This item has the following downloads:
1 A DISPERSION CORRECTED DFT STUDY OF H 2 OXIDATION ON THE PdO(101) SURFACE By RAHUL A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2013
2 2013 Rahul
3 To my parents
4 ACKNOWLEDGMENTS I would like to express m y sincere thanks to my family for their love and support th roughout. I also thank my advisor, Dr. Jason Weaver for his invaluable advic e an d suggestions for my research work at University of Florid a. I would like to extend my sincere gratitude to Abbin Antony and Dr. Arav ind Asthagiri for guiding and teaching me simulation technique VASP I extend my gratitude to Dr. Sergey Vasenkov for agreeing to evaluate my thesis and serve as my committee member. I also acknowledge Ohio supercomputing Center for providing computational resources for performing calculations reported in this thesis. I also thank faculty members, staff, and students of Che mical E ngineering Department for providing me for helping and guiding me through graduate program and providing me access to literature for my research work.
5 TABLE OF CONTENTS Page ACK NOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 6 LIST OF FIGURES ................................ ................................ ................................ .......... 7 LIST OF ABBREV IATIONS ................................ ................................ ............................. 8 ABSTRACT ................................ ................................ ................................ ..................... 9 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 10 2 COMPUTATIONAL PROCEDURE ................................ ................................ ......... 14 2.1 Density Functional Theory (DFT) ................................ ................................ ...... 14 2.2 Dispersion corrected DFT ................................ ................................ ................. 16 2.3 Computational Model ................................ ................................ ........................ 17 2.4 Analytical Methods ................................ ................................ ............................ 19 2.4.1 Normal Mode analysis ................................ ................................ ............. 19 2.4.2 Nudged Elastic Band Method ................................ ................................ .. 20 3 RESULTS AND DISCUSSION ................................ ................................ ............... 22 3.1 H 2 Adsorption and Dissociative Chemisorption on PdO (101) .......................... 22 3.2 H 2 O on PdO(101) Surface ................................ ................................ ................ 24 3.3 H atom Diffusion on PdO (101) Thin Film ................................ ......................... 27 3.4 Formation of H 2 O on PdO(101) Thin Film Surface at 0.08 ML H 2 Coverage .... 29 3.5 Coverage Effect of H 2 in Formation of H 2 O ................................ ....................... 32 3.5.1 Formation of Water at 0.35 ML H 2 (100% cus ) Coverage on PdO (101) Surface ................................ ................................ ................................ 32 3.5.2 Formation of Water at 0.26 ML H 2 (75% cus ) Coverage on PdO (101) Surface ................................ ................................ ................................ .......... 33 3.6 Effect of Presence of Hydroxyl Moiety and H atom Diffusion on PdO(101) Surface on water formation ................................ ................................ .................. 35 4 CONCLUSIONS ................................ ................................ ................................ ..... 45 LIST OF REFERENCES ................................ ................................ ............................... 48 BIOGRAPHIC AL SKETCH ................................ ................................ ............................ 51
6 LIST OF TABLES Table P age 3 1 Vibrational frequencies (wave numbers) of various stable and transition state configurations ................................ ................................ ................................ ..... 39 3 2 Comparison of binding energies of various configurations of H 2 on PdO(101) surface at low coverage on 4x1 and 4x2 PdO(101) slabs ................................ .. 40
7 LIST OF FIGURES Figure P age 2 1 Hard ball model representation of stoichiometric PdO(101) surface. .................. 21 3 1 Model representation of adsorption of H 2 on PdO(101) surface followed by dissociative adsorption and subsequent formation of H 2 O with adjacent cus vacancy on PdO(101) surface. ................................ ................................ ........... 37 3 2 Potential energy surface profiling of H 2 activity on PdO(101) surface to produce H 2 O at low H 2 coverage ................................ ................................ ........ 38 3 3 ZPC potential energy evolution of H 2 for H 2 O formation on PdO(101) surface at low H 2 coverage ................................ ................................ ............................. 41 3 4 H 2 O formation on PdO(101) surface via high energy pathway ........................... 42 3 5 H 2 O formation pathways through disproportionation of 2 adjacent cus OH at 0.26 ML of H 2 coverage on PdO(101) surface ................................ .................... 43 3 6 Effect of OH group presence in H 2 O formation ................................ ................... 44
8 LIST OF ABBREVIATIONS Angstrom (unit of length) BE Binding e nergy C# C onfiguration n umber, e.g. C1, C2 cs Coordinately saturated cus Coordinately unsaturated DFT Density f unctional t heory DFT D3 Dispersion c orrected d ensity f unctional t heory eV Electron volt (unit of energy) FS Final s tructure HRCLS High resolution core level shift IS Initial structure LEED Low e nergy e lectron d iffraction NEB Nudged e lastic b and NMC Normal m ode c alculation PAW Projector a ugmented w ave SCF Self c onsistent f ield TS Transition s tructure VASP Vienna a b initio s imulation p ackage XC Exchange c orrelation ZPC Zero p oint c orrected
9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master o f Science A DISPERSION CORRECTED DFT STUDY OF H 2 OXIDATION ON THE PdO(101) SURFACE By Rahul Aug ust 2013 Chair: Jason F. Weaver Major: Chemical Engineering We use dispersion corrected DFT to explore the H 2 chemi sorption process on PdO(101) surface leading to PdO(101) surface by H 2 O formation H 2 dissociates on the PdO(101) surface through a precursor mediated mechanism wherein the physisorbed H 2 acts as the precursor to dissociation. We find that t he H atom bound to the cus Pd site can diffuse freel y along the cus Pd row as the estimated energy barrier for diffusion is less than the barrier for H 2 dissociation on the surface. We also show that there is more than one pathway available for H 2 to reduce the PdO(101) surface by reacting with the lattice cus O and forming water. Disproportionation of two adjacent OH groups on PdO(101) surface leads H 2 O adsorbed on adjacent cus Pd sire lea ving a vacancy on the surface. The activation barrier for this reaction is ~82.8 kJ/mol H 2 which is almost independent o f the H 2 coverage on PdO(101) surface. However, we also find that at the saturation coverage, hydroxyl disproportionation is theoretically not feasible and H 2 O can only form through reaction of the surface OH moiety with the H atom chemisorbed on cus Pd at om, which has a high energy ba rrier of 136 kJ/mol H 2
10 CHAPTER 1 INTRODUCTION In past few decades, chemistry of late transition metal (TM) oxides has drawn much attention from the science community due to its extensive usage in the field of catalysis, gas sensors and electro chemistry. The primary goal of these studies is to gain atomic scale expertise into their surface structure, composition and the underlying mechanism of reactions involved. Surface science studies under a controlled environment such as ultra high vacuum (UHV) en able us to understand the underlying chemistry of these surfaces at the f undamental level. Although UHV studies have produced significant results, the knowledge is still limited due to structural complexities of the TM O surfaces and experimental limitations of UHV set up. Computational simulations have been proven to be an eff ective to fill these knowledge gaps to a good extent by providing predictions on structural, vibrational, energetic and electronic properties of adsorbate adsorbent system. Growing the oxide on the metal surface is a complex process because initial chemiso rption of oxygen can give rise to surface reconstruction, and may lead to formation of thin epitaxial surface oxides before the formation of bulk oxide. Characterization of these thin oxide films is a challenging task as the properties of the thin oxide fi lms may differ from that of the corresponding bulk oxides 1 The TM oxide surface properties are also found to be a function of oxygen coverage which also affects the surface reactivity. Pd based catalysts are commonly used in catalyt ic converters for CH 4 combustion at low temperatures to reduce CH 4 emission from automobiles and in gas turbines for alkane combustion PdO holds a unique place among late TM oxides due to its ability to completely oxidize low end alkanes. 2 12 Studies have shown that the
11 formation of PdO is respons ible for the favorable performance of supported Pd catalysts. The high activity of PdO as a complete oxidizing catalyst has motivated in depth investigation of the surface chemistry of PdO Recent studies have shown that the dissociation of C H bond in alk anes on PdO (101) surface at low temperatures occurs by a precursor mediated mechanism 7 in which adsorbed molecule serves a precursor to dissociation. S mall alkanes bind strongly on complexes on cus Pd sites 10 complexes involve electron donation from and back donation to the alkane molecule and d orbitals of the cus Pd atom which on one hand increases alkane adsorption energy on PdO(101), and weakens the C H bonds on the other, le ading to C H bond dissociation. The knowledge of dehydrogenation process of alkanes has motivated the study of the interactions of H 2 with TM oxide, where adsorption and diffusion of h ydrogen as intermediate steps. The simplest atomic structure of Hydrogen molecule make it suitable candidate for study of bond formation and subsequent surface r eactions. In addition to this, hydrogen oxidation is prevalent in catalytic oxidation of all alkanes and it may lead to PdO(101) surface reduct ion In this thesis, we examine the behavior of H 2 on PdO(101) surface computationally and concentrate on searching for low energy pathway for hydrogen o xidation by the surface oxygen. Previous studies of H 2 on RuO 2 (110) and PdO(101) have provided us a bas ic understanding of H 2 adsorption behavior on TM oxide surfaces 13 16 UHV experiments as well as DFT studies confirm the formation of complex between molecularly chemisorbed H 2 molecule and coordinatively unsaturat ed TM atom on the surface 14 17 H 2 /RuO 2 (110) studies present strong evidence of precursor mediated dissociation of H 2
12 over RuO 2 (110) 16 18 This complex weakens the H H bond leading to diss o ciation of H 2 on RuO 2 (110) These workers f urther report the formation of dihydride species (2H O br ) through reaction between molecularly adsorbed H 2 and bridging O atom (O br ) on RuO 2 (110) surface. These studies present the possibility of H 2 oxidation by t he coordinatively unsaturated oxyg en atom present on the surface. Molecule binding and surface reactions of PdO(101) has certain resemblance with RuO 2 (110) surface shares common characteristics. Cus Pd and cus O atoms in stoichiometric PdO(101) predominantly define the reactive properties of PdO(101) surface 7 8 14 19 22 Hakanoglu et al. showed that H 2 binds strongly on cus Pd sites by forming complexes. TPD experiments have shown that irrespective of the amount of H 2 exposed to the PdO(101) surface, most of the H 2 reacts with the surface to yield H 2 O 14 The TPD results presented by Hakanoglu et al. show that although most of the water desorbs in a sharp peak around 350 K after exposing PdO(101) to H 2 a small amount of water desorbs from the surface at higher temperatures as well 14 These results imply a possibility of multiple pathways for water production. Blanco Rey et al. predicted through DFT calculations that H 2 can form H 2 O by recombining with the cus O on PdO(101) though a high energy barrier of 1.94 eV was reported 13 The gr oup also suggested that the presence of neighboring hydroxyl group that is H atom adsorbed on the neighboring cus O atom dramatically lowers the barrier for water formation to 0.69 eV. The experimental results however show that the water formation on PdO(1 01) surface after H 2 exposure is nearly independent of the amount of exposure 14 These experimental and computation results motivate us to investigate the H 2 O formation process on PdO(101) after H 2 exposure in detail. In this document, we
13 examine the overal l process of H 2 adsorption on PdO(101) surface followed by reaction of hydrogen with the coordinatively unsaturated lattice oxygen to form H 2 O using dispersion corrected DFT (DFT D3). We explore different possible mechanisms for water formation and compare the obtained results with the already available experimental data. We learn that the H 2 dissociates on PdO(101) surface through a precursor mediated mechanism in which the H 2 complex serves as the precursor to dissociation, and t he dissociated H atom ca n move along the cus Pd row freely. We also find that there is more than one pathway available to H 2 to reduce the PdO(101) surface by forming H 2 O We present, in this thesis, a potent pathway with an energy barrier that is in a good agreement with the exp erimental results.
14 CHAPTER 2 COMPUTATION AL PROCEDURE 2.1 Density Functional Theory (DFT) We em ploy dispersion corrected Density Functional Theory (DFT D3 ) to obtain results discussed in this thesis. All DFT D3 calculations were performed using Vienna Ab initio Simulation Package (VASP ) 23 26 DFT is nowadays is commonly used computational quantum mechanical modeling approach to predict properties of isolated molecules, fluids, bulk solid surfaces such as metals, semiconduc tors and insulators and material interfaces. DFT utilizes functionals (functions of another function) of electron density to determin e various properties of matter. DFT computes the electronic structure of matter by solving the time dependent Schrdinger e quation which describes the spatial shape and temporal evolvement of a wave function in a given potential and for provided boundary conditions. The Schrdinger equation for a many body system is complex c ollection of atomic (electronic & nuclei) wave funct ions. Born Oppenheimer approximation (mass of nucleus >> mass of electrons ) is then applied to order to separate electronic wave function with nucleus wave functions. Electronic wave function is further approximated as Hartee Product which is the product o f individual wave function DFT simplifies the many body Schrdinger equation by representing it in terms of the spatial coordinate depen dent electron density function. The foundation of DFT is based on two theorems proven by Kohn & Hohenberg and the set o suggests that ground state energy from Schrdinger equation is a unique functional of electron density while the second theorem implies that the electron density
15 corresponding to the full solution of the Schrdinger equation minimizes the energy of overall functional. The second theorem is also called the variational principle. A functional is written in terms of single electron wave functions that collectively de fine the electron density. The energy functionals can describe the kinetic energy of electrons and various intra and inter atomic interactions. Kohn & Sham proved that solvi ng a set of equations in which each equation considers interactions of a single el ectron can yield the correct electron density. DFT applies an iterative algorithm to solve this closed loop problem. The process is usually referred as electronic relaxation or self consist e nt field (SCF) The user supplies the structure (number of atoms, atomic arrangement and coordinates) and sets required accuracy parameters for electronic and ionic optimization and an initial trial e lectron density is defined. Kohn Sham equation is then solved using the trial electron density to determine the Kohn Sham electron density. Next, the Kohn Sham electron density is compared with the trial electron density. If the energies are equal or difference is within the set parameter, the calculation is understood to have attained self consistency and the energy is consi dered as the ground state energy. If the difference is beyond the set parameter, calculation will be continued until the self consistency is achieved. Once ground state energy is acquired, inter atomic forces are calculated and tested to satisfy the force criterion. If the force criterion is not fulfilled, atoms are rearranged and SCF calculations are done for the new configuration followed by force optimization calculation. These steps are repeated until the configuration attains the minima based on the sp ecified parameters. DFT employs exchange correlational functionals to include quantum mechanical corrections pertaining to treatment of wave functions of electrons and electron densities.
16 The exchange i nteraction considers the energy change of the system d ue to electronic wave function being subjected to exchange symmetry while correlation port ion corrects the over estimation of the energy because of describing electrons as electron densities. Various approximations for XC functional s have been developed su ch as local density approximation (LDA), generalized gradient approximation (GGA), and meta GGA and hybrid functionals. GGA is found to be more accurate than LDA in cases of studying systems with sharper variation in electron densities such as solid surfac es. There are distinct GGA functionals available as the gradient in GGA can be treated differently. Calculations performed as part of thesis utilize GGA PBE exchange correlational functional 27 Electronic wave functions can be further simplified by knowing that core electrons of atoms do not participate in chemical bonds or defining material properties. Using this knowledge, pseudo potentials are used in DFT to eliminate core electron wave function without excluding their impact on total energy. We use projector augmented wave (PAW) 28 29 provided in VASP database. 2.2 Dispersion corrected DFT Accuracy of DFT calculated results is limited in case of insulators and semiconductors because of unaccountability for Van der Waals (dispersion) interactions However, inclusion of dispersion interactions is critical for the system being studied in this thesis. DFT D3 is combination of conventional DFT calculat ions implemented through VASP 23 26 and dispersion correction technique In this thesis, we use the method developed by Grimme et a l. to account for dispersion effects in the system 30 31 The method computes the dispersion effects separately by accounting for non linear exchange for the long range interactions ignored by conventional DFT.
17 Calculated dispersion energies and forces are then added to energies and forces determined by conventional DFT relaxation calculations. The total ionic forces are then relaxed using a limited memory Broyden Fletcher Goldfrab Shanno optimization method 32 until forces are less than the preset parameter. Incorporation of dispersion into DFT is still an active research area as DFT D3 sometimes overestimates the binding energy of dative bonds which are well characterized by conventional XC functionals 2.3 Computational Model As mentioned in previous section, we employed DFT D3 to obtain the results reported in this thesis. total ground state energy calculations are performed using projector augmented wave ( PAW ) pseudo potentials provide d in VASP database 28 29 along with Perdew Berke Ernzerhof (PBE) exchange correlation functionals 27 and plane wav e cut off of 400 eV and Fermi level smearing 33 with a Gaussian width of 0.1 eV. A force relaxation criterion of < 0.03 eV/ for unconstrained atoms was exercise d to obtain all stable structures. Experimentally, PdO(101) film is grown on Pd(111) surface at 500 K though oxidation of the metal surface in UHV by plasma generated atomic oxygen. A 13 thick PdO(101) film consisting of ~3 ML of oxygen atoms id generated. 14 21 22 We defined 1 ML as equivalent to the Pd(11 1) surface atom density of 1.53 x 10 15 cm 2 Fig ure 2 1 provides a top and isometric view of PdO(101) surface. The surface const itu tes alternating rows of coordinatively unsaturated ( cus also known as three fold or 3f ) and coordinatively saturated ( cs al so known as four fold or 4f ) Pd and O atoms. The cus sites (both Pd and O) have been shown to be highly active towards many molecular species. 7 9 12 14 20 21 34 36
18 The computation model surface of PdO(101) is obtained by cutting the relaxed bulk PdO structure along 101 crystal plane. The structure is then strained to meet the reported experimental l attice parameters ( a = 3.057 an d b = 6.352 ) followed by relaxation using DFT D3 relaxation criteria stated The model surface is characterized by a rectangular unit cell where and lattice vectors correspond to and [ direction of PdO(101) crystal, respectively. We use four layer ed PdO(101) which corresponds to a thickness of 9 . All SCF calculations were performed with the bottom PdO layer fixed to provide stability to the structure A 20 of vacuum spacing has been provided in the normal direction to the top surface l ayer to minimize spurious periodic interaction s. Our earlier studies have shown that the thickness of 9 is sufficient to ignore the effects of the metal substrate Pd(111) as the increasing slab thickness does not impact the results of the calculations. 21 A 4 x 8 x 1 Monkhorst Pack k point mesh was used for a 1x1 unit cell and the larger cells were scaled accordingly. A 4 x 2 x 1 and 2 x 2 x 1 k point mesh have been used for 4 x 1 and 4 x 2 unit cell sizes, respectively. The 4 x 1 unit cells were used to study the mo lecular and dissociative adsorption of H 2 on the surface H 2 O formation on PdO (101) due to reduction creates cus O vacancy on the surface which causes major distortion of the crystal; therefore we switched to the larger unit cell of 4 x 2 sizes to investig ate the H2O formation configurations and pathways. The adsorption/desorption/binding energies reported in this thesis are calculated by subtracting the ground state energies of the adsorbate system (adsorbed molecule on the surface) from the sum of total ground state energies for the isolated molecule in
19 as binding energy of the DFT D3 relaxed structure minus the dispersion energy. We calculate the dative energies by performing a single point conventional DFT calculation on the relaxed structure obtained using DFT D3. These calculations allow us to quantify contributions due to the dative and dispersion interactions. 2.4 Analytical Methods 2.4.1 Normal Mode analysis An isolated molecular system comprising of N atoms has total of 3N degrees of freedom 3 translational DOF, 3 rotational DOF and 3N 6 vibrational DOF. A linear molecule has 3N 5 vibrational DOF. The 3N 5(or 6 ) uncoupled DOF are periodic motions in the molecule and are referred as vibration modes, the corresponding frequencies are termed as vibrational frequencies. The translational and rotational modes in an isolated molecule are treated redundant. Although when the molecule is adsorbed on a surface, these 6 modes are no longer redundant Experimentally, these modes can be investigated by infrared spectroscopy or Raman spectroscopy. DFT utilizes numerical finite difference method to determine vibrational states. We use normal mode analyses to confirm whether or not the relaxed configuration is a true minima. The vibrational frequencies are also used to calculate the zero point correct ed energies of different configurations. Vibrational modes are usually calculated by fixing the surface atoms while leaving the molecule unconstrained. In our studies, a cus O atom on the surface is involved in the reaction with two H atoms to form water. However, we find that fixing the complete system but the 2 H atoms and a cus O (involved in H2O formation) does not yield us the imaginary frequencies at the transition state. The imaginary mode was found to be associated with the cus Pd atom with which H 2 O is bonded to after formation. Hence, all NMC calculations in this thesis are
20 performed by setting the adsorbed species (H atoms), cus O and cus Pd atom unconstrained while fixing the rest of the PdO(101) surface. 2.4.2 Nudged Elastic Band Method Nudged Elastic Band (NEB) is a computational method used to determine the saddle point and minimum energy pathway (MEP) between a pair of stable states. 37 The method utilizes the fact that any point on MEP is at an energy minimum in all direction perpendicular to the path. We employ the slightly modified version of NEB called climbing image NEB method which can also dete rmine the transition state for a MEP. 38 39 In the climbing image NEB method, the image with highest energy (climbing image) is identified after few iterations with conventional NEB method. The energy of the climbing image is the maximized while energy along the rest of the band is tried to be minimized. Climbing image calculations provide the saddle poin t which is further confirmed to be transition state through normal mode analysis by looking for an imaginary mode of vibration.
21 Figure 2 1 Hard ball model representation of stoichiometric PdO(101) surface. The surface con stitutes 0.35 ML of cus Pd, cus O, 4 f Pd and 4 f O each. The and lattice vector s correspond to  and [ 101] crystallographic directions of PdO. Atomic color code: Blue Pd, Orange O.
22 CHAPTER 3 RESULTS AND DISCUSSION 3.1 H 2 A dsorption and D issociative C hemisorption on PdO (101) Molecular chemisorption of H 2 and H 2 O has been extensively studied by our group experimentally as well as theoretically. From a previous study done for molecular H 2 adsorption on PdO(101), we gather that most stable configuration of molecular H 2 on PdO(101) is when H 2 molecule sits atop cus Pd and align s parallel to a & b lattice vectors ( parallel to cus Pd row and perpendicular to cus Pd row) DFT D3 predicts the adsorption energies of 56.9 and 48.7 kJ/mol for parallel and perpendicular H 2 configurations respectively Also we observe that the H H bond in the adsorbed species elongates 0.8 4 compared with the gas phase H2 at 0.75 . The Pd H bond length is 1.74 in parallel configuration. Previous studies 10 13 14 16 have well established the complex formation in adsorption of H 2 on a metal oxide surface. A similar type of dative bonding has been re ported in detail for H 2 on RuO 2 (110) surface by Sun Reuter and Scheffler. 16 The Pd H bonding con tains both charge donation and back donation between Pd and H atoms. We calculated the dative bond energy of the parallel and perpendicular configurations of adsorbed H 2 by performing single step DFT on DFT D3 relaxed configurations The dative bond energies of both configurations are estimated to be 47.2 and 39.2 kJ/mol H 2 The difference between DFT D3 energy and dative bond energy implies that dispersion forces stabilize the adsorbed species on the surface. Due to the small size of H 2 we expect a significant impact of zero point correction (ZPC) w hich is also a source of strong kinetic isotopic effects observed experimentally. 14 We calculated the vibration frequencies of both adsorption
23 configurations and it is observed that an imaginary frequency is associated with the H 2 perpendicular configuratio n which indicates that this configuration is not a stable minimum but a transition configuration for dissociation of adsorbed H 2 species aligned parallel to cus Pd row. The vibrational frequencies of the parallel configuration are listed Table 3 1 in column C1. Note that we set 4 atoms ( 2 H atoms, cus Pd and cus O) unconstrained while fixing the rest of PdO(101) surface to ensure that we maintain one reference, that is, isolated H 2 and bare PdO(101) surface, for all calculation performed at low co verage. We then studied the H 2 dissociation pathway described by Hakanoglu et al by using DFT D3. The adsorbed H 2 dissociates to yield the chemisorbed configuration wh erein one H atom remain s adsorb ed on the cus Pd and the other H atom adsorb s on adjacent cus O site (see configur ation C2 in Figure 3 1 ) The binding energy for this chemisorbed H 2 configuration is 132.8 kJ/mol with Pd H and Pd O bond lengths 1.55 and 0.98 , respectively. We estimate the dative bond energy for this structure using the method explained above to be 110.4 kJ/mol. The ZPC correc ted binding energy for the chemisorbed H 2 is estimated to be 108.5 kJ/mol To probe the H 2 dissociation pathway, we used 3 images in the climbing NEB calculation with accuracy limit of 0.0 3 eV/ . The H 2 molecule rotates on the cus Pd atom and instead of attaining a more symmetrical TS (perpendicular H 2 configuration), the H H bond begins to stretches at the same time from an initial value of 0.84 to 1 at the transition state TS1 T he d issociation barrier for this pathway is 39.7 kJ/mol. This low energy barrier assures the stability of the dissociated hydrogen configuration and that the dissociated H 2 will not recombine to yield molecular hydrogen. This result also suggest s the possibility of H 2 O
24 formation in the due process. We find that the ZPC dissociation barrier is 33.5 kJ/mol for H 2 using the vibrational frequencies of initial state and transition states Again, we observe that the DFT D3 predicted dissociation barrier for H2 is a lot lower than the desorption barrier (~42.8 kJ/mol) The vibrational frequencies for initial state (C1), transition state (TS1) and final state (C2) are liste d in Table 3 1 3.2 H 2 O on PdO(101) S urface Blanco Rey and c o workers have proposed a mechanism for H2O formation on PdO(101) which suggests that a water like dihydride structure can form on the PdO(101) surface that will be weekly bound to the cus Pd site and an adjacent vacancy at cus O site 13 The formation of water occurs from the initial state (C2) in which the one H atom is adsorbed on cus Pd site and the other is bonded to the adjacent cus O site. The cus OH group starts to leave the surface and move towards the Pd H group at the transition state giving birth to H2O adsorbed cus Pd atom creating a vacancy at cus O site on the PdO(101) surface However, the barrier for formation of this moiety is reported to be quite large (1.94 eV) which make s it an unlikely pathway for water formation. TPD spectra after exposing the PdO(101) thin film surface to various quantities of H 2 in Ultra high vacuum shows possibility of multiple pathways for water formation desorption limited water will be formed through low energy pathways whereas the water peak at high temperatures will be resulted from high energy pathways 14 Another suggested outlook was that water formed at low temperatures is desorbed at high temperature. Therefore, it becomes essential for us to understand the behavior of water on PdO(101) thin film surface. DFT calculations performed by Kan et al. 21 predict that the H 2 O molecule sits atop cus Pd site in the most preferred configuration with its O atom on top of cus Pd
25 atom forming H 2 O Pd bond and H O H plane aligned almost parallel to the PdO(101) surface. One of the H atoms in H2O points directly towards cus O whereas the other interacts with the 4 fold oxygen atom on the surface providing s tability to the configuration. The configuration has an estimated adsorption energy of 75.2 kJ/ mol with H 2 O molecule. It is suggested that adsorption of water on cus Pd sites is favored over other sites by at least 40 kJ/mol. These DFT results give us an initial understanding of how the water could form on the PdO(101) surface after H 2 adsorption. We now understand that in every possible pathway for H 2 oxidation H 2 O will be most likely to be bonded to cus Pd atom before desorp tion. To get the H 2 O configuration on PdO(101) surface with an adjacent cus O vacancy, we first relaxed structure using a 4x1 slab. We f i nd that the surface distortion was significant due to coverage effect of the cus vacancy and we could not obtain a configuration with H 2 O adsorbed on PdO(101) surface We then relaxed the same configuration on a 4x2 a relaxed configuration with H 2 O formed on the PdO(101) thin film surface with an adjacent vacancy at cus O site (refer C5 in Figure 3 1 ) Unlike H2O on bare PdO(101) surface where O atom lies directly atop cus P d t he O atom in reaction limited H 2 O molecule bonds with cus Pd atom at an bond angle of ~82 .5 PdO(101) surface plane and t he H O H plane rest s away from the PdO(101) surface at a very small angle from surface plane. We also observe from DFT D3 produced relaxed configuration that due to the nearby vacant cus O site, cus Pd atom shifts towards the vacant site distorting the original atomic arrangem ent of the surface. The H 2 O Pd bond length is 2.15 . Both H O bond lengths are 0.98 each and the H O H angle is
26 approximately 105.92 These values are very close to isolated gas phase H 2 O molecule. So, if we get a viable pathway linking configuration C 4 to C5, we can confirm the formation of reaction limited H 2 O on PdO (101) surface We calculated the vi brational frequencies of the relaxed C5 configuration using NM calculation. The frequencies are listed in Table 3 1 in column C5 Note that the configuration C5 has an imaginary mode with a very low frequencies, so we rotated the molecule alon g the imaginary mode and relaxed the structure to obtain a stable structure but a configuration without i maginary f requency could not be obta ined. We reckon that this small imaginary frequenc y indicates soft bond ing between the H 2 O moiety and cus Pd (bond length 2.15 ) and the possibility of immediate desorption of the H 2 O molecule as soon as it is formed. To begin with, we tested the viability of high energy Blanco Rey pathway using DFT D3 which links configuration C2 to C5 with a 5 image climbing NEB calculation Since the configuration C5 was obtained on a 4x2 slab, we relaxed configuration C2 on a 4x2 slab as well which almost has same adsorption energy as 4x1 surface (Table 3 2 provides a comparison of energies between 4x1 and 4x2 surface). The NEB calculation does not converge with an ac curacy 0.03 eV/ directly, therefore, we started the calculations at a low accuracy of 0.5 eV/ and used the results ob t a ined from it in a high accuracy calculation until the r equired accuracy of 0.03 eV/ was reached. In this mechanism, t he cus O on surface begins to leave the surface with H atom still attached to it and move towards the cus Pd atom creating a vacancy. The H atom on cus Pd site starts moving towards the cus OH group at the transition state as demonstrated in Figure 3 4A DFT D3 estimated activation barrier for this pathway is 168.8 kJ/mol H 2
27 (1.75 eV) which even though lesser than that predicted by Bl anco R ey et al is still too high to be practically possible. Another noticeable fact is that the apparent barrier for the reaction is positive which makes the transition state unlikely to occur. Above results forced us to look for other possible pathways for water formation on PdO(101) surface due to surface reduction. 3.3 H a tom D iffusion on PdO (101) T hin F ilm Pathway described by Blanco Rey & coworkers requires very high temperatures to form H 2 O from H2 induced PdO(101) surface reduction whereas TPD experiments of H2 chemisorption and dissociation over PdO(101) thin film have shown evidence s of H 2 O formation and desorption near 350 K 1 4 To probe alternative water formation pathway s we examined several other possibilities including the pathway reported by Blanco Rey and coworkers 13 In our attempts, we tested many distinct possible hopping paths H atom hoppin g (diffusion) along cus Pd row, along cus O row and hopping from cus Pd to adjacent cus O site. We found that H atom could move along the cus Pd row with considerably lower energy barrier as compared to that required to dissociative adsorption of H 2 molecule on PdO(101) while hopping from cus Pd to cus O site could be achieved within a reasonable energy barrier Probing the next stable configuration which can be attained by two H atoms in close proximity led us to configurations with one H atom atop cus O and the bonded to neighboring cus Pd (see configuration C3 in in Figure 3 1 ). The binding energy of this configuration is 115.0 kJ/mol with dative bonding contribution of 93.3 kJ/mol. We notice that binding energy of this configuration is lower than the C2 configuration (~132.0) however the bond lengths of cus PdH and cus OH bond in C3 are 1.54 and 0.98 which are as same as that in configuration C2. Figure 3 1 shows a model represen tation of both C2 and C3 configurations.
28 It is apparent that transition from C2 to C3 can occur by two possible pathways H atom hopping along cus Pd row and H atom hopping along cus O row. We investigated both these pathways. Transition structures for evolution from C2 to C3 are labeled as TS2.1 and TS2.2 in the Figure 3 1 N ote in TS2.1 that the H atom atop cus Pd begins to move towards the neighboring cus Pd, the Pd H bond length el ongates from 1.55 to 1.72 In TS2.2, the H atom bonded to cus O starts to shift toward neighboring cus O atom stretching the Pd O bond from 0.98 to 1.26 . DFT D3 predicts activation barriers for H atom hopping along cus Pd and hopping along cus O row are 26.0 and 132.4 kJ/mol, respectively. The reason behind the high hopping barrier along cus O row is strong electronic attraction between H atom and surface O atom which makes leaping along the cus O row impractical at lower temperatures. So we ca n omit the possibility of H atom hopping along cus O row to progress along C2 to C3. Using the vibrational frequencies from Table 3 1 for the initial state (C2) and transition states (TS2.1 and TS2.2), we obtain the ZPC ac tivation barriers of 20.5 and 117.5 kJ/mol H 2 respectively. We also note from the overall energy diagram shown in F igure 3 2 that the apparent barrier for H atom hopping along cus O row (TS2.2) is positive which makes the transition state unstable and practically unattainable. Thus, we conclude that H atom can move along the cus Pd row freely on the PdO(101) thin film surface due to its activation barrier lower than that for dissociation on adsorbed H 2 on PdO(101) surface. We examined the possibility of H atom hopp ing from cus Pd to cus O site and vice versa. However, since OH bond is stronger than Pd H bond, breaking the OH bond will require higher energy than breaking the Pd H bind. With this understanding, we
29 probed only the pathway for H atom hopping from cus Pd to cus O site. The final state ( labeled as C4) in the transition from C3 is shown in Figure 3 1 In this configuration, both H atoms bond to adjacent cus O sites, and the O H bond length is 0.98 . The configuration is estimated to be exothermic by 143.9 kJ/mol H 2 on a 4x1 PdO(101) surface. DFT D3 estimates a slightly higher adsorption energy (147.6 kJ/mol H 2 ) for the same atomic arrangement on a 4x2 surface. This stability is attributed to surface coverage effects. Using one image climbing NEB calculation we linked the C3 and C4 configurations. The Pd H bond elongates from 1.54 to 1.61 at the transition st ate ( labeled as TS3 in Figure 3 1 and in subsequent figures) as the H atom shift s toward neighboring cus O site. The energy barrier for t his pathway is estimated to be 88.5 kJ/mol H 2 After applying the zero point correction based upon vibrational frequencies of initial state (C3) and transition state (TS3) which are listed in Table 3 1 in corresponding columns we calculated the ZPC activation barrier for this path way to be equal to 82.4 kJ/mol H 2 Later in this chapter, we shall show that the energy required for H atom to hop from cus Pd site to cus O site is less than that to form water due to PdO(101) surface reduction. 3. 4 Formation of H 2 O on PdO(101) T hin F ilm S urface at 0.08 ML H 2 C overage To start with, we linked configurations C4 and C5 using 5 image climbing NEB calculation at a low accuracy of 0.5 eV / . The result obtained from the converged calculation was used in NEB calculation at higher accuracy until the required accuracy of 0.03 eV/ was achieved. The pathway shows the in presence of neighboring surface hydroxyl group, the cus OH on PdO(101) s urface begins to leave the surface. At the transition state, t he H atom on this cus OH rests away from the PdO (101) surface as well as the neighboring surface cus OH group while the H atom on the surface cus OH
30 group orients toward the cus OH. As the reac tion proceeds, H atom on surface cus O cus OH to form water. The activation b a rrier for the reaction pathway is found to be 88.0 kJ/mol H 2 An atomic model representation of transition state is show n in Figure 3 1 labeled as TS4 W e also notice that the pathway has a local m inim a after the tran sition state global maximum. We investigated the possibility of a stable intermediate state by relaxing t he local minima structure ( labeled as C4i in F i g ure 3 1 and onwards). The cus OH grou p relaxes on the cus Pd atom slightly oriented toward the surface cus OH group. The cus Pd atom shifts toward the adjacent vacancy created at cus O site. The binding energy of this conf ig uration is estimated to be 119.4 kJ/mol with reference as isolated H 2 and bare PdO(101) surface We ran t wo separate 3 image climbing NEB calculations linking C4 and C4i and C4i and C5. The activation barrier for C4 C4i pathway is estimated to be 88.2 kJ/mol H 2 with a transition state very similar to configuration TS4 depic ted above. T he activation barrier for C4i C5 pathway is found to be almost negligible 8.0 kJ/mol H 2 T he H atom bonded to the surface cus O begins to transfer to the cus OH atop cus Pd site at its transition state We observe that this mechanism for water formation requires only 88.0 kJ/mol as compared with gigantic 168 .8 kJ/mol barrier required for Blanco Rey mechanism. The mechanism also points to ward a possible effect introduced by the presence neighboring cus OH m o i e ty that lowers the barri er significantly. We shall discuss about the effect of presence of cus OH moiety later in this document. The vibr ation frequencies o f C4, C4i, TS4, TS4i and TS4ii are listed in Table 3 1 in corresponding columns. Using the vibrational frequencies obtained by NMC calculations on configuration s C4, C4i, TS4, TS4i and TS4ii, we estimate the activation
31 barrier of 82.8 83.0 kJ/mol H 2 to evolve from C4 to C5 producing water and a cus vacancy on the PdO(101) surface ZPC activation barrier for C4i to C5 step is ca lculated to be negative by 0.5 kJ/mol H 2 implying a n auto activated pathway. ZPC calculations for the C4 C5 pathway also provid e us a more profound view of the final water formation procedure by disproportionation of cus OH moiety The pathway can be viewe d as a two step process cleavage of cus OH moiety from PdO(101) surface in presence of a neighboring cus OH; following by H atom transfer moiety forming H 2 O The presence of local minima followed by a very low activation barrier pathway implies the possibility of a flat potential energy surface for the reaction. Figure 3 3 portrays t he overall zero point corrected potential energ y variations in ev o lution of H 2 into H 2 O. Configuration C6 corresponds to the water desorption from the PdO(101) surface leaving a vacancy behind at the cus O site. The activation barrier for C5 C6 evolution is estimated to be 85.1 kJ/mol H 2 which the difference between the ZPC energies of C5 and C6 configurations. It must be noted here that the value 85.1 kJ/mol H 2 is neither derived through NEB calculation nor it is the desorption energy of water on PdO(101) surface. 2 to keep a c onstant reference in the whole oxidati on cycle depicted in Figure 3 3 Although o ne must notice that the desorption barrier for water from PdO(101) surface with adjacent cus O vacancy is very close to the energy required to reduce the PdO(101) surface form ing water. This result may be significant because it indicates the possibility that experimentally, the water formed on the PdO(101) surface may desorb as soon as it forms. We can also infer from the diagram H 2 dissociation and diffusion of H atom along c us Pd rows can occur at very low temperature owing to low activation barrier,
32 nevertheless comparatively higher temperature will be required to facilitate water formation. These results do match well with the experimental results obtained by Hakanoglu et al 14 and Martin et al. 40 3.5 Coverage E ffect of H 2 in F ormation of H 2 O 3.5.1 Formation of W ater at 0.35 ML H 2 (100% cus ) C overage on PdO (101) S urface Experimental studies of H 2 over PdO (101) surface have shown that most of the H 2 reacts with the surface to produce H 2 O with an yield of ~76% of total product yield, and the fractional yield of H 2 O is almost independent of initial H 2 exposure 14 So, we studied the water formation mechanism at higher H 2 coverage on PdO (101) We popul ated 50% of the adjacent cus sites with H atom on a 4x2 PdO(101) surface which corresponds to 0.35 ML coverage of H 2 The relaxed configuration of dissociatively adsorbed H atom on PdO(101) is shown in Figure 3 4 B under the label IS Since we already know that 4 fold sites on the PdO(101) surface are neutral, and the remaining cus sites on the surface are too far for H atom to have any coverage effect we can safely assume this scenario to be equivalent to 100% cus site coverage by H 2 The structure is estimated to have binding energy of 110.1 kJ /mol H 2 with reference to bare PdO(101) surface and isolated H 2 molecule in gas phase. The lower bindin g energy of the structure compa red to the similar low coverage configuration is believed to be due to electronic repulsion among H at oms. HRCLS experimental results have also confirmed that binding energy of H 2 on PdO(101) surface decrea ses at the saturation coverage. We looked for all possible pathways for water formation and none but the Blanco 13 converged to yield the MEP. As explained earlier in the document and by Blanco Rey and coworkers 13 the cus OH moiety on PdO(101) moves out of the surface and move toward the adjacent cus P d atom with adsorbed H atom and the cus
33 Pd H bond stretches as the H atom begins to move toward OH group at the transition state. The activation barrier for MEP is estimated to be 136.6 kJ/mol H 2 at H 2 coverage equivalent to 0.35 ML which a lot lower than the 168.8 kJ/mol H 2 barrier calculated at a H 2 coverage of 0.04 ML on PdO (101) surface (see graphs in Figure 3 4 ) The presence of adjacent cus OH moieties promotes the cleavage of cus OH bonds with the surface hence lowering the e nergy barrier for the water formation. We reckon that even though interaction between H atom and cus O atom weakens bonds between cus OH moieties and surface leading to low adsorption energy of the initial configuration (shown under label IS in Figure 3 4 B ) steric hindrance and electronic repulsion among adsorbates at 100% cus site coverage by H atoms restrict the water f ormation via low barrier pathway i.e. cus OH disproportionation pathway. 3.5.2 Formation of W ater at 0.26 ML H 2 (75% cus ) C overage on PdO (101) S urface Our calculations show that only high energy pathway facilitates the formation of H 2 O at the saturation coverage of H 2 on PdO(101) surface. Since the energy barrier for water formation at 100% cus site coverage is very high, it is very much possible that molecular H 2 can desorb from the PdO(101) surface before the water is form on the surface creating few vacant cus sites. With this understanding, we studied the H2O formation reaction process on a surface with 75 % of cus site covered by H atom (0.26 ML coverage on a 4x2 PdO(101) surface). The 2 cus Pd sites are kept vacant as shown in Figure 3 5 (labeled as IS) The adsorption energy of this configuration is calculated to be 119. 7 kJ/mol H 2 The lo w binding energy of H 2 on Pd O(101) surface at high coverage has been confirmed through high resolution core level shift experiments as well. There can be three possible arrangements by which H 2 O can form on this surface These configurations have been labeled as C5 1, C5 2 and C5 3 in Figure 3 5.
34 Case 1 The cus OH 1 and H 2 atom adsorbed at cus O site react to form H 2 O creating an adjacent vacancy on the surface. ( Image C5 1 in Figure 3 5) Case 2 The cus OH 1 and H 3 atom adsorbed at cus O site react to produce adsorbed H 2 O on the PdO(101). ( Image C5 2 in Figure 3 5 ) Case 3 The cus OH 1 and H 4 atom adsorbed at cus Pd site react to produce adsorbed H 2 O species on the PdO(101). ( Image C5 3 in Figure 3 5 ) The energies of these configurations are very close to each other. The most stable configuration is the final state for case 1 with a binding energy of 121.52 kJ/mol H 2 whereas the case 3 has the least binding energy (119.8 kJ/mol H 2 ). Note that the energy of the initial state (dissociatively adsorbed H 2 on PdO(101)) surface) and the final states ( H2O on PdO(101) surface with adjacent vacancy) have their binding energies in very close proximity. We linked each final state with the initial state using 3 sepa rate climbing NEB calculations. Cus OH 1 moiety on PdO(101) splits from the surface and the cus OH* ( = 2,3,4) elongates as the H atom begins to move toward cus OH 1 moiety at the transition state. The reaction pathway from dissociatively adsorbed H 2 to adsorbed H 2 O on PdO(101) is shown in Figure 3 5 for all three possible pathways at high coverage of H 2 DFT D3 estimates energy barriers for case 1, case 2 and case 3 to be 91.4, 96.4 and 91.2 kJ respectively. As observed in case 3 the barrier for H atom to hop from cus Pd site diagonally and interact with the OH atop cus Pd to form H 2 O becomes negligible. Note that the energy barrier for H 2 O formation at high H 2 coverage is very close to that at low hydrogen coverage. These results are further strengthened by high resolution core level shift calculation s reported by Martin et al. 40
35 3.6 Effect of P resence of Hydroxyl Moiety and H atom D iffusion on PdO(101) S urface on water formation To understand whether or not the presence of OH group affects the energy barrier for H2O formation, we carried out few additional tests with H 2 over PdO(101) surface. In three separat e simu lat ions, we calculated the energy barrier for OH cleavage from PdO(101) surface in presence of H atom adsorb ed on neighboring cus Pd, the solo OH group pop out barrier and in presence of adjacent hydroxyl group We find an energy barrier of 101.7 kJ at the accuracy level 0.3 eV/ for the cus OH to cleave its bonds from the surface when all neighboring cus sites are vacant. The barrier is estimated to be higher by ~15 kJ when a H atom is adsorbed on the ne i ghboring (diagonally adjacent) cus Pd sites at same calculation accuracy of 0.3 eV /. We could not get the calculations converged for th ese test cases at our standard a ccuracy of 0.03 eV/. In our th i r d test, when the H atom is placed on the adjacent cus O atom, the energy barrier for cus OH group to pop out from the surface decreases significantly to 88.2 kJ (calculatio n accuracy 0.03 eV/). Figure 3 6 sh ows the initial, transition and final states for all three cases tested as part of this study It is observed that presence of H atom on cus O atom makes it less bound to the neighboring 4f Pd and cus Pd surface atoms as reflected by the increased bond len gths from ~0.203 and 0.197 to 0.214 and 0.208 , respectively. The cus Pd and cus O bond length increases slightly from 0.208 to 0.212 in presence of a neighboring hydroxyl group (shown in Figure 3 6C) The loss of most favorable bonding environment lowers the energy barrier for oxidation by facilitating the cus OH cleavage from the surface. We also notice that the energy barrier for the pathways first describe d by Blanco Rey and coworkers r educes signi fi cantly from ~187 kJ (168 .8 kJ predicted by
36 our DFT D3 calculations ) at low coverage to 136.5 kJ in presence of neighboring cus OH moieties at high H atom coverage These results present a strong case that H 2 O desorption experimentally observed at low er tem perature is facili t ated by pre sence of the adjacent OH group. Our calculations demonstrate a strong case that the presence of adjacent hydroxyl group on PdO(101) surface accelerates the PdO(101) surface reduction by lowering the barrier for cus OH cleavage H2O formation, though, by disproportionation of cus OH is almost unaffected by presence of more OH groups in the vicinity Note that the barrier estimated for the configuration shown in Figure 3 6A is ~115.5 kJ/mol H 2 which is significantly lower that th e high energy pathway described by Blanco Rey and coworkers. This lowered barrier may be a result of reduced electronic repulsion between the cus OH and the H atom adsorbed on the neighboring cus Pd that is observed in the high energy pathway.
37 Figure 3 1 Model representation of adsorption of H 2 on PdO(101) surface followed by dissociative adsorption and subsequent formation of H 2 O with adjacent cus vacancy on PdO(101) surface. Atomic color code: blue Pd, orange O and white H. Stable configuration s are labeled as C# and transition states as TS # where # = 1,2,3,4,5. C2 TS2.1 TS2.2 C3 C1 TS1 TS3 C4 TS4 TS4i C4i TS4ii C5
38 Figure 3 2 Potential e nergy surface profiling of H 2 activity on PdO(101) surface to produce H 2 O at low H 2 coverage The zero reference is taken as isolated H 2 in gas phase and bare PdO(101) surface. Atomic color codes: blue Pd, orange O and white H. NEB calculations R1, R2 and R3 are carried out in cell size 4x1 R 1 ( C 1 TS 1 C 2 ) : Dissociation of H 2 molecule chemisorbed atop cus Pd atom R 2 ( C 2 TS 2.1 TS 2.2 C 3 ) : Diffusion of H atom. Two possible pathways along cus Pd row and along cus O row. Lowe r barrier for diffusion along cus Pd row R 3 ( C 3 TS 3 C 4 ) : H atom hopping from atop cus Pd to cus O NEB calculations R4, R4i and R4ii have been carried out in cell size 4x2 d ue to coverage effects from cus O vacancy perpendicular to cus Pd row R 4 ( C 4 TS 4 C 5 ) : H 2 O formation which involves simultaneous hopping of cus OH and H onto cus Pd atom. Can be split into two steps R 4i ( C 4 TS 4i C 4i ) : cus OH hopping onto cus Pd. Note that TS 4 and TS 4i are the same states R 4ii ( C 4i TS 4ii C 5 ) : H hopping onto cus Pd and attaching onto already hopped OH from R 4i (C 2 TS*, C 5 ) : High energy Pathway C 1 C 2 TS 1 TS 2.2 C 3 TS 3 C 4 TS 4i C 4i C 5 TS 2.1 TS 4 TS 4ii Isolated H 2 & PdO(101) TS* R 1 R 2 R 3 R 4 R 4i R 4ii 0 50 100 150
39 Table 3 1. Vibrational frequencies (wave numbers) of various stable and transition state configurations. The values are reported in cm 1 and are imaginary frequencies. See text for more detail s. Vibration modes (cm 1 ) C 1 TS 1 C 2 TS 2.1 TS 2.2 C 3 TS 3 C 4 (4x1) C 4 (4x2) TS 4 TS 4i C 4i TS 4ii C 5 1 3070 1931 3728 3711 2109 3713 3717 3722 3717 3633 3632 3680 3704 3682 2 1520 1703 2079 1808 1509 2089 1805 3614 3608 3596 3597 3555 1571 3585 3 932 852 846 841 1144 835 832 810 819 824 823 849 1299 1539 4 556 848 772 710 766 732 714 781 792 716 716 821 768 631 5 485 497 736 703 505 715 489 709 718 657 659 648 571 569 6 471 483 505 449 500 485 446 657 667 519 518 554 470 375 7 402 405 472 360 431 451 359 429 423 327 325 175 365 192 8 350 318 353 327 332 353 323 354 361 317 312 155 172 153 9 286 176 332 176 179 325 171 315 315 167 166 151 156 126 10 179 172 180 124 152 175 119 169 170 113 113 96 139 73 11 168 102 154 113 117 122 85 115 120 64 66 61 100 49 12 106 1081 i* 87 427 i* 1331 i* 87 1259* i 82 84 129 i* 129 i* 92 i* 549 i* 131 i*
40 Table 3 2 C omparison of binding energies of various configurations of H 2 on PdO(101) surface at low coverage on 4x1 and 4x2 PdO(101) slabs. The zero reference is taken as isolated H 2 in gas phase and clean PdO(101) surface. Reference for ZPC is unrestricted H 2 and PdO(101) surface with all atoms but cus Pd and cus O atoms involved in H 2 O formation fixed All energies are reported in kJ/mol H 2 State Binding e nergy Activation b arrier ZPC binding energy ZPC activation barrier 4x1 4x2 4x1 4x2 C 1 56.9 39.7 42.8 33.5 TS 1 17.2 9.3 C 2 132.8 132.2 108.5 TS 2.1 104.3 28.5 85.4 23.0 TS 2.2 0.36 132.4 9.1 117.5 C 3 115.0 115.3 91.6 TS 3 26.5 88.5 9.2 82.4 C 4 143.9 147.6 110.5 113.9 TS 4 59.6 88.0 31.1 82.8 TS 4i 59.4 88.2 30.9 83.0 C 4i 119.4 92.0 TS 4ii 111.4 8.0 92.6 0.5 C 5 141.5 114.7 111.9 85.1 C 6 ** 26.8 26.8 The activation barrier is energy difference between C5 and C6 states ** C6 is a bare PdO(101) surface with vacancy at cus O site.
41 Figure 3 3 ZPC p otential energy evolution of H 2 for H 2 O formation on PdO(101) surface at low H 2 coverage. Zero point reference is isolated H 2 molecule and clean PdO(101) surface. Stable structures are shown at bottom (C0 C 6 ) where C0 is the bare PdO(101) surface and C7 is PdO(101) surface with a vacancy at cus O site. Transition structures are shown at the top. Atomic color codes: bl ue Pd, orange O and white H. Isolated H 2 + PdO(101) 0 100 200 100 108.5 114.0 112.0 42.8 91.6 26.8 33.5 20.5 82.4 82.8 85.1
42 Figure 3 4 H 2 O formation on PdO(101) surface via high energy pathway. A ) H 2 O formation at low H 2 coverage, activation barrier E a. =168.8 kJ/mol H 2 The pathway was first discussed by Blanco Rey et al. See text for details ; B ) H 2 O formation at 100% cus site coverage, E a = 136.5 kJ P otential energy profile comparison of H 2 O formation by high energy pathway at low coverage against saturation coverage of H 2 on PdO(101) surface Atomic color codes: blue Pd, orange O and white H. C 2 TS C 5 168.8 136.5 Low Coverage (A) High Coverage (B) 132.2 36.6 141.5 440.4 460.0 303.8 Isolated H 2 & PdO(101) Isolated H 2 & PdO(101) 500 450 400 350 300 0 10 150 100 50 0 50 B A
43 Figure 3 5 H 2 O formation pathways through disproportionation of 2 adjacent cus OH at 0.26 ML of H 2 coverage on PdO(101) surface. Three possible pathways have been shown for H 2 O formation at this coverage estimated barriers of 91.4, 96.4 and 91.2 kJ for case 1, case 2 and case 3, respectively. Atomic color codes: blue Pd, orange O and white H.
44 Figure 3 6 Effect of OH group presence in H 2 O formation A ) OH hopping with neighboring cus Pd H atom, E a = 115.5 kJ/mol H 2 (accuracy 0.3 eV/A); B ) OH hopping on clean surface, Ea = 101.2 kJ (0.3 eV/A); C ) OH hopping with neighboring cus OH, Ea = 88.2 kJ /mol H 2 (accuracy 0.03 eV/A). Atomic color codes: blue Pd, orange O and white H.
45 CHAPTER 4 CONCLUSIONS Using DFT D3 we studied H 2 adsorption process on PdO(101) surface and H 2 reaction with lattice cus O to yield H 2 O. We learn that H 2 adsorb s onto cus Pd sites of PdO(101) surface and readily dissociatively chemisorbs throu gh precursor mediated mechanism with energy barrier of 33.5 kJ/mol H 2 wherein the molecularly adsorbed H 2 s erves as the precursor. Our calculations predict that diffusion barrier for H atom along the cus O row is 117.5 kJ/mol H 2 whereas barrier for H atom diffusion from one cu s Pd site to another is only 20.5 kJ/mol H 2 which is lower than dissociation barrier of molecularly adsorbed H 2 on the surface. This implies that the H atom can diffuse freely along the cus Pd row of the PdO(101) surface but the diffusion barrier along cus O row is high enough to restrict H atom diffusion from one cus O to another The barrier for H atom from cus Pd to cus O is estimated to be 82.4 kJ/mol H 2 which yields the chemisorbed structure of H 2 with binding energy of 114.0 kJ/mol H 2 Observe that e v en though the binding energy of H atom chemisorbed on adjacent cus O is the highest among all other chemisorbed configurations, at very low temperatures, the H 2 will prefer to be dissociatively chemisorbed on cus Pd adjacent cus O being this the stable co nfiguration with low energy barrier of formation. Earlier experimental studies have shown possibility of multiple pathways for water formation. Using DFT D3, we confirm the presence of more than one pathway for water formation on PdO(101) surface leading t o its reduction. Our calculations predict the presence of a low barrier pathway for H 2 O formation. We find that H 2 O can form on the PdO(101) by disproportionation of two adjacent OH group on the surface. The barrier for this reaction is predicted to be 82. 8 kJ/mol H 2 The energy barrier for H 2 O C1 C2 C3 C4 C5 C6 C7 1.12 1.18 1.16 Isolated H 2 + PdO(101) 0.44 0.95 0.28 0.35 0.24 0.85 0.86 0.88
46 formation through this pathway is very close to the barrier for H atom to diffuse from cus Pd to cus O The low energy barrier is attributed to the presence of neighboring hydroxyl group which weakens the bond between cus OH and the surface facilitating the cus OH cleavage from the PdO(101) surface. Existence of this pathway also suggests the possibility tha t as the temperature of the H 2 adsorbed PdO(101) surface is increased, most of the hydrogen on the surface will combine with lattice cus O to produce water. This conclusion matches well with experimental results which suggest that approximately 76% of H 2 i s released from the PdO(101) surface as water. Another high energy pathway describ ed by Blanco Rey and coworkers is also confirmed by our DFT D3 calculation s though the 168.8 kJ/mol H 2 barrier estimated by our calculations is lower than that reported in B lanco Rey to taking dispersion effects into consideration. Our calculations to understand the coverage effects of H 2 on the surface reveal that only the high energy Blanco Rey pathway will be viable at the sa turation coverage, however, the barrier for the pat hway reduces significantly to 136.5 kJ/mol H 2 compared with the low coverage barrier. Once again the neighboring hydroxyl groups lower the barrier for cus OH cleavage from surface but the presence of H ato m on the adjacent cus Pd increases the electronic repulsion between H and OH moieties leading to a high energy barrier for H 2 O formation at saturation coverage. We find that if the H 2 coverage on PdO( 101) is slightly reduced to 0.26 ML, the hydroxyl disproportionation pathway once again becomes viable. Agreei ng to experimental results, our calculations predict that the H 2 O formation barrier is almost independent of the H 2 O coverage, the energy barrier for H 2 O formation at 0.26 ML H 2 coverage lies between 91.4 kJ/mol H 2 to 96.2
47 kJ/mol H 2 (not ZP corrected). These results suggest that at saturation coverage, chemisorbed hydrogen will prefer to desorb from the surface as molecular H 2 creating vacant cus si tes than to oxidize and produce water due t o its high formation barrier Since the Pd H bond weaker than the O H bond, H 2 desorption will be preferred from cus sites.
48 LIST OF REFERENCES 1. C. Stampfl, Catal Today 2005, 105 17 35. 2. R. B. Anderson, K. C. Stein, J. J. Feenan and L. J. E. Hofer, Ind Eng Chem 1961, 53 809 812. 3. A. Antony, A. Asthagiri and J. F. Weaver, Phys Chem Chem Phys 2012, 14 12202 12212. 4. R. Burch and F. J. Urbano, Appl Catal a Gen 1995, 124 121 138. 5. R. Burch, F. J. Urbano and P. K. Loader, Appl Catal a Gen 1995, 123 173 184. 6. C. F. Cullis and B. M. Willatt, J Catal 1983, 83 267 285. 7. J. F. Weaver, S. P. Devarajan and C. Hakanoglu, J Phys Chem C 2009, 113 9773 9782. 8. J. F. Weaver, S. P. Devarajan, C. Hakanoglu and H. H. Kan, Abstr Pap Am Chem S 2009, 237 875 875. 9. J. F. Weaver, C. Hakanoglu, A. Antony and A. Asthagiri, J Am Chem Soc 2011, 133 16196 16200. 10. J. F. Weaver, C. Hakanoglu, J. M. Hawkins and A. Asthagiri, J Chem P hys 2010, 132 11. J. F. Weaver, C. Hakanoglu, J. A. Hinojosa, A. Antony, J. M. Hawkins and A. Asthagiri, Abstr Pap Am Chem S 2011, 241 12. J. F. Weaver, J. A. Hinojosa, C. Hakanoglu, A. Antony, J. M. Hawkins and A. Asthagiri, Catal Today 2011, 160 213 227. 13. M. Blanco Rey, D. J. Wales and S. J. Jenkins, J Phys Chem C 2009, 113 16757 16765. 14. C. Hakanoglu, J. M. Hawkins, A. Asthagiri and J. F. Weaver, J Phys Chem C 2010, 114 11485 11497. 15. M. Knapp, D. Crihan, A. P. Seitsonen, E. Lundgren, A. Resta, J. N. Andersen and H. Over, J Phys Chem C 2007, 111 5363 5373. 16. Q. Sun, K. Reuter and M. Scheffler, Phys Rev B 2004, 70 17. J. H. Wang, C. Y. Fan, Q. Sun, K. Reuter, K. Jacobi, M. Scheffler and G. Ertl, Angew Chem Int Edit 2003, 42 2151 2154.
49 18. K. Jacobi, Y. Wang and G. Ertl, J Phys Chem B 2006, 110 6115 6122. 19. J. A. Hinojosa, A. Antony, C. Hakanoglu, A. Asthagiri and J. F. Weaver, J Phys Chem C 2012, 116 3007 3016. 20. J. A. Hinojosa, H. H. Kan and J. F. Weaver, J Phys Chem C 2008, 112 8324 8331. 21. H. H. Kan, R. J. Colmyer, A. Asthagiri and J. F. Weaver, J Phys Chem C 2009, 113 1495 1506. 22. H. H. Kan and J. F. Weaver, Surf Sci 2008, 602 L53 L57. 23. G. Kresse and J. Furthmuller, Phys Rev B 1996, 54 11169 11186. 2 4. G. Kresse and J. Furthmuller, Comp Mater Sci 1996, 6 15 50. 25. G. Kresse and J. Hafner, Phys Rev B 1993, 47 558 561. 26. G. Kresse and J. Hafner, Phys Rev B 1994, 49 14251 14269. 27. J. P. Perdew, K. Burke and M. Ernzerhof, Phys Rev Lett 1996, 7 7 3865 3868. 28. P. E. Blochl, Phys Rev B 1994, 50 17953 17979. 29. G. Kresse and D. Joubert, Phys Rev B 1999, 59 1758 1775. 30. S. Grimme, Wires Comput Mol Sci 2011, 1 211 228. 31. S. Grimme, J. Antony, S. Ehrlich and H. Krieg, J Chem Phys 2010, 132 32. D. Sheppard, R. Terrell and G. Henkelman, J Chem Phys 2008, 128 33. M. Methfessel and A. T. Paxton, Phys Rev B 1989, 40 3616 3621. 34. C. Hakanoglu, J. A. Hinojosa and J. F. Weaver, J Phys Chem C 2011, 115 11575 11585. 35. C. Hakanoglu and J F. Weaver, Surf Sci 2013, 611 40 48. 36. J. A. Hinojosa and J. F. Weaver, Surf Sci 2011, 605 1797 1806. 37. R. A. Olsen, G. J. Kroes, G. Henkelman, A. Arnaldsson and H. Jonsson, J Chem Phys 2004, 121 9776 9792. 38. G. Henkelman and H. Jonsson, J Ch em Phys 2000, 113 9978 9985. 39. G. Henkelman, B. P. Uberuaga and H. Jonsson, J Chem Phys 2000, 113 9901 9904.
50 40. N. M. Martin, M. Van den Bossche, H. Grnbeck, C. Hakanoglu, J. Gustafson, S. Blomberg, M. A. Arman, A. Antony, R. Rai, A. Asthagiri, J. F. Weaver and E. Lundgren, J Phys Chem C 2013.
51 BIOGRAPHICAL SKETCH Rahul, born in the state of Uttar Prade sh in India, completed bachelor degree in chemical engineering from Sardar Vallabhbhai National Institute of Technology, Surat. After gaining experience in information technology in banking and finance domain, he from UF in summer 2013 majoring in chemical engineering with thesis.