Direct Measurement of van der Waals and Diffuse Double-Layer

---

Article pubs.acs.org/JPCC

Direct Measurement of van der Waals and Diffuse Double-Layer Forces between Titanium Dioxide Surfaces Produced by Atomic Layer Deposition Rick B. Walsh,† Andrew Nelson,‡ William M. Skinner,§ Drew Parsons,† and Vincent S. J. Craig*,† †

Department of Applied Mathematics, Research School of Physics and Engineering, The Australian National University, Canberra ACT 0200, Australia ‡ Bragg Institute, Australian Nuclear Science and Technology Organization, Locked Bag 2001, Kirrawee DC NSW 2232, Australia § Ian Wark Research Institute, Mawson Lakes Campus, University of South Australia, Mawson Lakes SA 5095, Australia S Supporting Information *

ABSTRACT: The van der Waals forces between titanium dioxide surfaces produced by atomic layer deposition (ALD) at the isoelectric point have been measured and found to agree with the calculated interaction using Lifshitz theory. It is shown that under the right conditions very smooth ALD surfaces are produced. At pH values slightly below and above the isoelectric point, a repulsive diffuse double-layer repulsion was observed and is attributed to positive and negative charging of the surfaces, respectively. At high pH, it was found that the forces remained repulsive up until contact and no van der Waals attraction or adhesion was evident. The absence of an attraction cannot be explained by the presence of hydration forces.

■

index,10 it also finds application as a photocatalyst, in optical coatings, in solar cells, and as the outermost surface of titanium implant materials, where it is responsible for adsorption of the extracellular matrix that confers biocompatibility. It is also highly suitable for fundamental surface science studies because it is chemically stable, has low solubility, is nontoxic, and has an isoelectric point that is readily accessible.11 Our interest is in measuring the surface forces between TiO2 surfaces in aqueous solutions. Here the high refractive index (i.e., high dielectric strength) of TiO2 results in large dispersion forces (also known as van der Waals, Lifshitz, or Casimir forces), and the existence of an isoelectric point (iep) around the center of the pH range means that neutral, positively charged, and negatively charged TiO2 surfaces are easily accessible. The surface of TiO2 is dominated by three surface groups, the singly, doubly, and triply coordinated groups (TiO−4/3, TiO2−2/3, and TiO3, respectively).12 The proton affinity for the surface group TiO−4/3 is so high that when placed in contact with water it immediately attracts a proton to form the group TiOH−1/3. Adsorption of protons to the TiO3 surface group does not occur in the normal pH range due to the low proton affinity of the group.13 Therefore, charging of

INTRODUCTION Understanding the fundamental forces between matter has long been recognized as an important scientific endeavor and elucidating the relationship between the properties of materials and surface forces by direct force measurement is an ongoing pursuit.1 Direct force measurements have been conducted between mica,2 silica,3 gold,4 cellulose,5 polystyrene6 and polypropylene7 surfaces both pristine and modified. Notably, very few direct surface force investigations of mineral oxide surfaces have been reported8,9 despite strong interest in these materials, as they generally cannot be prepared in a suitable form. For precise quantitative force measurements, the surfaces must be prepared in a suitable geometry (typically a cylinder, flat or sphere) with minimal surface roughness. For this reason, a large number of investigations are performed using mica or silica. Here we have employed a technique that is widely used to produce electronic devices, atomic layer deposition (ALD), whereby a film of material is grown, atomic layer by atomic layer, to produce surfaces suitable for force and surface science measurements. Importantly, the growth is conformal and adds minimally to the roughness of the substrate material under appropriate growth conditions. Therefore, the ALD technique can be used to prepare many materials that previously were not available in a suitable form for surface force investigations. Here we describe measurements between titanium dioxide (TiO2) films produced by the ALD technique. TiO2 is a technologically and industrially important mineral oxide. Whereas it is primarily used as a light-scattering pigment utilizing the dielectric properties that result in a high refractive © 2012 American Chemical Society

Received: January 16, 2012 Revised: March 7, 2012 Published: March 16, 2012 7838

dx.doi.org/10.1021/jp300533m | J. Phys. Chem. C 2012, 116, 7838−7847

The Journal of Physical Chemistry C

Article

but silica or silicon surfaces coated with TiO2 films, we must also calculate the van der Waals interaction that is expected between the layered surfaces, which is sensitive to the thickness of the TiO2 films. Dispersion Force Theory for Layered Systems. The Hamaker van der Waals interaction energy between a TiO2− coated flat surface and a similarly coated silica probe, separated by distance L, is given (for flat surfaces) by

the surface is controlled by the protonation of the singly and doubly coordinated groups, which can be described by TiOH−1/3 + H+ ↔ TiOH2+2/3

(1)

Ti2O−2/3 + H+ ↔ Ti2OH+1/3

(2)

By comparison, silica which is the most commonly used surface for force studies, has an iep at very low pH, such that positively charged silica surfaces are not generally accessible and the Hamaker constant, which is a measure of the strength of the van der Waals forces, for silica surfaces is nearly an order of magnitude less.14 Moreover, the accessibility of the iep for TiO2 surfaces should allow the van der Waals interaction to be directly measured; this is not possible for silica surfaces. To our knowledge, there has been only one previous surface force investigation between unadulterated TiO2 surfaces in aqueous solutions. This was by Larson et al.9 in 1993 and was performed between rutile surfaces. We feel that TiO2 is worth revisiting because due to the inherent roughness of the rutile colloid probe employed, they were unable to investigate the short-range forces with certainty, and force measurements on positively charged surfaces were not presented. Indeed, it is difficult to obtain TiO2 surfaces that are suitable for force measurements at small separations. For quantitative AFM measurements a flat substrate and a spherical colloid probe are required, both with minimal roughness. To produce suitable surfaces, we employ a technique that was not available to Larson et al., ALD, whereby a TiO2 film can be grown conformally on a substrate a (sub) monolayer at a time. Here we take silica and silicon surfaces commonly used to study surface forces and coat them with TiO2 by the ALD method. By appropriate choice of deposition conditions, surfaces with roughness nearly equivalent to the underlying substrates can be produced on both flat and spherical substrates.15 We have several aims in this investigation. (I) We wish to measure the surface forces at the iep to ascertain the magnitude of the van der Waals forces between TiO2 surfaces across water and compare the measurements with theoretical estimates arrived at from spectral data. (II) We wish to compare the forces at short range with what is seen between silica surfaces. Measurements between silica surfaces do not follow the DLVO paradigm − no van der Waals attraction is seen. It is generally held that silica surfaces are strongly hydrated, and this gives rise to significant hydration or secondary hydration forces, which are strongly repulsive and sufficient in range to mask the van der Waals attraction.3,16 TiO2 surfaces are of interest in this regard because the magnitude of the van der Waals attraction is nearly an order of magnitude greater than that for silica surfaces; therefore, the measurable range of the van der Waals attraction should extend beyond the range of any hydration force, which decays exponentially with a characteristic length similar to the size of a water molecule.17 (III) Also, we wish to make measurements at pH values below the iep, where the surfaces are positively charged. Because both surfaces are identical, this should also give rise to repulsive forces that would stabilize a colloidal dispersion. Because ALD TiO2 surfaces have not previously been employed in these types of investigations, we have characterized them by atomic force microscopy, zeta potential measurement, X-ray reflection, X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), and nanoindentation. Additionally, because we are not using solid TiO2 surfaces

FHamaker(L) =

− A (L ) 12πL2

(3)

The separation-dependent Hamaker coefficient (not constant) A(L) is calculated with retardation (due to the finite speed of light) using the multilayer formula of Parsegian and Ninham.18 This calculation neglects nonadditivity effects,19 which may be large at large separations (L >100 nm, but at these large separations the Hamaker force is, in any case, small). The algorithm uses an effective dielectric coefficient for a multilayer structure and applies a recursive relationship to the dielectric properties of the component layers. We place one multilayer surface to the left and a second multilayer to the right, with an intervening medium m with thickness L. The Hamaker coefficient is given by A (L ) =

2kT 3

∞′

∑∫ r

n=0

∞

dx x{B + C}

n

(4)

where B = ln[1 − Δ̅ mL(i ωn)Δ̅ mR (i ωn)e−x]

and C = ln[1 − ΔmL(i ωn)ΔmR (i ωn)e−x]

The prime against the summation in n indicates that the zero frequency term in n = 0 is taken with a factor of 1/2. ωn are the so-called Matsubara frequencies, ωn = 2πkTn/ℏ . rn = 2Lωn(εm(iωn))1/2/c, where εm is the dielectric function of the intervening medium between the two multilayers and c is the speed of light. In terms of the underlying fundamental physics, Δ describes the reflection of the magnetic component of a photon passing across an interface and Δ̅ describes the reflection of the electric component. Here ΔmL and Δ̅ mL refer to the reflection coefficients between the medium and the left multilayer structure, and ΔmR and ΔmR are the reflection coefficients between the medium and the right multilayer. Taken between two whole materials j and k, the retarded diamagnetic reflection coefficient is defined at an imaginary frequency iω as Δjk (i ω) =

sjμj − sjμk sjμj + sjμk

where μ is the magnetic permeability of each material in turn. In our case, all materials are nonmagnetic, such that μ = 1. The dielectric reflection coefficient is Δ̅ jk (i ω) = 7839

sj εj − sj εk sj εj + sj εk dx.doi.org/10.1021/jp300533m | J. Phys. Chem. C 2012, 116, 7838−7847

The Journal of Physical Chemistry C

Article

where ε is the electric permittivity or dielectric function (frequency dependent) of each material in turn. The variable s is a retardation coefficient sj(i ω) =

ballotini was sieved to give a sample of lower polydispersity with radius ∼15 μm. Before being placed in the reaction chamber, both the spheres and wafers were treated with RF water plasma (30 W for 90 s, followed by 50 W for 30 s) using an in house system. Surface Characterization. The zeta potential of TiO2coated glass ballotini spheres was measured using a Zetasizer Nano ZS instrument (Malvern Instruments, U.K.) with 0.001 M NaCl as background electrolyte. The pH was adjusted using an autotitrator system attached to the ZetaSizer. HCl and NaOH solutions were used for the adjustments. The X-ray reflectivity (XRR) and XRD measurements were performed using a PANalytical X’Pert Pro diffractometer located at the Australian Nuclear Science and Technology Organization. This instrument employs a Cu Kα (8048 eV) Xray source. XRD measurements were measured over the 2θ range of 20−80°. XRR data were fitted using Motofit21 software. XRD spectra were compared with literature spectra.22 The XPS was carried out in the Ian Wark Research Institute using a Kratos Axis-Ultra spectrometer employing a monochromatic Al Kα X-ray source (1486.6 eV) operated at 130 W. All samples were cooled, in vacuum, using a liquid-nitrogen-fed cold stage (temperature < −100 °C) to minimize photoreduction during exposure to X-rays during analysis.23 Survey spectra were initially acquired with a pass energy of 160 eV, to quantify detected elements, in atomic %, using the appropriate elemental sensitivity factors. High-resolution spectra of the Ti 2p, O 1s, and C 1s were acquired with a pass energy of 20 eV, resulting in a fwhm at the Ti 2p3/2 of ∼1.2 eV. All spectral fitting was accomplished using the CasaXPS software package (http://www.casaxps.com/). Atomic force microscope height mode images of ALD surfaces were obtained using a Nanoscope Multimode IIIa AFM with phase extender box in tapping mode. Tap300Al cantilevers of nominal resonance frequency 300 kHz and spring constant 40 N m−1, supplied by Budget Sensors, were employed, and the feedback parameters were adjusted to minimize the features in the amplitude image. Measurement of the Young’s modulus of the films was performed using a Hysitron TI-950 TriboIndenter using a Berkovich tip with a Macor shaft. Samples were mounted in stainless-steel stubs with a recess cut into the top surface, which allowed the thin film and the nanoindentation tip to be immersed in electrolyte solution during measurements. Measurements were made by indenting a 5 × 5 grid covering 30 μm × 30 μm. Experimental results were analyzed using the technique described by Oliver et al.24 to determine the reduced modulus of the TiO2 thin films. Surface Force Measurements. The surface forces between a spherical TiO2-coated particle and a TiO2-coated surface were measured using a Digital Instruments Multimode Nanoscope III AFM. Use of a fluid cell enabled surface force measurements under aqueous conditions. The colloid probe technique developed by Ducker16 et al. and Butt25 was used. An ALD-coated sphere was attached to the end of a rectangular silicon nitride cantilever (CSG11 supplied by ND-MDT), using purified Epikote 1004 resin (Shell) via the use of a three-way stage manipulator. The interaction between the ALD TiO2 colloid probe and an ALD TiO2 surface was investigated. The colloid probes were imaged to determine the radius of the particle using a Zeiss UltraPlus analytical FESEM. The cantilevers were calibrated prior to particle attachment via the use of the thermal tune method using an Asylum Picoforce

p2 − 1 + εj(i ω)/εm(i ω)

where p=

xc 2ωL (εm(i ω))

in the nonretarded limit, neglecting the finite speed of light (i.e., taking the limit c → ∞), then the retardation coefficients si → ∞ and cancel the reflection coefficients, leaving Δjk → 0 (when μ = 1) and Δ̅ jk → (εj − εk)/(εj + εk). A recursion formula leads from the multilayer reflection coefficients ΔmL, Δ̅ mL, ΔmR, and Δ̅ mR down to the coefficients Δjk, Δ̅ jk between neighboring whole (nonlayered) materials ΔmL =

l / pL Δm ,1 + Δ1, Le−xs11 l / pL 1 + Δm ,1Δ1, Le−xs11

where lN is the thickness of the layer N adjacent to the central medium and Δj,L is the reflection coefficient of each layer j with j becoming larger with increasing distance from the central medium such that ΔjL =

Δj , j + 1 + Δj + 1, Le−xsj + 1l j + 1/ pL 1 + Δj , j + 1Δj + 1, Le−xsj + 1l j + 1/ pL

In the nonretarded limit the exponential term reduces to e−xlj−1/L, which means that in this limit the Hamaker coefficient of a multilayer system retains a dependence on separation L, depending on the relative thickness of the layers. That is, the van der Waals interaction of a multilayered system, in general, cannot be reduced to a single Hamaker constant independent of surface separation.

■

EXPERIMENTAL METHODS Atomic Layer Deposition. ALD of TiO2 was carried out using a Savannah 100 (Cambridge Nanotech) system with titanium isopropoxide and water as precursors. Depositions were carried out with the chamber at both 80 and 250 °C; this was found to have a significant influence on the nature of the film. Nitrogen was used as the carrier gas at a flow rate of 20 sccm (standard cubic centimeters per minute). A deposition cycle consists of a 0.015 s pulse of the TiO2 precursor, a purge with nitrogen for 10 s, and a 0.015 s pulse of water, followed by a second nitrogen purge for 10 s. A single cycle deposits one atomic layer of TiO2. The number of cycles employed varied between 280 and 2200 and was used to control the thickness of the resulting film. Substrates. ALD TiO2 films were grown on three different types of substrate. Flat surfaces were produced using borondoped silicon wafers (100) supplied by MEMC, U.S., with a native oxide layer of thickness ∼2 nm. Borosilicate spheres of monomodal size distribution and radius 10 ± 0.1 μm, supplied by Duke, (borosilicate glass 9020) were used as substrates for force measurements. These were chosen for their very low surface roughness,20 sphericity, and convenient size. Because these spheres are expensive and available in small quantities, inexpensive glass ballotini spheres were employed as substrates for zeta potential measurements. A very polydisperse sample of 7840

dx.doi.org/10.1021/jp300533m | J. Phys. Chem. C 2012, 116, 7838−7847

The Journal of Physical Chemistry C

Article

Figure 1. Tapping mode AFM images from left to right of type A, B, and C TiO2 surfaces. The surface roughness over an area of 1000 nm × 1000 nm was determined to be 8.2, 0.24, and 0.47 nm rms, respectively. In each image, the scale bar is 400 nm long.

AFM. The spring constant was corrected for off-end loading due to the attachment of the colloid particle to the tip using the correction detailed in Sader et al.26 Analytical grade chemicals were used in the experiments. NaCl was roasted at 400 °C for at least 12 h before use to remove organic contaminants. All solutions were prepared using Milli-Q water (conductivity ∼18.3 MΩ). All glassware were soaked in 10 w/w% NaOH solution for 10 min then rinsed with copious amounts of Milli-Q water before use. Before injection into the AFM fluid cell, solutions were bubbled with high-purity nitrogen gas to remove surface active contaminants following the method of Parkinson;27 because the surfaces are hydrophilic, the increase in dissolved gas should have no significant effect on our measurements. Immediately prior to force measurements, colloid probes and surfaces were cleaned using an RF water plasma system, (10 W, 45 s, 0.15 Torr) to clean and hydroxylate the surfaces.

Table 1. Properties of ALD Surfaces Investigated surface

substrate growth temperature (°C)

no. of cycles

thickness (nm)

A B

250 250

2200 280

82 5

C

80

1600

102

crystal structure

rms roughness (nm)

anatase none observed amorphous

8.2 0.24 0.47

■

RESULTS AND DISCUSSION Surface Characterization. It is well known that the properties of ALD films are strongly influenced by the deposition temperature employed.28,29 In the case of TiO2 films, amorphous films are formed at low temperatures, and higher temperatures favor crystalline films.30 In this study, we investigated three different TiO2 films. The first films (designated A) were formed from 2200 deposition cycles performed at 250 °C. Another set of films (designated B) was also produced at 250 °C using 280 deposition cycles, and a third set of films was produced at 80 °C using 1600 deposition cycles. Sample A was crystalline with a thickness of 82 nm (Supporting Information); however, the surface roughness at 8.2 nm rms, while better than that for the surfaces used in previous studies of TiO2, was still less than ideal. To minimize the roughness, thinner films were produced (B). The thinner films showed no sign of crystallinity and were considerably smoother with an rms roughness of