Functionalization of silicon oxide surfaces
Silicon is well known for the quality and the stability of its oxide. When properly grown silicon oxide is a stable dielectric that has very few defects and so is well suited for electronics application. We first review the different types of oxide encountered in this work and describe how these can be functionalized.
Different types of silicon oxide
Silicon dioxide can be encountered in many forms, crystalline or amorphous. In this thesis, we encounter primarily two types of amorphous SiO2. These differ in the way they were formed, which affects their electrical properties:
Pristine silicon surface in unstable in air: if not protected with a GOM or kept in Ultra High Vacuum (UHV), the surface oxidizes, forming a thin layer of SiO2 on the surface. Because it is a spontaneous reaction in an uncontrolled environment, the quality of this oxide is ill-defined. This type of oxide is typically ~2nm thick and not homogeneous. It may also have defects and contamination coming from the ambient air, which means that it is not suited for most electronic devices application18. Prior to the deposition of another material or the growth of another oxide, one often removes this “native” oxide first.
Thermal oxide is obtained by putting a cleaned sample in a furnace with a controlled atmosphere of oxygen (O2). This type of oxide is obtained in the UTD cleanroom and the equipment is operated by the cleanroom staff. The growth temperature is typically 900°C in a horizontal diffusion tube (see Figure 1.1). The oxidation of the silicon occurs at the interface of the silicon and the SiO2 through oxygen diffusion. High temperature is required to enhance the diffusion, as at room temperature the oxide cannot be thicker than 2nm. Thermally grown oxides, with a post-anneal in an inert gas, are of high quality with an excellent Si/SiO2 interface (low density of interface traps), compared to SiO2 deposited with other techniques like evaporation or sputtering. It can be used as a protective layer, as a dielectric or as a substrate.
Thermal silicon oxides are terminated with hydroxyl groups albeit with a variable density because of the amorphous nature of these oxides as shown in Figure 1.222. It is however possible to use a piranha cleaning solution (H2O2/H2SO4; 2/3; v/v) to eliminates hydrocarbons and metal ions contamination, and increase the density of hydroxyl groups on the SiO2 surface14,23,24.
Different types of molecules used
Glass or oxidized silicon are cheap flat substrates that can have excellent dielectric and optical properties For that reasons, silicon oxide functionalization is a natural topic that has been thoroughly studied for different applications like biosensors23 or solar cells development21. Since the pioneering work of Sagiv et al. in the 1980s25,26,27, two main types of molecules have been traditionally used. Silanization of organic chains with silane termination on SiO2 surface is the most common28,29 and it is the one used in this thesis. The attachment is made through the formation of a Si-O-Si bond between the silanol group and the surface. The reaction is performed in two steps: first there is the hydrolysis of the alkoxy group to form a hydroxysilane; this product is then covalently adsorbed on the surface30. The presence of water is needed for the first step to occur, but to avoid unwanted polymerization, it is best to perform the attachment in a water free environment and rely only on the water previously physisorbed on the hydrophilic sample2. A serious problem is the reproducibility of results because of the variable –OH termination and water concentration on the surface where the reaction takes place31,32. This is why usually a prior cleaning with a piranha solution is performed. To functionalize SiO2 surfaces, four silane molecules have been used in this work:
– 3-Aminopropyltriethoxysilane (APTES) a short molecule with an amine termination at the other end, used for nanoparticle attachment7,23.
– 11-aminoundecyltriethoxysilane (AUTES). This longer chain molecule typically leads to a better ordered monolayer and it is more stable as shown in Figure 1.323.
– 11-mercaptoundecyltrimethoxysilane (MUTS). This molecule has the same chain length as AUTES but with a thiol terminal group that has a special affinity with gold and ZnS and allows effective attachment of nanoparticles33,34.
– triethoxysilyl undecanal (C11-Ald). This is a similar molecule than AUTES, but with an aldehyde termination.
The experimental protocol is the same for all these molecules and is described in the beginning of CHAPTER 9.
Figure 1.3: Schematic figure highlighting the different configurations of APTES and AUTES attachment to the surface. While AUTES is mainly attached to the silicon oxide surface and can polymerize in a 2 dimensional direction, APTES molecules form a 3 dimensional attachment due to the 3-D polymerization. Such arrangement enables water molecules to reach the interface and can lead to (molecule) Si–O—Si–O (SiO2) bond breakage i.e. APTES layer removal. Seitz et al., J. Mater. Chem., 2011.
For completeness, phosphonate molecules need to be mentioned because, although they are not used in our work, they can be effectively grafted on silicon oxide. The grafting technique was developed by Hanson et al. in 20033 and recently optimized in Yves Chabal’s group by Vega et al. in 201235. They can be grafted on a SiO2 surface with a method called: “tethering by aggregation and growth” or T-BAG. Initially the molecules are physisorbed, forming an ordered array because of the interaction between phosphonate headgroups. A bond is then formed with the substrate after a long annealing (~2 days, 140°C). However, the GOM layer is not very stable because the Si-O-P bond is relatively weak and can be easily hydrolyzed especially in basic and neutral solutions.
Oxide-free silicon surface functionalization
Oxide-free silicon surface functionalization technology was developed in light of the previously mentioned issues regarding oxidized silicon. Most of the chemical routes involve the covalent attachment of molecules to hydrogen terminated Si surfaces. Comprehensive reviews have been done on this topic by Thissen et al. in 201215 and De Benedetti et al. in 201336. The following provides only a brief overview.
Oxide-free silicon surfaces
Hydrogen terminated silicon surfaces
Oxide-free Si functionalization takes advantage of the relatively easy chemical process (i.e. HF etching) that removes surface oxide and leads to hydrogen termination. Specifically, a aqueous solution of hydrofluoric acid (1-40%) efficiently removes the oxide. For a long time it was thought that the resulting surface was F-terminated37 because F was detected with x-ray photoelectron spectroscopy (XPS), but it has been shown using infrared spectroscopy in 1990 that it is H-terminated instead5. The etching mechanism is shown in Figure 1.438. Even if the strength of the Si-F (~5eV) bond is higher than the one of the Si-H bond (~3.5eV), the Si-F bond is highly polarized, weakening the underlying Si-Si bond (because of the eletronegativity of the neighboring bonded fluorine atom) and leading to the removal of the surface Si atom in the form of SiFX (mainly SiF4 or H2SiF6) molecules. The resulting surface is then H-terminated and is stable because the underlying Si-Si is less polarized. Figure 1.539 shows that the removal of the Si-F is only possible if the surface is rough (Figure 1.5a). If the surface is smooth, the Si-Si bond under the Si-F is protected because of the steric hindrance of neighboring atoms (Figure 1.5b).
Figure 1.4: Mechanism leading to the formation of H-terminated Si surface by HF etching: The last step of oxygen removal from SiO2 involves HF attack of the Si-O bond, with removal of OH as H2O and termination of the surface Si atom with fluorine. Further attack of the polarized Siδ–Siδ+ leads to H-termination. From Ubara et al. Solid State Communications (1984).
Figure 1.5: Mechanism of HF attack on (a) an atomically rough, partially F-terminated surface, and (b) an atomically flat, partially F-terminated surface. From Michalak et al. Nat Mater (2010).
Atomically smooth Si(111)surfaces
Using an HF solution removes the oxide and H-terminates the silicon surface, but the resulting oxide-free surface is atomically rough because the initial interface between the silicon and the native SiO2 is rough. An atomically rough surface is characterized by several hydrides: monohydride, dihydride and –rarely- trihydride5,4,40,41. With a basic solution of HF, however, it has been shown that an atomically flat Si(111) surface can be obtained by removing most of the defect on the (111) face because of preferential etching (through slow reoxidation/removal of defect Si atoms), leading to ideal monohydride termination5. Practically, 40% NH4F solution with a pH equal to 8 is optimum. Fourier Transform Infrared Spectroscopy (FTIR) is used characterize the treated sample, exhibiting a single stretch mode at 2083.7cm-1 with a very narrow Full width at half maximum (FWHM <1cm-1), characteristic of a high quality monohydride surface termination42. FTIR results are displayed in Figure 1.643.
Figure 1.6: FTIR Spectra and STM images of surfaces obtained by (a) low pH etching (HF, pH<3) leading to rough inhomogeneous surfaces with mono- and dihydride terminated silicon species, and (b) high pH etching (NH4F/HF buffer system, pH>8) leading to ideally H-terminated homogenous and atomically flat silicon surface. From M. A. Hines, Annu. Rev. Phys. Chem. (2003).
One of the main advantages of this process is that an ideal Si(111) hydrogen-terminated surface is a remarkably stable surface. It can remain stable indefinitely when kept in controlled atmosphere at room temperature. Controlled atmosphere means ultra-pure gases, including water vapor and oxygen (e.g. water or oxygen can only induce oxidation at high temperature, >300K44). Initial stages of oxidation are usually caused by free radicals and Ozone (O3) found in air (the concentration can vary from one place to another). Some initial surface degradation can occur right after exposition to air that can only detected by charge recombination techniques45. However, initial detectable oxidation (with FTIR or XPS, see CHAPTER 4) of an ideal surface only starts to appear after a few hours depending on the air composition.
Because of their remarkable stability, the atomically flat oxide-free silicon surfaces cannot easily be chemically modified. There have been several techniques that have been developed for that purpose, as describe in the section.
Linford et al. proposed in 1993 a novel method that enabled the hydrosilylation of a hydrogen terminated silicon surface. The product of such reaction was the covalent attachment of an alkyl group on a Si(111) surface though a Si-C bond6,46. This achievement offered surface scientists a way to protect a silicon surface with a stable molecular layer with a density close to that of crystalline hydrocarbons (~90%)6. This reaction can be done using a wide variety of activation: thermally17,24, catalytically47,48, photochemically8,49 or radically46. The reaction is as follow: — Eq 1.1
The radical R can be a long carbon chain terminated with an active group on top, as long as it does not affect the hydrosilylation itself16,17. Several processes, like the one involved in this thesis rely on the attachment of a protected group (like a protected carboxylic acid: an ester) which is then “deprotected” after the grafting on the surface7,16. Two techniques have been used in this thesis: thermal hydrosilylation and UV-activated hydrosilylation. The first one, which relies on elevated temperatures to start the hydrosilylation reaction, has been extensively used and reported in the literature mainly because of its relative ease of use. Contrary to lots of other techniques it does not require catalytic compounds that can oxidize and contaminate the surface. The experimental protocol is presented in CHAPTER 5. UV activated hydrosilylation is only slightly more complicated, involving the formation of a radical because the energy of the incoming light (~200-300nm so ~4-6eV) is higher than the energy of the Si-H bound (~3.5eV). The silyl radical created can then react with the alkene of the molecule inducing the formation of a Si-C bond. This technique has the advantage over the thermal one that it allows the attachment of smaller molecules that would evaporate during thermal hydrosililation. On the other hand, smaller molecules are prone to polymerization, so dilution, or introduction of inhibitors compounds is sometime needed. Chidsey et al. first reported the use of UV-initiated hydrosilylation to graft 1-pentene on a oxide free Si(111) surface in 199346. Possible mechanisms are shown in Figure 1.7 and the experimental protocol is shown in CHAPTER 6.
Table of contents :
0.1 General introduction
0.2 Oxide-free silicon functionalization for single electron transport
0.3 Oxide free silicon functionalization for NQDs based solar cells
CHAPTER 1 SILICON FUNCTIONALIZATION
1.2 Functionalization of silicon oxide surfaces
1.2.1 Different types of silicon oxide
1.2.2 Different types of molecules used
1.3 Oxide-free silicon surface functionalization
1.3.1 Oxide-free silicon surfaces
1.3.2 Functionalization methods
CHAPTER 2 GRAFTED ORGANIC MONOLAYER ON OXIDE-FREE SILICON FOR SINGLE ELECTRON TRANSPORT
2.2 Coulomb blockade and Coulomb staircase for single electron transistor
2.2.1 Orthodox theory
2.2.2 Single electron transistor (SET)
2.2.3 Experimental studies
2.3 Case of an electrode being a semiconductor
2.4 Specificity of using an organic dielectric
2.4.1 Electronic transport in molecules
2.4.2 Metal anchoring contact, the role of the GOM termination
2.4.3 Effective electron mass
2.4.4 How to more accurately model a molecular junction
CHAPTER 3 NANOQUANTUM DOTS ON SILICON FOR HYBRID SOLAR CELLS
3.1.1 Overview of solar cell technology
3.1.2 Hybrid nanoquantum dot technology
3.2 Properties of Nanoquantum dots
3.2.1 What is a Nanoquantum dots?
3.2.2 Energy transfer
3.3 Experimental progress toward NQDs based solar cells
3.3.1 Grafting of NQDs on planar surfaces
3.3.2 NQDs with silicon absorption
CHAPTER 4 EXPERIMENTAL TECHNIQUES
4.1 Fourier Transform Infrared Spectroscopy (FTIR)
4.2 X-ray Photoelectron Spectroscopy (XPS)
4.3 Scanning Tunneling Microscopy (STM)
4.3.2 Fabrication of STM tip
4.3.3 Scanning Tunneling Spectroscopy (STS)
4.3.4 Other modifications
4.4 Atomic Force Microscopy (AFM)
4.5 Scanning Electron Microscopy (SEM)
4.6 Photoluminescence spectroscopy (PL)
4.7 Transmission Electron Microscopy (TEM)
4.8 Other techniques
CHAPTER 5 GOLD NANOPARTICLES ON OXIDE-FREE SILICON−MOLECULE INTERFACE FOR SINGLE ELECTRON TRANSPORT
5.2 Experimental method
5.2.1 Preparation of Highly Ordered Monolayers on Si(111).
5.2.2 Attachment of gold nanoparticles
5.3.1 Characterization of the SAM
5.3.2 AuNP characterization
5.3.3 STM characterization of the assembled system
5.4 Results and discussion
5.4.1 STS measurements
5.4.2 Coulomb blockade
CHAPTER 6 CONTROLLING THE REPRODUCIBILITY OF COULOMB BLOCKADE PHENOMENA FOR GOLD NANOPARTICLES ON AN ORGANIC MONOLAYER/SILICON SYSTEM
6.2.1 Sample preparation
6.2.2 Characterization techniques
6.3.1 Nanoparticle size and shape
6.3.2 Crystallinity of nanoparticles
6.3.3 Nanoparticle coverage
6.3.4 Removal of ligand by annealing
6.3.5 Quality of Si/GOM interface:
CHAPTER 7 COULOMB BLOCKADE: CORRELATION BETWEEN EXPERIMENTAL AND SIMULATED DATA
7.2 IEF’s simulation software
7.2.2 Parameters used
7.2.3 Capacitances of the system
7.2.4 Current flow approximation
7.3 Additional consideration regarding simulations
7.3.1 HOMO-LUMO of the GOM
7.3.2 Band Bending
7.4 Experimental results analysis
7.4.1 Parasitic oscillations
7.4.2 Step width analysis
7.5 Simulation results
7.5.1 Example of simulation results using IEF’s software
7.5.2 Band bending calculation results
7.6 Experimental correlation with simulated data
7.6.1 Band bending correction
7.6.2 Nanoparticle size dependence
7.6.3 Future work for single electron simulation
7.7 Toward single electron transistor (SET)
7.7.1 Gold strip transistor
7.7.2 Silicon on insulator (SOI) based transistor
CHAPTER 8 SILICON PATTERNING: AN OPTIMIZATION PROCESS FOR HYBRID PHOTOVOLTAICS
8.2 Fabrication of silicon nanopillars
8.3 Nanosphere lithography
8.3.3 Plasma etching and metal deposition
8.4 Metal assisted etching
8.4.1 Basic principle
8.4.2 Selected recipe
8.4.3 Effect of H2O2 concentration
8.5 Achievable structures
8.5.1 Silicon nanopillars
8.5.2 Bent Silicon nanopillars
8.5.3 Silicon polydiameter pillars
8.6 Optimizing NRET in hybrid NQDs/silicon structures by controlled nanopillar architectures
8.6.3 Results and discussion
CHAPTER 9 DITHIOL AND DIAMINE BASED MULTILAYER CDSE NANOCRYSTAL QUANTUM DOTS FABRICATION
9.2 Experimental method
9.2.1 Surface preparation
9.2.2 Multilayer fabrication
9.3 Results and discussion:
9.3.1 Multilayer grafting
9.3.2 Effect of NQDs functionalization on PL measurement
9.3.3 Photoluminescence measurements of multilayers
9.3.4 Density of layers
CHAPTER 10 EFFICIENT DIRECTED ENERGY TRANSFER THROUGH SIZE-GRADIENT NANOCRYSTAL LAYERS INTO SILICON SUBSTRATES
10.2 Preparation method
10.3 Results and discussion
10.3.1 Bilayer on glass
10.3.2 Theoretical model
10.3.3 Bilayer on Silicon
GENERAL CONCLUSION AND PERSPECTIVE