ANCHORING TRANSITIONS OF NEMATIC LIQUID CRYSTALS ON LARGE ANGLE DEPOSITED SILICON OXIDE THIN FILMS

Transcription

1 ANCHORING TRANSITIONS OF NEMATIC LIQUID CRYSTALS ON LARGE ANGLE DEPOSITED SILICON OXIDE THIN FILMS A dissertation submitted to Kent State University in partial fulfillment of the requirements for the Degree of Doctor of Philosophy By Cheng Chen August 2006 i

2 UMI Number: UMI Microform Copyright 2006 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, MI

3 Dissertation written by Cheng Chen B.S., Peking University, China Ph. D., Kent State University, 2006 Approved by Chair, Doctoral Dissertation Committee, Philip J. Bos, Professor of Chemical Physics Interdisciplinary Program Members, Doctoral Dissertation Committee, John L. West, Professor of Chemistry Department, Deng-Ke Yang, Professor of Chemical Physics Interdisciplinary Program, David W. Allender, Professor of Chemical Physics Interdisciplinary Program, Kenneth K. Laali, Professor of Chemistry Department Accepted by, Oleg D. Lavrentovich, Director, Chemical Physics Interdisciplinary Program, John R.D. Stalvey, Dean, College of Arts and Sciences ii

4 TABLE OF CONTENTS TABLE OF CONTENTS...iii LIST OF FIGURES... v LIST OF TABLES...xiii ACKNOWLEDGEMENTS... xiv Chapter 1 Introduction Liquid Crystalline Materials Liquid Crystal Displays Liquid Crystal Alignment and the Method to Achieve the Same Overview of the Dissertation Chapter 2 Theory Introduction Review of Previous Theories Short Range Interactions Long Range van der Waals Potential Competition between Long Range and Short Range Forces Topography Our Theory Summary Chapter 3 Physical-chemical properties of LAD-SiO x thin films Introduction iii

5 3.2 Experimental Method Inorganic Alignment Layer Preparation Thin Film Characterization Method Experimental Results and Discussions Surface Topography and Anisotropy Stoichiometry and Surface Properties Summary Chapter 4 Anchoring Transitions on LAD-SiO x Due to the Change in Liquid Crystal Composition Introduction Experimental Methods Materials Sample Preparation General Examination Methods and Definition for Alignment Quality Pretilt Measurement Dielectric Anisotropy Measurement Method Birefringence Measurement Method Electro-Optical Curve and Response Time Measurement Methods Experimental Results The Effect of Large Longitudinal Dipole The Effect of Large Lateral Dipole The effect of varying the molecular structure of the additives iv

6 4.3.4 A Method to Make Improved Liquid Crystal Mixtures for Vertical Alignment Applications Discussions The Effect of Large Longitudinal Dipole The Effect of a Large Lateral Dipole The effect of molecular structure on liquid crystal anchoring on SiO x Summary Chapter 5 Temperature Dependence of the Anchoring Transitions on LAD-SiO x Introduction Experimental Methods Cell Preparation and Characterization Surface Adsorption and Thermal Desorption Results Thermal Induced Anchoring Transitions The Effect of Temperature on the Critical Concentration of 5CB Thermal Desorption Discussions Thermal Induced Anchoring Transitions The Effect of Temperature on the Critical Concentration of 5CB Summary Chapter 6 The Effect of LAD-SiO x Thickness on Liquid Crystal Anchoring Introduction v

7 6.2 Experimental Methods LAD-SiO x Sample Preparation Polyimide Sample Preparation Pretilt Measurement Experimental Results The Effect of LAD-SiO x Thickness on Liquid Crystal Alignment The Effect of LAD-SiO x Thickness on the Critical Concentration of 5CB Screening Effect Discussions The Effect of LAD-SiO x Thickness on the Alignment of Liquid Crystal The Effect of LAD-SiO x Thickness on the Critical Concentration of 5CB Screening Effect Summary Chapter 7 Conclusions and Suggestions for Future Work Summary of Dissertation Work Conclusions Suggestions for Future Work vi

8 LIST OF FIGURES Figure 1: Illustration of Dubois-Violette and de Gennes model in which long range van der Waals torque prefers planar alignment while short range forces prefer homeotropic alignment...14 Figure 2: The preference in LC orientation by long-range/short-range forces...18 Figure 3: The Working Principle of AFM...28 Figure 4: The working principle of XPS...28 Figure 5: AFM images of LAD-SiO x thermally evaporated at a medium angle. (a): 10µm x 10µm tapping mode 3D image (b): 5µm x 5µm tapping mode 3D image (c): 3µm x 3µm contact mode 2D image of friction (d): Cross-section analysis...32 Figure 6: (a): RMS Roughness of LAD-SiO x surface as a function of layer thickness (b): Anisotropy in surface roughness as a function of layer thickness...33 Figure 7: XPS spectrum of thermally evaporated LAD-SiO x and e-beam evaporated LAD- SiO 2, measured at 45º take-off angle. Atomic ratio of Si and O of the sample can be calculated from the corresponding area of the peak. Signal of carbon is from the residual of CO 2 or hydrocarbon contaminations on the sample surface...34 Figure 8: XPS spectrum analysis of silicon (Si2p) in (a) e-beam evaporated LAD- SiO 2 and (b) thermally evaporated LAD-SiO x. The blue line is the characteristic peak of Si in SiO 2 ; The cyanic line is the characteristic peak of Si in SiO; The magenta line is the characteristic peak of Si in Si crystal; The black line is the measured Si peak; The red line is the synthetic peak based on characteristic Si peak in SiO 2, SiO and Si crystal...35 vii

9 Figure 9: Chemical structure of (a) 5CB and (b) C Figure 10: Anchoring transitions from parallel to homeotropic to parallel again as the concentration of 5CB in the mixture with LC1 decreases. From top left to bottom right: pure 5CB, 50% 5CB, 25% 5CB, 10% 5CB, 5% 5CB, and pure LC1. Photo taken with cells placed between crossed polarizers on a light table Figure 11: Anchoring transitions of liquid crystal mixtures (5CB/LC1) on LAD-SiO x due to the change of the ratio of two components...48 Figure 12: The addition of C3 into LC2 leads to an anchoring transition of liquid crystal on LAD-SiO x from homeotropic to planar...51 Figure 13: The addition of 5CB into the mixture of C3 and LC2 causes an anchoring transition from planar to homeotropic on LAD-SiO x...52 Figure 14: The correlation between the concentration of C3 and the critical amount of 5CB that is needed to maintain homeotropic alignment of C3/5CB/LC2 mixture on LAD-SiO x...53 Figure 15: On E-beam evaporated SiO 2, more C3 is needed than on thermally evaporated SiO x to cause its mixture with LC2 to change from homeotropic alignment to planar alignment...54 Figure 16: Alignment of mixtures with different additives of LC1 on LAD-SiO x, photographed between crossed polarizers on a light table. From top left to bottom right cells are filled with: LC1; 10%C5-Ph-Ph-CN (5CB); 10%C5-Ph-Ph-O-C2; 5% C5-Ph- Ph-Br, 10% C3-Cyclohexyl-Ph-O-C2 (PCH302); 10% C5-Ph-Ph; 10%C6-Ph-Ph-C5..58 viii

10 Figure 17: The effect of cyano groups on the liquid crystal anchoring on LAD-SiOx. Left: 20% C7-Cyclohexyl-Ph-CN; Right: 5% C3 (C3- Cyclohexyl-COO-Ph(-2CN)-O-C2)..59 Figure 18: The addition of 5CB enables the mixture of LC2 and C3 to obtain uniform vertical alignment on LAD-SiO x with a greater negative dielectric anisotropy...62 Figure 19: The addition of 5CB also allows higher birefringence of the LC2/C3 mixture to be used for vertical alignment applications on LAD-SiO x...63 Figure 20: E-O curves of two identical LCoS devices filled with LC2 and improved mixtures (88% LC2, 10% C3 and 2% 5CB) respectively...64 Figure 21: Time response curves of two identical LCoS devices that used LAD-SiO x as alignment layers and were filled with LC2 and improved mixtures (88% LC2, 10% C3 and 2% 5CB) respectively...65 Figure 22: The addition of small amount of 5CB into a LC that has a large negative dielectric anisotropy also helps to produce uniform vertical alignment on polyimide alignment layers. Photo of SE-7511 coated cells purchased from EHC with ITO patterns. Left cell was filled with LC1. Right cell was filled with 10% 5CB +90% LC Figure 23: Dielectric anisotropy of 5CB/LCI mixtures as a function of 5CB concentration...71 Figure 24: A cartoon showing the effect of adding 5CB into LC1. Green and orange rods represent LC1 and 5CB molecules respectively. The blue surface represents the LAD- SiO x...72 Figure 25: A cartoon that shows the interaction between the LAD-SiO x and the cyano groups...73 ix

11 Figure 26: The working principle of a TDMS (thermal desorption mass spectroscopy)...79 Figure 27: Microscopic images of a LAD-SiO x cell filled with 1/3 LC1 and 2/3 LC2 at different temperatures. Left side photos were taken with crossed polarizers. Right side photos were taken with parallel polarizers Figure 28: Intensity of transmitted light as a function of temperature. Samples were held between crossed polarizers with evaporation direction 45º to the polarizer axis. All cells have the same cell gap ~20µm...83 Figure 29: Temperature dependence of the anchoring transitions of 5CB/LC1 mixtures on LAD-SiO x...84 Figure 30: XPS spectrum showing nitrogen atoms of 5CB on LAD-SiO x. On the spectrum of the original sample and the sample that has been baked at 49.5ºC, a peak of Nitrogen has been observed. This implies the existence of 5CB on the SiOx surface. However on the spectrum of the sample that has been baked at 100ºC the nitrogen peak no longer exists, indicating that the thermal deposption temperature of 5CB is between 49.5ºC and 100ºC...87 Figure 31: (a): Thermal desorption curve of 5CB (3 samples of 5CB absorbed on LAD- SiOx were prepared by the same methods). (b): Thermal desorption curve of C Figure 32: The critical concentration of 5CB in the planar-to-homeotropic anchoring transition of 5CB/LC1 mixtures as a function of temperature...94 Figure 33: The effect of LAD-SiO x thickness on the alignment of liquid crystal. A commercial liquid crystal mixture with a negative dielectric anisotropy was used in the experiment x

12 Figure 34: The anchoring transitions in a 5CB/LC1 mixture depend on the underlying LAD-SiO x layer thickness Figure 35: The effect of LAD-SiO x layer thickness on the alignment of 5CB screened by polyimide that prefers homeotropic anchoring Figure 36: Anchoring Transitions induced by the screening effect of polyimide on top of LAD-SiO x surface Figure 37: Two infinite surfaces separated by distance D Figure 38. The cross section of a half slab of a liquid crystal cell Figure 39: The critical concentration of 5CB in the homeotropic-to-planar anchoring transition of 5CB/LC1 mixtures (shown in Figure 34) depends on the thickness of underlying LAD-SiO x layer xi

13 LIST OF TABLES Table 1: The preference in LC orientation by long-range/short-range torques...17 Table 2: The refractive index and dielectric constant data of LC1 and LC Table 3 General Composition of LC1 and LC2. Column 2 and 3 show the gas chromatography retain time of LC1 and LC2. Void indicates the missing of this component. Column 4 shows the molecular weight of the component Table 4: Additives and their effects in determining the anchoring of their mixtures with LC2 on LAD-SiO x. Here NUP, UVA, UP stand for non-uniform planar, uniform vertical alignment (homeotropic) and uniform planar respectively xii

14 To my family xiii

15 ACKNOWLEDGEMENTS This work is dedicated to my wife Rong Luo and my parents. Without their ceaseless encouragement and support, the dissertation would not have been possible. I also feel deeply grateful to Dr. Philip J. Bos who has been my advisor on the dissertation work. His knowledge and enthusiasm have been a constant source of motivation to me during this endeavor. Being always very considerate and helpful, Dr. Bos has given me the most support, not only in my research but also in many other ways. I would like to thank Dr. James E. Anderson for his collaboration. He has provided me with numerous insightful suggestions and discussions. I also want to thank all my colleagues at LCI for their kind help and valuable discussions. My committee members deserve special thanks for their willingness to participate and for their valuable insights. Funding for my research was provided by HANA Microdisplay Technologies, Inc. xiv

16 Chapter 1 Introduction 1.1 Liquid Crystalline Materials Thanks to the blooming LCD market, the phrase liquid crystal has become more and more known to the public during the past decade. As told by its name, liquid crystal is an intermediate phase between isotropic liquid and crystal. Everyday experience has shown that materials undergo a single transition from solid to liquid. However, there are many organic materials that exhibit mesophases where the molecular ordering lies between that of a solid and that of an isotropic liquid. Of all the types of liquid crystal phases, nematic is one of the most important and also by far the most widely used in the LCD industries. Generally speaking, a nematic liquid crystal is composed of rod-like organic molecules trying to align parallel to each other. A nematic liquid crystal has long range orientational order, but not positional order. The average direction of the molecules is labeled by a unit vector n ρ, called the director. A typical nematic liquid crystal molecule should have a rigid elongated core and one or two flexible tails

17 The combination of molecular orientational order and fluidity in a single phase results in remarkable properties unique to liquid crystals. [1] Due to the anisotropy of nematic liquid crystal molecular shape, and the long range orientational order, the macroscopic dielectric anisotropy and optical birefringence are prevented from being averaged to zero. And because of the fluidity (within certain temperature ranges), nematic molecules are able to realign in an electric field to minimize the free energy. These two features make nematic liquid crystals very useful in making electrically switchable optical devices such as LCDs. Depending on the sign of the dielectric anisotropy, ε = ε ε, nematic liquid crystals can be divided into two categories. A nematic with a positive dielectric anisotropy has greater polarizability along the director axis than in the direction perpendicular to it, and the director tends to align in the direction of the external electric field. On the contrary, a nematic with a negative dielectric anisotropy is more polarizable in the direction perpendicular to the director axis, and its director tends to align perpendicular to the direction of the external electric field. In regards of optical anisotropy, n = n n, most nematic liquid crystals are positive, i.e., light sees a higher refractive index for the electric field of the light along the director direction than perpendicular to the director direction. When polarized light passes through a liquid crystal layer it splits into two parts: ordinary light and extraordinary light. These two may experience different optical retardation because of 2

18 the birefringence of liquid crystals. Also, the output state can be controlled by electrically adjusting the liquid crystal orientation. Therefore, both phase and amplitude modulation can be achieved using electrically addressed liquid crystal films. 1.2 Liquid Crystal Displays Liquid crystals have found applications in many electronic devices because of their unique electro-optical properties. Among all the applications, the Liquid Crystal Display (LCD) is no doubt the most famous. Usually an LCD is composed of a thin layer of liquid crystalline material sandwiched between two glass plates with transparent electrodes. By controlling the voltage on the electrodes we can control the amount of light transmitted or reflected by each pixel on the display. Thus, images/text can be produced. Several liquid crystal modes are commonly used in LCD industries, including TN (twisted nematic), STN (super twisted nematic), ECB (electrically controlled birefringence), Pi-Cell, VA (vertical alignment), IPS (in-plane switching) and others. These names refer to specific liquid crystal director orientation (alignment) configurations that will be introduced in the next section. Two very important characteristics for all LCDs are Contrast Ratio and Response Time. Contrast ratio refers to the ratio of light intensity between a bright state and a dark state of a LCD. Response time is essentially the time needed to switch the liquid crystal between bright and dark states. Those two characteristics have been proven to be critical to the performance of a LCD. 3

19 An LCD can be either transmissive or reflective, or as a combination transflective. Direct-view flat panel LCDs in the market are usually transmissive while LCDs on wrist watches and in Rear Projection TVs (RP-TVs) are reflective. Mobile devices such as cell phones, MP3 players and PDAs are designed for both indoor and outdoor use so most likely transflective LCDs are used. For the purpose of this dissertation, I want to emphasize a type of LCDs called LCoS (Liquid Crystal on Silicon). LCoS is a technology that incorporates reflective LCD technology onto a silicon chip with a CMOS (Complimentary Metal Oxide Semiconductor) active matrix lying underneath. LCoS may enable the industry to manufacture high resolution RP-TVs with lower cost and better performance. For LCoS technology, a high contrast ratio, a fast response and a long lifetime with high light throughput are critical. Currently, TN and VA technologies are the most widely used in LCoS. In a vertically aligned nematic liquid crystal (VAN) cell, liquid crystals with a negative dielectric anisotropy are utilized. The inner surfaces of the cell are pretreated with alignment layers that give a liquid crystal orientation normal to the surface. In a Normally Black mode, a VAN cell is sandwiched between two crossed polarizers. Without voltage, light that goes in normal to the surface will not be affected by birefringence. So, the black state can be really black. With voltage, the director falls down trying to be perpendicular to the electric field and the effective birefringence increases. The polarization of the light will be changed when passing through the cell so that light will pass through the analyzer. 4

20 One major advantage of the VAN mode is its superior high on-axis contrast ratio even without a retarder. With the help of a negative C plate, a high contrast ratio over big viewing angle can also be achieved. In most designs, VAN requires a small pretilt angle from the surface normal to prevent the formation of disclination lines, which seriously lower the display quality. A larger pretilt angle also allows the liquid crystal device to work at an increased speed. However, the pretilt angle must be small enough not to degrade the black state and hence the contrast ratio of the display. So, the pretilt angle has to be carefully chosen and controlled so that it balances both properties. 1.3 Liquid Crystal Alignment and the Method to Achieve the Same Traditionally, liquid crystal alignment is achieved by unidirectional rubbing of polyimide thin films on the surface of the electrodes. Polymer chains are believed to align along the rubbing direction and provide an anisotropy that aligns the liquid crystal director. Depending on the type of polyimide used, both planar and vertical alignment can be obtained. This technique has been widely adopted in LCD manufacturing. However, rubbing is at the same time thought to be dirty and not preferable in the clean room because it generates a lot of particles. Rubbing may also produce cosmetic defects such as scratches on the surface. This is very important to microdisplay applications where any defect will be magnified, sometimes with a factor of more than 40 when projected. What s more, the organic nature of the polyimide alignment layer makes it susceptible to damage from strong light intensity, especially when UV light is 5

21 considered. This leaves the lifetime of the device questionable. Because of all the reasons above, a rub-free, inorganic alignment layer is highly desired. In 1971, John L. Janning first reported that obliquely deposited inorganic layers are able to align liquid crystals. [2] Ever since, the topic has been extensively studied by numerous researchers. The scope of the research covers many inorganic materials (such as SiO, SiO 2, CaF 2, MgF 2, metals) and many deposition techniques (such as thermal evaporation, e-beam evaporation, sputtering, ion-beam etching, and chemical vapor deposition). The resulting alignments include planar, high pretilt and vertical alignment. The advantage of using inorganic alignment layers is not limited to a cleaner process and better UV stability. It also provides a reliable method to produce alignment that is very difficult to obtain using PI (such as 45 tilt and 3 pretilt of VA). Big efforts have been spent to understand the mechanism of the alignment, which has been found to be rather complicated. A detailed literature review of vertical alignment on inorganic layers is provided in Chapter Overview of the Dissertation In this dissertation we will first review previous work of the alignment of liquid crystal on inorganic thin films. This includes the methods to produce an alignment layer, the liquid crystal alignment behavior on inorganic alignment layers, and the mechanism of the alignment. 6

22 Following the literature review will be a theory section in which we followed and expanded a model proposed by Dubois-Violette and de Gennes to discuss the competition between long range van der Waals forces and short range dipolar forces in determining the liquid crystal alignment on SiO x. After the discussion of the theory, experimental data on this topic will be presented. The first part is a study on the physical-chemical properties of SiO x thin films and the effects on liquid crystal alignment. The second part discusses the how two types of materials in liquid crystal mixtures affect the alignment by shifting the balance between long range van der Waals interactions and short range dipolar interactions. Experimental results on the anchoring transitions caused by the shift of competition balance will be shown. The third part reports the temperature dependence of the observed anchoring transition. Surface adsorption and thermal desorption is believed to cause the change in short range interaction strength hence the balance between long range van der Waals potential. Surface thermal desorption experiments were conducted and results are used to add to our theory to explain the temperature dependence of anchoring transitions. The dependence of anchoring transition on SiO x thickness is also studied and explained by the correlation between van der Waals potential and alignment layer thickness. Finally, we will summarize all the experimental data and discuss how the theory explains the phenomena we have considered. 7

23 Chapter 2 Theory 2.1 Introduction Obliquely evaporated silicon oxide (SiO x ) thin films have been of great interest in the past decades for its use as liquid crystal (LC) alignment layers. It is produced by evaporating silicon oxide source onto the target surface in vacuum. The obtained silicon oxide thin film may vary in its Si/O ratio as well as its chemical state so it s generally called SiOx. Compared to the traditional rubbed polyimides (PI), SiO x is obtained using a non-contact method that produces less cosmetic defects as well as fewer particles that can contaminate the alignment surface. It is more UV stable. It has been found capable of producing a wide range of pretilt angles. These advantages have caused SiO x to be considered or implemented in applications like microdisplays and telecommunications devices. Particularly, Large-Angle-Deposited SiO x (LAD-SiO x ) has attracted interests for its capability of producing vertical (homeotropic) alignment of liquid crystals

24 Along with the increased interest have been efforts to understand the mechanism of the alignment. In this chapter we will give a review of some work in that area. Based on those previous theories and the assumptions that we found valid in our particular case we propose an expansion to a previous model and use it to explain the mechanisms of liquid crystal anchoring transitions on LAD-SiOx. 2.2 Review of Previous Theories Short Range Interactions Surface short range interactions have been found important in liquid crystal alignment. For example, it was discovered that the 5 degree (shallow angle deposition) SiO x column structures that stick out from the surface, are important for LC anchoring. Also important are molecular groups on SiO x that has been coated with alcohol, silane or other organic materials. The surface has been pictured as a comb with liquid crystal molecules embedded between the molecular groups sticking out of the surface. 1,2,3 Wu et al. have reported an interesting alignment phenomenon observed on SiO x and successfully explained it using this theory. 4 From a more chemical-physical point of view, some other researchers have demonstrated that the strong interfacial interactions between the surface and the surface liquid crystal molecules give rise to the anchoring energy that determines the bulk orientation. 5,6 Those interfacial interactions may include steric interaction, charge-charge interaction, charge-dipole interaction, dipole-dipole interaction, hydrogen bonding or 9

25 even chemical bonding. Since liquid crystals are often polar materials, the coupling between the permanent dipole of a liquid crystal molecule and the surface dipoles/charges can be significant. As a result, dipole moments may tend to be normal to the interface to maximize their interaction. 7 This effect is essentially short range and never goes beyond a few tens of angstroms but it could be a big contribution to the LC anchoring Long Range van der Waals Potential The van der Waals potential between liquid crystals and an anisotropic medium has been reviewed by previous researchers. In two classic papers 8,9 Dubois-Violette and de Gennes have shown that the more polarizable axis of liquid crystal will align parallel to the more polarizable directions of the surface and the angular dependence can be separated out in the potential by using a simple expression: 2 U = U 0 sin θ (1) Here U 0 denotes the van der Waals potential with liquid crystal aligned in its preferred direction, and θ is the angle between the two more-polarizable axes. Many other researchers have followed the same sin 2 θ model 10,11 or P 2 (cosθ) model 12, 13. More recently Lu 14, Vithana 15, and Kang 16 et al. have shown that LC with a positive ε (dielectric anisotropy) prefers parallel alignment (also called planar) on LAD-SiO x, while LC with a negative ε prefers perpendicular alignment (also called homeotropic). Lu et al. have explained the effect by considering the difference in van der Waals 10

26 potential between parallel and perpendicular states, caused by the dielectric anisotropy of LC Competition between Long Range and Short Range Forces In reference [9], E. Dubois-Violette and P. G. de Gennes discussed the local Fredericks transitions. A solid/nematic interface was considered, where long range van der Waals torques favor perpendicular anchoring, while short range effects tend to induce a parallel anchoring. The final anchoring depends on the relative strength of short range and long range interaction. The authors proposed to use equation (2.2) to express the total energy. 2 dθ 2 2 2F = u( z)sin θdz + 2 K( ) dz + 2W sin θ 0 dz δ δ (2.2) As shown in Figure 1, z is the distance from surface, δ is a small isotropic gap to prevent the energy from diverging, which is in the magnitude of the size of a liquid crystal molecule. θ 0 is the angle that director deviates from the short range torque preferred direction (surface normal) on the interface, θ is the actual anchoring angle. F is the total free energy; u(z) is the van der Waals potential, K is the elastic constant of liquid crystal and W is the surface anchoring energy that corresponds to short range interactions. Two anchoring transitions were predicted: parallel < > conical (tilted) and conical < > perpendicular. These anchoring transitions are called local Fredericks transitions 11

27 because they are caused by local (short range) forces. Sonin et al. have successfully demonstrated local Fredericks transitions on mica cleavages covered by an amorphous 17, 18 film Topography Topography is an important factor in liquid crystal alignment. A classic view describes the surface of SiO x as porous columns or periodic structures. LC molecules are believed to align parallel to the surface everywhere and the orientation of the director is determined when the elastic distortion energy is minimized. 19,20,21,22 A more recent study by Papanek and Martinot-Lagarde has shown that other factors such as order electricity are important in the case of porous SiO x surface. 23 However, studies have shown that the porous surface morphology exists only when the evaporation angle (the angle between the SiO x beam and substrate surface) is small. Evaporation at a medium or larger angle (e.g. >30º) results in a more compact structure and a smooth surface. 24, 25 We have confirmed this using AFM (Atomic Force Microscopy). In this paper we restrict our attention to the particular case of Large Angle Deposited SiO x (LAD-SiO x ), where we found that the elastic energy resulted from the topography is at least one order of magnitude smaller than the measured anchoring energy. In this case topography is unlikely to have a significant effect on the liquid crystal alignment. 12

28 The fact that different liquid crystal materials may choose completely different orientation on the same SiO x substrate also indicates a mechanism that cannot be explained solely by the elastic distortions of the director. 13

29 Z θ 0 δ Figure 1: Illustration of Dubois-Violette and de Gennes model in which long range van der Waals torque prefers planar alignment while short range forces prefer homeotropic alignment. 14

30 2.3 Our Theory Our ideas are based on the model proposed by de Gennes and Dubois-Violette in reference [9]. The theory is related to anchoring transitions that are seen on smooth LAD- SiO x, and is made with three assumptions that have previously been accepted by many others as discussed in the last section: a) Short range dipolar interactions tend to align dipole moments perpendicular to the SiO x surface. b) Long range van der Waals interaction tends to align the more polarizable direction of the liquid crystal with those of the alignment layer c) We can neglect surface topography and resulting steric forces for the case of large angle deposited SiO x alignment layers used in this study. From the first assumption it follows that for an LC with a positive ε (the dipole is more or less along the long molecular axis); a perpendicular boundary condition is preferred by short range dipolar interactions, while for an LC with a negative ε a parallel boundary condition is preferred because the dipole is more or less perpendicular to the long molecular axis. The second assumption gives the long range force preference of bulk LC orientation as a function of dielectric anisotropy. We assume here that the in-plane polarizability of LAD-SiO x is greater than the out-of-plane polarizability. This assumption is consistent with the molecular structure of SiO x thin films. According to Philipp 26,27 and Hohl et al. 28 the molecular structure of SiO x can be described in a Random Binding Model. In the 15

31 model every silicon atom is combined with four other atoms (either oxygen or another silicon) to form a matrix. Considering the dimensions of this matrix, electrons should be easier to move in-plane than out-of-plane. Therefore, LAD-SiO x should be more polarizable in the surface plane than along its normal direction. As a result, a liquid crystal with a positive ε tends to align parallel to the surface but a liquid crystal with a negative ε tends to align perpendicular to the surface. In both cases the electrically more polarizable direction of the liquid crystal is parallel to the more polarizable direction of SiO x. The third assumption holds true in our particular case of large angle deposited SiOx. This allows us to neglect the elastic energy distortion on SiO x surfaces. Based on the above assumptions we can list the orientational preferences of both the long range van der Waals forces and short range dipolar forces in Table 1. A cartoon illustration is also shown in Figure 2. It is clear that long range van der Waals forces and short range dipolar forces have opposite preference in the liquid crystal orientation direction. The final liquid crystal anchoring on SiO x is determined by the competition between the long range van der Waals forces and short range surface dipolar forces. This hypothesis may explain many effects that were hard to explain before. For example, it has been found that the orientation of the first layer of liquid crystal can differ appreciably from the orientation in the bulk. Resinikov et al. reported that the first layer (or a monolayer) of 5CB aligns perpendicularly at the liquid crystal/quartz surface, but 16

32 the bulk of 5CB shows parallel anchoring. 29 Similar phenomena have been reported on other substrates like polymers, crystals and glass. 30,31 Liquid crystal dielectric anisotropy Long range van der Waals force preferred liquid crystal orientation Short range dipolar force preferred liquid crystal orientation Positive Parallel to the interface Perpendicular to the interface Negative Perpendicular to the interface Parallel to the interface Table 1: The preference in LC orientation by long-range/short-range torques 17

33 + LC - LC van der Waals force + LC - LC Short range forces Figure 2: The preference in LC orientation by long-range/short-range forces 18

34 Following the model that Dubois-Violette and de Gennes proposed in reference [9] we start with equation (2.2) to described the free energy in the situation where long range van der Waals torque prefers parallel anchoring while short range torques prefer perpendicular anchoring. We limit the consideration to either planar or perpendicular anchoring ( θ 0 = θ ) so that the more complicated conical situation can be excluded. We further assume that there s no deformation of liquid crystal director orientation to eliminate the elastic energy. This assumption may not be completely true but it should give us a fairly good approximation since the short range interaction only works on the first layer of liquid crystal. Therefore, the formula is simplified to: 2 2F = u( z)sin θdz + W sin δ 2 θ (2.3) δ Let us define U = u( z) dz then 2 2 2F = U sin θ + W sin θ (2.4) Here θ can only be 0 (perpendicular) or π / 2 (parallel) from the surface normal. Let us use superscript + and to denote the material with positive and negative dielectric anisotropy respectively. Now consider the following situations: a) An LC with a positive ε + When θ = 0, F = 0 ; when θ = π /2, 2 F = W U 19

35 So, when W + < U +, the system has lower energy in the parallel state and when W + >U + perpendicular anchoring gives lower energy. b) An LC with a negative ε Similar to equation 2.4, for a liquid crystal that has a negative ε the total energy can be written as 2 2 2F = U cos θ + W cos θ (2.5) to reflect the preference of long range and short range torque. When θ = 0, 2 F = W U, when π / 2 θ =, F = 0 So if the short range interaction is strong enough, i.e., W > U, a planar anchoring is preferred. On another hand if W < U and van der Waals wins, a perpendicular anchoring is preferred. c) A mixture containing both negative and positive ε LCs In a mixture that contains liquid crystals with both positive and negative ε we have to take into consideration the distribution of each component in the bulk and on the surface. A simplified model would be two active components (one positive and one negative) in a neutral base. Here we use x to denote the concentration of one component in the mixture x = m /( m + m + neutral m ) (2.6) 20

36 + x = m /( m + m + neutral m ) (2.7) Here m is the amount of the component in the mixture. In a liquid crystal mixture sandwiched between two LAD-SiO x, for any component, it is safe to assume that the bulk concentration in the cell is the same as x. However, the surface concentration can deviate from x appreciably. The surface concentration of a component can be represented by its surface coverage ratio Θ defined as Θ + = + + n / N (2.8) Θ = n / N (2.9) Here n is the number of adsorbed molecules and N is the maximum number of the molecules of this component that can be adsorbed, i.e., the total available sites for this particular component. As can be seen we have assumed that the total available sites could be different for different components because of their very different properties. Therefore, the total energy can be expressed as F = x U cos θ + Θ W cos θ x U sin θ + Θ W sin θ (2.10) The difference in energy between perpendicular anchoring and parallel anchoring is F = 2[ F(0) F( π / 2)] = x U + Θ W + x U Θ W (2.11) An anchoring transition takes place at the critical point when F = 0, i.e., x U Θ W = x U Θ W (2.12) 21

37 2.4 Summary In this chapter we have reviewed some important work regarding the liquid crystal alignment on SiO x. With certain assumptions we showed that long range van der Waals forces and short range dipolar interactions have opposite preference in liquid crystal alignment directions. We expressed the competition between long range and short range interactions in the form of a model proposed by de Gennes et al. Further we expanded this model to the case where multiple components were present with different dielectric anisotropies. The contribution to the energy by long range and short range interactions of each active component is assumed to be proportional to its bulk concentration and surface coverage ratio respectively. As a result, change of the concentration or surface adsorption properties of any component may shift the balance between long range van der Waals interactions and short range dipolar interactions, leading to anchoring transitions. The point where an anchoring transition happens has been given in the model as a state where no energy difference exists between homeotropic alignment and planar alignment. 22

38 Chapter 3 Physical-chemical properties of LAD-SiO x thin films 3.1 Introduction The history of SiO x as a liquid crystal alignment material started with John Janning s report in 1971 that obliquely evaporated SiO films caused 5CB and MBBA to align in a preferred direction. Later it was discovered that the composition of the resulted thin film may deviate from SiO and become SiO x where x can be between 1 and 2. Janning s discovery inspired great interest in the research of SiO x thin films as alignment layers both in applications and in scientific understanding. More recently, LAD-SiO x alignment layers found application in producing high quality VAN (vertically aligned nematic) microdisplays for rear projection TVs. Companies such as Sony and JVC are using this technique in mass production of products. Many other companies are trying to develop new technologies and products using SiO x. After 20 years, SiO x alignment layers have become a hot spot of research in the display industry

39 SiO x alignment layers possess many unique merits when compared to other alignment layers. For instance, SiO x layers are able to produce a wide range of pretilt angle that are extremely difficult to produce on traditional polyimide alignment layers. Another desirable feature of SiO x alignment layers is that the deposition process is clean. Compared to rubbing polyimides, SiO x deposition doesn t generate so many particles that contaminate the alignment surface. A rub-free process also prevents the devices from cosmetic defects such as scratches that can be disastrous to microdisplay applications. Thanks to its inorganic nature, SiO x alignment layers are also less sensitive to UV. Because of these unique advantages, SiO x has been widely considered in applications such as STN, VAN, pi-cell and dual frequency liquid crystal devices. For VAN applications, silicon oxide films evaporated at a relatively large angle (>30º w.r.t. the surface) are normally used. Though applications have been successful, the properties of the LAD-SiO x alignment layers and their effects on liquid crystal alignment are still poorly understood. In this chapter we will discuss the properties of LAD-SiO x thin films used in our experiments. 3.2 Experimental Method Inorganic Alignment Layer Preparation Two types of silicon oxide films were used: thermally evaporated SiO x and e-beam evaporated SiO 2. For the purpose of simplicity, I will hereafter name them as SiO x and 24

40 SiO 2 respectively. But they should be strictly differentiated for reasons that will be discussed later. SiO x alignment layers were prepared by thermally evaporating silicon monoxide (SiO) powders (purchased from Kurt J. Lesker Company) onto substrates. Though the equipment is able to do oblique evaporation with any angle to the substrate surface we did all our depositions at a large angle of incidence (usually w.r.t. the substrate surface). This particular range of angles has been shown by previous researchers to be effective in producing vertical alignment of liquid crystal. The thickness of coating is measured in-situ by an oscillating quartz crystal thickness monitor. The reading of the thickness monitor has been calibrated by ellipsometry measurement data. The deposition rate was controlled to be 2~3Å/s. Residual pressure in the deposition chamber was controlled by back-bleeding air through a needle valve. Electron beam (e-beam) provides a source of heat with much higher temperature. So instead of silicon monoxide, silicon dioxide is typically used as the source of evaporation. In our experiments, SiO 2 films were prepared by evaporating quartz pellets by e-beam using the same process parameters as those used for thermal evaporation. The e-beam evaporator we used was of the same basic geometry as the thermal evaporator. Thus the two evaporators have almost the same geometry and should produce substrates for a fair comparison. E-beam evaporated SiO 2 thin films were used only in a few cases in our study, mainly to compare with SiO x. 25

41 3.2.2 Thin Film Characterization Method AFM AFM is a widely used technique for surface characterization. It consists of a micro scale cantilever with a sharp tip (probe) that is used to scan the specimen surface. When the tip is brought into proximity of a sample surface, forces between the tip and the sample lead to a deflection of the cantilever according to Hooke Law. Typically, the deflection is measured using a laser spot reflected from the top of the cantilever into an array of photodiodes. The sample is mounted on a piezoelectric tube that can move the sample in the z direction for maintaining a constant force, and the x and y directions for scanning the sample. The resulting map of s(x,y) represents the topography of the sample. Usually there are two types of scan methods: contact mode and tapping mode. The former uses static probe while the latter uses probe oscillating at close to its resonance frequency. In our measurements we used both contact and tapping modes to scan fresh thin film samples in order to obtain clear images of the surface morphology. Then the images were analyzed by software to obtain cross-section plots and statistical information such as surface roughness, average horizontal domain size, average peak-to-peak height, etc. In anisotropy measurement, samples were first scanned along the evaporation direction, then along the direction perpendicular to it. For each scan, surface roughness was calculated. Roughness anisotropy is defined as the difference between the results of the two scans. 26

42 XPS The XPS technique is based on the photoelectric effect that electrons eject from a surface when photons impinge upon it. Al Kα (1486.6eV) or Mg Kα (1253.6eV) are often the photon energies of choice. The energy of the photoelectrons leaving the sample is determined using a Concentric Hemispherical Analyzer and this gives a spectrum with a series of photoelectron peaks. The binding energies of the peaks are characteristic of each element and its local environment. The peak areas can be used (with appropriate sensitivity factors) to determine the composition of the material s surface. The shape of each peak and the binding energy can be slightly altered by the chemical state of the emitting atom. Hence, XPS can provide chemical bonding information as well. The XPS technique is highly surface specific due to the short range of the photoelectrons that are ejected from the solid. By using different incident angles of X-ray, photoelectrons excited from different depths under the surface can be collected. Thus a depth profile of the sample can be obtained using an Angular Resolved XPS. In our study samples of LAD-SiO x and LAD-SiO 2, thin films were deposited on glass substrates and measured using Al Kα as the photon source. Spectra were analyzed to give chemical state, atomic ratio and other information. Other than the depth profiling, all measurements were done using a 45 angle. 27

43 Figure 3: The Working Principle of AFM Electron Energy Analyzer X-ray source Sample Pump Figure 4: The working principle of XPS 28

44 3.3 Experimental Results and Discussions Surface Topography and Anisotropy Surface topography of obliquely evaporated LAD-SiO x was examined by AFM (Atomic Force Microscopy). Figure 5 shows some typical AFM images of LAD-SiO x thermally coated at a large angle of incidence. A few points can be seen from the images. First, the surface topography suggests that LAD-SiO x thin films are possibly composed of densely packed column structures and the direction of column growth is close to the surface normal. Second, the LAD-SiO x surface is very smooth. The cross section analysis (Figure 5(d)) of the sample shows that a typical topographic feature on the surface is around 5nm in height but 200nm in width (note that the horizontal and vertical scaling in the figures are very different). The measured RMS roughness is generally around 1nm. So, it s more close to reality to picture the LAD-SiO x surface as a smooth ground with pebbles on it, rather than hills and valleys that are typically seen in glancing angle (such as 5º) deposition. On the other hand, a typical liquid crystal molecule is only about 2.5nm in length and 0.5nm in diameter. With this kind of geometry it is difficult to produce any significant elastic distortion in LC director field. We also studied the evolution of surface topography as we increased the LAD-SiO x layer thickness. The results are shown in Figure 6. Other than a tiny decrease in the low thickness region, little change has been seen in either surface roughness or anisotropy (defined as the difference in RMS roughness when sample is scanned along evaporation 29

45 direction and perpendicular to evaporation direction) when the thickness increases from ~30nm to ~350nm Stoichiometry and Surface Properties The stoichiometry of LAD-SiO x /SiO 2 thin films was studied by XPS (X-ray photoelectron spectroscopy ). From each spectrum, atomic ratio of each element in the sample can be calculated. As shown in Fig. 5(a) e-beam evaporated LAD-SiO 2 has an O/Si atomic ratio very close to 2/1. But for thermally evaporated LAD-SiO x we have seen ratios from 1.2 to 1.7, depending on the deposition conditions. The data from atomic ratio shows that thermally evaporated SiO x has an oxygen deficient chemical structure. The Angle-Resolved XPS also allows us to do a depth profile of the stoichiometry. Increasing the photoelectron takeoff angle by rotating the sample in the energy dispersive plane of the analyzer reduces the sampling depth. Using this technique we were able to measure the atomic ratio from the top of the surface to ~1.5nm underneath. The results we obtained showed no significant difference in atomic ratio of Si and O. The analysis software of XPS has the capability to fit the Si2p peak with the characteristic Si2p peak from crystal SiO 2, SiO, and silicon. The results shown in Fig 5(b) imply that e-beam evaporated LAD-SiO 2 is more close to crystal SiO 2 in its chemical structure while thermally evaporated LAD-SiO x has a big contribution from SiO and even a small contribution from silicon. As we all know in a crystal SiO 2 each Si atom bonds with 4 oxygen atoms to form a network of tetrahedrons. However, in the case of SiO x, there will be many unoccupied silicon orbits due to the lack of oxygen. Since the depth 30

46 profile of atomic ratio shows no obvious difference between the top surface and underneath, we believe that the LAD-SiO x surface also has many dangling bonds or empty orbitals that may attract nearby dipoles. As a summary, the e-beam evaporated LAD-SiO 2 surface is more passive compared to the oxygen-deficient thermally evaporated LAD-SiO x surface, which may have lots of empty Si orbits and dangling bonds on the surface. 31

47 (a) (b) (c) (d) Figure 5: AFM images of LAD-SiO x thermally evaporated at a medium angle. (a): 10µm x 10µm tapping mode 3D image (b): 5µm x 5µm tapping mode 3D image (c): 3µm x 3µm contact mode 2D image of friction (d): Cross-section analysis 32

48 Roughness vs. Thickness 1.5 along evaporation 1.25 perpendicular to evaporation (a) SiOx Layer Thickness/nm RMS Roughness Anisotropy(Difference between RMS along and perpendicular to evaporation direction) 0.5 Delta RMS 0.4 Delta Ravg (b) SiOx Layer Thickness/nm Figure 6: (a): RMS Roughness of LAD-SiO x surface as a function of layer thickness (b): Anisotropy in surface roughness as a function of layer thickness 33

49 (a) SiO 1.95 (b) SiO 1.50 Figure 7: XPS spectrum of thermally evaporated LAD-SiO x and e-beam evaporated LAD-SiO 2, measured at 45º take-off angle. Atomic ratio of Si and O of the sample can be calculated from the corresponding area of the peak. Signal of carbon is from the residual of CO2 or hydrocarbon contaminations on the sample surface. 34

50 (b) Figure 8: XPS spectrum analysis of silicon (Si2p) in (a) e-beam evaporated LAD-SiO 2 and (b) thermally evaporated LAD-SiO x. The blue line is the characteristic peak of Si in SiO 2 ; The cyanic line is the characteristic peak of Si in SiO; The magenta line is the characteristic peak of Si in Si crystal; The black line is the measured Si peak; The red line is the synthetic peak based on characteristic Si peak in SiO 2, SiO and Si crystal. 35

51 3.4 Summary AFM data reveals that SiO x thin films evaporated at a medium or large angle exhibit densely packed columnar structures in the direction close to the surface normal. The surface roughness and anisotropy are so small that we believe surface topography and elastic distortion energy is unlikely to have a significant effect on the anchoring of the liquid crystals on LAD-SiO x thin films. It is also extremely hard to use topography to explain all the anchoring effects we observed in the experiments. A mechanism that shows a closer relationship between the physical-chemical properties of SiO x and the liquid crystal molecules must be considered. The stoichiometry of SiO x also plays an important role in the liquid crystal anchoring. Unoccupied orbits or dangling bonds on the LAD-SiO x surface tend to interact with the dipole moment strongly. So, a saturated surface will be more stable and less interactive to liquid crystal molecules compared to an unsaturated one. From XPS data we can see that on thermally evaporated the LAD-SiO x surface, silicon atoms are not saturated with oxygen, leaving many orbits accessible to liquid crystal dipoles. On the other hand, e- beam evaporated LAD-SiO 2 is more like a crystal structure with each Si bonded to 4 oxygen atoms. As a result we can expect a stronger short range surface interaction between the alignment layer and the liquid crystal on LAD-SiO x, compared to on LAD- SiO 2. 36

52 Chapter 4 Anchoring Transitions on LAD-SiO x Due to the Change in Liquid Crystal Composition 4.1 Introduction In the previous chapter we reviewed experimental data on LAD-SiO x alignment layers and concluded that the topography is unlikely to produce significant effects on liquid crystal alignment in our particular case of medium angle evaporation. Another piece of evidence that topography should not be held responsible for the entire alignment phenomenon on SiO x is the material dependence of the alignment. In other words, different liquid crystals tend to align in different ways on LAD-SiO x. This cannot be explained using the model of elastic energy minimization. Here in this chapter we will report experimental observations of the material dependence of liquid crystal alignment on SiO x. Further, we will demonstrate anchoring transition phenomena due to the change of the relative ratio of two components in liquid crystal mixtures. The effect will be explained using the theory we introduced in Chapter 2, by considering the competition between the long range van der Waals interactions and the short range dipolar interactions. A novel method that may produce improved liquid crystal mixtures

53 for vertical alignment applications will be proposed based on our discovery involving anchoring transitions. 4.2 Experimental Methods Materials Silicon monoxide powder (EVMSIO-1065B, >99.99% purity) purchased from Kurt J. Lesker was used for the evaporation. Commercial liquid crystals from Merck were used in the experiment. LC1 and LC2 (part number intentionally omitted for the proprietary of the research sponsor) are liquid crystal mixtures with negative dielectric anisotropy. Table 2 lists the refractive index and dielectric constant of these two mixtures. Table 3 lists the general composition of these two mixtures. Notice that LC1 has a very large negative value of dielectric anisotropy. Another liquid crystal known as 5CB or K15 (4-cyano-4-n-pentylbiphenyl), also purchased from Merck was used. 5CB, as shown in Figure 9(a) is a small linear molecule with a strong polar group on one end. Therefore, it possesses a strong longitudinal dipole and a positive dielectric anisotropy. All other materials used in the experiments were synthesized in-house. Among them, 1 ethoxy 4 (4 trans - propylcyclohexylcarboxy) - 2, 3 - dicyanobenzene (hereinafter referred to as C3) is of particular importance. As shown in Figure 9(b), each C3 molecule has 2 cyano groups on one side producing a large dipole moment in the direction perpendicular to the molecular long axis. Also because of the cyano groups and the conjugation with benzene rings, C3 and compounds that have similar structures have been reported to have huge negative ε 38

54 and are used in commercial liquid crystal mixtures as dopants to increase the magnitude of the negative dielectric anisotropy. 32, Sample Preparation 1350Ǻ-thick SiO x films were deposited onto clean glass substrates at 45º by thermal evaporation. Residual pressure was controlled to be around 1.0x10-5 torr by backbleeding air through a needle valve. Coated substrates were assembled into 20µm-thick cells with anti-parallel deposition directions on the top and bottom plates. Liquid crystal was forced into the cell under vacuum by capillary force at room temperature. Following filling, the cells were sealed General Examination Methods and Definition for Alignment Quality After filling, the liquid crystal cells were first examined on a light table between crossed polarizers. With vertically aligned liquid crystals, cells should always look dark when rotated. For planar cells, bright-dark alternation will be observed when rotated. A cell is defined as uniform if all of following criteria have been satisfied: 1) More than 80% of the cell area has uniform brightness or darkness observed by visually; 2) Choose 5 spots in the uniform area that are at the area s center and 4 corners. For a planar cell, measure the extinction angle on each spot. The maximum difference between two extinction angles should be smaller than 2. Or, for a vertically aligned cell, measure the 39

55 pretilt angle on each spot. The maximum difference between two pretilt angles should be smaller than 1. Otherwise a cell is defined as non-uniform Pretilt Measurement The pretilt angle of liquid crystals confined in a cell was measured by one of two methods: Conoscopy and Crystal Rotation. The Conoscopy method was mainly used to measure a homeotropic cell with a small pretilt in which case an off-centered uniaxial cross can be recognized under conoscopic observation. There s a simple relationship that determines the pretilt angle: (r/r)/n.a. = n o sinθ (4.1) Here r is the distance between the conoscopic image center and the cross center. R is the diameter of the conoscopy. Detailed discussion of this method can be found in reference [34]. The Crystal Rotation method has been used in our experiments to measure larger pretilt angles that the conoscopy method is not capable of measuring due to the limitation of microscope numerical aperture. Details of this method are available in reference [35] Dielectric Anisotropy Measurement Method 20µm-thick empty cells with 1.0 cm 2 patterned ITO electrodes were made in our lab. Accurate cell gap thickness was measured from the interference patterns formed by the 40

56 reflection from top and bottom surfaces of the gap. The cell uniformity was also carefully examined by measuring the gap thickness at the center and at four corners of the patterned electrode. Only those cells with less than 2% thickness variation were used in the experiments. Spin-coated polyimides were used as alignment layers using the standard soft bake-hard bake procedure. For the homeotropic cell, SE-7511 was used, and for the planar cell, SE-2555 was used. Cells were filled with liquid crystal and then examined for uniformity. Pretilt angle of each cell was measured on a center-plus-fourcorners basis, as described before. The results show pretilt angle to be less than 1 for planar cells and greater than 89 for homeotropic cells, all angles measured from the surface. For each material, the impedances (real and imaginary parts) of both planar cell and vertical cell were measured on a Hewlett Packard 4284A 20Hz-1MHz precision LCR meter as a function of frequency, ranging from 1 khz to 1 MHz. ε was calculated using the equations of (2), (3), (4), and (5): Z 1 d = Z r + iz i = = ; (4.2) iωc iωaε 0 ε ε d d = = ; iωaε Z iωaε Z r + iz ) (4.3) 0 0 ( i dz i dz r ε r = ; ε 2 2 i = ; (4.4) 2 2 ωaε ( Z + Z ) ωaε ( Z + Z ) 0 r i 0 r i ε = ε ε = ε r ( vertical) ε r ( planar), (4.5) 41

57 Here Z is the impedance, ω is the angular frequency, C is the capacitance, A is the area of the electrode, d is the cell gap, ε is the dielectric constant, and ε0 is the dielectric permittivity of the free space. Subscript r and i stand for real and imaginary part respectively. And subscript and stand for parallel and perpendicular to molecular long axis respectively Birefringence Measurement Method Birefringence of the liquid crystal mixtures were obtained from the optical retardation measurements on the planar cells. The same center-plus-four-corners examination on cell thickness uniformity and pretilt was performed and only cells with less than 2% thickness variation and less than 1 pretilt (from the surface) were allowed. The optical retardation of a cell was measured by the standard Senarmont Technique. Birefringence was calculated from the optical retardation using equation (4.6): λ δ n = (4.6) d Here λ is the wavelength of light, which is 632.8nm in our case, δ is the optical retardation and d is the cell gap Electro-Optical Curve and Response Time Measurement Methods Electro-optical curves and response times of tested cells were measured using a home-built setup and software. The test cell is placed between crossed polarizers with its 42

58 surface projection of the easy axis making a 45º angle with the polarization axis. Light coming out from a 632.8nm He-Ne laser passed through the setup and passed to a detector. For the E-O curve, 1 khz AC was applied to the cell with rms voltage ramping from 0 to 10V. The transmitted light intensity was detected as a function of the ramping voltage. For response time, the tested cell was switched between 0 and 5V at 1 khz. The detector recorded the transmitted light intensity as a function of time. All measurements were done at 50 C. Table 2: The refractive index and dielectric constant data of LC1 and LC2 n e n o n ε ε ε LC LC

59 Table 3 General Composition of LC1 and LC2. Column 2 and 3 show the gas chromatography retain time of LC1 and LC2. Void indicates the missing of this component. Column 4 shows the molecular weight of the component. 44

60 C N (a) N N C C O O (b) C O Figure 9: Chemical structure of (a) 5CB and (b) C3 45

61 4.3 Experimental Results The Effect of Large Longitudinal Dipole A commercial mixture LC1, which has a large negative dielectric anisotropy ( ε = - 5.7) was filled into LAD-SiO x cells. While a liquid crystal with a moderate negative dielectric anisotropy will typically align vertically on LAD-SiO x, LC1 on the contrary assumes an orientation parallel to it. 5CB was filled into identical cells and was found to align parallel to the SiO x surface as well. However in this case 5CB has a positive dielectric anisotropy. Next we mixed 5CB into LC1 and filled the mixtures into LAD-SiO x cells. At room temperature, when the mixture contains less than 3% 5CB (by weight, the same in the following) it aligns parallel to the LAD-SiO x surface. When the concentration of 5CB reaches about 3% an anchoring transition takes place that brings the mixture into homeotropic alignment. Not until we increase the concentration of 5CB to about 55% does another transition happen and switch the LC anchoring to parallel again. Figure 10 shows a photo of cells filled with liquid crystal mixtures of 5CB and LC1, observed between crossed polarizers on a light table. Figure 11 plots the tilt angle of the LC director (w.r.t. substrate surface) as a function of 5CB concentration. 46

62 Figure 10: Anchoring transitions from parallel to homeotropic to parallel again as the concentration of 5CB in the mixture with LC1 decreases. From top left to bottom right: pure 5CB, 50% 5CB, 25% 5CB, 10% 5CB, 5% 5CB, and pure LC1. Photo taken with cells placed between crossed polarizers on a light table. 47

63 Tilt angle from surface Concentration of 5CB (weight%) in the LC1/5CB mixture Figure 11: Anchoring transitions of liquid crystal mixtures (5CB/LC1) on LAD-SiO x due to the change of the ratio of two components 48

64 4.3.2 The Effect of Large Lateral Dipole As described in C3 has two cyano groups on one side of the molecule. It also has a huge negative dielectric anisotropy, which makes it desirable as an additive used in making negative ε commercial liquid crystal mixtures. A commercial liquid crystal LC2 from Merck ( ε = - 2.7) was used to mix with C3. The reason we chose LC2 is that C3 has a good solubility in LC2 so the anchoring transitions can be more clearly demonstrated. LC2 by itself chooses homeotropic orientation on LAD-SiO x. We were not able to know how C3 aligns on LAD-SiO x because it doesn t have a nematic phase by itself. We increasingly added C3 into LC2 and filled the mixtures into LAD-SiO x cells. When the concentration of C3 is equal to or less than 5% the mixture still aligns perpendicular to the surface. However, starting from 6% a transition takes place and finally changes the LC anchoring into planar when the concentration of C3 is equal to or larger than 8%, as can be seen in Figure 12. The LC anchoring can also be swung back to homeotropic with the addition of small amount of 5CB to the mixture composed of LC2 and more than 8% of C3. Figure 13 shows the transitions indicated by the change of LC tilt angle. Also seen from the Figure 14 is for mixtures with higher concentration of C3, larger amount of 5CB is needed to trigger the transition. For the same experiment we have also used e-beam evaporated SiO 2 as the alignment layer. The SiO 2 layer was produced with identical thickness and deposition angle to the thermally evaporated SiOx. Figure 15 shows that on e-beam evaporated SiO2, more C3 is 49

65 needed than on thermally evaporated SiOx to cause its mixture with LC2 to change from homeotropic alignment to planar alignment. 50

66 100 Tilt angle form substrate surface Concentration of C3 (weight%) in the mixture of C3 and LC2 Figure 12: The addition of C3 into LC2 leads to an anchoring transition of liquid crystal on LAD-SiO x from homeotropic to planar 51

67 100 Pretilt angle from substrate surface/degree % C3 7.5% C3 10.0% C3 12.5% C3 15.0% C Concentration of 5CB (weight%) in the mixture of C3, 5CB and LC2 Figure 13: The addition of 5CB into the mixture of C3 and LC2 causes an anchoring transition from planar to homeotropic on LAD-SiO x 52

68 5 Critical Amount of 5CB (weight%) Concentration of C3 (weight%) Figure 14: The correlation between the concentration of C3 and the critical amount of 5CB that is needed to maintain homeotropic alignment of C3/5CB/LC2 mixture on LAD- SiO x 53

69 100 Tilt angle from surface/degree Thermal Evaporated SiOx E-Beam Evaporated SiO Concentration of C3 (weight% ) Figure 15: On E-beam evaporated SiO 2, more C3 is needed than on thermally evaporated SiO x to cause its mixture with LC2 to change from homeotropic alignment to planar alignment 54

70 4.3.3 The effect of varying the molecular structure of the additives Through our experiments previously described in and we found that C3 and 5CB play a keen role in determining the alignment of liquid crystal mixtures on LAD-SiO x. To understand how the chemical structure and physical properties of the additives affect the anchoring, several compounds were carefully chosen to use in the experiments. Their chemical structures are shown in column 2 of Table 4. They have similarities as well as differences in many ways to 5CB and C3. Specifically, the first 5 compounds have a cyano group along the molecular long axis and thus strong longitudinal dipole moments; but have different molecular length, shape, number of rings, and types of linkage groups. Compound 6 has similar structure to compound 5 except that the cyano groups are in the direction perpendicular to the molecular long axis, so it has a strong lateral dipole moment. Compound 7 (C3) is the same as compound 6 except that it has a cyclohexyl ring instead of benzene. Compound 8 to 13 are all similar to compound 1 but their functional groups along the molecular long axis are less polar or even nonpolar. Compound 14 has not only less longitudinal polarity, but also a cyclohexyl ring instead of benzene. Except 5CB and 8CB (compound 1 and 2), all compounds are synthesized in our institute. With these dopants, comparisons become possible between their polarity, molecular shape, dielectric anisotropy, electronic conjugation, and other properties. Additives are mixed with LC1 to observe the effect on anchoring. The experimental results are listed in Table 4. 55

71 Two conclusions can be made from the results. First, a small amount of a material that has a cyano group at the end of its molecular long axis tends to promote homeotropic alignment. This is clearly demonstrated in Figure 16 where a bunch of additives that have similar structure to 5CB were tested for the effects on LC1 alignment. Second, a small amount of a material that has cyano groups on the side of the molecular axis tends to promote planar alignment. Figure 17 shows a photo of two cells observed between cross polarizers. The left one was filled with a mixture of LC1 and a material that has a longitudinal cyano end-group. The right hand side one was filled with a mixture of LC1 and a material that contains two lateral cyano groups. As a result, the left cell is uniform homeotropic and the right cell is uniform planar. 56

72 Table 4: Additives and their effects in determining the anchoring of their mixtures with LC2 on LAD-SiO x. Here NUP, UVA, UP stand for non-uniform planar, uniform vertical alignment (homeotropic) and uniform planar respectively. Compound Host Liquid Crystal Weight % of dopant Alignment LC1 0 NUP Dopant Weight % of dopant Alignment CN 1 (5CB) From 2.5% to 50% UVA CN 2 (8CB) 5%, 10% UVA CN 3 20% UVA 4 N CN 10% UVA O 5 O O O CN 10% UVA O 6 O C O C O O 5% UP NC CN N N C C 7 (C3) O O C 5%,10% UP O 8 Br 5%, 25% NUP 9 O 4%, 10% NUP 10 10%, 33.3% NUP 11 10%, 25% NUP 12 O O 4%, 10% NUP 13 O 5%, 10% NUP 14 O 4%, 10%, 33.3% NUP 57

73 Figure 16: Alignment of mixtures with different additives of LC1 on LAD-SiO x, photographed between crossed polarizers on a light table. From top left to bottom right cells are filled with: LC1; 10%C5-Ph-Ph-CN (5CB); 10%C5-Ph-Ph-O-C2; 5% C5-Ph- Ph-Br, 10% C3-Cyclohexyl-Ph-O-C2 (PCH302); 10% C5-Ph-Ph; 10%C6-Ph-Ph-C5. Here Ph represents a phenyl (benzene) ring; C stands for carbon; O stands for oxygen and Br stands for bromine. 58

74 Figure 17: The effect of cyano groups on the liquid crystal anchoring on LAD-SiO x. Left: 20% C7-Cyclohexyl-Ph-CN; Right: 5% C3 (C3- Cyclohexyl-COO-Ph(-2CN)-O-C2) 59

75 4.3.4 A Method to Make Improved Liquid Crystal Mixtures for Vertical Alignment Applications. Electro-optical devices using vertically aligned liquid crystals with a negative dielectric anisotropy (VAN) have been widely used in many applications because of their high contrast ratio. To achieve lower driving voltage and faster response, liquid crystals with large ε are preferred. Unfortunately, it is well known that these types of liquid crystals are very difficult to align vertically on SiO x, sometimes even on polyimides. However, the experimental discovery we discussed in previous sections points out a potential way to solve this problem. If an appropriate amount of a positive dielectric material, such as 5CB is added to the host material that has a large negative dielectric anisotropy, uniform vertical alignment can be easily achieved. Commercial liquid crystal mixtures that have negative ε are generally made by mixing highly negative dopants into a neutral or slightly positive base liquid crystal mixture. As an example, we will use a mixture of LC2 and C3 to explain how the method we propose may help to improve the liquid crystal properties for VAN applications. In section we discussed the effect of introducing C3 into base material LC2. We found that if greater than 4% of C3 was added, the alignment of the mixture deviated from homeotropic toward planar. This effect practically prohibited us from producing useful liquid crystal mixtures with LC2 and C3 with a larger negative ε. We also reported that the addition of 5CB allowed mixtures of C3 and LC2 to form vertical alignment that they couldn t do originally. Though 5CB exhibits a positive ε itself the overall effect still shows improved ability to produce vertical alignment with a larger 60

76 negative ε. Figure 18 shows that improvement can be achieved in the magnitude of the negative value of ε vs. the amount of added 5CB. Also found was an improvement in the birefringence as shown in Figure 19. We made two 1.3µm-thick reflective liquid crystal cells using LAD-SiO x as the alignment layer. One was filled in with LC2. Another was filled with the mixture of 88% LC2, 10% C3 and 2% 5CB. The electro-optical and time response curves were measured. As shown in Figure 20, the device with the improved liquid crystal mixture has a lower threshold voltage and a higher optical retardation. The response times shown in Figure 21 are roughly the same, but since the improved LC has higher birefringence, the device could be made thinner to achieve a faster response time. Another interesting disclosure is that the addition of a small amount of 5CB or similar materials into a liquid crystal with a large negative dielectric anisotropy also helps to produce uniform vertical alignment on polyimide alignment layers. We used two identical empty cells from EHC with homeotropic polyimide coatings inside. One was filled with LC1 and another one was filled with the mixture of 10% 5CB and 90% LC1. As shown in Figure 22, both cells look like vertical alignment. But closer examination reveals that the one filled with LC1 has higher pretilt (bright) region around the gasket, the ITO pattern and at the filling port. But the one filled with the 5CB mixture shows almost perfect uniform vertical alignment. 61

77 ε Without 5CB, uniform vertical alignment cannot be achieved for C3 > 4% With critical amount of 5CB Concentration (weight%) of C3 in the mixture of C3, 5CB and LC2 Figure 18: The addition of 5CB enables the mixture of LC2 and C3 to obtain uniform vertical alignment on LAD-SiO x with a greater negative dielectric anisotropy. 62

78 0.18 Without 5CB, uniform vertical alignment cannot be achieved for C3 > 4% With critical amount of 5CB 0.16 n Concentration (weight%) of C3 in the mixture of C3, 5CB and LC2 Figure 19: The addition of 5CB also allows higher birefringence of the LC2/C3 mixture to be used for vertical alignment applications on LAD-SiO x 63

79 Normalized Light Intensity LC2 Improved mixture of 10% C3, 2% 5CB and 88% LC Voltage/V Figure 20: E-O curves of two identical LCoS devices filled with LC2 and improved mixtures (88% LC2, 10% C3 and 2% 5CB) respectively. 64

80 Normalized light intensity LC2 switch on LC2 switch off Improved mixture switch on Improved mixture switch off Time/ms Figure 21: Time response curves of two identical LCoS devices that used LAD-SiO x as alignment layers and were filled with LC2 and improved mixtures (88% LC2, 10% C3 and 2% 5CB) respectively. 65

81 Figure 22: The addition of small amount of 5CB into a LC that has a large negative dielectric anisotropy also helps to produce uniform vertical alignment on polyimide alignment layers. Photo of SE-7511 coated cells purchased from EHC with ITO patterns. Left cell was filled with LC1. Right cell was filled with 10% 5CB +90% LC1. 66

82 4.4 Discussions The Effect of Large Longitudinal Dipole In 4.3.1, we showed that the addition of small amount of 5CB into LC1 (which has a large negative ε) changes the LC anchoring on SiO x from parallel to vertical. But, further increasing the amount of 5CB changes the anchoring back to parallel again. We have discussed the competition between long range van der Waals forces and short range dipolar forces in the chapter on Theory. Here we continue the discussion to explain the experimental data. Liquid crystal mixtures with a negative ε are usually obtained by mixing a base LC with materials that have very high negative ε, such as C3. Since LC1 has a large negative ε we expect it to contain a relatively large amount of negative additives. The effects of doing so on LC anchoring are two-fold. On one hand an increased ε will increase the anisotropy in van der Waals potential, making vertical alignment a more preferred LC anchoring on LAD-SiO x. On the other hand, the short range interaction between LAD-SiO x surface and negative additives such as C3 also increases, but with parallel orientation as its more favorable anchoring direction. In the case of LC1, short range dipolar interaction exceeds the long range van der Waals interaction so liquid crystal aligns parallel to the SiO x surface. 5CB is a small molecule with a large longitudinal dipole moment and a positive ε. It favors parallel anchoring by van der Waals forces but vertical anchoring by short range dipolar forces. When a small amount of 5CB is added to LC1 5CB molecules may bind 67

83 with the SiO x surface preferentially, which substantially changes the anchoring preference of short range forces towards favoring homeotropic alignment. Due to the limited amount of 5CB in the mixture, bulk properties are unlikely to be noticeably altered so the long range force still favors homeotropic anchoring. Hence both long range and short range interactions agree on the anchoring direction homeotropic alignment is obtained. Further increasing the amount of 5CB makes the ε of the bulk LC more positive and turns the van der Waals potential preference towards parallel anchoring. This can be seen in Figure 23, in which dielectric anisotropy of the mixture was measured as a function of 5CB concentration. However since the SiO x surface becomes more or less saturated with 5CB molecules the short range interaction doesn t increase. As a result long range van der Waals forces eventually prevail and the LC changes back to parallel anchoring. This process of anchoring transition is illustrated in the cartoons of Figure The Effect of a Large Lateral Dipole In the experiment described in , C3 is mixed with LC2 causing an anchoring transition of the liquid crystal from homeotropic to planar. Adding 5CB into the mixture shifts the anchoring back towards vertical. This effect can also be explained by the theory we proposed in Chapter 2. C3 has a two cyano groups hence a very large dipole moment perpendicular to its molecular long axis. The addition of C3 into LC2 leads to the increase of short range 68

84 dipolar interaction between the liquid crystal and the LAD-SiO x. It also contributes to the van der Waals potential due to its negative ε. But the increase in short range interactions must be more profound to cause the anchoring transition towards planar alignment. The effect of adding 5CB afterwards is the same as discussed in the previous section when 5CB is mixed with LC1. 5CB molecules bind with SiO x surface preferentially and make the short range dipolar interactions favor an anchoring direction perpendicular to the LAD-SiO x surface, while in the bulk, C3 still dominates the long range van der Waals interaction so a homeotropic alignment is preferred. Since both long range and short range interactions favor homeotropic alignment the anchoring is switched back to homeotropic. The reason why critical concentration of 5CB is quasi-proportional to the concentration of C3 lies in the short range forces. For C3 short range force prefers parallel anchoring while for 5CB it prefers vertical anchoring. The result is that to keep the surface short range interaction in favor of homeotropic anchoring the favorable interaction between 5CB and LAD-SiO x must exceed the unfavorable interaction between C3 and LAD-SiO x. Therefore, more 5CB is required in a system that contains more C3 to maintain homeotropic alignment. In the comparison of SiOx and SiO 2 we showed that the critical concentration of C3 needed to trigger the anchoring transition is much higher on LAD-SiO 2 than LAD-SiO x. We believe that this is also because e-beam evaporated LAD-SiO 2 has less surface 69

85 polarity than thermally evaporated LAD-SiO x, meaning the short range surface dipolar interactions (that prefer parallel anchoring) between LAD-SiO 2 and C3 is smaller than that between LAD-SiO x and C3. Therefore, more C3 is needed for LAD-SiO 2 to achieve the same magnitude of short range torque on SiO x to compete with long range van der Waals torque and change the anchoring direction The effect of molecular structure on liquid crystal anchoring on SiO x In section we tested several compounds for their ability to promote homeotropic alignment on SiO x. The experimental results show that a molecule with a cyano group at the end of its molecular axis helps to generate homeotropic alignment, while a molecule with cyano groups on its side helps to obtain planar alignment. We have already explained that this effect is due to the short range interactions between liquid crystal molecules (especially cyano groups) and the LAD-SiO x surface. Though we believe that any large dipole moment in general will cause similar effect, we give an explanation of why cyano group looks particularly effective in our experiments. Let us review the stoichiometry of LAD-SiO x described in the Chapter 2. LAD-SiO x is an oxygen-deficient structure. According to the random-binding model, an Si atom forms a tetrahedron with 4 randomly selected atoms (Si or O). At the surface, Si has nothing to bind hence has a vacant orbital. These orbitals are electron acceptors, which makes SiO x a Lewis acid. On the other hand, a cyano group has a pair of spare electrons, which makes it a strong electron donor, i.e., a Lewis base. The strong interaction between a Lewis acid and base makes the cyano-group orientated along surface normal as shown in Figure

86 Dielectric Anisotropy of LC1/5CB mixtures dielectric anisotropy % 20% 40% 60% 80% 100% CB concentration (wt%) Figure 23: Dielectric anisotropy of 5CB/LCI mixtures as a function of 5CB concentration 71

87 (a) (b) (c) Figure 24: A cartoon showing the effect of adding 5CB into LC1. Green and orange rods represent LC1 and 5CB molecules respectively. The blue surface represents the LAD-SiO x. 72

88 Figure 25: A cartoon that shows the interaction between the LAD-SiO x and the cyano groups. 73

89 4.5 Summary In this chapter we have shown our experimental results and explanations on the material dependence of liquid crystal alignment on LAD-SiO x and the anchoring transitions caused by the material dependence. The experimental results can be concluded as follow: A liquid crystal with a positive ε aligns parallel to LAD-SiO x surface while that with a moderate negative ε aligns vertical to LAD-SiO x surface. However, a liquid crystal with a large negative ε aligns parallel to LAD- SiO x surface. The addition of small amount of a material that has a large longitudinal dipole to a liquid crystal that has a large negative ε promotes perpendicular anchoring on LAD-SiO x. But further increasing the amount of additive leads to planar alignment. The addition of a material that has a large lateral dipole to a liquid crystal with a moderate ε leads to an anchoring transition from homeotropic to planar alignment on LAD-SiO x. Further introduction of a small amount of a material with a large longitudinal dipole can switch the anchoring back to vertical. Results were explained by the following points: Dipole moment (or cyano group) tends to align perpendicular to the LAD- SiO x surface. Therefore, a material with a longitudinal dipole prefers the 74

90 homeotropic boundary condition while a material with a lateral dipole prefers the planar boundary condition. Assuming that the small additive molecules with large longitudinal dipole moments will bind with the LAD-SiO x surface preferentially, they will favored in covering the LAD-SiO x surface hence change the overall orientational preference of surface short range dipolar interactions toward homeotropic alignment. Long range van der Waals interaction between LAD-SiO x and the positive ε additive favors planar anchoring. Therefore, the addition of the additive with a large longitudinal dipole also shifts the long range forces towards favoring planar alignment. The final alignment depends on the relative strength of these two opposite effects. 75

91 Chapter 5 Temperature Dependence of the Anchoring Transitions on LAD-SiO x 5.1 Introduction Liquid crystal alignment on SiO x has been found to be temperature dependent. But most commonly pretilt angle of liquid crystal has been mentioned in the literature without much regard to the influence of temperature. Some work has been published on the temperature behavior of liquid crystals 36, 37, 38, 39, 40, 41, 42, most of which proposed to use the temperature dependence of the order parameter (S) to explain the temperature dependence of liquid crystal orientation. The temperature-correlated term can come into the free energy either through the S-dependence of the van der Waals interaction, the S 2 - dependent elastic adaptation to the surface topology term, or through the order electricity term that depends on the gradient of S on the SiO x surface. Most of the reported experimental results show that anchoring transitions become obvious only when temperature is about 1 C below the T NI. However, Vithana et al. have reported that there is a gradual increase in the pretilt angle of homeotropic alignment on SiO x

92 starting from about 30 C below the T NI. 43 We observed similar effects in our experiments on LAD-SiO x. We found that a smooth anchoring transition from homeotropic to planar alignment can start at least 20 C below the clearing temperature. More interestingly, the transition can be in the opposite directions for different nematics. While the temperature dependence of the order parameter has been successful in explaining many orientational effects of liquid crystals we propose in this chapter another possible explanation that we found to be useful in discussing the observed phenomena in our particular case. 5.2 Experimental Methods Cell Preparation and Characterization SiO x was evaporated onto substrates at a large angle of incidence. Cells were assembled with anti-parallel evaporation directions on the top and bottom substrates. Cell thickness was ~20µm. Liquid crystals were filled by capillary force under vacuum. Pretilt angle was measured by conoscopy and crystal rotation Surface Adsorption and Thermal Desorption The combination of Thermal Desorption analysis and Mass Spectroscopy makes an effective tool for studying the surface interaction between LAD-SiO x and liquid crystal molecules. The equipment we used was a Thermo Electron Polaris Q GC-MS with Direct Exposure Probe (DEP). The principle of this method is illustrated in Figure 26. The 77

93 sample is placed on the probe and inserted into a vacuum chamber where the probe is heated up with a pre-set temperature ramping profile. Molecules that are originally absorbed on the sample surface are excited by the heat and leave the surface. Then the free molecules are bombarded by an electron beam and get ionized. Ions fly into the mass spectrometer and are analyzed. From the measurement, a time dependent mass spectrum is obtained. This can be translated into the relative abundance of certain chemicals with evolving temperature, from which we should infer the basic information of the binding properties between these chemical and the surface. In the experiments, LAD-SiO x was deposited onto both sides of aluminum foils and soaked into diluted liquid crystal solutions (0.3% by weight in isopropyl alcohol). After more than 8 hours the foils were taken out and gently dried. Dried foils were cut into small pieces around 1mm x 6mm size to be compatible with the crucible in the probe. This also helps with a good thermal conductivity between sample and the probe hence accurate temperature control on the samples can be achieved. Sample was heated up at 10 C/min. Real-time temperature measurement was done by a thermal coupler on the probe. The mass spectra of desorbed materials were recorded as a function of time. It is true that the LAD-SiO x surface will absorb many things other than the target liquid crystal molecules during the handling. When heated, all the absorbents tend to be set free from the surface and will be all recorded by the mass spectrometer. So before the measurement we had first obtained the standard mass spectrum of each target molecule so we can look only at their spectral signatures. 78

94 Figure 26: The working principle of a TDMS (thermal desorption mass spectroscopy) 79