Master Thesis
The selective functionalization of small focused-electron-beam-induced SiO x
deposits by self-assembly
Author:
Winand Slingenbergh s2039230
Supervisors:
Dr. W. F. van Dorp Prof. Dr.
J. De Hosson
A thesis submitted in fulfilment of the requirements for the degree of MSc Applied Physics
in the
Materials Science Department Zernike Istitute for Advanced Materials
June 2012
Abstract
Faculty of Mathematics and Natural Science Zernike Istitute for Advanced Materials
MSc Applied Physics
The selective functionalization of small focused-electron-beam-induced SiOx
deposits by self-assembly by Winand Slingenbergh s2039230
In this report I present my work on the selective functionalization of SiOx features. The SiOx is written by means of focused electron beam induced deposition (FEBID). Small SiOx patterns were written on gold, diamond like carbon (DLC) and Si wafers with a native SiOx layer and functionalized.
Surfaces with a native oxide are functionalized by means of either 3-aminopropyltriethoxysilane (APTES) or 3-aminopropyldimethylethoxysilane (APDMES) molecules, self assembled onto the surface from the liquid phase. The layers are functionalized attaching fluorescin- isothiocyanate (FITC), ATTO655 dye, Au nanoparticles or carboxyl-coated polystyrene spheres (PS). Selectivity was determined by observing fluorescence with optical mi- croscopy, and PS with a SEM.
Problems with pre-existing functionalization methods were identified and the method was adapted to allow for selective functionalization. The conditions necessary for perfect monolayer coverage of APTES are extremely difficult to achieve. lack of reproducibility is primarily due to cross polymerization. In order to determine a recipe that would allow for monolayer coverage, the APTES layer was labeled with PS and coverage obseerved.
Due to clustering of physisorbed PS each time a sample was removed from a solution, the washing of the excess PS spheres had to be carefully controlled. A new sample holder was constructed that allows the sample to be submerged between successive rinses to avoid clustering.
Using the new labeling technique, reducing the concentration of APTES, limiting water content during deposition and using fresh APTES, a regime was identified that would allow monolayer coverage. Attempts to selectively functionalize small deposits on gold still proved problematic due to the cross polymerization. In order to solve the polymer- ization issue, a switch was made from the tri-functional APTES to the mono-functional APDMES. The new surface modification reaction was found to not be self-limiting, most likely due to impurities. With the aim of limiting polymerization, the regime in which a monolayer is just formed was again identified.
The newly established recipe optimized for monolayer coverage was used to functionalize SiOx deposits with ATTO655 and polystyrene spheres on various substrates. Due to use of ultrasonication in the various washing procedures the SiOx deposits may detach, with the narrow pillars and deposits on gold being especially susceptible. Using the new APDMES recipe the PS spheres show no attachment to the SiOx deposits, presumably due to the morphology of the deposit, while deposits functionalized with ATTO655 show good and selective attachment when written on gold. To date, features with a lateral dimension as small as 24.4 nm have been successfully functionalized. From SEM images and optical images of samples functionalized with fluorophores, it is clear that the FEBID process not only deposits SiOx in the intended location but that the substrate surrounding the feature(s) is also covered in a functionalizable thin SiOx layer.
While this project has been a valuable opportunity to establish myself as a more inde- pendent researcher, the nature of science is collaborative and I would like to express my gratitude to all those who helped me bring the project to fruition:
Sanne de Boer - Introducing me to the labs and showing me the ropes Paul Zomer - Providing me with the EBL samples
Jochem Smit - SAM discussions and assistance operating the single molecule fluores- cent microscope
Thorben Cordes - Discussions about super-resolution microscopy application
Wesley Browne - SAM discussions, training and assistance operating fluorescent mi- croscope
Ben Feringa - SAM discussion
Robbie Roswanda - APTES discussions and Au nanoparticles Jort Robertus - Assistance obtaining NMR spectrum
Eric Detsi - Gold surface discussions Menno Eekma - AFM instruction
Willem Pier - Helping explain crystalline formations of polystyrene spheres Diego Martinez-Martinez DLC samples and discussions
YuTao Pei - Dual beam instruction
Mikhail Dutka - SEM instruction and assistance Madga Wojtaszek - O-plasma instruction Special thanks:
Willem van Dorp for his excellent guidance, expert advice and providing me a good insight into what it means to do professional research.
Jeff de Hosson for providing me with the opportunity to work on this project and experience working in a fantastic research group.
iii
Abstract i
Acknowledgements iii
List of Figures vii
Abbreviations ix
1 Introduction 1
1.1 Motivation . . . 1
1.1.1 Review of existing patterning techniques. . . 2
1.1.2 Lithographic . . . 4
1.1.3 Direct-write . . . 5
1.2 Critical assessment and reasons for choosing FEBID . . . 7
1.3 How FEBID works . . . 9
1.4 Theoretical model . . . 10
1.5 Small features . . . 11
1.6 Monolayer formation . . . 12
1.6.1 Mechanism . . . 13
1.6.2 Amine Catalysis . . . 15
1.6.3 Effect of water . . . 15
1.6.4 Density and coverage. . . 15
1.6.5 Comparing APDMES with APTES. . . 17
1.7 Functional molecules and particles . . . 18
1.7.1 Carboxyl-coated Polystyrene Spheres (PS). . . 19
1.7.2 Au nanoparticles . . . 20
1.7.3 Fluorescinisothiocyanate (FITC) . . . 20
1.7.4 ATTO655 red. . . 21
2 Experimental 22 2.1 Materials . . . 22
2.1.1 Solvents . . . 22
2.1.2 Particles, chemicals, compounds . . . 22
2.2 Substrates . . . 23
2.3 Patterning . . . 23
2.4 Sample handling . . . 25
iv
2.5 Surface activation. . . 26
2.5.1 Treatment with Piranha . . . 26
2.5.2 Treatment with O-plasma . . . 26
2.5.3 Untreated . . . 27
2.6 General procedure . . . 27
2.7 Silinization . . . 27
2.7.1 APTES deposition from ethanol . . . 28
2.7.2 Procedure APTES+FITC . . . 28
2.7.3 Deposition from anhydrous toluene . . . 28
2.7.4 APTES . . . 28
2.7.5 APDMES . . . 29
2.8 Attachment of functional molecules and particles . . . 29
2.8.1 FITC . . . 29
2.8.2 ATTO655 . . . 29
2.8.3 Au Nanoparticles . . . 29
2.8.4 Polystyrene Spheres (PS) . . . 30
2.9 Imaging . . . 30
2.9.1 Optical Microscope (Synthetic Chemistry) . . . 30
2.9.2 Optical Microscope (Single molecule microscopy) . . . 31
2.9.3 SEM . . . 31
3 Results 32 3.1 Writing SiOx Features . . . 32
3.1.1 Small Dots . . . 32
3.1.2 Lines. . . 33
3.1.3 Halo Deposition . . . 35
3.1.4 Resilience to Ultrasonication . . . 36
3.2 Functionalization with FITC . . . 38
3.2.1 Height Dependance. . . 41
3.2.2 Interference . . . 42
3.3 Functionalization with PS . . . 47
3.3.1 Optimization . . . 47
3.3.2 Selectivity to APTES . . . 48
3.4 Functionalization with Au nanoparticles . . . 54
3.4.1 Optimization . . . 54
3.4.2 Selectivity for APTES . . . 55
3.5 APTES optimization . . . 55
3.5.1 Lack of selectivity irrespective of substrate . . . 55
3.5.2 Possible reasons for lack of selectivity . . . 56
3.5.3 Concentration optimization for monolayer coverage . . . 57
3.6 APDMES . . . 59
3.6.1 Optimization . . . 60
3.6.2 Residues. . . 63
3.6.3 Confirming selectivity . . . 65
3.6.4 Smallest functionalized features . . . 70
4 Conclusions 72
5 Recommendations 74 5.1 Improvements . . . 74 5.2 Further investigation . . . 75
A Piranha solution safety information 77
B O-plasma procedure 79
C Optimizing APTES for monolayer coverage 82
Bibliography 84
1.1 Super-resolution microscopy . . . 2
1.2 DNA self assebly . . . 3
1.3 Lithography . . . 4
1.4 Nanografting . . . 6
1.5 Single molecule cut and paste . . . 6
1.6 Dip-pen Nanolithography . . . 7
1.7 Secondary electron emission . . . 10
1.8 Feature diameter . . . 12
1.9 Monolayer assembly . . . 14
1.10 Organosilane molecules. . . 14
1.11 Amine catalysis . . . 16
1.12 APTES attachment to suface . . . 17
1.13 APTES Polymerization . . . 17
1.14 Aminosilane occupyig two sites . . . 18
1.15 EDC/NHS reaction. . . 19
1.16 FITC . . . 20
1.17 Click chemistry . . . 21
1.18 NHS-ester linking to amine . . . 21
2.1 Electron microscope set-up . . . 24
2.2 Current vs PC number . . . 25
2.3 General outline of experimental procedure . . . 27
2.4 Schematic of Optical Microscope . . . 30
3.1 Dot on Si . . . 33
3.2 Dot on Au. . . 33
3.3 Lines on Si . . . 34
3.4 SiOx lines on Au . . . 34
3.5 narrowest SiOx line on Au . . . 35
3.6 SiOx squares on Si . . . 36
3.7 W and Pt squares on Si . . . 37
3.8 SiOx square on Au . . . 37
3.9 breaking of SiOx pillars . . . 38
3.10 SiOx pillars detach . . . 39
3.11 SiOx lines detach . . . 39
3.12 Luminosity dependance on deposit height . . . 41
3.13 Intensity vs. Height . . . 42
3.14 Interference in SiOx deposits . . . 42 vii
3.15 [Interference in EBL features . . . 43
3.17 Intensity vs. deposit height . . . 45
3.19 Interference effects on EBL edge . . . 47
3.20 PS selectivity to APTES. . . 48
3.21 Large PS clusters . . . 49
3.22 PS ring clustering. . . 50
3.23 Medium drying speed features. . . 51
3.24 fast drying speed features . . . 52
3.25 Effect of ultrasonication on PS attachment . . . 53
3.26 Attempt at removing physisorbed PS. . . 54
3.27 Custom teflon holder . . . 54
3.28 Au nanoparticle selectivity to APTES . . . 55
3.29 Lack of selectivity of APTES for SiOx . . . 56
3.30 Ageing effect on silinization of APTES solution . . . 57
3.31 Determining the APTES concentration for monolayer coverage . . . 58
3.32 Best results along edge of sample with optimized APTES procedure . . . 59
3.33 Best results in middle of sample with optimized APTES procedure . . . . 60
3.34 APDMES silinization not a self-limiting process. . . 61
3.35 PS coverage vs. time in APDMES with Piranha . . . 62
3.36 PS coverage vs. time in APDMES without Piranha. . . 62
3.37 APDMES agglomerate . . . 63
3.38 APDMES agglomerate at edge . . . 64
3.39 APDMES agglomeration around a crack . . . 65
3.40 small APDMES agglomerate . . . 66
3.41 larger APDMES agglomerate . . . 67
3.42 selectivity on DLC substrate . . . 68
3.43 ATTO655 functionalized SiOx lines on Si . . . 68
3.44 Photobleaching experiment results . . . 69
3.45 SiOx square on Au . . . 71
C.1 . . . 83
SAM Self Assembling Menolayer
DLC Diamond Like Carbon
FITC Fluorescin IsothioCyanate APTES Amino-propylttriethoxysilane APDMES Amino-propyldimethylethoxysilane PS Polystyrene Spheres
FEBID Focussed Electron Beam Induced Deposition DPN Dip-Pen Nanolithography
GIS Gas Injection System
SEM Scanning Electron Microscope EBL Electron Beam Lithography
ix
Introduction
1.1 Motivation
Top-down patterning has given, and continues to give us great control of various mate- rial properties allowing for the development of technologies such as integrated circuits and high speed processors. With considerable research being performed in the field of nanotechnology, the use of bottom-up self assembly techniques, resembling and inspired by processes commonly found in nature, is proving to be of value in producing a large variety of small scale structures. Combining the top-down and bottom-up techniques can offer increased control and variation in the types of nanoscale structures that may be fabricated and engineered. One way of combining the two methods is the function- alization of patterns through self-assembly.
Functionalization is the process of modifying a surface with the aim of giving it a partic- ular function. This could mean changing the surface groups or even attaching so called functional molecules or particles. The types of functional molecules and nanoparticles available today include molecular motors [1], molecular switches [2], uorescent molecules, and quantum dots [3] with many variants of each type and many more being developed.
Potential applications for these functional molecules and particles are vast and varied within the rapidly expanding field of nanotechnology and include molecular electronics, NEMS, biochips and super-resolution microscopy [4]. Most applications are contingent on the ability to precisely position these functional molecules in specific locations and/or in a desired pattern.
1
In the particular case of super-resolution microscopy, the field would greatly benefit from a nanoscale pattern that can be functionalized with various fluorophores for calibration purposes. Super-resolution microscopy is an optical microscopy technique that offers the possibility of resolving features beyond the diffraction limit, a hard limit determined by the wavelength of the light used to image it. The technique works by observing specially designed blinking fluorophores of known brightness and intensity distribution in a sensitive optical microscope. By collecting a time series of the blinking molecules, and using the known intensity distribution, the data can be used to recover the more precise location of the fluorophores and hence allow for super-resolution microscopy see (figure 1.1).
The resolution limit of this technique will depend on the type of fluorophore used, type of microscope set-up and buffer solution in which the imaging takes place. For this reason, a calibration sample with specially designed blinking fluorophores precisely placed on a sub-diffraction-limit scale would be useful to the further development of this technique.
Figure 1.1: Image detailing the process by which a resolution greater than the diffrac- tion limit can be reached by using blinking fluorophores.
1.1.1 Review of existing patterning techniques
There is a wide array of techniques available to create nanoscale patterns on a surface and some may be more suited for particular applications than others. High resolution,
precise positioning, pitch, speed of patterning and number of steps involved in the process are some of the factors that can determine the choice of technique.
Generally speaking, top-down techniques are employed for patterning and bottom up self-assembly is used for the functionalization of patterns. However, a pattern may also be created using bottom-up techniques. Such a pattern is created by a reversible reaction in which molecular units spontaneously assemble into ordered structures. An example of one such technique is known as DNA self-assembly.
Figure 1.2: Diagram showing a pattern that can be created via the DNA self assembly technique.
DNA origami uses the complementary interaction of the four base pairs that make up the structure of DNA [5]. It is possible to make strands of DNA with a predetermined sequence of bases. By carefully choosing the order of the bases and the length of the chain, complementary strands can assemble and fold into both two and three dimensional structures and patterns such as the one pictured in figure 1.2. The advantage of DNA origami is that it offers a method for parallel synthesis of small nanostructures at mild conditions.
While the bottom-up self assembly techniques are still in their infancy, difficult to control and nearly impossible to precisely position, top-down patterning is more often employed.
The top-down techniques tend to be categorized as either lithographic, where the pattern is transferred in a one step process via an etching step, or direct-write, where the pattern
is written directly onto the surface. A brief review of some of these techniques is given below.
1.1.2 Lithographic
Electron- beam or Photolithography
Figure 1.3: Diagram detailing standard lithography methods.
With electron-beam and photolithography a resist is deposited on the surface prior to being irradiated or exposed by an electron beam or light. In the case of photolithography this is done through a mask and in the case of electron lithography it is done via the scanning of an electron beam [6], figure1.3. Electron beam lithography has the advan- tage over photolithography in that it can increase the resolution beyond the diffraction limit of the light used for exposure. Subsequently either the exposed or unexposed re- gions are etched away, these are referred to as positive and negative resists respectively.
The main drawback of the technique is that before the pattern is transferred, it requires
the coverage of the entire substrate with resist which may result in unintended effects.
Furthermore, subsequent etching steps are quite aggressive and may result in damage to or destruction of other parts of the device.
Microcontact printing
Microcontact printing is known as soft-lithography technique whereby a PDMS stamp transfers an ink onto a surface through contact [7]. The ink used is usually a SAM chosen for its specific interaction with the surface on which it is placed. While this technique offers extremely high throughput there are some significant drawbacks; it is completely contingent on other patterning techniques to create a pattern from which the PDMS stamp may be cast. Additionally, stamps suffer from deformation and the transferred ink from edge effects which may be detrimental to certain applications.
1.1.3 Direct-write
Positive and negative e-beam writing
E-beam writing uses a focused electron beam to alter the tail groups of a previously deposited SAM. While being extremely similar to EBL, it differs in the lack of an etching step, and therefore can be referred to as a direct write technique. With positive e-beam writing the irradiated area is altered to become the active pattern while in negative e-beam writing the irradiated area is passivated [8].
Nanografting
With nanografting, a self-assembled monolayer (SAM) is deposited onto the substrate prior to patterning. Subsequently an AFM tip is used to scratch away the SAM to expose the substrate. All this is done while submerged in a solution of a different SAM resulting in the formation of a different type of monolayer on the exposed surface [9], figure 1.4. By careful selection of the tail groups in the two types of monolayer, the pattern can subsequently be functionalized for various applications.
Single molecule cut and paste
Figure 1.4: Schematic representation of the nanografting process. [9]
This technique uses an AFM tip to move functional units coupled to a DNA oligomer from a depot onto a target area into a particular pattern [10]. The removal from the de- pot and the position onto a substrate is done through specific interaction with a comple- mentary DNA strand bound to the AFM tip. The process is demonstrate schematically in figure 1.5 where A shows the removal from the depot and B shows the placement of the molecule in its target location. This technique allows for single molecule resolution but is very slow and not particularly suited to applications which require larger patterns.
Figure 1.5: Schematic showing the single molecule cut and paste technique. [10]
Nano-oxidation
Nano-oxidation by AFM is an effective technique for writing small features and requires no mask or photoresist. By using an AFM tip as a cathode a water meniscus between the tip and surface can serve as the electrolyte. The strong localization of the electrical field lines near the tip apex can give rise to nanometer-size oxide features. However, it is a relatively slow process and the material that makes up the deposits is limited to oxides [11].
Dip-Pen Nanolithography
Figure 1.6: Schematic showing the single molecule cut and paste technique.
Dip pen nanolithography is a direct-write process whereby an AFM tip is used to diffuse SAM molecules onto a surface via a meniscus [12] as is schematically represented in figure1.6. The width of the written pattern can be determined by controlling the shape of the meniscus through manipulation of the tip. DPN is capable of high resolutions and has the potential to be a massively parallel technique if multiple tips are used to write the pattern on a flat surface.
1.2 Critical assessment and reasons for choosing FEBID
The patterning techniques described in the previous section are all viable alternatives to patterning by focused electron beam induced deposition (FEBID) and either allow direct functionalization or with one additional step.
Though the lithographic techniques offer high throughput, they are not particularly suited for prototyping as a new mask or stamp needs to be created for a new pattern.
Another drawback, shared by the lithography techniques as well as some direct-write techniques, isnecessary coverage with a resist layer or SAM prior to patterning. If the desired pattern is part of a larger device the resist or the etching steps could influence the
integrity of other features and affect the functioning of the device. For this reason pref- erence is given to the resist-less direct write techniques when more complex applications are to be considered.
With the exception of e-beam writing, the direct-write techniques detailed in the ”Direct- write” section do not suffer from this problem. Thus, what determines the relative strength of a direct-write technique are the maximum resolution, the types of patterns that can be written (dimensionality and materials) and the ability to pattern different substrates. FEBID is a direct write technique requiring no additional treatment and can be performed using equipment already widely available in many research institutions, namely a SEM fitted with a gas injection system. Because of this, FEBID is a relatively simple and cheap way for rapid patterning and prototyping. Features can be written on virtually any solid surface, in a variety of materials (W, Pt, Fe, SiOx, Co [13] and in three spatial dimensions; allowing for the design of complex structures on the nanoscale. Using a SEM features as small as 2.9 nm (full-width at half maximum) have been reported [14].
While for the application of super resolution microscopy calibration samples other pat- terning techniques, such as nanografting, are perfectly suitable, FEBID fills a niche due to ability to reach sub 10 nm dimensions, pattern in three dimensions, write on any substrate in a prespecified location and in a wide range of materials.
Name Maximum Resolu- tion
Pitch Arbitrary Location
Arbitrary Pattern
Assesment
Nanografting 10nm [15] Yes Yes - 2D Requires SAM before
pattern can be written E-beam writ-
ing or EBL
15nm [6] Yes Yes - 2D Requires resist before
pattern can be written Single
Molecule cut and paste
Single Molecule [10]
50nm Yes Yes - 2D Extremely time con-
suming technique
DPN 15 nm [16] Yes Yes - 2D Excellent technique for
2D patterns
DNA self-
assembly
6.6 nm [5] NA No Yes Very high resolution but
not arbitrary location Microcontact
printing
2 nm [7] Depends Yes Dependant on other
patterning techniques to create stamp
Nano- Oxidation
10 nm [11] Yes Yes - 2D structures are limited to oxides
FEBID 2.9 nm [14] Yes Yes High resolution, no
SAM or resist needed, selectivity determined by materials used
1.3 How FEBID works
In FEBID, features are written when an adsorbed precursor molecule is dissociated by a focused electron beam, leaving a non-volatile deposit on the surface. The process can be done in a scanning electron microscope fitted with a system for letting in a precursor gas;
this can be achieved either by filling the entire chamber or by injecting the gas directly onto the substrate through a nozzle. The precursor gas dissociates due to energetic electrons but not exclusively due to those incident directly onto the surface (primary electrons). Due to inelastic scattering, these primary electrons also cause the emission
of secondary electrons from a volume below the surface (see figure 1.7) and are know to contribute heavily to the dissociation process [17]. Assuming sufficient gas flux, high growth rates are achieved by maximizing the emission of secondary electrons from the surface and small features are achieved by having a well focused beam.
Figure 1.7: Schematic illustrating the effect of acceleration voltage and substrate on the emission volume for secondary electrons
1.4 Theoretical model
A relatively simple model to describe the process has been proposed [13].
dN dt = gF
1 − N
N0
−σ(E)N J − N τ
where N is the number of molecules on the surface, N0is the total number of surface sites available, g is the sticking factor, τ is the residence time of the precursor gas molecules on the surface, J is the current density of the incoming electron beam and σ(E) is known as the dissociation cross section, effectively the probability that a molecule will dissociate as a function of electon energy. The steady state solution (dN
dt = 0) is given by:
N = N0
gF N0
gF N0
+ σJ + 1 τ
Where σ is the integral value of σ(E). From this, the steady state growth rate is determined to be:
R = VM oleculeN σJ = VM oleculeN0
gF N0
gF N0
+ σJ + 1 τ
σJ
Where VM olecule is the volume of a precursor molecule, and ignoring desorbtion, this allows us to identify two distinct regimes:
gF N0
> σJ, R = VM oleculeN0σJ gF
N0 < σJ, R = VM oleculegF
These regimes are known as electron limited and flux limited respectively. To ensure reproducibility, a constant growth rate is desired. As the gas flux is not know, it is easier to manipulate the number of incident electrons and thus guarantee being in the electron limited regime. This is achieved by having low dwell times and multiple successive scans, allowing time for the precursor gas to replete between successive doses. While this translates to longer writing times, it is worthwhile due to better reproducibility.
1.5 Small features
Features can be written either by keeping the beam stationary or by sweeping it across a surface. In order to write the smallest possible features a stationary beam must be used. Assuming sufficient flux, material will be deposited anywhere where secondary electrons reach the surface. If the incident electron beam is stationary and normal to the surface this consists of a circular area, with the highest probability of secondary electron emission being at the centre. Keeping the beam stationary leads to the formation of a conical pillar whose radial dimension grows till the radius of the pillar base approaches
Figure 1.8: Tip diameter as a function of exposure time. (a) is where deposition starts to occur. (b) indicates when the tip diameter is comparable with the inelastic mean free path of the secondary electrons, at which point the diameter will increase
very little until is reaches its maximum diameter at (c)
that of the inelastic mean free path of the secondary electrons (b) in figure1.8. At this stage the pillar diameter reaches a maximum and only the height increases.
As should be evident from figure 1.8, the key to writing the smallest possible features is by having a dwell time as close as possible to (a) but sufficiently long so as to have a reasonable (detectable or functionalizable) size deposit. A common and well documented manner of affixing functional molecules to a surface is by means of a self-assembling monolayer (SAM). It is in the combination of a top down patterning technique with the bottom up monolayer self assembly that bring potential applications within reach.
Each functional molecule, self assembling monolayer, patterning technique and substrate will have advantages and disadvantages highly dependant on the application. The aim of this project is to develop a method for selectively attaching functional molecules by means of a self assembling monolayer to a pattern written by means of focused electron beam induced deposition (FEBID).
1.6 Monolayer formation
Self-assembling monolayers are ordered molecular assemblies formed by the adsorbtion of an active surfactant onto a solid surface. The formation of such thin crystalline films is
a commonly employed technique for altering the surface properties of a material leading to many research and fabrication applications, such as: altering wetting propreties, labeling [3] and attachment of functional molecules [1], protein or DNA immobilization [18], biological activation/passivation [19] and electrical insulation [20]. Generally self assembling molecules will consist of a head group chosen for its affinity for a specific surface and a tail group that will allows for an alternative functionality. The most commonly used SAMs are the organothiolates on noble metals and organosilanes on quartz, glass and metal oxides [21]. This report will focus on the functionalition of silicon oxide surfaces with amine-teminated alkoxysilanes. There exists great variation in the possible types of alkoxysilanes with the main varying parameters being the amount of head groups that can interact with the surface or each other (polymerization), the length of the alkoxy chain and the type of head group. The number of functional groups on the head of the molecule is an important factor determining film formation and stability.
1.6.1 Mechanism
Monolayer formation is presumed to be a two step process whereby the molecules that are to form the layer are first adsorbed onto the surface before diffusing rearranging themselves more vertically into a closely packed structure.[21] This growth mechanism leads to the formation of islands which grow and fuse as more molecules diffuse along the surface to the more densely packed regions as seen in figure1.9.
We used organosilanes deposited exclusively from the solution phase, however, vapor phase deposition is also possible [22]. Where surface hydroxyl groups are present on the surface, the organosilane molecules attach covalently by forming Si-O-Si bonds.
The most common head groups in organosilanes that react with hydroxyl groups are ethoxy, methoxy and Cl as shown in figure1.10, named ethoxy- methoxy and chlorosilanes respectively. In addition to variation in the type of reactive groups, the number of reactive groups per molecule is also an important factor influencing film formation, and can vary from one to three, identified as mono- figure1.10(c) and trifunctional figure1.10 (d) respectively.
Depending on the molecule used, attachment and growth mechanisms will vary; one of the more complex is that of APTES [23–25]. In the particular case of APTES
(3-aminopropyl-triethoxysilane), detailing its growth mechanism provides a good in- troduction to the types of SAM used in this project, namely APTES and APDMES (3-aminopropyl-dimethyl-ethoxysilane).
Figure 1.9: Diagram illustrating the process of monolayer assembly as a two step process where the molecules are first adsorb and then rearrange themselves into an
ordered structure. [21]
Figure 1.10: Schematic representation of some different types of organosilanes; (a) trifunctional ethoxysilane, (b) trifunctional methoxysilane, (c) monofunctional chlorosi-
lane, (d) trifunctional chlorosilane
1.6.2 Amine Catalysis
APTES is a popular silinizing agent for modifying the surface of silica based materials to be amine terminated. The amine tail group can then be used to attach other molecules and particles in order to functionalize the surface. It is possible for the amine of one APTES molecule to bind to the Si of another, forming an intermediate that is highly reactive with a nucleophile such as a hydroxyl group, figure1.11(a). If the chain length is sufficiently long (>4 carbon atoms), as is the case with APTES, it is possible for the amine to bind with the Si of the same molecule and therefore act as its own catalyst figure 1.11 (b) [26]. Without the presence of an amine to catalyze the reaction, bond formation may still occur but requires water.
1.6.3 Effect of water
In most solutions a hydroxyl terminated surface acts as a dessicator and a monolayer of water is adsorbed or H-bonded to the surface. The presence of water on the surface causes the hydrolysis of the APTES followed by a condensation reaction resulting in covalent attachment to the surface, the reaction is detailed in figure1.12.
The hydrolysis of APTES does not happen exclusively at the surface and can also occur in solution. The hydrolysis in solution is the driver of the polymerization process as detailed in figure 1.13. After polymerization the APTES may attach to the surface via one of the remaining functional groups, so deposition is not necessarily molecule by molecule but may occur as larger plaques or agglomerates.
Research into the stability of different silane layers, including APTES, has shown that exposure to water results in a delamination of the film [22]. So while the presence of water can help with the attachment of organosilanes to surfaces, it can also remove the film.
1.6.4 Density and coverage
The possibility of plaques and agglomerates being deposited, as well the polymerization of already deposited APTES can result in changes to the deposition rate as well as the deposited silane layer to be something other than a monolayer. A roughening or
Figure 1.11: Illustration showing the way an amine tail group can catalyze the cova- lent bonding to the surface (a) as a result of the interaction between two alkoxysilanes,
(b) as a result of pentacoordinate intermediate being formed [26]
Figure 1.12: Schematic representation of the hydrolysis and condensation reactions that result in the covalent attachment of an APTES molecule to a hydroxyl terminated
surface of SiO2 [24]
Figure 1.13: Schematic representation of the polymerization reactions that result in the covalent attachment of an APTES molecules to each other in solution [24]
polymerization of the film may affect the amine density on the surface and in turn affect the extent to which the surface can be functionalized. Aside from these considerations, the amine present in silanes such as APTES and APDMES may H-bond to surface hydroxyl groups. This means that each molecule can occupy up to two adjacent hydroxyl sites as is shown in figure 1.14. As a consequence, the H-bonded amine is not available for functionalization and the number of sites available for further deposition is reduced.
If the amine density needs to be increased, preadsorbtion of ethylenediamine (EDA) can catalyze the formation of siloxane bonds while simultaneously preventing the amine on the silane to H-bond to the hydroxyl extending from the surface citepEDA.
1.6.5 Comparing APDMES with APTES
APDMES and APTES are the mono- and trifunctional variants of aminpropyl-ethoxysilanes.
They share many ways in which they can attach to a hydroxyl terminated surface,
Figure 1.14: Schematic of a molecule occupying two suface hyroxyl sites
namely: H-bonding of the amine, H-bonding after hydrolysis and covalent attachment through siloxane bond formation. The main differences arise from the polymerization of APTES, something that does not occur in a similar fashion with APDMES. APTES has the ability to polymerize and form a covalently cross-linked network, thus eliminating the necessity for all molecules to be bound to the substrate. In contrast, if the mono- functional APDMES polymerizes, it forms a dimer which then has no reactive head groups left to attach to the surface. Also, instead of forming a covalent cross-linked network, APDMES film stability largely depends on van der Waals (VDW) interactions between adjacent molecules. Thus, denser films will be more stable as molecules are stacked vertically on the substrate, maximizing the inter-molecule interactions.
Due to amine - Si interactions between molecules, the formation of agglomerates in solution cannot be avoided, however, it is only with APTES that these agglomerates can consist partially of covalent networks. As a consequence of this, the agglomerates formed in APTES cannot easily beremoved and thus the APTES solutions can be said to age as agglomerates and plaques become larger and larger with time.
1.7 Functional molecules and particles
The SAMs used in this project change the surface to become amine terminated. It is this amine group that is taken advantage of in the functionalization process. Over the course of the project two types of particles and two fluorescent molecules were used.
1.7.1 Carboxyl-coated Polystyrene Spheres (PS)
The carboxyl-coated spheres used in this project have diameter of 120 nm and are used primarily due to their visibility in a SEM. The attachment of carboxyl coated particles to an amine terminated surface is already well established [3]. It takes place by the carboxyl reacting with EDC (a zero-length cross linker) to form an intermediate ester and subsequently reacting with the amine to form an amide bond. The reaction is detailed in figure1.15. While EDC is capable of forming amide bonds (top pathway), the reaction is not very efficient due to the tendency of the intermediate ester to hydrolyze (middle pathway). By introducing NHS, the intermediate ester, often referred to as a succinate ester, is stabilized somewhat and the amide bond formation is sped up due to the reduced rate of hydrolysis. Given sufficient time the PS can attach covalently to an amine terminated surface in the presence of EDC/NHS, however, NHS esters hydrolyze rapidly with a half-life of 4-5 hours at a pH 7 [27].
Lacking the presence of EDC/NHS, or in the case of complete hydrolysis the amine and carboxyl groups can still interact by forming an electrostatic bond. Hydrolysis can be minimized by using a Sulfo-NHS instead of the NHS as the extra charge leads to an intermediate even more stable to hydrolysis.
Figure 1.15: Schematic representation of the EDC/NHS assisted linking of carboxyl and amide
1.7.2 Au nanoparticles
The gold nanoparticles differed in size but have a mean diameter of 15 nm. Particles of this size are both visible in a SEM and fluorescence microscope. The fluorescence of the Au nanoparticles arises from quantum confinement, with the wavelength of the emitted light dependent on the size of the particles. Additionally, they are not subject to photobleaching.
1.7.3 Fluorescinisothiocyanate (FITC)
Figure 1.16: FITC molecule
FITC, figure1.16is a fluorophore commonly used as a label in various biological imaging applications as it can bind to amine groups under mild conditions. Its absorption peak lies at 495nm and emits it emits light of wavelength 521 nm. Bonding to amine groups is done via the isothiocyanate group and the reaction, detailed in figure 1.17, proceeds with such high efficiency that it is commonly referred to as click chemistry. FITC is also known to be subject to quenching, caused by non-radiative energy transfer between nearby molecules. To prevent this, molecules must be separated by a distance of at least 100 angstrom.
Figure 1.17: Schematic representation of the linking between amine and isothio- cyanate
1.7.4 ATTO655 red
Atto655 is a fluorophore manufactured by ATTO-TEC GmbH in Siegen, Germany. The structure is currently unknown but it is sold as an ATTO655 NHS-ester and so will couple to an amine in a similar way as to the carboxyl-coated PS. An example of the attachment is shown in figure1.18 for the example of a bodipy fluorophore.
Figure 1.18: Schematic representation of a bodipy fluorophore linking to an amine via an NHS-ester
Experimental
2.1 Materials
2.1.1 Solvents
• Mili-Q water prepared in house
• Ethanol from Emsure (CAS: 64-17-5)
• Acetone form Lab-Scan (CAS: 67-64-1)
• Anhydrous toluene (CAS: 108-88-3)
2.1.2 Particles, chemicals, compounds
• H2SO4 from Emsure (CAS: 7664-93-9)
• H2O2 from Sigma-Aldrich (CAS: 7722-84-1)
• 2, 4, 6, 8, 10-pentamethylcyclopentasiloxane (SiOx precursor)
• trimethylplatinum (IV)methylcyclopentane (Pt precursor)
• tungsten hexacarbonyl (W precursor)
• APTES from Sigma-Aldrich (CAS: 919-30-2)
• APDMES from Acros-Organics (CAS: 18306-79-1) 22
• 120 nm diameter carboxyl-coated polystyrene spheres from Bangs Labs
• Colloidal Au Nanoparticles produced via the Frens method, average diameter 15 nm suspended in water. (Obtained from Robbie Roswanda, unknown concentra- tion)
• FITC from Acros-Organics (CAS: 3326-32-7)
• ATTO655 from ATTO-TEC GmbH (CAS: 485815-43-8)
• EDC (CAS: 1892-57-5)
• NHS from Fluka (CAS: 6066-82-6)
2.2 Substrates
• Si wafer from SQI with native oxide
• SiOx patterns deposited on Si wafer by means of EBL (from Paul Zomer)
• 40 nm Plasma enhanced vapor deposited Au on Si
• Physical vapour deposited Au on Si
• 100 nm Gold coated silicon wafer with Ti layer to promote adhesion from Sigma- Aldrich
• Au TEM grid from Agar Scientific
• 750 nm PVD deposited DLC on Si wafer
2.3 Patterning
Patterning is done in the TESCAN Lyra FIB/FEG which is fitted with a multi- gas- injection system (GIS). A schematic of the set-up can be seen in figure 2.1.
The precursor gas is heated in a containment vessel and subsequently introduced into the vacuum chamber through a nozzle with a working distance around 9 mm directed at the sample at a working distance anywhere from 9.1 to 10.4 mm. The precursor gas used is 2, 4, 6, 8, 10-pentamethylcyclopentasiloxane and requires an outgassing procedure
Figure 2.1: Schematic representation of the SEM used for patterning
prior to use. The pattern is written using TESCAN s proprietary patterning software without the use of a beam blanker and allows for the dwell time and number of scans to be specied.
Patterns were written on pieces of silicon wafer with typical dimensions 0.5 x 1 cm.
These were either untreated, or covered with another material, namely, DLC or Au.
With the exception of the DLC samples, the chamber, sample holder and to be written substrates are cleaned with an air plasma for 16 to 18 hours by an XEI Scientic Evactron 25 De-Contaminator prior to deposition.
The electron beam parameters used for deposition vary greatly on what type of deposit was required. A basic strategy of multiple scans and short dwell times (0.8-1.4 µs) was used to ensure that writing was done in the electron limited regime. This way results are not dependant on gas flux, which is challenging to measure and possibly quite variable.
Typically, low beam energies (10kV) were used to deposit large areas of SiOxdue to the
high deposition rates, and high beam energies (30kV) used for writing small features due to the reduced spot size. By increasing the PC number a smaller portion of the emitted electrons is focused onto the surface. This results simultaneously in a lower probe current and a smaller spot size. However, the spot size reaches a minimum for reasons currently not yet well understood and is dependant on acceleration voltage and working distance. For an acceleration voltage of 30kV, the minimum spot size is achieved at PC number 9 or 10 with giving a spot size of 3.45 nm as calculated by the TESCAN software.
Figure 2.2: Beam current measured in a faraday cup plotted against PC nuber as given by software
The probe current was not always measured directly but presented little variation in time or with changes in other parameters. In the rest of this report the beam current will be indicated with PC number unless explicitly measured. Figure 2.2shows a chart for approximate conversions.
2.4 Sample handling
Dry samples were always handled with clean tweezers and wet samples we handled using a custom made Teflon holder that allowed for easy transfer as well as draining of excess fluid. Treatment in liquids was done in new 20 ml scintillation vials as received. Surface
activation was done in approximately 4 ml; deposition was performed in a solution of volume of 2-10 ml and rinses in a volume of 10-18 ml unless otherwise specified.
2.5 Surface activation
Prior to the formation of a monolayer the surface of the deposit may be treated so as to have a reproducibly functionalizable surface. In the case of SiOx deposits being functionalized by APTES and APDMES this means the conversion of the surface oxides to be- OH terminated. This process is more generally referred to as surface activation.
Surface activation of the SiOxdeposits can be performed by either submersion in Piranha solution or with oxygen plasma.
2.5.1 Treatment with Piranha
Piranha solution is a 7:3 mixture of sulfuric acid (H2SO4) to hydrogen peroxide (H2O2) and is so named due its ability to aggressively dissolve organic species, somewhat resem- bling a piranha feeding frenzy. Because of its aggressive nature, it should be handled with extreme caution. A guide for making and handling the solution is given in Ap- pendix A. Samples were immersed in a 95 degrees Celsius solution for 30-45 minutes, then rinsed thrice with mili-Q water to remove any traces of Piranha. The sample was then rinsed thrice in ethanol and dried under a stream of nitrogen.
2.5.2 Treatment with O-plasma
Surface activation using oxygen plasma was done in the clean room of the Physics of Nanodevices group using standard equipment. Operating procedures can be found in the appendix. Samples were exposed to the plasma for no more than 3 min. If the samples were to be compared with those prepared in piranha solution, samples were also rinsed in mili-Q water and thrice in ethanol before being dried under a stream of nitrogen. An operation manual for the O-plasma can be found in Appendix B.
2.5.3 Untreated
For some experiments the sample surfaces were not activated, in this case the samples were first placed in acetone in an ultrasonic bath for 5 minutes to clean the surfaces.
Samples were then rinsed thrice in ethanol with a 5 min ultrasonic bath on the second rinse to remove remaining traces of acetone. They were then dried under a stream of nitrogen before undergoing any further processing.
2.6 General procedure
Figure 2.3
The basic procedure for functionalization is outlined schematically in figure 2.3. With the exception of the nitrogen drying, there exist variations of each step. The variants will be described below but discussion of the reasons for the choices will be left for the results and discussion section of this report. In the case of functionalization with FITC and APTES, the two were added together prior to deposition a this procedure was well established and has already been optimized for maximum fluorescence intensity in our imaging set-up [28]. This procedure is detailed below as procedure APTES+FITC.
2.7 Silinization
The formation of a monolayer of silane on surfaces is commonly referred to as silaniza- tion. In our experiments we used APTES and APDMES as the silanating agents, and deposition was always done from solution. As the silanization process was one of the main processes under investigation, many of the parameters influencing monolayer for- mation were studied. Throughout the project the effect of the age of the silane solution, concentrations of silane in solution, time in solution, temperature of solution and solvent used for deposition were all at some point varied. The procedures below will detail the different silanization procedures and corresponding rinses.
2.7.1 APTES deposition from ethanol
The sample is first placed in a 0.5mM solution of APTES in ethanol at room temperature for 1h then rinsed thrice in ethanol with 10s in an ultrasonic bath on the second rinse.
The sample dried under a stream of nitrogen before being functionalized.
2.7.2 Procedure APTES+FITC
In procedure APTES+FITC, the FITC is added to a 0.5mM solution of APTES in ethanol at a molecular ratio of 1:50 of FITC to APTES, sonicated to dissolve the FITC, and stirred overnight to allow the FITC to attach to the APTES. The sample is then placed in this solution at room temperature for 1h and subsequently rinsed thrice in ethanol with 10 seconds in an ultrasonic bath on the second rinse prior to being dried in a stream of nitrogen.
2.7.3 Deposition from anhydrous toluene
In the case of deposition from anhydrous toluene special care was taken to ensure the solution contained as little water as possible, all glassware was heated to at least 100 de- grees Celsius and deposition solutions were stored and used under nitrogen atmosphere.
All toluene based solutions were freshly prepared on the day of use.
2.7.4 APTES
Concentrations of APTES in dry toluene varied from 0.05 to 0.0005 mM. Additionally, two different types of APTES were used; one fresh and stored under nitrogen atmosphere and one 4 years old which had been exposed to a moisture rich environment. Samples were placed in solution at room temperature for 30 min and then rinsed thrice in ethanol with 10s in an ultrasonic bath on the second rinse. The samples were dried under a stream of nitrogen before being functionalized.
2.7.5 APDMES
Deposition of APDMES from anhydrous toluene was done for times varying from 1 minute to 20 hours in a 1% (v/v) solution pre-heated to between 65-70 degrees Celsius.
Samples were first placed under nitrogen atmosphere before the solution was added.
Samples were then rinsed thrice in toluene with 5 min in an ultrasonic bath on the second rinse and then rinsed thrice in ethanol, again with 5 min in an ultrasonic bath on the second rinse. Drying was done in a nitrogen stream.
2.8 Attachment of functional molecules and particles
2.8.1 FITC
Functionalization with FITC was achieved by placing the samples in a 0.01mM solution of FITC in ethanol (ultrasonicated so ensure it was properly dissolved) for 15 minutes at room temperature. Samples were then rinsed thrice in ethanol with 10 seconds in an ultrasonic bath on the second rinse prior to being dried in a stream of nitrogen.
2.8.2 ATTO655
Functionalization was done in a 1µM solution of ATTO655 in mili-Q water with a N aHCO3 buffer giving the solution a pH of 8.2. The samples were placed in 5 ml of this solution at room temperature for 1 minute and thne rinsed thrice in mili-Q water with 10s in an ultrasonic bath on each rinse. The sample was then rinsed once in ethanol to allow for quick drying in a stream of compressed air.
2.8.3 Au Nanoparticles
Samples were placed into the solution of Au nanoparticles as for 1 hour at room tem- perature and rinsed once with mili-Q water and dried under a nitrogen stream.
2.8.4 Polystyrene Spheres (PS)
In this project the method of labeling the silane layers with PS was found through much experimentation. The basic procedure consisted of placing the sample in a solution of PS in mili-Q water for a 15-45 minutes and then rinsing with water. Different solutions and rinses were used and are detailed in the results and discussion section of this report.
2.9 Imaging
2.9.1 Optical Microscope (Synthetic Chemistry)
Figure 2.4
Samples functionalized with FITC were analyzed by measuring fluorescence with an inverted optical microscope fitted with a CCD to obtain digital images. A schematic of the set-up is shown in figure2.4. The microscope used is a Nikon Eclipse Ti-E fitted with a 488 nm LED laser to illuminate and excite the FITC molecules. A band pass filter removes reflected light while allowing the transmission of 521 nm light emitted by the FITC. In addition to being able to adjust the laser intensity, exposure time and frames to be averaged, the CCD is capable of electron multiplier (EM) gain. Typical images
were taken with laser intensity at 100%, maximum exposure time and averaged over 16 frames. If images were still not bright enough, EM gain was used but was avoided as much as possible as to avoid the introduction of unwanted noise.
2.9.2 Optical Microscope (Single molecule microscopy)
Samples functionalized with ATTO655 were observed using a microscope very similar to that used for FITC. However the setup was specially made for single molecule microscopy and is extremely sensitive. The ATTO655 molecules were excited with a 643 nm laser with the power set to 8mW. Imaging was done with an exposure time of 100ms and an EM gain ranging between 100 and 220. Images were averaged over a minimum of 100 frames.
2.9.3 SEM
Three different SEMs were used for imaging depending on availability. The microscopes used were the Phillips XL30 ESEM, Phillips XL30 SEM+OIM and the Phillips XL30S SEM. Images are all taken with the SE detector with the exception of images from the XL30S which were taken using its ultra high-resolution (UHR) through-lens detector (TLD).
Results
3.1 Writing SiOx Features
While looking into the minimum lateral dimension that could be obtained with our experimental set-up both SiOx dots and lines were written. Having established that writing with 30kV and PC number 9 will yield the smallest spot size (typically 3.33-3.50 nm depending on the working distance) and writing in parallel with a low iterative dwell time (1µs) will keep us in the electron limited regime, the only parameter to be varied was the number of scans. Sizes of the features were measured from SEM images. Total exposure can easily be calculated as number of scans x dwell time, and electron dose can be calculated by further multiplication by the beam current.
3.1.1 Small Dots
To date the smallest dots have been obtained with a spot size of 3.33 nm and a dwell time of 1 µs. Reproducibility started becoming an issue once the number of scans was reduced to below 20,000 and contrast was not sufficient to reliably distinguish the features from the background. At 20,000 scans dots with a diameter of 25 nm were written, an SEM image of which is shown in figure 3.1. While no express effort has been made to limit the lateral dimension of dots written on a gold substrate, dots with a diameter of 60 nm can easily be written with 300,000 scans. See figure 3.2.
32
Figure 3.1: SEM image of a 25 nm dot of SiOx on Si substrate
Figure 3.2: SEM image of a 60 nm dot of SiOxon Au substrate
3.1.2 Lines
Lines were used as an alternative to dots for resolution analysis as it allowed for more time between successive scans for precursor gas to replete and fluctuations in gas flux are averaged over a larger distance allowing for better reproducibility. Additionally, it makes features easier to find and measure. Efforts were made to limit both the lateral dimension of the lines and pitch. Lines written with 10,000 scans provided good contrast with the Si wafer and have a width of 46 nm which give a good indication of how far
Figure 3.3: SEM image of 64nm wide SiOx lines on a Si substrate with pitch decreas- ing from 200 to 60 nm in 20 nm increments
the pitch can be reduced. As is evident from the results shown in figure 3.3, lines can still be distinguished with a pitch of 60 nm with no obvious proximity effects observed, suggesting that the pitch can be further reduced with narrower lines.
Figure 3.4: SEM image of SiOx lines on a Au substrate written with 4000, 3000, 2000 an 1000 scans with a dwell time of 1µs illustrating the difficulty in distinguishing
progressively smaller features
Narrower lines can be written by limiting the number of scans. Unfortunately, with less material deposited, the contrast with the substrate is lower and it is more difficult to distinguish features from the background. An example of this on a gold substrate is shown in figure3.4where the line made up of 1,000 scans is barely distinguishable from the gold surface. The narrowest line written to date was made up of 1,000 scans and measures 24.4 nm, an SEM image of which is shown in figureC.1(f). Further reduction of the lateral dimension might prove to be challenging in our current set-up as the beam is broadened by mechanical and electrical vibrations. As the amount of deposited material is reduced further still, the stochastics of the deposition process might start to play a role and influence the reproducibility of results.
Figure 3.5: SEM close-up of a SiOx line written on gold with the smallest achieved line width of 24.4 nm
3.1.3 Halo Deposition
In addition to the desired deposition, contiguous deposition around the target area is also observed. This halo deposition can be shown to depend on the dwell time (number of scans) used to write the feature as shown in figure 3.6. This effect is not unique to SiOx deposits and is also observed around W and Pt deposits, figure 3.7. The halo deposition is presumed to be due to secondary electrons emitted from the surface and dissociating the precursor gas. At such a distance from the area irradiated with the electron beam the emitted electron count is so low that it is often taken to be negligible.
However, with sufficient exposure enough material may be deposited to cover the surface.
This material, just as the intended deposit, can be functionalized, figure3.8(optical and SEM), and will negatively affect the selectivity of functional groups to the target area.
Figure 3.6: 1.5x1.5 µm SiOx squares written on Si substrate with beam energy of 10kV and an interative dwell time of 0.8 µs. Values indicate the number of scans.
3.1.4 Resilience to Ultrasonication
During the functionalization process the samples undergo various washes in an ultrasonic bath. As a consequence, deposits sometimes detach from the surface. How prone a certain feature is to removal in the ultrasonic bath is found to depend on two factors, namely deposit height and substrate.
The higher and narrower a feature is, the more likely it is to break off the substrate as is shown in figure3.9where pillar height increases from left to right. When detachment occurs, the halo deposit may either remain attached, figure 3.10 or may be detached together with the deposit 3.11. As in the case of SiOx, detachment was also observed for Pt and W deposits.
The deposits were less likely to survive the ultrasonic bath if they were attached to a gold, rather than a silicon substrate and if they were written close to the edge. The
Figure 3.7: W (top) and Pt (bottom) squares written with 30000 and 20000 scans re- spectively with an iterative dwell time of 1.14 µs, 15kVand PC number 5 on Si substrate
with ring of contiguous deposition visible
(a) (b)
Figure 3.8: SiOx square deposit written with 5kVand a probe current of 519.14 pA, functionalized with FITC in (a) fluorescence microscopy (scale bar is 1µm) with the contiguous deposition clearly functionalized, the darkened deposit is presumed to be due to interference effects and (b) the same square imaged in a SEM at a tilt of 60
degrees.
Figure 3.9: SEM image of SiOx dots/pillars of increasing height from left to right written on Au substrate and functionalized with polystyrene spheres. As is evident the higher deposits are more easily removed from the surface. Also visible adjacent to the
central pillar is one of the detached pillars on its side.
reduced affinity for gold surfaces can be explained by the lack of covalent bonds between gold and deposit as opposed to the same deposit on silicon. Samples written closer to the edge are believed to be more prone to stresses as the sample hits the side of the container when it is being jostled about in the ultrasonic bath.
3.2 Functionalization with FITC
Functionalization of the SiOx deposits was studied using a variety of different functional molecules and particles. In previous investigations the surface was always activated by means of treatment with Piranha solution so as to have a reproducibly functionalizable
Figure 3.10: SEM image at a tilt of 60 degrees of locations on a Au substrate where FITC functionalized SiOxpillars have detached. Clearly visible are the darker regions
of contiguous deposition.
Figure 3.11: SEM image at a tilt of 60 degrees of locations on a Au substrate where FITC functionalized SiOxpillars have detached. Clearly visible are the darker regions
of contiguous deposition.
surface. By functionalizing EBL patterned SiOx samples with the FITC + APTES procedure the effect of surface activation was studied by observing the fluorescence intensity of the features. Reported intensity values are in arbitrary units but have been normalized as:
Intensityof f eature Intensityof background − 1
By observing the fluorescence intensity of EBL samples that have undergone different surface activation treatments prior to functionalization with FITC, it was established that the SiOx is not fluorescent without functionalization regardless of activation treat- ment.
EBL SiOx Sample Intensity (arbitrary units ±0.2)
Untreated before functionalization 0
Piranha activated before functionalization 0
Utreated after functionalization 12.2
Piranha activated after functionalisation 14.5 O-plasma activated after functionalization 7.4
As it turns out, untreated deposits are also perfectly functionalizable, indicating there are sufficient hydroxyl groups on the deposit surface for silinization even without surface activation treatment. It is expected that treatment with piranha solution increases the number of hydroxyl groups on the deposit surface and so will lead to a higher degree of functionalization and thus higher intensity. This result is reflected in the table above.
What can also be concluded is that O-plasma is an alternative way to prepare the surface for functionalization. By comparing intensity values of the untreated and O- plasma activated functionalized samples, it may appear as if the O-plasma treatment removes hydroxyl groups from the deposit surface or in some other way decreases the degree to which the surface may be functionalized. However, this is attributed to a growth of an oxide layer on the surrounding Si substrate as a result of the O-plasma treatment. This hypothesis is backed by the nominal background intensity values prior to normalization.
3.2.1 Height Dependance
It has previously been established [28] that the emitted light from a SiOx deposit func- tionalized with the FITC fluorophore show a dependence on the height of the deposit.
This observation was reproduced by functionalizing 1x1µm squares with FITC according to the APTES + FITC procedure after having been activated in Piranha solution and is shown in3.12.
Figure 3.12: Fluorescence microscopy image of FITC functionalized SiOx deposits of increasing height (left to right) written on a Si substrate.
One possible explanation for this effect is that porosity of the SiOx deposits leads to a larger functionalizable surface. Immediately after deposition of SiOx no pores were observed in a SEM. This suggested that if porosity was the origin of the variation of intensity with deposit height, it is created during subsequent processing. It is known that the SiOxdeposits contain around 10% carbon [13], and the removal of these contaminants by the Piranha solution, known for its ability to remove carbon based contaminants, could lead to the formation of a porous structure.
If the above hypothesis is correct then SiOx deposits not treated with Piranha and functionalized with FITC should show no height dependence on luminosity. A pattern of squares of varying heights were written on a silicon wafer and functionalized according to procedure APTES + FITC without surface activation. The results are shown in3.13.
A dependence of the luminous intensity on the deposit height is still present, allowing us to conclude that Piranha is not responsible for this effect. As SEM imaging of deposits fails to reveal a porous structure, the pores can be assumed to be microporous (<2 nm in size). This limitation, together with the knowledge that an APTES and FITC molecule have a combined length on the same order of magnitude, makes it unlikely that the effect is caused by porosity. An alternate explanation is that the fluorescent intensity dependence on deposit height is due to interference effects.
Figure 3.13: Graph showing the intensity in arbitrary units plotted against the height (proportional to the number of scans) of untreated SiOxsquares functionalized accord-
ing to the FITC + APTES procedure
3.2.2 Interference
Figure 3.14: Fluorescence microscopy image of FITC functionalized SiOx deposits of increasing height (left to right) written on a Si substrate with the darkening presumed
to be due to interference effects
With a further increase in the height of the SiOxdeposits functionalized with the APTES + FITC procedure, we observed the darkening of squares, figure3.14. This observation, in combination with previously observed dark fringes on the sloping edges of EBL fea- tures functionalized in the same way, figure 3.15, suggests an interference effect is at work.
As the observed light intensity is that of the emitted light, it is unclear whether the interference is due to the absorbed or the emitted light. As interference effects of the
Figure 3.15: Fluorescence microscopy image of FITC functionalized square EBL SiOx features with a sloping edge
incoming light will dictate whether or not electrons in a fluorescent molecule are excited in the first place, we should start our investigation with absorbed rather than emitted light. The assumption that the observed darkening is caused by interference effects can be verified by demonstrating that the intensity minimum coincides with a deposit height that satisfies the condition for destructive interference.
Assuming that nair < nSiOx < nSi, incoming light traveling through the air/ SiOx de- posit interface will undergo a 180 degree phase transition, reflect off the silicon substrate undergoing another 180 degree phase transition before interfering with the incoming ra- diation. By combining this knowledge, an expression for the optical path length and Snells law, the condition for destructive interference of light incident normal to the substrate is found to be satisfied when:
h = 4nmλ
SiOx
where h is the height of the deposit,λ is the wavelength of the incoming light, n is the refractive index of the SiOx deposit and m is an odd integer. The wavelength of the incoming light is known to be 495nm and we are observing the first minimum (m=1), the only uncertain parameter is the refractive index of our deposit at that particular wavelength. The refractive index for various Si based glasses can reasonably be assumed to lie between 1.46 and 1.6 [29]. Thus the corresponding deposit height that would satisfy the condition for destructive interference would be between 77 and 85 nm. By similar considerations, the first maximum (m=2) is expected to lie between 155 and 170nm.
