Reactive-Layer-Assisted Deposition Mechanism and Characterization

Nov 12, 2012 - The growth mechanism of TiO2 films and their morphology are reported using the reactive-layer-assisted deposition (RLAD) method under ...
0 downloads 0 Views 5MB Size
Article pubs.acs.org/Langmuir

Reactive-Layer-Assisted Deposition Mechanism and Characterization of Titanium Oxide Films Liat Zilberberg and Micha Asscher* Institute of Chemistry, The Hebrew University, Jerusalem, Israel ABSTRACT: The growth mechanism of TiO2 films and their morphology are reported using the reactive-layer-assisted deposition (RLAD) method under ultrahigh vacuum conditions. The oxide film formation involves Ti atom deposition on top of amorphous solid water (ASW) condensed on a SiO2/Si(100) support at 90 K. Subsequent annealing leads to the desorption of all nonreacted buffer molecules, resulting in the deposition of the titanium oxide film. Employing mass spectrometry and using D2O as a buffer, we detected the evolution of deuterium molecules during titanium atom deposition. A solid state sol−gel-like formation mechanism of titanium oxide is proposed on the basis these observations. The morphology of the oxide films is characterized by AFM as a rather uniform amorphous thin film at room temperature. Upon further annealing above 750 K, crystallization of the titanium oxide film has set in, coinciding with a dewetting process of the oxide layer, and information obtained from similar growth procedure on an amorphous carbon-covered TEM grid. It was shown that these films are rather insensitive to the underlying substrate at temperatures below 500 K. solid water (ASW)20 as inert buffer materials to manipulate the size and density of metallic clusters, in RLAD the ASW serves as one of the reactants. The RLAD method was demonstrated for the first time in preparing magnesium oxide (MgO), with condensed oxygen as the buffer material,18 and later in producing TiO2 on top of the Au(111) substrate with ASW as the buffer.21 The main objectives of the current research have been to study the mechanism of preparation of thin TiO2 films by the RLAD method and to investigate the morphology of the obtained material grown on top of another oxide, SiO2/ Si(100). Understanding this process may open the way to manipulate the composition of TiO2 via atom doping and the controlled formation of defects in order to modify its optical and electronic properties.

1. INTRODUCTION The photoactivity of TiO2 has been the focus of many studies over the past two decades, primarily because of the fact that under UV irradiation efficient internal charge separation occurs. This effect has been employed for the photolysis of water1 and for the degradation of organic molecules via photocatalysis.2 These activities are utilized in several environmental applications, such as water decontamination,3 to produce fuels such as hydrogen,4−6 and antibacterial functions.7 All of these applications originate from the excitation of TiO2, leading to highly reactive surface sites. TiO2 is a wide band gap (∼3.1 eV) semiconductor; therefore, it cannot be efficiently excited by visible solar radiation. Different material modifications were tested to decrease the TiO2 band gap, for example, the formation of bulk and surface defects.8,9 The dominant type of defects arise from oxygen vacancies,10 resulting in reduced Ti ions typically from the Ti4+ to Ti3+ oxidation state. These reduced sites change the electronic structure of the material, leading to band gap narrowing and thus shifting of the absorption toward the visible region. Reduced Ti ion sites can be prepared, for example, by heating TiO2 in vacuum or by ion sputtering.11,12 Among the commonly used methods for the preparation of photocatalysts is the sol−gel process,13 which usually takes place in the liquid phase under ambient conditions. Other methods including CVD (chemical vapor deposition),14,15 plasma spraying,16 and the thermal oxidation of sputtered Ti films17 are carried out in the gas phase under vacuum conditions. An alternative way to prepare oxide nanostructures is RLAD (reactive-layer-assisted deposition),18 which is similar to the BLAG (buffer-layer-assisted deposition) method.19 Although employing the BLAG method requires Xe19 and amorphous © 2012 American Chemical Society

2. EXPERIMENTAL SECTION The growth of titanium oxide via the RLAD procedure has been performed within a Varian ion-pumped ultrahigh-vacuum (UHV) chamber at a base pressure of 1 × 10−10 Torr equipped with an e-beam evaporator (McAllister), a quartz microbalance (QMB, Sigma) and a quadrupole mass spectrometer (QMS, SRS-200). Various materials were used as substrates for the growth of the titanium oxide film: 12 × 6 mm2 (n-type) SiO2/Si(100), 12 × 6 mm2 aluminum oxide crystals (sapphire), Si3N4, and amorphous carbon (aC) on a copper grid, the standard sample holders for transmission electron microscopy (TEM) imaging. The SiO2/Si(100) and the sapphire substrates were sonicated for 5 min in isopropanol and then attached to a stainless steel sample holder and inserted into the vacuum chamber. The Si samples were inReceived: July 21, 2012 Revised: October 31, 2012 Published: November 12, 2012 17118

dx.doi.org/10.1021/la302957q | Langmuir 2012, 28, 17118−17123

Langmuir

Article

vacuum annealed to 1000 ± 30 K for 10 min on a separate holder with a heating capability of up to 1100 K. After annealing, the sample and its holder were in-vacuum transferred to the bottom of a liquidnitrogen Dewar and cooled to about 90 K. The temperature during the cooling process was measured by a W26%Re−W5%Re thermocouple spot welded to a thin tantalum foil attached between the substrate and one of the clips that holds the sample. D2O molecules were deposited on the substrate by backfilling the UHV chamber while keeping the sample temperature at 90 K. The thickness of the ASW layer, 20−300 ML, has been determined by recording the D2O pressure, 1.7 × 10−7 Torr, for different time intervals. The sticking probability of water molecules on all of our substrates at 90 K was assumed to be unity; therefore, 1 L (1 Langmuir = 10−6 Torr s) was considered to be equivalent to 1 ± 0.1 ML of D2O on the different substrates. Under these experimental conditions, the water buffer layer has the compact structure of ASW.22 Subsequently, titanium atoms were evaporated on the water buffer layer. This was performed by an e-beam evaporator. The titanium atom flux was calibrated in situ by the QMB. Deposition rates of 0.01−0.1 Å/s were employed. We have constructed a magnetically coupled shutter between the Ti evaporator and the sample, enabling us to modulate the flux of impinging titanium atoms with a period of 25 s. After the Ti deposition has been completed, the sample was heated to 300 K or higher in order to desorb the remaining water molecules and to heat treat the titanium oxide film. The oxidation states of the products were investigated in situ by Auger electron spectroscopy measurements. The products were subsequently exposed to the ambient environment and ex-situ examined by SEM, XPS, TEM, and AFM, all at room temperature. AFM measurements were performed in order to image the morphology of the titanium oxide on silicon oxide and sapphire substrates, and TEM enabled the determination of the morphology and internal structure of the titanium oxide nanoclusters on a-C. The chemical composition of the titanium oxide layers on both the silica and the sapphire substrates has been determined by XPS, and its optical properties have been determined by a UV−vis spectrometer.

To support the suggested mechanism, we have investigated the formation of monomers Ti(OD)n (n ≤ 4) using D2O as a buffer layer. By employing a magnetically driven shutter between the Ti evaporator and the sample, we could modulate the Ti flux every 25 s. The emitted flux of the sol−gel reaction products, D2, is therefore also modulated, as detected by QMS (Figure 1).

Figure 1. Evolution of D2 from D2O ASW at various thicknesses vs time. The Ti atom impingement flux is modulated every 25 s.

The average number of OD groups per Ti atom could be evaluated by estimating the number of D2 molecules emitted per Ti atom. This was done by measuring the pressure increase at mass 4 (D2), as detected by QMS, of (3 − 5)10−9 Torr. The flux of these products was calculated by the Hertz−Knudsen equation f = P/(2πmKT)1/2molecules/m2 s, with the pressure P corrected for the ionization gauge sensitivity factor.23 The contribution to the pressure rise due to emitted DH molecules during the Ti evaporation, as a result of some HDO and H2O contamination within the D2O layer, has been considered as well. The Ti flux was calculated from the evaporation rate determined by the QMB to be 0.017 ± 0.005 Å/s. Eventually, the resulting flux ratio was found to be 1.2 ± 0.4 Ti atoms per D2 (HD) molecule. Considering the density of a single Ti layer, 1.4 × 1019 atoms/m2, we conclude that a monolayer should be equivalent to 3.2 Å of Ti. The flux of impinging Ti atoms was further calibrated by measuring the step height formed by masking half a sample during the deposition of a 20 nm layer of Ti. This step was measured by employing ex situ AFM measurements, not shown here, that confirmed the QMB data to within the uncertainty of the QMB instrument. The obtained ratio indicates that a hot Ti atom striking the ASW top surface on average breaks one O−D bond in each pair of neighboring D2O molecules, releasing one D2 (DH) molecule. Therefore, in the first stage of the reaction the formal oxidation state of Ti is +2.

3. RESULTS AND DISCUSSION 3.1. Sol−Gel-like TiO2 Film Preparation Mechanism. Titanium oxide films were prepared in this study by employing the RLAD method, following Ti atom impingement on D2OASW acting as the reactive layers. The suggested mechanism for TiO2 formation is based on a sol−gel-like (SGL) reaction. Although the sol−gel mechanism has always been associated with liquid-phase chemistry,13 the SGL mechanism described here is basically a low-temperature, solid-phase process. The standard liquid-phase sol−gel mechanism is initiated with monomer precursors that are condensed to create a gel phase that is a mixture of polymers and a liquid solvent. A solid material is subsequently formed by evaporating the solvent upon heating the system. 13 Our low-temperature SGL mechanism is based on similar principles. As in the classical mechanism, the first stage in the low-temperature in-vacuum reaction is the formation of the monomers (Ti(OH)n (n ≤ 4)) by a redox reaction. This step occurs while the hot evaporated Ti atoms hit the surface of the ASW. The chemically reactive Ti atoms reduce the water molecules and break O−D bonds in the D2O molecules, resulting in the release of D2 molecules into the vacuum. Gradually heating the system from 90 to 250 K leads to two simultaneous processes while the D2O layer desorbs: (1) condensation of two neighboring titanium hydroxyls to produce an oxide bridge while releasing a single D2O molecule and (2) film aggregation or calcination. In contrast to the standard sol−gel processes, in the solidstate SGL mechanism these two processes occur simultaneously at low temperatures (