<%BANNER%>

An Ab Initio Study of Alpha-Quartz (0001) Surface and Water-Silica Interface Interaction

Permanent Link: http://ufdc.ufl.edu/UFE0042518/00001

Material Information

Title: An Ab Initio Study of Alpha-Quartz (0001) Surface and Water-Silica Interface Interaction
Physical Description: 1 online resource (164 p.)
Language: english
Creator: Chen, Yun-Wen
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: defect, dft, md, quartz, silica, surface, water
Physics -- Dissertations, Academic -- UF
Genre: Physics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Silica is an important mineral for many applications in our daily life. Thus, numerous studies have been done to understand its properties and its interaction with other materials/molecules. In this work, we apply first-principles density functional theory (DFT) calculations to the investigation of alpha-quartz (0001) surfaces and water clusters and water layer(s) adsorbed on those surfaces. Ten different surfaces are studied for exploring alpha-quartz (0001) surface properties. They include five reconstructed surfaces, one fully hydroxylated surface, and four defective surfaces. Water molecule(s) and layer(s) were deposited on seven of the surfaces to examine how surface character can affect the behavior of water molecule(s) or layer(s) adsorbed on the surface. A newly cleaved silica surface generally is thought to be hydroxylated quickly in atmosphere and become hydrophilic. However, the surface can be dehydrated under an Ultra High Vacuum (UHV) environment and the siloxane surface is then hydrophobic. We will discuss the five perfect surface reconstruction patterns of the alpha-quartz (0001) surface (called perfect surfaces frequently in the following text) and how they can explain the observed (2 x 2) diffraction spectrum in experiments. A fully hydroxylated alpha-quartz (0001) surface will be discussed in detail regarding its formation energy, structure and charge distribution. Two oxygen vacancy and two oxygen displacement defective surface patterns on (1 x 1) perfect surface were considered in this work. Their formation energies, structures, charge distributions and hydroxylation on the defect sites will be discussed. Water molecule adsorption on surfaces was investigated by depositing molecules one by one. The hydroxylated surface exhibits highly hydrophilic properties. Conversely, perfect surfaces are hydrophobic or interact with water molecules quite weakly. A defect on the surface does enhance water adsorption on that surface. However, the strong hydrogen bonds within a water cluster have a tendency to compromise the water-silica interaction. Ice XI sub-layers were deposited one by one on surfaces. The highly stable bilayer structures which result effectively shield the interaction between water and the silica surface.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Yun-Wen Chen.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Cheng, Hai Ping.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042518:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042518/00001

Material Information

Title: An Ab Initio Study of Alpha-Quartz (0001) Surface and Water-Silica Interface Interaction
Physical Description: 1 online resource (164 p.)
Language: english
Creator: Chen, Yun-Wen
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: defect, dft, md, quartz, silica, surface, water
Physics -- Dissertations, Academic -- UF
Genre: Physics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Silica is an important mineral for many applications in our daily life. Thus, numerous studies have been done to understand its properties and its interaction with other materials/molecules. In this work, we apply first-principles density functional theory (DFT) calculations to the investigation of alpha-quartz (0001) surfaces and water clusters and water layer(s) adsorbed on those surfaces. Ten different surfaces are studied for exploring alpha-quartz (0001) surface properties. They include five reconstructed surfaces, one fully hydroxylated surface, and four defective surfaces. Water molecule(s) and layer(s) were deposited on seven of the surfaces to examine how surface character can affect the behavior of water molecule(s) or layer(s) adsorbed on the surface. A newly cleaved silica surface generally is thought to be hydroxylated quickly in atmosphere and become hydrophilic. However, the surface can be dehydrated under an Ultra High Vacuum (UHV) environment and the siloxane surface is then hydrophobic. We will discuss the five perfect surface reconstruction patterns of the alpha-quartz (0001) surface (called perfect surfaces frequently in the following text) and how they can explain the observed (2 x 2) diffraction spectrum in experiments. A fully hydroxylated alpha-quartz (0001) surface will be discussed in detail regarding its formation energy, structure and charge distribution. Two oxygen vacancy and two oxygen displacement defective surface patterns on (1 x 1) perfect surface were considered in this work. Their formation energies, structures, charge distributions and hydroxylation on the defect sites will be discussed. Water molecule adsorption on surfaces was investigated by depositing molecules one by one. The hydroxylated surface exhibits highly hydrophilic properties. Conversely, perfect surfaces are hydrophobic or interact with water molecules quite weakly. A defect on the surface does enhance water adsorption on that surface. However, the strong hydrogen bonds within a water cluster have a tendency to compromise the water-silica interaction. Ice XI sub-layers were deposited one by one on surfaces. The highly stable bilayer structures which result effectively shield the interaction between water and the silica surface.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Yun-Wen Chen.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Cheng, Hai Ping.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042518:00001


This item has the following downloads:


Full Text

PAGE 1

1 A N AB INITIO STUDY O F ALPHA QUARTZ (0001) SURFACE AND WATER SILICA INTERFACE INTERACTION By YUN WEN CHEN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Yun Wen Chen

PAGE 3

3 To my parents and siblings

PAGE 4

4 ACKNOWLEDGMENTS I acknowledge Dr. Hai Ping Cheng for her advice in all this work and her great patience. None of this would be done without her teaching and inspiring tutelage in Physics. I also thank my former colleagues, Dr. Chao Cao and Dr. Alexander Kemper for their help in programming and technological suggestions. In particular, Dr. Cao shared diverse knowledge in Physics and numerous strategies in computational science. I would also like to thank my colleagues and peers in my gr oup for their helpful questions and discussions in the long journey of developing my research Funding from NSF under Gran t No. DMR 0325553 and DMR 0804407 has supported my PHD study for past 5 years. The DOE/NERSC and UF/HPC center s have provided computing resource s for this work. I would like to show my appreciation to Dr. Samuel B. Trickey for his extensive help with editing this work.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............. 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 ABSTRACT ................................ ................................ ................................ .................... 13 CHAPTER 1 INTRODUCTION AND BACKGROUND ................................ ................................ .. 15 1.1 Two Important Ingredients in Human Life ................................ .......................... 15 1.2 Silica Surfaces ................................ ................................ ................................ ... 17 1.3 Water at Interfaces and Water Silica Interaction ................................ ............... 18 2 COMPUTATIONAL METHODS ................................ ................................ ............... 22 2.1 Classical Molecular Dynamics ................................ ................................ ........... 22 2.1.1 Newtonian Equations ................................ ................................ ............... 22 2.1.2 Computational Algorithm (for NVE ensemble) ................................ ......... 23 2.1.3 Extension to Thermostats and Barostats ................................ ................. 24 2.1.3.1 Berendsen thermostats and barostats ................................ ............ 24 2.1.3.2 Nos Hoover thermostats and barostats ................................ ........ 26 2.2 Density Functional Theory ................................ ................................ ................. 27 2.2.1 Hartree Fock approximation ................................ ................................ .... 27 2.2.2 Kohn Sham Ansatz and Local Density Approximation ............................. 29 2.2.3 Generalized Gradient Approximation ................................ ....................... 32 2.2.4 Pseudopotential ................................ ................................ ....................... 34 3 ALPHA QUARTZ (0001) SURFACES ................................ ................................ ..... 37 3.1 Perfect Surface Reconstruction ................................ ................................ ......... 38 3.1.1 Introduction ................................ ................................ .............................. 38 3.1.2 Method ................................ ................................ ................................ ..... 39 3.1.3 Results ................................ ................................ ................................ ..... 40 3.1.4 Discussion and Conclusion ................................ ................................ ...... 42 3.2 Fully Hydroxylated Surface ................................ ................................ ................ 43 3.3 Defective Surfaces ................................ ................................ ............................ 45 4 WATER CLUSTER ADSORPTION ON SILICA SURFACES ................................ .. 56 4.1 Introduction ................................ ................................ ................................ ........ 56 4.2 Methods ................................ ................................ ................................ ............. 58

PAGE 6

6 4.3 Water Adsorption on Surfaces 1 and 2 ((1X1) and (2X1) Perfect Surface) ...... 59 4.4 Water Adsorption on Surface 3 (the Fully Hydroxylated Surface) ..................... 6 2 4.5 Water Adsorption on Surfaces 4 and 5 (Oxygen Vacancy Defect) .................... 64 4.6 Water Adsorption on Surfaces 6 and 7 (Oxygen Displacement) ....................... 66 5 WATER LAYER(S) ADSORPTION ON SILICA SURFACES ................................ .. 94 5.1 Introduction ................................ ................................ ................................ ........ 94 5.2 Methods ................................ ................................ ................................ ............. 95 5.3 Water Adsorption on Surfaces 1 and 2 ((1X1) and (2X1) Perfect Surface) ....... 95 5.4 Water Adsorption on Surface 3 (the Fully Hydroxylated Surface) ................... 100 5.5 Water Ad sorption on Surfaces 4 and 5 (Oxygen Vacancy Defect) .................. 102 5.6 Water Adsorption on Surfaces 6 and 7 (Oxygen Displacement) ..................... 105 6 HYDROXYLATION OF ALPHA QUARTZ (0001) SURFACES ............................. 145 6.1 Introduction ................................ ................................ ................................ ...... 145 6.2 Methods ................................ ................................ ................................ ........... 146 6.3 Results ................................ ................................ ................................ ............. 146 6.3.1 Hydroxylation on Surfaces 1 and 2 ((1x1) and (1x2) Pefect Surface) .... 146 6.3.2 Hydroxylation on Defective Surfaces ................................ ..................... 148 7 SUMMARY ................................ ................................ ................................ ............ 156 REFERENCES ................................ ................................ ................................ ............ 159 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 164

PAGE 7

7 LIST OF TABLES Table page 3 1 Experimental and calculated structure of bulk alpha quartz ................................ 49 3 2 2 ) of five reconstructed surfaces and the freshly cleaved surface, and the ener gy barriers (in meV/ 2 ) for transition from type I (E I b ) or type II (E II b ) surface to others. ................................ ............... 49 3 3 The average charge on each atom species of surface type I, type III and quartz bulk.. ................................ ................................ ................................ ......... 49 3 4 The average charge on each atom species of a hydroxylated quartz slab. ........ 50 3 5 The formation energy of a defective surface with respect to (1x1) perfect surface. ................................ ................................ ................................ ............... 50 3 6 The average charge transfer onto Si/O atoms at and around the defect site.. .... 50 4 1 The adsorption and bonding energy of water molecules on silica surface.. ........ 69 4 2 Charge transfer analysis of water molecules on surface. ................................ .... 69 4 3 The charge redistribution of water cluster adsorption systems. .......................... 70 5 1 The adsorption and bonding energy of water layer(s) on a silica surface. ........ 107 5 2 The charge transfer analysis of water layer(s) on a silica surface (Positive ................................ ................................ ............ 108 5 3 The charge redistribution of water layer(s) adsorption systems.. ...................... 109 6 1 Hydroxylation energy of surfaces 1.. ................................ ................................ 151 6 2 Hydroxylation energy of surfaces 2.. ................................ ................................ 151 6 3 Hydroxylation energy of defective surfaces.. ................................ .................... 151

PAGE 8

8 LIST OF FIGURES Figure page 3 1 Top and side views of reconstructed alpha quartz (0001) surfaces. ................... 51 3 2 Transformations from one surface type to another. a) From type I (left) to type II (right); b) from type I (left) to type III (right).. ................................ ............. 51 3 3 Statistics of bond length and bond angles of all perfect surfaces.. ..................... 52 3 4 Top view of the large scale structure of a reconstructed alpha quartz surface from MD simulation. ................................ ................................ ............................ 52 3 5 Top view (a) and side view (b) of a full hydroxylated alpha quartz (0001) surface. ................................ ................................ ................................ .............. 53 3 6 Top view (a) and side view (b) of detec tive surface 1 (oxygen vacancy) ........... 54 3 7 Top view (a) and side view (b) of detective surface 2 (oxygen vacancy). ........... 54 3 8 Top view (a) and side view (b) of detective surface 3 (oxygen displacement). ... 55 3 9 Top view (a) and side view (b) of detective surface 4 (oxygen displacement). ... 55 4 1 One water molecule adsorbed on surface 1 ((1x1) perfect surface) at the hollow site. ................................ ................................ ................................ .......... 71 4 2 One water molecule adsorbed on surface 2 ((2x1) perfect surface). .................. 72 4 3 One water molecule adsorbed on surface 1 ((1x1) perfect surface) at the silicon atom site. ................................ ................................ ................................ .. 73 4 4 Two water molecules adsorbed on surface 1 ((1x1) perfect surface).. ............... 74 4 5 Two water molecules adsorbed on surface 2 ((2x1) perfect surface). ................ 75 4 6 Two water molecules adsorbed on two silicon sites of surface 1 ((1x1) perfect surface). ................................ ................................ ................................ .............. 76 4 7 Three water molecules adsorbed on surface 1 ((1x1) perfect surface). .............. 77 4 8 Three water molecules adsorbed on surface 2 ((2x1) perfect surface).. ............. 78 4 9 One water molecule adsorbed on surface 3 (fully hydroxylated surface). ........... 79 4 10 Two water molecules adsorbed on surface 3 (fully hydroxylated surface).. ........ 80 4 11 Three water molecules adsorbed on surface 3 (fully hydroxylated surface). ...... 81

PAGE 9

9 4 12 One water molecule adsorbed on surf ace 4 (oxygen vacancy defect). ............... 82 4 13 One water molecule adsorbed on surface 5 (oxygen vacancy defect).. .............. 83 4 14 Two water molecules adsorbed on surface 4 (oxygen vacancy defect).. ............ 84 4 15 Two water m olecules adsorbed on surface 5 (oxygen vacancy defect). ............. 85 4 16 Three water molecules adsorbed on surface 4 (oxygen vacancy defect). ......... 86 4 17 Three water molecules adsorbed on surface 5 (oxygen vacancy defect). ......... 87 4 18 One water molecule adsorbed on surface 6 (oxygen vacancy displacement). ... 88 4 19 One water molecule adsorbed on surface 7 (oxygen vacancy displacement). ... 89 4 20 Two water molecules adsorbed on surface 6 (oxygen vacancy displacement). ................................ ................................ ................................ .... 90 4 21 Two water molecules adsorbed on surface 7 (oxygen vacancy displacement). 91 4 22 Three water molecules adsorbed on surface 6 (oxygen vacancy displacement). ................................ ................................ ................................ ..... 92 4 23 Three water molecules adsorbed on surface 7 (ox ygen vacancy displacement). ................................ ................................ ................................ ..... 93 5 1 One water layer adsorbed on surface 1 ((1x1) perfect surface) with protons pointing up. ................................ ................................ ................................ ........ 110 5 2 One water layer adsorbed on surface 1 ((1x1) perfect surface) with protons pointing down. ................................ ................................ ................................ ... 111 5 3 One water layer adsorbed on surface 2 ((2x1) perfect surface) with protons pointing up. ................................ ................................ ................................ ........ 112 5 4 One water layer adsorbed on surface 2 ((2x1) perfect surface) with protons pointing down. ................................ ................................ ................................ ... 113 5 5 Two water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing up in the first water layer.. ................................ ................................ ... 114 5 6 Two water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing down.. ................................ ................................ ................................ .. 115 5 7 Two water layers adsorbed on surface 2 ((2x1) perfect surface) with protons pointing up in the first water layer. ................................ ................................ .... 116

PAGE 10

10 5 8 Two wate r layers adsorbed on surface 2 ((2x1) perfect surface) with protons pointing down.. ................................ ................................ ................................ .. 117 5 9 Three water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing up in the first water layer. ................................ ....................... 118 5 10 Three water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing down. ................................ ................................ ...................... 119 5 11 Three water layers adsorbed on surface 2 ((2x1) perfect surface) with protons pointing up in the first water layer. ................................ ....................... 120 5 12 Three water layers adsorbed on surface 2 ((2x1) perfect surface) with protons point ing down. ................................ ................................ ...................... 121 5 13 Four water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing up in the first water layer.. ................................ ................................ ... 122 5 14 Four water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing down. ................................ ................................ ................................ ... 123 5 15 Four water layers adsorbed on surface 2 ((2x1) perfect surface) with protons pointing up in the first water layer.. ................................ ................................ ... 124 5 16 Four water layers adsorbed on surface 1 ((2x1) perfect surface) with protons pointing down. ................................ ................................ ................................ ... 125 5 17 One water layer adsorbed on surface 3 (fully hydroxylated surface). ............... 126 5 18 The other two possible orientations of one water layer adsorbed on surface 3 (fully hydroxylated surface). ................................ ................................ .............. 127 5 19 Two water layers adsorbed on surface 3 (fully hydroxylated surface).. ............ 128 5 20 One water layer adsorbed on surface 4 (oxygen vacancy defect) with protons pointing up. ................................ ................................ ................................ ........ 129 5 21 One water layer adsorbed on surface 4 (oxygen vacan cy defect) with protons pointing down. ................................ ................................ ................................ ... 130 5 22 One water layer adsorbed on surface 5 (oxygen vacancy defect) with protons pointing up.. ................................ ................................ ................................ ....... 131 5 23 One water layer adsorbed on surface 5 (oxygen vacancy defect) with protons pointing down. ................................ ................................ ................................ ... 132 5 24 Two water layers adsorbed on surface 4 (oxygen vacancy defect) with protons pointing up in the first water layer. ................................ ....................... 133

PAGE 11

11 5 25 Two water layers adsorbed on surface 4 (oxygen vacancy defect) with protons pointing down. ................................ ................................ ...................... 134 5 26 Two water layers adsorbed on surface 5 (oxygen vacancy defect) with protons pointing up in the frist water layer. ................................ ....................... 135 5 27 Two water layers adsorbed on surface 5 (oxygen vacancy defect) with protons pointing down. ................................ ................................ ...................... 136 5 28 One water layer adsorbed on surface 6 (oxygen displacement) with protons pointing up. ................................ ................................ ................................ ........ 137 5 29 One water layer adsorbed on surface 6 (oxygen displacement) with protons pointing down. ................................ ................................ ................................ ... 138 5 30 One water layer adsorbed on surface 7 (oxygen displacement) with protons pointing up. ................................ ................................ ................................ ........ 139 5 31 One water layer adsorbed on surface 7 (oxygen displacement) with protons pointing down. ................................ ................................ ................................ ... 140 5 32 Two water layers adsorbed on surface 6 (oxygen displacement) with protons pointing up in the first water layer.. ................................ ................................ ... 141 5 33 Two water layers adsorbed on surface 6 (oxygen displacement) with protons pointing down. ................................ ................................ ................................ ... 142 5 34 Two water layers adsorbed on surface 7 (oxygen displacement) with protons pointing up in the frist water layer. ................................ ................................ .... 143 5 3 5 Two water layers adsorbed on surface 7 (oxygen displacement) with protons pointing do wn ................................ ................................ ................................ .... 144 6 1 Two examples of the initial configurations for surface hydroxylation.. .............. 15 2 6 2 The hydroxylation sites on surface 1 ................................ ................................ 152 6 3 Two types of hydroxylation on surface 1.. ................................ ......................... 153 6 4 Two types of hydroxylation of surface 2 with higher hydroxylation energy. ...... 154 6 5 Hydroxylation on surface 4 and s urface 6.. ................................ ....................... 155

PAGE 12

12 LIST OF SYMBOLS AND ABBREVIATIONS AFM A tomic F orce M icroscopy B88 Becke 88 BKS van Beest, Kramer, and van Santen CPMD Car Parrinello Molecular Dynamics DFT Density functional theory GEA Gradient Expansion Approximation GGA Generalized gradient approximation HSC Hamann, Schluter and Chiang LDA Local density approximation LEED Low E nergy E le ctron D iffraction MD Molecular dynamics NPT Constant particle number, system pressure, and system temperature NVE Constant particle number, system volume, and total energy NVT Constant particle number, system volume, and system temperature OPW Orthogonalized Plane Wave PAW Projector Augmented Wave PBE Perdew Burke, and Enzerhof PW91 Perdew and Wang 91 UHV Ultra High Vacuum VASP Vienna Ab initio Simulation Package

PAGE 13

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirement s for the Degree of Doctor of Philosophy A N AB INITIO STUDY OF ALPHA QUARTZ (0001) SURFACE AND WATER SILICA INTERFACE INTERACTION By Yun Wen Chen December 2010 Chair: Hai Ping Cheng Major: Physics Silica is an important mineral for many applications in our daily life Thus, numerous studies ha ve been done to understand its properties and its interaction with other materials /molecules In this work, we appl y f irst principles density functional theory (DFT) calculations to the investigation of alpha quartz (0001) surfaces and water cluster s and water layer(s) adsorbed on those surfaces Ten different surfaces are studied for exploring alpha quartz (0001) surface properties. They include five reconstructed surfaces, one fully hydroxylated surface and four defective surfaces. Water molecule ( s ) an d layer(s) were deposited on seven of the surfaces to examine how surface character can affect the behavior of water molecule(s) or layer(s) adsorbed on the surface. A new ly cleaved silica surface generall y is thought to be hydroxylated quickly in atmosphere and become hydrophilic. However, the surface can be dehydrate d under an Ultra High Vacuum (UHV) environment and the siloxane surface is then hydrophobic. We will discuss the five perfect surf ace reconstruction patterns of the alpha quartz (0001) surface (called perfect surfaces frequently in the following text) and how they can explain the observed (2 x 2) diffraction spectrum in experiments. A fully hydroxylated

PAGE 14

14 alpha quartz (0001) surface wi ll be discussed in detail regarding its formation energy, structure and charge distribution. Two oxygen vacancy and two oxygen displacement defective surface pattern s on (1 x 1) perfect surface were considered in this work. Their formation energies, struct ures, charge distributions and hydroxylation on the defect site s will be discussed. Water molecule adsorption on surfaces was investigated by depositing molecules one by one The h ydroxylated surface exhibits high ly hydrophilic propert ies. Conversely perf ect surfaces are hydrophobic or interact with water molecules quite weakly. A d efect on the surface do es enhance water adsorption on that surface H owever, the strong hydrogen bonds with in a water cluster have a tendency to compromise the water silica interaction Ice XI sub layer s were deposited one by one on surfaces The h ighly stable bilayer structures which result effectively shield the interaction between water and the silica surface.

PAGE 15

15 CHAPTER 1 INTRODUCTION AND BACKGROUND 1. 1 Two Important I ngredient s in Human Life Water and s ilica, two important ingredients involved in daily human life both play irreplaceable roles in many aspects of biology, chemistry, geology weathering etc. The water silica interaction is a topic of extensive scientific attention from ancient time s. I t is our research interest in this work. Water is definitely rec ognized as the most important ma terial affecting all living beings on the Ea rth because it is involved in many life essential chemical processes T he existence of water on a remote plan e t is always the first index of hypothesizing other life form s on said planet Of course, the value of water is not just in biology; it is involved in myriad phenomena happening around us (on Earth, at least). The daily weather we must accommodate, the erosion of Earth crust, the rotting of food and even the type s of volcanic eruptions are all examples wherein the character of water plays a role On the other hand, a clea n water resource is so important to human activities that we can say that water also affects phenomena happening in civilization and economics Scien tists have exerted themselves for over a hundred of years to und erstand the behavior of water and its interaction with other materials. Water in solid phase has multiple polymorphs under different pressure and temperature conditions [1, 2] Various crystal types built by t he complex network of hydrogen bonds ha ve draw n enormous scienti fic effort to investigate their properties and the research progress continues Salzmann et al. found a new ice phase, Ice XV, in 2009 by neutron diffraction [3] Ice XV i s the hydrogen ordered form of ice VI and has antiferroelectric property. The trouble

PAGE 16

16 makers, hydrogen bonds, are also responsible for the strange behavior of water in liquid and gas phase [4] ; superheated, supercooled water and supercritical fluid phenomena enticed researchers to repeat experiments again and again. Because most life essential chemical processes always involve liquid water in participating or mediating the interaction between water and other material s and the behavior of liquid wate r itself comprise one of the most important modern fields of study Another reason for investing so much scientific effort in this topic is the potential for numerous applications in biology and industry to make human live s better. Silica is one of the most abundant minerals i n the Earth crust I ts various polymorphs and many applications in industry ha ve attracted sustained research on this material Silica exists in various solid forms i n the Earth crust : quartz, cristobalite, tridymite, coesite, stishovite and also amorphous silica In principle they are constructed from rigid SiO 4 tetrahedron units with shar ed corner s [5, 6] H owever, each has its special pro perties in density, bulk modulus, dielectric con stant, etc Silica appear s not only as a mineral in nature but is also found in cell walls of diatoms T he amazing structure of the diatoms cell walls not only draws biologis compels scientists to study its application in nanotechnology [7] Human kind already use s silica in many applications such as a substrate in the semiconductor industry len se s, glass, silica gel and nano particles. Similar to water, silica has been studied for a very long time in human history S uch study becomes more conspicuous in contemporary time because of growing applications. Recently, for example, silica based device s ha ve been developed for DNA/RNA sequencing [8 10]

PAGE 17

17 Understanding the water silica interaction thus is an important subject to which many research groups ha ve devoted prodigious efforts. For understanding how erosion ial ap plications i t is important to know the mechanism of interaction between water molecules and silica, the key procedure to prevent silica erosion with coating, etc. It also helps to understand the formation process of a diatom cell wall, the etiology of silicosis [11] and perhaps the application of silica in DNA/RNA sequencing 1. 2 Silica Surfaces S ince at room temperature, silica is in solid phase and water is liquid or vapor phase for many bi ological and c hemical processes of high interest investigating silica surface struictures will be the first step to take Surface properties of crystalline silica, amorphous silica and amorphous silica combined with other chemical components all have been the subjects of study by experiments or theory Basically, these surfaces are defined as either hydrophilic or hydrophobic depending on how they are prepared. There are many strategies in chemistry that can change the hydrophilic properties of silica surfaces by coating or chemically terminat ing the surface with other molecules. In this work, we only consider surface s terminated with hydroxyl groups ( Si OH, or silanol group) and bare surface s with or without defects In atmosphere, a new ly cleaved silica surface w ill interact exothermic ally with water molecules and form hydroxyl groups because of the dangling bond left after cleavage T he resulting surface has been shown to be hydrophilic in many stud ies [5, 6] On the other hand, if the surface is carefully prepared in an Ultra High Vacuum (UHV) environment, the silica surface is dehydrate d and hydrophobic [12 16] Many investigations ha ve been done both experiment ally [17 19] and computational ly [20 22]

PAGE 18

18 to answer various questions about how water is adsorbed on silica surface, how weather ing phenomena proce ed etc. Because silica has so many polymorphs with different sy mmetries on each facet and the structures of silica surface s vary from one typ e to the other the surface properties could vary greatly among them Amorphous silica is the form with the most numerous applications I ts surface properties and defect populations are main object s for the experimental studies Yet a local description of the structure and feature of amorphous silica surface has not been completed by either experiments or first principle calculations. The c rystalline silica surface could provide a relative ly simple model for underst anding the effects of silica surface on wa ter molecules and the hydrogen bonding network. In this work we investigate several alpha quartz (0001) surfaces and water cluster/layer(s) adsorption on the surface. 1. 3 Water at I nterface s and Water Silica Interaction B ecause many natural phenomena and technological applications depend upon the interfaces of water and other materials, it is a prominent research topic The hydrophilic /hydrophobic properties of a surface can dramatically alter the hydrogen bonds network of water near the surface and ad d complexity to the numerous known anomalies in the water phase diagram [1, 2, 4] The interaction strength varies from weak bonding between hydrophobic surface s and water molecules to strong chemical bonding of hydroxylation of surface or partially dissociat ion of water adsorbed on the surface [23] Recently, w ater confined in a nano scale space has commanded intense scrutiny from rese a rchers, especially in biological science Castellana and Cremer reported that a water layer can exist between lipid bilayers and a solid substrate at a thickness of 10 to 20 The existence of this water layer is important for hydration of lipid s and for enhancing lipid bilayers stabili ty [24] Janiak and Scharmann reported a

PAGE 19

19 neutron diffraction study on nearly two dimensio nal water and ice layers confined between organic layers of nickel(II) chelate complexes [25] Temperature dependence studies on the water/ice layers ranging from 20 K to 278 K showed both liquid solid and solid solid transitions. Antognozzi et al. measured the viscoelastic forc e of confined water layers on mica surface with transverse dynamic force microsopy [26] They measured the relationship between amplitude and phase of the oscillating force and observed a step like behavior; the elastic and visco us forces of the water layer were evaluated. For water silica interface s many groups investigated the mechanism of weathering (hydroxylation) of a silica surface water adsorbed on a silica surface and water confined in silica nanopores in e xperiments and also theoretical simulation via density functional theory ( DFT ) or classical molecular dynamics ( MD ) methods. A summary of these methods is given in Chapter 2. D u et al. studied the hydroxylation of the silica surface with a combination of cl assical MD and DFT calculations [27] They conclude d that water will attack regions with high local strain Du et al [21] and Adeagbo et al [28] studied the dissolution of an Si(OH) 4 unit on fully hydroxylated alpha quartz (0001) surfaces by using classical MD and DFT calculations respectively. On the other hand, Du et al. [14] and Rignanese et al. [15] showed that the perfect ly reconstructed alpha quartz (0001) surface is hydrophobic ; further, Rignanese et al. showed this surface is very hard to h ydroxylate Asay and Kim studied water layer s adsorbed on a hydrophilic silica surface with attenuated total reflection infrared spectroscopy [17] They suggested that icelike water will grow up to three layers at room temperature. Optiz et al. measured the friction force

PAGE 20

20 of a water thin film on silica surfaces T he measurement showed dramatic ally different tendenc ies on hydrophilic and hydrophobic surface s with high and low friction forces respectively [29] Ostroverkhov and Shen et al. studied a series of phase sensitive sum frequency spectroscop ies of the water quartz inter f ace [18, 19] Through that technology, they analyzed the icelike and liquidlike peaks in the spectrum and the polar orientation of water at the interface for different pH value s Yang et al. found water could form an icelike layer on full h ydroxylated beta cristobalite (100) and alpha quartz (0001) surface s in their DFT calculations [22, 30] Several groups stud ied the density distribution, dipole orientation, and drift velocity of water molecule s on silica surfaces with classical MD [31 33] Their result show ed water layers forming on the silica surface, and some of them showed that hydroxylation condition s can greatly alter the behavior of water Many group s experimentally observed confined water in silica nanopores [34 36] T he melting point of water is lowered depending on the size of the nanopores. In principle there may be two or three types of water existing in the pores: quasi liquid (or non freezing) water layer with low mobility next to the silica wall, co re water at the central region of the pore and perhaps shell water between the former two types. The thickness of the quasi liquid water layer was not well determined in these experiments but it wa s suggested to be between one to three monolayers in most of the papers Liu et al. clearly suggested that there will be three types of water, core water, shell water and bound non freezing water [36] From their experiment al results they suggested th at non freezing water formed a bila yer film next to the pore wall when water completely filled the nanopores. Confined water in a silica tube also was studied with classical MD

PAGE 21

21 by some groups [20, 37, 38] Their observations agree with the proposed mod el for experimental ly confined water in silica nanopores Both bound non freezing water and core water exist in the silica tube model. So far, first principle studies on the adsorption of water on silica surface only focused on one or two water molecules, and one monolayer adsorption [14, 15, 22, 30] M ore effort is warranted to investigate water silica interfacial properties and answer a long standing question Is the re an ice or quasi liquid film at the interface and if so, what is the thickness of such a film [17] ?

PAGE 22

22 CHAPTER 2 COM P UTATIONAL METHODS In this chapter, we are introduce the computational methods used in this work. Cla ssical molecular dynamics (MD) and density functional theory (DFT) method s are included. 2. 1 Classical Molecular Dynamics 2.1.1 Newtonian Equations In MD, t he many body dynamics of the simulat ed sys tem are s e quation s of motio n. (2 1 ) Here and are the momentum, force and position vectors of atom i. is the potential energy of the system, which is a function of the positions of all atoms. The corresponding Hamiltonian is: (2 2 ) Here is the mass of atom i and N is the total number of atoms in the system. The first term, is the kinetic energy of atom i I t is related to system temperature via is the total number of degrees of freedom and is the Boltzmann constant. In principle, t he potential energy can be expanded into the sum of an external potential two body potential s three body potential s etc.

PAGE 23

23 (2 3) In practical application s only the external potential and first few low order interaction potentials are considered This choice arises both because of the capability of contemporary computational power and because the system usually can be well described without the higher order terms. In our classical MD simulations, we used two body potentials for pure silica system s and include d three body potentials as well if water molecules were involved. 2.1. 2 Computational Algorithm ( for NVE ensemble) e quations of motion computational ly it is necessary either make some approximation except in the rare cases that they can be solved analytically W hen simulating a system with hundreds or thousands of atoms in general, it is not feasible to solve such hug e linear combinations of equations exactly Thus, in the contemporary classical MD simulation they are solved in a discrete time sequence with a tiny time step Verlet [39] proposed a particular algorithm which in mod ified form has been wide ly applied in classical MD simulations. The original version has a problem of accumulating numerical error in the term I t ha s been modified in two ways to eliminate the problem. The first way is the Verlet Leapfrog algorithm [40] : (2 4) The second way is the Velocity Verlet algorithm [41] :

PAGE 24

24 (2 5) In above equations, and are the velocity and acceleration of atom i. These two algorithms are basically equivalent In both, the numerical error in the term is eliminated. In our classical MD simulations, we always use the Verlet Leapfrog a lgorithm. 2.1. 3 Extension to Thermostats and Barostats The algorithms just given are for the microcanonical ensemble They keep the total number of particles, volume of the system and total energy constant I f the chosen time step is small enough the total energy will display minute fluctuat ion s around the constant with almost un noticeable error. To make the simulation resemble practical experiments the algorithm should be extended to simulate ensembles with constant temperature and /or pressure (NVT and NPT) Two kinds of algorithms are implemented most widely in contemporary classical MD simulation packages. 2.1. 3.1 Berendsen t hermostats and b arostats As noted, the s energies via It is very straight forward t o modify the system temperature to a target value by just linearly rescaling the atomic velocities : (2 6) Here, is the target temperature, is the temperature at time t. However, this simple rescaling is c rude in that it does not correspond to any physical process of

PAGE 25

25 heating or cooling Berendsen [42] therefore introduced a modified to modulate the system temperature : (2 7) Here, is a constant which char acterizes the response time of the system to temperature changes If the system is a non interacting ideal gas, then it is easy to see that the analytical solution of will be: (2 8) This results from taking to be infinitesimal and takes The equation (2 7) now transform s into: ( 2 9) So the idea l in contact with a reservoir of constant temperature However, in the real simulation systems, the potential is not zero Over time the temperature will be modulated by the system potential as well and become much more If we consider how to change the pressure of a free ideal gas system and maintain the temperature the simplest way w ould be b y chang ing the volume of the gas because of the ideal gas law, PV=NRT namely (2 10) As a result, all of the particle coordinates will be scaled by (2 1 1 ) To mim he Berendsen barostat follows a similar concept

PAGE 26

26 to that of the Berendsen thermostat Specifically, it assumes that the instant pressure change is proportional to its difference with a pressure reservoir (2 1 2 ) Here, is a constant which characterizes the response time of the system to pressure changes 2.1.3. 2 Nos Hoover t hermostats and b arostats The idea of the Nos Hoover algorithm is more sophisticated In it, the variables representing the temperature or pressure reservoir als o fluctuate as well when heat is transferr ed between the simulat ion system and the reservoir. In the Berendsen algorithm, on the other hand, the temperature and pressure of the reservoir are kept constant and the transfer rate s are controlled by and only. Consider the thermostats first [43] T he central parameter i s changed to (2 13) Here, i s t he effective thermal mass of the reservoir and is a specified time constant that reflect s the response time of the reservoir The heat transfers to the simulation system by accelerating the atoms, (2 14) The result is an extended Hamiltonian which is a conserved quantity (2 15) This equation represents a combined Hamiltonian of the simulat ed system and reservoir

PAGE 27

27 For the barostat [44] and both are changed to read (2 1 6 ) Here and is the same as before. The equations for changing the system volume, atomic velocities and accelerations then are (2 17) The extended Hamiltonian (2 18) Is conserved. 2. 2 Density Functional Theory Density Functional Theory (DFT) is a powerful tool for investigating nano scale system s properties and behaviors in computational study today It replaces the problem of solving for the many electron ground state wave function with the problem of solving for the electron density distribution In practice, it produces structural parameters and system cohesive energies of high accuracy in many successful cases We will start from th e Hart r ee Fock approximation and then introduce Density Functional Theory. 2.2.1 H a r tree Fock approximation The Hamiltonian of a many electron system with in the Born Oppenheimer non relativistic approximation can be written as

PAGE 28

28 (2 19) Here the indices and are for ions i and j are for electrons. is the charge of ion and is the distance between the and ions. Similarly, is the distance of electron i from The electron mass and charge are and respectively. In the Born Oppenheimer non relativistic approximation, the ions are treated as point charges while the electrons are treated as charge d ensit y distributions which adjust instantaneously to the ionic positions. Under this approximation, the problem can be separate d into solving first for the electron ic ground state, then solving for the ionic motion The effective many electron Hamiltonian then is (2 20) in Hartree atomic unit s ( = = = 1). According to the Pauli Exclusion Principle the overall electron wave fun ction has to be antisymmetrized The simplest possible form is a Slater determinant of N single electron orbit al s, (2 21) Then the expectation value of the electronic Hamiltonian is: (2 22) Here means the distance from ion to the position In this expression, the electron electron potential energy is partitioned into two contributions. are called

PAGE 29

29 Coulomb potential energies and are called exchange potential energies. has the form of a classica l Coulomb interaction (or Hartree energy) wherein electrons i and j with charge distribution and interact with each other. However, originate s from the antisymmetr y of the overall electron wave function and has no direct classical analogue. It is esse ntial, however, for the stability of matter. 2.2.2 Kohn Sham Ansatz and Local Density Approximation Hohenberg and Kohn [45] prove d that th e electron density distribution alone determine s the system ground state and hence the positions of ions They also demonstrated that the ground state energy can be o btained by minimizing a suita ble electron density functional (2 23) Here is the total electron kinetic energy includes the e nergy from the interaction between electrons and ions and any other external potential and is the energy of electron electron interaction. The proof penned by Hohenberg and Kohn is really simple. Consider that there are two external potentials and which differ by only a constant. Suppose the two corresponding Hamiltonians and have two different ground state wave functions and These will lead to the same charge density since the two Hamiltoni ans just differ by a constant. If the ground states of two Hamiltonians are non degenerate and is not the ground state of and vice versa then the following inequality is true :

PAGE 30

30 (2 2 4 ) A correponding inequality can be derived for and then we get the contradictory inequality This means there cannot be two external potentials differ ing by only a constant which give the same non degenerate ground state density. Although Hohenberg and Kohn proved the existence of the universal electron density functi onal to determine the ground state, they did not give a construction. T he universal funct ional still needs to be determined. DFT did not become a feasible tool u ntil Kohn and Sham [46] proposed an ansatz to make the whole theory practical with two assumptions: 1. The exact ground state electron density can be found as the ground state electron density of a no n interacting reference system with an auxiliary potential 2. The auxiliary potential of the non interacting reference system is local. Hence, the Kohn Sham ansatz simplifies the ground state problem of the original many body system into an independent particle sys tem problem with an auxiliary potential In equation (2 2 3 ), the term can be decomposed into and is the functional for the Coulomb ic electron electron potential, is for the exchange energy, and is for the Coulomb ic correlation energy Compared with equation (2 2 2 ) the forms of and are rather straight forward. (2 2 5 )

PAGE 31

31 However, the explicit forms of the kinetic energy exchange energy and Coulomb ic correlation energy are not known in general The next step in the Kohn Sham ansatz is to introduce and rearrange the true kinetic energy (2 2 6 ) Combining the exchange and Coulomb ic correlation energ ies with the difference of and into the exchange correlation energy t he energy functional becomes: (2 2 7 ) All of the trouble in finding the universal functional is shifted over to the exchange correlation energy which will be expressed as t he auxiliary potential of a non interacting reference system However, u ntil now there is no exact expression for this part. Note that Kohn and Sham provided a method to solve the complicated many body equations by replacing an interacting V representable potential into a non interacting V representable potential It is known that this is not always possible. Fortunately, the success of Local Density Approximation (LDA) and Generalized Gradient Approximation (GGA) made the Kohn Sham ansatz the most widely used approach today in DFT calcu lations for solving electronic structure. The recent studies for improving the accuracy of DFT approximations are devoted to find better exchange correlation functional s which are feasible in a wide range of applications. The simplest success ful functional approximation is the Local Density Approximation (LDA). LDA simply represent s the exchange correlation energy

PAGE 32

32 as that of a homogeneous electron gas at each spatial point. T hat is, the exchange correlation potential energy just depends on the local density of charge. (2 2 8 ) Here and are spin and electron densities at position The exchange correlation potential can be further divide d into exchange and correlation part s The exchange potential of a homogeneous electron gas has an explicit form from the Hartree Fock approximation namely (2 29) The correlation part has been calculated to a high accuracy with Monte Carlo methods [47] 2.2. 3 Generalized Gradient Approximation The next step to improve LDA is to introduce density gradients into the exchange correla tion energy The idea of expanding the exchange correlation energy in terms of local density gradient (GEA, gradient expansion approximation) was suggested in the original Kohn and Sham paper and was developed by Herman et al. [48] However, the simple GEA does not improve upon LDA result s consistently and can even leads to worse ned results because it violates the sum rules and the large density variation i n real materials caus es the expansion to fail Instead, generalized gradient approximation s (GGA) were developed. A GGA exchange correlation energy can be express ed as: (2 3 0 )

PAGE 33

33 For the exchange energy, only the spin unpolarized form is necessary The enhancement factor is naturally expressed in terms of the dimensi onless reduced density gradients (2 3 1) The low order expansion of is known analytically [49] ( 2 3 2 ) N um erous forms of ha ve been proposed notably the widely used forms of Becke (B88) [50] Perdew and Wang (PW91) [51] and Perdew, Burke and Enzerhof (PBE) [52] In the range of 0 < s 1 < 3, these thr ee forms of have values close to each other This range corresponds to most of the environments in physical and chemical applications, and this is why different GGA forms give similar results in many cases (see the review of Martin [53] in section 8.2). But in the range of s 1 > 3, which is relevant to large density gradient s the three forms converge to differ ent limiting values with different choice of physical conditions. B88 GGA chooses to give the correct exchange energy density, PW91 GGA and PBE GGA exchange obey the Lieb Oxford bound but PBE GGA approaches a constant for large s 1 while PW91 GGA decreases to zero asymptotically. The differences occur because not all asymptotic constraints on can be satisfied by a GGA. One form successfully predicts the energy and structure of some systems, but may fail in other cases. Correlation energy is ty pically much smaller in magnitude than the exchange energy but more difficult to express. The lowest order expansion has been calculated by Ma and Brueckner [54]

PAGE 34

34 ) ( 2 3 3 ) At large density gradient limit, the correlation energy diminishes to zero as 2.2. 4 Pseudopotential The motivation for using with a pse udopotential is to replace the problem of solving the all electron Kohn Sham problem with either LDA or GGA with the task of diagonalizing only an effective valence Hamiltonian The latter task needs smaller basis expansion s Since in most of the environment s (either in a molecule or in a solid), the distribution of core electrons is almost unchanged com par ed to the distribution in an individual atom, only the valence electrons will contribute to chemical bonding or Van der Waals interactions. An important idea for si mplifying a n one electron wave function is the orthogonalized plane wave (OPW) approach [55, 56] It decomposes the eigenvalue problem by separating the valence wave functions into smooth parts and core like functions. Consider a wave function and a plane wave The n a wave function orthogonal to each can be formulated as (2 3 4 ) Here, is the volume of the integration space. From the Kato cusp condition, an one electron orbital will approach to zero in a form of (n, Z, and a 0 are the principal quantum number atomic number and Bohr radius) and the radial part of the orbital has high radial derivatives localized near the nucl eus If the are chosen to be well localized, then the wave function can be partitioned into a smooth part and a

PAGE 35

35 localized part. Herring [55] suggested that the eigenfunction in a crystal can be approximated by a linear combination of a few plane waves, plus a linear combination of a few functions centered around each nucleus and obeying the wave equations: (2 3 5 ) Note that the OPWs are not orthonormal: (2 3 6 ) Thus, expansion of orbitals in OPWs will lead to a generalized eigenvalue problem with a non diagonal overlap matrix. In modern DFT calculations, several popular forms of pseudopotential are available, like HSC [57] Troullier Martins [58] Ultrasoft [59] and PAW [60] Different type s of pseudopotential s follow from different criteria, thus pseudopotential s are classified as hard or soft, and as transfer able or not. Hard or soft means that the pseudo wave functions must be expanded in plane waves including many high frequency terms or just a few low frequency terms respectively. If the pseudopotential is suitable for every environment, then it is transferable Usually, a harder pseudopotential is more transferable than a soft one In general, there are two important aspects to consider in constructing a pseudopotential. O ne is shape consisten cy and the other is norm conserv ation In the paper of Ha mann, Schluter and Chiang (HSC) [57] their pseudopotential obeys the following properties (in their original words) : 1. Real and configuration. c

PAGE 36

36 3. The integrals from 0 to r c of the real and pseudo charge density agree for each valence sta te (norm conservation). 4. The logarithmic derivatives of the real and pseudo wave function and their first energy derivatives agree for r > r c Here r c plays an important role for offering flexibility to make a pseudopotential harder or softer. It will be easy to choose a large r c and keep norm conservation, since atomic wave functions fluctuate less at large distance from the nucleus However, the price for choosing a large r c and making the pseudopotential soft is that the pseudopotential is made non tr ansferable. Ultrasoft and PAW pseudopotential s are two powerful and widely used forms implemented in [61] The u ltrasoft pseudopotential does not respect norm conservation criterion because it uses a larger r c However, it uses auxiliary functions and overlap operators to maintain high accuracy. On the other hand, the PAW method reformulates the O PW s equations and keeps the full wave function s with some auxiliary transformation.

PAGE 37

37 CHAPTER 3 ALPHA QUARTZ (0001) SURFACES Alpha quartz, the most stable structure among numerous silica polymorphs, has its SiO 4 tetrahedra units arra nged in a trigonal symmetric crystal Am ong facets of alpha quartz, the (0001) surface is the most stable one. A new ly cleaved alpha quartz (0001) surface will be hydroxylated quickly under atmosphere condition and is hydrophilic. However, if the surface is prepared under UHV condition s the sur face can reconstruct itself and result in a hydrophobic surface. In section 3.1, we present our study on the r econstruction of alpha quartz ( 0001 ) surfaces using combined classical molecular dynamics and density functional theory. Five reconstruction patterns are identified, including three (2x1) patterns and two (1x1) patterns [62] The energetically most stable surface structure is found to be a (2x1) reconstruction pattern, though several patterns can coexist on a large scale surface. A combination of structures can explain the experimen t ally observed (2x2) diffraction pattern [62] In section 3.2, we analyze a full y hydroxylated alpha quartz ( 0001 ) in respect to its structure and charge distribution The system is found to form alternati ng strong and weak hydrogen bond chain s within the surface. In section 3.3, we study four defective alpha quartz ( 0001 ) (1x1) surface s Two of them have non equivalent oxygen vacancy point defect s, while the other two have non equivalent oxygen displacement point defect. Their structure and charge distribution are discussed.

PAGE 38

38 3. 1 Perfect Surface Reconstruction 3 1 1 Introduction S ilica is an important mineral that has wide application in various fields in industry [5, 6, 63] Due to the close relationship between the structure of a surface and its chemical activit y it is crucial to determine the silica surface structure accurately so that one can understand its interactions with other molecules, such as water, oxide compounds, or proteins [5, 6] Although as mentioned, a newly cleaved silica surface will be hydroxylated rapidly when it is exposed to atmospheric moisture [5] a dry surface (not hydroxylated) can be prepared either in vacuum or via heat treatment. The activity of dry silica surfaces is considered to be important for its biological toxicity [11] Among all different dry surfaces, the alpha quartz ( 0001 ) surface has been widely studied in experiments and theoretically [12 14, 16, 64 67] because of its stability. Previous low energy electron diffraction ( LEED ) experiments claimed a ( 1 x 1 ) pattern for alpha quartz ( 0001 ) surfaces [12, 64] which undergoes a ( reconstruction with a R 11 rotation at a temperature above 600 C [12] A dense surface with a ( 1 x 1 ) pattern ( which from now on we call surface type I ) identified and studied using density functional theory ( DFT ) calculations [14, 65, 66] or classical molecular dynamics ( MD ) calculations [13, 14, 66] can explain the ( 1 x 1 ) pattern in the LEED experiments well In recent helium atom scattering experiments, Steurer et al [16] observed weak ( 2 x 2 ) diffraction peaks, which also app ear in the LEED pattern of Bart and Gautier [12] but were not explicitly mentioned. On the other hand, atomic force microscopy ( AFM ) studies of the surface

PAGE 39

39 failed to give further detail about the surface structure due to its limited resolution [16, 67] 3 1 2 Method This section present s a combined classical MD and first principles study of the alpha quartz ( 0001 ) surface reconstruct ion. We find five stable surface patterns with similar surface energies. Two of them are dense surfaces with ( 1 x 1 ) patterns while the other three possess ( 2 x1) patterns A ll are different from the semi dense surface in the work of Rignanese et al [65] Classical MD simulations [68] using the potential [69] developed by van Beest, Kramer, and van Santen ( BKS ) were employed to provide a reasonable initial approximation of the reconstructed surface structure. To simulate the surface reconstruction, a 4 x 4 x 5 supercell a nd 720 atoms were included in MD simulations. The resulting structures have either a ( 1 x 1 ) pattern or a (2x 1 ) pattern in the X Y plane These are truncated into a supercell of 2 x 2 x 5 for further optimization via DFT calculations. We use d the VASP package, which employs a plane wave basis set and the projector augmented wave potential [60, 61] The PW91 generalized gradient a pproximation [51] ( GGA ) wa s used as the exc hange correlation energy functional A kinetic energy cutoff of 400 eV and a 2 x 2 centered K grid were used to ensure energy convergence. The lattice constant was optimized in bulk simulations The result is in good agreement with the experimental value ( see Table 3 1) To form the surface t he system wa s cleaved between two oxygen sublayers to maintain stoichiometry T his is done for both top and bottom surfaces, and con sequently the supercell has no dipole moment. A vacuum layer of 50 wa s inserted in the z direction between two adjacent

PAGE 40

40 slabs. We fix ed the middle of the slab ( 1 unit cell thick ) at a geometry which mimic s the quartz bulk and let the top and bottom layers relax ( each side two unit cell s thick ) 3 1 3 Results The type I surface observed at low temperature ( T ~ 100 K ) is obtained in a constant particle number, system volume and temperature (NVT) ensemble in classical MD calculations ( Figure 3 1 (a) ). To generate other surfaces, we heat ed the system ( using the Berendsen thermostat [42] ) with relaxati o n constant 0.1 ps f rom 100 to above 1400 K with an increment of 100 K and a duration of 20 ps at each stage. For each 100 K, we perform ed a separate calculation to quench the system to 0 K and analyze the structure. Three more surface structures were identified via this procedure ( type III, IV and V ) The t ype II surface is constructed by manual tuning the orientations of the topmost layer SiO 4 tetrahedra ( Fig uire 3 1 ( b ), ( c ) and 3 2 ) Type I and II surface s can transform between each other through rotating all SiO 4 tetrahedra in the topmost layer ~ 35 about the z axis, as shown in Figure 3 2 (a) The underlying alpha quartz bulk feature ( right hand ed helix ) distinguishes these two as different surfaces without ambiguity. All five surfaces feature six membered rings in the topmost l ayer and three membered rings right below it, and have the same topological character. The stability of these surface structures is confirmed through DFT optimization calculations until the total energy converges within 0.2 meV. Their surface energies, given by = ( E surface E bulk ) / area, are listed in Table 3 2. The statistic al variations among the five surfaces (Figure 3 3) show that the changes in O Si O bond angles and Si O bond lengths are less than 4%, but the Si O Si bond angles reach 18%. This analysis suggests that the

PAGE 41

41 SiO 4 tetrahedra are rigid, and that the different surface structures can be regarded as rotations of SiO 4 tetrahedral units. The type III surface has the lowest surface energy, 27.3 meV/ 2 in our calculations. It features zigzag shaped six membered rings in the topmost layer and three membered rings underneath ( Fig ure 3 1 ( b ) and ( c ) ). As shown in Fig ure 2 (b) the type III surface can be regarded as rotating the topmost SiO 4 tetrahedra of the type I surface in certain lines along the [ 110 ] direction. As a result, the type III surface undergoes a ( 2 x 1 ) reconstruction that has two unit cell patterns in the [ 100 ] and [ 010 ] directions, but one unit cell pattern in the [ 110 ] direction. The type IV surface has a surface energy similar to type I and II. It can be regarded as rotating the SiO 4 tetrahedra in the topmost layer of type I surface in a similar way as for the type III surface, but along the [ 100 ] direction. Correspondingly, the type IV surface undergoes a different ( 2 x 1 ) reconstruction pattern that has a one unit cell pattern in the [ 100 ] direction and two unit cell patterns in the other two directions. The type V surface also has a ( 2 x 1 ) reconstruction pattern but has the highest surface energy 34.3 meV/ 2 in the series of surfaces. It can be regarded as rotating the topmost SiO 4 tetrahedra of type I surface along [ 010 ] direction in a similar way as wit h the type III and IV surfaces. Hence, it has a one unit cell pattern in the [ 010 ] direction and a two unit cell pattern in the other two directions. Even though the types III V surfaces have very similar ( 2 x 1 ) reconstruction patterns, the underlying alpha quartz bulk feature ( right hand ed helix ) distinguishes their orientations ( Fig ure 3 1(c) )

PAGE 42

42 3 1 4 Discussion and Conclusion To determine the energy barriers between surface structures, we perform ed nudged elastic band calculations [70] in VASP with five intermediate images The results are listed in Table 3 2 The barrier from type II to III is the lowest, 5.67 meV/ 2 relative to type II. The number of degree s of freedom ( N f ) the barrier height E and the temperature T of a structural transition are related by N f /2 k B T = E where k B is the Boltzmann constant. As we mentioned above, the SiO 4 tetrahedra are essentially rigid therefore the rigid unit mode model [71] can be used. In our simulations, 24 tetrahedra ( 56 O and 24 Si ) are involved in the structural change. However, 24 Si and 4 O atoms remain rigid and all interatomic bonds are unchanged. This leads to N f =12. The energy barrier corresponds to a transition temperature of 913 K. Steurer et al. [16] proposed that the combination of the dense surface and semi dense surfaces [65] can explain the ( 2 x 2 ) observed diffract ion patterns. However, the transition from a dense surface to a semi dense surface will include a bondbreaking process since the semidense surface has three membered rings in the topmost layer hence has di fferent topological cha racter from the dense surface Therefore this transition will occur only at a much higher temperature. Conversely the transformations between the five surface patterns in our study involve no bond breaking process, similar to the alpha beta incommensurate phase transform in which a quartz twinning phenomenon was observed. [72] Bart and Gautier [12] also explained the observed ( x ) surface reconstruction pattern using the quartz twin argument. Since the surface energies of these five surfaces are very close to each other, coexistence of several surface types is very probable, as seen in our large scale classical MD

PAGE 43

43 calculations ( a 10 x 10 x 5 ) supercell with 4500 atoms ). In a MD simulation, types I III surfaces coexist at 400 K (Fig. 3 4 ). On the one hand, the coexistence of surface types III V provides a good explanation of the ( 2 x 2 ) pattern in experiments in a defect free condition. On the other hand, Steurer et al [16] observed flat terraces oriented 60 relative to each other in their AFM image, which can be explained by different steps in alpha quartz due to imperfect cutting of the surface. Such steps offer three different surface orientations automatically T herefore a combination of type III surfaces along three orientations also can explain the experimental results. Based on energetic, we argue that the latter explanation is more convincing. Due to computational restrict ions, it is not feasible to examine the stability of such large scale structure in a DFT calculation. In summary we have studied the reconstruction of alpha quartz ( 0001 ) surfaces through DFT and classical MD calculation. Three surfaces with a ( 2 x 1 ) patte rn and two surfaces with a ( 1 x 1 ) pattern are found in our calculation. Two of the ( 2 x 1 ) pattern surfaces are energetically more stab le than the dense surfaces. The experimentally observed ( 2 x 2 ) pattern can be explained by combinations of ( 2 x 1 ) surfaces. 3. 2 Full y Hydroxylated Surface As noted earlier, a new ly cleaved silica surface will be hydroxylated quickly under atmosphere with a var ing degree of hydrophilic ity depending on surface preparation [5, 6] Multiple groups ha ve studied the mechanics of silica surface hydroxylation by using DFT or Classical MD. Du et al. combined two to study possible hydroxylation pathways of the amorphous silica surface [27] The y concluded that the high ly strained sites such as two member ed rings are much eas ier to interact with water T hey mapped the energy landscape along the interaction pathway and observed a barrier free hydroxylation process. Adeagbo et al. used Car Parrine llo molecular dynamics (CPMD)

PAGE 44

44 combin ation with a coordination constraint to study the dissociation o f one Si(OH) 2 unit on full y hydroxylated alpha quartz (0001) surface [28] The calculated energy barrier height is about 186 kJ/mol which is roughly comparable to experimental data if app ly a temperature dependence with an Arrhenius slope Ma et al. [73] and Rimola et al. [74] studied the hydroxy lation process of silica surface by using silica cluster models. They both observed that the presence of more than one adsorbed water molecule could accelerate the hydroxylation of a defective surface. The full y hydroxylated alpha quartz (0001) surface had been studied by Murashov [75] and Yang et al. [22] with DFT Yang et al. found that the hydroxyl groups (sil anol groups) will form hydrogen bond chain s with alternat ng strength on the surface itself H find the same phenomenon : the weak hydrogen bonds were not form ed Yang et al. had reported the geometry of this surface and also studie d the adsorption of water molecule s and one water layer on it. In this section, we will only report how the full y hydroxylated alpha quartz (0001) surface is prepared for the water adsorption study which will be discussed in the following chapters and als o give a charge analysis of the surface. In our simulation, t he new ly cleaved silica surface has a 2x2x5 supercell and is cut in a way that maintain s its stoichiometry on the top and bottom surfaces as described in section 3.1.2. We hydroxylated the newly cleaved surface by attaching a proton to each dangling oxygen bond and a hydroxyl radical to each dangling silicon bond. The structure is optimized by using DFT calculations with the same parameters used in section 3.1.2 ; the cent er layer of the silica slab is still fixed during optimization

PAGE 45

45 The optimized structure has a (1x1) symmetry pattern and the hydrogen bond chain s with alternati ng strength form on the surface (Figure 3 5) The hydroxylation energy is 14.5 meV/ 2 calculat ed from E surf = ( E tot E AQ n E H 2 O )/ area. Here, E tot is the total energy of the surface system; E AQ is the Alpha Quartz bulk energy with a 2x2x5 supercell and E H 2 O is the energy of a single structure optimized water molecule n is 8 and the value of the area is 169.8 2 for both top and bottom surfaces in total. The calculated hydroxylation energy is larger than in absolute value( 10.0 meV/ 2 ) [75 ] T his is reasonable since weak hydrogen bonds form in our model but not Yang et al. did not give the hydroxylation energy in their work for compar ison with The hydrogen bond lengths (OH distance s ) of strong an since we use a larger lattice constant set The charge on each atom wa s calculated by using Bader analysis with a program developed by Henkelman et al. [76, 77] The silicon atoms on the topmost layer have an average charge similar to silicon atoms on perfect surface, but the average charge on those oxyg en atoms in hydroxyl group is obviously smaller than that of oxygen atoms on perfect surfaces (Table 3 3, 3 4). We note that the charge on hydroxyl groups has slight vari ation depending on its proton is involv ed in a strong or weak hydrogen bond. Table 3 4 gives a summ ary of the charge analysis. 3. 3 Defective Surface s Defects in silica (especially in amorphous silica) draw attention because they affect many properties : stiffness, transparency, refraction index, generation of additional defects etc [78, 79] In optics, high energy radiation and astronomy, equipment with lens es demand high accuracy and lo ng lifetime Recognizing and controlling the number

PAGE 46

46 and properties of defects is important. In the semiconductor and telecommunication business, the quality of wafer s optical fiber s and optical lens es as well also depen d upon by the defect population in silica. Researchers in biochemistry and the medical fi el d are concern ed with the quality of glassware and its chemical activity with respect to other atoms/molecules. The point defect types of silica have been studied and classified for years [78, 79] R oughly they can be categorized as Frenkel defects, dangling bonds, strained bonds and d opants [79] Each category includes more detailed classification of defect types and each has its special properties and population in different silica form s Among those defect types, the center (dangling silicon bond), the oxygen vacancy and the NBOCH (dangling oxygen bond) ha ve been studied extens ively and are much better understood than are other defects. Many investigations ha ve been devoted to the optical adsorption spectra in different charge state, defect formation energy, defect structure and the diffusion of defects [80 87] In this section, we report our study of the point def ect on the (1x1) perfect surface (surface type 1 in section 3.1) T hose surfaces will be involved in the study of water adsorption in subsequent chapters. Two types of point defect are studied and two surfaces with non equivalent point defect site are prep ared for each type. The first type of defect is an oxygen vacancy which is made by dropping one bridge oxygen atom from the surface. Depending on the environment of the oxygen vacancy, two kinds of defective surface are prepared. Defective surface 1 has the oxygen vacancy involved in a three member ed ring while defective surface 2 does not (Figure 3 6, 3 7). The second type of defect is oxygen displacement, made by moving one bridge oxygen

PAGE 47

47 atom to the other bridge oxygen site next to the original one. In another word s it consists of one oxygen vacancy and one peroxy linkage [79] Depending on the defect environment, there are three non equivalent surfaces H owever, in this work, we treat just two of them The oxygen vacancy on each surface is not involved in a three member ed ring Defective surface 3 h as its peroxy linkage involved in a three member ed ring and defective surface 4 does not (Figure 3 8, 3 9). To reduce the spurious interaction between defects under periodic boundary condition s the size of the quartz slab wa s increase d compared to previou s calculations to contain a 3x3x5 supercell T otal of 403 or 405 atoms are included as a result The center layer still was fixed to mimic the alpha quartz bulk A ll of the DFT calculation parameters used wer e the same as in section 3.1.2. The optimized surface structures are shown in Figure 3 6 to 3 9 Table 3 5 lists the defect formation energ ies It is necessary to consider the defect involved in a three member ed ring because the local strain will be higher in such ring s than in six memb er ed ring s Prior studies showed that water will more likely attack the high st r ain sites [27, 74] E ven though they were investigating the high strain site s of two member ed ring s it is important to consider the effect of a three member ed ring on the formation of defects and the subsequent effect on water adsorption. The defect involved in a three member ed ring do es enhance the formation energies (less stable) when compar ed to its counterpart of the same defect type but the energy difference between surface s 1 and 2 is not so big as between surface s 3 and 4 (Table 3 5) Note that the oxygen atoms in the peroxy linkage o n defective surface 3 are not at the same height s in Z direction for they are involved in a three member ed ring with a higher strain than the peroxy linkage o n

PAGE 48

48 defective surface 4 Charge transfer analysis is given in Table 3 6 Silicon atoms at the defect site general ly gain 0.77 to 0.79 more electrons compared to the silicon atoms on the (1x1) perfect surface This charge is ga ined from other atoms on the surface and bulk as well. On defective surface s 1 and 2, the charges on the oxygen atoms around the defect do not differ from those on the (1x1) perfect surface so much On the other hand, the peroxy linkage on surface 3 and 4 tend to lose electrons Those losses compensate for the electrons gained by silicon atoms of the oxygen defect site

PAGE 49

49 Table 3 1. Experimental and calculated structure of bulk alpha quartz Experiment GGA PW91 a(c) () 4.916 (5.405) 4.950 (5.445) d Si O () 1.605 1.623, 1.627 < Si O Si (deg) 143.73 142.83, 142.96 < O Si O (deg) 110.52, 108.81, 108.93, 109.24 110.72, 108.66, 108.75, 109.22, 110.67 Table 3 (in meV/ 2 ) of five reconstructed surfaces and the freshly cleaved surface, and the energy barriers (in meV/ 2 ) for transition from type I ( E I b ) or type II ( E II b ) surface to others. The transition between surface type I and II is mediated by type III, and is therefore not listed. Type I Type II Type III Type IV Type V C leaved 31.0 30.8 27.3 29.8 34.3 166.6 E I b N/A N/A 5 .71 6.16 7.55 E II b N/A N/A 5 .67 6.57 8.67 Table 3 3 The average charge on each atom species of surface type I, type III and quartz bulk. The unit is in electron. Si O Bulk Surface Bulk Surface Surface Type I 0.81 0.84 7.59 7.59 Surface Type III 0.81 0.83 7.59 7.60 Quartz Bulk 0.81 7.60

PAGE 50

50 Table 3 4 The average charge on each atom species of a hydroxylated quartz slab The unit is electron s Si Si O O O H H Bulk Surface Bulk Strong acceptor Weak acceptor Weak donor Strong don o r Hydroxylated surface 0.8 0 0.8 3 7.60 7.45 7.50 0.32 0.3 0 Table 3 5 The formation energy of a defective surface with respect to (1x1) perfect surface Here is the total energy of the defective surface system, is the total energy of (1x1) perfect surface. is the energy of one oxygen atom. n is 1 on defect 1 and 2, it is 0 on defect 3 and 4. The unit is eV. Defect 1 Defect 2 Defect 3 Defect 4 E defect 8.10 8.02 7.16 6.48 Table 3 6 The average charge transfer onto Si/O atoms at and around the defect site. The se are calculated with reference to the average charge s of Si and O atoms on (1x1) perfect surface (Table 3 3). The positive numbers mean electron gain and negative mean electron loss In Figure 3 6 to 3 9, t he atoms included in the calculation of average charge transfer are labeled with the same mark as in th e parentheses in the first column of this table. Defcet 1 Defcet 2 Defcet 3 Defcet 4 Si(*) 0.79 0.77 0.78 0.77 O(+) <0.01 <0.01 0.32 0.31 O peroxy linkage N/A N/A 0.79 0.77

PAGE 51

51 Fig ure 3 1. Top and side views of reconstructed alpha quartz (0001) surfaces. The larger spheres represent silicon and the smaller ones oxygen. The positions of some atoms in three membered rings are circled. The arrow shows the [110] direction. a) Top view of a type I surface. b) T op view of a type III surface. c) Side views of the surfaces. From top to bottom: type I and II; type III; type IV; type V. The blue circles in the top views indicate the location of three membered ring underneath. Fig ure 3 2. Transformations from one surface type to another. a) From type I (left) to type II (right); b) from type I (left) to type III (right). The circled atoms are shifted along the [ ] direction.

PAGE 52

52 Fig ure 3 3. S tatistic s of bond length and bond angles of all perfect surfaces a) Si O bond length distribution. b) O Si O bond angle distribution. c) Si O Si bond angle distribution. Fig ure 3 4 Top view of the large scale structure of a reconstructed alpha quartz surface from MD simulation. The sha ded areas are types I (pink) and II (blue) as indicated in the figure, and the unshaded area is type III. A 10x10x5 supercell is included in this system.

PAGE 53

53 a b Fig ure 3 5 Top view (a) and side view (b) of a full hydroxylated alpha quartz (0001) surface. The silicon and oxygen atoms in the second level are represented in thinn er lines in the pa nel b. One of the hydrogen bond chains on the surface is shown in blue dash ed lines; dark blue means stronger hydrogen bonding and light blue is for weaker bonding. The big blue arrow indicates the direction of side view.

PAGE 54

54 a b Fig ure 3 6. Top view (a) and side view (b) of detective surface 1 (oxygen vacancy). The defect site (Si Si) is presented as yellow color ed sph eres. a b Fig ure 3 7. Top view (a) and side view (b) of detective surface 2 (oxygen vacancy).

PAGE 55

55 a b Fig ure 3 8. Top view (a) and side view (b) of detective surface 3 (oxygen displacement). a b Fig ure 3 9. Top view (a) and side view (b) of detective surface 4 (oxygen displacement).

PAGE 56

56 CHAPTER 4 WATER CLUSTER ADSORPTION ON SILICA SURFACES Water molecules wer e deposited one by one on seven of alpha quartz (0001) surfaces. These were the (1X1), (2X1) perfect surface (type I, type III surface in Chapter 3) the fully hydroxylated surface and four defective surfaces For convenience, they are named as surface s 1 through 7 in the discussions and in Chapter 5. T he structure s of one, two and three water mol ecules adsorbed on surfaces were optimized by DFT calculations The adsorption energies and bonding energ ies of water for all these systems were calculated to reveal the interaction strength of the water silica interface The charge densit y difference of each system was calculated to serve as an indicator for the location of the water silica interaction Th ough the net charge transfer between water and a silica surface is tiny the charge redistribution is obvious. On perfect surfaces, the bonding energies of water dimer adsorption were obviously enhanced as compared to one or three water molecule adsorption and it exhibits a quadrup o le pattern in charge density difference on ea ch perfect surface In Chapter 4 and Chapter 5, the methods section skips most of the details of using classical MD and DFT because most of the parameters and criteria are the same as in Chapter 3. We will only mention the equation s and the changed paramet er s if it is necessary 4.1 Introduction Asay and Kim studied water layers growing on a hydrophilic silica surface and suggested that there will be about three icelike layers built on a silica surface [17] Ostroverkhov and Shen et al. measured phase sensitive sum frequency spectroscopy of the water quartz interface They observed both icelike and liqui d like peaks in the

PAGE 57

57 spectra [18, 19] Water confined in silica nanopores ha s been studied experimentally by many groups [34 36] I n the theory, it is belie ved that two or three types of water exist in the pores. A q uasi liquid layer with low mobility stays next to the silica wall, core water with bulk water like behavior stays in the central region and there may be shell water existing between the q uasi liq uid layer and core water. M any groups did computational studies of the hydroxylation mechanism of silica surface s [27, 28, 73, 74] water molecule /layer adsor ption on silica surface s [14, 15, 22, 30, 88] and the statistical behavior of bulk water when it is adsorbed on a silica surface and fil l s in a silica nanopore [20, 21, 31, 32, 37, 38, 89] After those studies, i t wa s recognized that surface s having hydroxyl group s ( Si OH, silanol group s ) are hydrophilic a nd siloxane surface s (bare silica surface s ) are hydrophobic. Silica s urface s with defects of dangling bonds and high local st r ain are hydroxylated easily, while a dry surface with perfect reconstruction (like surf ace 1 and 2) will be hydrophobic and hydroxylation is not energetically preferred [ 15] Yang et al studied water molecule(s)/layer adsorption on a fully hydroxylated alpha quartz (0001) surface with DFT methods and found a n icelike layer to form the surface [22, 30, 88] First principle studies o f the water adsorption on silica surface s had only been done for one or two water molecules and one monolayer adsorption More effort was need ed regarding multi water layer adsorption on more species of silica surface s Thus w e studied water adsorption on seven alpha quartz (0001) surfaces This Chapt er 4 addresses the water molecules adsorption and Chapter 5 treats water layers adsorption.

PAGE 58

58 4.2 Methods We use d c lassical MD to look for possible adsorption sites on surface s for water molecule s then optimize d those target structure s by DFT calculations. Instead of using the BKS potential [69] used in Chapter 3, we use d the CLAFF interac tion developed by Cygan et al. [90] which allow s the simulation of a system with coexisting wa ter, hydroxyl groups and silica. In this chapter, the adsorption and bonding energies are calculated according t o ( 4 1 ) ( 4 2 ) In equation (4 1) is the adsorption energy of the m th water molecule, is the total energy of the optimized system with m water molecules adsorbed, is the total energy of the optimized system with ( m 1) water molecules adsorbed, and is the energy of one optimized water molecule. In equation (4 2), is the bonding energy of the m water molecule cluster, and is the energy of the individual surface o r water cluster subsystem which has the same configuration as in the optimized total system. The charge density difference is calculated with the definition ( 4 3 ) Here, and represent the total charge of the optimized system, the charge of the unrelaxed surface only and the charge of the unrelaxed water cluster only. The charge redistribution is defined as the sum of the absolute value of charge density difference

PAGE 59

59 ( 4 4 ) Here, the indices i, j and k are indices of real space grids in DFT calculations. 4. 3 Water Adsorption o n Surface s 1 and 2 ( (1X1) and (2X1 ) P erfect S u rface ) When one water molecule is adsorbed on a perfect surface two kinds of adsorption sites are observed in the calculation O ne is the silicon atom site, the other is the hollow site. In our calculat ions, one water molecule adsorb ed on a hollow site is preferred more than on a silicon atom site on both surface 1 and 2. This result is different from what Du et al. reported [21] T hey claimed that the silicon site is the most stable adsorption site. In our calculations, t he bonding energy on the hollow site is 18.8 meV, 2.9 meV lower than on the silic on atom site on s urfaces 1 and 2. The adsorption energies are smaller than bonding energies (Table 4 1) because of the energy cost for deformation of the perfect surfaces and the water molecule. The bonding energy is only about 19 % of one hydrogen bond in Ice XI (350 meV in DFT calculation) [91] In Fig ure s 4 1 and 4 2 we show the side view s and top view s of the relaxed structure s as well as the charge density difference The charge transfer from the water molecule to each surface ( Table 4 2 ) is small, but the charge redistribut ion is larg e (0 .45 e on surface 1 and 0.31 e on surface 2 Table 4 3 ) and concentrat ed around the water molecule and the adsorption site. The charge density d ifference indicate s that the proton(s) of water is/are attract ed to the bridge oxygen atom(s) on the silica surface s The bonding energy, charge transfer and charge redistribution values all indicat e that surface 1 has a stronger interaction with a water molecule (the difference between bonding energies is tiny) Howev er, the adsorption energy of one water molecule on surface 1 is smaller than for surface 2. This is because surface 1 is less stable than

PAGE 60

60 surface 2 and the deformation of surface 1 is larger when adsorb ing a water molecule on it. On the other hand, the cha rge density difference for one water molecule adsorb ed on the silicon site of surface 1 ( Figure 4 3) show s another kind of charge redistribution. The o xygen atom in the water molecule gains electron s and the silicon atom on the adsorption site los es electron s. The charge transfer to the water molecule is smaller than for the hollow site adsorption ( 1.4 x E 3 e ) and with opposite sign. The charge density difference indicate s that the oxygen atom of water is attractive to the silicon atom on the surfa ce and that the hydrogen atom has an interaction with the bridge oxygen atom on the surface as well T he charge redistribution is smaller (0.31 e) than for the hollow site adsorption. When the second water molecule is adsorbed on the surfaces, the first water molecule stay s on the hollow site and the second one is absorbed on the silicon atom site with one hydrogen bond formed between two water molecules (Figure 4 4, 4 5) The bo n ding energy is greatly enhanced compar ed to single water molecule adsorption ( Table 4 1 ) so that the average bo n ding energy is also much higher than E b1w on both perfect surfaces C onsideration of the adsorption energy and bonding energy ( E a2w and E b2w ) indicates that the water silica interaction is comparable with the strength of one hydrogen bond in Ice XI ( E b2w on either surfaces is about 50 to 60% of one hydrogen bond of Ice XI). The charge density difference show s a much larger range of redistribution s on both surface 1 and 2 than for one water molecule adsorption The charge density difference is characterized by a quadrup o le like distribution the alternating enriched and depleted zone s l ying in the Z direction (Figure 4 4 (c) ,(d) and 4 5 (c), (d)). T his special phenomenon is observed in water layer(s ) adsorption on silica

PAGE 61

6 1 surface s also and it could be the reason for the enhanced water silica interaction The charge transfer on water molecules is still small (Table 4 2) but changes the sign of the total (the first one is positive and the second one is negative). The total charge redistribution is 0.85 e on surface 1 and 0.68 e on surface 2 E ach of them shows again a significant charge redistribution happening when water is adsorbed on the silica surface. On the other hand, if two water molecules are adsor bed on two adjacent silicon sites the bonding energy is lower On surface 1, we obtained a result that the first water molecule was slightly dragged off the silicon site and the total bonding energy is 53 meV lower (Figure 4 6) The charge d ensity di fference show s that the interaction between the oxygen atom of the first water molecule and the silicon site is weak er than that of one water molecule adsor bed on a silicon site due to the displacement of water molecule from the original position On the o ther hand, t he interaction between the proton of the first water molecule and the bridge oxygen atom is unaltered or slightly strengthened (with a larger isosurface region near the atoms). Interestingly, the charge density difference still shows a quadrup o le pattern after structur al relaxation When a third water molecule is adsorbed on the surface s after the adsorption of a water dimer in the most stable state three water molecules form hydrogen bonds with each other and the third water molecule is absor bed on the other silicon site (Figure 4 7, 4 8) Contradicting intuitive thinking the total bonding energ ies are decreased and the average bonding energ ies are even lower than the bonding energies of one water molecule adsorption on both perfect surfaces. Compar ed to the adsorption energies, the bonding energies are much smaller T his indicates that the strength of hydrogen bond s overwhelm s the water silica interaction and direct s the optimization of a water cluster

PAGE 62

62 adsorbed on the perfect surfac e s On the other hand, the charge density difference on each surface also show s smaller distribution range s than in the dimer adsorption case Similar to the two silicon site adsorption case, the second water molecule is lifted from the silicon site, and t he water silica interaction between the second third water molecules and silica surfaces is much smaller than the case of dimer adsorption in the most stable state (consider ing the distribution of charge density difference ). However, the first water molec ule still has a strong interaction with the silica surface. The bonding energies of one water molecule adsorbed on both surfaces are similar H owever, the bonding energies in two and three water molecules adsorption cases on surface 2 are obviously smaller than those on surface 1 There could be two reasons O ne is because surface 2 is more stable than surface 1, hence surface 2 deforms less when water water and water surface interactions compete each other during structur al optimization T he other reason c ould be that surface 1 has (1x1) hexagonal symmetry which is more preferred for water arrangement, since they tend to form hexagonal arrangement s a kin to in one sub layer of Ice XI. Because of the similarity between six member ed ring geometry on surface 1 and the hexagonal arrangement of one Ice XI sub layer, we suggest that there are two kind s of one water layer a d sorption on surface 1. One arrangement has all of the oxygen atoms of water molecules are a d sorbed on silicon atom s ites T he other arrangement ha s half on silicon atom sites and half on hollow sites (Fig ure 5 1, 5 2 ). We discuss water layer adsorption in Chapter 5 4. 4 Water Adsorption o n Surface 3 (the F ull y H ydroxylated S urface) Water adsorption on a fully hydroxylated surface has been stu died by Yang et al. already [22] T hey studied the adsorption of one, two water molecule s and one water

PAGE 63

63 layer on surface 3 We describe the results of our calculations in the following. When one water molecu le is adsorbed, the water molecule will break one weak hydrogen bond form ed within the hydroxyl groups on the surface and form two strong hydrogen bonds with two of them ( Figure 4 9 and R ef. [22] ) The bond length we found differs slight set of larger lattice constant s and also a larger energy cutoff in the calculation s (recall Chapter 3 ). The adsorption energy of one water molecule adsor bed on the surface is 618.9 meV which is ~14% larger t han [22] The charge density difference ( Figure 4 9 (c), (d)) shows a much larger range of charge redistribution than the same case on perfect surfaces and indicate s a much stronger water silica surface interaction. The adsorpt ion and bonding energies are about 14 and 12.5 times larger respectively than those on surface 1. Table 4 2 shows that the water molecule tends to lose some electron s to the surface. There is 0.68 e of charge redistribution in the system a value which is larger than that of one water molecule adsorption on perfect surfaces I t is surprising that the charge redistribution does not show a much larger number than for perfect surfaces since the water silica interaction on the fully hydroxylated surface i s much stronger than on the perfect surfaces When the second water molecule is adsorbed, the exist ing hydrogen bonds on the first water molecule do not break as Yang et al. reported T he second water molecule donates one proton to form a weak hydrogen bond with the surface and forms one slightly stronger hydrogen bond with the first water molecule ( Figure 4 10 ). We also tr ied to optimize a structure which Yang reported [22] T he adsorption energy of water dimer is slightly l ess by 5.4 meV. The charge density difference shows that the second

PAGE 64

64 water molecule does not have as strong an interaction with the surface as the frist water molecule does since the charge re distribution range is much smaller around the second water molecule When the third water molecule is adsorbed, it forms three hydrogen bonds in total with the second water molecule and hydroxyl groups on the surface I ts position relative to the second water molecule is similar to the first one ( Figure 4 11). The charge d ensity difference shows there are strong interactions between the first third water molecules and the surface T he interaction of the second water molecule with the surface is enhanced compar ed to two water molecule adsorption. Because of the hydrogen bonds form ed between water molecules and the s ilica surface, the bonding energies on the fully hydroxylated surface ar e always much higher than for water molecule adsorption on perfect surfaces (Table 4 1). The total charge redistribution is increased to 0 .98 e, which is somewhat larger than the cases on perfect surfaces 4. 5 Water Adsorption o n Surface s 4 and 5 ( O xygen V acancy D efect ) The strongest one water molecule adsorption site on defective surf aces with one oxygen vacancy is the center of the defect formed by two silicon atoms ( Figure 4 12 4 13 ) T he oxygen atom of the water molecule is almost equally distant to the two silicon atoms. The adsorption energies on surface 4 and 5 are about 4 to 4.8 times larger than that those on surface 1 (Table 4 1 ) and about 50 to 60% of the strength of one hydrogen bond in Ice XI ( in DFT calculation s) In our calculation, surface 5 has a larger bonding energy for a water molecule than surface 4 by 39 meV even though surface 5 is a little mo re stable than surface 4 To test the effect of a defect on neighbor adsorption sites, we optimize the structure of one water molecule adsorbed on hollow and silicon atom site s as well (hollow site next to the defect and silicon atom on defect)

PAGE 65

65 T he bondin g energies are 52.3 and 5.6 meV higher than for the low energy state However, t he hollow site adsorption still has higher adsorption and bonding energy than those on surface 1 by 84.1 and 78.7 meV. The distance between oxygen atom s of water molecules adsorbed on the hollow site and the defect site is about 1.8 This indicates that the oxygen vacancy defect does enhance water molecule attraction to the silica surface T he interaction strength is about 50 to 60% of one hydrogen bond with ~ 2 interact ion range. The charge density difference of the low energy state shows a larger range of charge redistribution than on surface 1 as well. The silicon silicon bond of the defect gains electron s but just above the defect there is a large region of electron depletion T he charge redistributes around the water molecule in a large range as well. Some electron transfers from the water molecule to the surface, which is in opposite sign with that of the water molecule adsorbed on perfect surfaces (Table 4 2) Ther e are 0.75 e and 0.83 e charge redistribution s on surface 4 and 5, which is reasonable, since the bonding energies are larger than on the perfect surfaces. When the second water molecule is adsorbed, it is at the hollow site and form s a hydrogen bond with the first one ( Figure 4 14, 4 15 ). The bonding energ ies are weakened to be ing slightly smaller than that of a water dimer adsorbed on perfect surfaces T he adsorption energies of the second water molecule obviously are smaller than the counterpart on perfect surfaces The charge density difference show s the second water molecule on each surface has a very weak interaction with the surface compar ed to the first water molecule Compar ed with surface 1 the lowered bonding energy and adsorption energy ind icate the silica water interaction is compromised for the strong hydrogen bond within the water dimer. As in the one water molecule

PAGE 66

66 adsorption case, the w ater clusters lose electrons and surfaces gain electrons, but the absolute value is smaller than befor e. The charge redistributions are reduced to 0.59 e and 0.76 e respectively. On the other hand, the adsorption and bonding energies of water dimer adsorbed on perfect surfaces are much enhanced after one water molecule adsorption T he charge densit y difference show s quadrup o le pattern on each perfect surface ( Figure 4 4, 4 5) which may be the reason for the interaction strengthening We will discuss the quadrup o le pattern later in Chapter 7 When the third water molecule is a d sorbed, it forms hydrogen bonds with the two previous water molecules on both surfaces (Figure 4 16, 4 17) The third water molecule is not adsorbed on the silicon atom site or on the hollow site. The total bonding energ ies are increased compar ed to the previous cases. The isosurf aces of charge density difference show that the first water molecule still has a strong interaction with the defect on surfaces but the interaction of the other two is much smaller T he interaction between the second wa ter molecule and the surface is increased slightly T his could be the reason for the r ise in bonding energy. The charge redistributions are increased to 1.02 e and 1 e respectively. Compar ed with the second water molecule adsorption energy, the third adsorption energy is much higher. Thi s is because there are two new hydrogen bond s form ed within the water trimer. 4. 6 Water Adsorption o n Surface s 6 and 7 ( O xygen D isplacement) When one water molecule comes near to the surface, it will be attracted to th e peroxy linkage of the defect with on e of the proton s pointing to the oxygen atom of the peroxy linkage ( Figure 4 18, 4 19). The bonding energy on surface 6 is ~ 73% larger than that on surface1 ; but on surface 7, it is about the same value. Because the peroxy linkage on surface 6 is involved in a three member ed ring ( which has a higher strain

PAGE 67

67 environment than in six member ed rings and quartz bulk), the interaction between the peroxy linkage and the proton in the water molecule is stronger. Du. et al. studied the relationship of the water silic a interaction with the surface structure and reached a similar conclusion [27] The charge density difference show s the interaction is located at the peroxy linkage and the proton of water on each surface. There are 0.43 e and 0.39 e charge redistribution for the t wo systems respectively. When the second water molecule is a d sorbed, it is located on the hollow site. On surface 6, the second water molecule forms a hydrogen bond to the first water molecule (Figure 4 20) On surface 7, the second water molecule interacts with the oxygen peroxy linkage and alters the orientation of the first water molecule to form a hydrogen bond within the water dimer ( Figure 4 21 ). Compared to water dimer adsorption on surface 4 and 5, the total energies are not reduced but increased. The reason could be that the defect has a larger interaction area than the oxygen vacancy defect on surface s 1 and 2, since this defect type is composed of one oxygen vacancy and one peroxy linkage. T he average bonding energies are d ecreased by about 10% of one water molecule adsorption The effect of hydrogen bonds making the water silica interaction to compromise does not appear yet, it will be revealed in three water molecules adsorption The charge redistributions are increased with values of 0.69 e and 0.68 e respectively. When the third water molecule is adsorbed, it forms a hydrogen bond to the second one on surface 6 and to both pre ceding molecules on surface 7 (Figure 4 22, 4 23) The bonding energies are decreased and the a ds orption energies are increased, similar to the water trimer adsorption on perfect surface s (Table 4 1). The charge

PAGE 68

68 density difference still show s that the water silica interactions on each surface are located at the first water molecule and peroxy linkag e T he charge redistributions are slightly re duced to 0.59 e and 0.56 e. In table 4 1, we can see that the adsorption energ ies on surfaces increase with increasing number of adsor bed water molecules H owever, the total bonding energies rise a little or des cend. This behavior is an indication that the interaction of hydrogen bonds within the water entity gradually controls the optimization direction when more water molecules accumulate We notice that t he charge transfer behaviors are different on two types of silica defects. The oxygen vacancy always will gain electrons from water molecule(s), while the oxygen displacement defect always gives electrons to water molecule(s). On the other hand, the perfect surface loses electrons at one water molecule adsorption and gains electrons when more water molecules are adsorbed.

PAGE 69

69 Table 4 1 The adsorption and bonding energy of water molecules on silica surface. Adsorption energy, E total is total system energy, E o is total energy of a system with one fewer water molecule. Bonding energy, All energies are given in meV. Surface E a 1 w E a2w E a3w E b1w E b2w E b3w 1 43.8 343 481 67.0 207 146 2 58.8 327 485 66.9 176 120 3 619 365 779 838 1148 2120 4 180 224 597 198 175 267 5 213 183 590 237 192 251 6 108 248 414 116 184 158 7 63.0 291 523 70.3 125 118 Table 4 2. Charge transfer analysis of water molecules on surface (P ositive means gaining electrons ) Unit s : ( electron number/1000 ) The number is giv en as the electron s transferred on to the whole water molecule(s) cluster. The charge on surface means the charge transferred on the topmost layer of the surface, not including the whole slab. The charge transfer is calculated from the charge distribution of [whole system static surface a lon e static cluster of water molecule(s)]. Surface 1 water molecule 2 water molecules 3 water molecules water surface water surface water surface 1 8.9 8.6 3.9 5.2 3.3 2.6 2 2.9 2.5 7.7 7.7 3.3 2.9 3 27 23.6 31 31.3 20 18.5 4 15.7 13.4 12.3 7.7 19.1 13 5 22.9 19.1 18.5 16.9 26.6 16.3 6 5.9 4.8 6.7 4.9 3.3 2.2 7 4.4 4 8 8.7 0.1 0.6

PAGE 70

70 Table 4 3. The charge redistribution of water cluster adsorption systems. See equation (4 4). The values are in electron s Surface 1 water molecule 2 water molecules 3 water molecules 1 0.45 0.85 0.65 2 0.31 0.68 0.56 3 0.68 0.89 0.98 4 0.75 0.59 1.02 5 0.83 0.76 1.00 6 0.43 0.69 0.59 7 0.39 0.68 0.56

PAGE 71

71 a b c d Fig ure 4 1 One water molecule adsorbed on surface 1 ( (1x1) perfect surface ) at the hollow site a ) Top view. b ) Side view. c ) and d ) are the top view and side view of the system with charge density difference draw n on isosurface of 0.005/ 3 The yellow balloon is the zone gaining electron s and blue is that losing electron s

PAGE 72

72 a b c d Fig ure 4 2. One water molecule adsorbed on surface 2 ( (2x1) perfect surface). a ) T op view. b ) Side view. c ) and d ) are the top view and side view of the system with charge density difference drawn on isosurface of 0.005/ 3

PAGE 73

73 a b c d Fig ure 4 3. One water molecule adsorbed on surface 1 ( (1x1) perfect surface) at the silicon atom site. a ) Top view. b ) Side view. c ) and d ) are the top view and side view of the system with charge density difference draw n on isosurface of 0.00 3 / 3

PAGE 74

74 a b c d Fig ure 4 4. Two water m olecules adsorbed on surface 1 ( (1x1) perfect surface). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 5 / 3

PAGE 75

75 a b c d Fig ure 4 5. Two water m olecules adsorbed on surface 2 ( (2x1) perfect surface). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 5 / 3

PAGE 76

76 a b c d Fig ure 4 6. Two water molecules adsorbed on two silicon s ites of surface 1 ( (1x1) perfect surface). a) Top view. b) Side view. c) and d) are t he top view and side view of the system with charge density difference drawn on isosurface of 0.00 5 / 3

PAGE 77

77 a b c d Fig ure 4 7. Three water m olecules adsorbed on surface 1 ( (1x1) perfect surface). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.00 5 / 3

PAGE 78

78 a b c d Fig ure 4 8. Three water m olecules adsorbed on surface 2 ( (2x1) perfect surface). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 5 / 3

PAGE 79

79 a b c d Fig ure 4 9. One water molecule adsorbed on surface 3 ( fully hydroxylated surface). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.0 3 / 3

PAGE 80

80 a b c d Fig ure 4 10. Two water molecules adsorbed on surface 3 ( fully hydroxylated surface). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.0 3 / 3

PAGE 81

81 a b c d Fig ure 4 11. Three water molecules adsorbed on surface 3 ( fully hydroxylated surface). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.0 3 / 3

PAGE 82

82 a b c d Fig ure 4 12. One water molecule adsorbed on surface 4 (oxygen va cancy defect). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.0 05 / 3

PAGE 83

83 a b c d Fig ure 4 13. One water molecule adsorbed on surface 5 (oxygen vacancy defect). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.0 05 / 3

PAGE 84

84 a b c d Fig ure 4 14. Two water molecule s adsorbed on surface 4 (oxygen vacancy defect). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.0 05 / 3

PAGE 85

85 a b c d Fig ure 4 15. Two water molecules adsorbed on surface 5 (oxygen vacancy defect) a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.005/ 3

PAGE 86

86 a b c d Fig ure 4 16. Three water molecules adsorbed on sur face 4 (oxygen vacancy defect). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.005/ 3

PAGE 87

87 a b c d Fig ure 4 17. Three water molecules adsorbed on surface 5 (oxygen vacancy defect). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.005/ 3

PAGE 88

88 a b c d Fig ure 4 18. One water molecule adsorbed on surface 6 (oxygen vacancy displacement). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.005/ 3

PAGE 89

89 a b c d Fig ure 4 19. One water molecule adsorbed on surface 7 (oxygen vacancy displacement). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference draw n on isosurface of 0.005/ 3

PAGE 90

90 a b c d Fig ure 4 20. Two water molecules adsorbed on surface 6 (oxygen vacancy displacement). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density diffeence draw n on isosurface of 0.005/ 3

PAGE 91

91 a b c d Fig ure 4 21 Two water molecules adsorbed on surface 7 (oxygen vacancy displacement). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density distribution draw n on isosurface of 0.005/ 3

PAGE 92

92 a b c d Fig ure 4 22. Three water molecules adsorbed on surface 6 (oxygen vacancy displacement). a) Top view. b) Side view. c) and d) are the top view and side view of the syste m with charge density distribution draw n on isosurface of 0.005/ 3

PAGE 93

93 a b c d Fig ure 4 23. Three water molecules adsorbed on surface 7 (oxygen vacancy displacement). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density distribution draw n on isosurface of 0.005/ 3

PAGE 94

94 CHAPTER 5 WATER LAYER(S) ADSORPTION ON SILICA SURFACES After the test of water cluster adsorption on alpha quartz (0001) surfaces, we deposit ed Ice XI sub layers one layer at one time onto the su rfaces. We tr ied two strategies of water layer depositi on on both perfect surfaces and defective surfaces. According to the observation of water cluster adsorption on perfect surfaces in Chapter 4, o ne arrangement could be that all of the water molecules are absorbed on silicon atom sites; the other one is half on silicon atom sites and half on hollow sites Regarding water layer adsorption on hydroxylated surface, we follow ed the ice layer structure proposed by Yang et al [2 2] and then deposit ed more water layers on it. 5.1 Introduction In the introduction of Chapter 4, we reviewed briefly the recent studies of water adsorption on silica surfaces in experiment and in computational calculations. So far, the first principles studies on water silica interaction were focus ed on the hydroxylation of silica by water molecule s [27, 28, 73, 74] water molecule adsorption on the (1x1) perfect surface and the fully hydroxylated surface [14, 15, 22, 30, 88] and one water layer adsorption on the fully hydroxylated surface s [22, 30, 88] More efforts still is need ed on water multi layer adsorption on silica surface s The most important issue may be the understanding of the water silica interaction on an amorphous silica surface. While a full understanding of an amorphous silica surface is not available experimentally or theoretically, the alpha quartz (0001) surface offer s a simplified model for investigating water silica interaction and its effect on the water hydrogen bonding network.

PAGE 95

95 5.2 Method s The methods us ed are the same as in Chapter 4 except for the equations for calculating the adsorption and bonding energies namely (5 1) (5 2) In equation (5 1), is the adsorption energy of the nth water layer is the total energy of the optimized system with n water layer s adsorbed, is total energy of the optimized system with (n 1) water layers adsorbed, and is the energy of one optimized water molecule. m is the number of water molecules in one w ater layer, thus it is 8 on surface s 1 through 3 and 18 on surface s 4 through 7. The a rea is 84.89 2 for surface 1 through 3 and is 191.01 2 for surface s 4 through 7. In equation ( 5 2), is the bonding energy of n water layer and is the energy of the individual surface or water layer(s) subsystem which ha s the same configuration as in optimized total system. 5. 3 Water Adsorption on Surface s 1 and 2 ( (1X1) and (2X1) P erfect S u rface ) According to the discussion in section 4.3 there are two kinds of water molecule adsorption sites: silicon atom sites and hollow sites. All of the silicon atom sites form a hexagonal 2D matrix and all of the hollow sites combin ed with half the silicon ato m sites also form the same arrangement. Because one sub layer of the most stable ice structure (Ice XI) and surface 1 have the same hexagonal geometry we deposit ed the sub layer of Ice XI one by one onto the surfaces to see how the surface interact s with the water layer. Figure 5 1 and 5 2 show the optimized structures of two types of one water layer deposition on surface 1 Each of t he relaxed water layer s adsorbed on

PAGE 96

96 surface 1 has a flat ice feature with two group s of hydrogen bond lengths (~1.75 and ~1.96 ). In each water layer, h alf of the water molecules have two protons forming hydrogen bonds in the water layer plane T he other halves have one proton form ing hydrogen bond s and one proton pointing out of the water layer plane. For the water layer with the proton pointing up, the water molecules prefer to be adsorbed on all silicon sites. For the proton pointing down structure, th e water molecules with proton pointing down will be adsorbed on hollow site s whil e the other water molecules are adsorbed on silicon atom sites. The ground state is the water layer with the proton pointing down It has a total surface energy lower than pointing up system by 3.58 meV/ 2 The bonding energy of the low energy state is abo ut twice of its isomer but the adsorption energies of two states are similar (Table 5 1 ). The proton pointing down case is considered as a hydrophilic adsorption case, which has an average bonding energy (divided by the number of water molecules) of 40.8 m eV. The distance between the water monolayer to the surface is shortened in the proton pointing down case by ~16% compar ed to the proton pointing up case ( isomer case ) Comparing the charge density difference of these two cases, the proton pointing down structure has a larger range of charge redistribution (lo w energy state: 0.71 e isomer state: 0.11 e) which also form s quadrup o le matrix on the surface (Figure 5 2) T he proton pointing up case has a smaller range of charge redistribu tion and the distribution is not equivalent on the adjacent silicon atom sites. In section 4 3 we discussed the two cases of water dimer adsorption on surface 1 O ne case has water molecules adsorbed on one hollow and one silicon site, the other has both water molecule adsorbed on adjacent silicon sites. T he resulting charge densit y difference form s quadrup o le arrangements for both cases

PAGE 97

97 H owever, in the t wo silicon sites case, the oxygen atom of one water molecule is lifted off the silicon site and one proton has an interaction with the oxygen atom on the surface (Figure 4 6). The charge density difference for the water layer isomer case shows alternating interaction strength of water molecules to silica surface (Figur e 5 1) A dsorption of water molecules on adjacent silicon sites seems not preferred according to the two cases described On the other hand, the average bonding energy of the water layer adsorption in proton pointing down case is obviously lowe r than that of the water dimer adsorption, even though the charge density difference forms a hexagonal quadrup o le matrix. Because the hydrogen bonds are much stronger than the water silica interaction t he best adsorption position for the individual water molecule is compromised. We will see this point much clearer in adsorption on defective surfaces. We also treated t wo types of water mo nolayer adsorption on surface 2. Compared to the counterparts on surface 1, in the surface 2 case, each water mono layer is deformed a little with a larger variation on the longer hydrogen bond group ( 1.93 ~ 2.01 in Figure 5 3, 1.95~1.98 in Figure 5 4 ). T he distance between the water layer and silica surface is similar to cases on surface 1 but the water molecules on silic on sites deviate from the silicon sites. The charge densit y difference of each case is different from the counterpart on surface 1 but a quadr up o le matrix still form ed in the proton pointing down case The low energy state is still the proton pointing dow n configuration and the proton pointing down case is isomer state. The charge redistributions of the proton pointing down and up cases are 0.58 e and 0.11 e respectively The adsorption and bonding energ ies are a little bit weaker than for the correspondin g cases with surface 1 for dissimilar symmetries of surface 2 and water

PAGE 98

98 layer. Because the hydrogen bonds are much stronger than the interaction of water silica interface, the water layer hexagonal structure is deformed just a little on surface 2 and the adsorption energy is surprisingly similar. For the second water layer adsorption on surface 1, we deposit a corresponding Ice XI sub layer on top of the first layer structures The optimized structures of water layers are very different for the two ad sorption cases. For the p roton pointing up case, the out of plane protons in the second layer will reverse their orientations to point ing down. The up and down water layers form a membrane like bilayer thin film and all hydrogen bonds of the water thin fil m are saturate d within itself ( Figure 5 5). The thickness of the thin film is 2.71 and the distance between the water film and surface are similar to the isomer state of monolayer adsorption case The charge density difference is localized at the interf ace which indicates that the second water layer does not have a strong (or observable) interaction with the silica surface. The bonding energy is only increased by 6.8% from monolayer adsorption. In the proton pointing down case, there is no direction revers al for protons out of the water plane ( Figure 5 6 ). The charge density difference is also localized at the interface as in the bilayer adsorption case but it still forms a quadrup o le matrix as with the proton pointing down case in monolayer adsorpti on. Not like th bilayer adsorption case, t he b onding energy decreases by 15% from monolayer adsorption The charge redistributions are 0.11 e and 0.54 e for the systems with proton s point ing up and down in first water layer. The ground state of two water l ayer adsorption is now switched to the bilayer adsorption case, not the case with proton pointing down. The bilayer adsorption is 37.9 meV/ 2 lower than the proton pointing down configuration in total surface energy. The

PAGE 99

99 main energy gain is from the form ation of more hydrogen bonds I n the bilayer structure, all protons form hydrogen bond s with the other water molecule s but in the proton pointing down case they are not saturated Although the bonding energy is still larger in the proton pointing down cas e, the h ydrogen bonds strongly direct the total energy minimization and the ground state should be the bilayer adsorption case. On surface 2, the bonding energ ies of the adsorption of two water layers are similar to those on surface 1. The deviation of the hydrogen bond length s within water layers become smaller compa red to the monolayer case s The lower energy state is the bilayer adsorption case as well The charge redistributions are 0.12 e and 0.46 e for the bilayer and the protons pointing down case For the third and fourth water layer adsorption, the s ame way of Ice XI sub layer deposition is applied on to two adsorption case s on each perfect surface (Figure 5 9 to 5 16) When the third layer is added on top of the bilayer structure, hydrogen bonds fo rm and break between the second and third water layers ( Figure 5 9, 5 11 ). The water layers pucker in the vertical direction. But for the proton pointing down case, there is no hydrogen bond breaking ; only the third layer forms hydrogen bonds with the seco nd layer. The result is like a piece of Ice XI place d on a silica surface. The ground state is still the case with the first layer pointing up w hich has a total surface energy lower than the proton pointing down configuration on surface 1 by 35.3 meV/ 2 T he result for the fourth water layer adsorption is similar to that for the third water layer adsorption (Figure 5 10, 5 12) The first layer pointing up case has a total surface energy 47.1 meV/ 2 lower than the proton pointing down configuration on surface 1 The adsorption energies

PAGE 100

100 of the water monolayer to four layers (ground state) show an odd even oscillation on both surfaces. It is remarkable that the adsorption and bonding energies are so similar on both perfect surfaces, even though their surface symmetries are different. The other phenomenon which deserves notice is that the bonding energies increase a large amount for four water layer with proton pointing down on defect surfaces when compared to the adsorption of ewer water layers (~5.3 e V vs ~3.0 eV) The charge density difference shows that a small amount of charge redistribution is localized at the fourth layer T he weird phenomenon of charge accumulation at the topmost water layer actually starts from the third water layer adsorption ( see Table 5 2) At the same time, the total charge redistributions are also increased (from 0.58 e to 1.7 e, see Table 5 3) This behavior seems to be originat ing from the interaction between the surface and the dipole of Ice XI. Even though the proton poi nting down arrangement s are not the ground state s for 2, 3, 4 water layers adsorption it is still interesting to study the existence of a stable dipole moment of ice adsorbed on a silica surface and the interaction s with in But in this dissertation we do not go beyond four water layers adsorption and further efforts need to be done in the future work. 5. 4 Water Adsorption o n Surface 3 (the F ully H ydroxylated S urface) The interaction of the hydroxyl group s with water molecule s is very strong and comparabl e with hydrogen bonds be tween water molecules in Ice XI. The hydrogen bond network could be complicated and without any symmetry T here are many ways for water molecules to be absorbed on the fully hydroxylated quartz surface. For one water layer adsorptio n, the way Yang reported [22] will saturate all of the hydrogen bonds of water molecules and hydroxyl groups on the surface ( Figure 5 17). We did not

PAGE 101

101 find another way for water layer adsorption to occur with the same coverage and lower tota l system energy. However, we point out that this mode of water layer adsorption has three kinds of polymorphic states with different dipole orientations (Fig ure 5 18 ). The adsorption energies are 61.5, 62.2, 61.7 meV/ 2 respectively The energy difference s are not big, but the relative dipole orientations differ by 120 relative to each other The existences of these three types of adsorption need to be clarified experimentally. I this work, w e only pick the first configuration for discussion and consider t he adsorption of more water layers The charge density difference shows a large range of charge redistribution at the interface again, and it form s a hexagonal quadrup o le matrix on the surface ( Figure 5 17 (c), (d) ). The charge transfer is much larger than that on perfect surfaces but it is still a small amount in absolute value The total charge redistribution is 1.85 e which is also much larger than for perfect surfaces, and about doube the value for three water molecule adsor ption on surface 3 (8 water molecules in one water layer for water layer adsorption ). When the second water layer is deposit ed on the first layer, there is no strong bonding between those two as indicated by the adsorption energy of the second water layer and also the sligh t increase in bonding energy. The adsorption energy is slightly larger than that of 1 water layer adsorption on perfect surfaces T he main contribution to the adsorption energy is from the formation of hydrogen bonds with in the second water layer. We tried several water layer deposition positions for the second water layer adsorption. Figure 5 19 shows the structure with the lowest total energy in which the second layer form s weak (long, ~ 2 ) hydrogen bonds to the first water layer. The charge density difference is still loca lize d at the interface and is not even disturbed

PAGE 102

102 when compar ed with the monolayer adsorption. To look for the possible adsorption site, we deposit ed one water molecule at several positions on top of the first water layer. The bondin g energies are about 100 meV /molecule, which is much lower than the hydrogen bond s within Ice XI. Th e se data show that the hydrophilic property of the surface is shield ed by the bilayer structure which consists of the first water layer and the hydroxyl gro ups. On surface 1 and 2, the first two water layers also form a membrane like bilayer and all of the hydrogen bonds are saturated. The second water layer adsorption on surface 3 is similar to the third water layer adsorption on surface 1 and 2 H owever, th e adsorption energy is dramatically different. The third water layer adsorption on surface 1 or 2 results in hydrogen bonds forming and breaking at the second and third water layer s. It also causes the whole water film to pucker. On the other hand, the bil ayer structure on surface 3 almost does not deform for the adsorption of the second water layer. This is because the oxygen atoms o f hydroxyl groups are almost fixed by the silicon oxygen chemical bond on the surface and make the bilayer structure much mor e stable and inert. We deposit one and two more water layers onto the second water layer and the optimized structures consist of surface the first layer structure and water bilayer, trilayer structures on perfect surfaces. The adso rption energies are similar to those for the second, third water layer adsorption on perfect surfaces (Table 5 1) This test proved that the bilayer structure formed by the hydroxyl groups and the first water layer is truly inert and stable. 5. 5 Water Adsorptio n on Surface s 4 and 5 ( O xygen V acancy D efect ) Even though the interaction between the surface defect and one water molecule is stronger than water silica interaction on perfect surfaces, the interaction decays

PAGE 103

103 quickly as the distance between water and defect increases. I n section 4.5 we discuss ed the adsorption of water molecules on defective surfaces and the water silica interaction compromises for the strong hydrogen bond s within the water cluster. Since the hydrogen bonds are very strong and affect the structure of wa ter cluster so much in the previous test of adsorption of water molecules w e deposit ed Ice XI sub layers on defective surfaces and then stud ied the effect of the defect on the adsorption of water layer(s) M onolayer and two water layer adsorption are included For one water layer adsorption, the proton pointing down structure is still more stable than the proton pointing up structure on both surfaces (Figure 5 20 to 5 23 ) and the total surface energy difference between two states are 3. 31 meV/ 2 on surface 4 and 3.16 meV/ 2 on surface 5; which are slightly smaller than that on surface 1 (3.58 meV/ 2 ) The adsorption and bonding energ ies of the proton pointing down state on defective surfaces are slightly smaller than for the perfect surf ace as well, which may result from the competition between the minimum energy of the water layer and the water silica defect interaction. As a result, the water layer distorts a little from the one adsorbed on surface 1. The water molecules adsorbed on the defect sites are pulled slightly close r to the surface because of the stronger interaction. Th e data of adsorption and bonding energies (Table 5 1) show that the effect of the defect is compromised for the appea rance of the hydrogen bonds network T he strong hydrogen bonds will control the direction of energy minimization. On both defective surfaces, the charge density difference of the proton pointing down state form s a hexagonal quadrup o le matrix (Figure 5 21, 5 23) as well as on perfect surfaces even though the distribution s are distorted near the surface defect s On the other hand,

PAGE 104

104 the charge densit y difference of each isomer state ha s an arrangement (Figure 5 20, 5 22) similar to the counterpart s on perfect surfaces except the region near the wa ter molecules adsorbed on the defect. In particular t here is a large zone of electron enrichment between two water molecules on the defect on both defective surfaces The effect of the defect is a short range interaction for each the charge densit y difference that quickly recover s to the corresponding distribution on (1x1) perfect surface ( Figure 5 1, 5 2) about 2 to 3 away from the defect site. One special phenomenon is that the observed charge enrichment on the defect (Si Si bond) in water molecu le(s) adsorption cases disappears or diminishes very much in the water layer adsorption on both defective surfaces. In the two water layer adsorption case, t he bilayer structures on surfaces are the lower energy states again on both defective surfaces wit h total surface energy 37.6 meV/ 2 and 37.7 meV/ 2 lower than the proton pointing down states. As on perfect surfaces, even though the bonding energy is higher for the proton pointing down states, the minimum of total energy requires that the structure transform to a bilayer state. The existence of defect s did not change the water layer(s) adsorption tendency; and the adsorption and bonding energies are very similar to the numbers on perfect surfaces (Table 5 1) The charge density differe nce in each case is localized at the silica water interface as well (Figure 5 24 to 5 27) S imilar to water monolayer adsorption cases t he disappearance of charge enrichment on defect for both water layer states and electron enrichment between two water m olecules on defect for bilayer cases are also observed.

PAGE 105

105 5 6 Water Adsorption o n Surface s 6 and 7 ( O xygen D isplacement) We use the same strategy for depositing Ice XI sub layer(s) on surface 6 and 7 and optimize the structure s to see the effect of defect s on the water film. The low energy state for one water adsorption is still the one with the proton pointing down structure and the other one is the isomer state on both surfaces Compar ed to the water films on surface 1, two types of adsorbed water film ha ve some distortion near the defect as well T he water molecule close to the defect will be pushed up or pulled down a little bit. The adsorption energies show similar value s to those of surface 1. The bonding energy on surface 7 is slightly lower than that on surface 1 H owever, on surface 6 it is apparently lower by 27% of the value on surface 1. The charge density difference shows the water silica interaction is concentrated at the peroxy linkage of the defect for two s tates of adsorption on surfaces Unl ike on surface 4 and 5 the large range charge redistribution above oxygen vacancy disappears and the charge redistribution is more concentrat ed on the peroxy linkage now (Figure 5 28 to 5 31) Again, the interaction with the defect is a short range effec t; t he pattern of charge redistribution recovers quickly to the corresponding distributions on surface 1 just as with surface s 4 and 5. In the water cluster adsorption section s surface 6 has larger water molecules bonding energies than surface 7; however, it is reversed for 1 water layer adsorption. The optimization follow s the energy minimum of the total system, so the strong hydrogen bonds tend to keep the hexagonal structure of the water layer but compromise the interaction between silica surface and wa ter molecules at the same time. Though surface 6 has higher tension on peroxy linkage and should attract the

PAGE 106

106 water molecule s more One oxygen atom of the peroxy linkage is out of the surface plan e which will make the water layer to distort more than on sur face 7. The result is the bon ding energy has to be weakened for maintain ing water layer hexagonal structure. For two water layers adsorption, the water bilayer is still more stable than the proton pointing down configuration. Again, the adsorption energie s on two surfaces are similar to those on surface 1, but the bonding energies are smaller than those on surface 1 O n surface 6 the number is even smaller. We can use the same reason to explain the similar tendency for the bonding energies as in 1 water la yer adsorption case.

PAGE 107

107 Table 5 1. The adsorpt ion and bonding energy of water layer (s) on a silica surface. Adsorption energy, E total is total system energy, E o is total energy of one fewer water layer system. m is the number of water molecules in one water layer. Bonding energy, The numbers out/in parentheses are for proton pointing up/down systems. All energies are given in meV/ 2 Surface E a 1 L E a 2L E a 3L E a 4L E b1L E b2L E b3L E b4L 1 44.3 (46.1) 68.7 (48.0) 46.9 (48.2) 56.5 (50.6) 1.48 (3.84) 1.58 (3.26) 1.51 (3.35) 1.56 (5.60) 2 43.9 (45.6) 68.8 (48.4) 47.0 (48.0) 58.3 (51.5) 1.43 (3.01 ) 1.44 (2.70) 1.56 (2.85) 1.56 (5.15) 3 62.2 48.5 64.1 46.1 34.6 35.9 35.3 36.4 4 44.4 (46.0) 68.9 (48.4) 1.42 (3.53) 1.45 (3.04) 5 44.4 (46.0) 68.9 (48.4) 1.43 (3.33) 1.45 (2.79) 6 44.3 (45.6) 68.9 (48.6) 1.41 (2.81) 1.39 (2.58) 7 44.4 (46.0) 68.9 (48.4) 1.45 (3.22) 1.42 (2.82)

PAGE 108

108 Table 5 2. The charge transfer analysis of water layer ( s ) on a silica surface (Positive means gaining electrons ). Unit: ( electron number/1000 ) The number given is the electron s transferred to the whole water layer(s) system. The charge on surface means the charge transferred to the topmost layer of the surface, not including the wh ole slab.The charge transfer is defined as the charge distribution of [whole system static surface alon e static water layer(s)]. The numbers out/in parentheses are for proton pointing up/down systems. For systems with more than two water layers, the ch arge transfer is presented in a way of charge transfer in first layer, second Surface 1 water layer 2 water layers 3 water layers 4 water layers water surface water surface water surface water surface 1 5.7 ( 18.2) 5 (15.5) 5.4, 0.2 ( 16.2, 0.2) 5 (14.3) 4.6, 0.7, 0.2 ( 16.4, 1.3, 21.3) 5 (21.3) 5, 0.2, 0.1, 0.3 ( 187.2, 1.8, 1.5, 35.7) 4.6 (134) 2 5.4 ( 16.1) 4.7 (14.4) 4.5, 0.9 ( 15.1, 0.2) 4.3 (13.5) 5.1, 0.1, 0.4 ( 18.9, 0.5, 38.1) 4.7 (33.1) 5.1, 0.2, 0.2, 0.3 ( 189.9, 4.1, 0.2, 39.8) 5 (135.2) 3 21 17.3 46, 19.1 22.8 N/A N/A N/A N/A 4 12.1 ( 42.7) 11.1 (35.1) 10.4, 0.4 ( 32.5, 0.4) 9.4 (26.1) N/A N/A N/A N/A 5 13.7 ( 43.6) 12.1 (39.2) 10.5, 1.1 ( 32.7, 0.1) 9.5 (28.4) N/A N/A N/A N/A 6 13.6 ( 55.7) 13.6 (50.6) 12.2, 0.7 ( 40.3, 0.3) 10.9 (35.6) N/A N/A N/A N/A 7 13.9 ( 42) 13.6 (35.5) 11.5, 0.1 ( 32.5, 0) 10.4 (28.1) N/A N/A N/A N/A

PAGE 109

109 Table 5 3. The charge redistribution of water layer(s) adsorption systems. The definition is at equation (4 4) with unit s in electron. The numbers out/in parentheses are for proton pointing up/down systems. Surface 1 water layer 2 water layers 3 water layers 4 water layers 1 0.11 (0.71) 0.11 (0.54) 0.15 (0.83) 0.15 (1.61) 2 0.11 (0.58) 0.12 (0.46) 0.18 (0.89) 0.18 (1.70) 3 1.85 2.05 N/A N/A 4 0.29 (1.58) 0.30 (1.26) N/A N/A 5 0.27 (1.44) 0.25 (1.10) N/A N/A 6 0.33 (1.21) 0.33 (1.06) N/A N/A 7 0.33 (1.37) 0.31 (1.09) N/A N/A

PAGE 110

110 a b c d Fig ure 5 1. One wat er layer adsorbed on surface 1 ( (1x1) perfect surface) with protons pointing up a ) Top view. b ) Side view. c ) and d ) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 05 / 3

PAGE 111

111 a b c d Fig ure 5 2. One wat er layer adsorbed on surface 1 ( (1x1) perfect surface) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 3 / 3

PAGE 112

112 a b c d Fig ure 5 3. One water layer adsorbed on surface 2 ((2x1) perfect surface) with protons pointing up. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 05 / 3

PAGE 113

113 a b c d Fig ure 5 4. One water layer adsorbed on surface 2 ((2x1) perfect surface) with protons p ointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 2 / 3

PAGE 114

114 a b c d Fig ure 5 5. Two water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing up in the first water layer a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 05 / 3

PAGE 115

115 a b c d Fig ure 5 6. Two water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 2 / 3

PAGE 116

116 a b c d Fig ure 5 7 Two water layers adsorbed on surface 2 (( 2 x1) perfect surface) with protons pointing up in the first water layer a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 05 / 3

PAGE 117

117 a b c d Fig ure 5 8. Two water layers adsorbed on surface 2 ((2x1) perfe ct surface) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 2 / 3

PAGE 118

118 a b c d Fig ure 5 9. Three water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing up in the first water layer a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 05 / 3

PAGE 119

119 a b c d Fig ure 5 10. Three water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 2 / 3

PAGE 120

120 a b c d Fig ure 5 11. Three water layers adsorbed on surface 2 ((2x1) perfect surface) with protons pointing up in the first water layer a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 05 / 3

PAGE 121

121 a b c d Fig ure 5 12. Three water layers adsorbed on surface 2 ((2x1) perfect surface) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 2 / 3

PAGE 122

122 a b c d Fig ure 5 13. Four water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing up in the first water layer a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 05 / 3

PAGE 123

123 a b c d Fig ure 5 14. Four water layers adsorbed on surface 1 ((1x1) perfect surface) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 2 / 3

PAGE 124

124 a b c d Fig ure 5 15. Four water layers adsorbe d on surface 2 ((2x1) perfect surface) with protons pointing up in the first water layer a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 05 / 3

PAGE 125

125 a b c d Fig ure 5 16. Four water layers adsorbed on surface 1 ((2x1) perfect surface) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.00 2 / 3

PAGE 126

126 a b c d Fig ure 5 17. One water layer adsorbed on sur face 3 ( fully hydroxylated surface). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 3 / 3

PAGE 127

127 a b c d Fig ure 5 18. The other two possible orientation s of one water layer adsorbed on surface 3 ( fully hydroxylated surface). a) and b) are the top and side view of the first orientation. c) and d) are the top and side view of the second orientation.

PAGE 128

128 a b c d Fig ure 5 1 9 Two water layer s adsorbed on surface 3 ( fully hydroxylated surface). a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 3 / 3

PAGE 129

129 a b c d Fig ure 5 20. One water layer adsorbed on surface 4 (oxygen vacancy defect) with protons pointing up. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 005 / 3

PAGE 130

130 a b c d Fig ure 5 21. One water layer adsorbed on surface 4 (oxygen vacancy defect) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 03 / 3

PAGE 131

131 a b c d Fig ure 5 22. One water layer adsorbed on surface 5 (oxygen vacancy defect ) with protons pointing up. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 005 / 3

PAGE 132

132 a b c d Fig ure 5 23. One water layer adsorbed on surface 5 (oxygen vacancy defect ) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 02 / 3

PAGE 133

133 a b c d Fig ure 5 24. Two water layers adsorbed on surface 4 (oxygen vacancy defect) with protons pointing up in the first water layer. a) Top view. b) Side view. c) a nd d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 005 / 3

PAGE 134

134 a b c d Fig ure 5 25. Two water layers adsorbed on surface 4 (oxygen vacancy defect) with protons pointing down. a) Top view. b) Side view. c) and d) are the top vi ew and side view of the system with charge density difference drawn on isosurface of 0.0 02 / 3

PAGE 135

135 a b c d Fig ure 5 26. Two water layers adsorbed on surface 5 (oxygen vacancy defect ) with protons pointing up in the frist water layer. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 005 / 3

PAGE 136

136 a b c d Fig ure 5 27. Two water layers adsorbed on surface 5 (oxygen vacancy defect ) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and sid e view of the system with charge density difference drawn on isosurface of 0.0 02 / 3

PAGE 137

137 a b c d Fig ure 5 28. One water layer adsorbed on surface 6 (oxygen displacement) with protons pointing up. a) Top view. b) Side view. c) and d) are the top view and side view o f the system with charge density difference drawn on isosurface of 0.0 005 / 3

PAGE 138

138 a b c d Fig ure 5 29. One water layer adsorbed on surface 6 (oxygen displacement) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 02 / 3

PAGE 139

139 a b c d Fig ure 5 30. One water layer adsorbed on surface 7 (oxygen displacement) with protons pointing up. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 005 / 3

PAGE 140

140 a b c d Fig ure 5 31. One water layer adsorbed on surface 7 (oxygen displacement) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 02 / 3

PAGE 141

141 a b c d Fig ure 5 32. Two water layers adsorbed on surface 6 (oxygen displacement) with protons pointing up in the first water layer. a) Top view. b) Side view. c) and d) are the top view and side view of the system with charge density difference drawn on isosurface of 0.0 005 / 3

PAGE 142

142 a b c d Fig ure 5 33. Two water layers adsorbed on surface 6 (oxygen displacement) with protons pointing down. a) Top view. b) Side view. c) and d) are the top view and side vie w of the system with charge density difference drawn on isosurface of 0.0 02 / 3

PAGE 143

143 a b c d Fig ure 5 34. Two water layers adsorbed on surface 7 (oxygen displacement) with protons pointing up in the frist water layer. a) Top view. b) Side view. c) and d) are th e top view and side view of the system with charge density difference drawn on isosurface of 0.0 005 / 3

PAGE 144

144 a b c d Fig ure 5 35. Two water layers adsorbed on surface 7 (oxygen displacement) with protons pointing down a) Top view. b) Side view. c) and d) are the top view and side vi ew of the system with charge density difference drawn on isosurface of 0.0 0 2 / 3

PAGE 145

145 CHAPTER 6 HYDROXYLAT IO N OF ALPHA QUARTZ (0001) SURFACES 6.1 Introduction We have already mentioned the easy hydroxylation of a newly cleaved silica surface under atmosphere conditions [5, 6] and the contrast hydrophilic properti es with the hydrophobic properties of such a surface prepared in high vacuum [12, 16] In simulation, t he hydroxylation of silica surfaces ha s been studied by sever al groups by using silica cluster models [73, 74] and amorphous, hysdroxylated or crystalline silica slab s [13 15, 22, 27, 28, 30, 75, 88] On a newly cleaved silica surface, the oxygen and silicon dangling bond s interact exothermically with water and form hydroxyl groups [13 15, 28, 75] Du et al. concluded that the high strain site s like two member ed rings are easy to be hydroxylated [27] Rignanese e t al. studied the hydroxylation of a perfect reconstructed alpha quartz (0001) surface (as surface 1 in this work) with DFT calculations [15] They concluded that surface 1 is hydrophobilc and stable, since surface hydroxylation by one water molecule is not energetically preferred. Rimola and Ugliengo studied the hydroxylation mechani sm of a two member ed ring defect with a cluster model [74] The vibrational spectrum of the defect and the hydroxyl groups were calculated to compare with experiment. They also found that the hydroxylat ion had been speeded up when there was more than one water molecule adsorbed. Each of the silica polymorphs possesse s different chemical properties and defects in silica bulk and on silica surfaces increase the variety of chemical activity There is still need for effort devoted to the understanding of the silica hydroxylation mechanism at atomic scale. In this chapter we report on the hydroxylation of six alpha quartz (0001) surfaces studied in the preceding chapters Two perfect surfaces and four

PAGE 146

146 defecti ve surfaces are included T he same nomenclature as used in C hapter 4 and 5 still applies in this chapter. 6.2 Method s The initial configurations were prepared by breaking some selected Si O, Si Si or O O bonds on surfaces and insert ion of one piece of a d issociated water molecule ( a proton or a hydroxyl radical ) on to the dangling bonds created. Here are the examples. ( 6 1 ) ( 6 2 ) ( 6 3 ) Figure 6 1 shows two examples of the initial configurations for two surfaces. DFT optimization calculations were done on those prepared initial configurations and the hydroxylation energy is calculated by the following equation ( 6 4 ) Here, is the total energy of the hydroxylated surface system, is the energy of one optimized alpha quartz (0001) surface (See Table 3 2, 3 5), and is the energy of one optimized water molecule. The supercell vectors of the systems were the same as those surface systems in Chapter 3 and all of the DFT parameters for optimization calculations were the same as reported in section 3.2. 6.3 Results 6 3 1 Hydroxylation on Surface s 1 and 2 ( (1x1) and (1x2) Pefect Surface ) On surface 1, there are three in equivalent hydroxylation sites O ne of them is involved in a three member ed ring the other two are not. On surface 2, there are six in equivalent hydroxylation sites T wo of them are involved in two separated three

PAGE 147

147 member ed rings and the others are not. In Figure 6 2 we label the hydroxylation sites on surface 1 and 2 for the calculations For each site two or three slightly different initial configurations of hydroxylation are tried Different kind s of optimized configurations were observed. Some have one hydroxyl group with its oxygen atom three coordinated some have two of them and some recover the status of separate surface and water molecule Table s 6 1 and 6 2 give the hydroxylation energies of those systems which keep hydrox yl group s According to Table 6 1, all of the optimized hydroxylation processes are endothermic and most of the hydroxylation energies range from 1.29 eV to 1.68 eV, except that the larger energies corre sponding to site 3 on surface 2 (~2.5eV). There are two major type s of hydroxylation structures One type has two hydroxyl groups with their oxygen atoms bonding to t he same two silicon atoms In this type the oxygen atoms of the hydroxyl groups are three coordinated and the both bonded silicon atoms are f ive coordinated T he other type has just one hydroxyl group behaving this way (Figure 6 3 ) and only on e silicon atom is five coordinated Figure 6 4 show s the other two types of hydr oxylation observed on site 3 of surface 2 with higher hydroxylation energi es The first one has one hydroxyl group bonding to the two silicon atoms on site 3 of surface 2 The other hydroxyl group transfers to site 2, which results in two silicon atoms on site 2 being five coordinated. In the second hydroxylation type, the proton of one hydroxyl group is transferred to the oxygen atom on site 4; only one silicon atom on site 3 is five coordinated. The re is no obvious difference between the hydroxylation on a site involved in a three member ed ring a nd not. It seems that the three member ed rings are not highly

PAGE 148

148 strained compared to other six member rings on surface 1 and 2 Rignanese et al. reached the same conclusion in their investigation [15] They reported that the hydroxylation structure has two hydroxyl groups without three coordinated oxygen atom and fi ve coordinated silico n atoms on surface 1 and the ir hydroxylation energy (4.69 eV) is much higher than our calculation s The discrepancy remains to be clarif ied but our results and those of Rignanese have the same conclusion namely that the hydroxylation on perfect surfaces is not energetically preferred. And indeed, in many cases which we tried the optimized system s recover ed the states with the separate surface and water molecule. The charge on each atom varies a little at the hydroxylation site. In general, the three coordinated oxygen atom of the hydroxyl group ha s charge ranging from 7.43 e to 7.47 e, but not the three coordinated oxygen atom which of hydroxyl group has less charge ranging from 7.38 e to 7.40 e. 6 3 2 Hydroxylation on Defective Surfaces We se t up the initial configurations for surface 4 and 5 according to equation (6 2) that break s the Si Si bond of the defect and attaches a hydroxyl radical and a proton onto the silicon dangling bonds For surface 6 and 7, we set up the initial configurations according to equation (6 3); we did not try the setup of equation (6 2), since we already showed that the peroxy linkage is more active than the oxygen vacancy site (see the charge density differenc e of water adsorption on surface 6 and 7). On surface 5 and 7, the optimized structures relaxed to separate surfaces and water molecules. In particular, the defect on surface 7 wa s repaired and surface 7 relaxed to surface 1. This phenomenon ha d been disc ussed by Lee et al. and Lockwood et al. [92, 93] In their stud ies water play ed an important role in mediating the processes of defect repair. On surface 4, the hydroxylation is stable with hydroxyl group and proton

PAGE 149

149 terminated silicon coexist ing after optimization (Figure 6 5 (a) and (b) ). The hydroxylation is still not energetically preferred (note its positive hydroxylation energy ) but it has much smaller hydroxylation energies than for perfect surfaces (more than 10 times smaller) This indicates that an oxygen vacancy is a potential site for hydroxylation. The charge on the hydroxyl group is about 0.1 to 0.2 e more than the three coordinated hydroxyl group on the perfect surface. On the other hand, the H S i terminated group possesses 2.6 e I t seems that most of the electron s residing in the Si Si defect site transfer to H Si group after hydroxylation and the silicon atom on h ydroxyl group gains only 0.85 e which is slightly higher than the average charge gained by silicon atoms in alpha quartz bulk (0.81 e) On surf ace 6, the optimized structure has both hydroxyl an d hydroperoxyl groups bonding t o the same pair of silicon atoms T wo oxygen atoms are three coordinated and two silicon atoms are five coordinated (Figure 6 5 (c) and (d)). The hydroxylation energy sh ows t hat this is a n exothermic process and it is much more preferred compared to water molecules adsorption on surface 6 (Table 4 1). The three coordinated hydroxyl group gains similar charge as on surface 4. On the other hand, the oxygen atoms on the hydropero xyl group gain only 6.6 and 6.8 e respectively, which is less than the average number of electrons on the bridge oxygen of the surface (7.59 e). One of the five coordinated silicon atom gains 0.84 e and the other one involved in the Si Si site gains 1.44 e The other silicon atom on the Si Si site gains more electron s (1.75 e) than before hydroxylation. The defect on each surface 4 and 6 is involved in a three member ed ring; the defects on surface 5 and 7 are not. On the other hand, surface s 4 and 6 are ind ividually

PAGE 150

150 less stable than surface d 5 and 7 when comparing their defect formation energies (Table 3 5). Combining the results in this section, we conclude that the defect involved in a three member ed ring on (1x1) perfect surface is less stable and likely to trigger the hydroxylation with water molecules.

PAGE 151

151 Table 6 1 Hydroxylation energy of surfaces 1. Unit: eV. Site/type 1/two three coordinated O atoms 1/one three coordinated O atom 2/two three coordinated O atoms 3/two three coordinated O atoms 3/one three coordinated O atom 1.49 1.41 1.38 1.41 1.68 Table 6 2 Hydroxylation energy of surfaces 2. Unit: eV. Site/type 1/two three coordinated O atoms 2/two three coordinated O atoms 2/one three coordinated O atom 3/two three coordinated O atoms* 3/one three coordinated O atom* 5/two three coordinated O atoms 6/one three coordinated O atom 1.29 1.31 1.35 2.54 2.49 1.32 1.54 Table 6 3 Hydroxylation energy of defective surfaces. Unit: eV. Surface/type 4/ separate hydroxyl group and proton termination 6/ three coordinated O atoms in hydroxyl and hydroperoxyl group 0.12 0.33

PAGE 152

152 a b Fig ure 6 1 Two examples of the initial configurations for surface hydroxylation. a) an attempt site for hydroxylation on surface 1 which is i nvolved in a three member ed ring. b) an attempt site for hydroxylation on the surface 4 which is the oxygen vacancy site. The silicon atoms are represented in yellow balls, oxygen in red, and hydrogen in light grey. a b Fig ure 6 2. The hydroxylation sites on surface 1 (a) and surface 2 (b). a) Site 1 is involved in a three member ring underneath the surface; the other two sites are not. b) Site 1 and 4 are involved in separate three member ring s, other sites are not.

PAGE 153

153 a b c d Fig ure 6 3. Two types of hydroxylation on surface 1. a) and b) are the side view and top view of the first type in which the oxygen atoms of two hydroxyl groups bond to the same two silicon atoms. c) and d) are the side view and top view of the second type in which only one hydroxyl group bond s to two silicon atoms.

PAGE 154

154 a b c d Fig ure 6 4. Two types of hydroxylation of surface 2 with higher hydroxylation energy. a) and b) are the side view and top view of the first type; in which, the oxygen atoms of two hydroxyl groups bond to two silicon atoms (not the same pair). c) and d) are the side view and top view of the second type in which only one hydroxyl group bond s to two silicon atoms and two hydroxyl group s do not bond to the same silicon atom

PAGE 155

155 a b c d Fig ure 6 5. Hydroxylation on surface 4 and surface 6. a) and b) are the side view and top view of hydroxylation on surface 4. c) and d) are the side view and top view of hydroxylation on surface 6.

PAGE 156

156 CHAPTER 7 SUMMARY With the power of DFT and Classical MD computational calculations, we have investigated the properties of several alpha quartz (0001) surfaces and water silica interaction on those surfaces. There are five dehydrated surface reconstruction s found with close surface energies. Two of the surfaces have (1x1) symmetry and the other three possess (2x1) symmetry on the surface plane. All of them are characterized by six member ed rings on the topmost layer and three member ed rings un derneath. The SiO 4 tetrahedra within surface slabs are perfectly conne cted without any defects. T hese five surfaces possess the same topological structure such that they can transform among each other without bond breaking. T he low energy state of the perf ect surface with (2x1) symmetry can explain the observed (2x2) diffraction pattern on alpha quartz (0001) surface when combined with what Steurer et al proposed [16] That is a surface cleavage may not be perfect and may, instead, consist of several steps which display three orientations of the (2x1) surface. Th e energy barriers from (1x1) to (2x1) perfect surfaces we re calculated The lowest one explains well why the (2x2) diffraction spectrum was observed after heat treatment at a temperature roughly higher than when quartz twinning happen s [62] Water adsorption wa s investigated on one (1x1), one (2x1) perfect surfaces, one hydroxylated and fo ur defective surfaces. Water molecule s were dropped on to surfaces one by one to observe the variation of water silica interaction. The interaction between water molecule(s) and hydroxyl group(s) on the fully hydroxylated surface is much stronger than other types of surface and is comparable to the strength of hydrogen bonds between water molecules. The water silica interaction on perfect surfaces is

PAGE 157

157 much weaker than on the fully hydroxylated surface, but an enhancement wa s obs erved when two water molecules were adsorbed The charge density difference show s a quadrup o le pattern on the perfect surfaces with two water molecules adsorbed The defects made on the (1x1) perfect surface do enhance water molecule adsorption. However, on both perfect surfaces and four defective surfaces, the water silica interaction will compromise to hydrogen bonds forming within water cluster. The water silica interaction investigation was done in another way by depositing water layer s (one sub layer of Ice XI) one by one onto surfaces. On perfect surfaces and defective surfaces, two types of water layer deposition were tested. One has out of plane protons pointing down and the other has protons pointing up. T he ground state of water layer adsorption changes from pr otons pointing down depositing to protons pointing up when there are two or more water layers adsorbed. The hydrogen bonds between water layers strongly affect the optimization direction and result in the formation of a stable water bilayer [94] As a consequence, the silica surface property changes f ro m hydrophilic (protons pointing down) to hydrophobic (pr otons pointing up ) The charge density difference shows that the effect of defects on water adsorption is short range and will be shielded by water layer/molecules. The interaction between one water layer and a full hydroxylated surface is much stronger th an on other surfaces. However, the first water layer and hydroxyl groups on the surface form a stable bilayer structure which effectively shield s the interaction between the silica surface and more following adsorbed water molecule experiment [17] must result from the vicinity of the surface [94]

PAGE 158

158 The hydroxylation on perfect surfaces was tested on many oxygen sites All of the results show that the perfect ly reconstructed alpha quartz (0001) surface s are stable and not eas ily hydroxylated. The hydroxylation energies on defective surfaces are much lower than that on perfect surfaces. In particular, an exothermic process is observed on the hydroxylation of a peroxy linkage which is involved in a three member ring. The surface defects are the possible triggers for surface hydroxylation. However, many cases of the hydroxylation on defective surface s were not successful; the optimized systems with separate surface and water molecule were observed The mechanism of silica su rface hydroxylation/repair will need more careful molecular dynamics modeling A system consisting of two defects and more than one water molecules would be a modeling candidate for investigating the complex interactions.

PAGE 159

15 9 REFERENCES [1] F. Franks, Water: A Comprehensive Treatise (Plenum Press, New York, 1972) [2] V. F. Petrenko and R. W. Whitworth, Physics of Ice (Oxford University Press, Oxford ; New York, 1999) [3] C. G. Salzmann, P. G. Radaelli, E. Mayer, and J. L. Finney, Physical Review Letters 103 105701 (2009). [4] P. G. Debenedetti and H. E. Stanley, Phys ics Today 56 40 (2003). [5] R. K. Iler, The Chemistry of Silica : Solubility, Polymerization, Colloid and Surface Properties, and Biochemistry (Wiley, New Y ork, 1979) [6] A. P. Legrand, The Surface Properties of Silicas (John Wiley, Chichester ; New York, 1998) [7] J. Bradbury, Plos Biology 2 1512 (2004). [8] E. N. Fung and E. S. Yeung, Analytical Chemistry 67 1913 (1995). [9] A. T. Woolley and R. A. Math ies, Analytical Chemistry 67 3676 (1995). [10] D. Schmalzing, A. Adourian, L. Koutny, L. Ziaugra, P. Matsudaira, and D. Ehrlich, Analytical Chemistry 70 2303 (1998). [11] T. Nash, A. C. Allison, and J. S. Harington, Nature 210 259 (1966). [12] F. Bart and M. Gautier, Surface Science 311 L671 (1994). [13] N. H. de Leeuw, F. Manon Higgins, and S. C. Parker, Journal of Physical Chemistry B 103 1270 (1999). [14] Z. M. Du, and N. H. de Leeuw, Surface Science 554 193 (2004). [15] G. M. Rignanese, J. C. Charlier, and X. Gonze, Physical Chemistry Chemical Physics 6 1920 (2004). [16] W. Steurer, A. Apfolter, M. Koch, T. Sarlat, E. S nderg rd, W. E. Ernst, and B. Holst, Surface Science 601 4407 (2007). [17] D. B. Asay and S. H. Kim, Journal of Physic al Chemistry B 109 16760 (2005). [18] V. Ostroverkhov, G. A. Waychunas, and Y. R. Shen, Physical Review Letters 94 046102 (2005). [19] Y. R. Shen and V. Ostroverkhov, Chemical Reviews 106 1140 (2006). [20] P. Gallo, M. A. Ricci, and M. Rovere, Journal of Chemical Physics 116 342 (2002). [21] Z. M. Du and N. H. de Leeuw, Dalton Transactions, 2623 (2006). [22] J. J. Yang and E. G. Wang, Physical Review B 73 035406 (2006). [23] P. J. Feibelman, Science 295 9 9 (2002). [24] E. T. Castellana and P. S. Cremer, Surface Science Reports 61 429 (2006).

PAGE 160

160 [25] C. Janiak, T. G. Scharmann, and S. A. Mason, Journal of the American Chemical Society 124 14010 (2002). [26] M. Antognozzi, A. D. L. Humphris, and M. J. Miles, Applied Physics Letters 78 300 (2001). [27] M. H. Du, A. Kolchin, and H. P. Cheng, Journal of Chemical Physics 119 6418 (2003). [28] W. A. Adeagbo, N. L. Doltsinis, K. Klevakina, and J. Renner, Chemphyschem 9 994 (2008). [29] A. Opitz, S. I. U. Ahmed, J A. Schaefer, and M. Scherge, Wear 254 924 (2003). [30] J. J. Yang, S. Meng, L. F. Xu, and E. G. Wang, Physical Review B 71 035413 (2005). [31] M. C. F. Wander and A. E. Clark, Journal of Physical Chemistry C 112 19986 (2008). [32] D. Argyris, D. R. Co le, and A. Striolo, Journal of Physical Chemistry C 113 19591 (2009). [33] J. C. Fogarty, H. M. Aktulga, A. Y. Grama, A. C. T. van Duin, and S. A. Pandit, Journal of Chemical Physics 132 174704 (2010). [34] G. H. Findenegg, S. Jhnert, D. Akcakayiran, and A. Schreiber, Chemphyschem 9 2651 (2008). [35] S. Jhnert, F. V. Ch vez, G. E. Schaumann, A. Schreiber, M. Schnhoff, and G. H. Findenegg, Physical Chemistry Chemical Physics 10 6039 (2008). [36] X. X. Liu Q. Wang, X. F. Huang, S. H. Yang, C. X. Li, X. J. Niu, Q. F. Shi, G. Sun, and K. Q. Lu, Journal of Physical Chemistry B 114 4145 (2010). [37] M. Rovere, M. A. Ricci, D. Vellati, and F. Bruni, Journal of Chemical Physics 108 9859 (1998). [38] E. B. Moor e, E. de la Llave, K. Welke, D. A. Scherlis, and V. Molinero, Physical Chemistry Chemical Physics 12 4124 (2010). [39] L. Verlet, Physical Review 159 98 (1967). [40] R. W. Hockney, Methods in Computational Physics 8 136 (1970). [41] W. C. Swope, H. C. A ndersen, P. H. Berens, and K. R. Wilson, Journal of Chemical Physics 76 637 (1982). [42] H. J. C. Berendsen, J. P. M. Postma, W. F. v an G unsteren, A. Dinola, and J. R. Haak, Journal of Chemical Physics 81 3684 (1984). [43] S. Nos Molecular Physics 52 255 (1984). [44] W. G. Hoover, Physical Review A 31 1695 (1985). [45] P. Hohenberg and W. Kohn, Physical Review 136 B864 (1964). [46] W. Kohn and L. J. Sham, Physical Review 140 A 1133 (1965).

PAGE 161

161 [47] D. M. Ceperley and B. J. Alder, Physical Review Letters 45 566 (1980). [48] F. Herman, J. P. Van Dyke, and I. B. Ortenburger, Physical Review Letters 22 807 (1969). [49] J. P. Perdew and K. Burke, International Journal of Quantum Chemistry 57 309 (1996). [50] A. D. Becke, Physical Review A 38 3098 (1988). [ 51] Y. Wang and J. P. Perdew, Physical Review B 43 8911 (1991). [52] J. P. Perdew, K. Burke, and M. Ernzerhof, Physical Review Letters 77 3865 (1996). [53] R. M. Martin, Electronic Structure : Basic Theory and Practical Methods (Cambridge University Press, Cambridge ; New York, 2004) [54] S. K. Ma and K. A. Brueckner, Physical Review 165 18 (1968). [55] C. Herring, Physical Review 57 1169 (1940). [56] C. Herring and A. G. Hill, Physical Review 58 132 (1940). [57] D. R. Hamann, M. Schlter, and C. Chiang, Physical Review Letters 43 1494 (1979). [58] N. Troullier and J. L. Martins, Physical Review B 43 1993 (1991). [59] D. Vanderbilt, Physical Review B 41 7892 (1990). [60] P. E. Blchl, Physical Review B 50 17953 (1994). [61] G. Kresse, and J. Fu rthmller, VASP, http://cms.mpi.univie.ac.at/marsweb/ (Institut fr Materialphysik, Universitt Wien Vienna, Austria, 1999 ) [62] Y. W. Chen, C. Cao, and H. P. Cheng, Applied Physics Letters 93 181911 (20 08). [63] R. B. Sosman, The Properties of Silica; an Introduction to the Properties of Substances in the Solid Non conducting State (Book Dept. Chemical Catalog Co., New York, 1927) [64] I. J nossy and M. Menyh rd, Surface Science 25 647 (1971). [65] G. M. Rignanese, A. De Vita, J. C. Charlier, X. Gonze, and R. Car, Physical Review B 61 13250 (2000). [66] M. V. Koudriachova, J. V. L. Beckers, and S. W. de Leeuw, Computational Materials Science 20 381 (2001). [67] M. Kawasaki, K. Onuma, and I. Sunagaw a, Journal of Crystal Growth 258 188 (2003). [68] W. Smith and T. R. Forester, Journal of Molecular Graphics 14 136 (1996). [69] B. W. H. van Beest, G. J. Kramer, and R. A. van Santen, Physical Review Letters 64 1955 (1990).

PAGE 162

162 [70] G. Mills, H. J nsson, a nd G. K. Schenter, Surface Science 324 305 (1995). [71] M. T. Dove, A. P. Giddy, and V. Heine, Ferroelectrics 136 33 (1992). [72] C. Frondel, American Mineralogist 30 447 (1945). [73] Y. C. Ma, A. S. Foster, and R. M. Nieminen, Journal of Chemical Physics 122 144709 (2005). [74] A. Rimola and P. Ugliengo, Journal of Chemical Physics 128 204702 (2008). [75] V. V. Murashov, Journal of Physical Chemistry B 109 4144 (2005). [76] R. F. W. Bader, Atoms in Molecules : A Quantum Theory (Clarendon Press, Oxford ; New York, 1990) [77] G. Henkelman, A. Arnaldsson, and H. J nsson, Computational Materials Science 36 354 (2006). [78] G. Pacchioni, L. Skuja, and D. L. Griscom, Defects in SiO2 and Related Dielectrics : Science and Technology (Kluwer Academic Publishers, Dordrecht, Netherlands, Boston, MA, 2000) [79] L. Skuja, M. Hirano, H. Hosono, and K. Kajihara, Physica Status Solidi. (C) 2 15 (2005). [80] C. M. Carbonaro, V. Fiorentini, and F. Bernardini, Physical Review Letters 86 3064 (2001). [81] G. Roma, Y. Limoge, and S. Baroni, Physical Review Letters 86 4564 (2001). [82] Y. Duval, J. A. Mielczarski, O. S. Pokrovsky, E. Mielczarski, and J. J. Ehrhardt, Journal of Physical Chemistry B 10 6 2937 (2002). [83] G. Roma and Y. Limoge, Physical Review B 70 174101 (2004). [84] L. Martin Samos, Y. Limoge, J. P. Crocombette, G. Roma, N. Richard, E. Anglada, and E. Artacho, Physical Review B 71 014116 (2005). [85] A. K. Verma and B. B. Karki, Phy sical Review B 79 214115 (2009). [86] S. R. George, S. C. Langford, and J. T. Dickinson, Journal of Applied Physics 107 033108 (2010). [87] L. Martin Samos, G. Roma, P. Rinke, and Y. Limoge, Physical Review Letters 104 075502 (2010). [88] J. J. Yang, S. Meng, L. F. Xu, and E. G. Wang, Physical Review Letters 92 14602 (2004). [89] R. Notman and T. R. Walsh, Langmuir 25 1638 (2009). [90] R. T. Cygan, J. J. Liang, and A. G. Kalinichev, Journal of Physical Chemistry B 108 1255 (2004). [91] D. Pan, L. M. L iu, G. A. Tribello, B. Slater, A. Michaelides, and E. Wang, Physical Review Letters 101 155703 (2008).

PAGE 163

163 [92] J. W. Lee, M. Tomozawa, and R. K. MacCrone, Journal of Non Crystalline Solids 354 1509 (2008). [93] G. K. Lockwood and S. H. Garofalini, Journal of Nuclear Materials 400 73 (2010). [94] Y. W. Chen and H. P. Cheng, Applied Physics Letters 97 161909 (2010).

PAGE 164

164 BIOGRAPHICAL SKETCH Yun Wen Chen was born in Taipei, Taiwan, Republic of China in 1975. He was highly in terested in nature science since his childhood and loved to read the stories about Newton, Einstein, Curie, Bell, Stephenson, and also Edison. Somehow before he knew who made more money, he chose to register in the physics department for his bachelor degre e because he thought physics looks more fun than engineering and it did. He was working with Dr. Yu on a topic of the quantum interference of electromagnetic field to cold 87 Rb spectroscopy and then he got his master degree in physics in Tsing Hua Universi ty, Taiwan. After another couple years of military service and working experience, Yun wen joined the Department of Physics in UF in 2004 to pursue his Ph.D degree. He met Dr. Hai Ping Cheng and started working on the computational simulation on water sili ca interface. He learned the nature of water and silica interaction via simulation and also tools that apply physics and chemistry ideas in simulations. He will like to d evote himself in computational applied physics or chemistry for future academic research.