This item has the following downloads:
1 CATALYST DEVELOPMENT ON NANOPARTICLE OXIDE SUPPORTS FOR THE WATER SPLITTING AND OXIDATIVE COUPLING OF 4 METHYLPYRIDINE REACTIONS By JUSTIN JEFFREY DODSON 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 2013
2 2013 Justin Jeffrey Dodson
3 To my mom and dad
4 ACKNOWLEDGEMENTS I have been fortuna te to be surrounded by such a wonderful support system to be part of the graduate program at the University of Florida and during my time here. I thank Dr. Helen Hagelin Weaver for allowing me to pursue my PhD after completing my lways been available and made time to guide and help my researc h and has been extremely patient in all aspects. Lab members Luke Neal, Trent Elkins, and Samantha Roberts assisted me with experiments over the years. I and advice early on with the oxidative coupling of 4 methylpyridine experiments. The Particle Engineering Research Center and Major Analytical Instrumentation Center at the University of Florida and their staff provided insight assistance in characterizin g my catalysts. I am grateful for the support of Erdem Uguz and Sam Gause for listening to all the difficulties that arose. Finally but not least, I thank my parents Jeffrey and Sherry Dodson for their constant encouragement and unwavering support in any a nd all situations that have occurred over the years.
5 T A B L E O F C O N T E N T S Page ACKNOWLEDGEMENTS ................................ ................................ ............................... 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 11 ABSTRAC T ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 14 1.1 Overview ................................ ................................ ................................ ........... 14 1.1.1 Catalyst Development ................................ ................................ ............ 14 1.1.2 Nanoparticle Oxide Supports ................................ ................................ 14 1.2 Water Splitting Reaction ................................ ................................ ................... 15 1.2.1 Background ................................ ................................ ........................... 15 1.2.2 Reactor System ................................ ................................ ..................... 19 1.2.3 Objective ................................ ................................ ................................ 21 1.3 Oxidative Coupling of 4 Methyl pyridine ................................ ............................. 21 1.3.1 Background ................................ ................................ ........................... 21 1.3.2 Reactor System ................................ ................................ ..................... 23 1.3.3 Objective ................................ ................................ ................................ 23 1.4 Catalyst Preparation Techniques ................................ ................................ ...... 23 1.5 Catalyst Characterization Techniques ................................ .............................. 24 1.5.1 Brunauer Emmett Teller (BET) Surface Area Measurements ............... 24 1.5.2 Chemisorption Measurements ................................ ............................... 25 1.5.3 X ray Photoelectron Spectroscopy (XPS) ................................ .............. 25 1.5.4 X ray Diffraction (XRD) ................................ ................................ .......... 26 2 TWO STEP THERMOCHEMICAL WATER SPLITTING F OR IRON OXIDE SUPPORTED ON NANOPARTICLE ZIRCONIA AND YTTRIA STABILIZED ZIRCONIA ................................ ................................ ................................ ............... 33 2.1 Background ................................ ................................ ................................ ....... 33 2.2 Experimental ................................ ................................ ................................ ..... 34 2.2.1 Catalyst Preparation ................................ ................................ .............. 34 2.2.2 Reaction Experiments ................................ ................................ ............ 35 2.2.3 Cat alyst Characterization ................................ ................................ ....... 37 2.3 Results and Discussion ................................ ................................ ..................... 37 2.3.1 Catalytic Activity Measurements ................................ ............................ 38 2.3.2 BET Surface Area Analysis ................................ ................................ ... 39
6 2.3.3 SEM Analysis ................................ ................................ ........................ 40 2.3.4 XRD Analysis ................................ ................................ ......................... 41 2.4 Summary ................................ ................................ ................................ .......... 42 3 EFFECT OF TITANIA STRUCTURE ON PALLADIUM OXIDE CATALYST FOR THE OXIDATIVE COUPLING OF 4 METHYLPYRIDINE ................................ ....... 53 3.1 Background ................................ ................................ ................................ ....... 53 3.2 Experimental ................................ ................................ ................................ ..... 55 3.2.1 Catalyst Preparation ................................ ................................ .............. 55 3.2.2 Reaction Conditions ................................ ................................ ............... 55 3.2.3 Catalyst Characterization ................................ ................................ ....... 56 3.3 Results and Discussion ................................ ................................ ..................... 57 3.3.1 Surface Areas Measurements ................................ ............................... 57 3.3.2 Reaction Results ................................ ................................ .................... 59 3.3.3 XRD Measurements ................................ ................................ .............. 62 3.3.4 XPS Measurements ................................ ................................ ............... 64 3.4 Summary ................................ ................................ ................................ .......... 67 4 THE INFLUENCE OF ZnO, CeO 2 AND ZrO 2 ON NANOPARTICLE OXIDE SUPPORTED PALLADIUM OXIDE CATALYSTS FOR THE OXIDATIVE COUPLING OF 4 METHYLPYRIDINE ................................ ................................ .... 76 4.1 Background ................................ ................................ ................................ ....... 76 4.2 Experimental ................................ ................................ ................................ ..... 78 4.2.1 Catalyst Preparation ................................ ................................ .............. 78 4.2.2 Reaction Co nditions ................................ ................................ ............... 79 4.2.3 Catalyst Characterization ................................ ................................ ....... 80 4.3 Results ................................ ................................ ................................ .............. 81 4. 4 Discussion ................................ ................................ ................................ ........ 81 4.4.1 PdO Supported on Nanoparticle Alumina ................................ .............. 81 220.127.116.11 Precalcination of n Al 2 O 3 (+) ................................ ..................... 81 18.104.22.168 ZnO addition ................................ ................................ ............. 83 22.214.171.124 CeO 2 addition ................................ ................................ ........... 84 126.96.36.199 ZrO 2 addition ................................ ................................ ............. 84 188.8.131.52 Modified preparation method ................................ .................... 85 4.4.2 PdO Supported on Nanoparticle Magnesia ................................ ........... 85 184.108.40.206 Precalcination of n MgO ................................ ........................... 86 220.127.116.11 ZnO addition ................................ ................................ ............. 87 18.104.22.168 CeO 2 addition ................................ ................................ ........... 88 22.214.171.124 ZrO 2 addition ................................ ................................ ............. 89 4.4.3 PdO Supported on Nanoparticle Titania ................................ ................ 90 126.96.36.199 Precalcination of n TiO 2 ................................ ............................ 90 188.8.131.52 ZnO addition ................................ ................................ ............. 91 184.108.40.206 CeO 2 addition ................................ ................................ ........... 92 4.4.3. 4 ZrO 2 addition ................................ ................................ ............. 9 2 4.5 Summary ................................ ................................ ................................ .......... 94
7 5 REGENERATION OF PALLADIUM OXIDE SUPPORTED NANOPARTICLE ALUMINA FOR THE OXIDATIVE COUPLIN G OF 4 METHYLPYRIDINE ............ 106 5.1 Background ................................ ................................ ................................ ..... 106 5.2 Experimental ................................ ................................ ................................ ... 107 5.2.1 Catalyst Preparation ................................ ................................ ............ 107 5.2.2 Reaction Conditions ................................ ................................ ............. 108 5.2.3 Regeneration Conditions ................................ ................................ ..... 108 5.2.4 Catalyst Characterization ................................ ................................ ..... 109 5.3 Results and Discussion ................................ ................................ ................... 110 5.3.1 Reaction Data ................................ ................................ ...................... 111 5.3.2 BET Surface Area Measurements ................................ ....................... 111 5.3.3 DRIFTS Investigation ................................ ................................ ........... 112 5.3.4 CO Chemisorption Measurements ................................ ....................... 114 5.3.5 XRD Analysis ................................ ................................ ....................... 115 5.3.6 XPS Investigation ................................ ................................ ................ 116 5.4 Summary ................................ ................................ ................................ ........ 117 6 THE INFLUENCE OF PALLADIUM OXIDE LOADING FOR THE OXIDATIVE COUPLING OF 4 METHYLPYRIDINE ................................ ................................ .. 129 6.1 Background ................................ ................................ ................................ ..... 129 6.2 Experimental ................................ ................................ ................................ ... 130 6.2.1 Catalyst Preparation ................................ ................................ ............ 130 6.2.2 Reaction Conditions ................................ ................................ ............. 131 6.2.3 Catalyst Characterization ................................ ................................ ..... 131 6.3 Results and Discussion ................................ ................................ ................... 132 6.3.1 PdO loading on Nanoparticle Alumina ................................ ................. 132 6.3.2 ZrO 2 Loading on PdO Supported on Nanoparticle Alumina ................. 133 6.3.3 PdO Loading on Titania ................................ ................................ ....... 134 6.4 Su mmary ................................ ................................ ................................ ........ 134 7 CONCLUSION ................................ ................................ ................................ ...... 137 LIST OF REFERENCES ................................ ................................ ............................. 142 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 148
8 LIST OF TABLES Table Page 2 1 H 2 production per cycle for the prepared water splitting catalysts. ..................... 43 2 2 BET surface area of prepared water splitting catalysts. ................................ ..... 44 3 1 Catalyst surface areas, CO adsorption, palladium dispersions, yield, and turnover numbers of PdO/TiO 2 support catalysts. ................................ .............. 68 4 1 Catalys t support properties. ................................ ................................ ................ 97 4 2 Catalytic activities of PdO/MeO x /support catalysts. ................................ ............ 98 4 3 Catalyst surface areas, CO adsorptio n, palladium dispersions and turnover numbers of PdO/MeO x /support catalysts. ................................ ........................... 99 5 1 Fresh and regenerated PdO/n Al 2 O 3 (+) catalyst yields. ................................ .. 119 6 1 CO adsorption, palladium dispersions, yield, and turnover numbers of prepared loading series catalysts. ................................ ................................ .... 136
9 LIST OF FIGURES Figure Page 1 1 Water splitting process flow diagram ................................ ................................ .. 28 1 2 Palladium catalyzed oxidative coupling of 4 methylpyridine. .............................. 29 1 3 Oxidative coupling of 4 methylpryidine batch reactor. ................................ ........ 30 1 4 X ray photoelectron spectroscopy. ................................ ................................ ..... 31 1 5 X ray diffraction. ................................ ................................ ................................ 32 2 1 Activation step for the two step water splitting reaction ................................ ...... 45 2 2 Water splitting step for the two step water splitti ng reaction .............................. 46 2 3 H 2 production per cycle for the water splitting step. ................................ ............ 47 2 4 Images of the water splitting catalysts befo re and after reaction cycles ............. 48 2 5 SEM images of the 10FeZrO 2 catalyst ................................ ............................... 49 2 6 SEM images of the 10Fe8YSZ catalyst ................................ .............................. 50 2 7 XRD data for ZrO 2 support water splitting catalysts ................................ ........... 51 2 8 XRD data for YSZ support water splitting catalysts ................................ ............ 52 3 1 Activity data for PdO catalysts supported on n TiO2 pretreated at different temperatures, r TiO2, and p TiO2. ................................ ................................ ..... 69 3 2 X ray diffractio n (XRD) data TiO 2 supports and PdO/TiO 2 catalysts. .................. 70 3 3 X ray diffreaction (XRD) narrow scans for PdO/TiO 2 catalysts. .......................... 71 3 4 X ray photoelectron spectroscopy (XPS) survey scan spectra for PdO/TiO 2 catalysts. ................................ ................................ ................................ ............ 72 3 5 X ray photoelectron spectroscopy (XPS) Ti 2p spectra for PdO/TiO 2 catalysts. ................................ ................................ ................................ ............ 73 3 6 X ray photoelectron spectroscopy (XPS) Pd 3d spectra for PdO/TiO2 catalysts. ................................ ................................ ................................ ............ 74 3 7 X ray photoelectron spectroscopy (XPS) O 1s spectra f or PdO/TiO 2 catalysts. ................................ ................................ ................................ ............ 75
10 4 1 Oxidative coupling of 4 methylpyridine over PdO/MeO x /nano oxide support catalysts ................................ ................................ ................................ ........... 100 4 2 Acti vity data from prepared PdO/MeO x /nano oxide support catalysts. ............. 101 4 3 Thermogravimetric analysis (TGA) data obtained from n TiO 2 ,n Al 2 O 3 ( +) and n MgO supports. ................................ ................................ ........................ 102 4 4 X ray diffraction (XRD) data obtained from PdO/ZnO/n Al 2 O 3 (+) catalysts. ..... 103 4 5 X ray diffraction (XRD) data obtain ed from PdO/ZnO/n MgO catal ysts ............ 104 4 6 X ray diffraction (XRD) data obtained from PdO/ZrO 2 /n TiO 2 catalysts ............ 105 5 1 DRIFTS data obtained from the PdO/ n Al2O3 (+) regenerated catalysts. ........ 120 5 2 XRD data obtained from the PdO/n Al 2 O 3 (+) regenerated catalysts. ............... 121 5 3 XRD nar row scan data obtained from the PdO/n Al 2 O 3 (+) regenerated catalysts. ................................ ................................ ................................ .......... 122 5 4 XPS survey spectra obtained from the PdO/n Al 2 O 3 (+) static air regenerated catalysts. ................................ ................................ ................................ .......... 123 5 5 XPS survey spectra obtained from the PdO/n Al 2 O 3 (+) He regenerated catalysts. ................................ ................................ ................................ .......... 124 5 6 XPS survey spectra obtained from the PdO/n Al 2 O 3 (+) O 2 regenerated catalysts. ................................ ................................ ................................ .......... 125 5 7 XPS Pd 3d spectra obtained from the PdO/n Al2O3 (+) static air regenerated catalysts. ................................ ................................ ................................ .......... 126 5 8 XPS Pd 3d spectra obtained from the PdO/n Al2O3 (+) He regenerated catalysts. ................................ ................................ ................................ .......... 127 5 9 XPS Pd 3d spectra obtained from the PdO/n Al2O3 (+) O 2 regenerated catalysts. ................................ ................................ ................................ .......... 128
11 LIST OF ABBREVIATIONS BET Brunauer Emmett Teller DRIFTS Diffusion reflectance i nfrared s pectroscopy EDX Energy dispersive x ray spectroscopy FWHM Full width at half maximum MCT Mercury cadmium telluride MFC Mass flow controller OD Outside d iameter OPEC Organization of the petroleum exporting countries QGA Quantitative gas analysis SA Surface area SCCM Standard cubic centimeter per minute SDTA Scanning differential thermal analysis SEM Scanning electron microscopy TCD Thermal conductivity det ector TEM Transmission electron microscopy TGA Thermogravimetric analysis TON Turnover number UHV Ultra high vacuum XPS X ray photoelectron spectroscopy XRD X ray diffraction YSZ Yttria stabi lized zirconia
12 Abstract of Dissertation Presented to the Gradua te School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CATALYST DEVELOPMENT ON NANOPARTICLE OXIDE SUPPORTS FOR THE WATER SPLITTING AND OXIDATIVE COUPLING OF 4 METHYLPYRIDIN E REACTIONS By Justin Jeffrey Dodson August 2013 Chair: Helena E. Hagelin Weaver Major: Chemical Engineering Catalyst development is a formulation balance for high activity, selectivity, and stability. The three essential components that comprise hetero geneous catalyst are (1) the active component, (2) the support, and (3) additives or promoters. The two reactions, thermochemical water splitting and oxidative coupling of 4 methylpyridine, were studied using nanoparticle oxide supports. The water splittin g reaction requires a temperature of at least 1 500 C to thermally activate an iron oxide catalyst. The extremely high temperature sinters the catalyst significantly decreasing the surface area. Z irconia and yttr i a stabilized zirconia are considered high temperature stable materials. In order t o improve the stability of iron oxide it was deposited on zirconia and yttria stabilized zirconia The se iron oxide catalysts were subjected to cyclic reaction experiments which consisted of a thermal heating step t o create an oxygen deficien t iron oxide followed by feeding water vapor to generate hydrogen. The n ZrO 2 and YSZ supports stabilize the iron oxide on the providing a consistent hydrogen yield over mu ltiple cycles.
13 Palladium oxide deposited on nanoparticle alumina (n Al 2 O 3 (+)) and titania supports for the oxidative coupling of 4 methylpyridine has yielded highly active catalysts A series of calcination temperatures were investigated for the nano parti cle titania (n TiO 2 ) demonstrating the titania phase is important with the the anatase phase having the best yield The optimum yield of 5wt% Pd /n TiO 2 with a support pretreatment at 450 C was 2.2 g/g catalyst. A series of additive s ceria, zinc oxide, and zirconia were investigated in an attempt to improve the yield. The product yields were dependent on the additive, support, and preparation method showing the delicate balance with the synthesis process and in the formulation. A 5wt% Pd, 5wt% Zr, 90wt% n A l 2 O 3 (+ ) catalyst consistently produced a yield of 3.4 g/g catalyst which is the best catalysts reported to date. Regeneration studies were performed to increase the catalyst life time and increase the yield per catalyst. These catalysts were subjected to a series of temperature treatments under air, helium, and oxygen gas. X ray photoelectron spectroscopy (XPS) and X ray diffraction (XRD) showed a reduction from PdO to palladium metal while diffuse reflectance infrared spectroscopy (DRIFTS) showed carbon d eposits on Static air and oxygen with heat re stored most of the catalyst activity but the heat treatments caused sintering preventing complete regeneration
14 CHAPTER 1 INTRODUCTION 1.1 Overview 1.1.1 Catalyst Development Catalyst design is a delicat e formulation balance that evolves slowly over time to satisfy a process need whether it is an increase in activity, an improvement in selectivity, and/ or a decrease in deactivation. Heterogeneous catalyst formulation is a compromise between three componen ts : high activity i.e. good chemical activity and high specific active surfaces, material characteristics i.e. the strength, size, and shape of particles, and long life i.e. stability by being resistant to sintering, fouling, and poisoning . The process begins with selecting appropriate catalyst candidates to screen for those components Each heterogeneous catalyst candidate typically consists of three components : a support an active metal, and an additive or promoter The support provides a high surfac e area and stability by functioning as a stable surface to disperse the active met al which in turn contributes to reducing sintering. A dditive s exist in small amounts to promote a desired activity, selectivity, and/or stabili ty Additives can promote the s upport structurally or the active metal through changing the electronic environment, morphology, or poising undesirable sites . In this research, catalysts are developed by dispersing an active metal onto nanoparticle oxide supports and subsequently imp roving the formulation through support improvements and the promotion via additives. 1.1.2 Nanoparticle Oxide Supports Nanoparticles are particles with diameters less than 100 nm. A large advantage of the nanoparticles is the very high surface area and large sur face to volume ratio.
15 Consequently, the smaller size leads to higher number of corner and edge sites with low coordination sites . These low coordination sites suggest the active metal interact more strongly with the nanoparticle oxide compared to tradi tional porous oxide catalyst supports. The stronger interactions may lead to higher dispersions and/or alter the electronic environment of the active metal particles. 1.2 Water Splitting Reaction 1.2.1 Background Hydrogen is an important chemical feed stock with a reported 20 billion kilograms generated in the United States in 2008 mostly for ammonia production in fertilizers and fossil fuel productions [3 5]. The hydrogen manufacturing rate is growing at roughly 10% annually [3,5] with demand expect ed to grow from increases in food production and energy requirements to support an ever growing population. Concerns exist over energy supply and the environmental impact from fossil fuels, thus, creating a shift towards cleaner, more sustainable energy sources. Hydrogen provides an en vironmentally attractive energy alternative that may be used in the transportation sector and in power generation to replace fossil fuels [3,6 16]. Current methods to generate hydrogen produce greenhouse gases with reformation of methane or natural gas and electrolysis of water being the two more common methods. Reformation separates hydrogen from water and a carbon species, typically natural gas, under pressure and heat. Steam methane reforming is one such example and has a relatively high e fficiency of approximately 75% in comparison to other methods . Although economically preferred because of the large H 2 production, CO and CO 2 production from the associated water gas shift reaction is detrimental [3,4,6,17,18]. Water electrolysis decom poses water into hydrogen and oxygen with
16 electricity. Since electricity is mostly generated from fossil fuels, greenhouse gas emissions indirectly remain a factor in this method. However, when strictly considering the reaction, electrolysis is more energy intensive and costs three to ten times more 25% as a consequence of the limitation converting heat to electricity [6,7,13,19 21]. Thermochemical water splitting is viewed as a competitor to electrolysis . These thermochemical routes are of particular interest when heat is supplied via a renewable energy source such as solar radiation to dissociate water [8,22]. Additionally, this is an environmentally friendly reaction where water, hyd rogen, and oxygen are the only chemical components involved. However, the direct thermal dissociation of water is an extremely endothermic and thermodynamically unfavorable reaction (Equation 1 1). (1 1) Water begins dissociating around 1,100 C [ 2 3], hydrogen production becomes significant above 2,200 C [13,22 ,24 26 ] and the Gibb s free energy becomes zero at 4,000 C [14,21, 27 ]. Maintaining such high operating temperatures is difficult, expensive, and has low efficiencies cause d by heat losses. The Organization of Petroleum Exporting Countries (OPEC) oil embargos, specifically 1973 oil crisis and 1979 energy crisis, began a remarkable research push into thermochemical water dissociation in the late 1970s and early 1980s [3,13]. The research focused on breaking the reaction into multiple steps to lower the operating temperature but this introduc ed contaminants and required additional separation steps [3,13]. Early multi step cycle research employed nuclear energy for the primary heat source and limiting the operating temperature to 900C [ 28 ]. In order to meet the
17 temperature requirements, cycles of three steps or more are necessary. Several cycles were proposed such as UT 3, Iodine Sulfur (I S), and Mark Process but are 35 50% efficient plus contain hazardous and corrosive compounds [6,7,13,2 9 31]. These processes demonstrate d generating hydrogen thermochemically was feasible. Another proposed heat source was solar energy for metal oxide cycle s [4,12]. Advances in solar technol ogy with heliostats available to direct and concentrate solar radiation to 1,500 C inside a solar receiver [ 11,20,32 ] allow ed this route to be commercially viable. Although more than 280 of these metal oxide cycles have been proposed, extensive research o n the technical feasibility and potential efficiency has been performed only on a few cycles such as Fe 3 O 4 /FeO and ZnO/Zn [13,21] with the majority done via thermodynamic computer simulations. Two and three step metal oxide cycles are predominately attrac tive for simplicity and efficiency . In the two step cycle, the initial activation reaction (Equation 1 2, where M is the metal) reduces the metal oxide thermally. (1 2) This highly endothermic reaction releases oxygen gas and leaves a reduced metal oxide for the water splitting reaction step. The water splitting reaction (Equation 1 3) reoxidizes the metal oxide and produces hydrogen from steam. (1 3) The water dissociation reaction is slightly exothermic and occurs at or below 1,000C. However, temperature still remains a n issue with materials. The activation reaction, also, require s a quench such as inert gas [33 ] to prevent oxygen recombining with the reduced metal oxide forming a thin passiva ting oxide layer .
18 Two and three step metal oxide cycles are classified according to the reduction temperature and volatility. Volatile metal oxides such as Ce 2 O 3 /CeO 2 Mg/MgO, and Zn/ZnO may lose mass and a separation is necessary to remove the metal vapor from the product gas. Ceria possess oxygen mobility and potentially may improve the reaction kinetic but has a reduction temperature above 2,000 C [6,20,30]. Magnesia has excellent thermal shock and a high reduction temperature [25 ]. The z inc oxide cycle is heavily studied because zinc nanoparticles are formed when an inert gas quenches the vapor [6,20,34 ]. Nanoparticle formation is advantageous because the (1) high specific surface area increase reaction kinetics, heat transfer, and mass transfer a nd (2) the large surface to volume ratio favors nearly complete oxidation [10,34 ]. Therefore, nanoparticles may enhance the reaction while increasing the number of operating cycles. The non volatile Fe 3 O 4 /FeO is another extensively studied redox pair becau se of the variety of oxide states produced during the activation step [7,19,2 9 ,30,3 5 ]. The exact composition of the Fe 3 O 4 depends on the reaction conditions. The conversion to FeO is 1% from 900 1,000 C, 15% conversion between 1,000 1,200 C, and 98% near 2,100C [7,35 ]. An inert flowing atmosphere prevents an oxide layer formation and further drives the reaction toward FeO by lowering the partial pressure of oxygen [20,30,32]. The higher conversion to FeO may generate more hydrogen, yet sintering may be m ore pronounced. Above 1,400 C, iron oxide begins to melt and solidifies into a non porous, dense, hard mass when quenc hed [20,30 32 ]. In order to perform multiple runs, the catalyst is removed and physically broken apart after each activation step to maxi mi ze hydrogen production [20,30,35 ]. For a continuous operation to occur,
19 sintering needs to be minimized, and the addition of a high temperature stable support improves the catalyst. Zirconia and yttria stabilized zirconia (YSZ) are high temperature suppo rts with good thermal shock that are known to suppress aggregation of the iron oxide particles [9,17] by enhancing chemical, structural, and the rmal stability up to 2,300 C [36 ]. This minimizes sintering and maintains a higher surface area during repeated cycles with good activity. A stable, inert support shows reproducible yields over numerous cycles without processing the cataly st between reaction steps [28 ,30 ,32 ]. The inclusion of a support in the iron oxide cycle significantly alleviates the sintering and allows both reaction steps to occur in a single reactor without the need to remove and break apart the catalyst. 1.2.2 Reactor System A reactor system was designed and constructed for analyzing the catalytic activity. Figure 1 1 shows the process flow diagra m for the reactor system. The main components of the system include mass flow controller s (MFCs), a syringe pump, two furnaces, an ice bath condenser, and a mass spectrometer. The MFCs control the gas flow rate s for calibrating the mass spectrometer, which is used to analyze the gaseous products, and provide an internal standard during r eaction. The syringe pump supplies water during the water splitting reaction step to be converted to steam in the evaporator furnace at 120 C. The second furnace is a high temperature furnace where the two step water splitting reactions occur and generate gaseous products. The mass spectrometer was calibrated for oxygen, hydrogen, nitrogen, helium, argon, and water. Three components, argon, helium, and oxygen, were passed t hrough the system at 47.5 standard cm 3 /min (sccm), 47.5 sccm, and 5.0 sccm, respectively,
20 and the Quantitative Gas Analysis (QGA) software on the mass spectrometer was set for the expected the percentage of the reaction components at 47.5% Ar, 47.5% He, 5% O 2 0% N 2 0% H 2 and 0% H 2 O in the gas flow. A calibration program is generated and saved based on the partial pressure percentages of the actual flow controlled by the MFCs and the expected input percentages. The flow rates we re changed to verify the ac curacy of the calibration. A catalyst is packed in 99.8% alumina OD reactor tube using ceramic wool to hold the catalyst in place. The ends of the high temperature reactor are packed with ceramic wool to minimize the heat loss during operati on. Argon an d helium flow rates we re set at 20 sccm each, the high temperature furnace was set at 120 C, and the inlet and exit lines to the high temperature reactor are covered in heating tape for 8 hours to minimize the water in the system. The high temperature fur nace is programed to ramp the temperature at 10 C/min to 1,500 C, dwell at this temperature for 45 min, ramp down to 1,000 C at 10 C/min, and dwell at 1,000 C for 90 min. The ramp and dwell times we re repeated for each cycle. The QGA program on the ma ss spectrometer wa s started followed by the high temperature furnace heatin g program. The activation step wa s the initial ramp and dwell time, and oxygen wa s the main product measured. Once the oxygen signal on the mass spectrometer return ed to baseline, t he QGA program wa s stopped, and the system wa s prepared for the water splitting step. While the high tempera ture furnace wa s cooling the evaporator furnace wa s set to 120 C, the condenser wa s filled with ice water to condense any unreacted water vapor, a nd a 20 c m 3 syringe wa s filled with deionized water. The QGA program wa s started again and syringe pump was started at
21 a rate of 0.3 mL/min once the high temperature furnace reache d 1,000 C. Once the hydrogen signal wa s seen, the syringe pump was turn ed off to prevent saturating the s ystem with water. The sequence wa s repeated for each cycle and for as many cycles as necessary. After the last cycle, all the heat sources were turned off, and the system was allow ed to return to room temperature before remov ing the spent catalyst. 1.2.3 Objective The objective of this research is to identify candidates for the two step metal oxide cycle to thermally decompose water. Since extremely high temperatures are necessary, thermally stable catalysts at high temperatures tha t significantly lower the activation temperature are desired. The non volatile iron oxide cycle will be studied becau se iron is abundant, affordable material and has many oxidation states. Since iron sinters easily, nanoparticle oxides are used as supports for their high surface area and to stabilize the iron. In this research, iron oxide is deposited on zirconia and YSZ nanoparticle supports to reduce the sintering by enhancing temperature stability and thermal shock resistance leading to retention of high er surface areas and more reaction cycles and improve the reaction yield. 1.3 Oxidative Coupling of 4 Methylpyridine 1.3.1 Background Bipyridines are important as they are excellent chelating agents with transition metals complexes due to their robust redox stabil ity and ease of functionalization [ 3 7]. This is advantageous in numerous applications such as luminescent devices, photonic and optoelectronics, and electrochemistry. Ruthenium bipyridine c omplexes adsorb visible light [3 8, 3 9] displaying potential for arti ficial photosynthesis [ 4 0] and chemiluminescence detection systems [ 4 1]. Iron and cobalt bipyridine complexes
22 catalyze the reduction of CO 2 and O 2 [4 2]. Palladium bipyridines are useful in numerous organic reactions including the carbon monoxide and styre n e copolymerization reaction [43,44 ]. Despite their utility, bipyridines are expensive and demanding to synthesize leading to a relatively high cost of m ore than $7,600 per kilogram [45 ]. Several synthesis routes include the building a second pyridine ring from a substituted pyridine and the catalyzed coupling of a halogenated pyridine [4 3 46 48 ]. These routes require purification steps and disposal of environmentally hazardous materials. An alternative route utilizes low v alent metal catalyst such as Ra ney nickel or palladium on activated carbon in a simple one step reaction [ 43, 4 9 52 ]. Raney nickel is relatively inexpensive but is pyrophoric in nature requiring pretreatment and special handling. On the other hand, the reaction with palladium is environmenta lly friendly contain ing only the bipyridine and water as the byproduct. Despite this being a slow reaction and catalyst activity losses during reaction, improving the palladium on nanoparticle oxide support dimethyl bypryidine more economical. The oxidative coupling of 4 methylpyridine over palladium (Figure 1 2) has shown good activity on nanoparticle oxide catalyst supports and is a suitable reaction for further developing these palladium catalysts. Earlier research showed pa lladium on activa ted carbon has good activity [49 52 ] while palladium on alumina was poor [4 9 ]. The two main reasons are activated carbon has a high surface and tendency for high metal dispersions and the act ive phase is palladium oxide [43 ]. Palladium on alumina tends to be pre reduced to palladium metal, hence being a poor catalyst. However, it has been shown palladium deposited onto nanoparticle alumina is a good catalyst with
23 yields slightly above Pd/C . The nanoparticle alumina impart s a high surfa ce area and the oxidized palladium is on the prepared catalyst. 1.3.2 Reactor System The oxidative coupling reaction utilizes a batch reactor system. The catalyst powder is placed in a round bottom flask with the liquid 4 methylpyridine reactant in a 1:10 mass ratio of catalyst to reactant with a stir bar The flask is connected to a reflux condenser as seen in Figure 1 3. The reflux system is evacuated to introduce an oxygen environment and is lowered into oil bath to react under reflux for 72 at the boiling p oint of 4 methylpyridine (145 C) under agitation. 1.3.3 Objective Early research has shown palladium on activated carbon to have yields upwards of 2.2 g/g catalyst [43 ] and palladium on various metal oxide supports equaling or exceeding that value with nanopart icle alumina and titania being the best at 2.5 g/g catalyst and 2. 6 g/g catalyst, respectively [53 ]. In this research, the nanoparticle alumina and nanoparticle titania catalyst are improved through the deposition of additives to the catalysts and altering the preparation conditions to further improve the product yield. Furthermore, the palladium loading and deactivation and regeneration of the catalyst are studied to lower ing the cost associated with palladium and extend the catalyst lifespan. 1.4 Catalyst Pre paration Techniques Sever al preparation techniques exist such as precipitation, impregnation, incipient we tness impregnation and many variations of these techniques. In this research, precipitation is primarily used to fabricate these catalysts. A brief ov erview of
24 the precipitation method is given below with preparation specifics for each study provided in their respective chapters. Precipitation imparts a uniform coverage of the metal precursor on the support Metal salts, nitrates in these studies as to avoid poison such as sulfates and chlorides, act as the active phase and/or promoter. The dissolved nitrate salts are added to an aqueous dispersion of the nanoparticle oxide support. The aqueous suspension is titrated with sodium hydroxide to a catalyst s pecific pH and aged ove rnight forming metal hydroxide(s) on the support. The catalyst solution is filtered and rinsed with deionized water to remove the remaining salts. After a second filtration, the catalyst is dried at 105 C in the oven overnight befor e calcining. 1.5 Catalyst Characterization Techniques Numerous characterization techniques were used to characterize the catalysts in this research. Detailed descriptions and the purpose can be found in many of t extbooks and handbooks Masel , Cullity [55 ], and the Handbook of X ra y Photoelectron Spectroscopy [56 ] to name a few. An overview of the more common techniques found in this work is given below. 1.5.1 Brunauer Emmett Teller (BET) Surface Area Measurements The overall surface areas of the supports and cata lysts were determined using BET adsorption i sotherms, described in Masel [54 ]. The measurements were performed on Quantachrome Nova 1200 instrument. Essentially, an evacuated sample cell cont aining the catalyst or support wa s immersed in liquid nitrogen to maintain a constant tempe rature of ~123 K. Nitrogen gas wa s pulsed into the sample cell and adsorbed onto the catalyst or support. An isotherm plot relating cell pressure to the volume of gas pulsed into the cell is generated. The volume of nitrogen neces sary to
25 adsorb one monolaye r of nitrogen onto the surface wa s calculated from the slope and zero intercept of the isotherm plot. The surface area is related to the area of one monolayer through the properties of the nitrogen adsorbing gas. 1.5.2 Chemisorption Me asurements molecules. The palladium metal surface area is commonly measured with carbon monoxide gas. The palladium metal surface area was measured on a ChemBET 3000. Initially, the palla dium oxide catalyst was reduced under mild conditions (170 C and 5% H 2 in He) for 60 minutes then outgassed with N 2 CO was then pulsed and adsorbed onto the surface Pd atoms. A thermal conductivity detector (TCD) determined the saturation amount through comparing the CO pulse gas to the He carrier gas Once Pd surface atoms are completely saturated, the number of Pd atoms on the surfac e is calculated from a known stoichiometry of CO to Pd surface atom. The metal surface area and crystallite size may be ca lculated by estimating the particle shape. 1.5.3 X ray Photoelectron Spectroscopy (XPS) The Handbook of X ra y Photoelectron Spectroscopy [56 ] provides a more detailed overview of XPS. Briefly, XPS operates under ultra high vacuum (UHV), which is less than 10 9 T orr, providing information about the solid surface. An X ray source excites a solid surface (Figure 1 4) ejecting the electrons that can escape within the first 10 nm of the surface of the solid by excitation. XPS measures the kinetic energy of the ejected electrons. The kinetic energy of the ejected electrons equal s energy of the X ray source minus the binding energy and the work function of the spectrometer. The binding energy is the energy necessary to remove an electron from the surface which differs be tween orbitals elements and oxidation states of the same element. The exciting X
26 ray source, typically magnesium or aluminum K is known, and the use of a standard establishes the work function of the spectrometer. Thus, a binding energy is calculate d to obtain inform ation about the elemental composition of the surface and oxidation state of the surface atoms. 1.5.4 X ray Diffraction (XRD) XRD measures the bulk cry stal structure of samples. X rays penetrate several micrometers into the sample providing a pa ttern of diffracted crystalline atoms. Although some waves cancel one another out through destructive interference, the constructive waves (Figure 1 5) are described by the Bragg equation ( E quation 1 4). (1 4) The spacing bet ween diffraction planes, d hkl of the incident beam where n is any integer. If the composition of the sample is known then the crystal phase present can be determined by comparing the diffraction p attern to a known sample, and the average particle diameter, d av may be calculated from the peak widths ( E quation 1 5) (1 5) The Scherrer equation correlates the average particle diameter to the peak width at half maximum (F WHM), the angle of reflection, and wavelength of the incident beam with K being a prefactor determined the instrument and typically taken as unity. More detailed inf ormation is found in Cullity [55 ]. Other limitations besides being a bulk technique, XRD on ly detects crystalline phases i.e. amorphous samples are undetectable and crystalline planes may overlap
27 causing the appearance of one phase. In spite of these limitations, XRD is beneficial especially when used in conjunction with other surface techniques
28 Figure 1 1 Water splitting process flow diagram
29 Figure 1 2 Palladium catalyzed oxidative coupling of 4 methylpyridine.
30 Figure 1 3 Oxidative coupling of 4 methylpry idine batch reactor. Photo courtesy of Justin Dodson.
31 Figure 1 4 X ray photoelectron spectroscopy.
32 Figure 1 5 X ray diffraction.
33 CHAPTER 2 TWO STEP THERMOCHEMICAL WATER SPLITTING FOR IRON OXIDE SUPPORTED ON NANOPARTICLE ZIRCONIA AND YTTRIA STABILIZED ZIRCONIA 2.1 Background Hydrogen is a widely produced and consumed chemical with approximately 20 million metric tons produced annually in the US mostly for commercial usage in hydro treating crude oil, fertiliz ers, and hydrogenation of food [ 3 5 ]. Hydrogen is, also, considered a promising energy carrier especially in the transportation sector with fuel cells. The current p roduction methods occur primarily from steam reforming of natural gas. Although natural ga s is considered a cleaner fuel stock than either petroleum or coal pollutants such as CO, CO 2 and NO x are still emitted. Thermochemical water splitting is an environmentally attractive alternative to natural gas steam reforming if solar energy is used as an energy source T hermochemical water splitting uses heat and water as inputs to produce high purity hydrogen and oxygen streams while eliminating air pollutants It is difficult to obtain an appreciable hydrogen yield though directly dissociating water because extremely high temperatures are necessary with only 10% hydrogen production at 2,200 C [ 13,22,24 26 ] and thermodynamic favorability occurring above < 0 kJ/mol [ 14,21,27 ] Separating the hydrogen and oxygen product stream is also an issue to prevent the explosive oxidative recombination. Early research shows two and three step cycles such as Fe 3 O 4 /FeO and ZnO/Zn are attractive to se parating the products and lowering the temperature [ 13 21 ]. These cycles utilize metal oxides to alternate between an oxygen rich and oxygen deficient metal and the reduced form i s either the metal or an oxygen deficient metal oxide Initially, the highly endothermic activation step uses heat to release oxygen and create an oxygen deficient species for the water
34 splitting step The second step is slightly exothermic where water vapor passes across the sample generating hydrogen and restoring oxygen to the sample The non volatile Fe 3 O 4 /FeO cycle has been studied extensively. Nakamura and Steinfeld have suggested that the abundance of oxidation states make s it a good candidate for water splitting [ 7,11,19,29,30 ]. The oxidation state varies with activation te mperature as higher temperatures convert a larger percentage to FeO and an inert flow can prevent oxygen from easily recombin ing with FeO [ 7,11,19,29,30 ]. These variations cause fluctuations in the amount of FeO present and consequently the hydrogen prod uction volume. Furthermore, iron oxide sinters into a dense, non porous mass at these temperatures [ 20 30 32 ]. Sintering arises from the formation of an ox ygen deficient metal oxide forming on the outer most layers of the surface and causes a decrease in the surface area and deactivates the catalyst. In this study, Fe 2 O 3 was deposited on different ZrO 2 based supports to investigate the effect of the support on the hydrogen production and the catalyst properties specifically sintering of the catalyst Zirc onia is known to have good thermal and shock resistance at high temperature, which minimizes sintering allowing a longer catalyst lifespan and more hydrogen production. The objective of this study was to determine the degree of sintering and what effect si ntering had on the catalytic activity in the water splitting reaction over ten cycles of catalyst activation and oxidation 2.2 Experimental 2.2.1 Catalyst P reparation The catalysts were prepared with commercially available nanoparticles zirconia (ZrO 2 Nanostruct ur ed and Amorphous Materials), 3% yttria stabilized zirconia (3YSZ, Nanostruct ur ed and Amorphous Materials), and 8% yttria stabilized zirconia (8YSZ,
35 Nanostructured and Amorphous Materials). A precipitation method was employed to deposit iron onto the suppor ts. In this method, an aqueous solution of iron nitrate (Fe(NO 3 ) 3 9H 2 O Alfa Aesar) was added to an aqueous dispersion of the nanoparticle oxide support The iron loading was either 10% or 20% by weight on a metal basis. The catalysts are listed according to the iron loading amount and support i.e. 10Fe8YSZ for a 10wt% loading of iron on 8YSZ support. The mixture was titrated with sodium hydroxide ( NaOH, Sigma Aldrich) to a pH ~9.5 forming iron hydroxide(s) on the support. The resulting solution was aged ov ernight at room temperature before checking the pH and adjusting, if necessary. Once the pH is stable, the solution is filter, rinsed in deionized water overnight, and filtered again. The material was then dried at 105 C overnight and calcined in static a ir at 800 C for 4 h. 2.2.2 Reaction E xperiments diameter pellets under two tons of force with a Carver press. The pellet was crushed using a mortar and pestle and sieved for particle si ze Approximately 3 g of catalysts were packed between two ceramic blanket plugs inside a measurements. A flow reactor system was designed for evaluating th e activity of the two step water splitting reaction. AliCat mass flow controllers (MFCs) were used to regulate the the system. Deionized water was feed from a 20 stran d ard cm 3 syringe by a KDS 100 syringe pump vaporized using a MTI tube furnace as an evaporator. The reactor furnace was immediately downstream of the evaporator. A Carbolite STF 16/450 acted
36 as the high temperature reactor furnace for both the activation s tep and water splitting step. Since stainless steel melts at the operating conditions, Swagelok UltraTorr fittings were used to connect the stainless steel tubing to the alumina ceramic reactor tube. A condenser was placed in an ice bath after the reactor to condense the unreacted water before the product stream was analyzed using a Hiden Analytical mass spectrometer with a Quantitative Gas Analysis (QGA) software program. Each reaction experiment consisted of ten cycles which repeated an activation step an d water splitting step. Once the catalyst i s packed into the reactor tube, the reactor temperature i s set to 120 C and 20 sccm Ar flow and 20 sccm He flow degassed the system overnight Ar gon and helium acted as a quench for the catalyst and used as an in ternal reference for the mass spectrometer. The first step is activating the catalyst The high temperature reactor furnace is programmed to ramp the temperature at a rate of 10 C / min to 1,500 C then dwell at this temperature for 45 min. Ice is placed in to the condenser (which remove d any water that is present during the oxidation step) the QGA program is started, as well as the activation temperature program. Once the activation step finishes, the high temperature furnace is cooled to 1,000 C at a rate of 10 C / min. While the reactor is cooling down, the system is being prepared for the water splitting reaction step. The syringe pump is filled with deionized water, the evaporator is set to 120 C and the condenser is replenished with ice. Once t he temperature reaches 1,000 C, deionized water is fed into the system at 0.3 mL/min from the syringe pump and the QGA program is started again for the water splitting step.
37 The activation and water splitting step s conclude the first cycle. After this cycle, the furnace is heated again to 1,500 C following the same procedure as for the activation step. The only difference is the catalyst material is at a lower oxidation state (Fe 3 O 4 rather than Fe 2 O 3 ) after the first cycle. During this step oxygen is evolved and the catalyst is thermal ly re duced which is why this step is referred to as catalyst regeneration or reduction step. During the water splitting step, the catalyst is oxidized and hydrogen is generate d One complete cycle involves a catalyst red uction/regeneration and an oxidation step. Each catalyst is on stream for ten cycles o f activity measurements before characteriz ation 2.2.3 Catalyst C haracterization The as received support surface areas, the prepared catalysts surface areas, initial activated catalyst surface area, and the surface areas of the spent catalysts (after ten cycles) were determined by multipoint Brunauer Emmett Teller (BET) isotherms on a Quantachrome Nova 1200 instrument as described previously [ 57 ]. Scanning electron micrographs ( SEM) were obtained on a FEI XL 40 FEG SEM. The samples were placed on an aluminum mount with double sided conductive carbon tape and coated with a carbon film. ray diffractometer using Bragg Brentano ge = 1.54 ). Selected supports and catalysts powders (before and after reaction) were secured onto a glass slide with double sided sticky tape prior to the measurements. 2.3 Results and Discussion The four catalysts were subjected to activity measurements in the water splitting reaction and a nalyzed with BET, SEM and XRD.
38 2.3.1 Catalytic A ctivity M easurements Each catalyst was cycled ten times which consisted of the thermal activation and water splitting step s The activation produced an oxygen deficient iron oxide while the water dissociation oxidized the iron and generated hydrogen. Three different commercial zirconia based nanoparticle supports were analyzed along with a higher Fe loading for the monoclinic zirconia support. Pure ZrO 2 changes phase from monoclinic to tetragonal at 1,000 C and to cubic at 2,370 C. Yttria stabilize s the different zirconia phases and improves the high temperature stability of ZrO 2 The 3YSZ stabilizes ZrO 2 in the tetragonal phase while 8YSZ gives the cub ic phase. The initial activation cycle (Figure 2 1) contains two peaks because the catalyst is calcined at 800 C in air and thus the catalyst consist of Fe 2 O 3 on the support The first feature is more intense revealing that more oxygen is released compare d to the second feature. The maximum oxygen evolution is observed around 1380 C and is due to reduction of Fe 2 O 3 to Fe 3 O 4 The second feature peak s around 1490 C and is due to reduction of Fe 3 O 4 to FeO. This feature is present on all subsequent cycles. The YSZ support ed catalysts consistently evolve more oxygen which means that more hydrogen is generated in the following step compared to the zirconia support ed catalysts. The higher iron loading produced less oxygen than the other catalyst after the firs t activation as the additional Fe loading is more susceptible to sintering because there are fewer iron support interactions present The water splitting step immediately followed each activation step with Figure 2 2 showing the hydrogen production for the corresponding activation cycles in Figure 2 1. The highest hydrogen production rate occurred during the first cycle. This is expect ed as the starting material was calcined at a lower temperature The change in oxidation
39 states results from the extent of r eduction to FeO followed by the formation of an oxygen passivation on the FeO surface making the complete regeneration to Fe 2 O 3 unlikely. The 10Fe8YSZ catalyst had the highest hydrogen production rates at 0.8 3 cm 3 /g material in each cycle despite a slight decline in subsequent cycles. The higher iron loading catalyst, 20FeZrO 2 declined dramatically after the first cycle and had almost no hydrogen production by the fifth cycle. Since iron sinters easily, the additional loading actually hindered the producti on. The total hydrogen production for each cycle is presented in Table 2 1 and Figure 2 3. The general trend shows a decline in the H 2 productivity until cycle 4 except with 10FeZrO 2 Again this productivity decline indicates sintering. Interestingly, 10Fe ZrO 2 consistently generates the same amount of H 2 The iron oxide stabilizes the ZrO 2 [ 58 ] in a similar fashion to yttria stabilizing the cubic ZrO 2 phase. Despite the constant yield for the 10FeZrO 2 catalyst 10Fe8YSZ produced almost two and half times as much hydrogen as 10FeZrO 2 49.35 cm 3 /g material and 20.71 cm 3 /g material, respectively. A partial ZrO 2 stabilization increased the hydrogen production as 10Fe3YSZ generated 32.64 cm 3 /g material. 2.3.2 BET S urface A rea A nalysis The BET surface area (Table 2 2 ) is presented for the commercial nanoparticle supports, fresh, first activation step and spent catalyst s The surface area of the fresh catalyst s remains relatively unchanged compared to the nanoparticle supports SA. This indicates ZrO 2 has some temperature resistance to the 800 C calcination temperature. However, there was a slight SA decline on the 20FeZrO 2 sample likely caused by sintering from the additional iron after calcination. As the catalysts were calcined at a temperature lower than the reaction t emperature, the SA of the first activation step was measured. There was a drastic decline in the SA compared to the fresh sample. The
40 10Fe8YSZ had a SA of 1.8 m 2 /g, which was the highest SA amongst the four samples. This is contributed to preventing ZrO 2 p hase changes by stabilizing the cubic phase with yttria. It is interesting that the spent 10FeZrO 2 and 10Fe8YSZ have almost no SA change 1.6 m 2 /g each compared to the SA after the first activation step. Despite the significant decline in SA these values r emain about an order of magnitude higher in comparison to literature values [ 32 ] Additionally, the samples remain as powder s (Figure 2 4) unlike the dense, hard material s that require processing between steps. This is somewhat surprising since nanoparticl es are more susceptible to sinter ing at these high temperatures because of their smaller size. 2.3.3 SEM A nalysis SEM images were obtained from 10FeZrO 2 (Figure 2 5 ) and 10Fe8YSZ (Figure 2 5 ) for the fresh catalysts, as well as, after the first activation step, the first water splitting step, and after 10 cycles. E nergy dispersive x ray spectroscopy (ED X ) line scans identified and confirmed the Fe particles which are circled in the figures. Both fresh catalysts show small Fe particles covering the larger support particles. As the reaction cycles progress ed the sharp edges and corners of the support on the fresh samples become round and almost spherical in shape. The rounding of the particles indicates sintering of the support. The sintering is further supported in the growth of the Zr O 2 (after 10 cycle reactions) catalyst. This is also seen for the 10Fe8YSZ catalyst as there is no defined morphology for the spent sample (Figure 2 5 D). The Fe particles are circled for identification purposes. The Fe particles develop into defined shapes rectangular and pyramidal for 10FeZrO 2 and 10Fe8YSZ, respectively, after the first activation step. Interestingly, the Fe particles are arranged in between or partially cove r
41 by the support particles instead of on top. After the first water splitting reaction, the support covers more of the Fe particle and almost completely encases the Fe particle in Figure 2 5 C. Considering the H 2 production for cycle 10 on 10Fe8YSZ catalyst it is surprising Fe is undetectable by ED X and not visible in the SEM images 2.3.4 XRD A nalysis XRD measurements were performed on the commercial supports, fresh catalysts, and spent catalysts to determine the crystal phase and particle size. As with BET surf ace area analysis and SEM images, sintering was confirmed through the sharp en ing of the peaks between fresh and spent catalysts The crystallite size is inversely related to the peak width at half maximum via the Scherrer equation. Therefore, the sharper t he peak translates to a narrower peak width and thus a larger particle size. Since Fe 2 O 3 and support peaks are convoluted, the extent to particle growth cannot be calculated. SEM confirmed the particle growth through sintering. In a dditional to the convolu ted peaks XRD measures the bulk crystal lattice of the sample which causes difficultly in determining the iron oxide phase(s), Fe 2 O 3 or Fe 3 O 4 present on the surface.The bulk Fe 2 O 3 phase is present for both the fresh and spent sample narrow scans on 10FeZ rO 2 and 108YSZ (Figure 2 7 B and Figure 2 8 B, respectively). The Fe 2 O 3 peak at 33.4 is present on both the 10FeZrO 2 fresh and 20FeZrO 2 fresh samples and is undetected on the spent samples for both loading Interestingly, a tetragonal ZrO 2 (t ZrO 2 ) peak dev elops on the spent 20FeZrO 2 This is in agreement with iron partially stabilizes a higher temperature tetragonal ZrO 2 phase suggesting a larger amount of the iron was trapped in the bulk participate in the water splitting reacting causing the lower H 2 yiel d. A monoclinic ZrO 2 (m ZrO 2 ) phase is detected on the 3YSZ support at 31.6 again supporting as well as indicating a partial stabilization
42 of ZrO 2 Once again Fe 2 O 3 and Fe 3 O 4 are difficult to detect as SEM shows iron oxide in small quantities on the sam ple surface. 2.4 Summary Several powder catalysts were identified for the water splitting reaction for hydrogen production. Sintering is a foremost cause of deactivation as shown through the decrease in surface area and particle growth from SEM and XRD. Zir conia displays good thermal stability over the reaction cycles with the surface area remaining at 1.6 m 2 /g after the initial activation step The 10FeZrO 2 catalyst produced a consistent amount of H 2 after the first cycle of almost 2 cm 3 H 2 /g material Howe ver, a higher iron loading generate d half the hydrogen as the iron was trapped in the bulk of the support to stable the tetragonal ZrO 2 phase. The YSZ supports were already stabilized for the higher temperature ZrO 2 phase allowing for greater H 2 production Despite a declining H 2 production between cycles 10Fe8YSZ consistently produced more than double the H 2 per cycle as 10FeZrO 2 with the partially stabilized 3YSZ support having a yield between these two catalysts.
43 Table 2 1 H 2 production per cycle for the prepared water splitting catalysts. Cycle # 10FeZrO 2 H 2 Production [cm 3 /g material ] 20FeZrO 2 H 2 Production [cm 3 /g material ] 10Fe3YSZ H 2 Production [cm 3 /g material ] 10Fe8YSZ H 2 Production [cm 3 /g material ] 1 2.30 2.97 6.86 8.16 2 2.50 1.42 5.34 5.79 3 2.46 1.13 4.21 4.98 4 2.22 0.78 2.08 4.42 5 1.82 0.90 1.72 4.87 6 2.30 0.90 1.28 4.26 7 1.53 0.81 1.63 3.95 8 1.93 0.93 3.30 4.07 9 1.89 0.82 3.33 4.58 10 1.77 0.84 2.90 4.26 Overall 20.71 11.51 32.64 49.35
44 Table 2 2. BET surface area of prepar ed water splitting catalysts. Entry Catalyst Support SA [m 2 /g] Fresh SA [m 2 /g] Activation #1 SA [m 2 /g] Spent SA [m 2 /g] 1 10FeZrO 2 10 10 1.6 1.6 2 20FeZrO 2 10 7.8 1.3 0.7 3 10Fe3YSZ 15 17 1.3 0.5 4 10Fe8YSZ 13 12 1.8 1.6
45 Figure 2 1 Activation ste p for the two step reaction for cycle 1, 2, 5, and 10.
46 Fi gure 2 2. Water splitting step for the two step reaction for cycles 1, 2, 5, 10.
47 Figure 2 3. H 2 production per cycle for the water splitting step.
48 Figure 2 4. Images of the water spl itting catalysts before and after reaction cycles (A) fresh 10FeZrO 2 (B) spent (after 10 cycles) 10FeZrO 2 (C) fresh 10Fe8YSZ, and (D) spent (after 10 cycles) 10Fe8YSZ. Photo courtesy of Justin Dodson. A. fresh 10FeZrO2 B. spent 10FeZrO2 C. fresh 10Fe8YSZ D. spent 10Fe8YSZ
49 Figure 2 5 SEM images of the 10FeZrO 2 catalyst with the Fe x O y particles indicated for (A) fresh, (B) after the first activation step, (C) after the first water splitting step, and (D) spent (after 10 cycles). A. fresh 10FeZrO2 B. activation 1 10FeZrO2 C. water splitting 1 10FeZrO2 D. spent 10FeZrO2
50 Figure 2 6. SEM images of the 10Fe8YSZ catalyst with the FexOy particl es indicated for (A) fresh, (B) after the first activation step, (C) after the first water splitting step, and (D) spent (after 10 cycles). A. fresh 10Fe8YSZ B. activation 1 10Fe8YSZ C. water splitting 1 10Fe8YSZ D. spent 10Fe8YSZ
51 Figure 2 7. XRD data for ZrO 2 supported water splitting catalysts for (A) supports, fresh, and spent Fe 2 O 3 depos ited catalysts and (B) narrow scan regions for fresh and spent Fe 2 O 3 deposited catalysts.
52 Figure 2 8. XRD data for YSZ support water splitting catalysts for (A) supports, fresh, and spent Fe 2 O 3 deposited catalysts and (B) narrow scan regions for fresh and spent Fe 2 O 3 deposited catalysts.
53 CHAPTER 3 EFFECT OF TITANIA STRUCTURE ON PALLADIUM OXIDE CATALYST FOR THE OXIDATIVE COUPLING OF 4 METHYLPYRIDINE 3.1 Background Bipyridines coordinate well to transition metal cations forming complexes with distinct photochemical and catalytic properties [ 40,53,59,60 ]. For example, bi p yridine ruthenium complexes are appealing for applications as organic light emitting diodes [ 40 ] and chemiluminescence detection systems [ 38,39,41 ] and organometallic complexes with bipyridines are very common in homogeneous catalysis [ 61 ]. However, large scale usage is partially limited by the cost associated with bipyridine compounds. For dimethyl bipyridine exceeds $7,600 per kilogram [ 46 ]. Therefore, finding a mor e economical synthesis pathway for bipyridines is desirable. Environmental impacts are, also, an important consideration. Hence, reactions with no solvents and no halogenated compounds are advantageous The oxidative coupling of 4 d imethyl bipyridine over palladium is a simple one step process that uses neither solvents nor halogenated precursors ( Figure 1 2 ) with water and terpyridine as the only by products. Although this is an environmentally friendly process, the reaction ra te is slow, and the catalyst undergoes deactivation which limits the product yield [ 52 ]. Early research for this reaction focused on using palladium on carbon (Pd/C) catalysts with yields for 5 wt% and 10 wt% Pd loadings varying between 1.5 and 2.0 g of p roduct per g catalyst [ 52,62 64 ]. In our previous research it was shown that both 5 wt% palladium (metal basis) precipitated onto pellet titania ( PdO/p TiO 2 ) and nanoparticle alumina ( PdO/n Al 2 O 3 (+) ) yield 2.6 and 2.5 g product/ g catalyst, respectively [ 44 ]. However, the yield for palladium precipitated onto nanoparticle titania ( PdO/n TiO 2 ) was remarkably lower, at only 1.6 g
54 product per g catalyst compared with the PdO/p TiO 2 despite a significantly larger support surface area [4 4 ]. Apart from the TiO 2 surface area, t he main difference between the two supports was the crystal structure, as the n TiO 2 was nearly amorphous, while anatase was identified as the phase of the p TiO 2 [ 4 4 6 5 ]. Anatase and rutile are the main forms of titanium dioxide [6 6] Tita nia is formed through a solution that precipitates as either amorphous or metastable anatase phase and then through heat treatment transforms into the more stable rutile phase [ 6 7 ]. The a natase phase is widely used as catalyst supports because of the larg e overall surface area and thus a high number of available surface sites onto which a metal can be deposited Evidence suggests that the anatase phase leads to strong favorable metal support interactions especially when using noble metal s such as palladi um [ 6 8 ]. However, the main disadvantage with anatase is its low thermal stability and the significant loss in surface area which is observed during phase transformation to rutile [6 8 ]. In this study, the effect s of different TiO 2 supports on the catalyti c activity and the properties of the different TiO 2 catalysts are investigated. As previous research indicated that heat treatment of the n ano particle TiO 2 support increases the contribution from the anatase phase, the n TiO 2 support was treated at differe nt temperatures prior to palladium deposition to investigate in detail the effects of the titania structure on the catalyst properties Since rutile is the stable phase at higher temperatures, temperatures up to 800 C were included to investigate in detai l the effects of both anatase and rutile phases and the support surface area on these catalysts. The
55 objective s of this study are to determine how the morphology of titania effects the palladium support interactions and how this influence the product yield s 3.2 Experimental 3.2.1 Catalyst P reparation The catalysts were prepared using commercially available titanium dioxide nanoparticles (n TiO 2 NanoScale Corporation), p orous titania pellets ( p TiO 2 Alfa Aesar ), and rutile titania powder (r TiO 2 Sigma Aldrich ). Th e nanoparticles were pretreated at 10 5 C, 350 C, 450 C, 600 C, or 800 C for 3 h in an air muffle furnace. A precipitation method was used to deposit palladium onto the support s For this method, an aqueous solution of palladium nitrate (Pd(NO 3 ) 2 2H 2 O Sigma Aldrich) was added to an aqueous dispersion of the titania support. The mixture was titrated with sodium hydroxide (Sigma Aldrich), 50% stoichiometric ex cess based on the amount of palladium nitrate used to form palladium hydroxide(s) on the suppo rt [ 6 9 ]. The resulting mixture was aged overnight at room temperature before being filtered. The filtered material was rinsed in deionized water overnight and filtered again. The recovered material was then dried at 105 C overnight and calcined in air at 350 C for 3 h. The palladium loading was 5% by weight on a metal basis unless otherwise noted. 3.2.2 Reaction C onditions The reactant, 4 methylpyridine (Acros), was doubly distilled over KOH prior to use. In a typical reaction, 0.7 g of catalyst was placed in a round bottom flask with 7 g of distilled 4 methylpyridine. The reaction mixture was evacuated before introducing an oxygen atmosphere. The mixture was then heated to the boiling point (145 C) under continuous agitation. After refluxing for 72 hours, the flask contents were filtered using a glass micro fiber filter and washed with chloroform to dissolve the product. The product
56 was obtained by removing the chloroform, water, and unreacted 4 methylpyridine using a rotary evaporator. 3.2.3 Catalyst C haracterizati on The as received support surface areas, the pretreated support surface areas, and the surface areas of the prepared catalysts were determined by multipoint Brunauer Emmett Teller (BET) isotherms on a Quantachrome Nova 1200 instrument as described in prev ious work [ 57 ]. Chemisorption measurements were performed in a Quantachrome ChemBET 3000 instrument and used to characterize the active metal surface area. The catalysts were first reduced with 5% hydrogen in nitrogen at 170 C for 1 h, and then outgassed in nitrogen at 170 C for another hour. The mild conditions were used to limit the sintering of the Pd particles on the surface. This was followed by pulse titrating carbon monoxide over the reduced catalysts measuring the CO adsorbed, and thus, determine the palladium dispersion of the catalysts. Detailed descriptions of the procedure and calculations are given in previous work [ 70 ]. While PdO is believed to be the active phase, or at least a necessary precursor, the CO chemisorption measurements on reduc ed catalysts are important since the PdO surface cannot be measured directly. It is assumed there is a correlation between the original PdO surface area and the Pd surface area of the reduced catalyst. Previous XRD and TEM data support this assumption [ 70, 71 ]. ray diffractometer using Bragg Brentano geometry with Cu K = 1.54 ). Selected supports and catalyst powders were secured onto a glass slide with double sided sticky tape prior to the measurements.
57 XPS was performed on selected fresh catalysts using a custom built instrument equipped with a double pass cylindrical mirror analyzer (PHI model 25 270 AR). The catalyst powders were pressed into aluminum cups prior to insertion into the ult ra high vacuum (UHV) chamber (base pressure 1 x 10 10 Torr). Spectra were taken in retarding mode with a pass energy of 50 eV for survey spectra and 25 eV for high resolution spectra using a Mg K x ray source (PHI04 151). Data were collected using a computer interface and then digitally smoothed. A value for the C 1s binding energy of 284.6 eV was assigned to correct for static charging and work function differences [ 72 ]. 3.3 Results and Discussion A ll of the catalysts prepared using the different TiO 2 supports were tested for activity in the oxidative coupling of 4 methylpyridine and subjected to a number of catalyst characterization techniques. 3.3.1 Surface Areas Measurements As the support surface are a (SA) is one of the more important support properties, the surface areas were measured for the different TiO 2 samples. For the commercial anatase and rutile supports (p TiO 2 and r TiO 2 respectively) the surface area was measured after drying at 105 C fo r 3 hours (the pretreatment before palladium deposition), while for the n anoparticle TiO 2 the surface area was measured as a function of heat treatment temperature between 105 C and 800 C. The results are presented in Table 3 1 As expected, the n TiO 2 surface area is significantly higher than the surface areas of the p TiO 2 and r TiO 2 Furthermore, the p TiO 2 SA is almost a factor of five higher than the r TiO 2 SA. However, the surface area of the n TiO 2 is evidently unstable. The original surface area of this support was 505 m 2 /g, but after exposure to the atmosphere for several months the SA decreased to 190 m 2 /g.
58 Additionally, the surface area of the n TiO 2 decreases as a function of heat treatment temperature. The SA at 450 C (96 m 2 /g) is less than half of t hat at 105 C (190 m 2 /g) Then there is a drastic decrease in SA between heat treatment temperatures of 450 C and 600 C, which is likely due to the beginning of a phase change from anatase to rutile ( vide infra ). After calcination at 800 C the n TiO 2 surface area (2.3 m 2 /g) is almost two orders of magnitude smaller than the original surface area. In general, the surface areas of the catalysts (after PdO deposition) are slightly lower than that of the support SA. This is expected since the PdO can block some of the pores in the support during deposition. However, for the lowest support surface area, the n TiO 2 heat treated at 800 C (2.3 m 2 /g), adding palladium actually results in an increase in the overall surface area to 5 m 2 /g. The catalyst surface areas measured after palladium deposition reveal that the palladium does not completely block the pores of the support as would lead to a drastically lower catalyst surface area compared with the support surface area. In an attempt to measure the PdO surface area on these catalysts, CO adsorption was measured after reduction of the catalysts. This works reasonably well on non reducible supports, as it would be expected that the PdO surface area correlates fairly well with the measured Pd surface ar ea on the reduced catalysts. However, this is not necessarily the case for reducible supports like TiO 2 Due to strong metal support interactions even low temperature reductive treatment can result in partial reduction of the support, which in turn can lea d to CO adsorption on the support as well as on the palladium. As strong metal support interactions have been shown to be beneficial for this reaction [71 ], overestimating the CO adsorption and thus the Pd surface area due to
59 adsorption on the support woul d not be too detrimental. Therefore, CO adsorption was measured after reduction in hydrogen at 170 C, without adjusting for potential adsorption on the support. The catalyst and support surface areas as well as the CO adsorbed on the reduced catalyst de crease with increasing pretreatment temperature for n TiO 2 supported catalysts, which indicate that the Pd surface area also decreases. As can be seen from Table 3 1 the Pd surface area o n the p TiO 2 support is similar to that on the PdO/n TiO 2 450 C cata lyst. The Pd surface area of the r TiO 2 support is lower than that obtained from the catalysts supported on p TiO 2 and n TiO 2 which is mainly due to the lower surface area of the r TiO 2 support compared to the other TiO 2 supports. 3.3.2 Reaction Results The 4 dimethyl bipyridine product yields for the different PdO/TiO 2 catalysts are summarized in Table 3 1 and Figure 3 1 The best performing catalyst is the PdO/p TiO 2 with a pro duct yield of 52 g /g palladium. As noted previously, this is a much higher yield compared with the PdO/n TiO 2 105 C catalyst despite the lower CO adsorption on the PdO/p TiO 2 catalyst. One explanation for this could be a higher CO adsorption contribution from the n TiO 2 support compared with the p TiO 2 as the surface area of th e n TiO 2 support is higher than that of the p TiO 2 support. However, for the yield to be 40% lower on the PdO/n TiO 2 105C compared with the PdO/p TiO 2 catalyst, the Pd surface area would have to be considerably lower on the n TiO 2 support and the CO adsor ption on the Pd grossly overestimated. Considering the higher surface area of the n TiO 2 support, a substantially lower Pd surface area on the n TiO 2 compared with the p TiO 2 is not likely, and it appears more plausible that other factors,
60 such as the TiO 2 crystal phase as noted previously [ 6 5], influence the yields on the TiO 2 supported catalysts. To investigate the influence of the TiO 2 crystal phase in detail, a rutile TiO 2 support was included in the study. The yield from the PdO/r TiO 2 catalyst is ro ughly half of that obtained from the PdO/p TiO 2 catalyst, which is even lower than the yield obtained from the PdO/n TiO 2 105C catalyst. However, considering that the surface area of the p TiO 2 support is five times that of r TiO 2 a t an even lower yield from the r TiO 2 supported catalyst would be expected. The PdO/r TiO 2 CO adsorption measurements indicate that the Pd surface area is approximately a third of the Pd surface area on the PdO/p TiO 2 catalyst. Since the yield from the PdO/r TiO 2 catalyst is r oughly half of the yield obtained from the p TiO 2 supported catalyst, the turnover number (TON), defined as the number of moles of product formed per mole of surface palladium, is larger on the PdO/r TiO 2 catalyst compared to the other PdO/TiO 2 catalysts. Hence, the PdO on the surface of the r TiO 2 support appears to be more active than PdO on the other TiO 2 supports, but the r TiO 2 support is limited by its low surface area. The PdO/n TiO 2 catalyst prepared using the n TiO 2 support dried at 105 C result s in a higher yield than the PdO/r TiO 2 but lower than the PdO/p TiO 2 catalyst. Considering the high CO adsorption compared to the other TiO 2 supported catalysts, the turnover number on th e PdO/n TiO 2 105C catalyst is amongst the lowest of the catalysts under investigation. The catalyst prepared using the support pretreated at 350 C results in a similar product yield to the one obtained from the PdO/n TiO 2 105C catalyst, despite a lower support surface area. Since the CO adsorption is also lower on
61 the PdO/n Ti O 2 350C, the turnover number is higher compared with the PdO/n TiO 2 105C catalyst. The maximum yield from the n TiO 2 supported catalysts is obtained after support treatment at 450 C. The CO adsorption measured over the PdO/n TiO 2 450C catalyst is similar to the CO adsorption on the PdO/p TiO 2 catalyst. While the product yield is slightly lower for the PdO/n TiO 2 450C, the turnover number for this catalyst is close to the TON obtained from the PdO/p TiO 2 catalyst. At support treatment temperatur es above 450C a drastic drop in product yield i s observed with only 15 g/g Pd at 600 C and 8 g /g Pd at 800 C. Table 3 1 reveals that the main reason for the drastic drop in product yield is due to the loss in support surface area associated with the hig her pretreatment temperatures. At a pretreatment temperature of 800 C, the CO adsorption is an order of magnitude lower than for the PdO/n TiO 2 450C catalyst. The product yields obtained from the PdO/n TiO 2 600C and PdO/n TiO 2 800C catalysts are theref ore higher than would be expected from the estimated Pd surface area (obtained from the CO adsorption measurements). In fact, the turnover numbers for the PdO/n TiO 2 600 C and PdO/n TiO 2 800 C catalysts are closer to the TON for the PdO/r TiO 2 catalyst tha n the PdO/p TiO 2 catalyst. Thus, the turnover number increases with increasing pretreatment temperature, and this is true for all tem peratures under investigation. Support pretreatments at 600 C and 800 C evidently cause a severe reduction in support sur face area (Table 3 1 ), which in turn limits the active PdO surface area that can be obtained on these supports. Therefore, despite the higher turnover numbers on these catalysts, the higher activity cannot counterbalance the significantly lower Pd surface area on the PdO/n TiO 2 600C and PdO/n TiO 2 800C catalysts. To increase
62 the Pd surface area and at the same time improve the turnover number of the PdO/n TiO 2 catalyst, one catalyst was prepared using a n TiO 2 support pretreated at 105 C and then calcine d at 600 C after palladium deposition. The product yield obtained from the PdO/n TiO 2 M6 00 C is almost double that obtained from the PdO/n TiO 2 600C catalyst. This is due to a higher Pd surface area as indicated by the CO adsorption on this catalyst. It is very interesting to note that the turnover numbers are the same for the PdO/n Ti O 2 6 00 C and PdO/n Ti O 2 M6 00 C catalysts. Additionally, another catalyst was prepared to determine if the TON on a high yield catalyst (PdO/n TiO 2 450C) can be increased b y calcin ation at 800 C. The product yield of PdO/n TiO 2 M800C is only a fourth of the yield obtained over the PdO/n TiO 2 450C but it is slightly higher than that obtained over the PdO/n TiO 2 800C catalyst Since the measured Pd surface area is slightl y lower on the PdO/n TiO 2 M800C compared with the PdO/n TiO 2 800C catalyst, the TON is indeed higher on the PdO/n TiO 2 M800C catalyst. However, the catalysts calcined at 600 C and 800 C cannot compete with the other PdO/n TiO 2 catalysts due to the low surface areas. 3.3.3 XRD Measurements T he XRD patterns obtained from the titania supports before palladium deposition are presented in Figure 3 2 A The rutile phase of the r TiO 2 and the anatase phase of the p TiO 2 supports are labeled in the figure. The crysta l sizes of the r TiO 2 and p TiO 2 supports according the Scherrer equation are 12 and 20 nm, respectively. As the pretreatment temperature increases, the initially amorphous n TiO 2 begins to crystallize into an anatase phase. Between 105 C and 450 C the a natase crystal size grows from 6 nm to 12 nm. At the 600 C heat treatment the peaks due to the anatase phase are very sharp indicating large crystal sizes, and peaks due to the rutile phase appears in
63 t he XRD pattern. C omplete transformation to rutile oc curs at the 800 C heat treatment The formation of the rutile phase at 800 C has been reported along with a severe decrease in surface area [ 6 7, 73 ], which is consistent with the XRD data in Figure 3 2 and in the TiO 2 surface area decreasing from 190 m 2 g 1 to 2.3 m 2 g 1 The XRD measurements on the PdO/n TiO 2 catalysts and the reaction data support the previous conclusion that the anatase phase is advantageous for these PdO/TiO 2 catalysts. The turnover number increases with the contribution from the a natase phase, suggesting that the palladium on the surface is more active on anatase TiO 2 compared with amorphous high surface area TiO 2 This is particularly evident when comparing the PdO/n TiO 2 105 C and PdO/n TiO 2 35 0 C catalysts since these two cataly sts have different properties yet yield the same amount of product. While both catalysts were calcined at 350 C after palladium deposition, the XRD pattern reveals that the anatase phase is more developed on the PdO/n TiO 2 350C compared with the PdO/n Ti O 2 105C catalyst as the anatase crystal size grew from 6 nm to 10 nm. Since the n TiO 2 support surface area decreases with increasing heat treatment temperature and increasing anatase phase, the PdO particle size grows from 7 nm to 28 nm for PdO /n TiO 2 10 5C and PdO /n TiO 2 600C, respectively. Therefore, there is a trade off between a higher Pd surface area and a more active surface Pd species. The maximum yield is obtained at a n TiO 2 heat treatment of 450 C Above this temperature there is a drastic d ec rease in support surface area, which appears to coincide with the introduction of the rutile TiO 2 phase. It is interesting to note that the most active palladium species, i.e. the palladium species with the highest TON, is observed on the rutile TiO 2 phase This is the TiO 2 phase with the lowest surface area, which of course
64 l imits the Pd surface area and thus the ability to compete with the PdO supported on anatase TiO 2 The turnover numbers of the PdO/n TiO 2 600 C and PdO/n TiO 2 800 C catalysts are consis tent with the turnover number obtained over the rutile phase of TiO 2 and the XRD confirms the presence of a ru tile phase on these catalysts. The XRD pattern obtained from the PdO/n Ti O 2 M6 00 C catalyst is somewhat surprising since the rutile TiO 2 phase is not observed on this catalyst (Figure 3 3 ). This is despite the fact that the TON is the same as for the PdO/n Ti O 2 6 00 C catalyst, which indicates the presence of a rutile phase. Evidently, the presence of PdO during high temperature calcination prevents formation of bulk rutile, which is why the overall surface area is higher on the PdO/n Ti O 2 M6 00 C compared with the PdO/n Ti O 2 6 00 C catalyst. On the PdO/n Ti O 2 M8 00 C catalyst, w here the support was pretreated at 4 50 C and the catalyst calcined at 800 C after PdO deposition, no rutile phase can be observed, while on the PdO/n Ti O 2 8 00 C this is the only evident TiO 2 phase. This indicates that palladium addition stabilizes the anatase phase of the TiO 2 support. Of course, the high temperature treatment also results in PdO particle growth, which is evident in the XRD patterns. According to the Scherrer equation, the crystal PdO sizes are 18 nm and 35 nm on PdO/n Ti O 2 M6 00 C and PdO/n Ti O 2 M8 00 C respectively. Despite the lack of peaks due to rutile in th e XRD pattern obtained from PdO/n Ti O 2 M6 00 C, the TON indicates that undetectable with XRD rutile has formed at the surface of this catalyst. 3.3.4 XPS Measurements Selected catalysts were analyzed using XPS to obtain detailed information on the oxidation s tates of the surface species and palladium titania interactions. Since the survey spectra obtained from these catalysts are similar, only the region from 500 to 0
65 eV is displayed in Figure 3 4 The survey spectra have been normalized to give Ti 2p peaks of similar intensity to facilitate comparisons between catalysts. The most evident difference between the spectra obtained from the catalysts is due to variations in the Pd 3d peak intensities. The highest Pd 3d peak intensities are observed on the PdO/n TiO 2 600C and PdO/n TiO 2 800C catalysts and reveal palladium does not penetrate the heat treated n TiO 2 support s and is deposited on the outermost surface of these catalysts. This is expected due to the low surface areas of the n TiO 2 600C and n TiO 2 800C supports. For the same reason, the Pd 3d peak intensities are higher on the PdO/r TiO 2 compared to the other PdO/n TiO 2 catalysts. The Pd 3d, Ti 2p and O 1s peaks obtained from these PdO/n TiO 2 catalysts as a function of support pretreatment temperatur e are presented in Figure 3 5 to Figure 3 7 together with the XPS data obtained from the PdO/p TiO 2 and PdO/r TiO 2 catalysts. The Ti 2p 3/2 peaks for PdO/p TiO 2 and PdO/r TiO 2 catalysts are both located at a binding energy of 458.6 eV consistent with TiO 2 (Figure 3 5 ). In contrast, the binding energy of the Ti 2p 3/2 peak obtained from PdO/n TiO 2 105C catalyst is 458.4 eV, which could indicate the presence of more hydroxyl groups on this catalyst. As the support pretreatment temperature increases, the Ti 2p peaks become broader, suggesting the presence of additional TiO x species on the PdO/n TiO 2 catalysts. The Ti 2p peaks are also shifted to lower binding energies with increasing support treatment temperature. After a pretreatment at 800 C, the Ti 2p 3/2 bi nding energy is 458.2 eV. This trend suggests that the palladium titania interactions strengthen with increasing support treatment temperature. The Pd 3d peaks confirm this trend, and indicate that electrons are transferred from palladium to titania.
66 Pre vious research revealed that the Pd 3d 5/2 binding energies of the PdO/p TiO 2 and PdO/n TiO 2 105C catalysts are slightly higher (336.3 eV) compared to the Pd 3d 5/2 bin ding energy of PdO (336.1 eV) [70 ], suggesting strong palladium titania interactio ns. The Pd 3d 5/2 binding energy obtained from the PdO/r TiO 2 catalyst is even higher at 336.6 eV (Figure 3 6 ), which confirms the conclusions from the previous section that the rutile phase results in stronger palladium titania interactions compared to the anatas e phase. As the support pretreatment temperature increases from 105 C to 450 C, the Pd 3d peaks broaden like the Ti 2p peaks. The Pd 3d 5/2 peak obtained from the PdO/n TiO 2 600C has a binding energy of 336.4 eV indicating a contribution from a PdO speci es on rutile TiO 2 in agreement with the XRD data. On the PdO/n TiO 2 800C catalyst the Pd 3d 5/2 binding energy is 336.6 eV as expected from the PdO/r TiO 2 catalyst. The O 1s binding energies on the PdO/p TiO 2 and PdO/n TiO 2 105C catalysts are the same at 529.9 eV, while the O 1s BE of the PdO/r TiO 2 is slightly lower at 529.7 eV. As the n TiO 2 support heat treatment increases to 350 C the O 1s peak shifts to 529.6 eV, suggesting that changes occur at the surface of these catalysts even at 350 C. The s hift towards the O 1s BE of rutile is clearly advantageous, as the turnover number for this catalyst is higher than that of the PdO/n TiO 2 105C catalyst. As expected from the higher Pd 3d peak intensities in the survey spectra obtained from the PdO/n TiO 2 600C and PdO/n TiO 2 800C catalysts, the contribution from the Pd 3p peak at 533 eV increases with increasing support pretreatment temperature. This is clearly evident in the shoulder at 533.3 eV in Figure 3 7 As with the Pd 3d peaks, it
67 appears that th e Pd 3p peak shifts to higher binding energies with increasing support treatment temperature. 3.4 Summary Heat treating the support induces a phase change in n TiO 2 as anatase forms at 350 C and rutile begins to emerge at 600 C. T he phase changes from a morphous to anatase and then rutile are associated with a continuous decrease in the n TiO 2 surface area from 190 m 2 /g for the mostly amorphous n TiO 2 105C to 2.3 m 2 /g for rutile n TiO 2 800C. Despite the decrease in support surface area, and thus also pa lladium surface area, the product yield increases with support pretreatment temperature up to 450 C, which reveals that the anatase phase is important to obtain high yields. The drastic decline in surface area with heat treatment above 450 C causes lower product yields, but the TON is higher once the rutile phase is present. The two catalysts, PdO/n TiO 2 800C and PdO/r TiO2, have two of the higher TON of 372 and 292, respectively. The XPS data shows that PdO interacts more strongly with anatase TiO 2 comp ared with amorphous TiO 2 and the interactions increase with introduction of a rutile TiO 2 phase. This is evidenced in a shift in the Pd 3d 5/2 BE from 336.3 eV on PdO/n TiO 2 105 C to 336.6 eV on PdO/n TiO 2 800C. These stronger palladium titania interactio ns in the rutile phase are favorable and can be seen in the increase in TON with increasing palladium titania interactions. However, these catalysts are limited by their low support surface areas. Thus, the best PdO/n TiO 2 catalyst is the PdO/n TiO 2 450C, which gives a yield (44 g product/g Pd) that almost matches the best catalyst; PdO/p TiO 2 with a product yield of 52 g product/g Pd.
68 Table 3 1 Catalyst surface areas, CO adsorption, palladium dispersions, yield, and turnover numbers of PdO/TiO 2 support catalysts. Entry PdO/TiO 2 Catalyst a Support Treatment Temp [C] b Support SA [m 2 /g] c Catalyst SA [m 2 /g] d CO ads. cat] Dispers ion [%] Product Yield [g/g cat] Product Yield [g/g Pd] T ON e 1 p TiO 2 105 C 105 120 f 115 f 1450 14 2.6 f 52 f 209 2 n TiO 2 105 C 105 190 170 2960 28 1.6 f 31 f 70 3 g n TiO 2 350 C 350 133 110 1720 16 1.5 31 110 4 n TiO 2 4 50 C 450 96 80 1530 15 2.2 44 175 5 n TiO 2 600 C 600 21 6 350 3.0 0.8 15 256 6 n TiO 2 800 C 800 2.3 5 135 1.3 0.4 8.3 372 7 r TiO 2 105 C 105 24 26 520 5.0 1.3 25 292 8 h n TiO 2 M600 C 105 30 660 6.0 1.4 28 255 9 i n TiO 2 M800 C 450 23 110 1.0 0.6 12 680 a 5wt% Pd on a metal basis. b Calcination temperature of the support prior to Pd deposition. c BET surface area of the support after 3 h calcination. d BET surface area after Pd deposition and 350 C calcination for 3 h. e Turnover number in m ol product per mol surface Pd. f Results from previous work [ 44 ]. g Results fro m previous work [ 65 ]. h Modified calcination where the support was pretreated at 105 C and calcined at 600 C for 3 h after Pd deposition. i Modified calcination where the support was pretreated at 450 C and calcined at 350 C for 3 h after Pd deposition then re calcined at 800 C for 3 h.
69 Figure 3 1. Activity data for PdO catalysts supported on n TiO 2 pretreated at different temperatures, r TiO 2 and p TiO 2
70 Figure 3 2 X ray diffraction (XRD) data where A = anatase TiO 2 R = rutile TiO 2 and o = PdO for (A) TiO 2 supports after temperature treatment and (B) PdO/TiO 2 catalysts.
71 Figure 3 3 X ray diffreaction (XRD) narrow scans where A = anatase TiO 2 R = rutile Ti O 2 and o = PdO for PdO/TiO 2 catalysts.
72 Figure 3 4 X ray photoelectron spect roscopy (XPS) survey scan spectra for PdO/TiO 2 catalysts.
73 Figure 3 5 X ray photoelectron spectroscopy (XPS) Ti 2p spectra for PdO/TiO 2 catalysts.
74 Figure 3 6 X ray ph otoelectron spectroscopy (XPS) Pd 3d spectra for PdO/TiO2 catalysts.
75 Figure 3 7 X ray photoelectron spectroscopy (XPS) O 1s spectra for PdO/TiO 2 catalysts.
76 CHAPTER 4 T HE INFLUENCE OF ZnO, CeO 2 AND ZrO 2 ON NANOPARTICLE O X IDE SUPPORTED PALLADIUM OXIDE CATALYST S FOR THE OXIDATIVE COUPLING OF 4 METHYLPYRIDINE 1 4.1 Back ground Bipyridines possess the ability to coordinate to transition metal cations and form complexes with distinct photochemical and catalytic properties [40,53,59,60] Bipyridine complexes with ruthen ium are especially interesting in applications for organic light emitting diodes  and chemiluminescence detection systems [38,39,41] Various catalyst systems also use transition metal complexes with bipyridine ligands [42,74] including oxidative carb o n ylation  the Kumada Corriu reaction  and the Suzuki cross coupling reaction  Yet, a widespread, large scale usage is, in part, limited by the cost associated with bipyridine compounds. dimethyl bipyridine exceeds $5,600 per kilogram  Therefore, finding a more economical synthesis pathway for bipyridines is desirable. Environmental impacts are also an important concern. Hence, reactions with no so lvents and no halogenated compounds are preferred. The oxidative coupling of 4 dimethyl bipyridine over palladium is a simple one step process that uses neither solvents nor halogenated precursors ( Figure 4 1 ) with water and ter pyridine as the only observed by products. While this is an environmentally friendly process, the reaction rate is slow, and the catalyst undergoes deactivation which in turn limits the product yield  Early research for this reaction focused on using palladium on carbon (Pd/C) catalysts with 1 Reprinted from Journal of Molecular Catalysis A: Chemical Vol. 341 pp. 42 50, Justin J. Dodson, Luke M. Neal, Helena E. Hagelin Weaver (2011) with permission from Elsevier.
77 yields for 5 wt% and 10 wt% Pd loadings varying between 1.5 and 2.0 g bipyridine per g catalyst [ 52,62 64] In our previous research, it was shown that both 5 wt% and 10 wt% palladium (metal basis) precipitated ont o nanoparticle alumina [PdO/n Al 2 O 3 ( +)] yield approximately 2.5 g product per g catalyst  As the product yield per gram of palladium for the catalyst with 5% palladium loading is almost twice that of the catalyst with 10% palladium loading, the 5 wt% palladium catalyst is more economical and is the best performing palladium catalyst reported to date [44,57] Several other oxide supports including porous titania (p TiO 2 ) and nanoparticle magnesia (n MgO) with a 5 wt% palladium loading produce reasonabl e yields, similar to or higher than those obtained from a Pd/C catalyst  In addition, catalysts supported on nanoparticle ceria (n CeO 2 ), nanoparticle zinc oxide (n ZnO), and nanoparticle zirconia (n ZrO 2 ) were found to produce moderate yields (>1.5 g per g catalyst) despite relatively low support surface areas (<70 m 2 /g)  The moderate yields and high dispersions on the latter nanoparticle oxides indicate favorable metal support interactions  XPS confirms strong palladium support interactions as the Pd 3d peaks are shifted to higher binding energies indicating electron deficient PdO species on the surface  The favorable interactions are further evidenced by the large number of studies on palladium promoted with or supported on CeO 2 [39,5 9,60,79 83] ZrO 2 [84,85] and Zn/ZnO [40,86 88] These oxides are therefore potential promoters for palladium in the oxidative coupling of 4 methylpyridine. As it has been shown previously that the PdO on the surface of the catalysts is reduced to Pd metal after reaction, the reactions are likely limited by the reoxidation of palladium [52,70,71] Consequently, the oxygen mobility and storage capacities of CeO 2 and ZrO 2 are particularly interesting for the oxidative
78 coupling reaction. Assistance in oxygen t ransfer between palladium and the support could potentially produce highly effective catalysts. In this study, the effects of ZnO, CeO 2 and ZrO 2 on a few selected nanoparticle oxide supported catalysts were investigated. The n Al 2 O 3 ( +), n TiO 2 and n MgO s upports were selected due to very high surface areas (>500 m 2 /g) and, consequently, a strong potential for producing catalysts with high palladium dispersions. The additives were deposited on the support using two methods, (1) along with the active metal ( co precipitation) or (2) before the active metal (in a two step sequential precipitation method). The main objectives of the study were to determine (1) if CeO 2 ZrO 2 and ZnO can promote palladium catalysts supported on n Al 2 O 3 ( +), n TiO 2 and n MgO, (2) i f the effects of the additive are dependent on the nanoparticle oxide support used, (3) which is the more effective preparation method, co or sequential precipitation of the additive and palladium, and (4) if a catalyst with a greater activity than the be st to date, i.e. the PdO/n Al 2 O 3 ( +), can be prepared by adding one of these oxides. 4.2 Experimental 4.2.1 Catalyst P reparation The catalysts were prepared using commercially available nanoparticles, n Al 2 O 3 ( +), n MgO and n TiO 2 supplied by NanoScale Corporation  ( Table 4 1 ). A precipitation method was used to deposit palladium and additive onto the supports. Using this method, aqueous solutions of metal nitrate(s) (Pd(NO 3 ) 2 2H 2 O: Fluka, Ce(NO 3 ) 3 6H 2 O: Sigma Aldrich, Zn(NO 3 ) 2 6H 2 O: Alfa Aesar, and ZrO(NO 3 ) 2 6H 2 O: Sigma Aldrich) were added to an aqueous dispersion of the support (either n Al 2 O 3 (+), n MgO or n TiO 2 ). The mixture was titrated with sodium hydroxide to form metal hydroxide(s) on the support  The amount of NaOH corresponded to a 50%
79 stoichiometr ic excess, based on the amount of metal nitrate(s) used. The resulting mixture was aged overnight at room temperature before being filtered. The filtered material was rinsed in deionized water overnight and filtered again. The material was then dried at 10 5 C overn ight and calcined in air at 350 C for 3 h. All of the catalysts had loadings of 5% palladium and 5% additive by weight on a metal basis, unless otherwise noted. Two different precipitation methods were used; co precipitation and sequential precip itation, to determine the effects of the additive. For the co precipitation method, cerium nitrate, zinc nitrate or zirconium nitrate was dissolved with palladium nitrate in deionized water before being added to an aqueous dispersion of the support. The me tals were then precipitated together onto the support by titration with a sodium hydroxide solution. In the sequential precipitation method, the metal oxide additive was deposited first by precipitation onto the support, aged, rinsed, dried and calcined be fore repeating the process to deposit the palladium. 4.2.2 Reaction C onditions The reactant, 4 methylpyridine (Acros), was doubly distilled over KOH prior to use. In a typical reaction run, 0.7 g of catalyst was placed in a round bottom flask with 7 g of the dis tilled 4 methylpyridine. The reaction mixture was evacuated followed by the introduction of an oxygen atmosphere. The mixture was then heated to the boiling point ( 145 C) under continuous agitation. After refluxing for 72 h, the flask contents were filter ed using a glass micro fiber filter and washed with chloroform to dissolve the product. The product was obtained by removing the chloroform, water, and unreacted 4 methylpyridine using a rotary evaporator.
80 4.2.3 Catalyst C haracterization The as received support surface areas and the surface areas of the prepared catalyst were determined by multipoint Brunauer Emmett Teller (BET) isotherms on a Quantachrome Nova 1200 instrument as described in previous work  Chemisorption measurements were performed in a Quan tachrome ChemBET 3000 instrument and used to characterize the active catalyst surface area. The catalysts were fir st reduced with hydrogen at 170 C for 1 h, and th en outgassed in nitrogen at 170 C for another hour. The mild reduction conditions were used to limit sintering of the Pd particles on the surface. This was followed by pulse titration with carbon monoxide to characterize the palladium dispersions of the catalysts. Detailed descriptions of the procedure and calculations are given in previous work  While PdO is believed to be the active phase, or at least a necessary precursor, the CO chemisorption measurements on reduced catalysts are important since the PdO surface area cannot be measured directly. It is assumed that there is a correlation b etween the original PdO surface area and the Pd surface area of the reduced catalyst. Previous XRD and TEM data support this assumption [70,71] The XRD data was gathered on a Philips powder X ra y diffractometer using Bragg = 1.54 ). Selected catalyst powders were secured onto a glass slide with double sided sticky tape prior to measurements. The thermogravimetric analysis (TGA) of the three nanoparticle supports was carried out on a Mettler Toledo TGA/SDTA (scanning differential thermal analysis) with sample weights between 4 and 20 mg, under a 40 mL min 1 air flow. The samples were heated from room temper ature to 105 C at 20 C min 1 and held for 5 min bef ore again being heated from 105 C to 1000 C at 10 C min 1
81 4.3 Results All prepared catalysts were subjected to activity measurements in the coupling reaction of 4 methylpyridine. The results are summarized in Table 4 2 and Fig ure 4 2 The catalysts with out additives from previous research, PdO/n Al 2 O 3 (+), PdO/n TiO 2 and PdO/n MgO, as well as some additional catalysts prepared using a slightly modified preparation procedure, are included for comparison. The properties of the supports have been presented in a previous paper  but are summarized in Table 4 1 as they are important in the current study. As seen in Table 4 2 the effects of the additives are dependent on the identities of the additive and support as well as the preparation method. To obta in more information about the effects of the oxide additives and the preparation methods on the different supports, the overall surface areas were determined, and CO chemisorption measurements were performed to estimate the Pd surface areas of the catalyst s. The results from these measurements are presented in Table 4 3 4.4 Discussion 4.4.1 PdO S upported on N anoparticle A lumina Previous research showed that the PdO/n Al 2 O 3 (+) catalyst is the most active and reproducible catalyst amongst the various catalysts examin ed for this reaction. Addition of ZnO, CeO 2 and ZrO 2 to this catalyst was investigated in an attempt to improve the product yields. 220.127.116.11 Precalcination of n Al 2 O 3 ( +) As Fig ure 4 2 indicates, the sequential precipitation method is inferior to co precipitation f or catalysts supported on n Al 2 O 3 (+). Thus, the effect of the first
82 calcination treatment was explored. A n Al 2 O 3 (+) sample was calcined in air at 350 C for 3 h before palladium deposition and then tested for activity. This support pretreatment results in a significantly reduced product yield ( Table 4 2 ). The difference between the catalysts prepared using heat treated and untreated supports is attributable to the lower Pd surface area on the heat treated n Al 2 O 3 (+). The lower Pd surface area is not cau sed by a drastic reduction in support surface area, as the surface area of the n Al 2 O 3 (+) is still above 500 m 2 /g even after calcination for 24 h at 350 C. Previous research revealed that n Al 2 O 3 (+) mainly consist s of poorly crystalline Al 2 O 3 between 350 C and 450 C . From TGA measurements ( Figure 4 3 weight loss (weight lost after drying at 105 C for 5 min) oc curs between 105 C and 350 C. As roughly 4% out of the total 12% dry weight loss is lost during the ramp to 350 C (a 15% total weight loss would be expected for complete conversion of a 100% Al 2 O 3 ), a significant fraction of the AlO(OH) is expected to transform into Al 2 O 3 during the 3 h calcination at 350 C. Even though the XRD spectra of catalysts calcined at 350 Al 2 O 3 ), TGA strongly indicates that a significant fraction of the original AlO(OH) has decomposed after calcination at this temperature. The XRD measurements cannot detect the developing Al 2 O 3 phase. The low Pd surface area measured on the catalysts prepared using the precalcined n Al 2 O 3 (+) therefore suppor ts our previous conclusion that the hydroxyl groups in the n Al 2 O 3 (+) (or rather the n AlO(OH)) support are important to obtain a high Pd surface area. It is our belief that these hydroxyl groups
83 interact with the Pd 2+ ions during the precipitation proces s and thereby result in high Pd dispersions on these catalysts. 18.104.22.168 ZnO addition While ZnO has been shown to result in favorable Pd Zn interactions in PdO/n ZnO catalysts  adding ZnO to the PdO/n Al 2 O 3 ( +) catalyst does not have a positive effect on the p roduct yield irrespective of catalyst preparation method used ( Fig ure 4 2 and Table 4 2 ). The co precipitation of ZnO and PdO precursors only slightly decreases the yield, but the sequential precipitation method results in a significant decrease to half th e original yield from PdO/n Al 2 O 3 ( +). This is even lower than the yield obtained from the PdO supported on precalcined n Al 2 O 3 ( +). Previous results have shown that deposition of ZnO onto n Al 2 O 3 ( +) support under similar preparation techniques results in facile ZnAl 2 O 4 formation  Therefore XRD spectra were obtained for co precipitated (CP) and sequentially precipitated (SQ) PdO/ZnO/n Al 2 O 3 ( +) catalysts, and reveal additional peaks which are reasonably consistent with ZnAl 2 O 4 ( Fig ure 4 4 ). The peaks due to ZnAl 2 O 4 are more pronounced on the SQ catalyst indicating that the presence of palladium may disrupt the Zn Al interactions during co precipitation, possibly due to Pd Zn interactions. The lower yields obtained from the ZnO containing catalysts corr elate with the reduction in Pd surface area, as both the CP and SQ Al 2 O 3 ( +) catalyst without added ZnO. In the case of the sequentially prepared PdO/ZnO/n Al 2 O 3 ( +) catalyst, the lower Pd surface area is attributed to the smaller support surface area and the removal of the surface hydroxyl groups during ZnO addition and calcination. In the case of the CP catalyst, a slightly lower Pd surface area is observed, while the overall
84 surface a rea is higher, compared to the PdO/n Al 2 O 3 ( +), which could be due to some ZnO covering the PdO on the surface. 22.214.171.124 CeO 2 addition Co precipitation of CeO 2 and PdO precursors on the n Al 2 O 3 ( +) support does not affect the average product yield significantly, ev en though the measured Pd surface area does increase ( Table 4 2 and Table 4 3 ). As with the ZnO additive, the sequential addition of CeO 2 produces an inferior catalyst compared to the co precipitated PdO/CeO 2 /n Al 2 O 3 ( +) catalyst and gives a significantly lower yield than PdO supported on a precalcined n Al 2 O 3 ( +) support. One reason for the low yield is a lower Pd surface area for the SQ PdO/CeO 2 /n Al 2 O 3 ( +) than the PdO/n Al 2 O 3 ( +) catalyst with no CeO 2 additive. However, as the Pd surface area of the SQ PdO/CeO 2 /n Al 2 O 3 ( +) catalyst is higher than that of PdO supported on the precalcined n Al 2 O 3 ( +), the turnover number for the SQ catalyst is lower than expected from CeO 2 containing catalysts. Upon examining the results from the SQ PdO/CeO 2 /n Al 2 O 3 ( +) ca talyst series, the yields vary significantly from 2.1 to 3.1 (i.e. 2.6 0.44) g product/g catalyst. As strong PdO CeO 2 interactions leading to migration of CeO x over the palladium have been observed on these and other catalysts [70,91,92] the yield varia tion in this case may be due to varying degrees of CeO 2 coverage of Pd active sites. CeO 2 migration during reaction would explain the lower than expected yields despite the reasonably high Pd surface areas from the PdO/CeO 2 /n Al 2 O 3 ( +) catalysts. 126.96.36.199 ZrO 2 addi tion In contrast to the ZnO and CeO 2 addition of ZrO 2 via the co precipitation method consistently demonstrates a significant improvement in the catalyst performance compared to the PdO/n Al 2 O 3 ( +) catalyst. The co precipitated method produces a yield
85 of 3.4 0.1 g/g catalyst, which is approximately a 36% increase over the highest yield reported to date for this reaction. As with the other oxides added to the PdO/n Al 2 O 3 ( +) catalyst, sequential precipitation results in an inferior catalyst. However, the yield from the SQ PdO/ZrO 2 /n Al 2 O 3 ( +) catalyst is only slightly lower than that from the PdO/n Al 2 O 3 ( +) catalyst, and, more importantly, is significantly higher than the yield obtained from PdO on a precalcined n Al 2 O 3 ( +) support. This indicates that Zr O 2 is acting as a promoter for the PdO/n Al 2 O 3 ( +) catalyst regardless of the preparation method. 188.8.131.52 Modified preparation method To further investigate the differences between the co precipitation and the sequential precipitation methods, the sequential preci pitation procedure was modified for the PdO/ZrO 2 /n Al 2 O 3 ( +) catalyst. Instead of calcining the catalyst after deposition of ZrO 2 one catalyst was simply aged, rinsed, and filtered before re dispersion in an aqueous Pd 2+ solution for PdO deposition. This results in a catalyst with almost the same activity as a co precipitated catalyst ( Table 4 2 Entries 7 and 9) and confirms that the calcination step between depositions has an undesirable effect on the catalysts. In summary, the co precipitation method is preferred over the sequential precipitation on n Al 2 O 3 ( +) supported palladium catalysts. The calcination step between depositions is detrimental and results in lower Pd surface areas which are attributed to removal of OH groups on the support. Only ZrO 2 is promoting the activity of a PdO/n Al 2 O 3 ( +) catalyst. 4.4.2 PdO S upported on N anoparticle M agnesia While the n MgO support has a surface area comparable to the n Al 2 O 3 ( +) support (> 650 m 2 /g), the PdO/n MgO catalyst is not as active as the PdO/n Al 2 O 3 ( +) ca talyst in the coupling of 4 methylpyridine, mainly due to a lower Pd surface area. The
86 effects of ZnO, CeO 2 and ZrO 2 addition to the PdO/n MgO catalyst were therefore investigated to determine if the yields of this MgO supported PdO catalyst could be impro ved. 184.108.40.206 Precalcination of n MgO The behavior of the PdO supported on the precalcined n MgO is similar to that of PdO on precalcined n Al 2 O 3 ( +) as a significant reduction in product yield is observed ( Fig ure 4 2 ). However, in contrast to the PdO catalysts sup ported on n Al 2 O 3 ( +), the lower yield is not due to a decreased Pd surface area on the precalcined n MgO support ( Table 4 3 ). TGA analysis reveals that the n MgO support exhibits a similar weight loss behavior upon heating as seen in the n Al 2 O 3 ( +) suppo rt ( Figure 4 3 ). This suggests that the n MgO support consists of a significant amount of Mg(OH) 2 While most of the weight loss in the AlO(OH) containing n Al 2 O 3 ( +) support occurs in the temperature range between 350 C and 450 C, the n MgO support lose s weight over the whole temperature range from 100 C to 800 C, although a very rapid weight loss between 300 C and 400 C is observed. The total weight loss on a dry basis is 21% (i.e. the weight lost after the drying period). As a total weight loss of 31% is expected for converting a sample consisting of 100% Mg(OH) 2 to MgO, the TGA measurements indicate that approximately 70% of the n MgO is Mg(OH) 2 It is expected that a significant fraction of the Mg(OH) 2 remains in the sample after calcination at 35 0 C for 3 h. This is consistent with an XRD spectrum obtained from the PdO/n MgO catalyst after calcination, which reveals peaks from both MgO and Mg(OH) 2 ( Figure 4 5 ). Consequently, after precalcination, a large number of hydroxyl groups are likely still present, which can interact with the Pd 2+ ions during deposition to yield a high Pd dispersion. Despite similar initial support surfaces areas, the Pd surface area is not as
87 high on the n MgO as the n Al 2 O 3 ( +) support, which may be due to more acidic hyd roxyl groups on the alumina support compared with the basic magnesia support ( Table 4 1 ). 220.127.116.11 ZnO addition Similar to the n Al 2 O 3 ( +) supported catalysts, addition of ZnO to the PdO/n MgO catalyst results in a significant reduction in the product yield irrespe ctive of the preparation method. As expected from the results over the PdO supported on precalcined n MgO, the sequential precipitation method is inferior to the co precipitation method. The lower yields obtained from the PdO/ZnO/n MgO catalysts are largel y attributed to the lower Pd surface area compared to the PdO/n MgO catalyst with no added ZnO. The larger overall surface area and the smaller Pd surface area of the CP compared to the SQ PdO/ZnO/n MgO catalyst indicates more ZnO covering the surface PdO during co precipitation. Despite the lower Pd surface area of the CP PdO/ZnO/n MgO catalyst, the product yield is higher than that obtained from the SQ catalyst, resulting in a higher CP catalyst turnover number ( Table 4 3 ). Thus, the palladium on the surf ace of the CP PdO/ZnO/n MgO catalyst is more active, or at least more stable ( vide infra ), than on the SQ catalyst. According to XRD the main differences between the SQ and the CP PdO/ZnO/n MgO catalysts appear to be (1) a more crystalline PdO on the CP Pd O/ZnO/n MgO catalyst and (2) less Mg(OH) 2 on the SQ PdO/ZnO/n MgO catalyst ( Fig ure 4 5 ). A lower Mg(OH) 2 content in the SQ PdO/ZnO/n MgO catalyst is expected with the second calcination. While no other crystal phases are visible in the XRD spectra, the for mation of poorly crystalline mixed oxides between ZnO and n MgO (Mg x Zn x O)  on this catalyst surface cannot be excluded. Strong interactions between ZnO and n MgO
88 could disrupt favorable palladium support or Pd ZnO interactions leading to low Pd surfa ce areas and catalytic activities. 18.104.22.168 CeO 2 addition CeO 2 addition to the MgO supported palladium catalyst via the co precipitation method results in a slightly lower yield compared to the PdO/n MgO catalyst. The lower yield is not related to a decrease in Pd surface area as it is slightly higher on the CP PdO/CeO 2 /n MgO catalyst compared with the PdO/n MgO catalyst without added CeO 2 ( Table 4 3 ). However, this preparation method causes leaching of palladium into the reaction mixture, which is evident as a Pd m irror on the reaction flask wall after reaction. The palladium leaching indicates strong Pd CeO 2 interactions during the co precipitation which undermine the Pd interactions with the support. Leaching, in turn, lowers the activity despite a higher initial Pd surface area compared to the PdO/n MgO catalyst without added CeO 2 On the contrary, no palladium leaching was observed in the reactions using the SQ PdO/CeO 2 /n MgO catalyst. This catalyst has a higher Pd surface area than the co precipitated catalyst, increases the product yield 17% over the PdO/n MgO catalyst without added CeO 2 and increases the yield 60% over PdO supported on precalcined n MgO. However, similar to the SQ PdO/CeO 2 /n Al 2 O 3 ( +) catalyst, there appears to be a larger variation in yields compared to other catalysts. A catalyst prepared using the modified sequential precipitation method results in an even broader yield distribution and is thus not a very stable catalyst. It is possible that the catalytic activities of the n MgO supported ca talysts are very dependent on the drying and calcination procedure as Mg(OH) 2 decomposes to MgO beginning at 320 C  and continues to decompose up to 800 C ( Figure 4 2 ). The large variations in yields from these catalysts could be
89 due to varying amoun ts of palladium leaching during reaction, which may be challenging to detect in some cases. It is also possible that CeO 2 migration over the Pd/PdO on the catalyst surface occurs during reaction contributing to lower than expected yields from the measured Pd surface area. 22.214.171.124 ZrO 2 addition Unlike the results for the n Al 2 O 3 (+) support, adding ZrO 2 to the PdO/n MgO catalyst yields a slightly lower activity for both preparation methods. This is despite the fact that both catalysts, CP and SQ, have higher Pd surf ace areas than the PdO/n MgO catalyst without added ZrO 2 The CP catalyst gives slightly better average yields than the SQ catalyst, which is attributed to a higher Pd surface area since the turnover numbers are the same. As palladium leaching is observed in some runs using the CP PdO/ZrO 2 /n MgO catalyst, this can account for the lower activity compared with the reference PdO/n MgO catalyst. For the SQ PdO/ZrO 2 /n MgO catalyst, both the yields and the measured Pd surface areas vary significantly between runs indicating an unstable catalyst. Although the measured Pd surface areas are higher, the turnover numbers of the PdO/ZrO 2 /n MgO catalysts are lower in comparison to the PdO/n MgO catalyst revealing less active catalysts. The co precipitation method is sli ghtly better than the sequential precipitation method for n MgO supported catalysts due to removal of hydroxyl groups during the first calcination treatment. However, the preferred preparation method is dependent on the added oxide, as the sequential prepa ration method gives a higher yield for the PdO/CeO 2 /n MgO catalyst. Compared to the other catalysts under investigation, the BET surface areas of the n MgO supported catalysts are very low. The significant reduction in MgO surface area during catalyst prep aration is not directly due to
90 calcination of the support ( Table 4 1 ). Instead the drastic reduction in surface area appears to originate during the palladium deposition step. This may be attributed to partial dissolution of the MgO (or Mg(OH) 2 ) support in the slightly acidic Pd(NO 3 ) 2 (aq) solution  CeO 2 addition increases the yield of the PdO/n MgO catalyst, mainly due to an increased Pd surface area. However, a sequential precipitation method is necessary to prevent leaching of palladium. 4.4.3 PdO S upport ed on N anoparticle T itania The n TiO 2 has a very high surface area ( 500 m 2 /g), but the PdO/n TiO 2 catalyst is only moderately active, comparable to the typical Pd/C in the oxidative coupling of 4 methylpyridine. Therefore, the addition of oxides with fav orable properties has the highest potential for improving the catalytic performance of the catalysts under investigation. Moreover, the variation in yield of the PdO/n TiO 2 catalyst has been shown to be larger than for PdO/n Al 2 O 3 ( +). Thus, the addition o f ZnO, CeO 2 and ZrO 2 was investigated to determine if more reproducible and active catalysts can be obtained from n TiO 2 supported PdO. 126.96.36.199 Precalcination of n TiO 2 Precalcination of the n TiO 2 support before PdO deposition does not notably affect the yield of the resulting catalyst ( Fig ure 4 2 ). However, the measured Pd surface area is significantly lower on the precalcined n TiO 2 support compared to the reference PdO/n TiO 2 catalyst ( Table 4 3 ), which is likely due to the drastic reduction in BET surface area with calcination of the n TiO 2 support ( Table 4 1 ). The heat treatment causes a 74% reduction in surface area, from 505 to 133 m 2 /g, compared to a 59% reduction in Pd surface area observed on the calcined support. In contrast to the n Al 2 O 3 ( +) and n MgO supports, the n TiO 2 does not appear to contain a high number of
91 hydroxyl groups. The TGA reveals a dry weight loss of only 3.4% for the n TiO 2 support with most of this weight loss occurring before the 350 C calcination temperature. It appears that the d rastic change in BET surface area with calcination of the n TiO 2 support is due to a phase change ( Figure 4 6 ). The as received n TiO 2 is nearly amorphous with an indication of an anatase phase present, but there is no significant long range order and the presence of other phases cannot be excluded. The XRD spectra of the PdO/n TiO 2 catalyst clearly reveal the growth of anatase crystallites with calcination. Precalcination followed by another calcination treatment after palladium deposition likely increases the anatase phase contribution, as seen for the SQ PdO/ZrO 2 /n TiO 2 catalyst in Figure 4 6 Anatase is known to result in strong Pd support interactions  and has been shown to result in highly active PdO catalysts in the oxidative coupling of 4 methylp yridine  Therefore, development of the anatase phase during calcination is likely the reason the lower Pd surface area on the precalcined support does not result in a reduction in yield. 188.8.131.52 ZnO addition Similar to the behaviors of the n Al 2 O 3 ( +) and n MgO supported catalysts, addition of ZnO results in lower yields than those obtained from the PdO/n TiO 2 without ZnO. In this case, the sequential precipitation method produces a higher yield than the co precipitation, the latter reducing the yield 31% com pared to the n TiO 2 without added ZnO. As for the other CP ZnO containing catalysts, it appears that ZnO is covering the palladium on the CP PdO/ZnO/n TiO 2 catalyst, as the overall surface area is larger, but the Pd surface area is lower compared to the Pd O/n TiO 2 catalyst. In contrast, on the SQ PdO/ZnO/n TiO 2 catalyst the Pd surface area is larger than on the PdO/n TiO 2 catalyst, but the yield is lower. It is likely that the presence of ZnO disrupts the favorable
92 PdO anatase TiO 2 interactions, either by s imply blocking the n TiO 2 from PdO or via the formation of a mixed oxide between ZnO and n TiO 2 (such as ZnTiO 4  ). 184.108.40.206 CeO 2 addition CeO 2 addition to PdO/n TiO 2 has minimal effect on the yield if sequentially precipitated and decreases the yield if co prec ipitated onto the support with palladium. Similar to the behavior of CP PdO/CeO 2 /n MgO, the CP PdO/CeO 2 /n TiO 2 catalyst results in a significant amount of palladium leaching. Therefore, it appears that co precipitation of Ce and Pd precursors on the n MgO and n TiO 2 supports weakens the Pd support interactions. The Pd Ce interactions are likely stronger than the Ce Ti and Pd Ti interactions [70,91,92,98] Since leaching is not observed on the CP PdO/CeO 2 /n Al 2 O 3 ( +) catalyst, this may indicate that the Pd i nteractions with the hydroxyl groups of AlO(OH) in the n Al 2 O 3 ( +) are stronger than the Pd Ce interactions during deposition. For both the SQ and CP PdO/CeO 2 /n TiO 2 catalysts, addition of CeO 2 increases the amount of CO adsorbed on the catalysts compared to the PdO/n TiO 2 without CeO 2 indicating that the Pd surface areas are larger for the CeO 2 containing catalysts. However, as the yields are not significantly higher, the turnover numbers are lower for the PdO/CeO 2 /n TiO 2 catalysts. While palladium leachi ng from the catalyst surface likely leads to lower than expected yields, it is also possible that CeO x species migrate over the palladium surface species thereby reducing the active metal surface area and the yield. This has been observed previously for Pd O/CeO 2 catalysts  220.127.116.11 ZrO 2 addition The sequentially precipitated PdO/ZrO 2 /n TiO 2 catalyst has a higher activity than the catalyst prepared using the co precipitation method, which is similar to the behavior
93 of the other n TiO 2 supported catalysts (PdO/Zn O/n TiO 2 and PdO/CeO 2 /n TiO 2 ). The SQ PdO/ZrO 2 /n TiO 2 catalyst results in a 20% higher yield, while the CP PdO/ZrO 2 /n TiO 2 catalyst yield is significantly lower in comparison to the PdO/n TiO 2 catalyst. Table 4 3 reveals that the Pd surface areas are almos t the same on the CP and SQ PdO/ZrO 2 /n TiO 2 catalysts and, thus, cannot account for the observed differences in activity. While leaching of palladium is common on the n TiO 2 supported catalysts, it does not explain the low yield of the CP catalyst. It appe ars that palladium leaching affects the SQ PdO/ZrO 2 /n TiO 2 catalyst more than the CP PdO/ZrO 2 /n TiO 2 catalyst, as evidenced in a palladium mirror covering the reaction flask after completed reactions. Compared to the n Al 2 O 3 ( +) and n MgO supported catalys ts, the product yields and Pd surface areas vary more for the n TiO 2 supported catalysts, which indicates these catalysts are not stable and/or are very sensitive to preparation conditions. To further probe the differences between the SQ and CP PdO/ZrO 2 /n TiO 2 catalysts, they were subjected to XRD analysis ( Figure 4 6 ). A significantly higher contribution from the anatase phase is observed on the SQ PdO/ZrO 2 /n TiO 2 catalyst indicating the presence of larger anatase crystallites on this catalyst. The XRD spe ctra obtained from the CP PdO/ZrO 2 /n TiO 2 and the PdO/n TiO 2 catalysts are very similar. While the two heat treatments could be the sole cause of the higher anatase contribution on the SQ PdO/ZrO 2 /n TiO 2 catalyst, the addition of Zr 4+ to TiO 2 possibly obst ructs rutile phase formation and results in more anatase phase  As the anatase phase has been shown to result in more favorable Pd support interactions compared to other crystal phases or amorphous TiO 2  the higher yield obtained from the SQ PdO/ ZrO 2 /n TiO 2 catalyst is attributed to the increased anatase phase. In fact, the
94 sequential precipitation method for all the n TiO 2 supported catalysts is likely more active due to the increased anatase phase. As the activities of these catalysts are affect ed by the n TiO 2 support crystal structure, and the amount of crystalline anatase phase is dependent on the calcinations, the n TiO 2 supported catalysts are likely very sensitive to pretreatment conditions, such as calcination temperature and time. Since t wo calcinations at 350 C result in more anatase phase, this indicates that the n TiO 2 support is still changing after 3 h at this temperature. This can explain the reproducibility issues for these catalysts. Furthermore, the added ZrO 2 on these catalysts likely disrupts the favorable Pd TiO 2 interactions resulting in an unstable catalyst where palladium leaches into the reaction solution and causes a reduction in yield despite the high palladium surface areas of the PdO/ZrO 2 /n TiO 2 catalysts. In contrast t o the n Al 2 O 3 ( +) supported catalysts, the sequential preparation is the preferred method for the n TiO 2 supported catalysts. This is likely due to an increased contribution from the anatase TiO 2 phase with calcination thus increasing favorable PdO TiO 2 in teractions. However, the addition of another oxide to the PdO/n TiO 2 catalysts appears to disrupt the favorable Pd TiO 2 interactions and leads to increased Pd leaching in several cases. Only ZrO 2 results in a significant increase in the yield obtained from n TiO 2 supported PdO catalysts, if the catalyst is prepared using the sequential precipitation method. The effect of ZrO 2 is mainly to increase the Pd surface area, as the turnover numbers for the PdO/n TiO 2 and PdO/ZrO 2 /n TiO 2 catalysts are very similar. 4.5 Summary It is evident that there is no correlation between the measured Pd surface area and the catalytic activity over these catalysts. As the measured Pd particle size is
95 reasonably consistent with the PdO particle sizes in the previous TEM results [70,71] it seems likely that the reaction is structure sensitive. However, migration of oxides to cover the active Pd species as well as palladium leaching also contribute to the lack of correlation with measured Pd surface area. The optimal preparation m ethod and the best additive to use for PdO catalysts in the oxidative coupling of 4 methylpyridine is dependent on the support used. In general, the co precipitation method is favored for the n Al 2 O 3 ( +) support, as the first calcination treatment removes support hydroxyl groups which are important to give a high Pd dispersion. The sequential precipitation yields better results for the n TiO 2 support, since the heat treatment induces crystal growth of the anatase phase, which in turn leads to favorable PdO support interactions. The n MgO support is not as sensitive to heat treatment as the n Al 2 O 3 ( +) support, so the best preparation method depends on the additive. There are complex interactions between the PdO, the added oxide, and the nanoparticle oxide su pport in these catalysts. If the PdO additive oxide interactions are too strong, they undermine the PdO support interactions and lead to Pd leaching. If the PdO support interactions are stronger, addition of another oxide is ineffective or reduces the PdO support interactions leading to Pd leaching and a lower activity. Therefore, only a few PdO/MeO x /n support catalysts are more active than the corresponding PdO/n support catalyst. As an additive, ZnO generally decreases the activity of the catalysts. This is accredited to strong ZnO PdO interactions that either block active PdO sites or undermine the PdO support interactions, or to strong ZnO support interactions leading
96 to mixed oxide formation, which reduces favorable ZnO PdO and/or PdO support interactio ns. CeO 2 increases the Pd surface area of all catalysts and does promote the reaction in some cases. However, the yields obtained from CeO 2 containing catalysts vary significantly. Strong PdO CeO 2 interactions resulting in disrupted PdO support interaction s and leaching of palladium and/or CeO x migration to cover active PdO species on the surface are identified as potential reasons for the reproducibility issues observed on these catalysts. Addition of ZrO 2 also increases the Pd surface area on most catalys ts. While ZrO 2 is a true promoter for the PdO/n Al 2 O 3 ( +) catalyst, i.e. not only the yield but also the turnover number increase, this is not the case for the other supports. The yields obtained from the CP PdO/ZrO 2 /n Al 2 O 3 ( +) catalyst are consistently 3 6% higher at 3.4 0.1 g product per g catalyst than the previously reported best yield in this reaction (2.5 g/g catalyst).
97 Table 4 1 Catalyst support properties. Support o xide NanoScale p roduct a SA [ m 2 /g ] b SA [m 2 /g] a fter c alcination c NH 3 S cm 3 /g d CO 2 S cm 3 /g e n Al 2 O 3 ( +) NanoActive a luminum o xide p lus 695 515 f 9.0 0.75 n MgO NanoActive m agnesium oxide p lus 685 616 3.4 2.0 n TiO 2 NanoActive t itanium d ioxide 505 133 4.0 0.6 a Nanoparticles purchased from NanoScale corporation: http://www.n anoscalecorp.com/content.php/chemicals/powders/ accessed on 11/19/2010. b BET surface area after drying at 105 C for 3 hours. c BET surface area after calcination at 350 C for 3 hours. d Adsorbed amount of NH 3 on support. Measured by pulse titr ation after outgassing in flowing nitrogen at 105 C for one hour [ 44 ]. e Adsorbed amount of CO 2 on support. Measured by pulse titration after outgassing in flowing ni trogen at 105 C for one hour [44 ]. f Surface area after calci nation at 350 C for 24 hours.
98 Table 4 2 Catalytic activities of PdO/MeO x /support catalysts. Entry Preparation m ethod a Support Additive Yield [g product/ g catalyst] Yield [g product / g Pd] 1 b PT n Al 2 O 3 ( +) 2.5 50 2 PT c n Al 2 O 3 ( +) 1.8 36 3 CP n Al 2 O 3 ( +) ZnO 2.4 47 4 SQ n Al 2 O 3 ( +) ZnO 1.6 30 5 CP n Al 2 O 3 ( +) CeO 2 2.5 51 6 SQ n Al 2 O 3 ( +) CeO 2 1.4 26 7 CP n Al 2 O 3 ( +) ZrO 2 3.4 68 8 SQ n Al 2 O 3 ( +) ZrO 2 2.5 49 9 SQ d n Al 2 O 3 ( +) ZrO 2 3.3 66 10 b PT n MgO 2.3 46 11 PT c n MgO 1.7 32 12 CP n MgO ZnO 1.4 27 13 SQ n MgO ZnO 0.9 18 14 CP n MgO CeO 2 2.2 43 15 SQ n MgO CeO 2 2.7 53 16 SQ d n MgO CeO 2 2.6 52 17 CP n MgO ZrO 2 2.1 43 18 SQ n MgO ZrO 2 1.8 37 19 b PT n TiO 2 1.6 31 20 PT c n TiO 2 1.5 31 21 CP n TiO 2 ZnO 1.1 23 22 SQ n TiO 2 ZnO 1.2 24 23 CP n TiO 2 CeO 2 1.2 25 24 SQ n TiO 2 CeO 2 1.6 32 25 CP n TiO 2 ZrO 2 0.6 13 26 SQ n TiO 2 ZrO 2 1.9 37 27 SQ d n TiO 2 ZrO 2 2.0 41 a Preparation methods: PT: precipitation, CP: co precipitation, SQ: s equential precipitation. b Results from previous work [44 ]. c Catalyst prepared using a precalcined support. d Catalyst prepared using a modified sequential precipitation, where the catalyst is simply filtered (not calcined) before the second deposi tion.
99 Table 4 3 Catalyst surface areas, CO adsorption, palladium dispersions and turnover numbers of PdO/MeO x /support catalysts Catalyst [5/5/90%] a Deposition m ethod b SA [ m 2 /g ] CO ads. [ cat ] Dispersion [%] Pd SA [ m 2 /g ] Turn over # c PdO/n Al 2 O 3 ( +) d PT 180 4600 44 9.7 66 PdO/n Al 2 O 3 ( +) PT e 215 1800 17 3.8 120 PdO/ZnO/n Al 2 O 3 ( +) CP 245 4490 43 9.5 64 PdO/ZnO/n Al 2 O 3 ( +) SQ 235 2800 27 5.9 66 PdO/CeO 2 /n Al 2 O 3 ( +) CP 220 538 0 51 11.4 57 PdO/CeO 2 /n Al 2 O 3 ( +) SQ 250 2960 28 6.3 54 PdO/ZrO 2 /n Al 2 O 3 ( +) CP 220 4830 46 10.2 85 PdO/ZrO 2 /n Al 2 O 3 ( +) SQ 215 1730 16 3.7 172 PdO/ZrO 2 /n Al 2 O 3 ( +) SQ f 214 3940 37 8.3 103 PdO/n MgO d PT 85 2350 22 5.0 119 PdO/n MgO PT e 101 2490 24 5.3 78 PdO/ZnO/n MgO CP 140 780 7 1.7 213 PdO/ZnO/n MgO SQ 50 1520 14 3.2 70 PdO/CeO 2 /n MgO CP 64 2560 24 5.4 101 PdO/CeO 2 /n MgO SQ 120 4190 40 8.9 77 PdO/CeO 2 /n MgO SQ f 49 2290 22 4.8 137 PdO/ZrO 2 /n MgO CP 165 3740 35 7.9 70 PdO/ZrO 2 /n Mg O SQ 125 3200 30 6.8 71 PdO/n TiO 2 d PT 210 4150 39 8.8 46 PdO/n TiO 2 PT e 110 1720 16 3.6 110 PdO/ZnO/n TiO 2 CP 230 2870 27 6.1 48 PdO/ZnO/n TiO 2 SQ 190 4620 44 9.8 31 PdO/CeO 2 /n TiO 2 CP 230 5050 48 10.7 30 PdO/CeO 2 /n TiO 2 SQ 170 5030 48 10. 7 39 PdO/ZrO 2 /n TiO 2 CP 120 4860 46 10.3 16 PdO/ZrO 2 /n TiO 2 SQ 110 5160 49 10.9 44 PdO/ZrO 2 /n TiO 2 SQ f 186 5780 55 12.2 43 a Weight percent of different components on a metal basis. b Preparation methods: PT: precipitation, CP: co precipitation, S Q: sequential precipitation, SQ filt.: modified sequential precipitation, where the catalyst is simply filtered (not calcined) before the second deposition. c Turnover number: number of product molecules formed per palladium surface atom. d Results f rom previ ous work [44 ]. e Catalyst prepared using a precalcined support. f Catalyst prepared using a modified sequential precipitation, where the catalyst is simply filtered (not calcined) before the second deposition.
100 Figure 4 1 Oxidative coupl ing of 4 methylpyridine over PdO/MeO x /nano oxide catalysts. MeOx: ZnO CeO 2 or ZrO 2 Nano oxide: Al 2 O 3 ( +), n MgO or n TiO 2
101 Figure 4 2 Activity data from prepared catalysts. Nn = PdO/support only, no additive oxide. Additive oxide: Zn = ZnO, Ce = CeO 2 Zr = ZrO 2 Filled symbols are results from the co precipitation method (Zn, Ce or Zr), as well as the precipitated reference PdO/support catalysts (Nn) and open symbols are results from the sequential precipitation method (Zn, Ce or Zr), or for PdO/supp ort catalysts (no additive) prepared using a precalcined support (Nn). The open grey symbols are results from the modified sequential precipitation method (filtration only, no calcination).
102 Figure 4 3 Thermogravimetric analysis (TGA) data obtained fro m n TiO 2 (red), n Al 2 O 3 ( +) (black) and n MgO (blue) supports. Dashed lines mark the drying temperature (105 C) and calcination temperature (350 C).
103 Figure 4 4 X ray diffraction (XRD) data obtained from PdO/ZnO/n Al 2 O 3 ( +) catalysts prepared using t he co precipitation (CP) and the sequential precipitation (SQ) methods, and PdO/n Al 2 O 3 ( +) catalyst with no added ZnO (Nn).
104 Figure 4 5 X ray diffraction (XRD) data obtained from PdO/ZnO/n MgO catalysts prepared using the co precipitation (CP) and the sequential precipitation (SQ) methods, and PdO/n MgO catalyst with no added ZnO (Nn).
105 Figure 4 6 X ray diffraction (XRD) data obtained from PdO/ZrO 2 /n TiO 2 catalysts prepared using the co precipitation (CP) and the sequential precipitation (SQ) metho ds, PdO/n TiO 2 catalyst with no added ZrO 2 (Nn), and the n TiO 2 support.
106 CHAPTER 5 REGENERATION OF PALLADIUM OXIDE SUPPORTED NANOPARTICLE ALUMINA FOR THE OXIDATIVE COUPLING OF 4 METHYLPYRIDINE 5.1 Background High cost and scarcity limit the usage of noble metal base d catalysts. Palladium catalyzed oxidative coupling of bipyri dines has been shown to be a simple one step, environmentally friendly process [ 44,52,57,6 4 ,70,71 ]. Bipyridines coordinate well to transition metal cations forming complexes with distinct photoch emical and catalytic properties [ 40,53,59,60 ]. Bi p yridine ruthenium complexes are appealing for applications as organic light emitting diodes [ 40 ] and chemiluminescence detection systems [ 40,44,6 4 ]. However, large scale usage is partially limited by the co st associated with dimethyl bipyridine exceeds $7,600 per kilogram [ 46 ]. Therefore, finding a more economical synthesis pathway for bipyridines is desirable. Environmental impacts are also an impor tant consideration. tion decreases the process cost and improv es the reaction economic s Palladium catalyst deactivation and regeneration has been studied in detail [ 100 103 105 ] Coke deposition is th e most common reason for deactivation where a carbonaceous species eventually deactivating the catalyst [ 100 104 ]. Washing the catalyst and burn ing o ff the carbonaceous species generally lead to the regeneration of the initial catalytic activity. However, other deactivation mechanism s are also common [ 100 105 ]. Thermal degradation changes the particles size typically through crystallite growth due to sintering. While sintering is very common at hi gh temperatures, particularly under reducing conditions, p alladium oxide is fairly stable with minimal sintering under mild
107 temperature and high partial pressures of oxygen. Support modifications can also lead to deactivation and encompass a variety of eff ects such as loss of surface area, changes in composition, and/or changes in particle shape or porosity  Additionally, there can be alterations at the palladium surface because of changes in the valence state  Different process types, batch or flow, are susceptible to different types of deactivation. Batch processes suffer more deactivation due to localized excess of poisons on the catalyst [ 102 104 ]. In our previous research, poisoning from product build up on the surface and the reduction of P dO were identified as likely causes for deactivation [ 52,57 ] In this study, the effect of temperature and carrier gas were investigated to regenerate the PdO/n Al 2 O 3 (+) catalyst to determine if these catalysts can be regenerated to their original activity and also determine the optimal preparation conditions i.e. atmosphere and temperature. 5.2 Experimental 5.2.1 Catalyst P reparation The catalysts were prepared using a commercially available nanoparticle aluminum oxide (n Al 2 O 3 (+), NanoScale Corporation) to deposit palladium via precipitation. In this method, an aqueous solution of the palladium nitrate (Pd(NO 3 ) 2 2H 2 O, Sigma Aldrich) was dispersed into an aqueous solution of the alumina support. The palladium amount was 5% by weight on a metal basis unless otherwis e stated. The mixture was titrat ed with sodium hydroxide (NaOH, Sigma Aldrich) to form palladium hydroxide(s) on the support [69 ]. The amount of NaOH corresponds to a 50% stoichiometric excess, based on the amount of palladium nitrate used. The resulting m ixture was aged overnight at room temperature before being filtered. The filtered
108 material was rinsed in deionized water overnight and filtered again. The material was then dried at 105 C overnight and calcined in air at 350 C for 3 h. 5.2.2 Reaction C ondition s The reactant, 4 methylpyridine (Acros), was doubly distilled over KOH prior to use. The reaction contained a 1:10 ratio by mass of catalyst to reactant, distilled 4 methylpyridine, placed in a round bottom flask. The reaction mixture was evacuated befor e introducing an oxygen atmosphere. The mixture was then heated to the boiling point (145 C) under continuous agitation. After refluxing for 72 hours, the flask contents were filtered using a glass micro fiber filter and washed with chloroform to dissolve the product. The product was obtained by removing the chloroform, water, and unreacted 4 methylpyridine using a rotary evaporator. 5.2.3 R egeneration C onditions After separating the spent catalyst from product and reactants, the spent catalyst wa s drie d under a tmospheric conditions to allow chloroform and remaining solvents to evaporate. The spent catalyst wa s then treated in static air, with an inert helium flow, or with an oxygen flow. When using air, the spent catalyst wa s placed in a crucible and heated in a muffle furnace for 2 h. The other two methods involve d a treatment under a gas flow of either helium or oxygen For the flow system, the spent catalys t wa s packed n AliCat mass flow c ontroller. The packed catalyst wa s heated in Lindberg/Blue tube furnace and an Omega K type thermocouple was used to monitor the temperature under 50 standard cm 3 /min ( sccm ) of gas flow for 2 h. Each catalyst was labeled according to regeneration conditio ns i.e. Air 350C 2 for the second treatment under static air in the oven at 350 C.
109 5.2.4 Catalyst C haracterization The as received support surface areas, the surface areas of the prepared catalysts and the treated support surface areas were determined by multi point Brunauer Emmett Teller (BET) isotherms on a Quantachrome Nova 1200 instrument as described in previous work [57 ]. Chemisorption measurements were performed in a Quantachrome ChemBET 3000 instrument and used to characterize the active catalyst surface area. The catalysts were first reduced with 5% hydrogen in nitrogen at 170 C for 1 h, and then outgassed in nitrogen at 170 C for another hour. The mild conditions were used to limit the sintering of the Pd particles on the surface. This was followed by pulse titrating with carbon monoxide to characterize the palladium dispersion of the catalysts. Detailed descriptions of the procedure and calculations are given in previous work [ 70 ]. While PdO is believed to be the active phase, or at least a necessary precursor, the CO chemisorption measurements on reduced catalysts are important since the PdO surface area cannot be measured directly. It wa s assumed that there is a correlation between the original PdO surface area and the Pd surface area of the reduced catalyst. Previous XRD and TEM d ata support ed this assumption [70,71 ]. ray diffractometer using Bragg Brentano geometry with Cu K = 1.54 ). Selected supports and catalyst powders were sec ured onto a glass slide with double sided sticky tape prior to the measurements. XPS spectra were collected using a Mg K X ray source (PHI model 04 151) and a double pass cylindrical mirror analyzer (PHI model 25 270 AR) on selected fresh catalyst s The c atalyst powders were pressed into aluminum cups prior to insertion into
110 the ultra high vacuum (UHV) chamber (base pressure 1 x 10 10 Torr). Spectra were taken in retarding mode with a pass energy of 50 eV for survey spectra and 25 eV for high resolution sp ectra Data were collected using a computer interface and then digitally smoothed. A value for the C 1s binding energy of 284.6 eV was assigned to correct for static charging and work function differences [ 72 ]. Diffuse reflectance infrared transform spectr oscopy (DRIFTS) was gathered on a Thermofisher Scientific LLC Nicolet 6700 spectrometer equipped with a MCT detector cooled by liquid N 2 Powdered catalyst samples were mixed with KBr in a 1:10 ratio by mass and placed into the cell. Spectra were collected at a spectra l resolution of 4 cm 1 accumulating 256 scans per spectrum. 5.3 Results and Discussion Three deactivation pathways for this reaction ha ve been identified in previous research; 1) reduction of active PdO species to Pd metal, 2) poisoning of active sites due to product formed or product decomposition, and 3) palladium lea ching into the solution [5 2 ,57 ]. Of these three possible pathways, the third, palladium leaching into the solution, is the worst as it would not be possible to regenerate a catalyst to its original activity if palladium is removed from the surface. However, when leaching is a severe issue, inactive palladium species are deposited as a palladium mirror onto the reaction flask and this was not observed in any of the experiments in the current study. To investigate which of the remaining two deactivation pathways, reduction of PdO and site blocking by product, is the more limiting, three treatment conditions were investigated, 1) heating in sta tic air, 2) heating under a He flow, and 3) heating under an O 2 flow. Heating in sta tic air is the simplest treatment, but may not be very efficient in removing adsorbed product or reoxidizing the palladium. Heating in a flow of He is likely to
111 remove adsorbed species, but will not cause reoxidatio n of palladium. Heating in an O 2 flow has potential to both remove adsorbed species and reoxidize palladium. The main consideration here is if this can be done at a low enough temperature to prevent other catalyst degradation pathways. For example, sinteri ng of Pd or PdO can occur at higher temperature. The support surface area may also decrease with increasing temperature. 5.3.1 Reaction Data The prepared catalysts and catalysts after the different treatment methods were subjected to catalytic activity measureme nts. The results of these experiments are presented in Table 5 1 The result from the initial reaction on the fresh catalyst has been included in the table for comparison. Each catalyst was exposed to the same treatment method two times to determine if mul tiple regeneration cycles would be possible. A few general trends can be observed in Table 5 1 First, not many of the treatment conditions tested give a yield close to the original yield. It is also evident that the yield from the second regeneration is l ower than those of the first regeneration, which indicates that these catalysts have a finite and rather short life time. Regenerative treatments above 200 C are not recommended as the yields in all cases decrease with increasing temperature above 200 C. Heating a spent catalyst in a flow of He yields a lower product yield compared to the oxidative treatments. 5.3.2 BET Surface Area Measurements The BET surface areas were determined, and CO chemisorption measurements were performed to evaluate the heat treatmen t effects on the overall catalyst surface area and the Pd metal surface area. These results have been added to Table 5 1 In general a small decrease in surface area is observed with the heat treatments. It is interesting to note that the lower treatment t emperatures yields smaller BET surface
1 12 areas compared to the higher treatment temperatures. This is likely an effect of product blocking the pores of the catalysts (as indicated by the lower surface area of the washed and dried catalyst) The overall surfa ce areas are in most cases smaller after the second generation compared with the first generation. This could be due to a slight deterioration of the support due to the heat treatment or due to more product blocking pores. At the higher heat treatment temp eratures, the support effect is more prominent, while at the lower temperatures pore blocking by product is likely more prevalent. 5.3.3 DRIFTS Investigation DRIFTS was performed on the fresh and selected regenerated catalysts to investigate the nature of the sp ecies on the surface and how they vary before and after reaction and afte r the different heat treatment ( Figure 5 1 ) A fresh catalyst saturated or doped with the liquid reactant and dr ied under atmospheric conditions overnight was included for comparison. The doped sample reveals a significant absorption band at 1616 cm 1 that is not present on the fresh undoped catalyst. This absorption band is associated with pyridine on a Lewis acid site together with the band at 1454 cm 1 [ 106 ] The band at 1557 cm 1 i s a pyri dine on a Brnsted acid site [106 ]. After reaction, the catalyst is washed with chloroform to dissolve and recover the product. Before DRIFTS measurement this catalyst was dried in static air at 25 C. Evidently, this treatment does not remove all the product or reactant from the surface (Figure 5 1 A ). The resolution of the DRIFTS measurements is not sufficient to differentiate between the product and the dimethyl bipyridine produ ct probably acts as a poison by blocking the active sites, thus deactivating the catalyst. As the air treatment at 25 C did not remove the adsorbed product, it is not surprising that the yield from this catalyst is lower than the original yield. Furthermo re,
113 the intensity of this feature increases after the second regeneration indicating a further buildup of product, which can explain the lower yield after the second treatment compared to the first treatment By heating the spent catalyst in air at 200 C, some species on the surface appear to have been removed (Figure 5 1 B ). However, the adsorption feature is broader, potentially indicating the presence of decomposition products. The spent catalyst after the first treatment preparation again exhibit s the s harp feature due to the pyridine moiety, but the feature is broader compared to the feature on the washed catalyst. Also, the bands due to the pyridine moiety are larger on the spent catalyst compared with the doped one. As for the heat treatment in stat ic air at 200 C, heating the recovered spent catalyst in inert He flow at 200 C does not remove the species which gives rise to the adsorption feature at 1616 cm 1 (Figure 5 1C ). The feature is a little more intense and broader in the spectrum obtained a fter the He flow compared to that obtained from the washed and dried catalyst, but not as broad as the feature on the catalyst regenerated in air at 200 C. The spent catal yst from the first regeneration ex hibit s several absorption band s in the region betw een 1300 and 1700 cm 1 These are consistent with the pyridine moiety on the doped catalyst, but are more intense and likely indicate a strongly adsorbed product. The second regeneration is less efficient in removing these species compared to the first reg eneration. The DRIFTS spectra obtained after heat treatment in flowing O 2 are very different from the spectra obtained after the He and air treatments at the same temperature s (Figure 5 1 D ). The absorption band at 1616 cm 1 is significantly smaller than af ter any of the other treatments, indicating that this temperature is sufficient to remove most of the
114 species on the surface. The band at 1072 cm 1 due to carbonates and the region for hydroxyl groups, between 3100 and 3500 cm 1 [ 107 ] are also smaller. Th is is consistent with the removal of product or reactant and hydroxyl groups. The spent catalyst after the O 2 regeneration has a significantly smaller contribution from the pyridine moiety compared to the catalysts after the other heat treatments. I n contr ast to the other catalysts, the second regeneration appears to be as effective as the first regeneration in removing the adsorbed species. It is also noteworthy that a significant fraction of hydroxyl groups on the surface area is removed with this treatme nt. 5.3.4 CO Chemisorption Measurements The CO adsorption measurements indicate that the accessible Pd surface area is lower on all the heat treated and the washed catalysts. This is consistent with the DRIFTS observations, as the species on the catalysts are li kely associated with the palladium on the surface. However, leaching of palladium into the solution cannot be excluded. Furthermore, the lower yields observed are not consistent with the apparent decrease in available Pd surface area on most catalysts. The yields are higher than expected from the measured CO chemisorption. Only for the heat treatment done in sta tic air does there appear to be a correlation with the decrease in yield and the loss in Pd surface area, but only above 200 C. Compared to the fir st regeneration treatment, the CO adsorption is lower after the second regeneration over all catalysts, and the apparent loss in Pd surface area is the greatest for the He heat treatments. For the air and He treatments this is consistent with the DRIFTS da ta, which indicate s accumulation of adsorbed species on the surface. During heat treatment in pure O 2 the adsorbed species are oxidized to CO 2 a highly exothermic reaction, which is likely to c reat e localized hot spots
115 surface significa ntly higher than the heat treatment temperature s, and this can cause severe sintering and result in a lower CO adsorption compared to the other heat treated catalysts. 5.3.5 XRD Analysis XRD data was collected from the fresh and selected regenerated catalysts ( Figure 5 2 and Figure 5 3). As has been noted previously, the fresh catalyst displays the active PdO phases at 33.8 and 34.1, while only Pd metal is observed on the washed spent catalyst. As expected, no PdO is observed on the catalyst heat treated in He which explains the lower activity of these catalysts. Of course, XRD is a bulk technique, which would not detect a thin PdO layer on the surface that may be present after air exposure Heat treatment in sta tic air at 200 C is sufficient to result in an XRD visible oxide, but does not fully oxide the Pd metal on the surface as is evident since both Pd phases are present in the XRD pattern obtained from the catalyst after this heat treatment Only at a temperature of 350 C in sta tic air is Pd metal no lon ger visible on the surface. However, t he peak due to PdO is more intense after heat treatment at this temperature compared to 20 0 C heat treatment which indicat es sintering of the PdO particles As for the DRIFTS data, the XRD patterns obtained after hea t treatment in an O 2 flow are very different from those obtained from heat treatment in sta tic air. This is most evident in the significantly higher intensit ies of palladium related peaks. Perhaps the most surprising feature s in the XRD patterns obtained f rom the catalysts heat treated in O 2 at 200 C and 500 C are the sharp peak s due to palladium metal. Evidently, s ignificant sintering of palladium species has occurred over this catalyst. In fact, the sintering is so severe that the Pd after reaction is not fully oxidized even after treatment
116 at 500 C. This is in stark contrast to the air treatment in sta tic air, where 350 C was sufficient to fully oxidize the Pd metal on the surface. This is consistent with the conclusion from the CO adsorption measure ments that hot spots are formed due to an exothermic oxidation reaction of the carbon containing surface species in pure oxygen. 5.3.6 XPS Investigation The XPS survey spectra obtained from the fresh, the washed spent catalyst, and selected catalysts after diffe rent heat treatments are presented in Figure 5 4 thru Figure 5 6 The higher C 1s and lower Pd 3d peak intensities on the washed and dried spent catalyst compared with the fresh catalyst ha ve been reported previously  The surface of these catalysts al so contains some sodium and potassium, which may be residues from the catalyst preparation and the reactant pretreatment. Consistent with the DRIFTS data, some carbon is removed after air treatment at 200 C, but the C 1s peak intensity is still significan t. The high resolution spectra of the Pd 3d peaks reveal that Pd metal with a binding energy of 334.0 eV is the dominant species on the washed spent catalysts. However, the air exposure at 25 C to dry the catalyst is sufficient to cause a small amount of surface oxidation (Figure 5 7 ), as a shoulder is present on the higher binding energy side of the Pd 3d 5/2 peak. Increasing temperature increases the contribution from PdO at a binding energy of 336.1 eV. However, both Pd 3d peaks have shoulders of lower b inding energy, indicating that Pd metal is present underneath the oxide layer. As this was not detected with XRD, the Pd crystals must be small or amorphous. After He treatment at 200 C, the surface carbon content is even higher than on the washed cataly st. The higher carbon content is relate d to the product (or reactant) which is evident in the higher intensity of the nitrogen peak in the survey spectrum
117 obtained from this catalyst. Heat treatment in He at 350 C appears to remove most of the nitrogen co ntaining species on the surface E ven though the carbon peak has a lower intensity compared to the catalyst heat treated in He at 200 C i t is still higher than on the washed spent catalyst. The surface of the catalyst treated in He consists almost exclus ively of Pd metal. Consistent with the DRIFTS data, the carbon content is lower on the catalysts heat treated in O 2 compared to the catalysts treated in air or He, although the carbon content appears to be a little higher than on the fresh catalyst. From t he high resolution scans of the Pd 3d peaks, it is evident that 200 C is not sufficient to fully oxidize the Pd on the surface. A significant shoulder is present at lower binding energies compared to the main PdO related peak at 336.1 eV. After heat treat ment in O 2 at 350 C, PdO is the main component according to XPS. However, the Pd 3d 5/2 peak is broad and a shoulder at binding energies higher than 336.1 eV is evident. As electron deficient Pd(II) species have been shown to be favorable in this reaction [52,57 ], this could be the reason why this regeneration is so effective and the original yield is achieved over this catalyst. However, the sintering that occurred during the heat treatment, as evidenced by XRD, caused significant unrecoverable damage to the catalyst explaining why the second generation is ineffective. 5.4 Summary S everal factors contribut e to the deactivation of PdO/n Al 2 O 3 (+) catalyst s Both XRD and XPS show the active PdO on the catalyst surface ha s been reduced to Pd metal after reacti on Exposure to an oxygen atmosphere oxidize s the Pd metal even in the washed spent catalyst which was allowed to dry in air at room temperature. However, the batch process system facilitates poisons such as product or reactants to
118 build up on the cataly i s not sufficient to remove all of the product s from the surface and completely restore the activity. A heat treatment temperature at or above 200 C in oxygen is needed to remove the carbon with a flow feed being the most efficient. Al thoug h a PdO phase is present and the carbon is removed, sintering of the Pd particles prevents complete regeneratio n. This is especially noticed for the second oxygen treatments at or above 200 C as localized hot spots from the carbon removal causes a significant decline in the Pd SA from the fresh and first oxygen heat treat ed catalysts. The best heat treatment condition i s in static air at 200 C which yield ed 50 g / g Pd and 46 g /g Pd for the first and second treatments, respectively. Although the tre atments restore almost all the activity, the c atalyst has a limited life time The inability to restore the active PdO phase to its original condition results in a continuous diminishing yield for each subsequent reaction.
119 Table 5 1 Fresh and regenerated PdO/n Al 2 O 3 (+) yields. Regeneration #1 Regeneration #2 Preparation a Yield [g product/g Pd] SA b [m 2 /g] CO ads. Pd SA [m 2 /g] Yield [g product/g Pd] SA b [m 2 /g] CO ads. Pd SA [m 2 /g] Fresh 50 c 180 c 3030 6.7 Washed d 41 140 1000 2.1 37 100 580 1.2 105 Air 26 100 800 1.7 42 125 560 1.2 200 Air 50 115 775 1.6 46 100 510 1.1 350 Air 47 160 730 1.5 43 115 580 1.2 500 Air 36 160 660 1.4 21 145 320 0.7 50 He 39 130 675 1.4 40 125 410 0.9 105 He 34 110 580 1.2 34 90 285 0.6 200 He 30 120 545 1.2 25 105 275 0.6 350 He 32 150 800 1.7 5 120 350 0.7 500 He 13 150 1050 2.2 2 100 790 1.7 50 O 2 47 130 650 1.4 40 140 525 1.1 105 O 2 44 90 765 1.6 37 95 605 1.3 200 O 2 40 160 960 2.0 16 115 280 0.6 350 O 2 50 150 840 1.8 23 100 380 0.8 500 O 2 35 120 835 1.8 20 110 390 0.8 a 5wt% Pd on a metal basis b BET surface area of the prepared catalyst c Results from previous work [71 ] d Results are of regenerated catalyst where support was only washed and solvents were allowed to dried unde r atmospheric conditions.
120 Figure 5 1 DRIFTS data obtained from the PdO/n Al 2 O 3 (+) rege ne rated catalysts.
121 Figure 5 2 XRD data obtained from the PdO/n Al 2 O 3 (+) regenerated catalysts.
122 Figure 5 3 XRD narrow scan data obtained from the PdO/n Al 2 O 3 (+) regenerated catalysts.
123 Figure 5 4 XPS survey spectra obtained from the PdO/n Al 2 O 3 (+) static air regenerated catalysts.
124 Figure 5 5 XPS survey spectra obtained from the PdO/n Al 2 O 3 (+) He regenerated catalysts.
125 Figure 5 6 XPS survey s pectra obtained from the PdO/n Al 2 O 3 (+) O 2 regenerated catalysts.
126 Figure 5 7 XPS Pd 3d spectra obtained from the PdO/n Al 2 O 3 (+) static air rege ne rated catalysts.
127 Figure 5 8 XPS Pd 3d spectra obtained from the PdO/n Al 2 O 3 (+) He rege ne rated cata lysts.
128 Figure 5 9 XPS Pd 3d spectra obtained from the PdO/n Al 2 O 3 (+) O 2 rege ne rated catalysts.
129 CHAPTER 6 THE INFLUENCE OF PALLADIUM OXIDE LOADING FOR THE OXIDATIVE COUPLING OF 4 METHYLPYRIDINE 6.1 Background Bipyridines are valuable precursor chemical building blocks and are often used in the creation of herbicides and heart treatment medications. The 2, 2 bipyridine coordinates well to transition metal cations forming complexes with distinct photochemical and catalytic properties [40,53,59,60]. Bipryidine ruth enium complexes are have applications as organic light emitting diodes  and chemiluminescence detection systems [40,44,65]. The cost associated with bipyridines limits their usage on dimethyl bip yridine exceeds $7,600 per kilogram  with smaller quantities selling for $15 44 per gram depending on the quantitiy (1 50 g) and purity . However, the oxidative coupling of 4 methylpyridines with palladium has been shown to be a simple one step en vironmentally friendly process [44,52,57,65,70,71]. However, the cost and scarcity of noble metal based catalysts limit the usage of palladium. Finding a more economically efficient synthesis for bipyridines by optimizing the loading amounts of the active metal (PdO) and additive (Zr O 2 ) to obtain a highly active catalyst is desirable. The major cost of the palladium catalyst is the high price of the palladium precursor, and therefore, optimizing the yield per gram of palladium is important. Previous researc h has shown that 5% and 10% palladium loading on n Al 2 O 3 (+) gives approximately the same yield per gram of catalyst despite significantly different Pd surface areas per gram of catalyst . It has also been shown that the measured Pd surface areas of va rious catalysts in the oxidative coupling of 4 methylpyridine do not exhibit a strong correlation with the catalytic activity . It could be expected that lower
130 Pd loadings potentially generate similar yields as with the 5% and 10% loadings. In order to improve the cost efficiency, it is important to optimize the yield per unit weight of the palladium and to investigate the influence of other supports and additives. Therefore, nanoparticle alumina [n Al 2 O 3 (+)], porous titania (p TiO 2 ), and nanoparticle titania treated at 450 C (n TiO 2 450C) supports and a zirconium (ZrO 2 ) additive were included in this study. These supports and the ZrO 2 additive were selected as previous studies have shown high product yields per unit mass of palladium. The effects of palladium loading on the product yields for the PdO/n Al 2 O 3 (+), PdO/p TiO 2 and PdO/n TiO 2 450C catalysts were investigated. The influence of zirconium loading on the best catalyst PdO/ZrO 2 /n Al 2 O 3 (+) to date was also examined to determine if the product yield could be further improved. The objective of this study was to optimize the PdO and ZrO 2 loadings to give the maximum yield with the minimum amounts of active metal, and thus, reduce the cost and improve the economics for production. 6.2 Experimental 6.2.1 Cat alyst P reparation The catalysts were prepared using a commercially available nanoparticle aluminum oxide (n Al 2 O 3 (+), NanoScale Corporation), nanoparticle titanium dioxide (n TiO 2 TiO 2 Alfa Aes ar) supports Palladium was deposit ed onto these supports via precipitation. In this method, an aqueous solution of palladium nitrate (Pd(NO 3 ) 2 2H 2 O, Sigma Aldrich) was added to an aqueous dispersion of the desired support. For co precipitation of the act ive metal and additive, zirconium nitrate (ZrO(NO 3 ) 2 2H 2 O, Sigma Aldrich), was dis solved along with the palladium nitrate in the aqueous solution of the support. Each catalyst was
131 labeled according to the lo ading and support i.e. 5 Pd 5 Zr Al 2 O 3 CP for a 5wt% palladium, 5wt% zirconium additive on n Al 2 O 3 (+) support and CP for co precipitated. The mixture was titrated with 50% stoichiometric excess of sodium hydroxide (NaOH, Sigma Aldrich) to form metal hydroxide(s) on the support . The resulting mixture w as aged overnight at room temperature before being filtered. The filtered material was rinsed with deionized water overnight before being filtered again. The material was then dried at 105 C overnight and calcined in a muffle furnace at 350 C for 3 h. 6.2.2 Re action C onditions The reactant, 4 methylpyridine (Acros), was doubly distilled over KOH prior to use. In a typical reaction the catalyst and reactant, distilled 4 methylpyridine, were placed in a round bottom flask in a 1:10 mass ratio The reaction mixtu re was evacuated followed by the introduction of an oxygen atmosphere. The mixture was then heated to the boiling point (145 C) under continuous agitation. After refluxing for 72 hours, the flask contents were filtered using a glass micro fiber filter and washed with chloroform to dissolve the product. The product was obtained by removing the chloroform, water, and unreacted 4 methylpyridine using a rotary evaporator. 6.2.3 Catalyst C haracterization Chemisorption measurements were performed in a Quantachrome Che mBET 3000 instrument and used to characterize the active catalyst surface area. The catalysts were first reduced with 5% hydrogen in nitrogen at 170 C for 1 h, and then outgassed in nitrogen at 170 C for another hour. The mild conditions were used to lim it the sintering of the Pd particles on the surface. This was followed by pulse titrating carbon monoxide to determine the palladium surface area and dispersion of the catalysts. Detailed descriptions of the procedure and calculations are given in previous work .
132 While PdO is believed to be the active phase, or at least a necessary precursor, the CO chemisorption measurements on reduced catalysts are important since the PdO surface area cannot be measured directly. It is assumed that there is a correlat ion between the original PdO surface area and the Pd surface area of the reduced catalyst. Previous XRD and TEM data support this assumption [70,71 ]. 6.3 Results and Discussion Four different palladium loading amounts 1, 2.5, 5, and 10% were chosen to test the activity of catalysts supported on n Al 2 O 3 (+), p TiO 2 and n TiO 2 450 C. Since the 5/5/90 PdO/ ZrO 2 /n Al 2 O 3 (+) is the best preforming catalysts to date, the zirconium loading amounts at 1, 2.5, 5, and 10% were test ed for activity on a 2.5 and 5% palladiu m on n Al 2 O 3 (+). 6.3.1 PdO loading on N anoparticle A lumina The reaction results and Pd dispersions obtained over the PdO/n Al 2 O 3 (+) catalyst with p alladium loadings of 1, 2.5, 5, and 10% are presented in Table 6 1. As expected, the product yield per gram of ca talyst decreases with increasing palladium loading. However, while the palladium loading decreased by 50%, from a 10% to 5% Pd loading and from a 5% to 2.5% Pd loading the yield per unit mass of catalyst only decreased by 24 28%. Thus, the yield per uni t weight of palladium increased with a decrease in loading. In fact, reducing the palladium loading by 50% (from 10% to 5% as well as from 5% to 2.5%) increased the product yield per unit mass of palladium by approximately 50% (Table 6 1). This suggests a palladium surface area effect, and in fact, for the 10PdAl 2 O 3 and 5PdAl 2 O 3 catalysts, the turnover number (TON) is almost the same. Interestingly, the catalyst with 1% palladium loading gives almost the same yield of product per unit mass of catalyst as th e 2.5PdAl 2 O 3 catalyst. Therefore, the yield
133 per unit weight of palladium for the 1PdAl 2 O 3 catalyst is more than 100% the yield obtained for the 2.5PdAl 2 O 3 catalyst. The turnover number for the 1PdAl 2 O 3 catalyst is therefore significantly higher than the TO Ns for 5PdAl 2 O 3 and 10PdAl 2 O 3 catalysts, indicating the palladium on the surface of the 1PdAl 2 O 3 catalyst is more active. 6.3.2 ZrO 2 L oading on PdO S upported on N anoparticle A lumina Zirconia additive loadings were performed on bot h the 2.5 PdAl 2 O 3 and 5Pd Al 2 O 3 c atalyst The reaction results and Pd dispersions for these catalysts are presented in Tabl e 6 1. Addition of the ZrO 2 to the 2.5PdAl 2 O 3 catalyst only slightly increased the product yield from 1.8 to 1.9 2.2 g/g catalyst. No trend with ZrO 2 loading was ob served over the 2.5PdAl 2 O 3 catalysts. The effect of adding ZrO 2 to 5PdAl 2 O 3 catalyst is more pronounced. Increasing the Zr loading from 0% to 5% increased the yield obtained over the 5PdAl 2 O 3 catalyst from 2.5 to 3.4 g/g catalyst. The CO adsorption measure ments indicate that this is due to an increase in Pd surface area with Zr loading. However, the increase in yield over the 5Pd5ZrAl 2 O 3 compared to the 5PdAl 2 O 3 catalyst is not as high as would be expected from the CO adsorption measurements. The TON is onl y 85 mol product per mol surface Pd over 5Pd5ZrAl 2 O 3 while the TON for the 5PdAl 2 O 3 5Pd1ZrAl 2 O 3 and 5Pd2.5ZrAl 2 O 3 catalysts are all between 104 mol product per mol surface Pd and 109 mol product per mol surface Pd. Therefore, it is possible the higher CO adsorption measured on the 5Pd5ZrAl 2 O 3 catalyst compared with the 5PdAl 2 O 3 catalyst is due to some contributions from the ZrO 2 The ZrO 2 may be partially reduced on these catalysts after the reductive pretreatment, and can thus result in an overestimation in the Pd surface area due to some CO adsorption on the reduced ZrO 2 Increasing the Zr content to 10% is detrimental as the yield obtained over the 5Pd10ZrAl 2 O 3 is slightly lower than on the 5PdAl 2 O 3 catalyst. According to the CO
134 adsorption measurements, the Pd surface areas are the same on 5Pd5ZrAl 2 O 3 and 5Pd10ZrAl 2 O 3 catalysts. This could be a combination of two effects: Zr covering some of the active palladium sites and an increased CO adsorption from the reduced ZrO 2 on the 5Pd10ZrAl 2 O 3 compared with the 5Pd5ZrAl 2 O 3 catalyst. While the palladium on the surface does not appear to be significantly more active after addition of the ZrO 2 it appears that ZrO 2 increases the Pd dispersion, which is beneficial for these catalysts as it increases the yield per unit weight of palladium. 6.3.3 PdO L oading on T itania Palladium loading of 1, 2.5, 5, and 10wt% of were deposited onto a p TiO 2 support and a n TiO 2 support the latter pretreated at 450 C prior to palladium deposition. The reaction results and Pd dispersions have been added to Table 6 1. With the exception of the 10PdpTiO 2 the trends with palladium loading for the PdO/p TiO 2 catalysts are similar to those observed on the PdO/n Al 2 O 3 (+) catalysts More specifically, the yields per unit weight of catalyst dec reased while the yield per unit weight of palladium increased in going from 5% to 1% Pd loading. As expected, t he Pd dispersions increase, i.e. larger fractions of the Pd atoms are present at the surface, with a decrease in Pd loading. While the dispersion s on the PdO/n TiO 2 450C catalysts are very similar to those on the PdO/p TiO 2 catalysts, the product yields are lower on the PdO/n TiO 2 450 C catalysts. In comparison to the PdO/n Al 2 O 3 (+) catalysts, the 5% Pd loading on the PdO/p TiO 2 catalysts results in the highest yield per unit weight of catalyst while the 1% loading gives the highest yield per unit mass of palladium. 6.4 Summary For the same amount of catalyst, the 5% Pd loading appears to be optimum on all the catalysts under investigation. Howeve r, the yield per unit weight of palladium is
135 the highest on 1% Pd loading on both PdO/n Al 2 O 3 (+) and PdO/p TiO 2 catalysts. Addition of ZrO 2 appear s to increase the Pd surface area (and thus the dispersion) of the catalyst, and in most cases this result s i n an increased product yield. However, there is an optimum ZrO 2 loading, above which partial coverage of palladium may occur. It is possible that CO adsoption is overestimated on the ZrO 2 containing catalysts, and if this is the case, then the TONs are low er than expected. Therefore, the data in the current study does not unambiguously reveal whether the ZrO 2 containing catalysts are more active due to a higher Pd (and thus also PdO) surface area, or as a result of more active PdO species. Nevertheless, the CO adsorption measurements are important as they do reveal the benefits of adding ZrO 2 to the catalysts.
136 Table 6 1 CO adsorption, palladium dispersions, yield, and turnover numbers of prepared loading series catalysts. Catalyst n ame Yield [g product/g cat] Yield [g product/g Pd] CO adsorbed Dispersion [%] TON a 1PdAl 2 O 3 1.6 156 1100 52 173 2.5PdAl 2 O 3 1.9 74 1300 25 172 5PdAl 2 O 3 2.5 b 50 b 3 14 0 30 10 5 10PdAl 2 O 3 3.3 33 4170 20 96 2.5Pd 1ZrAl 2 O 3 CP 2.2 88 2000 38 133 2.5 Pd2.5 ZrAl 2 O 3 CP 1.9 73 2280 43 98 2.5Pd 5ZrAl 2 O 3 CP 1.9 7 4 2410 46 96 2.5Pd 10ZrAl 2 O 3 CP 2.1 82 2680 51 93 5Pd 1ZrAl 2 O 3 CP 2.7 53 3100 29 104 5Pd 2.5ZrAl 2 O 3 CP 3.0 59 3300 31 109 5Pd5Zr Al 2 O 3 CP c 3.4 68 4830 46 85 5Pd 10ZrAl 2 O 3 CP 2.4 48 4800 46 61 1 PdpTiO 2 1.0 98 1130 54 105 2.5PdpTiO 2 2.1 80 1810 34 134 5PdpTiO 2 2.6 b 52 b 1450 14 209 10PdpTiO 2 1.8 18 2220 11 97 1Pd n TiO 2 450C 0.3 25 925 44 32 2.5Pd n TiO 2 450C 0.9 37 1685 32 67 5Pd n TiO 2 450C 2.2 44 1530 15 175 10Pd n TiO 2 450C 2.0 19 1 955 9 120 a Turnover number in mol produ ct per mol surface Pd. b Results from previous work . c Results from previous work .
137 CHAPTER 7 CONCLUSION The research presented here studied the development on nanoparticle oxide support ed catalysts using two pr obe reactions. The water splitting reaction used iron oxide deposited onto commercial ZrO 2 based supports to investigate the H 2 production through a flow reactor system. The second reaction the oxidative coupling of 4 methylpyridine reaction continued th e development by investiga ting other aspects of the catalytic system Since the deposition of the active phase on a variety of supports was previously investigated in this reaction support treatments and additives or promoters were investigated to improve the catalytic activity. The complexity of the 4 methylpyridine coupling reaction presented many challenges for improving economic favorability through yield increases, optimizing the PdO and ZrO 2 promoter loadings, and increasing the catalyst life span via regeneration. A reactor system was designed and constructed for the two step water splitting reaction to analyze the feasibility of (1) hydrogen generation ove r iron oxide based catalysts and (2) operating th ese catalyst s for several cycles without proce ssing between steps or cycles and without catalyst deterioration with increasing cycles Considerable amount of the early research focused on the theoretically identif ication of potential candidates for the two step process, and iron oxide was proven exper imentally to be a promising candidate despite rapid deactivation. The introduction of an inert, high temperature stable zirconia based supports was investigated. The nanoparticle zirconia significantly increased the support surface area providing activity for multiple cycles. The addition of yttria to stabilize zirconia support further increased the hydrogen production. The stabilization of the cubic zirconia phase more than double the hydrogen
138 production over the monoclinic phase of the nanoparticle zircon ia. For both supports, the surface area remained relatively high, about an order of magnitude higher than reported in literature, over the course of 10 cycles. Increasing the Fe loading to 20 wt% for ZrO 2 support was detrimental as the H 2 production decli ned significantly after the first cycle and was consistently half that of the 10FeZrO 2 catalyst. The best catalyst was a 10 wt % loading on 8YSZ support which generated 49 cm 3 of H 2 /g of material over the ten cycles. The second probe reaction oxidative cou pling of 4 methylpyridine was used to continue the catalyst development of PdO supported on nanoparticle oxide supports for this reaction As previous research already identified several viable nanoparticle oxide supports, the support ed catalyst s with the best yields pellet titania and nanoparticle alumina, were investigated to improve the product yield and catalytic activity. Since the nanoparticle titania support surface area was appreciably higher than the pellet titania, support pretreatment conditio n s were studied to explain the cause of the low yield on the nanoparticle support and to obtain a yield similar to the pellet titania and nanoparticle alumina supports. The heat treatment conditions of the support were critical in improving the reaction yi eld and increasing the activity. The as purchased nanoparticle titania support phase was mostly amorphous, and a support heat treatment at 450 C prior to palladium deposition had an anatase phase similar to the pellet titania support. However, the heat tr eatment decreased the surface area in compar ison with the pellet titania leaving PdO/p TiO 2 with a slightly higher 52 g product/g Pd compared to the 44 g product/g Pd obtained for the PdO/n TiO 2 450C.
139 S elected supports were choosen to deposite additives i n order to promote a higher yield catalyst. PdO deposited on zirconia, ceria, and zinc oxide supports showed potential but had low surface areas and therefore were selected as the additives. Zinc oxide caused a decline in the yield through blocking active PdO sites, undermining PdO support interactions, or forming mixed metal oxides with the support The oxygen mobility of ceria was initially thought to be a benefit since PdO is the active phase. However, ceria is also known to migrate, and this migration covered the PdO resulting in large yield variations and thus causing reproducibility issues Zirconia, on the other hand, proved to a true promoter increasing the yield and turnover number of all the investigated catalyst. In fact, 5/5/90 PdO/ZrO 2 /n Al 2 O 3 (+) had the highest reported yield to date of any catalyst at 68 gram product/g Pd. The spent PdO/n Al 2 O 3 (+) catalyst underwent different treatment conditions in vity, and thus, extend the life sp an of the catalyst. S everal different deactivation mechanisms have been identified The majority of the surface by slowing down the reaction. The primary cause of deactivation was shown t o be the reduction of the active phase from PdO to Pd metal. The spent catalysts were treated under static air, He flow, and O 2 flow at different temperature to (1) regenerate the PdO phase and /or (2) remove the carbon species on the surface. The static ai r and oxygen were able to restore PdO, but only an O 2 flow was able to remove the carbon from the surface in any significant amount Yet, the activity could not be completely restored. In burning off the carbon, localized hot spots reduced the palladium su rface area causing severe sintering of the catalyst. A mild heat treatment of 200 C in static
140 air showed the best results with 50 g product/g Pd and 46 g product/g Pd for the first and second regenerations, respectively. Another method to further improve the economics was to optimize the palladium loading. L owering the palladium loading while maintaining a comparable yield decreases material cost Early research showed a 5 wt % palladium loading was more active than a 10 wt % loading as indicated by a simila r turnover number and higher yield. A 5 wt% loading had the best activity for the p TiO 2 and n TiO 2 450 C supports for the different loadings. T he yield per gram Pd increases with decreasing Pd loading, a 5 wt % PdO/p TiO 2 was optimal loading since half the loading only increases the yield by 35% while doubling the loading has a 65% increase. The PdO/n Al 2 O 3 (+) catalyst showed a yield increas e of 48 % in reducing the Pd loading from 5% to 2.5% and 110 % increase in reducing the Pd loading from 2.5% to 1% Alt hough a 1 % Pd loading gave a better yield per gram Pd (156 g/g Pd) the 2.5% Pd loading had the same TON as the 1% Pd loading The Zr loading on the best catalyst to date was also investigated. The yield was slightly increased for a 1% Zr loading along wi th the TON. Although the 2.5/1/96.5 PdO/ZrO 2 /n Al 2 O 3 (+) catalyst gave a better yield of 88 g/g Pd compared to 74 g/g Pd on the 2.5% PdO/n Al 2 O 3 (+) catalyst, the 1% PdO/n Al 2 O 3 (+) catalyst had the best yield to date of 156 g/g Pd. A systematic procedure for designing and developing a catalyst was demonstrated between these two different reactions. Support materials were used initially to impart mechanical strength a span by increasing the surface area and reducing sintering. Promoters were added to stabiliz e the support, increas e the oxygen mobility, and increas e the product yield. Preparation conditions
141 were explored via support heat treatments and optimizing the loading of active metal to discover a balance between support m orphology, active metal and/or additive loading. Regeneration treatments were performed to increase the life span of a catalyst after deactivation in the case of the palladium catalysts. The work presented here only identified two practical iron oxide cata lyst candidates for the two step water splitting reaction and improved a palladium oxide catalyst for one particular reaction, the oxidative coupling of 4 methylpyridine. As for the iron oxide catalysts, future work continues to develop a high temperature stable material that minimizes sintering. Additionally, these catalysts would be impregnated onto a foam support to increase the surface area and facilitate oxygen transport to improve the reaction kinetics. A stable foam catalyst would expedite the transi tion to a commercially viable process for a sustainable, environmentally friendly production of H 2 As for the palladium oxide catalyst, future work would expand this catalyst to other coupling reactions. In changing the reaction, the optimal formulation m ay vary leading to the adjustment in palladium loading and need for different additives to promote a different reaction. Ultimately, catalysis is a delicate formulation balance that requires experimentation to find the correct combination of ingredients (a ctive metal, additives, and support materials) to provide the high activity (and selectivity) for the desired reaction.
142 LIST OF REFERENCES 1. J.T. Richardson, Principles in Catalyst Development, Plenum Press: New York (1989) 2. L. Guczi, A. Beck, A. Horvth an d D. Hor vth, Top. Catal. 19 (2002) 157. 3. M.A. Rosen, Energy 35 (2010) 1068 4. A. Kogan, Int. J. Hyd ro En. 23 (1998) 89 5. G. Bromaghim, K. Gibeault, J. Serfass, P. Serfass, E. Wagner, Hydrogen and Fuel Cells: The U.S. Market Report (2010) 6. S. Abanades, P. Cha rvin, F. Lemont, G. Flamant, Int. J. Hyd ro En. 33 (2008) 6021 7. P. Charvin, S. Abanades, F. Lemont, G. Flamant, Energy 32 (2007) 1124 8. L. Nalbandian, A. Evdou, V. Zaspalis, Int. J. Hydro. En. 34 (2009) 7162 9. D.H. Lee, K.S. Cha, Y.S. Lee, K.S. Kang, C.S. Pa rk, Y.H. Kim, Int. J. Hydro. En. 34 (2009) 1417 10. S. Ab anades, G. Flamant, Sol. En. 80 (2006) 1611 11. N. Gokon, T. Hasegawa, S. Takahashi, T. Kodama, Energy 33 (2008) 1407 12. S. B. Han, T.B. Kang, O. S. Joo, K.D. Jung, Sol. En. 81 (2007) 623 13. S. Abanades, P. Char vin, G. Flamant, F. Lemont, Energy 31 (2006) 2805 14. A. Evdou, L. Nalbandian, V.T. Zaspalis, J. Membr. Sci. 325 (2008) 704 15. P. Charvin, S. Abanades, L. Florent, F. Gilles, Energy Con. Man. 49 (2008) 1547 16. H. Kaneko, T. Yokoyama, A. Fuse, H. Ishihara, N. Hase gawa, Y. Tamaura, Int. J. Hydro. En. 31 (2006) 2256 17. K. Urasaki, N. Tanimoto, T. Hayashi, Y. Sekine, E. Kikuchi, M. Matsukata, App. Catal. A 288 (2005) 143 18. K. Otsuka, C. Yamada, T. Kaburagi, S. Takenaka, Int. J. Hydro. En. 28 (2003) 335 19. Y. Tamaura, N. Ko jima, N. Hasegawa, M. Inoue, R. Uehara, N. Gokon, H. Kaneko, Int. J. Hydro. En. 26 (2001) 917 20. T. Kodama, Nobuyuki Gokon, Chem. Rev. 107 (2007) 4048
143 21. J.E. Funk, Int. J. Hydro. En. 26 (2001) 185 22. N. Gokon, T. Kodama, N. Imaizumi, J Umeda, T. Seo, Int. J. H ydro. En. 35 (2010) 7441 23. H. Naito, H. Arahi, Sol. St. Ion. 79 (1995) 366 24. T. Sano, M. Kojima, N. Hasegawa, M. Tsuji, Y. Tamaura, Int. J. Hydro. En. 21 (1996) 781 25. M.E. Glvez, A. Frei, G. Albisetti, G. Lunardi, A. Steinfeld, Int. J. Hydro. En. 33 (2008) 28 80 26. C. Agrafiotis, M. Roeb, A.G. Konstandopoulos, L. Nalbandian, V.T. Zaspalis, C. Sattler, P. Stobbe, A.M. Steele, Sol En 79 (2005) 409 27. N. Gokon, H. Murayama, J. Umeda, T. Hatamachi, T. Kodama, Int. J. Hydro. En. 34 (2009) 1208 28. J. E. Miller, M.D. Allen dorf, R.B. Diver, L.R. Evans, N. P. Siegel, John N. Stueck er, J. Mat. Sci. 43 (2008) 4714. 29. T. Kodama, Y. Kondoh, R. Yamamoto, H. Andou, N. Satou, Sol. En. 78 (2005) 623 30. N. Gokon, T. Mizuno, Y. Nakamuro, T. Kodama, J. Sol. En. 130 (2008) 1 31. T. Kodama, Y. N akamuro, T. Mizuno, J. Sol. En. Eng. 128 (2006) 3 32. T. Kodama, N.Gokon, R. Yamamoto, Sol En 82 (2008) 73 33. I. Vishnevetsky, A. Berman, M. Epstein, Int. J. of Hydro. En. 36 (201 1 ) 2817 34. K. Wegner, H.C. Ly, R.J. Weiss, S.E. Pratsinis, A. Steinfeld, Int. J. H ydro. En. 31 (2006) 55 35. N. Gokon, S. Takahashi, H. Yamamoto, T. Kodama, Int. J. Hydro. En. 33 (2008) 2189 36. A. Kogan, Int. J. Hydro. En. 25 (2000) 1043 37. F. Fr esno, R. Fernndez Saavedra, M. B. Gmez Mancebo, A. Vidal, M. Snchez, M.I. Rucandio, A.J. Quejido, M. Romero, Int. J. Hydro. En. 34 (2009) 2918 38. V. Balzani, G. Bergamini, F. Marchioni and P. Ceroni, Coord. Chem. Rev. 250 (2006) 1254 39. C.D. Clark, M.Z. Hoffman, Coord. Chem. Rev. 159 (1997) 359
144 40. R.C. Evans, P. Douglas and C.J. Winscom, Coord. Chem. Rev. 2 50 (2006) 2093 41. Campaa, J.J. Soto Chinchilla, J.F. Huertas Prez and A. Gonzlez Casado, Trends Anal. Chem. 24 (2005) 927 42. D.L. Feldheim, C.J. Baldy, P. Sebring, S.M. Hendrickson, C.M. Elliott, J. Electrochem. Soc. 142 (1995) 3366. 43. B. Milani, A. Anzilutti, L. Vicentini, A. Sessanta o Santi, E. Zangrando, S. Geremia, G. Mestroni, Organometallics 16(1997) 5064 44. L. M. Neal, D. Hernandez, H.E. Hagelin Weaver, J. Mol. Catal. A 307 (2009) 29 45. L.A. Summers, Adv. Heterocycl. Chem. 35 (1984) 281. 46. bipyridine, GFS Chemicals: http://gfschemicals.com/statics/productdetails/cHJvZHVjdGRldGFpbHMxMjc=.ht ml, accessed 5/2 8 /2013. 47. G. Cravotto, M. Beggiato, A. Penoni, G. Palmisana, S. Tollari, J. Lvque, W. Bontrath, Tetrahedron Lett. 46 (2005) 2267. 48. M. Heller, U Schubert, J. Org. Chem. 67 (2002) 8269. 49. P.E. Rosevear, W.H.F. Sasse, J. Heterocycl. Chem. 8 (1971) 483 50. P.E. Rosevear, W .H. Sasse, Patent App. No. 40930/72, 1973, p. 1 51. S. Munavalli, M. Gratzel, Chem. Ind. (1987) 722 52. H. Hagelin, B. Hedman, I. Orabona, T. kermark, B. kermark, C.A. Klug, J. Mol. Catal. A 164 (2000) 137 53. C. Kaes, A. Katz and M.W. Hosseini, Chem. Rev. 100 (2 000) 3553. 54. R. I. Masel, Principles of Adsorption and Reaction on Surfaces. Wiley: New York (1996). 55. B.D. Cullity, S. R. Stock, Elements of X ray Diffraction, 3rd Ed. Prentice Hall: New York (2001). 56. J.F. Moulder, W.F. Stickel, P.E. Sobol, K.D Bomben, Handb ook of X ray Photoelectron Spectroscopy, Physical Electronics Inc.: Eden Prairie, MN (1995). 57. L.M. Neal, H.E. Hagelin Weaver, J. Mol. Catal. A 284 (2008) 141. 58. A. Lamperti, E. Cianci, R. Ciprian, D. Sangalli, A. Debernardi, Thin Sol. Films 533 (2013) 83.
145 59. U.S Schubert, C. Eschbaumer, Angew Chem. Int. Ed. 41 (2002) 2892. 60. V. Balzani, A. Juris, M. Venturi, S. Campagna, S. Serroni, Chem. Rev. 96 (1996) 759. 61. G. Chelucci, R.P. Thummel, Chem. Rev. 102 (2002) 3129. 62. P.K. Ghosh, T.G. Spiro, J. Am. Chem. Soc. 102 (1980) 5543. 63. P.A. Adcock, F.R. Keene, R.S. Smythe, M.R. Snow, Inorg. Chem. 23 (1984) 2336. 64. G. Sprintschnik, H.W. Sprintschnik, P.P. Kirsch, D.G. Whitten, J. Am. Chem. Soc. 99 (1977) 4947. 65. J.J. Dodson, L.M. Neal, H.E. Hagelin Weaver, J. Mol. Catal. A 341 (2011) 4 2. 66. Y. Li, B. Xu, Y. Fan, N. Feng, A. Qiu, J.M.J. He, H. Yang, Y. Chen, J. Mol. Catal. A 216 (2004) 107. 67. J. Nair, P. Nair, F. Mizukami, Y. Oosawa, T. Okubo, Mat. Res. Bulletin 34 (1999) 1275. 68. L.M. Martinez, C. Montes de Correa, J.A. Odriozola, M.A. Centeno, J. Mol. Catal. A 253 (2006) 252. 69. B. Didillon, E. Merlen, T. Pages, D. Uzio, Stud. Surf. Sci. Catal. 118 (1998) 41. 70. L.M. Neal, M.L. Everett, G.B. Hoflund, H.E. Hagelin Weaver, J. Mol. Catal. A 335 (2011) 210. 71. L.M. Neal, S.D. Jones, M.L. Everett, G.B. Hoflu nd, H.E. Hagelin Weaver, J. Mol. Catal. A 325 (2010) 25. 72. C.D. Wagner, W.M. Riggs, L.E. Davis, J.F Moulder, G.E. Muilenberg, Eds., Handbook of X ray Photoelectron Spectroscopy, Perkin Elmer Corp.: Eden Prairie, MN (1979). 73. J. Hong, W. Chu, M. Chen, X. Wang, T. Zhang Catal. Comm. 8 (2007) 593. 74. N.M.L. Hansen, K. Jankova, S. Hvilsted, Eur. Polym. J. 43 (2007) 255. 75. H. Ishii, M. Goyal, M. Ueda, K. Takeuchi, M. Asai, Appl. Catal. A 201 (2000) 101. 76. F. Tsai, B. Lin, M. Chen, C. Mou, S. Liu, Tetrahedron 63 (2007) 430 4. 77. W. Wu, S. Chen, F. Tsai, Tetrahedron Lett. 47 (2007) 9267.
146 78. bipyridine, GFS Chemicals: http://gfschemicals.com/statics/productdetails/cHJvZHVjd GRldGFpbHMxMjc=.ht ml accessed 12/15/2010 79. C.E. Gigola, M.S. Moreno, I. Costilla, M.D. Sanchez, Appl. Surf. Sci. 254 (2007) 325. 80. W. Shen, Y. Matsumura, J. Mol. Catal. A 153 (2000) 165. 81. J. Kaspar, P. Fornasiero, M. Graziani, Catal. Today 50 (1999) 285. 82. J.M Padilla, G. Del Angel, J. Navarrete, Catal. Today 133 135 (2008) 541. 83. A. Trovarelli, C. de Leitenburg, M. Boaro, G. Dolcetti, Catal. Today 50 (1999) 353. 84. R. Zhou, B. Zhao, B. Yue, Appl. Surf. Sci. 254 (2008) 4701. 85. L.M. Martinez, T.C. Montes de Correa, J. A. Odriozola, M.A. Centeno, Catal. Today 107 108 (2005) 800. 86. L. Bollmann, J.L. Ratts, A.M. Joshi, J. Catal. 257 (2008) 43. 87. G.R. Newkome, A.K. Patri, E. Holder, U.S. Schubert, Eur. J. org. Chem. (2004) 235. 88. S. Liu, K. Takahashi, H. Eguchi, K. Uematsu, Catal Today 129 (2007) 287. 89. NanoScale Corporation http://www.nanoscalecorp.com/content.php/chemicals/powders/ accessed on 11/19/2010, n Al 2 O 3 (+): NanoActive Aluminum Oxide Plus, n Mg O: NanoActive Magnesium Oxide Plus, and n TiO 2 : NanoActive Titanium Dioxide. 90. S.D. Jones, L.M. Neal, H.E. Hagelin Weaver, Appl. Catal. B 84 (2008) 631. 91. A. Badri, C. Binet, J.C. Lavalley, J. Chem. Soc. Faraday Trans. 92 (1996) 1603. 92. S. Naito, T. Kasahara, T. Miyao, Catal. Today 74 (2002) 201. 93. R.N. Gayen, S.N. Das, S. Dalui, R. Bhar, A.K. Pal, J. Crystal Growth 310 (2008) 4073. 94. K. Itatani, K. Koizumi, F. Scott Howell, A. Kishioka, M. Kinoshita, J. Mater. Sci. 23 (1998) 3405. 95. D.R. Lide (Ed.), CRC Handbook of Ch emistry and Physics, 90 th ed., CRC Press Taylor & Francis Group, Boca Raton, FL, 2010, pp. 8 137.
147 96. Y. Li, Y. Fan, H. Yang, B. Xu, L. Feng, M. Yang, Y. Chen, Chem. Phys. Lett. 372 (2003) 160 97. C.T. Wang, J.C. Lin, Appl. Surf. Sci. 254 (2008) 4500. 98. H. Zhu, Z. Q in, W. Shan, W. Shen, J. Wang, J. Catal. 225 (2004) 267. 99. N. Venkatachalam, M. Palanichamy, B. Arabindoo, V. Murugesan, J. Mol. Catal. A 266 (2007) 158. 100. N. Madaan, S. Gatla, V.N. Kalevaru, J. Radnik, B. Lucke, A. Bruckner, A. Martin, J. Catal. 282 (2011) 10 3. 101. N. Concibido, T. Okuda, W. Nishijima, M. Okada, Appl. Catal. B 71 (2003) 64. 102. Y. Diao, P. Yang, R. Yan, L. Jiang, L. Wang, H. Zhang, C. Li, Z. Li, S. Zhang, Appl. Catal. B (2013) In Press 103. F.J. Urbano, J.M. Marinas, J. Mol. Catal. A 173 (2001) 329. 104. J.A. Moulijn, A.E. van Diepen, F. Kapteijn, Appl. Catal. A 212 (2001) 3. 105. V. Ferrer, A. Moronta, J. Sanchez, R. Solano, S. Bernal, D. Finol, Catal. Today 107 (2005) 487. 106. M. Konsolakis, I.V. Yentekakis, G. Pekridis, N. Kaklidis, A.C. Psarras, G.E. Marnellos, Appl Catal. B 138 (2013) 191. 107. M. Domingo Garcia, F.J. Lopez Garzon, M.J. Perez Mendoza, J. Coll. Int. Sci. 248 (2002) 116. 108. Dimethyl bipyridine, 99% Sigma Aldrich: http://www.sigmaaldrich.com/catalog/product/aldrich/245739?lang=en®ion=U S accessed 6/18/2013.
148 BIOGRAPHICAL SKETCH Justin Jeffrey Dodson was born in Saint Petersburg, FL to Jeffrey and Sherry Dodson in December 1983. He was raise d in Pinellas Park, FL. He attended the Center for Advanced Technology (CAT, a science and technology magnet program) at Lakewood High School. After graduating high school in 2002, he attended the University of South Florida in Tampa, FL. While there, he w orked as an undergraduate research assistant for Dr. Venkat Bhethanabotla focusing on characterizing the oxygen diffusion of fiber reinforced composite materials. He graduated from the University of South Florida in May 2007 with a B achelor of S cience in c hemical engineering and minor in mathematics. In August 2007, he moved to Gainesville, FL to pursue a M aster of S cience in chemical engineering at the University of Florida. He joined Dr. Helena Hagelin After graduating with a M aster of S cience in August 2008, he continued to work in Dr. Hagelin D octor of P hilosophy program in the fall 2009. His work has focused on two projects the two step thermal chemical water splitting and palladium support on nanoparticle oxide supports for use as an oxidative coupling catalyst.