference in magnitude of adsorption between 15 and 150 mM: the binding is not as strong as for citrate. However, on rinsing, nearly all the succinate remains adsorbed, so the binding is much stronger than for the acetate. Thus, simply having two carboxylates separated by two carbons is not sufficient to produce citrate-like affinity for the (0001) face, but is much better than a single carboxylate. Like citrate, succinate also shows selectivity in binding to the different faces of ZnO: the succinate rinses off the (101̅0) face 7193
dx.doi.org/10.1021/la2050674 | Langmuir 2012, 28, 7189−7196
Langmuir
Article
Figure 7. FTIR spectrum of Na2succinate in D2O solution, and adsorbed on (0001) and (101̅0) ZnO. 15 mM (light gray lines) and 150 mM (dark gray lines). Dotted lines indicate the spectrum after desorption. Dashed vertical lines added as a guide for the peak positions of Na2succinate in solution. The background spectrum for all spectra is 1 mM NaOH in D2O.
Figure 8. FTIR spectrum of Na2glutarate in D2O solution and adsorbed on (0001) and (1010̅ ) ZnO. 15 mM (light gray lines) and 150 mM (dark gray lines). Dotted lines indicate the spectrum after desorption. Dashed vertical lines added as a guide for the peak positions of Na2succinate in solution. The background spectrum for all spectra is 1 mM NaOH in D2O.
but remains on the (0001) face. However, previous work by Yahiro et al.24 showed that succinate does not promote the formation of large areas of (0001) faces. The irreversible binding is clearly not the key signature for aspect ratio control; the control of the ZnO morphology is correlated with changes in the wavenumber of absorption and changes in adsorption at low concentration, both indicators of strong adsorption. Succinate bonds to the surface, but the bonding is not strong enough to significantly hinder growth in the [0001] direction. Citrate also has carboxylic acid groups that are separated by three carbon atoms (Table 1). Glutarate is the simplest dicarboxylate with the same spacing, so its adsorption was also studied. Figure 8 shows the ATR-IR spectra of 15 mM and 150 mM glutarate on (101̅0) and (0001) plane ZnO as well as the transmission spectrum. For both the (0001) and (101̅0) surface there are large changes in the magnitude of the antisymmetric stretch between 15 mM and 150 mM showing that the adsorption is much weaker than for citrate, which had already reached saturation at about 10 mM. There are no peak shifts in the adsorption to (101̅0) plane suggesting that there is no specific interaction between (1010̅ ) ZnO and glutarate ions. The antisymmetric stretch is changed on the (0001) plane, it is split into four distinct peaks at 1533, 1571, 1596, and 1624 cm−1. The loss of the peak at 1556 cm−1 demonstrates that there is no free carboxylate: both carboxylate groups are interacting with the surface. (Recall from the acetate spectrum that a single carboxylate group does not interact strongly with the (0001) plane.) The presence of four distinct antisymmetric peaks may indicate that different succinate ions are in different conformations or bind to different types of surface groups. However, as for citrate, the majority of the new peaks are at higher wavenumber compared to aqueous glutarate ions, which again is consistent with the formation of monodentate complexes. The peak at 1533 cm−1 would be more consistent, however, with either a bridging or bidentate complex. It is interesting to note that the peak at 1533 cm−1 increases after
rinsing, while the peaks at 1596 and 1624 cm−1 decrease. This shows that once the glutarate ions are removed from bulk solution, the surface ions change conformation so that both of the carboxylate oxygen atoms are bound to the surface. Also looking again at the spectrum for citrate adsorbed onto (0001) ZnO (Figure 4), the symmetric COO− stretch appears to have a shoulder between 1450 and 1500 cm−1, which could also be caused if some of the carboxylate groups were forming bridging or bidentate complexes. Considering now the adsorption of each of the carboxylates, it is clear that citrate adsorbs to the (0001) surface at the lowest concentration and does not desorb as easily as the other carboxylates. We conclude that the effects of citrate arise from more than just two carboxylates: the third carboxylate and possibly the hydroxyl group help cause the strong adsorption to the (0001) ZnO plane. The glutarate (1,3 dicarboxylate) also experiences a large change in the spectrum on adsorption, so adsorption of both the terminal carboxylate groups of the citrate ion are probably the key to citrate adsorption. Adsorption of Ethylenediamine to (0001) and (101̅0) ZnO. As stated earlier, crystals with large areas of (1010̅ ) plane have been grown in the presence of ethylenediamine (en). Therefore, we also tested whether en adsorption differed on the (0001) and (101̅0) planes. Each of the (0001) and (101̅0) plane ZnO were exposed to 10 mM then 100 mM solutions of en, and the resultant IR spectra are shown as Figure 9. The most obvious result is that the absorbance intensity in 10 mM en to (101̅0) ZnO is already high enough to clearly measure a spectrum, and is about 50% of the intensity at 100 mM. In contrast, adsorption of 10 mM to (0001) is so low that absorbance is not resolved. Only at 100 mM does the en adsorption to (0001) become great enough for en to be observed. So en adsorbs much more strongly to the (1010̅ ) surface than the (0001) surface. Given that en is present when high aspect-ratio crystals are grown,16,17 this is consistent with the hypothesis that specific 7194
dx.doi.org/10.1021/la2050674 | Langmuir 2012, 28, 7189−7196
Langmuir
Article
Table 2. IR Peak Assignments for en between 1650 and 1300 cm−1 ethylenediamine (en) peak
aqueous
τ(CH2)/τ(−NH2) ω(CH2) ω(CH2) δ(CH2) γ(CH2) γ(CH2) γ(−NH2) γ(-+NH3)b γ(−NH2)
1325 1365 1410 1460 1502 1512 1584
(0001)
l
−a 1402a 1460a
1358 1415 1460
1551
1551
1608
a
Peak assignment ambiguous due to spectral noise. bPeak consistent with scissoring of ammonium ion.
combination of hydroxide (−OH) and hydrate (−OH2) species. Therefore, it is possible that the amine groups of en are bonding to either a hydroxyl group or replacing a water molecule to bind with a Zn atom, or possibly both. To check whether en binds to Zn2+ in solution, the transmission spectra of en in a 100 mM Zn(NO3)2/1 mM NaOH solution and 1 mM NaOH solution with no Zn(NO3)2 were recorded. The addition of Zn ions did not affect the spectrum, so the results in Figure 9 must be attributed to surface adsorption.
Figure 9. FTIR spectrum of en in D2O solution and on (0001) or (1010) ZnO. 10 mM (light gray lines) and 100 mM (dark gray lines). Dotted lines are en spectrum after desorption.
adsorption to a face leads to the growth of that face, but control of all other variables during the synthesis would be required to prove this relationship. However, unlike citrate on the (0001) plane, the en is not irreversibly bound to the (1010̅ )plane, as most of the en is removed by rinsing. Given the weaker adsorption of en, we would predict that the effect of en would be weaker than the effect of citrate on the hydrothermal synthesis of ZnO. From a practical viewpoint, the weaker binding of en may not be significant, because ZnO crystals already have a high aspect ratio when they are grown without any organic additives. The spectrum of en in solution was also examined. Even though the en is in D2O, complete proton−deuteron exchange with solution does not occur: the broad NH2 scissoring peaks are still present at 1584 and 1608 cm−1. While some exchange should occur, we could not monitor the N-D peak as it is
