Redox Catalysis for Environmental Applications

REDOX CATALYSIS FOR ENVIRONMENTAL APPLICATIONS

DISSERTATION

Presented in Partial Fulfillment of the Requirements for the Degree

Doctor of Philosophy

in the Graduate School of The Ohio State University

Preshit Vilas Gawade, M.S.

Chemical Engineering Graduate Program

*****

The Ohio State University

2012

Dissertation Committee:

Umit S. Ozkan, Advisor

Jeffrey J. Chalmers

Kurt W. Koelling

James Coe

Preshit Vilas Gawade

2012

ABSTRACT

The presented work comprehends a broad spectrum of redox catalysis for various environmental applications, such as i) hydrogen production via water-gas shift reaction, ii) hydrogen purification for fuel cell applications and iii) catalytic after-treatment of lean-burn engines. This dissertation involves, but is not limited to catalyst development, reaction studies and catalyst characterization for the above-mentioned environmental applications, which can be summarized as follows.

(i) Water-gas shift (WGS) remains an essential step in integrated gasification combined cycle (IGCC) for hydrogen production, as it forms a link between the gasification process and fuel cell operations. The current catalysts for WGS application are based on Fe-Cr and Cu/ZnO/Al2O3, as a high temperature (HT-WGS) and low temperature (LT-WGS) catalysts, respectively. This two-stage WGS process is a consequence of several operational drawbacks of the current catalyst formulations including Cr+6 being carcinogenic. Hence the presented WGS project has a two-fold purpose. First, Cr-free Fe-based catalyst development and second, Cu supported catalyst development for WGS that can be operated over a wide temperature range.

In this dissertation, Cr- free Fe-Al-Cu catalyst prepared through “one-step” sol- gel method using propylene oxide as a gelation agent has been reported. Steady state reactions demonstrated that WGS performance of Fe-Al-Cu was superior as compared to

ii commercial Fe-Cr catalyst. The reaction studies along with complementary catalyst characterization indicated that the amount of copper in iron oxide matrix played a crucial role. The optimized ratio of Fe to Cu was found to be five and any further increase in copper loading resulted in copper segregation from the iron oxide matrix. Thus various catalyst characterization techniques were exploited to understand this phenomenon.

Furthermore, a detailed study was performed to comprehend the formation of surface species during WGS reaction and to evaluate the reaction mechanism over Fe-Al-Cu.

In the quest of exploring Cu-based catalyst for WGS system, Cu supported over various CeO2 nano-morphologies were investigated. Here, nanoparticles (NP) and nanorods (NR) of CeO2 were prepared through hydrothermal precipitation method and copper was supported on these morphologies using a wet impregnation method. In the current findings, copper was more finely and uniformly dispersed over CeO2 nano- particles compared to nanorods, resulting in better WGS activity compared to particle- based samples. Catalyst characterization indicated finely dispersed copper particles in close interaction with ceria nanoparticles, whereas isolated bulk-like copper species were formed over the ceria nanorods. Finally, the formation of surface species during WGS reaction delineated the redox reaction mechanism over Cu/CeO2.

(ii) Hydrogen produced via WGS reaction may contain up to 1-2% CO in stream which can be poisonous to proton exchange membrane (PEM) fuel cell. Preferential oxidation of carbon monoxide (PROX) is considered as an effective and economical way

iii to purify the hydrogen stream for PEM fuel cell applications. The major challenge in this process is to selectively oxidize CO with minimum loss of hydrogen. Hence a non- precious metal catalyst such as, cobalt supported over ceria with a special focus on cobalt loading has been utilized. Both, activity and selectivity were found to be a strong function of cobalt content. In addition, CO and hydrogen oxidation kinetics was studied as a function of cobalt loading. The higher activation energy barrier for hydrogen oxidation compared to CO oxidation indicated higher temperature sensitivity for hydrogen oxidation. The cobalt phase was identified, as Co3O4 and it remained stable under PROX atmosphere. Time-on-stream experiment along with various catalyst characterization techniques indicated no significant contribution from lower valency cobalt species.

Finally, the formation of surface species during PROX reaction demonstrated conversion of carbonate species to more stable polydentate carbonates and formate type species with increase in reaction temperature.

(iii) Lean-burn natural gas fired engines remains a popular choice in the energy market. Despite emission being greatly reduced, exhaust still contains considerable amount of NOx, CO and hydrocarbons. Hence after-treatment to clean up the exhaust is essential. The selective catalytic reduction using hydrocarbons is considered as a promising alternative for conventional after-treatment technology and is well suited, especially for natural gas lean-burn engines.

For this purpose, a “single-stage” de-NOX system composed of a physical mixture of dual-catalyst bed has been developed. This dual catalyst bed was a physical mixture of reduction (Pd/SZ) and oxidation (CoOx/CeO2) catalysts capable of performing three distinct functions, NO oxidation, NOx reduction, and CO and hydrocarbon oxidation.

Here, oxidation catalyst was assumed to play multi-functional role in dual-catalyst bed.

These include oxidizing NO or re-oxidizing partially reduced NOx species, CO oxidation and catalyzing the combustion of un-burned hydrocarbons, which have not participated during SCR reaction.

In this dissertation, the role of oxidation catalyst in dual-catalyst bed was addressed. NO oxidation was studied as a function of cobalt loading in CoOx/CeO2 formulation. The dual-catalyst bed was optimized by varying the reduction to oxidation catalyst ratio in order to achieve significantly high NOx conversion during hydrocarbon-

SCR. The lower cobalt loading in oxidation catalyst in a dual-bed resulted in higher NOx conversion. This observation was associated with lower hydrocarbon oxidation and hence increased hydrocarbon availability for NOx reduction. Kinetic study along with catalyst characterization confirmed the activation of methane molecule via hydrogen abstraction, consequently participating in either in NOx reduction or directly oxidizing over the oxidation catalyst. Moreover, the effect of water vapor was thoroughly investigated over the optimized dual-catalyst bed. The primary focus of this work was to improve the hydrothermal stability of the dual-catalyst bed by changing the various engine exhaust parameters.

Dedicated to my parents and my wife

ACKNOWLEDGMENTS

I would like to thank Prof. Umit S. Ozkan for her support and guidance throughout my Ph.D. Her dedication to work, endless energy, support and respect to fellow co-workers has always inspired me.

I have always believed that research is a team-effort. I thank all the current and former group members for contributing directly or indirectly to my research. Especially, I would like to thank Burcu Bayram and Anne-Marie Alexander for their contributions towards my projects. Some of the work presented here was completed with their help.

I thank Ohio State University for providing excellent research facility. I thank department staff and faculty members for their support and help. Especially, I would like to thank Prof. Jeffrey Chalmers and Prof. Kurt Koelling for serving on my committee and providing valuable recommendations.

Financial support from Ohio Coal Development Office and Department of Energy is greatly appreciated. I also thank Caterpillar for funding the emission clean up project.

I would like to acknowledge Ronald Silver (Technical Centre, Caterpillar) for the valuable discussion on the NOx reduction project.

I would like to thank my friends, especially “Columbus-Gang” for their friendship and making my stay at Columbus memorable. I think I will really miss you guys. Finally,

I thank my parents and beautiful wife for their endless support and love.

vii

VITA

December 1, 1982..………………………………………Born, Mumbai, India

October 2004..……………………………………………B.S. Chemical Engineering U.D.C.T., Mumbai, India

December 2007..…………………………………………M.S. Chemical Engineering University of Toledo, Ohio

December 2007-Present..………………………………..Graduate Research Associate. Ohio State University, Ohio

INVENTION DISCLOSURES

I. U.S. Ozkan, P. Gawade and A.M. Alexander, “Addition of alumina binder during palladium over sulfated zirconia preparation for monolith coating: Implications NOx selective catalytic reduction from lean-burn natural gas engines”, Invention Disclosure, Technology Commercialization and Knowledge Transfer, Ohio State University, (2012), (ID: 2012-294) Pending

II. U.S. Ozkan, P. Gawade and A.M. Alexander, “Novel methods to coat supported metal catalysts on monolith cores”, Invention Disclosure, Technology Commercialization and Knowledge Transfer, Ohio State University, (2012), (ID: 2012-293) Pending

PUBLICATIONS

1. I. Soykal, B. Bayram, H. Sohn, P. Gawade, J. Miller and U.S. Ozkan, “Ethanol steam reforming over Co/CeO2 catalyst: Investigation of effect of ceria morphology”, Applied Catalysis A, Submitted

viii

2. P. Gawade, A.M. Alexander, R. Silver and U. S. Ozkan, “Effect of various engine exhaust parameters on the hydrothermal stability of hydrocarbon-SCR catalysts for lean-burn systems”, In Preparation

3. P. Gawade, A.M. Alexander, R. Clark, and U. S. Ozkan, “The Role of oxidation catalyst in dual-catalyst bed for after-treatment of lean burn natural gas exhaust”, Catalysis Today, Submitted

4. P. Gawade, B. Bayram, A.M. Alexander and U.S. Ozkan, “Preferential oxidation of CO (PROX) over CoOx/CeO2 nanoparticles in hydrogen-rich stream: Effect of cobalt loading”, Applied Catalysis B: Environmental, Submitted

5. P. Gawade, B. Mirkelamoglu, and U. S. Ozkan, “The role of support morphology and impregnation medium on the water gas-shift activity of ceria-supported copper catalysts”, Journal of Physical Chemistry: C, (2010), 114, 18173–18181

6. P. Gawade, B. Mirkelamoglu, B. Tan, and U. S. Ozkan, “Cr-free Fe-based water-gas-shift catalysts prepared through propylene-oxide sol-gel assisted technique”, Journal of Molecular Catalysis A: Chemical 321, (2010), 61-70

7. M. P. Woods, P. Gawade, B. Tan, and U. S. Ozkan, “Preferential oxidation of carbon monoxide on Co/CeO2 nanoparticles”, Applied Catalysis B: Environmental, 97, (2010), 28- 35

FIELDS OF STUDY

Major Field: Chemical Engineering

Area of Interest: Heterogeneous Catalysis

TABLE OF CONTENTS

ABSTRACT …………………………………………………………………………….. ii

DEDICATION…………………………………………………………………………... vi

ACKNOWLEDGMENTS……………………………………………………………… vii

VITA …………………………………………………………………………………... viii

TABLE OF CONTENTS………………………………………………………………….x

LIST OF TABLES.……………………………………………………………………. xvii

LIST OF FIGURES……………………………………………………………………xviii

PART 1: HYDROGEN PRODUCTION via WATER-GAS SHIFT (WGS) REACTION

OVER Fe and Cu-BASED CATALYSTS………………………………………………...1

CHAPTER 1: Introduction to Hydrogen Production via Water-Gas Shift Reaction ……..2

CHAPTER 2: Literature Review.…………………………………………………………4

2.1 Fe-based Catalysts for Water-Gas Shift……………………………………….4

2.2 Cu-based Catalysts for Water-Gas Shift………………………………………6

CHAPTER 3: Experimental Methods……………………………………………………..9

3.1 Steady State Reaction System…………………………………………………9

3.2 Catalyst Characterization…………………………………………………….12

3.2.1 N2- Physisoprtion for BET Surface Area Analysis………………...12 x

3.2.2 Temperature Programmed Reduction (TPR)………………….…...12

3.2.3 Temperature Programmed Oxidation (TPO)………………………13

3.2.4 X-ray Photoelectron Spectroscopy (XPS)…………………………13

3.2.5 X-ray Diffraction (XRD)…………………………………………..14

3.2.6 Diffuse Reflectance Infra-red Fourier Transform Spectroscopy

……………………………………………………….…………...16

CHAPTER 4: Chromium-free Fe-based Water-Gas-Shift Catalysts Prepared Through

Propylene Oxide-Assisted Sol–Gel Technique…………………….……18

4.1 Overview of Cr-free Catalyst Prepared through Propylene Oxide Sol-gel

Technique…………………………………………………………………….18

4.2 Experimental Procedure…………………………………………………...…20

4.2.1 Catalyst Preparation……………………………………………..…20

4.2.2 Catalyst Activity Testing………………………………………..…21

4.2.3 Catalyst Characterization…………………………………………..22

4.3 Results and Discussion……………………………………………………....27

4.3.1 Effect of Gelation Agent on WGS Activity………………………..27

4.3.2 Effect of Cu-loading on WGS Activity…………………………....29

4.3.2.1 Catalyst Activity………………………………………....31

4.3.2.2 X-ray Diffraction………………………………………...34

4.3.2.3 X-ray Photoelectron Spectroscopy………………………41

4.3.2.4 Temperature-Programmed Reduction……………...……48

4.3.2.5 In situ DRIFTS Study Using CO………………………..50

4.3.2.6 Temperature Programmed Re-oxidation………………...53

4.4 Conclusions …………………………………………………………………57

CHAPTER 5: The role of Support Morphology and Impregnation Medium On Water

Gas-Shift Activity of Ceria-Supported Copper Catalysts…………………………….….58

5.1 Overview of Cu/CeO2 Catalyst: Effect of Ceria Morphology and

Impregnation Medium………………………………………………………58

5.2 Experimental Procedure…………………………………………………..…59

5.2.1 Catalyst Preparation……………………………………………….59

5.2.2 Catalyst Activity Testing………………………………………….62

5.2.3 Catalyst Characterization………………………………………….63

5.3 Results and Discussion…………………………………………………...…66

5.3.1 Catalyst Activity…………………………………………………..66

5.3.2 Transmission Electron Microscopy ………………………………69

5.3.3 X-ray Photoelectron Spectroscopy………………………………..71

5.3.4 X-ray Diffraction………………………………………………….77

5.3.5 Temperature-Programmed Reduction…………………………….82

5.3.6 N2O Chemisorption……………………………………………….86

5.3.7 In situ DRIFT During WGS………….……………………………89

xii

5.4 Conclusions …………………………………………………………………94

PART 2: PREFERENTIAL OXIDATION OF CARBON MONOXIDE (PROX) IN

HYDROGEN-RICH STREAMS ………………………………………………………..95

CHAPTER 6: Introduction to Preferential Oxidation of CO (PROX) in Hydrogen…….96

CHAPTER 7: Literature Review……………………………………………………...…98

7.1 Precious Metal Catalysts……………………………………………………..98

7.2 Non-Precious Metal Catalyst……………………………………………...…99

CHAPTER 8: PROX Over CoOx/CeO2 In H2-Rich Streams: Effect of Cobalt Loading

………………………………………………………………....………..101

8.1 Overview of PROX over CoOx/CeO2…….………………………………...101

8.2 Experimental Procedure…………………………………………………….102

8.2.1 Catalyst Preparation………………………………………………102

8.2.2 Catalyst Activity Testing…………………………………………103

8.2.3 Catalyst Characterization..………………………………………..106

8.3 Results and Discussion …………………………………………………….109

8.3.1 Surface Area and Dispersion Measurements……………………..109

8.3.2 Steady State Activity of Performance of CoOx/CeO2…………….112

8.3.3 Oxidation of CO and H2 over CoOx/CeO2………………………..115

8.3.4 Effect of CO2 and water vapor over CoOx/CeO2…………………120

xiii

8.3.5 Catalyst Characterization…………………………………………124

8.3.5.1 XRD…………………………………………………….124

8.3.5.2 TPR……………………………………………………..128

8.3.5.3 XPS……………………………………………………..130

8.3.5.4 In situ DRIFTS during PROX………………………….133

8.3.5.5 In situ X-ray absorption near edge spectra (XANES)

………………………………………………………….139

8.4 Conclusions…………………………………………………………………141

PART 3: CATALYTIC AFTER-TREATMENT OF NATURAL GAS ENGINE

EXHAUST-A DUAL STAGE APPROACH………………………………………..…143

CHAPTER 9: Introduction to Catalytic NOx Reduction……………………………….144

CHAPTER 10: Literature Review: NOx Reduction with Hydrocarbons………………146

CHAPTER 11: The Role of Oxidation Catalyst in Dual Catalyst Bed………………...150

11.1 Overview of Role of Oxidation Catalyst in Dual-Catalyst Bed…………..150

11.2 Experimental………………………………………………………………151

11.2.1 Catalyst Preparation……………………………………………..151

11.2.2 Catalyst Characterization………………………………………..151

11.2.3 Catalyst Activity Testing………………………………………..152

11.2.3.1 NO oxidation over CoOx/CeO2……………………….153

11.2.3.2 Effect of Reduction-to-Oxidation Catalyst Ratio …….154

xiv

11.2.3.3 Hydrothermal Stability: Effect of Co Loading……….154

11.2.3.4 Kinetics of Hydrocarbon Oxidation…………………...155

11.3 Results and Discussion……………………………………………………157

11.3.1 NO Oxidation over CoOx/CeO2…………………………………157

11.3.2 Effect of Reduction- to- Oxidation Catalyst Ratio……………...160

11.3.3 Hydrothermal Stability: Effect of Co Loading………………….164

11.3.4 Kinetics of Hydrocarbon Oxidation……………………………..167

11.3.5 Investigation of Surface Species during CH4-TPD……………...175

11.4 Conclusions………………………………………………………………..183

CHAPTER 12: Effect of Various Engine Exhaust Parameters to Improve the

Hydrothermal Stability during Hydrocarbon-SCR……………………………………..184

12.1 Overview of Improving Hydrothermal Stability…………………………..184

12.2 Experimental and Reaction Studies……………………………………….185

12.2.1 Effect of Hydrocarbon Concentration…………………………..186

12.2.2 Effect of Reaction Temperature………………………………....187

12.3 Results and Discussion……………………………………………...….…188

12.3.1 Effect of Hydrocarbon Concentration……………………….….188

12.3.2 Effect of Reaction Temperature…………………………………200

12.4 Conclusions………………………………………………………………..205

CHAPTER 13: Summary and Future Work …………………………………………...206

13.1 Water Gas-Shift…………………………………………………………...206

13.1 Fe-based Catalyst………………………………………………….206

13.1.2 Cu-based Catalyst……………………………………………….208

13.2 Preferential Oxidation of Carbon Monoxide……………………………...210

13.3 Selective Catalytic Reduction……………………………………………..212

REFERENCES…………………………………………………………...…………….216

APPENDIX A: List of Acronyms………………………………………………………232

APPENDIX B: Sample Calculations………………………………………………...…234

APPENDIX C: Sample Calculations for Methane/Ethane Oxidation Kinetics. ……….236

xvi

LIST OF TABLES

Table 4.1: BET surface area, and pore volume of Fe-Al-Cu catalysts at various Cu- loadings…………………………………………………………………………………..30

Table 4.2: Crystallite size of Fe-Al-Cu catalysts at various Cu-loadings calculated from the XRD patterns through Scherrer equation…………………………………………….37

Table 5.1: Specific surface areas of CeO2-based samples calculated using BET method

……………………………………………………………………………………...…….61

Table 5.2: XPS analysis of the catalysts……………………………..…………………..76

Table 5.3 Cu-dispersion values estimated from N2O chemisorption…………………….88

Table 8.1: Specific surface areas and Co dispersions over CoOx/CeO2 catalysts….…..111

Table 8.2: Activation energies of CO and H2 oxidation over CoOx/CeO2 catalysts……117

Table 8.3: CO conversion (%) over CoOx/CeO2 in the presence of H2O and CO2. Values in parentheses show O2 selectivity to CO2 (%) over CoOx/CeO2 in the presence of H2O and CO2…………………………………………………………………………………122

Table 11.1: Summary of kinetic parameters during CH4 and C2H6 oxidation over oxidation, reduction and dual catalyst bed catalyst…………………………………….171

Table 11.2: Summary of DRIFTS study during CH4-TPD over oxidation, reduction and dual catalyst bed catalyst ………………………………………………...…………….182

xvii

LIST OF FIGURES

Figure 3.1: Schematic of Steady State Reaction System for Water-Gas Shift…………..11

Figure 4.1: WGS activity (CO% conversion) of Fe-Al-Cu catalysts prepared by propylene oxide-assisted sol-gel method with different Cu loadings at (a) S/C = 1 and (b) S/C = 2.

Inset: time-on-stream performance of Fe-Al-Cu (Fe/Cu=5) at 400o C at S/C = 1. Reaction

Conditions: [CO] = 10%, [H2O] = 10% or 20%, [CO2] = 5%, [H2] = 7.5% and balance

3 -1 -1 N2, WHSV = 0.06 m (g cat) h , P = 1 atm……………………………………………33

Figure 4.2: XRD patterns of Fe-based samples. (a) Fe-only (b) Fe-Cu (Fe/Cu=10) (c) Fe-

Al-Cu (Fe/Cu = 10), (d) Fe-Al-Cu (Fe/Cu = 5) and (e) Fe-Al-Cu (Fe/Cu = 2)………….38

Figure 4.3: Transmission electron micrograph (TEM) of Fe-Al-Cu (Fe/Cu = 5)………..39

Figure 4.4: XRD patterns of Fe-Al-Cu samples reduced in syngas mixture

(a) Fe/Cu=10, (b) Fe/Cu=5 and (c) Fe/Cu=2……………………...……………………..40

Figure 4.5: X-ray photoelectron spectra of Cu 2p region of pristine and reduced Fe-Al-Cu samples: (a) Fe/Cu = 10, (b) Fe/Cu = 5 and (c) Fe/Cu = 2……………………………....44

Figure 4.6: X-ray photoelectron spectra of Fe 2p region of pristine and reduced Fe-Al-Cu samples: (a) Fe/Cu = 10, (b) Fe/Cu = 5 and (c) Fe/Cu = 2……………………….……...47

Figure 4.7: Temperature-programmed reduction profiles of Fe-Al-Cu catalysts………..49

Figure 4.8: DRIFTS study of CO interaction with Fe-Al-Cu (Fe/Cu=5)………………..52

xviii

Figure 4.9: Temperature-programmed re-oxidation of reduced Fe-Al-Cu samples at different ramp rates: (a) Fe/Cu = 10, (b) Fe/Cu = 5 and (c) Fe/Cu = 2. Insets show determination of activation energies of re-oxidation following Kissinger method……...56

Figure 5.1: Activity of (a) CeO2-NP, (b) CeO2-NR based WGS catalysts: (●) CeO2, (■)

Cu/CeO2(E) and (▲) Cu/CeO2(T) and (c) time-on-stream activity of Cu/CeO2-NP(E) at

400°C. Reaction conditions: [CO] = 10%, [H2O] = 10%, [CO2] = 5%, [H2] = 7.5% and

3 -1 -1 balance N2, WHSV = 0.06 m (g cat) h , P = 1 atm…………………………………..68

Figure 5.2: TEM images of (a) CeO2-NR and (b) CeO2-NR at higher magnification, (c)

CeO2-NP and (d) CeO2-NP at higher magnification…………………………………….70

Figure 5.3: Ce 3d region of the X-ray photoelectron spectra of (a) CeO2-NP, (b) CeO2-

NR, (c) Cu/CeO2-NP(E), (d) Cu/CeO2-NP(T), (e) Cu/CeO2-NR(E) and (f) Cu/CeO2-

NR(T)……………………………………………………………………………………74

Figure 5.4: Cu 2p region of the X-ray photoelectron spectra in (a) Cu/CeO2-NP(E), (b)

Cu/CeO2-NP(T), (c) Cu/CeO2-NR(E) and (d) Cu/CeO2-NR(T)…………………………75

Figure 5.5: X-ray diffraction patterns of (a) CeO2-NR, (b) CeO2-NP, (c) Cu/CeO2-NR(E)

(d) Cu/CeO2-NR(T) (e) Cu/CeO2-NP(E) and (d) Cu/CeO2-NP(T)………………………80

Figure 5.6: X-ray diffraction patterns collected during in-situ reduction with 5%H2/N2 of

(a) Cu/CeO2-NP(E), (b) Cu/CeO2-NR(E) and (c) Cu/CeO2-NP(T)

……………………………………………………..…………………………………….81

xix

Figure 5.7: Temperature-programmed reduction profiles of (a) CeO2-NR, (b) CeO2-NP,

H2/He. Inset: Temperature-programmed reduction profile of bulk CuO………………...85

Figure 5.8: In-situ DRIFT spectra collected during WGS reaction over Cu/CeO2-NP(E)

(a)high wavenumber region and (b)low wavenumber region……………………………92

Figure 5.9: In-situ DRIFT spectra collected during WGS reaction over Cu/CeO2-NR(E)

(a)high wavenumber region and (b)low wavenumber region……………………………93

Figure 8.1: Schematic of Steady State Reaction System for PROX……………………105

Figure 8.2: PROX activity and selectivity of () 1% CoOx/CeO2, () 2% CoOx/CeO2, and () 10% CoOx/CeO2. (Reaction conditions: 1% CO, 1% O2, 60% H2 in He.

3 -1 -1 WHSV=15,000 cm (g cat.) h ). Inset: Time-on-stream CO conversion and O2

o selectivity to CO2 over 10% CoOx/CeO2 at 175 C under above mentioned reaction conditions……………………………………………………………………………….114

Figure 8.3: CO conversion over () 1% CoOx/CeO2, () 2% CoOx/CeO2, and () 10%

CoOx/CeO2 during CO oxidation. (Reaction conditions: 2% CO and 2% O2 in helium).

Inset shows Arrhenius plot for determination of the activation energy for CO oxidation

…………………………………………………………………………………….…….118

Figure 8.4: H2 conversion over () 1% CoOx/CeO2, () 2% CoOx/CeO2, and () 10%

CoOx/CeO2 during H2 oxidation. (Reaction conditions: 2% H2 and 2% O2 in argon) Inset shows Arrhenius plot for determination of the activation energy for H2 oxidation

…………………………………………………………………………………………..119

Figure 8.5: CO conversion over () 1% CoOx/CeO2, () 2% CoOx/CeO2, and () 10%

CoOx/CeO2 during water-gas-shift. (Reaction conditions: 1% CO and 1% H2O in helium)

…………………………………………………………………………………………123

Figure 8.6: X-ray diffraction patterns of (a) CeO2, (b) 1% CoOx/CeO2, (c) 2%

CoOx/CeO2, and (d) 10% CoOx/CeO2…………………………………………………126

Figure 8.7: In situ X-ray diffraction patterns collected during reduction of 10%

CoOx/CeO2 with 5% H2/He…………………………………………………………….127

Figure 8.8: Temperature-programmed reduction profiles of (a) CeO2, (b) 1% CoOx/CeO2,

Figure 8.9: X-ray photoelectron spectra in the Ce 3d and Co 2p regions of (a) 1%

CoOx/CeO2 (b) 2% CoOx/CeO2 and (c) 10% CoOx/CeO2…………………………...…132

Figure 8.10: In situ DRIFT spectra collected during preferential oxidation of CO over 2%

CoOx/CeO2 (a) high wavenumber region and (b) low wavenumber region……………137

Figure 8.11: In situ DRIFT spectra collected during preferential oxidation of CO over

10% CoOx/CeO2. (a) high wavenumber region and (b) low wavenumber region. Inset: spectra taken at 125°C after the reaction gases are flushed……………………….….138

Figure 8.12: Normalized Co K-edge XANES spectra of 10% CoOx/CeO2. (a) Pristine and

o (b) In situ under PROX at 175 C (Spectra for Co3O4 is included for comparison)…...140

Figure 11.1: Schematic of Steady State Reaction System for HC-SCR………………..156

xxi

Figure 11.2: NO2 yield during NO oxidation over 2%CoOx/CeO2 () and

10%CoOx/CeO2 (). Reaction conditions: 1000ppm NO, 10% O2; 1 atm; GHSV :

-1 100,000 h , (---) thermodynamic equilibrium conversion of NO to NO2…….………..159

Figure 11.3: (a) NOx conversion (b) CH4 conversion over dual catalyst bed as a function of Pd/SZ-10%CoOx/CeO2 ratios, 1:0 ( ), 2:1(), 4:1() and 8:1() under simulated lean exhaust conditions. Reaction conditions: [NO2]=180 ppm, [CH4]=1737 ppm,

[C2H6]= 208 ppm, [C3H8]= 104 ppm, [CO]= 650 ppm, [CO2]= 6.5%, [O2]=10%; 1 atm;

GHSV : 32,000 h-1……………………………………………………………………..162

Figure 11.4: NOx (bold) and CH4 conversion (void) over Pd/SZ: CoOx/CeO2=8:1 at different cobalt loadings; 2%CoOx/CeO2 (,) and 10%CoOx/CeO2 (,) under simulated lean exhaust. Reaction conditions: 180 ppm [NO2]=180 ppm, [CH4]=1737 ppm, [C2H6]= 208 ppm, [C3H8]= 104 ppm, [CO]= 650 ppm, [CO2]= 6.5%, [O2]=10%; 1 atm; GHSV : 32,000 h-1…………………………………………………………….….163

Figure 11.5: Reversibility of the effect of water vapor on (a) NOx (b) CH4 conversion over Pd/SZ: CoOx/CeO2=8:1 at different cobalt loadings 2%CoOx/CeO2 (,) and

10%CoOx/CeO2 (,) under simulated lean exhaust. Reaction conditions: [NO2]=180 ppm, [CH4]=1737 ppm, [C2H6]= 208 ppm, [C3H8]= 104 ppm, [CO]= 650 ppm, [CO2]=

o -1 6.5%, [O2]=10%, [H2O]= 0% or 7%; 450 C; 1 atm; GHSV : 32,000 h …………..…166

xxii

Figure 11.6: (a) Rate of methane oxidation Vs. methane concentration and (b) fractional ethane conversion Vs. W/FA,0 over 2%CoOx/CeO2. Inset: comparison of power-law model predictions for hydrocarbon oxidation rate against experimentally determined rates over 2%CoOx/CeO2…………………………………………………………………….172

Figure 11.7: (a) Fractional methane conversion Vs. W/FA0 and (b) fractional ethane conversion Vs. W/FA0 over Pd/SZ. Inset: comparison of power-law model predictions for hydrocarbon oxidation rate against experimentally determined rates over Pd/SZ

…………………………………………………………………………………………..173

Figure 11.8: (a) Fractional methane conversion Vs. W/FA0and (b) fractional ethane conversion Vs. W/FA0 over Pd/SZ: 2%CoOx/CeO2= 8:1. Inset: comparison of power-law model predictions for hydrocarbon oxidation rate against experimentally determined rates over Pd/SZ: 2%CoOx/CeO2= 8:1……………………………………………………….174

Figure 11.9: In situ DRIFT spectra collected during CH4-TPD over i) 2% CoOx/CeO2, ii)

Pd/SZ and iii) 1:8 Dual-catalyst bed 2%CoOx/CeO2-PdSZ a) high wavenumber region b) low wavenumber region………………………………………………………………..181

Figure 12.1: Effect of hydrocarbon concentration () CHx=2050 ppm, () CHx=2500 ppm and () CHx=3050 ppm on (a) NOX and (b) CH4 conversion during NO2 reduction in the presence of H2O. Reaction conditions: [NO2]= 180 ppm, [CHx]= varying, [CO]=

o -1 650 ppm, [CO2]= 6.5%, [O2]=10%, [H2O]=7% ; 450 C ; 1 atm ; GHSV ̴ 32,000h

…………………………………………………………………………………………..191

xxiii

Figure 12.2: Effect of hydrocarbon concentration on hydrothermal stability of a dual catalyst bed Pd/SZ and CoOx/CeO2 under simulated lean exhaust. (a) NOx conversion (b)

CH4 conversion. ()[CHx]=2050 ppm, ()[CHx]= 2500 ppm. Reaction conditions:

[NO2]= 180 ppm, [CHx]= varying, [CO]= 650 ppm, [CO2]= 6.5%, [O2]=10%,

o -1 [H2O]=10% ; 450 C ; 1 atm ; GHSV ̴ 32,000h ………………………………………193

Figure 12.3: The effect of either a methane only hydrocarbon feed or a mixed higher hydrocarbon feed on the hydrothermal stability of a dual catalyst bed Pd/SZ and

CoOx/CeO2 under simulated lean exhaust. (a) NOx conversion (b) CH4 conversion. ()

[CH4]= 2050 ppm, ()[CHx]= 2050 ppm. Reaction conditions: [NO2]=180 ppm, [CHx]=

o varying, [CO]= 650 ppm, [CO2]= 6.5%, [O2]=10%, [H2O]=10% ; 450 C ; 1 atm ; GHSV

-1 ̴ 32,000h …………………………………………………………………………...…..195

Figure 12.4: The effect of different hydrocarbons, as a function of equal carbon basis, on the hydrothermal stability of a dual catalyst bed Pd/SZ and CoOx/CeO2 under simulated lean exhaust. (a) NOx conversion (b) CH4 conversion. () [CH4]= 2050 ppm, ()

[C2H6]= 1025 ppm, ()[C3H8]= 683 ppm. Reaction conditions: [NO2]=180 ppm, [CHx]=

o varying, [CO]= 650 ppm, [CO2]= 6.5%, [O2]=10%, [H2O]=10% ; 450 C ; 1 atm ; GHSV

-1 ̴ 32,000h ……………………………………………………………………………….197

xxiv

Figure 12.5: The effect of supplementary ethane addition on the hydrothermal stability of a dual catalyst bed Pd/SZ and CoOx/CeO2 under simulated lean exhaust. (a) NOx conversion (b) CH4 conversion. () [CHx]= 2050 ppm, () [CHx]= 2050 ppm +

100ppm C2H6. Reaction conditions: [NO2]=180 ppm, [CHx]= varying, [CO]= 650 ppm,

o -1 [CO2]= 6.5%, [O2]=10%, [H2O]=10% ; 450 C ; 1 atm ; GHSV ̴ 32,000h

………………………………………………………………………………………….199

Figure 12.6: Effect of reaction temperature () 425oC () 450oC, and () 475oC on (a)

NOX and (b) CH4 conversion during NO2 reduction in the presence of H2O. Reaction conditions: [NO2]=180 ppm, [CHx]= 2500 ppm, [CO]= 650 ppm, [CO2]= 6.5%,

-1 [O2]=10%, [H2O]=7% ; 1 atm ; GHSV ̴ 32,000h ………………...…………………..201

Figure 12.7: The effect of temperature on the hydrothermal stability of a dual catalyst bed

Pd/SZ and CoOx/CeO2 under simulated lean exhaust. (a) NOx conversion (b) CH4

o o o conversion. () 450 C, () 475 C, () 500 C. Reaction conditions: [NO2]=180 ppm, [CHx]= 2050 ppm, [CO]= 650 ppm, [CO2]= 6.5%, [O2]=10%, [H2O]=10% ; 1 atm ;

-1 GHSV ̴ 32,000h ……………………………………………...……………………….204

Figure 13.1: Various nano-morphologies of ceria prepared through hydrothermal- precipitation (Image Courtesy: Bing Tan)……………………………………...………210

xxv

PART 1:

HYDROGEN PRODUCTION via WATER-GAS SHIFT (WGS) REACTION

OVER Fe and Cu-BASED CATALYSTS

CHAPTER 1

INTRODUCTION TO HYDROGEN PRODUCTION via WATER-GAS-SHIFT

REACTION

Hydrogen production via gasification of coal and biomass is a promising and clean technology to produce hydrogen, electricity and other valuable products. Coal is the nation’s most abundant fossil fuel resource, which can last up to several hundred years. It accounts for around 50% of electricity production in United states [1]. On the other hand, power generation via biomass gasification has received a considerable attention in recent years to replace fossil fuels. The current estimated biomass worldwide energy potential is around 100 EJ/year (1EJ= 1018 J) [2], which accounts for 25% of the total energy requirement. However, Fischer, et al. [3] and Hoogwijk, et al. [4] estimated that it will increase to 675-1135 EJ/year in next 40-50 years.

WGS unit remains as an essential step in the overall integrated gasification process for hydrogen generations from syngas regardless of the gasification feedstock and operating conditions. Typically, the syngas mixture derived from gasification processes consists of CO, CO2, H2O, H2, CH4 and traces of other hydrocarbons. It may also content traces of H2S and COS depending upon feedstock, operating conditions and clean-up processes.

Water-gas-shift reaction (CO+H2O = CO2 + H2, ΔH = −40.6 KJ/mol) is a slightly exothermic reaction which is favored at low reaction temperature and high steam to carbon ratio (S/C) [5]. The current WGS unit is operated in a “two-stage” mode, namely, high-temperature WGS (HT-WGS) and low-temperature WGS (LT-WGS).

Commercially, HT-WGS operated in the temperature range of 320-450oC over Fe-Cr catalyst, which brings down CO concentration from 10% to 3-5%. HT-WGS is followed

o by LT-WGS, operated around 200-250 C over Cu/ZnO/Al2O3 catalyst which can further decrease CO concentration to 0.5-1% [6-9]. A “two-stage” operation mode is a consequence of several drawbacks of the current catalytic system. For example, Fe-Cr catalyst is practically inactive at low reaction temperature whereas, Cu/ZnO/Al2O3 has sintering issues during HT-WGS.

* Information in this chapter is adapted/taken from P. Gawade, B. Mirkelamoglu, B. Tan and U.S. Ozkan, Journal of Molecular Catalysis A: Chemical 321, (2010), 61-70

CHAPTER 2

LITERATURE REVIEW

2.1 Fe-based Catalysts for Water-Gas Shift:

Fe-Cr catalyst is used at commercial level during HT-WGS due to its high stability and reasonable activity. Cr acts as a structural promoter and stabilizer to prevent the particle sintering during high temperature operations. Commercially, 8-14% (wt%) of Cr is incorporated in Fe catalyst as a stabilizer [5, 10, 11]. In Fe-Cr catalyst, iron is mostly in the form of α-Fe2O3 (hematite), which gets converted to WGS active Fe3O4

(magnetite) phase under the reducing environment. However, over-reduction of active phase (Fe3O4) may occur due to exothermicity of the reaction that may lead to the formation of less active metallic Fe [5, 11]. In addition, metallic Fe is believed to promote the side reaction such as, methanation reaction and carbon formation [12].

Hence higher amount of steam is usually used during WGS over Fe-Cr to avoid the over- reduction. Cu-promoted Fe-Cr catalysts have shown promising WGS performance in the past even in the presence of low steam content without forming metallic Fe [10, 13].

Andreev and co-workers reported that copper promoted Fe-Cr had much better WGS activity than cobalt and zinc promoted catalysts [14, 15]. However, as mentioned earlier,

Cr+6 in Fe-Cr is highly carcinogenic and posses environmental and health hazards.

Therefore, the substitution of Cr from Fe-based catalysts is essential.

In the past, several attempts have been reported to substitute Cr from Fe-Cr. The oxides of Pb, La, Ca, Th, Al, Mg, Mn and Zr have shown promising potential to replace

Cr [11, 16]. 10% (wt%) PbO doped in Fe-catalyst showed comparable WGS activity to that of commercial Fe-Cr. However, this formulation failed in the presence of sulfur [11].

In the quest of replacing Cr, de Araújo, G.C. and do Carmo Rangel, M. [17] showed that aluminum could be a promising option for Fe-based catalyst. In their findings, a small addition of copper (3wt%) and aluminum significantly improved Fe-catalyst WGS activity even at low steam to gas ratio (S/G=0.4). In addition the catalyst properties and activity was similar to that of commercial Fe-Cr-Cu catalyst. The same research group [5] extended this work by replacing Al with thorium (Th) in Fe-Cu catalyst. In theses findings, thorium acted as a stabilizer and prevented the particle sintering at high reaction temperatures whereas; Cu played the role of structural promoter. The increase in surface area due to thorium addition in Fe-Cu-Th resulted in higher WGS activity than commercial Fe-Cu-Cr catalyst at 370oC.

Ozkan research group extensively studied the effect of incorporation of Al, Ga and Mn as replacement for Cr in Fe-Cr catalyst as well as catalyst preparation variables and methods [18]. The catalyst was prepared using co-precipitation and impregnation method. In these finding, Al emerged as a potential substitute for Cr and the further catalyst improvements were achieved by addition of Cu as a promoter. In the following years, Ozkan, et al. [19, 20] extended this Cr-free work and found that Fe-Al-Cu catalyst prepared using sol-gel method had much improved activity due to uniform distribution of

5 copper in iron oxide matrix than its previous counterpart prepared using precipitation method. Fe-Al-Cu catalyst was prepared using NaOH as a gelation agent and its performance was significantly higher than commercial Fe-Cr-Cu catalyst over a temperature range of 250-400oC. However, as mentioned earlier in spite of excellent

WGS performance, catalyst preparation method was tedious, required critical pH, several washing steps. Therefore, in the present work we report “single-step” sol-gel preparation method using propylene oxide as a gelation agent to obtain highly active and stable Fe-

Al-Cu catalyst.

2.2 Cu-based Catalysts for Water-Gas Shift:

Cu-based catalysts are as extensively investigated as Fe-based catalysts for WGS reactions. Cu supported over various oxides such as, Al2O3, ZrO2, SiO2-Al2O3, CeO2,

MgO, ZnO-Al2O3, Cr2O3, CeO2-ZrO2, CeO2-Al2O3 and Ce(La)Ox have been reported as

WGS catalysts [21-33]. Among these systems, Cu/CeO2 is probably the most widely investigated formulation due to unique nature of CeO2.

CeO2 is widely reported as a support and catalyst for various applications such as

WGS [22, 27, 28, 32, 34-38], preferential oxidation [39-43], NO reduction [44] and methane oxidation [45] due to its excellent oxygen storage capacity and redox property

[35]. It is also known to improve the catalyst reducibility and prevent particle sintering

[46]. Both, precious/noble [38, 47-49] as well as non-precious metals [21, 27, 34, 50,

51] supported over ceria are widely reported as WGS catalyst.

Excellent WGS activity and stability has been reported over Cu/CeO2 catalyst due to close interaction between Cu and CeO2 that resulted into improved catalyst reducibility. The formation of Cu0/Cu+1 over ceria support has been considered as an active site for WGS whereas; ceria plays a direct or indirect role in overall reaction process [27, 32, 47, 48, 51]. The exceptional oxygen storage capacity (OSC) of ceria is known to assist the redox cycle of WGS reaction which can be further improved by the incorporation of promoters such as, ZrO2 and La2O3 [46, 52]. The studies [27, 46] have shown the addition of 8% (wt%) La2O3 in Cu-Ce(La)Ox can significantly improved WGS activity and thermal stability up to 600oC.

Various catalyst preparation techniques have been reported for ceria and Cu/CeO2 catalysts preparation such as, co-precipitation, hard-template method, urea gelation co- precipitation, hydrothermal and sol-gel methods ([34, 46, 50, 53] and references therein).

Da Silva and Assaf [21] reported that highly active 5%Cu-CeO2 catalyst could be prepared through hydrothermal and urea co-precipitation method. In their findings, high surface area and better copper dispersion was believed to provide superior WGS performance which can be further enhanced by the preparation of nano-structures of ceria

(cited in [34]).

In recent years, numerous nano-sized morphologies of ceria are studied and debated for structure sensitive reactions such as CO oxidation and preferential oxidation of CO (PROX). It is believed that ceria morphologies affect the redox property and OSC

[54-59]. Mai et al. [54] showed that ceria nanorods and nanocubes were more active towards CO oxidation. According to their claim, the more exposed planes [1 1 0]/[1 0 0] of ceria were responsible for higher oxygen storage. Tana, et al. [60] further supported

7 these finding in similar nano-shaped ceria study for CO oxidation. Pan et al. [57] showed that CeO2 nano-wires prepared through hydrothermal method had better CO oxidation performance than nano-particles and nano-cubes. Yi et al. [61] extended ceria morphology work over Au/CeO2 catalyst for PROX. In their observations, 1%Au/CeO2

(particles and rods) had better PROX activity and O2 selectivity than 1%Au-CeO2-cubes in the temperature range of 30-100oC.

Although broad literature is available on ceria morphology for CO oxidation processes, few studies have been reported on the effect ceria morphology for WGS system. Si and Flytzani-Stephanopoulos [62] have reported the effect of ceria

o morphology over 1%Au/CeO2 catalyst for WGS system over 150-350 C. In these findings, 1%Au-CeO2-nanorods had better WGS activity due to more exposed [1 0 0]/[1

1 0] crystal plane which assisted in stabilizing gold. To the best of our knowledge, Ozkan research group will be one of the first to report and investigate the effect of various ceria morphologies over Cu/CeO2, which will be discussed in details in the subsequent section.

* Information in this chapter is adapted/taken from 1. P. Gawade, B. Mirkelamoglu, B.

Tan and U.S. Ozkan, Journal of Molecular Catalysis A: Chemical 321, (2010), 61-70 2.

P. Gawade, B. Mirkelamoglu, U.S. Ozkan, Journal of physical Chemistry C 114 (2010)

18173-18181.

CHAPTER 3

EXPERIMENTAL METHODS

3.1 Steady State Reaction System

The schematic of steady state reaction system for WGS is shown in Figure 3.1.

The system was equipped with four independent mass controllers (Brooks Instruments,

5850E) to regulate the gas flow rates. Two micron particle filters were installed before every mass flow controllers (MFC) to avoid any particulate contamination reaching to

MFC. In addition, either Oxy-trap© or Drierite© was installed in some of the lines (Ar,

H2 and N2) to remove moisture or oxygen contamination. Water vapor was introduced to the system by bubbling nitrogen through heated de-ionized water bubbler. The temperature of water bubbler was monitored and controlled by temperature controller.

The gas lines were heated (c.a. 120o C) to avoid any condensation of water vapor in the lines.

All the steady-state reactions were conducted in a fixed-bed flow reactor using ¼“

O.D. stainless steel rector tube. The catalyst was packed and held inside the reactor using stainless steel frit and quartz wool plug. The reactor was placed in homemade resistively

9 heated furnace using Omega (model CS232) PID temperature controller and the reaction temperature was monitored using Omega K-type thermocouple.

For most of the WGS reactions, 100 mg of catalyst sample was used for each run and the experiments were carried out at a WHSV (weight hourly space velocity) of 0.06 m3 (g cat)-1 h-1. In addition, some steady-state experiments were conducted on equal surface area to ensure that the observed differences were not simply due to differences in surface areas. In order to assure that the activity testing experiments were run in the kinetically controlled regime, reaction conditions were maintained away from equilibrium in all of the activity testing experiments.

Prior to catalytic activity testing, catalyst samples were pretreated in-situ

o o in N2 (67.5 sccm) for 30 minutes at 350 C and then reduced at 350 C for 2 hours in syngas mixture (10% CO, 10% (or 20%) H2O, 7.5% H2, 5% CO2 balanced with 67.5%

o N2). Finally, reaction studies were performed over the wide temperature range (250 C-

400o C) in the above mentioned syngas mixture.

The feed as well as the reactor effluents were analyzed by an on-line gas chromatograph (Shimadzu, GC-14A series) equipped with a thermal conductivity detector (TCD). Argon was used as the carrier gas and the gas separation was performed using a molecular sieve column 13X (5 ft x 1/8 in. SS, 60/80 mesh) and porapak Q column (12 ft x 1/8 in. SS, 80/100 mesh).

Figure 3.1: Schematic of Steady State Reaction System for Water-Gas Shift

3.2 Catalyst Characterization

3.2.1 N2- Physisorption for BET Surface Area Analysis

The specific surface areas of the samples were measured on either Micromeritics

ASAP 2010 or 2020 accelerated surface area and porosimetry instruments using nitrogen physisorption technique. Prior to analysis, samples were degassed overnight at 130oC under a vacuum of 3µmHg. The specific surface areas were determined by the Brunauer-

Emmett-Teller (BET) method using the adsorption branch of the isotherm. Pore volumes of the samples were determined by the Barret-Joiner-Halenda (BJH) method using the adsorption branches of the N2 physisorption isotherms.

3.2.2 Temperature Programmed Reduction (TPR)

Temperature programmed reduction (TPR) is a bulk characterization technique, which assists to identify the reduction temperature of the sample, metal-support interaction, presence of various components and oxidation states of the sample.

For H2-TPR experiments, usually 50-100 mg of calcined catalyst sample was packed in a 0.25” O.D. U-tube quartz reactor using quartz wool. The catalyst was pretreated in He at 350o C for 30 minutes followed by cooling at room temperature in He.

Then either 5% H2/He or 5% H2/N2 was introduced at room temperature before ramping the temperature at 10° C/min. Hydrogen consumption during the temperature ramp was monitored either using a thermal conductivity detector (TCD) or an online mass

12 spectrometer (Cirrus II, MKS Instruments, 1–300 amu) operated in selective ion detection mode.

3.2.3 Temperature Programmed Oxidation (TPO)

TPO experiments were conducted to derive re-oxidation kinetic parameters over

Fe-Al-Cu catalysts during WGS using Kissinger’s method [63]. This method correlates the temperature at which the maximum rate of oxidant consumption occurs with the activation energy for re-oxidation through a series of temperature-programmed experiments ran at different heating rates. The following equation is used to calculate the

2 activation energy for re-oxidation, ln(φ /Tm ) = (−E /RTm ) + Constant .

Where, Φ is the heating rate and Tm is the temperature at which the oxidant consumption rate is a maximum€ . The details of this study are presented in Section 4.3.2.6

3.2.4 X-ray Photoelectron Spectroscopy (XPS)

XPS is a surface characterization technique, which can provide elemental compositions, oxidation state and dispersion of one component over other [64]. This technique works on photoelectric effect in which desired atom absorbs a photon (hν) and the photoelectrons (core/valance) with binding energy Eb are ejected with kinetic energy

Ek. The following equation is used to calculate binding energy of photoelectrons:

Ek = hν- Eb- φ where; φ is a work function

XPS analysis was carried using Kratos AXIS Ultra X-ray photoelectron spectrometer operated under high ultra-high vacuum. The monochromatized Al Kα

(1486.7eV) X-ray source operated at 13 kV and 10 mA. When necessary, the charge neutralizer was operated at a current of 2.1A and a filament bias of 1.3V. For each sample a survey scan was collected from 1400 to 0 eV. Concurrent sweeps of Cu 2p, Fe

2p, Ce 3d, O 1s and C 1s envelopes were collected at an electron pass energy of 20 eV. In addition, the collected spectra were corrected using C 1s envelope located at 284.5 eV.

Analysis of the collected data was performed on XPS Peak 4.1.

3.2.5 X-ray Diffraction (XRD)

XRD is a bulk characterization technique which is used to identify the crystal structures and average crystal sizes. It can be also operated in situ to monitor the changes in crystal structures as a function of temperature and reaction environment. This technique is based on constructive interference of scattered X-ray radiations by the sample [64]. The following Bragg relation is used to derive the lattice spacing:

n λ= 2 d sinθ

Where; n= Order of reflection,

λ= Wavelength of X-rays,

θ= Angle between incoming X-rays and normal to lattice plane,

d= Distance between lattice planes

Furthermore, Scherrer formula can be used to calculate the average crystal size using the

peak width

K" < L >= #cos$

Where; = Average crystal size

! K= Constant (usually 1)

λ= Wavelength of X-rays

θ= Angle between incoming X-rays and normal to lattice plane,

β= Peak Width

XRD patterns of the pristine samples were collected on a Rigaku X-ray

diffractometer (X-ray source: Cu Kα radiation, λ=1.5418 Å) operated at 40 kV and 25

mA. The diffraction patterns were collected in the range of 20o to 60o Bragg angle values.

Capillary XRD was used to understand the transformation of crystal phases

during reduction. For capillary XRD analysis of Fe-Al-Cu samples, Bruker D8 Advance

X-ray diffractometer equipped with a capillary sample holder was used. Following in situ

reduction treatment inside the reactor (10% CO, 10% H2O, 7.5% H2, 5% CO2 balanced

o with 67.5% N2 for 2 h at 350 C), the samples were transferred to an Ar-purged glove box

that contains minimal residual air (< 5ppm). The samples were transferred to capillary

sample tubes (0.2 mm O.D.) and the ends of the capillary tube were sealed inside the

glove box with glass sealant. The samples were then transferred air-free to capillary

sample holder accessory on the diffractometer. The diffraction patterns were collected in

15 the 20°-70° Bragg angle range using a Cu Kα X-ray source (λ=1.5418 Å) operated at 40 kV and 50 mA.

For in situ analysis of Cu/CeO2 samples under a flow of 5% H2/N2 (15 ccm), the same

Bruker D8 Advance X-ray diffractometer equipped with an Anton Paar HTK1200 oven was used. The samples were heated from room temperature to 450°C under a flow of 5%

H2/N2 (15 ccm) and diffraction patterns were collected at regular temperature intervals.

3.2.6 Diffuse Reflectance Infra-red Fourier Transform Spectroscopy (DRIFTS)

DRIFTS is a vibrational spectroscopy technique used to identify the adsorbed species on catalyst surface and their interactions. It is an absorption technique, which can be operated in diffuse reflectance mode for the powder samples. For vibrations to be IR active dipole moment of the molecule must change during the vibrations. The FT-IR spectrometer (Thermo Nicolet 6700) equipped with an MCT detector and a DRIFTS chamber with ZnSe windows was used for DRIFTS study. Liquid nitrogen was used throughout the experiments to cool the MCT detector. The DRIFTS spectra were collected in the mid-IR range at a nominal resolution of 4cm-1 and the presented spectra were averaged over 500 scans.

For Fe-Al-Cu catalysts, the samples were reduced in reactor system under the conditions mentioned in Section 3.1. The reduced samples then were recovered under nitrogen atmosphere and transferred to DRIFTS cell. The details of reaction studies over

Fe-Al-Cu catalyst are given in Section 4.3.2.5. For Cu/CeO2 samples, spectra were

16 collected during WGS reaction in 1%CO and 1%H2O. The detailed description of which can be found in Section 5.3.7

* Information in this chapter is adapted/taken from 1. P. Gawade, B. Mirkelamoglu, B.

Tan and U.S. Ozkan, Journal of Molecular Catalysis A: Chemical 321, (2010), 61-70. 2.

P. Gawade, B. Mirkelamoglu, U.S. Ozkan, Journal of physical Chemistry C 114 (2010)

18173-18181.

CHAPTER 4

Cr-FREE Fe-based WATER-GAS SHIFT CATALYST PREPARED THROUGH

PROPYLENE OXIDE-ASSISTED SOL-GEL TECHNIQUE

4.1 Overview of Cr-free Catalyst Prepared through Propylene Oxide Sol-gel Technique

The sol-gel method for catalyst preparation is considered to be more effective in terms of offering improved homogeneity, high purity, low calcination temperature and ability to tailor the desired metal oxide phase over traditional preparation methods [65,

66]. The sol-gel method is basically a two-step process in which metal salt undergoes hydrolysis reaction followed by the condensation to form nano-scale structures.

Formation of iron oxide matrix using sol-gel techniques in which the gel formation was induced by heating, adding a base or a gelation agent was reported (cited in [67]). In the past, several researchers synthesized nano-scale Fe as well as spinel oxides using epoxides such as ethylene oxide and propylene oxide [66-68]. These gelation agents were reported to provide high homogeneity, high purity and high surface area [66-68]. The ratio of epoxide to metal salt [67, 68] as well as the choice of solvent play a critical role in gel formation rate and determining the particle size. The epoxides are believed to act as proton scavengers that cause iron to undergo hydrolysis and condensation processes to

18 form an iron oxide matrix [67]. However, it is worthwhile to note that ethylene oxide is highly carcinogenic and toxic which defeats the purpose of environmentally benign catalyst preparation methods. Citric acid is also another gelation agent, which is widely studied for the preparation of highly dispersed mixed-oxides such as bismuth oxide, cerium-zirconium oxide and spinel oxides of cobalt [65, 69-72].

In this chapter, we report development of Cr-free Fe-based formulations using two gelation agents, namely citric acid and propylene oxide. The developed sol-gel preparation methods using these gelation agents are simpler, less time consuming and above all, environmentally benign. Propylene oxide based Fe-Al-Cu system demonstrated significant WGS performance which was comparable to the previously reported Fe-Al-

Cu sol-gel combination [19, 20]. This propylene oxide based gelation technique is a simple “one-pot” technique and requires neither catalyst washing nor critical pH adjustments. Copper is an essential promoter in the Fe-based catalysts [5, 10, 13, 19], therefore, the effect of Cu-loading for catalysts prepared by this technique has also been investigated. In addition to steady-state experiments for WGS activity evaluation, numerous catalyst characterization techniques such as X-ray diffraction (XRD), temperature-programmed reduction (TPR) and oxidation (TPO), X-ray photoelectron spectroscopy (XPS), diffuse reflectance infrared Fourier transform spectroscopy

(DRIFTS) and Brunauer-Emmett-Teller method (BET) were implemented in order to evaluate the surface as well as bulk properties of the catalysts which could be correlated to their WGS performance.

4.2 Experimental Procedure

4.2.1 Catalyst preparation

Fe-based catalysts with Cu and Al employed as promoters were prepared through a sol-gel route using propylene oxide and citric acid as gelation agents. Aluminum nitrate

[Al(NO3)3.9H2O] (crystals, Fisher), iron (III) nitrate nanohydrate [Fe(NO3)3.9H2O]

((98+%) ACS reagent, Sigma-Aldrich) and copper (III) nitrate trihydrate [CuN2O6.3H2O]

(Fluka) were used as metal precursors in catalyst preparation.

For preparation of catalysts with the propylene oxide route, metal salt(s) in quantities to yield the desired atomic ratios of Fe, Al and Cu in the final catalyst were dissolved in ethanol at ambient temperature. Propylene oxide was added at once to this solution in excess quantities while continuously stirring the solution. The addition of propylene oxide resulted in rapid heat evolution accompanied by gel formation. The gel was kept at 50°C for 4 hours and then dried overnight at 110°C in a convection oven.

Dried samples were ground to a fine powder and calcined in air for 4 hours at 450°C

(heating rate: 10°C/min). The samples were crushed to a fine powder following calcination. The propylene oxide method was utilized for preparation of Fe-only, Fe-Cu bi-metallic and Fe-Al-Cu ternary samples. The atomic ratio of Fe to Cu in Fe-Cu bi- metallic catalyst was 5, whereas Fe to Cu ratio was varied in Fe-Al-Cu catalysts

(Fe/Cu=10, 5 and 2) while keeping Fe/Al=10 constant.

Another gelation agent that was utilized for the preparation of Fe-based WGS catalysts was citric acid. For the preparation of catalyst sample through the citric acid

20 route, metal salts to yield the desired atomic ratios of the metals and citric acid were dissolved in a 1:1 (wt/wt) solution of ethanol and water at ambient temperature while stirring continuously. The ratio of moles of citric acid to the moles of total nitrate (NO3) contained in the metal salts was kept constant at 1. The solution was kept at 80°C for 3 hours to slowly drive off ethanol and then dried in a convection oven overnight at 110°C.

The sample was then calcined in air at 450°C (heating rate: 10°C/min) for 4 hours. The sample was ground to a fine powder following calcination.

4.2.2 Catalytic activity testing

All the steady-state experiments were conducted in a fixed-bed flow reactor system using 0.25” OD stainless steel reactor. Catalyst samples were held in place by the stainless steel frit and a quartz wool plug and the temperature of the reactor bed was measured by an Omega K-type thermocouple. The reactor was placed inside a homemade, resistively heated furnace, the temperature of which was controlled by an Omega

(model CS232) PID temperature controller. Reagent gases were supplied from Praxair and were used without any further treatment. Independent gas mass flow controllers

(Brooks, 5850E) connected to an electronics control box (Brooks, 0154) were used to control gas flows to the system. Water vapor was supplied to the system by bubbling nitrogen through a heated bubbler containing de-ionized water. The temperature of the bubbler was monitored and controlled precisely by a temperature controller. Gas lines in contact with water vapor containing stream were heated using heating tapes in order to avoid water condensation in the lines.

Unless otherwise stated, 100 mg of catalyst sample was used for each run and the experiments were carried out at a WHSV (weight hourly space velocity) of 0.06 m3 (g cat)-1 h-1. In addition, steady-state experiments were conducted on equal surface area to ensure that the observed differences were not simply due to differences in surface areas.

In order to assure that the activity testing experiments were run in the kinetically controlled regime, reaction conditions were maintained away from equilibrium in all of the activity testing experiments.

Prior to catalytic activity testing, catalyst samples were pretreated in-situ in N2

(67.5 sccm) for 30 minutes at 350oC and then reduced at 350o C for 2 hours in syngas mixture (10% CO, 10% (or 20%) H2O, 7.5% H2, 5% CO2 balanced with 67.5% N2).

Finally, reaction studies were performed over the wide temperature range (250oC-400oC) in the above mentioned syngas mixture.

4.2.3 Catalyst characterization

The specific surface areas of the samples were measured on Micromeritics ASAP

2010 surface area analyzer using N2 physisorption technique. Prior to analysis, samples were degassed overnight at 130oC under a vacuum of 3µmHg. The specific surface areas

22 were determined by the Brunauer-Emmett-Teller (BET) method using the adsorption branch of the isotherm. Pore volumes of the samples were determined by the Barret-

Joiner-Halenda (BJH) method using the adsorption branches of the N2 physisorption isotherms.

X-ray diffraction (XRD) patterns of the pristine samples were collected on a

Rigaku X-ray diffractometer (X-ray source: Cu Kα radiation, λ=1.5418 Å) operated at 40 kV and 25 mA. The diffraction patterns were collected in the range of 20o to 60o Bragg angle values. The X-ray diffraction patterns of the reduced catalyst samples were collected on a Bruker D8 Advance X-ray diffractometer equipped with a capillary sample holder. Following reduction treatment, the samples were transferred to an Ar-purged glove box that contains minimal residual air (< 5ppm). The samples were transferred to capillary sample tubes (0.2 mm OD) and the ends of the capillary tube were sealed inside the glove box with glass sealant. The samples were then transferred air-free to capillary sample holder accessory on the diffractometer. The diffraction patterns were collected in

the 20°-70° Bragg angle range using a Cu Kα X-ray source (λ=1.5418 Å) operated at 40 kV and 50 mA.

X-ray photoelectron spectroscopy was used for the investigation of the chemical states of copper and iron in the calcined (pristine) and reduced samples. The samples referred to as ‘pristine’ were ground into double sided carbon tape directly following calcination and transferred to the vacuum chamber of the XPS instrument. The reaction system described in Section 4.2.2 was used for preparing the pretreated samples. The

‘reduced’ samples were prepared through treatment in syngas mixture containing 10%

CO, 10% H2O, 7.5% H2 and 5% CO2 in N2 (100 sccm) at 350°C for 2 hours. Following

23 pretreatment, the reactor was flushed with N2 for 30 minutes at 350°C and the reactors were allowed to cool down to room temperature under N2 flow before sealing under N2.

The sealed reactors were transferred to a glove box kept under Ar atmosphere with moisture content lower than 5 ppm. The samples were ground into a carbon tape inside the glove box and then transferred air-free to the vacuum chamber of the XPS instrument.

The X-ray photoelectron spectra were collected on a Kratos Axis Ultra spectrometer

using Kα radiation through Al anode operated at 13kV and 10mA. When necessary, the charge neutralizer was operated at a current of 2.1A and a filament bias of 1.3V.

Concurrent sweeps of Cu 2p, Fe 2p, O 1s and C 1s envelopes were collected. The collected spectra were corrected using C 1s envelope located at 284.5 eV. For comparability of the signal intensities of the envelopes, counts per second, were divided by the number of sweeps and the transmission value at the specific binding energy.

Background correction and peak fitting were done using the XPSPeak 4.1 software.

The adsorption and thermal transformations of CO on Fe-Al-Cu based catalysts was investigated using an FT-IR spectrometer (Thermo Nicolet 6700) equipped with an

MCT detector and a DRIFTS chamber with ZnSe windows. The DRIFT spectra were collected in the mid-IR range at a nominal resolution of 4 cm-1 and the final spectra were averaged over 500 scans. For each experiment, catalyst sample was reduced in syngas mixture in a reactor system at 350o C for 2 hours as mentioned in Section 4.2.1. The reduced catalysts were recovered under N2 atmosphere and transferred to DRIFTS cell to acquire spectra. The sample was pretreated in He (30 sccm) at 400o C for 30 minutes, followed by background spectra collection in He (30 sccm) from 400oC to room temperature while cooling down the sample. Helium was switched off after background

24 collection and system was flushed with CO (30 sccm) for 30 minutes at room temperature. Catalyst temperature was raised under the flow of CO (30 sccm) from room temperature to 400o C stepwise and spectra were collected at regular temperature intervals.

Transmission electron microscopy (TEM) characterization was performed using a

Phillips Tecnai F20 instrument with FEG operated at a voltage of 200 kV. The catalyst images were collected in brightfield mode. Catalyst sample was dispersed in ethanol and the mixture was sonicated for 20 minutes. The sample was then deposited onto 200 mesh copper grid coated with lacey carbon.

Temperature-programmed reduction (TPR) and temperature-programmed re- oxidation experiments were conducted in order to provide insights to the oxidation/reduction characteristics of the samples. For the temperature-programmed reduction experiments, 50 mg of calcined catalyst sample was packed in a 0.25” O.D. U-

o tube quartz reactor. Catalyst was pretreated in He (35 sccm) at 350 C for 30 min and then cooled down to room temperature under the same atmosphere. Finally, temperature of the catalyst bed was raised to 880oC at 10oC/min under the flow of the reducing agent,

5%H2/N2 (20 sccm). Hydrogen consumption during the temperature ramp was monitored by a thermal conductivity detector (TCD).

For the temperature-programmed re-oxidation experiments, 100 mg of sample was packed in a fixed bed quartz reactor with a quartz frit, which is placed inside a fast response furnace (Carbolite, MTF 10/15/130). The dead volume of the reactor was filled with quartz wool. The reactor effluents during pretreatment and temperature-programmed re-oxidation stages were monitored via a residual gas analyzer (MKS – Cirrus II)

25 operated in scanning ion mode. An electron multiplier detector was used to trace mass signals (m/z) 2, 16, 18, 32 and 44. The gas lines from the reactor to the mass spectrometer capillary inlet were heated via a heating cord to prevent condensation during pretreatment and the experiment. Each experiment consisted of a reduction stage where the sample was heated at 10°C/min to 300°C under He atmosphere and then, reduced with 5%H2/He (40 sccm) at the same temperature for an hour. Reduction was followed by He flush (40 sccm) at 300°C prior to cooling to room temperature under helium. Once the sample was at room temperature, 5% O2/He (40 sccm) introduced to the reactor. The mass traces on the mass spectrometer were allowed to stabilize at room temperature and then, a linear temperature program from 30°C to 450°C was enacted with isothermal stages lasting 20 minutes at 30°C and 450°C. The same temperature- programmed re-oxidation experiment was repeated with different heating rates for the determination of activation energies of re-oxidation of Fe-Al-Cu samples. The samples were treated following the same pretreatment procedure and heated at 10, 14 or 18°C/min during the temperature-programmed re-oxidation stage. Kissinger’s method [63] was used for estimating the activation energies of re-oxidation. Kissinger’s method correlates the temperature at which the maximum rate of oxidant consumption occurs with the activation energy for re-oxidation through a series of temperature-programmed experiments ran at different heating rates. The temperature at which the maximum rate of oxidant consumption occurs corresponds to the sample temperature at which the deflection of the oxidant consumption peak takes place (Tm). The activation energy can

2 be estimated through a plot of 1/Tm versus ln (Φ/Tm ) which yields a straight line with the

26 slope being equal to –Ea/R where, Φ is the heating rate in K/min; -Ea is the activation energy in J/mol and; R is the universal gas constant in J/K.mol.

4.3 Results and Discussion

4.3.1 Effect of gelation agent on WGS activity

In our previous publications, we have reported the development of Cr-free Fe-Al-

Cu catalyst formulations which exhibited superior WGS performance over the commercial Fe-Cr catalyst in a wide temperature range [19, 20]. We also showed that the preparation method played a key role on the WGS performance of these catalysts. Our initial work involved catalyst preparation through co-precipitation of Fe and Al from nitrate precursors followed by impregnation of Cu onto the calcined precipitate (2-step) and co-precipitation of Fe-Al-Cu using nitrate precursors and sodium carbonate as the precipitating agent (1-step) [18]. In a later publication, we reported catalyst development through a ‘one-pot’ sol-gel-precipitation route using organic precursors [19]. The Fe-Al-

Cu samples prepared through this sol-gel/precipitation route showed significant enhancement in the WGS activity compared to the 2-step and 1-step routes. However, this sol-gel/precipitation method required critical pH adjustment, which was achieved through addition of NaOH and involved extensive catalyst washing to remove the sodium ions afterwards.

To address the issues associated with catalyst preparation, we have investigated the applicability of a gelation agent-assisted sol-gel ‘one-pot’ route using propylene oxide

27 as the gelation agent. Fe-only, Fe-Cu and Fe-Al-Cu samples were prepared by the propylene oxide assisted sol-gel route. The monometallic Fe catalyst prepared by propylene oxide-assisted sol-gel route exhibited a reasonable CO conversion of 30% at

400oC in a syngas mixture. The textural promoter (Al) and structural promoter (Cu) were incorporated in the catalysts in order to improve their WGS activity. Al and Cu have been reported to be effective promoters [5, 10, 11, 13, 17, 19, 20]. Among the promotional effects that are reported are structural and textural promotion, which include improved surface area, resistance to sintering and change in the electronic properties of the magnetite phase. As expected, further improvement was achieved in the WGS activity by incorporation of Cu and Al in the catalysts prepared by propylene oxides-assisted sol gel technique, achieving CO conversions of 40% and 57% over Fe-Cu (Fe/Cu=5) and Fe-Al-

Cu (Fe/Al=10 and Fe/Cu=5), respectively. The equilibrium CO conversion under these conditions is 64%.

Similar Fe-based water gas shift catalysts were also prepared by using another commonly used gelation agent, citric acid, and tested with regards to their activity in the syngas mixture. The Fe-only catalyst prepared through the citric acid route showed slightly higher WGS activity than the Fe-only catalyst prepared through the propylene oxide route however, the activity was not stable with time-on-stream. Further investigation of the WGS activity of Fe-Al-Cu (Fe/Al=10 and Fe/Cu=5) catalyst prepared through the citric acid route showed that, the activity of this catalyst was not stable either and the initial CO conversion of 40% over this catalyst decreased to less than 20% with time on stream over a period of 10 hours. Although the citric acid-assisted route was shown to be unsuitable for preparation of active and stable Fe-Al-Cu catalysts, the initial

28 catalyst testing results showed that it was possible to prepare active and stable Fe-Al-Cu catalysts through the propylene oxide-assisted sol-gel route. Therefore, the structural properties of Fe-based catalysts prepared through the propylene oxide route and the effect of copper loading on their performance was further investigated

4.3.2 Effect of Cu loading on the WGS activity

Sol-gel Fe-based catalysts with different Cu loadings (Fe/Cu = 10, 5 and 2) and constant aluminum loadings (Fe/Al = 10) were prepared using the propylene oxide- assisted route in order to investigate the effect of copper loading on the WGS activity of these catalysts and to elucidate the structural changes induced by copper incorporation.

Table 4.1 presents the specific surface areas and pore volumes of these Fe-Al-Cu catalysts. The surface area was found to be strongly dependent on copper loading and it decreased with increasing Cu content of the catalysts. A similar trend was observed in pore volume. As it will be discussed in the following sections, incorporation of copper in

Fe-based catalysts resulted in significant changes in the catalyst structure, especially in the major crystalline phases and particle size, as well as catalytic performance.

BJH adsorption BET surface area Catalyst pore volume (m2/g) (cm3/g)

Fe/Cu = 10 99 0.1830

Fe/Cu = 5 46 0.1669

Fe/Cu = 2 25 0.0421

Table 4.1: BET surface area, and pore volume of Fe-Al-Cu catalysts at various Cu- loadings

4.3.2.1 Catalyst Activity

Figure 4.1a shows the WGS activity of the three Fe-Al-Cu samples a wide temperature range (250-400°C) in syngas mixture at WHSV of 0.06 m3 (g cat)-1 h-1 at a steam-to-carbon (S/C) feed ratio of 1. WGS activity was significantly improved when Cu content was increased from Fe/Cu=10 to 5. The Fe-Al-Cu (Fe/Cu=5) sample was able to achieve CO conversion in excess of 55% in a broad temperature window of 300-400oC.

The equilibrium CO conversion at 400oC under these reaction conditions was 64%.

However, Fe-Al-Cu (Fe/Cu=10) sample was able to achieve a maximum CO conversion of slightly lower than 40%. Further increase in Cu content affected WGS activity adversely, resulting in considerable drop in CO conversion. This catalyst was able to achieve only 27% CO conversion in 300-400°C. When a commercial Fe-Cr catalyst was tested under identical conditions, the CO conversions achieved were much lower, even when compared to the least active of the formulations tested, i.e., Fe-Al-Cu (Fe/Cu=2).

Since the catalysts with different Cu loadings were compared at equal WHSV, a valid question would be if the differences in activities could be due to the differences in surface areas. To address this question, WGS activity of Fe-Al-Cu (Fe/Cu=2) and Fe-Al-

Cu (Fe/Cu=5) samples were also tested at 400°C on an equal surface area basis and the

Fe-Al-Cu (Fe/Cu=2) sample was found to exhibit significantly lower CO conversion than

Fe-Al-Cu (Fe/Cu=5). The CO conversion over Fe-Al-Cu (Fe/Cu=2) sample was 30%, whereas it was 41% over Fe-Al-Cu (Fe/Cu=5) at 400oC on equal surface area basis.

Therefore, decrease in CO conversion over Fe-Al-Cu (Fe/Cu=2) cannot be explained with the decrease in surface area.

The Fe-Al-Cu (Fe/Cu=5) catalyst was further tested for stability. The inset to

Figure 4.1a shows CO conversion over this catalyst during WGS in syngas mixture as a function of time-on-stream at 400oC. The catalyst was observed to exhibit stable WGS activity with time-on-stream over a period of 40 hours.

The WGS activity of the Fe-Al-Cu catalysts was further investigated at a higher steam-to-carbon ratio of the reaction feed. Figure 4.1b shows the CO conversions achieved as a function of temperature over Fe-Al-Cu catalysts during WGS reaction at a steam-to-carbon ratio of 2. At low temperatures no significant effect of increasing steam- to-carbon ratio on the WGS activity was observed; however, at higher temperatures, increased steam content of the feed resulted in a significant increase in CO conversion.

Although an increase in the stream content resulted in a shift in CO conversion curves to higher values over all of the samples by accelerating the forward reaction, it had a more pronounced effect over Fe-Al-Cu (Fe/Cu=10) and Fe-Al-Cu (Fe/Cu=2) than Fe-Al-Cu

(Fe/Cu=5). At 400°C, the CO conversion over Fe-Al-Cu (Fe/Cu=5) increased from 57% to 72% with increasing steam-to-carbon ratio, whereas the CO conversions over Fe-Al-

Cu (Fe/Cu=10) and Fe-Al-Cu (Fe/Cu=2) almost doubled. As it will be discussed in

Section 4.3.2.6, the re-oxidation of Fe-Al-Cu (Fe/Cu=10) and Fe-Al-Cu (Fe/Cu=2) is expected to proceed at a much slower rate than the re-oxidation of Fe-Al-Cu (Fe/Cu=5) during reaction. The fact that Fe-Al-Cu (Fe/Cu=10) and Fe-Al-Cu (Fe/Cu=2) catalysts benefit more from the increased steam content of the feed as compared to Fe-Al-Cu

(Fe/Cu=5) implies that the re-oxidation of surface through water splitting during WGS reaction is the rate limiting step over these catalysts under these conditions.

Figure 4.1: WGS activity (CO% conversion) of Fe-Al-Cu catalysts prepared by propylene oxide-assisted sol-gel method with different Cu loadings at (a) S/C = 1 and (b) S/C = 2.

Inset: time-on-stream performance of Fe-Al-Cu (Fe/Cu=5) at 400o C at S/C = 1. Reaction

Conditions: [CO] = 10%, [H2O] = 10% or 20%, [CO2] = 5%, [H2] = 7.5% and balance

3 -1 -1 N2, WHSV = 0.06 m (g cat) h , P = 1 atm.

4.3.2.2 X-ray diffraction

X-ray diffraction was used to examine the effect of copper incorporation and copper loading on the crystal phases formed during catalyst preparation. The X-ray diffraction pattern of the Fe-only sample showed the presence of α-Fe2O3 (hematite) phase, as evidenced by diffraction features located at Bragg angle (2θ) values of at 24.2o,

33.2o, 35.7o, 40.9o and 49.5o (Figure 4.2a) which can be associated with (012), (104),

(110), (113) and (024) crystal planes, respectively. Hematite phase disappeared with the incorporation of Cu into the iron oxide matrix. The X-ray diffraction pattern of Fe-Cu

(Fe/Cu=5) reflected the formation of γ-Fe2O3 (maghemite) phase instead of hematite phase. The diffraction lines at Bragg angle (2θ) values of 30.1o, 35.6o and 43.2o (Figure

4.2b) were associated with (220), (311) and (400) planes of maghemite, respectively.

This maghemite phase was retained even after the addition of Al (Figure 4.2c-e).

Maghemite is considered to be more active than hematite phase for WGS [11]. In addition, maghemite phase is in an imperfect cubic spinel structure, which is similar to the WGS active phase, Fe3O4 (magnetite) [6, 11, 73, 74]. This imperfect structure of maghemite allows better incorporation of active promoters than hexagonal hematite phase. Consequently, active promoter incorporated in maghemite is less prone to sintering effects [11, 75]. Also Figure 4.2a-c showed that the addition of promoters lead to broadening of diffraction lines which could be correlated to a decrease in the crystallite size.

The X-ray diffraction patterns of Fe-Al-Cu (Fe/Cu=10) and Fe-Al-Cu (Fe/Cu=5) samples (Figure 4.2c and 4.2d, respectively) showed diffraction lines located at 2θ values

34 of 30.1o, 35.6o and 43.2o which can be associated with (220), (311) and (400) planes of maghemite. No other phases that could be related to the presence of copper in the structure were detected in either of the samples. Unlike these two catalysts, the sample with the highest amount of copper loading (Fe/Cu=2) consisted of peaks that could be associated with maghemite as well as a separate CuO phase, which was identified through peaks located at 38.6o and 48.6o. The development of a separate CuO phase in the Fe-Al-Cu (Fe/Cu=2) sample but not in Fe-Al-Cu (Fe/Cu=5) suggests that there is a certain amount of copper atoms that could be incorporated into the iron oxide matrix. We have previously observed the separation of copper in the form of CuO domains on Fe-Al-

Cu samples prepared through the sol-gel route and suggested that there was a limit on the amount of copper atoms that could be incorporated into the iron oxide matrix [20]. We have further suggested that the amount of copper that could be incorporated into the iron oxide matrix lied in between Fe/Cu=5 and Fe/Cu=2, with the latter being an upper limit for Cu incorporation. Fe-Al-Cu samples prepared through the propylene oxide-assisted sol-gel route described in this study showed similar behavior, also suggesting that the maximum amount of copper that could be incorporated in the iron oxide matrix is a function of the charge structure of the system, rather than being dependent on the catalyst preparation route. Cu is an active promoter and modifies electronic properties of iron catalyst [8, 19, 20, 76] as long as it is incorporated in the iron oxide matrix. The optimum amount of Cu in iron oxide matrix facilitates the formation of WGS active maghemite phase and improves surface area of Fe-based catalysts, as shown earlier. However, metallic Cu on the surface is more prone to sintering at high operating temperatures [19,

20] which can result in poor WGS activity. The catalytic activity results presented in

Section 4.3.2.1 are in agreement with these findings.

The crystallite size in Fe-Al-Cu samples was calculated through the Scherrer equation. The crystallite size of Fe-Al-Cu samples increased with increasing copper loading (Table 4.2) indicating the formation of larger, ordered crystals with the incorporation of increasing amounts of copper into the structure. The formation of crystals with longer range order was also evidenced with the decrease in the specific surface area of the samples with increasing copper loading. The crystallite size calculations were in line with transmission electron microscopy (TEM) images, which show the presence of uniform, nano-sized (~12 nm) particles of Fe-Al-Cu sample, as shown in Figure 4.3. The lattice fringes for (400) plane of maghemite with a d-spacing of

1.70Å is visible in the TEM image of Fe-Al-Cu (Fe/Al=10, Fe/Cu=5) catalyst.

The capillary XRD technique was used to study the reduced Fe-Al-Cu samples in order to avoid oxidation of the catalyst surface during the sample transfer. The reduced samples were pretreated in nitrogen and then reduced in syngas mixture (refer to Section

4.2.2). Figure 4.4 shows XRD patterns of the reduced samples of Fe-Al-Cu at different

Cu-loadings. Magnetite phase (resulting from reduction of maghemite) was observed through peaks associated with (220), (311), (400), (511) and (440) planes at Bragg angle

(2θ) values of 30.1o, 35.4o, 43.3o, 57.4o, and 62.7o, respectively, in Fe-Al-Cu (Fe/Cu=10) and Fe-Al-Cu (Fe/Cu=5) samples. The XRD pattern of Fe-Al-Cu (Fe/Cu=2) sample, along with maghemite phase, also contained intense peaks of metallic copper (Cuo) at

43.2o and 50.3° which correspond to diffraction from (111) and (200) planes.

Crystallite Size Catalyst (nm)*

Fe/Cu = 10 11

Fe/Cu = 5 12

Fe/Cu = 2 15

Table 4.2: Crystallite size of Fe-Al-Cu catalysts at various Cu-loadings calculated from

the XRD patterns through Scherrer equation

Figure 4.2: XRD patterns of Fe-based samples. (a) Fe-only (b) Fe-Cu (Fe/Cu=10) (c) Fe-

Al-Cu (Fe/Cu = 10), (d) Fe-Al-Cu (Fe/Cu = 5) and (e) Fe-Al-Cu (Fe/Cu = 2).

Figure 4.3: Transmission electron micrograph (TEM) of Fe-Al-Cu (Fe/Cu = 5)

Figure 4.4: XRD patterns of Fe-Al-Cu samples reduced in syngas mixture (a) Fe/Cu=10,

(b)Fe/Cu=5 and (c)Fe/Cu=2

4.3.2.3 X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy was used to investigate the chemical states of iron and copper over the pristine samples and the changes that take place during catalyst activation via reduction in syngas. Figure 4.5 shows the Cu 2p envelopes of pristine and reduced samples of Fe-Al-Cu catalysts with Fe/Cu ratios of 10, 5 and 2. In order to prevent interaction of the reduced samples with ambient air, the reduced samples were transferred air-free from the reactor to the vacuum chamber of the XPS instrument under an inert atmosphere.

The Cu 2p envelopes of the pristine samples resembled each other regardless of the copper content of the catalyst. Figure 4.5 also shows the curve fitting for identification of photopeaks. In all of the samples, the Cu 2p3/2 component of Cu 2p envelope was located at 933.3eV with shake-up satellite lines at 941.0eV (ΔEB = 7.7 eV) and 943.1eV (ΔEB = 9.8eV). The spin orbit splitting (SOS) of Cu 2p3/2 and Cu 2p1/2 was

19.8eV, and the satellite peak for 2p1/2 was located at 961.7eV (ΔEB = 8.6 eV). The appearance of shake-up satellite peaks is common for paramagnetic species and the observed displacement in these together with the position of the photoelectron lines in the

Cu 2p region suggests the presence of Cu2+ ions in CuO environment as the major component [77, 78].

Following reduction treatment, lower oxidation state copper species were observed together with cupric oxide over all of the samples. It is well known that reduction of copper oxide occurs during irradiation with the X-ray beam in the XPS

41 instrument [79, 80] and it is worthwhile to note that, we have observed the presence of

Cu2+ over pristine samples under similar experimental conditions and irradiation periods in the XPS instrument as the sole copper species. It is plausible that interaction of copper species with iron oxide network leads to increased stability of the CuO species over Fe-

Al-Cu samples compared to pure cupric oxide. In line with the above discussion, the presence of copper species with lower oxidation states over the reduced samples is likely to be associated with the preceding treatment steps rather than reduction during X-ray irradiation.

Curve fitting of Cu 2p3/2 region of the reduced samples suggested the presence of two components –located at 933.3 eV and 932 eV- contributing to this photopeak in all of the reduced samples. Over Fe-Al-Cu (Fe/Cu = 5) and Fe-Al-Cu (Fe/Cu = 10) catalysts, the shake-up lines that were displaced by 7.8eV and 9.8eV with respect to 2p3/2 were also observed. The higher binding energy component together with the shake-up lines was associated with Cu2+ ions whose appearance have been attributed to stabilization through interaction with iron oxide matrix. The location of the lower binding energy component is characteristic of copper species at lower oxidation state, however differentiation of Cu1+ and Cu0 by the photoelectron lines is not possible as the difference in the core level binding energies of copper atoms in these environments is well within experimental uncertainty. Over Fe-Al-Cu (Fe/Cu = 2) the Cu 2p envelope was dominated by lower oxidation state copper species (i.e. Cu1+ or Cu0) with little contribution from a higher binding energy component. Although the binding energy difference between

+1 metallic copper and Cu ions in Cu2O environment is too low to allow spectral differentiation of these species, complementary evidence from the X-ray diffraction

42 patterns presented in Section 4.3.2.2 allows to associate this photopeak with the presence of metallic copper on reduced Fe-Al-Cu (Fe/Cu = 2). The relative contributions of CuO and the lower oxidation state component of the Cu 2p envelope were calculated using the areas under the fitted components. The relative contribution from Cu2+ to the Cu 2p envelopes of reduced Fe-Al-Cu (Fe/Cu=10), Fe-Al-Cu (Fe/Cu=5) and Fe-Al-Cu

(Fe/Cu=2) was 72%, 74% and 18%, respectively. The presence of higher amount of lower oxidation state components over Fe-Al-Cu (Fe/Cu=2) in comparison to Fe-Al-Cu

(Fe/Cu=5) and Fe-Al-Cu (Fe/Cu=10), can be attributed to the copper species which have not been incorporated into the iron oxide matrix over this catalyst. Deprived of the stabilizing effect of iron oxide matrix, this free CuO phase is more prone to reduction and sintering at elevated temperatures and harsher reducing conditions.

Figure 4.5: X-ray photoelectron spectra of Cu 2p region of pristine and reduced Fe-Al-Cu samples: (a) Fe/Cu = 10, (b) Fe/Cu = 5 and (c) Fe/Cu = 2

Figure 4.6 shows the X-ray photoelectron spectra in the Fe 2p region of pristine and reduced Fe-Al-Cu samples with the three different Fe/Cu ratios of 2, 5 and 10. The

X-ray photoelectron spectra of the pristine samples were similar and contributions of four different photopeaks were resolved. The photopeak located at 711.1 eV is characteristic of core level Fe 2p3/2 photoelectrons of ferric oxide. The 2p3/2 – 2p1/2 spin orbit splitting was 13.5 eV. The high degree of multiplet splitting of the Fe 2p core line resulted in the broad photopeaks observed in this region. Also observed were the satellite lines which were displaced by 8 eV. These satellite peaks were characteristic of Fe2O3. These satellite peaks are commonly used as fingerprints for different oxides of iron as the intensity and the position of the satellite peak is dependent on the environment of the Fe3+ ions.

However, Fe3+ satellites were not the only photoelectrons that have contributions in this

region in the case of Fe-Al-Cu samples. Since monochromatic Al Kα radiation which resulted in higher spectral resolution was used for obtaining well defined peak shapes, this region had contribution from Cu LMM Auger electrons at c.a. 719 eV. The core

3+ level Fe 2p spectra of Fe species in α-Fe2O3 and γ-Fe2O3 environments are almost identical to each other. McIntyre and Zetaruk [81] used multiplet splitting patterns of Fe

2p core levels for identification of different oxides of iron. Through curve fitting of the

Fe 2p3/2 peak these authors showed that the splitting of the two most intense components of the envelope was 1eV for γ-Fe2O3 which was 0.2 eV lower than α-Fe2O3. In line with their findings, the curve fitting of the Fe 2p core level spectra of the pristine Fe-Al-Cu samples using the same parameters that these authors have used yielded a displacement of 1 eV between the two most intense components (not shown). Although more indirect than XRD, X-ray photoelectron spectroscopy provides complementary evidence for the

45 presence of iron in the form of γ-Fe2O3 in the pristine samples. Following reduction treatment, the Fe 2p envelopes of all of the samples were similar to each other; the Fe

2p3/2 photopeak was located at 710.8 eV (SOS = 13.5 eV) with a weakly resolved shoulder that was displaced by 2.5 eV towards the lower binding energy side (Figure

4.6). Another important feature of the Fe 2p envelope of the reduced samples was the disappearance of the shake-up lines, further indicated the transformation of Fe2O3 to

Fe3O4 during reduction treatment. The X-ray photoelectron spectra suggested the presence of magnetite as the major phase and that the adopted reduction procedure resulted in partial reduction of iron as the presence of metallic iron (c.a. 707 eV) was not detected in any of the samples.

Figure 4.6: X-ray photoelectron spectra of Fe 2p region of pristine and reduced Fe-Al-Cu samples: (a) Fe/Cu = 10, (b) Fe/Cu = 5 and (c) Fe/Cu = 2

4.3.2.4 Temperature-programmed reduction

The effect of copper loading on the reduction characteristics of Fe-Al-Cu samples was investigated with temperature programmed reduction (TPR) in 80°C-880°C region.

Figure 4.7 shows the H2 consumption traces from these experiments. All of the samples exhibited a well-resolved consumption feature in 170-400°C region and another broad consumption band that corresponded to higher temperatures. The low temperature H2 consumption band, which can be associated with reduction of maghemite to magnetite, corresponded to the same temperature range for Fe-Al-Cu (Fe/Cu=10) and Fe-Al-Cu

(Fe/Cu=5) (Tmax =270°C and 261°C, respectively), whereas the peak shifted to higher temperatures (Tmax =296°C) over Fe-Al-Cu (Fe/Cu=2) sample. The shoulder peak at

o 230 C in H2 consumption trace of Fe-Al-Cu (Fe/Cu=2) was assigned to reduction of CuO to metallic Cu, presence of which has been confirmed through XRD pattern of the reduced sample. The reduction peaks above 600o C (in all samples) were assigned to over-reduction of active magnetite to WGS-inactive sub-oxides or to metallic Fe. The

TPR studies indicated that optimum Cu content (Fe/Cu=5) eased the reduction characteristic of iron oxide matrix (here, maghemite) to form active magnetite phase, whereas excess Cu content retarded the formation of the active phase during reduction.

The segregation of a separate CuO phase was also demonstrated in these experiments.

Figure 4.7: Temperature-programmed reduction profiles of Fe-Al-Cu catalysts.

4.3.2.5 In situ DRIFTS studies using CO

The interaction of CO with Fe-Al-Cu catalysts was investigated using in-situ

DRIFTS. Figure 4.8 shows the DRIFTS spectra of CO adsorbed on the reduced Fe-Al-Cu sample in the temperature range of 30o C- 400oC. Bands in the range of 1000-1700 cm-1 correspond to OCO symmetric as well as asymmetric stretching vibrations. However, carbonate and formate both contain OCO stretching and therefore it is difficult to distinguish between them. One possible way to is look for bands in the region 2800-3000 cm-1 for C-H stretching [82] as formate species contain both OCO and C-H stretching.

Fe-Al-Cu sample showed C-H stretching vibrations at 2845 and 2950 cm-1, both could be due to bidentate formate or bidentate and bridging type formate species [36]. These C-H stretching features increased in intensity with increasing adsorption temperature. The bands at 2110 and 2173 cm-1 assigned to weakly adsorbed CO and gas phase CO,

-1 respectively. The doublet of weakly adsorbed CO2 was observed at 2327 and 2360 cm

[18].

The bands in between 3500-3800 cm-1 were characteristic of OH groups associated with the catalyst. The bands at 3643 and 3685 cm-1 were assigned to Type II bridging OH group (geminal) [36, 83] on the “reduced” Fe-Al-Cu catalyst. Jacobs et al.

[36] reported that these Type II OH groups reacted with CO to form formate species for

Pt/CeO2 catalytic system. For Fe-Al-Cu, we observed that Type II OH groups diminished as temperature increased under the flow of CO while formate bands increased in intensity. Peaks at 3611 and 674 cm-1 were the characteristic of hydroxyl vibrations

50 associated with Cu [84]. The band at 3713 cm-1 was assigned to Type I OH group

(terminal) [36]. The OH stretching of adsorbed water was observed at 3400 cm-1 [85] at lower temperatures, which disappeared above 100oC.

While the presence of formate species may suggest an involvement as an intermediate in the reaction mechanism, it is also possible that they may be acting merely as spectators [86, 87], and that the redox pathway is more likely to be the mechanism in operation in WGS reactions over these catalysts [18, 19]. Temperature-programmed re- oxidation studies discussed in the next section also support this hypothesis. The catalyst re-oxidation studies suggest that a redox pathway may be the operative mechanism over these catalysts

Figure 4.8: DRIFTS study of CO interaction with Fe-Al-Cu (Fe/Cu=5)

4.3.2.6 Temperature-Programmed Re-oxidation

Re-oxidation of the catalyst surface constitutes an important step in water-gas shift reaction. Usually, surface oxygen vacancies are created through the oxidation of adsorbed CO and these vacancies act as sites for water adsorption to replenish surface oxygen [18, 88, 89]. Temperature programmed re-oxidation experiments were performed following reduction of Fe-Al-Cu samples to study the re-oxidation kinetics of the catalysts and to determine the apparent activation energies of re-oxidation of the Fe-Al-

Cu samples with Fe/Cu ratios of 2, 5 and 10.

The samples were reduced in 5%H2/He at 300°C and then, re-oxidized in

5%O2/He while being heated at 10, 14 or 18°C/min to 450°C. The reduction temperature for these experiments was chosen in light of the TPR results discussed in Section 4.3.2.4, to enable reduction of γ-Fe2O3 to Fe3O4, but to prevent magnetite from being over- reduced. Figure 4.9 shows oxidant consumption profiles during the temperature programmed re-oxidation of reduced Fe-Al-Cu catalysts. At a heating rate of 10°C/min, maximum oxidant consumption temperature corresponded to 201°C, 169°C and 180°C for Fe-Al-Cu (Fe/Cu = 2) Fe-Al-Cu (Fe/Cu=5) and Fe-Al-Cu (Fe/Cu=10), respectively.

For Fe-Al-Cu (Fe/Cu = 2), the maximum oxidant consumption temperature shifted from

201°C to 211°C as the heating rate was increased from 10°C/min to 18°C/min (Figure

4.9c). A shoulder in oxygen consumption profile was also resolved at 135°C. In line with the H2-TPR results, this O2 consumption feature could be associated with oxidation of

53 metallic copper species on the surface which have not been incorporated to the iron oxide matrix. Over Fe-Al-Cu (Fe/Cu = 5) and Fe-Al-Cu (Fe/Cu = 10), the temperature for maximum oxidant consumption shifted from 168°C to 200°C and from 180°C to 188°C, respectively. The oxidant consumption traces were composed of a single component, therefore no evidence for the formation of metallic copper domains through diffusion of copper species from the iron oxide matrix to the surface during reduction was present over these samples.

In a plot of oxidant consumption as a function of temperature, the deflection point gives the temperature (Tm) at which the oxidant consumption rate is a maximum. The temperature of maximum oxidant consumption rate shifted to higher temperatures as the

2 heating rate (Φ) was increased and a plot of 1/Tm versus ln(Φ/Tm ) was used to determine the activation energy for the re-oxidation process. In line with the above discussion on the selection of the reduction temperature, the apparent activation energies for re- oxidation calculated through Arrhenius plots were linked to re-oxidation of Fe3O4. The

2 inset to Figure 4.9c shows the plot of 1/Tm versus ln (Φ/Tm ) which gives a straight line with a slope of -12.48 that corresponded to activation energy of 103.76kJ/mol for re- oxidation of Fe-Al-Cu (Fe/Cu=2). Similarly, the activation energies for re-oxidation were calculated as 25kJ/mol and 115kJ/mol for Fe-Al-Cu (Fe/Cu=5) and Fe-Al-Cu

(Fe/Cu=10), respectively. Based on the activation energies, the re-oxidation of Fe-Al-Cu

(Fe/Cu = 2) and Fe-Al-Cu (Fe/Cu = 10) is expected to take place at a much slower rate than Fe-Al-Cu (Fe/Cu = 5).

Iron oxide-chromium catalysts have been shown to have apparent WGS activation energies in the order of 110-130 kJ/mol [8, 90, 91]. Hutchings et al. [8, 13] further

54 showed that incorporation of copper into the Fe-Cr system reduced the apparent activation energy for WGS reaction and reported activation energies in the order of 80 kJ/mol for this system. The apparent activation energies of re-oxidation of Fe-Al-Cu samples showed that while the re-oxidation of Fe-Al-Cu (Fe/Cu=2) and Fe-Al-Cu

(Fe/Cu=10) samples proceed at a rate comparable to (if not slower than) the reduction of the surface during oxidation of adsorbed CO with surface oxygen, the re-oxidation of Fe-

Al-Cu (Fe/Cu=5) proceeded at a much faster rate than the reduction. Therefore, WGS reaction over the Fe-Al-Cu (Fe/Cu=5) sample would solely be controlled by reduction kinetics of the surface whereas for the Fe-Al-Cu (Fe/Cu=2) and Fe-Al-Cu (Fe/Cu=10) availability of surface oxygen for CO oxidation would be a rate determining factor.

These results are consistent with the reaction experiments which showed a much more pronounced improvement over the Fe-Al-Cu (Fe/Cu=2) and Fe-Al-Cu (Fe/Cu=10) when steam-to-CO ratio was doubled. The catalysts which had the higher activation energy barriers for re-oxidation benefited more from the increased water concentration than the catalyst which had a lower activation energy for re-oxidation with water. Although the presence of formate species have been observed in the DRIFT spectra which does not allow us to rule out the formate mechanism during WGS reaction, the catalyst re- oxidation studies suggest that a red-ox pathway may be the operative mechanism over these catalysts.

4.4 Conclusions

Gelation agent-assisted one-pot sol-gel technique was used for preparation of Fe- based WGS catalysts. Propylene oxide and citric acid were investigated as the gelation agents. Propylene oxide emerged as a more suitable gelation agent than citric acid. The effect of Cu-loading study showed that Fe-Al-Cu (Fe/Cu=5) was the optimum Cu-loading and the excess amount of copper in Fe-Al-Cu (Fe/Cu=2) caused significant decrease in the WGS activity. With increasing copper content, segregation of a copper-containing phase, which is more prone to reduction under reaction conditions, was observed through

XRD and TPR.

The apparent activation energies of re-oxidation of Fe-Al-Cu (Fe/Cu=10) and Fe-

Al-Cu (Fe/Cu=2) were found to be much higher than Fe-Al-Cu (Fe/Cu=5). The apparent activation energies of re-oxidation of the Fe-Al-Cu catalysts suggested that WGS reaction over Fe-Al-Cu (Fe/Cu=5) was controlled by the rate of reduction of the surface through oxygen removal by CO, whereas for Fe-Al-Cu (Fe/Cu=10) and Fe-Al-Cu (Fe/2) reduction and re-oxidation of the surface would proceed at comparable rates. Increasing water concentration in the feed showed a more pronounced improvement in CO conversion over the catalysts with high activation energy barriers for re-oxidation.

* Information in this chapter is adapted/taken from P. Gawade, B. Mirkelamoglu, B. Tan and U.S. Ozkan, Journal of Molecular Catalysis A: Chemical 321, (2010), 61-70

CHAPTER 5

THE ROLE of SUPPORT MORPHOLOGY and IMPREGNATION MEDIUM on

WATER-GAS SHIFT ACTIVITY of CERIA-SUPPORTED COPPER CATALYSTS

5.1 Overview of Cu/CeO2 Catalyst: Effect of Ceria Morphology and Impregnation

Medium

In the present chapter, we report the effect of ceria nanostructure on the WGS activity of Cu catalysts. Two different nanostructures of ceria, nano-particles (NP) and nanorods (NR) were prepared using precipitation-hydrothermal method. Wet impregnation method was used to incorporate copper over ceria using, ethanol (E) or tetrahydrofuran (T) as the impregnation medium. To the best of our knowledge, the effect of CeO2 morphology on WGS reaction over Cu/CeO2 catalytic system has not been previously reported.

Steady-state reaction experiments showed that the catalysts supported over CeO2 nano-particles achieved significantly higher CO conversions than catalysts supported over nano-rods, which achieved only marginally higher conversions than the bare support. The catalysts were characterized by transmission electron microscopy, X-ray diffraction, X-ray photoelectron spectroscopy, N2O chemisorption and H2-temperature-

58 programmed reduction. Characterization of the catalysts showed that copper dispersion and catalytic activity were strongly correlated. Copper was observed to be better dispersed over CeO2 nano-particles, while larger, crystalline CuO domains with properties closer to that of bulk CuO were identified over CeO2 nano-rods. The finely dispersed copper species constitute the active sites for WGS reaction over Cu/CeO2 catalysts.

5.2 Experimental Procedure

5.2.1 Catalyst Preparation

Both of the CeO2 supports (nano-particles and nano-rods) were prepared via precipitation from an aqueous solution of 0.4 M cerium (III) nitrate hexahydrate

(99.999% trace metals basis, Sigma-Aldrich), by an aqueous solution of sodium hydroxide (2.7M) at room temperature under constant stirring. For the preparation of ceria nano-particles (NP), the precipitates were collected by filtration and washed with deionized water until the pH of the filtrate was 7. The particles were, then, dried at 110°C overnight in a convection oven followed by calcination at 400°C for 3 hours in air at

10oC/min. For the preparation of nano-rods (NR), the precursor solution and precipitates were transferred to a Pyrex© bottle, sealed and kept at 110°C for 24 hours. The precipitates were then filtered, washed, dried and calcined following the above- mentioned procedure for the preparation of the nano-particles.

The Cu/CeO2 catalysts were prepared through impregnation of the nano-structured

CeO2 supports with copper via a conventional wet-impregnation method. Two different organic solvents, namely ethanol (E) and tetrahydrofuran (T) were used as the impregnation media. For impregnation, 1.9 g of ceria support was added to a 40 ml solution of 0.04 M copper (II) nitrate tri-hydrate (98%, Fluka) in ethanol or tetrahydrofuran and kept overnight at room temperature while stirring continuously. The catalysts were dried at 100°C overnight in a convection oven and calcined at 400°C for 3 hours at 10oC/min under a flow of air.

The specific surface areas of the catalysts were measured using N2 physisorption technique on a Micromeritics ASAP 2020 accelerated surface area and porosimetry instrument. Before measurement, samples were degassed for 12 hours at 130°C under a vacuum better than 3 µmHg. The surface areas of the samples were determined using the

Brunauer-Emmett-Teller (BET) method and presented in Table 5.1.

Sample Surface Area (m2/g)

CeO2-NP 123

CeO2-NR 115

Cu/CeO2-NP(E) 112

Cu/CeO2-NP(T) 94

Cu/CeO2-NR(E) 106

Cu/CeO2-NR(T) 103

Table 5.1: Specific surface areas of CeO2-based samples calculated using BET method

5.2.2 Catalyst Activity Testing

The steady-state reaction experiments were conducted in a fixed-bed flow reactor system using ¼” OD stainless steel reactor with a stainless steel frit. The reactor was placed inside a resistively heated furnace and the reactor temperature was controlled by an Omega CSC232 PID temperature controller. Water vapor was supplied to the reactor system by bubbling nitrogen through a heated bubbler containing de-ionized water. Gas lines in contact with water-vapor-containing stream were heated in order to avoid water condensation in the lines. 100 mg of catalyst sample was packed inside the reactor for each run and the samples were pretreated in-situ in N2 (67.5 sccm) for 30 minutes at

o 350 C and then reduced in a syn-gas mixture with a composition of 10% CO, 10% H2O,

7.5% H2, 5% CO2 in N2 for 2 hours at 350°C.

Reaction studies were performed in 250-400°C range at a WHSV (weight hourly space velocity) of 0.06 m3 (g cat)-1 h-1. The feed syn-gas mixture constituted of 10% CO,

10% H2O, 7.5% H2, 5% CO2 and balance N2. The experiments were run in the kinetically controlled regime by maintaining the reaction conditions away from equilibrium during all of the activity testing experiments. The effluents from the reactor were monitored by an on-line gas chromatograph (Shimadzu, GC-14A series) equipped with molecular sieve

13X (5 ft x 1/8 in. SS, 60/80 mesh) and porapak Q (12 ft x 1/8 in. SS, 80/100 mesh) columns, and a thermal conductivity detector (TCD).

5.2.3 Catalyst Characterization

The X-ray diffraction patterns of the Cu/CeO2 catalysts as well as the CeO2 supports were collected in the 20-60° diffraction angle range on a Rigaku X-ray diffractometer using Cu Kα radiation (λ=1.5418 Å) through a tube operated at 40 kV and 25 mA. The transformation of crystal phases during reduction of the catalysts was monitored by in- situ XRD. A Bruker D8 Advance X-ray diffractometer using monochromatic Cu Kα radiation (λ=1.5418 Å) through a tube operated at 40 kV and 50mA and equipped with an

Anton Paar HTK1200 oven was used for collection of the X-ray diffraction patterns during in-situ reduction of Cu/CeO2 samples. The samples were heated from room temperature to 450°C under a flow of 5% H2/N2 (15 ccm) and diffraction patterns were collected at 30°C, 100°C, 200°C, 300°C, 400°C and 450°C. Following the collection of the XRD pattern at one temperature, the samples were heated under 5%H2/N2 flow at a rate of 12°C/min to the next set point where they were kept isothermally under the same flow for 30 minutes prior to collection of the XRD pattern. The identification of the crystalline phases through the collected diffraction patterns was done by using

International Center for Diffraction Data (ICDD) database.

Transmission electron microscopy (TEM) images were collected in brightfield mode on a Phillips Tecnai F20 instrument with FEG operated at a voltage of 200 kV. The samples were dispersed in ethanol and sonicated for 20 minutes before being deposited onto 200 mesh copper grid coated with lacey carbon/formvar.

The chemical states of cerium and copper in the Cu/CeO2 catalysts were investigated on a Kratos AXIS Ultra X-ray photoelectron spectrometer, using monochromatized Al Kα (1486.7eV) X-ray source operated at 13 kV and 10 mA. For each sample a survey scan was collected from 1400 to 0 eV. Next, concurrent sweeps of

Cu 2p, Ce 3d and O1s regions were collected at an electron pass energy of 20 eV.

Analysis of the collected data was performed on XPS Peak 4.1.

The temperature-programmed reduction profiles (H2-TPR) of CeO2 supports and

Cu/CeO2 catalysts were collected on an online mass spectrometer (Cirrus II, MKS

Instruments, 1–300 amu) operated in selective ion detection mode. 0.1 g of catalyst sample was packed inside a ¼” OD U-tube reactor made of quartz using quartz wool plugs. The catalyst was pretreated in He at 350oC for 30 minutes followed by cooling at room temperature in He. Then 5% H2/He (35 ccm) was introduced to the reactor at room temperature and the mass signals were allowed to stabilize for at least 30 minutes before ramping the temperature at 10°C/min to 550°C.

N2O chemisorption technique outlined earlier by Jensen et al. [92] was used to determine the dispersion of Cu over CeO2 supports by oxidation of surface Cu atoms to

Cu2O. For each run, 150 mg of sample was packed in a fixed bed quartz reactor with a quartz frit bed which is placed inside a fast response furnace (Carbolite, MTF 10/15/130) and pretreated with 5%H2/He (30 ccm) at 250°C for 2 hours. The reactor was then flushed with He at the same temperature and cooled under the same atmosphere. N2O chemisorption was carried out by introduction of a 4%N2O/He containing stream to the reactor at 40°C. The effluents in the m/z=12 to m/z=46 range were monitored via an on- line mass spectrometer (MKS – Cirrus II) and throughout the experiments N2O and N2

64 were the only species detected in the reactor effluent. The mass spectrometer was calibrated for instrumental sensitivity factors and the contribution of m/z=28 fragment of

N2O to the m/z=28 trace. For the quantitative determination of N2 evolving from the surface, a known amount of N2, which approximately matches the amount of N2 obtained during N2O chemisorption runs, was injected under the same flow conditions using an external sampling loop and a six-port valve using the same experimental set-up with a blank reactor. The number of moles of oxygen consumed is calculated through N2 evolution and gives the total amount of oxygen consumed via the surface reaction and oxygen uptake through Fickian diffusion of oxygen atoms from the surface to the bulk. A plot of Cu conversion versus square root of time gives a straight line once the surface is covered, and can be used for diffusion correction [92].

In-situ diffuse reflectance Fourier transform spectroscopy (DRIFTS) experiments were performed during WGS reaction on Cu/CeO2-NP(E) and Cu/CeO2-NR(E) catalysts to study the effect of support morphology on the formation of adsorption products and reaction intermediates. The FT-IR spectrometer (Thermo Nicolet 6700) was equipped with an MCT detector and a DRIFTS chamber with ZnSe windows. The DRIFTS spectra were collected in the mid-IR range at a nominal resolution of 4cm-1 and the presented spectra were averaged over 500 scans. The catalyst was pretreated in He (30 sccm) for 30

o min at 350 C, then reduced in 5%H2/He (30 sccm) at the same temperature for 1h.

Afterwards, the reaction chamber was purged with He (30 sccm) at 400oC for 30 min.

The background spectra were collected in He at regular temperature intervals while cooling down. The sample spectra were acquired under a stream of 1% CO and 1% H2O at temperature intervals of 50oC up to 400 oC.

5.3. Results and Discussion

5.3.1 Catalyst Activity

Figure 5.1a shows CO conversions as a function of temperature during water gas- shift reaction over Cu-based catalysts supported on CeO2 nano-particles. Copper was impregnated over CeO2 support via a wet-impregnation route using either ethanol (E) or

THF (T) as the impregnation medium. The bare CeO2-NP support was also tested under the same conditions for WGS activity and was observed to show no activity for WGS.

The best performing Cu/CeO2 catalyst achieved CO conversions that are comparable to our previously reported Fe-Al-Cu catalyst [93], reaching 36% and 49% at 400°C for THF and ethanol impregnated samples, respectively. Furthermore, the CO conversions observed over Cu/CeO2-NP(E) were, on the average, 40% higher than those achieved over the commercial Fe-Cr-Cu catalyst tested under the same conditions in 300-400°C range [19, 20].

The WGS activity (% CO conversion) achieved over CeO2-NR based samples was observed to be significantly lower than that of the CeO2-NP supported catalysts

(Figure 5.1b). The bare CeO2 nano-rod support showed no activity. Below 350°C, copper impregnation onto CeO2 nano-rods had no practical effect on the WGS activity and above that temperature the catalysts were able to achieve only marginally higher activity than the bare support. A maximum CO conversion of 15% was achieved over the nano-rod supported samples at 400°C, regardless of the impregnation medium. The rates of reaction were calculated at 400°C. In line with the above results, the NP-based samples

-2 -1 showed the highest rates of reaction (0.29 µmol m s for Cu/CeO2-NP(T) and 0.33

-2 -1 µmol m s for Cu/CeO2-NP(E)), while the rates of reaction over Cu/CeO2-NR(T) and

-2 -1 -2 -1 Cu/CeO2-NR(E) were calculated to be 0.10 µmol m s and 0.11 µmol m s , respectively.

The best performing catalyst, Cu/CeO2-NP(E), was further tested for stability with time-on-stream. Figure 5.1c shows the CO conversion attained over this catalyst as a function of time-on-stream at 400°C under the same feed conditions used in catalyst screening studies. The CO conversion over this catalyst was stable at 46% over a period of 55 hours.

As it will be discussed in the following sections, the support morphology has a significant effect on both the size of copper clusters on the surface, their redox behavior and resistance to sintering under reducing conditions that accounts for the significant difference in the WGS activity of Cu/CeO2 catalysts. The impregnation medium was also observed to have an effect on the copper segregation over these catalysts that could be correlated with the activity differences.

Figure 5.1: Activity of (a) CeO2-NP, (b) CeO2-NR based WGS catalysts: (●) CeO2, (■)

Cu/CeO2(E) and (▲) Cu/CeO2(T) and (c) time-on-stream activity of Cu/CeO2-NP(E) at

400°C. Reaction conditions: [CO] = 10%, [H2O] = 10%, [CO2] = 5%, [H2] = 7.5% and

3 -1 -1 balance N2, WHSV = 0.06 m (g cat) h , P = 1 atm.

5.3.2 Transmission Electron Microscopy

Figure 5.2 shows the TEM images of the cerium oxide nano-particles and nano- rods used as the supports. The CeO2-NR is comprised of nano-rods with diameters of

5nm. Figure 5.2b shows a high resolution TEM image of a representative CeO2 nano-rod with the lattice fringes for the [2 0 0] plane of CeO2 with an interplanar spacing of 2.7Å.

The CeO2-NP is comprised of nano-particles with diameters approximately 8nm (Figure

5.2c). Figure 5.2d shows a representative CeO2 nanoparticle with [1 1 1] plane orientation with an interplanar spacing of 3.1Å. Copper impregnation and subsequent heat treatments did not result in any change in the morphology of any of the supports that was observable through TEM (not shown).

Figure 5.2: TEM images of (a) CeO2-NR and (b) CeO2-NR at higher magnification, (c)

CeO2-NP and (d) CeO2-NP at higher magnification.

5.3.3 X-ray Photoelectron Spectroscopy

X-ray photoelectron spectroscopy was utilized to investigate the chemical states of cerium and copper. Figures 5.3a and 5.3b show the Ce 3d core level spectra of the

CeO2 nano-particles and nano-rods, respectively. The fairly complex structure of the core level Ce 3d spectra is due to the presence of both Ce4+ and Ce3+ derived modes, showing deviations from the ideal stoichiometry. The structures labeled as v (882.6 eV), v´´(889.1 eV) and v´´´ (897.1 eV) and the corresponding u modes – u (901.2 eV), u´´ (907.7 eV) and u´´´ (915.1 eV) – are characteristic of CeO2 while v0 (880.8 eV)and v´ (886.9 eV) and the corresponding u modes – u0 (899.3 eV) and u´ (905.8 eV) are associated with

Ce2O3 [94-96]. No changes were observed in the core level Ce 3d spectra following the impregnation of copper on to the CeO2 supports (Figures 3c through 3f). Table 2 shows the relative contributions of Ce3+ and Ce4+ that were calculated through peak fitting (not shown) and using the areas under the fitted components. The Ce3+-to-Ce4+ ratio was observed to be independent of the particle morphology and was fairly constant over the copper impregnated samples.

The core level Cu 2p spectra of the Cu/CeO2 catalysts are shown in Figure 5.4.

For the CeO2 nano-particle supported catalysts, curve fitting in the Cu 2p3/2 region showed the presence of four components located at, 932.3 eV, 933.2 eV, 941.0 eV and

943.1 eV. The lowest binding energy component can be assigned to Cu1+ or Cu0 species, however exact spectral identification of this species through the photoelectron lines is not possible as the difference in the core level binding energies of copper atoms in Cu2O and metallic copper environments is well within experimental uncertainty. The 2p3/2 -2p1/2

71 spin-orbit-splitting (SOS) for this mode was 19.8 eV. The higher binding energy components with 933.2 eV being the main photopeak and 941.0 eV and 943.1 eV being the shake-up satellite lines, are characteristic of Cu2+ ions in CuO environment [77, 78].

The Cu 2p1/2 component for this species was located at 953.0 eV (SOS=19.8 eV) with a satellite peak resolved at 961.7eV (ΔEB = 8.7 eV). Over the CeO2 nano-rod-supported catalysts, the Cu 2p3/2 region was composed of three components located at 933.2 eV,

941.0 eV and 943.1 eV, which are all, in line with the above discussion, assigned to Cu2+ species with no contribution from lower oxidation state copper species.

The relative contributions of the two copper species to the Cu envelope was calculated from the areas under the fitted components and presented in Table 5.2.

Reduction of copper oxide under the X-ray beam in the high vacuum environment of the

XPS instrument is a well known phenomenon [79, 80]. It is plausible that reduction of

CuO taking place under the X-ray beam resulted in the appearance of the lower oxidation state copper species over the CeO2 nano-particle-supported samples. As will be discussed below, the relative contribution of the reduced copper phase over these samples is well correlated with the dispersion of copper over the samples and the WGS activity. The nano-particle supported catalysts were comprised of well-dispersed copper domains, which tend to get reduced more easily than larger particles, while larger copper crystallites were present over the nano-rod supported sample. A control experiment was carried out by collecting the X-ray photoelectron spectra over bulk CuO using the same experimental parameters as the Cu/CeO2 catalysts and formation of the reduced copper phase was not observed over this sample (not shown). The absence of a reduced copper component over bulk CuO further supports the conclusion that the presence of a reduced

72 copper phase over the nano-particle supported catalysts was due to the reduction of small clusters of copper in the high vacuum environment of the XPS instrument.

Figure 5.3: Ce 3d region of the X-ray photoelectron spectra of (a) CeO2-NP, (b) CeO2-

NR, (c) Cu/CeO2-NP(E), (d) Cu/CeO2-NP(T), (e) Cu/CeO2-NR(E) and (f) Cu/CeO2-

NR(T)

Figure 5.4: Cu 2p region of the X-ray photoelectron spectra in (a) Cu/CeO2-NP(E), (b)

Cu/CeO2-NP(T), (c) Cu/CeO2-NR(E) and (d) Cu/CeO2-NR(T)

Ce3+ Ce4+ Cu2+ Cu1+(or Cu0)

CeO2-NP 49 % 51 % - -

CeO2-NR 48 % 52 % - -

Cu/CeO2-NP(E) 47 % 53 % 42 % 58 %

Cu/CeO2-NP(T) 49 % 51 % 55 % 45 %

Cu/CeO2-NR(E) 46 % 54 % 100 % -

Cu/CeO2-NR(T) 44 % 56 % 100 % -

Table 5.2: XPS analysis of the catalysts

5.3.4 X-ray diffraction

Figures 5.5a and 5.5b shows the X-ray diffraction patterns of the CeO2-NP and

CeO2-NR supports collected under ambient conditions. Both of the patterns show diffraction lines at 2Θ values of 28.8°, 32.9°, 47.4°, 55.9° and 58.6° that could be associated with [1 1 1], [2 0 0], [2 2 0], [3 1 1] and [2 2 2] planes of CeO2 in face- centered cubic cerianite structure (ICDD #34-394), respectively. A comparison of the relative intensities of the diffraction peaks show that the relative intensity of the [1 1 1]

(at 2Θ = 28.8º) peak is larger in the XRD pattern of nano-rods as compared to nano- particles. The ratio of the intensities of peaks associated with [1 1 1] (at 2Θ = 28.8º) and

[2 2 0] (at 2Θ = 47.4º) calculated from the XRD patterns of CeO2-NP and CeO2-NR showed that the [1 1 1] plane exposure in the nano-rods is 1.5 times higher than in nano- particles. This observation is in agreement with the literature [54, 62], where different ceria nano-structures have been reported to have different crystal plane exposures: [1 1 1] and [1 0 0] are the exposed crystal planes for nano-polyhedra, while nano-rods have [1 1

0] and [1 0 0] plane exposures. The crystallite sizes of the CeO2 samples were calculated from the broadening of the [1 1 1] diffraction peak of CeO2 (at 28.8°) using Scherrer’s method. The crystallite sizes of CeO2-NP and CeO2-NR were calculated as 8 and 6 nm, respectively, which are in line with TEM results.

Figures 5.5c and 5.5d show the X-ray diffraction patterns of Cu/CeO2-NR catalysts impregnated in ethanol or THF media, respectively. The cerianite phase was preserved after impregnation of copper on to the support material and following heat

77 treatment. In addition to the cerianite phase, both of the diffraction patterns showed the formation of copper oxide domains which are observed through diffraction lines at 2Θ values of 35.9° and 38.7°. These diffraction peaks can be associated with [1 1 ] and [1

1 1] planes of CuO in monoclinic tenorite structure (ICDD # 48-1548). Similarly,

Figures 5.5e and 5.5f show the diffraction patterns of Cu/CeO2-NP catalysts impregnated in ethanol and THF media, respectively. Over these catalysts cerianite phase was observed to be the major crystalline phase after copper impregnation. Unlike the nano- rod supported catalysts, diffraction peaks associated with a separate CuO phase were either non-existent or very weakly resolved over these catalysts suggesting the presence of smaller CuO particles.

Neither crystallite size calculations using Scherrer equation nor TEM images of the Cu/CeO2 catalysts (not shown) showed changes in the ceria particle size upon impregnation of the CeO2 supports with copper. Calculation of CuO crystallite size in

Cu/CeO2-NR sample was not attempted because of the low intensity of the CuO diffraction peaks.

The evolution of crystal phases during reduction of Cu/CeO2-NP(E), Cu/CeO2-

NP(T) and Cu/CeO2-NR(E) catalysts was studied by in-situ XRD as a function reduction temperature and the diffraction patterns collected during these experiments are presented in Figure 5.6. In agreement with the foregoing discussion, the peaks at 28.8º, 32.9º, 47.4º,

55.9º and 58.6º were associated with diffractions from [1 1 1], [2 0 0], [2 2 0], [3 1 1] and

[2 2 2] crystal planes of ceria (cerianite, ICDD #34-394), respectively. The CeO2 phase was observed to be the major crystalline phase present over all the catalysts throughout the temperature range studied and no contribution from sub-oxides or metallic cerium

78 was identified. In line with the XRD patterns collected under ambient conditions, an additional peak that has been associated with the presence of crystalline CuO domains

(tenorite, ICDD # 48-1548), was resolved at 35.9º [1 1 ] in the XRD patterns collected at 30ºC. As the reduction proceeds, a new diffraction peak was resolved at 43.5º above

200°C. This new peak can be associated with diffraction from [1 1 1] plane of Cuº (ICDD

# 4-836). This peak is well-resolved over Cu/CeO2-NR(E) and Cu/CeO2-NP(T) catalysts and grows in intensity up to 300ºC, suggesting the formation of larger, more ordered crystallites of metallic copper with increasing reduction temperature. The metallic copper phase is likely to be formed through the reduction of the crystalline CuO particles that were already present over these catalysts rather than agglomeration of dispersed CuO species as this phase is only very weakly resolved over Cu/CeO2-NP(E) catalyst where

CuO is more finely dispersed relative to the other two catalysts.

Figure 5.5: X-ray diffraction patterns of (a) CeO2-NR, (b) CeO2-NP, (c) Cu/CeO2-NR(E)

(d) Cu/CeO2-NR(T) (e) Cu/CeO2-NP(E) and (d) Cu/CeO2-NP(T).

Figure 5.6: X-ray diffraction patterns collected during in-situ reduction with 5%H2/N2 of

(a) Cu/CeO2-NP(E), (b) Cu/CeO2-NR(E) and (c) Cu/CeO2-NP(T)

5.3.5 Temperature-Programmed Reduction

The reduction behaviors of CeO2 supports and Cu/CeO2 catalysts impregnated in ethanol and THF media are studied using temperature-reduction technique and the results are presented in Figure 5.7. In the 50ºC-550ºC range, no H2 consumption features are observed in the H2 consumption traces of the CeO2 supports showing that, regardless of the particle morphology, CeO2 is not reduced under these conditions (Figures 7a and 7b).

This finding is in agreement with the in-situ XRD-TPR results discussed in the previous section where cerianite (CeO2) was observed to be the only crystalline phase in the 30-

450°C range. Regardless of the impregnation media, Cu/CeO2-NR supported copper catalysts exhibit a broad H2 consumption band in the 180ºC-450ºC range (Tmax = 285ºC) with a well-resolved shoulder in the high temperature range (Figures 7c and 7d). Figures

7e and 7f show the H2 consumption traces of Cu/CeO2-NP catalysts impregnated in ethanol and THF media. Both of the traces show a single H2 consumption feature in the

150ºC-250ºC region. The maximum corresponds to 205°C and 225°C for Cu/CeO2-

NP(E) and Cu/CeO2-NP(T), respectively and both traces show a shoulder in the low temperature range. The complex structure of the H2 consumption profiles of both of the samples suggests the presence of multiple species with different reduction potentials.

H2-TPR is extensively utilized for characterization of copper catalysts over ceria and the nature of copper species giving rise to the H2 consumption maxima observed in the TPR profiles is widely debated [26, 28, 35, 39, 51, 97-99]. The H2-consumption features located at lower temperatures (150ºC - 200ºC) are usually associated with well- dispersed copper oxide particles that are in close interaction with CeO2 and are

82 amorphous in nature [97, 98]. The H2 consumption peaks observed above 200ºC are attributed to the presence of larger and isolated crystalline CuO particles [26, 35, 39, 51,

97, 98]. Based on these studies, the low temperature H2 consumption features observed over both Cu/CeO2-NP(E) and Cu/CeO2-NP(T) were assigned to the reduction of finely- dispersed CuO species in close interaction with the CeO2 support. Avgouropoulos and

Ioannides [51] showed that the reduction temperature and the crystallite size of the CuO domains are correlated and that reduction temperature of the copper oxide domains increase with increasing cluster size. The lower temperatures at which the H2 consumption features were observed over CeO2-NP supported catalysts suggest the formation of smaller CuO particles compared to the catalysts supported over CeO2-NR.

The inset of Figure 5.7 shows the H2 consumption profile of bulk CuO during H2-

TPR carried out under the same experimental conditions. Under these experimental conditions, reduction of CuO takes place in 270ºC-600ºC region and goes through a broad maximum around 450ºC. The high temperature shoulder (300ºC-450ºC) in the H2 consumption profile of Cu/CeO2-NR can be associated with the reduction of crystalline copper oxide domains to form metallic copper. The transformation of these copper oxide particles to metallic copper was also observed through the in-situ XRD patterns collected during the reduction of Cu/CeO2-NR(E) (see Figure 5.6b). The slightly lower temperature for the appearance of metallic copper phase in the in-situ XRD patterns when compared to the temperature for the H2 consumption feature associated with the reduction of bulk- like CuO phase is due to the truly transient nature of the TPR experiment versus the stepped temperature profile utilized for the collection of in-situ XRD patterns. These

83 findings are in agreement with various studies in the literature that document the effect of ceria in enhancing the reducibility of copper [28, 35, 99, 100].

Figure 5.7: Temperature-programmed reduction profiles of (a) CeO2-NR, (b) CeO2-NP,

H2/He. Inset: Temperature-programmed reduction profile of bulk CuO.

5.3.6 N2O Chemisorption

The dispersion of Cu over the Cu/CeO2 catalysts was studied using N2O as a probe. The results of the N2O chemisorption measurements are presented in Table 5.3.

The results of these measurements further support the results from H2-TPR and XRD studies showing the presence of well-dispersed copper species over Cu/CeO2-NP(E) and

Cu/CeO2-NP(T). The Cu dispersion values are significantly lower for the CeO2 nano-rod- supported catalysts suggesting the formation of larger copper domains over this support.

The catalyst characterization studies showed the presence of copper species with different degrees of dispersion over CeO2 surface, ranging from large copper oxide crystallites to finely dispersed copper oxide particles. They also pointed to a correlation between copper dispersion and catalytic activity. It is also important to note that the possibility of copper occupying the interstitial sites within the ceria matrix cannot be ruled out; however, the preparation technique, i.e. wet impregnation of copper over the ceria support and the processing temperatures make copper migration into the ceria matrix is rather unlikely.

The presented results suggest that CeO2 nano-particles are capable of stabilizing copper oxide as finely-dispersed copper oxide particles which can be transformed into finely- dispersed metallic copper particles under reducing conditions. These finely dispersed clusters harbor the active sites for WGS reaction over Cu/CeO2 catalysts. Over the CeO2 nano-rods, copper oxide forms larger domains and is also present as bulk-like CuO species that get converted to large Cuº particles under reducing conditions. These particles exhibit poor WGS activity. Similar results on the dependence of WGS activity

86 of supported copper catalysts on the size of CuO domains, hence on the reducibility of

CuO has been reported previously [26, 33].

Cu dispersion Cu surface atoms

(%) (µmol/g cat)

Cu/CeO2 –NP(E) 18.0 129.6

Cu/CeO2 –NP(T) 13.8 99.4

Cu/CeO2 –NR(E) 1.2 8.6

Cu/CeO2 –NR(T) 1.0 7.2

Table 5.3 Cu-dispersion values estimated from N2O chemisorption

5.3.7 In-situ DRIFTS during WGS

The effect of ceria morphology on WGS reaction over Cu/CeO2 catalysts was further investigated by in-situ DRIFT spectroscopy during WGS reaction over Cu/CeO2-

NP(E) and Cu/CeO2-NR(E). Introduction of WGS reaction feed mixture over a pre- reduced Cu/CeO2-NP(E) sample at 50°C resulted in the formation of a sharp absorption band at 2105 cm-1 which can be associated with ν(C=O) vibrations of Cu-carbonyl species [101, 102]. The 2105 cm-1 band disappeared above 200°C after going through a maximum between 100°C-150°C. Due to their high thermal stability, the species were assigned as Cu+ carbonyls.

The 1700-1200 cm-1 region is the fingerprint region for asymmetric and symmetric stretching modes of O-C-O moieties which are present in formate, carbonate and carboxylate species. The complex structure of the spectrum in this region suggests the formation of several different surface species. Following introduction of WGS reaction feed at 50°C, bands located at 1638 cm-1 (shoulder), 1550 cm-1, 1390 cm-1

(shoulder), 1313 cm-1, 1050 cm-1 and 865 cm-1 were resolved. The 1638 cm-1 band decreased significantly in intensity with increasing temperature and disappeared above

100°C. This band was accompanied with a broad band centered around c.a. 3400 cm-1 in the hydroxy vibration region (νOH) which followed the same trend with temperature.

-1 Therefore, the shoulder band at 1638 cm was associated with deformation modes (δHOH) of undissociated water molecules on the surface, rather than νas(C=O) mode of bidentate carbonates. In the absence of bands in 2950-2850 cm-1 region, the absorption bands in

1600-1200 cm-1 region were associated with carbonate species rather than formates. The

89 bands located at 1055 cm-1, 1313 cm-1 and 1550 cm-1 were assigned to bidentate carbonate species [22, 103, 104] while the 1390 cm-1 band was associated with carbonates that are likely to be uni-dentate in nature [22]. The 1390 cm-1 band increased in intensity with increasing temperature and above 100°C the νs(OCO) mode of these

-1 -1 species were resolved at 1084 cm [105]. The 865cm band was assigned to π(CO3) mode of carbonates. 1215 cm-1 band observed in 150°C-250°C region may be attributed to formation of hydrogen carbonate species on the surface [103]. These results suggest that WGS reaction over Cu catalysts supported on nano-structured CeO2 particles proceeds through a red-ox mechanism rather than a formate mechanism and are in agreement with the findings of Wang et al. [32].

Figure 5.9 shows in-situ DRIFT spectra collected during a similar experiment over Cu/CeO2-NR(E). The significantly lower amount of surface species over this catalyst relative Cu/CeO2-NP(E) catalyst was apparent from the fewer absorption bands resolved in the DRIFT spectra. Over this catalyst the temperature window for the formation of copper carbonyl species was shifted by 100°C to higher temperatures and the intensity of the 2105 cm-1 band was significantly smaller, suggesting formation of a lower amount of copper carbonyl species. It is important to note that, the intensities and the temperature windows of appearance of the Cu+-carbonyl bands observed during WGS reaction over Cu/CeO2 catalysts is well correlated with the catalytic activities of these catalysts. Over Cu/CeO2, Gamarra et al. [102] observed a similar correlation between the intensity of the Cu+-carbonyl band and CO oxidation rate during preferential oxidation of

CO. These authors further suggested that the intensity of Cu+-carbonyl band was an indication of the reduction potential of the interfacial sites over dispersed copper oxide,

90 which serve as the active sites for CO oxidation. In line with the findings of these authors, the well dispersed copper entities, the presence of which have been shown by various characterization techniques over Cu/CeO2-NP(E) catalyst, were associated with high WGS activity. The high WGS activity of the catalyst with high copper dispersion together with the results of the DRIFTS experiments imply the presence of a red-ox mechanism which is facilitated by the higher reduction-oxidation potential of small, well- dispersed copper species.

Figure 5.8: In-situ DRIFT spectra collected during WGS reaction over Cu/CeO2-NP(E)

(a)high wavenumber region and (b)low wavenumber region.

Figure 5.9: In-situ DRIFT spectra collected during WGS reaction over Cu/CeO2-NR(E)

(a)high wavenumber region and (b)low wavenumber region.

5.4. Conclusions

Nano-structured CeO2 supports with two different particle morphologies, i.e. nano- particles and nano-rods, were prepared and wet impregnated with copper in two different impregnation media, ethanol and tetrahydrofuran. The morphology of the support was observed to have a strong effect on the WGS activity of the catalysts while the impregnation medium did not have a significant effect on the activity. Copper-catalyst supported over CeO2 nano-particles showed significantly higher activity than its CeO2 nano-rod-supported counterpart while neither bare CeO2 nano-particles, nor CeO2 nano- rods showed activity towards the WGS reaction. The effect was associated with the capability of CeO2 nano-particles in stabilizing CuO in a highly-dispersed state as the

Cu/CeO2-NP catalyst was found to have finely-dispersed CuO particles whereas large, crystalline CuO domains were present over ceria nano-rod-supported samples. Under reducing conditions, the bulk-like CuO domains were transformed into large metallic copper crystallites, which were associated with poor WGS activity of this catalyst.

* Information in this chapter is adapted/taken from P. Gawade, B. Mirkelamoglu and

U.S. Ozkan, Journal of physical Chemistry C 114 (2010) 18173-18181.

PART 2:

PREFERENTIAL OXIDATION OF CARBON MONOXIDE (PROX) IN

HYDROGEN-RICH STREAMS

CHAPTER 6

INTRODUCTION TO PREFERRENTIAL OXIDATION OF CO (PROX) IN

HYDROGEN

Proton exchange membrane (PEM) fuel cells are popular for their energy efficiency, low operating temperature, compactness and zero-emission. However, the Pt anode used in PEM fuel cells is highly sensitive to CO poisoning and even ppm level CO can significantly compromise PEM performance. Preferential oxidation of CO (PROX) is an important reaction to serve as a connecting link between low-temperature water-gas-shift

(LT-WGS) and the proton exchange membrane (PEM) fuel cell. It offers an economical and efficient technique for bringing down the carbon monoxide (CO) concentration from

0.5-1% to parts per million (ppm) level and purifies the hydrogen stream. Apart form this technique, there are other methods such as methanation, selective membrane separation and pressure swing adsorption, which are proposed for hydrogen purification application.

However, the ease of operation and implementation associated with PROX method makes it more promising than its other counterparts.

PEM fuel cell is operated around 80 oC [106] whereas, LT-WGS operating range is

200-250 oC. Therefore, 80-200 oC is an ideal temperature window for intermediate

PROX step. However, one of the challenges in this temperature range is to selectively

oxidize CO with minimal loss of H2 during side reactions. The main as well as the

possible side reactions during PROX can be summarized as follow [107]:

1 KJ CO oxidation: CO + O →CO ΔH 0 = −283 (1) 2 2 2 mol

1 0 KJ H2 oxidation: H + O →H O ΔH = −242 (2) 2 2 2 2 mol € € KJ WGS: CO + H O ↔ CO + H O ΔH 0 = −41.3 (3) 2 2 2 mol € € KJ Methanation: CO + 3H ↔ CH + H O ΔH 0 = −206 (4) 2 4 2 mol € € KJ Methanation: CO + 4H ↔ CH + 2H O ΔH 0 = −165 (5) 2 2 4 2 mol € €

€ Amongst the above-mentioned€ reactions, reaction (1) and (3) are desirable

reactions, whereas, reactions (2) and (4-5) are considered unfavorable for PROX

operations. It should be noted that although methanation reaction assists further removal

of CO from the reformate feed, it also causes the loss of hydrogen and hence brings down

the energy value from the reformate feed [108].

* Information in this chapter is adapted/taken from, P. Gawade, B. Bayram, A.M.

Alexander and U.S. Ozkan, Applied Catalysis B, Environmental, In Review.

CHAPTER 7

LITERATURE REVIEW

7.1 Precious Metal Catalysts

Noble metals such as Pt [107, 109-111], Rh/Ru [112], and Ir/Pd [113] are widely studied as PROX catalysts. Pt and Pt-X (X= Sn, Ce, Co and Ni) supported on various supports, such as Al2O3, CeO2, ZrO2, CeO2-ZrO2, MgO, activated carbon, La2O3, SiO2 and SiO2-Al2O3 [107, 111, 113-122] are widely investigated for their PROX activity.

Pt/CeO2 is a promising formulation in terms of PROX performance over Pt/Al2O3 due to the excellent oxygen storage capacity (OSC) of CeO2 [107, 115], which contributes to better CO oxidation. Apart from noble metals, gold is probably as widely studied as Pt- based catalysts for PROX [123-127] due to its higher availability and lower price than Pt

[128]. Au supported on α-Fe2O3, CeO2, γ-Al2O3, ZrO2, TiO2, Co3O4, NiO, MgO and

MnO have been investigated for their PROX activity [125, 126, 128-130]. Among these supports, Au/TiO2, Au/CeO2 and Au/α-Fe2O3 have promising PROX activity [125, 126,

131], especially Au/α-Fe2O3 prepared through co-precipitation method. This catalyst can achieve comparable PROX activity to that of commercially available Pt/γ-Al2O3 even at

98 low temperatures (80 oC) [126, 128]. In spite of the promising performance of both Pt- and Au-based catalysts for PROX, they need to be replaced due to cost factor.

7.2 Non-Precious Metal Catalysts

Copper and cobalt are probably the most commonly used non-precious transition metals as PROX catalyst. CuO/CeO2 and mixed oxides of CuO/ CeO2 are the most widely reported formulations for PROX [132-138]. The synergetic effect between CuO and CeO2 is known to provide high PROX activity [136, 138]. The excellent redox property of ceria support assists the CO oxidation, thereby provides higher activity. The

+4 addition of ZrO2 and Sn in CeO2 further improves its redox property and thermal resistance of the support [136, 139-141]. Chen, et al. [140, 141] found that

CuO/Ce0.9Zr0.1O2 and CuO/Ce0.9Sn0.1O2 had better PROX activity than CuO/ CeO2 due to improved redox property of CeO2.

Cobalt oxide (Co3O4) based catalysts are widely studied for CO oxidation in the absence of hydrogen [142-147]. Both, pure Co3O4 and Co3O4/Al2O3 have been reported for high CO oxidation in the absence of hydrogen [145, 146, 148]. When compared with other transition metals, cobalt oxide has better CO oxidation activity as well as selectivity over a wide temperature range in the presence of hydrogen [149]. Several reports in the past [127, 150-154] have suggested that cobalt oxide can be a promising candidate for

PROX application. It can either provide active sites or can be used as co-dopant in other catalyst formulations. As a co-dopant, it is known to enhance the PROX activity of

Au/CeO2 and Pt/ CeO2. In addition, Omata et al. [155] showed that cobalt supported on

99 alkali earth metal carbonate, such as Co/SrCO3 has excellent PROX activity in the presence of excess hydrogen.

Both, experimental and DFT studies have shown that Co+3 is an active site for CO

+3 oxidation [146, 156, 157]. Co occupies the octahedral positions in Co3O4, which can provide oxygen for CO oxidation resulting in the formation of Co+2. This lower oxidation state cobalt is known to occupy tetrahedral position of Co3O4, which may be re-oxidized

+2 in the presence of oxygen. However, the de-activation of Co3O4 may be observed, if Co fails to re-oxidize [146].

* Information in this chapter is adapted/taken from, P. Gawade, B. Bayram, A.M.

Alexander and U.S. Ozkan, Applied Catalysis B, Environmental, In Review.

100

CHAPTER 8

PROX OVER CoOx/CeO2 IN HYDROGEN-RICH STREAMS: EFFECT OF COBALT

8.1 Overview of PROX over CoOx/CeO2

Our group [151, 152] in the past had studied the various potential supports for Co, such as ZrO2, CeO2, Al2O3, SiO2 and TiO2 for their PROX performance. 10% (%wt)

CoOx catalysts were prepared over commercially available supports using wet- impregnation method. Interestingly, in spite of having extremely low surface area (7

2 m /g) and low Co-dispersion (<0.1%), Co/CeO2 had shown promising PROX activity.

Woods et al. [158] further extended this effort to prepare a high surface area 10%

CoOx/CeO2 using precipitation and incipient wetness impregnation (IWI) method. The catalyst had high specific surface area (78 m2/g) and significantly higher Co-dispersion

(3.4%) over nanoparticles of CeO2, thereby providing a much-improved PROX activity than its previous counterpart.

This work is an extension of our previous studies on PROX catalysis over Co/CeO2 catalysts. The catalyst preparation method used offers a simpler and cleaner way to prepare high surface area nanoparticles of CoOx/CeO2. The effect of Co-loading is

101 examined under hydrogen rich conditions as well as in the presence of CO2 and water vapor.

8.2 Experimental Procedure

8.2.1 Catalyst Preparation

High surface area ceria nano-particles were prepared through a precipitation technique using cerium (III) nitrate hexahydrate (99.999%, Sigma-Aldrich) and sodium hydroxide. An aqueous solution of cerium nitrate hexahydrate (0.4 M) was mixed with an aqueous solution of sodium hydroxide (2.7 M) at room temperature to form precipitates of Ce(OH)2. The resulting precipitate was aged at room temperature for 24 h followed by washing in de-ionized water until the pH of the filtrate was neutral. The precipitate was, then, dried at 110 oC overnight, followed by calcination at 400 oC for 3 h. Cerium oxide support prepared through the precipitation method outlined above yields nanoparticles approximately 8 nm in diameter [159].

CoOx/CeO2 catalysts with different cobalt loadings were prepared through a wet- impregnation method using aqueous medium. Aqueous solutions of cobalt nitrate hexahydrate of desired concentrations corresponding to 1%, 2% and 10% CoOx (%wt) loading on the resultant catalyst were prepared at room temperature followed by addition of ceria support into it under constant stirring. The solution was stirred rigorously for 24 h followed by drying at 110 oC. Finally, the catalysts were calcined at 400 oC for 3 h, 10 oC/min ramp.

102

8.2.2 Catalyst Activity Testing

The schematic of steady state reaction system for PROX is shown in Figure 8.1.

Steady-state reactions were performed in a fixed bed reactor flow system using ¼’ OD stainless steel reactor. For each run, the catalyst was packed inside the reactor using quartz wool and was heated using a homemade, resistively-heated furnace. The reaction temperature was measured and controlled by using a K-type thermocouple and an Omega

CSC232 PID temperature controller. The reaction feed as well as effluent from the reactor was monitored using on-line micro-GC (Agilent 3000A) equipped with 0.32 mm

PLOT mol-sieve and PLOT Q columns with thermal conductivity detectors (TCD) using helium (or argon, when appropriate) as the carrier.

Prior to catalytic activity testing, catalysts were pretreated in situ in 10% O2/He

(30ccm) at 300 oC for 30 min, followed by cooling to the desired reaction temperature under He flow. Steady-state PROX reaction studies were performed in the temperature range of 100-175 oC at a WHSV (weight hourly space velocity) of 15,000 cm3 (g cat-1) h-

1 . The feed composition was 1% CO, 1% O2, 60% H2 and balance helium.

CO and H2 oxidation experiments were performed over CoOx/CeO2 catalysts using 2% CO (or H2), 2% O2 in balance helium (or argon). For these experiments, 50 mg sample was packed inside the reactor and the total flow rate was maintained at 130 ccm to ensure the differential reactor operating conditions. For H2 oxidation experiments, carrier gas to the micro-GC was replaced with argon to enable detection of hydrogen by the TCD.

103

The effect of water vapor and CO2 on the PROX activity of the Co/CeO2 catalysts was investigated in 125-175 oC range at a WHSV of 15,000 cm3 (g cat-1) h-1. The feed composition for these studies was 1% CO, 1% O2, 60% H2, 0.5% H2O (or 0.5% H2O and

5% CO2) in balance helium. Water-gas-shift (WGS) and reverse water-gas shift (RWGS) activity of the CoOx/CeO2 formulations were studied in the temperature range of 125-200 oC at a WHSV of 15,000 cm3 (g cat-1) h-1. The feed stream for these studies constituted of

1% CO, 1% H2O or 5% CO2, 60% H2 in helium.

104

Figure 8.1: Schematic of Steady State Reaction System for PROX

105

8.2.3 Catalyst Characterization

Micromeritics ASAP 2020 accelerated surface area and porosimetry instrument was used to measure the surface area of support as well as CoOx/CeO2 catalysts through

N2 physisorption method at liquid nitrogen temperature (77 K). Before N2-physisorption,

o the samples were de-gassed at 130 C for 12 h at a vacuum better than 3 µm Hg.

Brunauer-Emmett-Teller (BET) method was used to determine the specific surface areas of the samples.

The N2O chemisorption technique outlined earlier in detail [160] was used to obtain the dispersion of cobalt in the CoOx/CeO2 catalyst (shown in Table 1). This technique is based on the method developed by Jensen et al. [92] to study the dispersion of Cu over CeO2. For each run, 100 mg of sample was packed in a fixed bed quartz reactor with a quartz frit bed that is placed inside a fast response furnace (Carbolite,

MTF 10/15/130) and pretreated with 5% H2/He (30 ccm) at 400 °C for 2 h. The reactor was then flushed with He for another 30 minutes at the same temperature and cooled under helium to 40 °C. Chemisorption of N2O was carried out by introduction of 4%

N2O/He to the reactor at 40 °C. An on-line mass spectrometer (MKS-Cirrus II) was utilized to monitor the reactor effluents in the m/z = 12 to m/z = 46 range. Prior to the experiment, the mass spectrometer was calibrated for instrumental sensitivity factors and the contribution of the m/z = 28 fragment of N2O to the m/z = 28 trace. Throughout the experiments, N2O and N2 were the only species detected in the reactor effluent. The number of moles of N2 evolved was related to the moles of oxygen consumed and moles of cobalt oxidized by assuming a Co:O stoichiometry of 1:1. As described by Jensen et

106 al. [92], the oxygen uptake is not only due to surface oxidation of Co, but also due to subsurface diffusion. The technique includes a correction for the Fickian diffusion.

Rigaku X-ray diffractometer equipped with Cu Kα radiation source (λ=1.5418 Å) operated at 40 kV and 25 mA was used to obtain X-ray diffraction patterns of Co/CeO2

o catalysts and CeO2 support in the range of 20-70 . In situ X-ray diffraction was performed during temperature-programmed reduction of 10% CoOx/CeO2 on a Bruker-

D8 advanced X-ray diffractometer equipped with an Anton Paar HTK1200 oven using

Cu Kα radiation source (λ=1.5418 Å) operated at 40 kV and 50 mA. The diffraction

o patterns were collected in the temperature range of 30-700 C under a flow of 5% H2/He.

International Center for Diffraction Data (ICDD) database was used for the identification of the crystalline phases from the collected diffraction patterns.

Temperature-programmed reduction (TPR) of CoOx/CeO2 catalysts, as well as the support, was carried out under a flow of 5% H2/N2 (40ccm). For each run, 100 mg of sample was packed in a quartz reactor with a porous quartz frit. The sample was

o pretreated in 10% O2/He (40 ccm) at 400 C for 1 h, followed by purging with helium at the same temperature for 30 min. The sample was then cooled down to room temperature under the same He flow. Finally, H2-TPR profile was collected using online mass spectrometer (Cirrus II, MKS Instruments) in 5% H2/N2 while increasing the temperature at a rate of 10 oC/min.

Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was performed over CoOx/CeO2 samples during PROX. The surface species over CoOx/CeO2 as well as the gaseous components were detected using Thermo Nicolet 6700, FT-IR spectrometer equipped with an MCT detector. 28 mg of the sample was loaded to the

107 controlled atmosphere chamber of the DRIFTS instrument for each run. DRIFTS spectra were collected in mid-IR range with resolution of 4 cm-1 and averaged over 500 scans.

o The catalyst was pretreated in 10% O2/He (40 ccm) for 30 min at 300 C, followed by cooling of the sample under helium to the desired temperature. Then the background spectrum was collected under helium. Finally, PROX gases (1% CO, 1% O2, 60% H2 and helium) were introduced at the same temperature and sample spectra were collected after

10 min under the gas flow.

Kratos AXIS Ultra X-ray photoelectron spectrometer (XPS) equipped with a monochromatic Al-Kα radiation source (1486.7 eV) operated at 13 kV and 10 mA was used to study the chemical states of cobalt and/or cerium in the ceria support as well as the CoOx/CeO2 catalysts. A survey scan was collected in the range of 1400-0 eV for every sample at an electron pass energy of 80 eV before sweeping in the Co 2p, Ce 3d and O 1s regions at an electron pass energy of 20 eV. Analysis of the collected X-ray photoelectron spectra was performed using XPS Peak 4.1.

In situ X-ray absorption near edge (XANES) data over 10% CoOx/CeO2 was collected at the Co K-edge (7709 eV) at the insertion device beamline (10ID-B) of the

Materials Research Collaborative Access Team (MR-CAT) of the Advanced Photon

Source, Argonne National Laboratories. The measurements were made in transmission mode and the Si(111) monochromator was tuned down to eliminate the higher order harmonics in the beam. The catalysts were mixed with SiO2 at a ratio of 1:2 and finely ground to obtain a homogeneous mixture. A ~30 mg of the mixture was then pelletized in a 6mm polished steel die and placed inside a 5 cm long quartz tube (6.5 mm. ID) and supported with quartz wool plugs. The sample was then centered in an in situ XAFS

108 chamber that was fitted with Kapton® windows and allowed continuous flow of the reactants as well as the isolation of the catalyst sample.

XANES data were collected over the pristine 10% CoOx/CeO2 sample in the range of 7500-7750 eV at room temperature. Then the catalyst was pretreated in the

o presence of 5% O2/He (20 ccm) at 300 C for 30 min followed by purging the catalyst with helium at the same temperature for 10 min. The sample was then cooled down to

o 175 C under the same He flow. Finally, PROX feed, 0.5% CO, 0.5% O2, 40% H2, balance helium (200 ml/min) was introduced at 175 oC for 30 min, followed by in situ

XANES data collection at the same temperature. It should be noted that a similar experiment was also attempted over 2% CoOx/CeO2 however, lower cobalt concentration coupled with the highly absorbing CeO2 matrix did not allow collection of XAFS data with sufficient resolution for analysis.

8.3 Results and Discussion

8.3.1 Surface Area and Dispersion Measurements

Table 8.1 presents the specific surface area of CoOx/CeO2 catalysts with different cobalt loadings. The surface area was found be dependent on the amount of cobalt content and it decreased with increase in cobalt loading. Cobalt dispersion was ∼20 % regardless of cobalt loadings, suggesting a similar particle size of the Co phase over the support. It should be noted that these dispersion measurements were within ± 2% error.

Although the particles size seemed to be similar regardless of cobalt loadings, the extent

109 of the support surface populated with these particles or the size of the clusters made of these particles increased with loading.

110

Surface Area Co Dispersion Co Surface Atoms

(m2/g) (%) (µmol/g cat)

CeO2 123 - -

1%CoOx/CeO2 111 17 ± 2 29

2%CoOx/CeO2 106 18 ± 2 62

10%CoOx/CeO2 84 21 ± 2 362

Table 8.1: Specific surface areas and Co dispersions over CoOx/CeO2 catalysts

111

8.3.1 Steady State Activity of Performance of CoOx/CeO2

Figure 8.2a shows the CO conversion as a function of temperature over

CoOx/CeO2 catalysts with different cobalt-loadings during preferential oxidation of carbon monoxide. CO conversion was found to increase monotonically with increasing temperature regardless of the cobalt-content. 10% CoOx/CeO2 showed slightly better CO oxidation activity than the 2% CoOx/CeO2 catalyst especially at temperatures below 150 o C while, the PROX activity of 1% CoOx/CeO2 was significantly lower than its counterparts. The difference in the PROX activity diminished at higher temperatures, especially in case of 10% CoOx/CeO2 and 2% CoOx/CeO2, reaching to approximately

94% of CO conversion at 175 oC over both of these catalysts. In addition, long-term

o reaction study was performed over 10% CoOx/CeO2 for over 40 h at 175 C as shown in

Figure 8.2 (inset). The data indicated no significant drop in activity or selectivity after the steady state was established in the first 10 h. Co3O4 has been considered an excellent candidate for CO oxidation in the presence of hydrogen. However, excess amount of hydrogen under the PROX condition may result in lower oxidation state cobalt phases, including metallic cobalt, which may promote side reactions [149, 151, 152, 155, 157,

161]. Both these side reactions could result in a drop in the CO conversion as well as O2 selectivity to CO2. However, no such phenomenon was observed during the time-on- stream experiment indicating that Co3O4 phase was preserved even in the presence of excess hydrogen. This claim was further supported by the X-ray absorption study

(Section 8.3.5.5) where no lower oxidation states of cobalt phases were observed during

PROX reaction.

112

The selectivity of O2 towards CO2 is illustrated in Figure 8.2b. O2 selectivity was around 100% in the temperature range of 100-125 oC regardless of cobalt content, indicating negligible hydrogen combustion over these catalysts. However, above 125 °C,

O2 selectivity was observed to decrease for all of the three catalysts, indicating the onset

o of hydrogen combustion. At 175 C, O2 selectivity decreased with increased cobalt loading, indicating that hydrogen combustion was more favored at higher cobalt loadings.

o For PROX reaction over CoOx/CeO2 catalyst, 175 C was a more relevant reaction temperature as CO conversion reached nearly 100%. Both 10% CoOx/CeO2 and 2%

o CoOx/CeO2 showed similar CO conversion at 175 C, however, O2 selectivity over 2%

CoOx/CeO2 was much better (71%) than 10% CoOx/CeO2 (58%). Therefore, under the dry PROX conditions, 2% CoOx/CeO2 could be a better choice than 10% CoOx/CeO2 at

175 oC.

It should be noted that the current CoOx/CeO2 catalyst showed superior PROX performance compared to our previously developed [151] cobalt catalysts supported on various commercial supports such as Al2O3, SiO2, ZrO2, TiO2 and CeO2 (not shown).

Furthermore, the PROX performance over the current CoOx/CeO2 catalyst was comparable to our previously developed [158] CoOx/CeO2 formulation (not shown).

However, the current catalyst formulation holds advantages over the previous generation catalysts in terms of ease of synthesis and environmentally benign catalyst preparation.

113

Figure 8.2: PROX activity and selectivity of () 1% CoOx/CeO2, () 2% CoOx/CeO2, and () 10% CoOx/CeO2. (Reaction conditions: 1% CO, 1% O2, 60% H2 in He.

3 -1 -1 WHSV=15,000 cm (g cat.) h ). Inset: Time-on-stream CO conversion and O2

o selectivity to CO2 over 10% CoOx/CeO2 at 175 C under above-mentioned reaction conditions 114

8.3.2 Oxidation of CO and H2 over CoOx/CeO2

The kinetics of CO and H2 oxidation over the CoOx/CeO2 catalysts were studied to investigate the intrinsic differences between these catalysts in the presence of excess oxygen. The kinetic data for CO and hydrogen oxidation was collected separately in the presence of 2% CO (or H2) and 2% O2. The results obtained during separate CO and H2 oxidation complement the activity and selectivity data shown in Figure 8.2.

Figure 8.3 shows the CO conversion over CoOx/CeO2 catalysts with different cobalt loadings as a function of temperature during CO oxidation. As expected, CO conversion was found to increase with temperature and cobalt loading. The inset shows the Arrhenius plot for CO oxidation, which was used to obtain activation energies shown in Table 8.2. The activation energy of CO oxidation was found to increase with decreasing cobalt content. Activation energy numbers indicated that CO oxidation became more difficult in lower cobalt content samples. Therefore, the difference in CO conversion among different CoOx/CeO2 catalysts was more prominent at lower PROX temperatures. However, with increase in reaction temperature, sufficient energy appears to have provided in order to overcome the activation energy barrier, consequently difference in activity was diminished at 175 oC.

Figure 8.4 shows the H2 oxidation over CoOx/CeO2 catalysts. H2 oxidation increased with temperature and cobalt content. Arrhenius plot for H2 oxidation is shown in inset, which was used to obtain activation energies shown in Table 2. The activation energy of H2 oxidation was found to increase with decreasing cobalt content indicating

H2 oxidation was not favored in samples with low cobalt loading, and hence better O2

115 selectivity was achieved over 2% and 1% CoOx/CeO2 even at higher reaction temperatures. Furthermore, higher activation energies for H2 oxidation compared to CO oxidation showed that H2 oxidation was more temperature sensitive and became more significant at higher PROX temperatures.

116

CO Oxidation H2 Oxidation

(KJ/mol) (KJ/mol)

1%CoOx/CeO2 58.4 93.2

2%CoOx/CeO2 52.7 80.2

10%CoOx/CeO2 40.8 65.5

Table 8.2: Activation energies of CO and H2 oxidation over CoOx/CeO2 catalysts.

117

Figure 8.3: CO conversion over () 1% CoOx/CeO2, () 2% CoOx/CeO2, and () 10%

CoOx/CeO2 during CO oxidation. (Reaction conditions: 2% CO and 2% O2 in helium).

Inset shows Arrhenius plot for determination of the activation energy for CO oxidation

118

Figure 8.4: H2 conversion over () 1% CoOx/CeO2, () 2% CoOx/CeO2, and () 10%

CoOx/CeO2 during H2 oxidation. (Reaction conditions: 2% H2 and 2% O2 in argon) Inset shows Arrhenius plot for determination of the activation energy for H2 oxidation

119

8.3.3 Effect of CO2 and water vapor over CoOx/CeO2

Table 8.3 compares the CO conversion among CoOx/CeO2 catalysts in the

3 - presence of water vapor as well as CO2. The WHSV was maintained at 15,000 cm (g cat

1 -1 ) h for all the runs. The introduction of 0.5% H2O in the PROX feed showed the negative effect on CO conversion over CoOx/CeO2 catalysts.

Several studies [107, 135, 153, 162] have shown that water vapor can significantly affect the CO conversion due to various reasons, such as blockage of active sites, reverse water-gas-shift (RWGS) and formation of a less active CO-H2O complex.

In our case, this effect was more prominent over catalysts with lower cobalt content and at low reaction temperatures. 10% CoOx/CeO2 was the least affected catalyst in the presence of water vapor at any given temperature between 125-175 oC, followed by 2%

CoOx/CeO2. Performing a water-gas-shift (WGS) reaction over these catalysts in the presence of 1% CO and 1% H2O supported this observation. As shown in Figure 8.5, CO conversion increased with increasing cobalt loading and reaction temperature. 10%

CoOx/CeO2 had a much better WGS activity compared to other catalysts due to which

95% CO conversion was observed during PROX even in the presence of water.

The combination of 0.5% H2O and 5% CO2 in the feed had more significant impact on CO conversion than water alone. Again, 1% CoOx/CeO2 was the most affected catalyst followed by 2% CoOx/CeO2. CO conversion of 2% CoOx/CeO2 dropped to 76%

o in the presence of CO2 and H2O at 175 C whereas, 10% CoOx/CeO2 was still able to achieve around 90% CO conversion. It should be noted that reverse-water-gas-shift

(RWGS) was carried out in the presence of 60% H2 and 5% CO2 over CoOx/CeO2

120 catalysts (not shown). However, CO2 conversion was below 3-4% in the temperature

o range of 100-200 C indicating that WGS was still a dominant reaction over CoOx/CeO2 even in the presence of CO2, especially over 10% CoOx/CeO2.

O2 selectivites are also presented in the Table 8.3 and were found to decrease in

o the presence of H2O and CO2 at 125 C indicating that hydrogen oxidation was not as affected as CO oxidation at lower reaction temperatures. However, as the reaction temperature was increased, O2 selectivites were seen to converge and in the case of 10%

o CoOx/CeO2 and 2% CoOx/CeO2 at 175 C, they showed an increase in the presence of

H2O and CO2. This observation could be due to an increase in CO oxidation and higher

WGS reaction rates at higher temperatures, as discussed earlier.

121

Temperature 1%CoOx/CeO2 2%CoOx/CeO2 10%CoOx/CeO2

No H2O-CO2 32 (100) 43 (93) 48 (99)

o 125 C 0.5% H2O 19 (82) 31 (91) 34 (93)

0.5%H2O-5%CO2 0 (N/A) 9 (90) 17 (73)

No H2O-CO2 61 (100) 77 (92) 79 (83)

150oC 0.5% H2O 50 (94) 70 (92) 78 (83)

0.5%H2O-5%CO2 16 (81) 36 (84) 53 (85)

No H2O-CO2 89 (86) 95 (70) 93 (58)

o 175 C 0.5% H2O 80 (87) 93 (70) 96 (59)

0.5%H2O-5%CO2 53 (86) 77 (83) 88 (65)

Table 8.3: CO conversion (%) over CoOx/CeO2 in the presence of H2O and CO2. Values in parentheses show O2 selectivity to CO2 (%) over CoOx/CeO2 in the presence of H2O and CO2

122

Figure 8.5: CO conversion over () 1% CoOx/CeO2, () 2% CoOx/CeO2, and () 10%

CoOx/CeO2 during water-gas-shift. (Reaction conditions: 1% CO and 1% H2O in helium)

123

8.3.5 Catalyst Characterization

8.3.5.1 XRD

Figure 8.6 shows the X-ray diffraction patterns over CoOx/CeO2 catalysts and the bare ceria support collected under ambient conditions. The diffraction lines observed at

o o o o o o 2Θ values of 28.44 , 32.96 , 47.56 , 56.2 , 58.6 and 68.9 were assigned to cerianite structure and associated with [1 1 1], [2 0 0], [2 2 0], [3 1 1], [2 2 2] and [4 0 0] planes, respectively. In addition to the cerianite phase, 10% CoOx/CeO2 showed distinct diffraction peaks corresponding to cobalt oxide phase (Co3O4) located at 2Θ values of

36.7o and 65.15o associated with [3 1 1] and [4 4 0] planes, respectively. Unlike 10%

CoOx/CeO2, weakly resolved peaks for Co3O4 were observed over 2% and 1%

CoOx/CeO2 along with the cerianite phase. This observation along with dispersion measurements indicated that increased cobalt loading did not change the particle size, but the extent of the support surface populated with cobalt particles or the sizes of the clusters made up of these particles. Hence the surface more populated with Co3O4 particles and larger clusters demonstrated diffraction lines with measureable intensities.

The evolution of the crystal phases under a reducing environment over 10%

CoOx/CeO2 is shown in Figure 8.7. XRD patterns were collected in situ in the presence of

o 5% H2/N2 over calcined 10% CoOx/CeO2 in the temperature range of 30-700 C. The

o o o o o peaks located at 2Θ values of 28.44 , 32.96 , 47.56 , 56.2 and 58.6 are associated with

124

[1 1 1], [2 0 0], [2 2 0], [3 1 1] and [2 2 2] planes, respectively and are characteristic of the cerianite phase. The intensity of cerianite peaks increased with temperature, indicating an increase in average particle size under the reducing environment, especially

o o above 500 C. The peak at 2Θ value of 36.7 associated with Co3O4 [3 1 1] plane started

o o to disappear at 250 C and a new crystal phase, CoO was observed at 2Θ value of 42.1 .

With further temperature increase, CoO started to disappear at 300 oC and metallic cobalt

o (Co) was observed at 2Θ value of 44.07 .

125

Figure 8.6: X-ray diffraction patterns of (a) CeO2, (b) 1% CoOx/CeO2, (c) 2%

CoOx/CeO2, and (d) 10% CoOx/CeO2

126

Figure 8.7: In situ X-ray diffraction patterns collected during reduction of 10%

CoOx/CeO2 with 5% H2/He

127

8.3.5.2 TPR

The reduction characteristics of CoOx/CeO2 catalysts and CeO2 support under 5%

H2/He is shown in Figure 8.8. A bare CeO2 support did not exhibit H2 consumption showing that CeO2 could not be reduced under the conditions that the TPR was carried

o o out. 10% CoOx/CeO2 showed three distinct reduction features at 290 C, 325 C and

o o around 500 C. The reduction peak at 290 C was assigned to the reduction of Co3O4 to

CoO whereas the peak at 325 oC was assigned to reduction of CoO to Co [163]. These observations were in good agreement with in situ XRD over 10% CoOx/CeO2.

Unlike 10% CoOx/CeO2, lower cobalt content catalysts, 2% CoOx/CeO2 and 1%

o CoOx/CeO2 showed a broad feature in the temperature of 230-390 C, which could be correlated to reduction of Co3O4 to CoO and CoO to Co. The weakly resolved reduction features observed over these catalysts could be due to low cobalt concentration over

CeO2 support.

o The broad peak observed over 10% CoOx/CeO2 around 500 C was assigned to reduction of surface ceria. The transition metals are known to improve the ceria reducibility. In fact, the absence of this reduction feature in 2% CoOx/CeO2 and 1%

CoOx/CeO2 supported the assertion that cobalt was able to improve the reducibility of ceria.

128

Figure 8.8: Temperature-programmed reduction profiles of (a) CeO2, (b) 1% CoOx/CeO2,

129

8.3.5.3 XPS

X-ray photoelectron spectroscopy was used to examine the chemical nature of cerium and cobalt species in CoOx/CeO2 catalysts. Figure 8.9a-c shows Ce 3d envelope of 1% CoOx/CeO2, 2% CoOx/CeO2 and 10% CoOx/CeO2, respectively. All three spectra showed the contribution from both Ce+3 and Ce+4. The peaks labeled as v (884.1 eV), vʹʹ

(889.2 eV) and vʹʹʹ (898 eV) and their corresponding u modes, u (902.7 eV), uʹʹ (907.6

+4 eV) and uʹʹʹ(916.4 eV) were the characteristic features of Ce whereas, v0 (882 eV), vʹ

(887.2 eV) and their corresponding u modes, u0 (900.5 eV) and uʹ (906.6 eV) were the characteristic features of Ce+3 species [94-96]. Ce 3d envelope showed no significant change upon changing the cobalt loading of the CoOx/CeO2 catalysts.

The core-level Co 2p3/2 spectra of CoOx/CeO2 catalysts are shown in Figure 8.9.

The curve fitting in Co 2p3/2 region for 10% CoOx/CeO2 (Figure 8.9c) showed two components located at 779.1 eV and 780.4 eV. The lower binding energy (B.E.) component was assigned to the main photopeak of cobalt oxide (Co3O4) whereas, the higher B.E. shoulder at 780.4 eV was associated with Auger LMM transition [164]. For

2% CoOx/CeO2 (Figure 8.9b), both these peaks were shifted approximately by 1 eV towards higher B.E. values. It is a known fact that binding energy of core-level electrons of Co+3 is smaller than that of Co+2. Bonelle et al. [165] explained this phenomenon by

+3 +2 postulating that Co ions in Co3O4 have smaller effective charge than Co ions in CoO.

Therefore, this final state effect results in a higher binding energy value (usually 0.9 eV apart) of Co+2 ions than Co+3 ions [166]. In addition, the final state effects in Co 2p region accompany a shift in B.E. of O 1s core-level electrons. The binding energy value

130 of O 1s core electrons in Co3O4 is at higher B.E. than CoO [164]. The peaks fitting (not shown) in O 1s region for 10% CoOx/CeO2 showed the higher B.E. values for O 1s core electron than 2% CoOx/CeO2 supporting this assertion. In addition, the satellite feature was observed at 785.6 eV in 2% CoOx/CeO2 along with main photopeak and Auger

LMM transition. The presence of the satellite feature in 2% CoOx/CeO2 could be due the presence of the paramagnetic Co+2 ions [164]. The reduction of cobalt oxide species under X-ray beam in the high-vacuum environment may occur. Therefore it is plausible

+3 that some of Co ions in Co3O4 were reduced to lower oxidation state cobalt oxide.

Hence 2% CoOx/CeO2 may have some contribution from CoO along with Co3O4. In contrast, 10% CoOx/CeO2 showed no such evidence of the presence of paramagnetic

Co+2, indicating a strong presence of Co+3.

131

Figure 8.9: X-ray photoelectron spectra in the Ce 3d and Co 2p regions of (a) 1%

CoOx/CeO2 (b) 2% CoOx/CeO2 and (c) 10% CoOx/CeO2

132

8.3.5.4 In situ DRIFTS during PROX

The investigation of surface species formed during PROX over CoOx/CeO2 was studied using in situ DRIFTS. Figures 8.10 a-b show the in situ DRIFTS spectra collected over 2% CoOx/CeO2, at various temperatures, in the presence of 1% CO, 1% O2, 60% H2 and helium. The bands associated with gaseous and weakly adsorbed CO were observed at 2152 and 2095 cm-1, respectively, and were seen to decrease with increasing

-1 temperature. The doublet of CO2 at 2358 and 2330 cm was found to increase indicating significant CO oxidation at higher PROX temperatures. The strong positive peaks at

1022, 1290 and 1581 cm-1 indicate the formation of bidentate carbonate species [167-

170] at 30 oC. This observation is consistent with the study reported by Luo. et al. [167] over Co3O4-CeO2. In their findings, the formation of bidentate carbonates was observed at 1590-1582, 1291-1272 and 1026 cm-1 during CO oxidation at room temperature, it was also reported that the intensities of these peaks decreased with increasing temperature. A weak positive feature at 1520 cm-1 accompanied by a broad shoulder at 1260-1220 cm-1 was also observed at 30 oC, which may be attributed to monodentate carbonate species.

Takeguchi et al. [171] also reported monodentate carbonate formation on CeO2 supports

-1 at 1520 and 1270 cm after adsorption of both CO and CO2. Although the presence of monodentate species is to be expected, hydrogen carbonate formation at room temperature cannot be ruled out. Hydrogen carbonate bands have also been reported to occur within the 1270-1220 cm-1 region. However these bands are generally accompanied by the additional bands at 1600, 1390-1413 and 1025-1045 cm-1 [168, 170]. It is possible that these bands are masked due to other carbonates and gas phase species. As the

133 reaction temperature was increased under the PROX feed, the intensities of carbonate bands decreased, possibly due to transformation of carbonate species. The presence of negative bands at 1520, 1220 and 1028 cm-1 could suggest that some of the carbonate species were present on the surface when background spectra were collected. The diminished carbonate bands were accompanied by the formation of strong bands at 1463,

1060 and 854 cm-1, which could be assigned to polydentate carbonates. This observation was in agreement with results obtained in a study by Binet et al. [172]. It was reported that adsorption of CO onto ceria at room temperature predominantly results in the formation of bidentate carbonate species, which upon heating convert to polydentate carbonates [172]. In the same study the thermal stability of various carbonate species was discussed and it was found that, the thermal stability increased from bidentate carbonates

< monodentate carbonates

-1 band at 1380 cm , which is attributed to the symmetric stretch of CO2 [168, 169]. This was confirmed in a separate experiment (inset in Figure 8.11a) where the sample chamber was flushed with He after the catalyst (10% CoOx/CeO2) was kept in a PROX reaction medium at 125°C for 30 minutes. After the gas phase species were flushed, a sharp band

-1 - that was observed at 1380 cm along with the CO2 doublet at 2358 and 2330 cm

1diminished with time. Therefore it is plausible that 1380 cm-1 feature was due to

-1 symmetric CO2 stretch. A well-defined positive band at 1211 cm was observed at higher

134 temperatures, which could be assigned to the formation of intermediate carbonate species

-1 due to adsorption of gas phase CO2 [167]. The peak at 1403 cm usually accompanies this feature, however, the presence of the strong band evidenced at 1380 cm-1 due to gaseous CO2, probably masks this feature. Apart from the carbonates, the formate species were observed with increasing temperature at 2929, 2840, 1560-1540 and 1361 cm-1

[152, 170, 173] which could be due to the interaction OH species at 3712 cm-1 (OH-I),

3665 cm-1 (OH-II) and 3610 cm-1 (OH-III) [170, 174] with CO. In addition to the above mentioned formate bands, some negative features were observed at low temperatures around 1548 and 1357 cm-1, which could be assigned to formate species on ceria support

[175]. One possibility is that these formate species were present originally on the catalyst when background spectra were collected and were transformed to other species upon introduction of the PROX reaction medium. Nevertheless, there is still some ambiguity in assigning these two negative bands and the possibility of a mathematical artifact due to positive/negative background subtraction cannot be ruled out.

A similar in situ DRIFTS experiment was performed during PROX over 10%

CoOx/CeO2 as shown in Figures 8.11 a-b. No significant difference was observed between DRIFT spectra collected over 10% and 2% CoOx/CeO2. However it was noted

-1 that the CO2 doublet occurring at 2356 and 2329 cm was more prominent over the 10%

CoOx/CeO2 sample, suggesting significantly higher CO oxidation. When compared with

2% CoOx/CeO2 on common Y-axis scale, the intensities of CO2 doublet were higher in

o the temperature range of 100-125 C over 10% CoOx/CeO2. However, the difference

o between CO2 intensities diminished from 175 C onwards, in agreement with data presented in Figure 8.2. It should be noted that background spectra were collected under

135 helium at each temperature before collecting the sample spectra under PROX feed in order to correct for background CO2. Therefore, it is reasonable to correlate CO2 intensities observed during the DRIFTS with CO oxidation activity presented in Figure

8.2.

136

Figure 8.10: In situ DRIFT spectra collected during preferential oxidation of CO over 2%

CoOx/CeO2 (a) high wavenumber region and (b) low wavenumber region.

137

Figure 8.11: In situ DRIFT spectra collected during preferential oxidation of CO over

10% CoOx/CeO2. (a) high wavenumber region and (b) low wavenumber region. Inset: spectra taken at 125°C after the reaction gases are flushed

138

8.3.5.5 In situ X-ray absorption near edge spectra (XANES)

The transformation of cobalt coordination environment during PROX was monitored using X-ray absorption near edge spectra (XANES). Figure 8.12a presents the

XANES spectra collected over the pristine 10% CoOx/CeO2 in the range of 7500-7750 eV together with Co3O4. The XANES spectrum of pristine 10% CoOx/CeO2 was well aligned with Co3O4. Figure 8.12b presents the XANES spectrum collected during PROX

o at 175 C over 10% CoOx/CeO2. The XANES spectrum remains similar to that of Co3O4 species and reduction of cobalt was not observed.

Several studies in the past have indicated that Co3O4 can be an excellent candidate for CO oxidation in the presence of hydrogen. However, the excess hydrogen content during PROX may reduce Co3O4 to cobalt oxide phases with lower oxidation states, including metallic cobalt, which may promote the side reactions [149, 151, 152, 155, 157,

161]. The present study suggests that Co3O4 phase remains intact in 10% CoOx/CeO2 and no lower valency cobalt was observed during PROX. Having said all this, the possibility of a partial reduction of cobalt species if the catalyst was kept on-line for much longer period or under higher reaction temperatures can not be ruled out.

139

Figure 8.12: Normalized Co K-edge XANES spectra of 10% CoOx/CeO2. (a) Pristine and

o (b) In situ under PROX at 175 C (Spectra for Co3O4 is included for comparison)

140

8.4 Conclusions

CoOx/CeO2 catalyst was investigated for PROX with a special focus on the effect of cobalt loadings. While higher cobalt loadings favored CO oxidation at low PROX temperatures (T ≤ 150 oC) under the dry conditions, oxygen selectivity was seen to suffer

o at higher cobalt loadings. At 175 C, both 10%CoOx/CeO2 and 2%CoOx/CeO2 showed nearly 100% CO conversion however, O2 selectivity over 2% CoOx/CeO2 was much better than 10% CoOx/CeO2. Furthermore, higher activation energies for H2 oxidation compared to CO oxidation were observed, regardless of the Co loading, implying a higher temperature sensitivity for the H2 oxidation reaction. The presence of water and

CO2 in the feed resulted in a drop in PROX performance, possibly due to the inhibition effect over CoOx/CeO2 catalyst, regardless of the cobalt loadings. However, higher cobalt loading was seen to demonstrate better water /CO2 tolerance and assisted water gas-shift reaction.

The cobalt phase was identified, as Co3O4 and was stable in the reducing environments present in PROX reactions. A stable time-on-stream performance over

CoOx/CeO2 along with catalyst characterization using several analytical techniques indicated no significant contributions from lower valency cobalt species, including metallic cobalt under the reducing environment. From the in situ DRIFTS studies conducted on CoOx/CeO2 upon the introduction of PROX reaction gases at room temperature several species were formed on the catalyst surface such as linearly adsorbed

CO, monodentate carbonate and bidentate carbonate species. As the PROX reaction

141 temperature increased, these carbonate species were converted into more stable polydentate carbonates and formate type species. In addition, the intermediate carbonate species were also observed between 100-150 oC, probably due to the interaction of adsorbed gas phase CO with the hydroxylated CoOx/CeO2 surface.

* Information in this chapter is adapted/taken from, P. Gawade, B. Bayram, A.M.

Alexander and U.S. Ozkan, Applied Catalysis B, Environmental, In Review.

142

PART 3

CATALYTIC AFTER-TREATMENT OF NATURAL GAS ENGINE EXHAUST-A

DUAL STAGE APPROACH

143

CHAPTER 9

INTRODUCTION TO CATALYTIC NOX REDUCTION

In the context of evermore stringent requirements for fuel efficiency and CO2 emissions, natural gas-fired reciprocating engines represent an important and increasingly popular choice within the distributed energy market place. Lean-burn natural gas reciprocating engines offer a simple, well-understood technology and have several advantages over stoichiometric gasoline engines [176]. Lean burn conditions are associated with higher engine efficiencies and significantly cleaner engine out emissions.

However, despite emissions being greatly reduced, engine exhaust still contains considerable amounts of NOx species, carbon monoxide and un-burned hydrocarbons.

Some NOx emissions can be reduced through the modification of fuels or by altering the combustion parameters¸ however in order to reach acceptable emission levels the need for effective after-treatment is evident.

Among current catalytic NOx reduction control technologies, three way catalysts and ammonia- or urea–based selective catalytic reduction (SCR) are the most prevalent.

Both methods are highly effective for the current combustion technologies, which operate over a wide temperature window (200-400 oC), yet are unsuitable for the next generation of lean-burn engines. For example, three way catalysts, which are mostly used in mobile

144 applications, quickly deactivate under lean conditions, while the use of ammonia in emission abatement poses hazards in itself, mainly related to ammonia slip, direct ammonia oxidation, corrosion of equipment due to the formation of ammonium salts, and the additional issues related to ammonia storage and handling [177].

The selective reduction of NOx species by hydrocarbons has attracted significant attention as a promising alternative to conventional after-treatment technologies [178-

180]. This type of application is particularly suited to natural gas engines, in which unburned hydrocarbons, particularly methane, are already present in the exhaust stream and capable of acting as the reducing agent. Although methane, which is a major component of natural gas, is readily available, its effective use as a reducing agent in hydrocarbon-SCR systems is somewhat limited. This is mainly due to the competition from the combustion of the hydrocarbon in the presence of excess oxygen and the inherent difficulty of methane activation [181, 182].

* Information in this chapter is adapted/taken from P. Gawade, A.M.C Alexander, R.

Clark and U.S. Ozkan, Catalysis Today, Submitted (2012).

145

CHAPTER 10

LITERATURE REVIEW: NOX REDUCTION WITH HYDROCARBONS

Early work by Iwamoto et al. [183] and Held et al. [184] demonstrated the possibility of selective catalytic reduction of NOx using hydrocarbons (HC-SCR) over zeolite-based catalysts. Since then, numerous studies and catalyst formulations have been reported on HC-SCR with variety of reaction conditions. Typically, the catalysts for HC-

SCR can be classified into two categories based on the support materials, namely zeolites and metal oxides.

Both noble (Pt, Pd, Rh, Ag) and non-noble (Cu, Fe, Co, In Ni) metals supported on zeolites are widely reported for this application [185-195] ([196, 197] and reference therein). Among these formulations, Cu-ZSM-5 is the probably the oldest and most studied formulation until now [198-200] ([196, 197] and reference therein). Despite being the most popular catalyst for HC-SCR for academic purpose, the performance of

Cu-ZSM-5 suffers in the presence of water and SO2 ([197] and references therein). Apart from Cu-ZSM-5, Pt and Pd-based zeolite catalysts are the most commonly used catalysts under the lean conditions. However, the major drawback of using Pt-ZSM-5 is the formation of N2O during HC-SCR, which is a strong greenhouse gas ([197] and references therein). Among the catalysts which have been reported to date, palladium- based catalysts supported on zeolites, such as ZSM-5 and mordenite, have shown the

146 highest activity and selectivity for catalytic NOx reduction under lean conditions [201].

Despite their high activity and selectivity in HC-SCR, zeolitic catalysts suffer from poor hydrothermal stability [193, 194, 202]. Consequently dealumination of the material may occur, resulting in the formation of PdO aggregates, which are thought to be active for methane combustion, and loss of metal dispersion [203].

Apart from zeolite-based catalysts, metal oxide supported catalysts such as, alumina-based and zirconia-based formulations are commonly used for HC-SCR. Ag and

Pt supported on alumina is widely reported for this application [204-210] ([197, 211] and references therein). Despite being widely studied, alumina-based catalysts are known to lose SCR performance in the presence of water and SO2 ([197] and reference therein). In recent years, sulfated-zirconia supported catalysts have attracted attention for HC-SCR due their promising performance under lean and wet/SO2 conditions [212-216]. The development of these acidic non-zeolitic catalysts, for the HC-SCR of NOx species, has therefore attracted much interest, mainly due to an increase in hydrothermal stability compared to zeolitic counterparts.

Early studies by Resasco and co-workers compared the activity of various zeolithic and non-zeolithic supported Pd catalysts [217]. It was reported that Pd- loaded sulfated zirconia has similar activity for NOx reduction by methane to that of

Pd-supported zeolites. The advantage of using zirconia over other well-established supports is its enhanced thermal and chemical stability. The increased acidity of sulfated zirconia is significant since the catalytic activity and selectivity of De-NOx are thought to be a consequence of the metallic phase and acidity of the support.

Studies by Ohtsuka and co-workers have shown that the addition of platinum into a

147

Pd/SZ catalyst [218] improves both long-term hydrothermal stability and NOx reduction activity. Later studies by the same group have shown that NOx reduction activity can be further improved through the use of a Fe dopant [219]. In these studies NOx conversion was reported to be 60% after 700 hours on stream in the presence of 9% water vapor. Similar co-operative effects have also been observed on Co promoted Pd supported catalysts [220-222]. Quincoces et al. have reported that a Co-Pd/SZ catalyst undergoes a reversible decrease of both NOx reduction and

CH4 oxidation activity in the presence of water vapor; it was also noted that in the presence of water vapor the optimum operating temperature shifts to higher temperatures.

Previously we have described an integrated oxidation and selective reduction catalytic system [223-227] where NO was first oxidized to NO2, then reduced selectively with hydrocarbons. This method combines separate oxidation and reduction catalyst components in order to perform three distinct catalytic functions, specifically NOx reduction, CO oxidation and hydrocarbon combustion. The dual-catalyst approach for lean-burn exhaust after-treatment takes advantage of the stronger oxidizing potential of

NO2 compared to NO, which in turn helps to utilize the reducing capability of unburned hydrocarbons in the exhaust. Despite the exothermic oxidation of NO to NO2 being thermodynamically limited at high temperatures, the dual-catalyst mechanism helps to drive the NO oxidation reaction in the forward direction, as a result of NO2 being continuously removed via NO2-SCR. Any gaseous NO that is present in the system, a consequence of partially reduced NO2, can be re-oxidized by the oxidation catalyst. It is clear that the oxidation catalyst assumes a multi-functional role in the dual-catalyst

148 scheme. In addition to oxidizing NO or re-oxidizing partially reduced NO2 species, the oxidation catalyst also plays a role in catalyzing the combustion of un-burned hydrocarbons and the oxidation of carbon monoxide, which have not been consumed during the SCR reaction [228].

* Information in this chapter is adapted/taken from 1. P. Gawade, A.M.C Alexander, R.

Clark and U.S. Ozkan, Catalysis Today, Submitted (2012). 2. P. Gawade, A.M.C

Alexander, R. Silver and U.S. Ozkan, In Preparation (2012).

149

CHAPTER 11

THE ROLE OF OXIDATION CATALYST IN DUAL-CATALYST BED

11.1 Overview of the Role of Oxidation Catalyst in Dual-Catalyst Bed

In this contribution, the multifunctional role of a metal oxide-supported cobalt oxidation catalyst, namely CoOx/CeO2, in a dual-catalyst system is addressed. The effect of Co-loading in NO to NO2 oxidation has been investigated through steady state experiments. In order to optimize the dual bed composition, the ratio of reduction to oxidation catalyst was varied and investigated via steady state experiments under simulated engine exhaust conditions. Cyclic experiments were conducted in the presence of water vapor to understand the effect of different cobalt loadings on the overall hydrothermal stability of the dual-bed system. Both kinetic and in-situ DRIFTS studies are reported with the aim of understanding the behavior of the oxidation catalyst with respect to hydrocarbon oxidation, which consequently affects the overall NOx reduction performance.

150

11.2 Experimental

11.2.1 Catalyst Preparation

Palladium catalyst supported on sulfated zirconia with 0.3 wt% loading, which was employed as the NOx reduction catalyst for the dual-catalyst system, was prepared using a one-pot sol-gel method, details of which can be found elsewhere [223, 224]

Cobalt catalysts supported on ceria were prepared through an incipient wetness impregnation technique and were used as the NO oxidation catalyst component in the dual-catalyst system. The nanoparticle ceria support was prepared using a precipitation method described previously [159]. The calcined ceria support was impregnated with aqueous cobalt nitrate hexahydrate so as to yield catalysts with Co loadings of either 2 or

10wt% [229].

11.2.2 Catalyst Characterization

Surface area measurements were obtained using a Micromeritics ASAP

2010/2020 accelerated surface area and porosimetry instrument using a N2-physisorption method. Prior to analysis, the samples (0.1 g) were degassed overnight at 130 oC under a vacuum of 3µmHg to remove any adsorbed moisture. The specific surface areas of the catalysts, 0.3% Pd/SZ, 10% CoOx/CeO2, and 2% CoOx/CeO2, were determined by applying the Brunauer, Emmett and Teller (BET) method to the nitrogen physisorption

151 isotherms which were determined at liquid nitrogen temperatures (77K) and found to be

38 m2/g, 84 m2/g and 106 m2/g respectively.

Diffuse Reflectance Fourier Transform Infrared Spectroscopy (DRIFTS) was performed over individual reduction (0.3% Pd/SZ) and oxidation (2% CoOx/CeO2) catalysts and a dual-catalyst bed (0.3% Pd/SZ: 2% CoOx/CeO2 =8:1) during methane temperature programmed desorption (CH4-TPD). The DRIFTS instrument (Thermo

Nicolet 6700) was equipped with a liquid nitrogen-cooled mercury-cadmium-telluride

(MCT) detector and a DRIFTS chamber with ZnSe window. The DRIFTS spectra were collected in the mid-IR range with a resolution of 4 cm-1 averaged over 500 scans.

o Samples were pre-treated in situ at 500 C in 10% O2/He for 30 min at a flow rate of

30ccm, followed by helium flush at the same temperature for an additional 30 min. The background spectra were collected while cooling, under a He flow, at regular temperature

o intervals. Adsorption of 10% CH4/He was performed for 30 min at 50 C followed by

He flush for 30 min at the same temperature, to remove physisorbed CH4. Finally, the spectra were acquired, in He, at 50 oC increments up to a maximum temperature of 500 oC.

11.2.3 Catalyst Activity Testing

Steady-state reaction experiments were conducted at ambient pressure in a ¼”

O.D. stainless steel fixed bed reactor system. The schematic of NOx system is shown is

Figure 11.1. A physical mixture of reduction and oxidation catalysts were packed in desired quantities inside the reactor and held in place using quartz wool plugs. The reactor bed temperature was monitored and controlled using an Omega K-type

152 thermocouple, which was upstream of the catalyst bed and an Omega (model CS232) PID controller. The reactor was kept in a resistively heated homemade furnace. Brooks mass flow controllers (5850E) were used to regulate the gas flows to the reactor system. Water vapor was introduced to the reactor by saturating a stream of helium gas through a heated water bubbler containing de-ionized water. A chemiluminescence NOx analyzer

(Thermo-Scientific 42i-HL) was coupled with a micro-GC (Agilent 3000A) equipped with 0.32mm mol-sieve and PLOT Q columns with thermal conductivity detector to analyze the product gases. NO2 yield and conversions of hydrocarbons (XCHx) and NOx

(XNOx) were calculated using the following relationships:

NO2 yield = (moles of NO2 produced) /(moles of NO fed)

(XCHx) or (XNOx)= ([moles in]-[moles out])/(moles in)

11.2.3.1 NO Oxidation Over CoOx/CeO2

o The catalyst samples were pre-treated in 10%O2/He (40ccm) for 30 min at 450 C prior to catalytic activity testing. Steady-state NO oxidation experiments were performed on CoOx/CeO2 with two different cobalt loadings, namely 2 and 10% Co. These experiments were conducted in the temperature range of 200-400oC and at a gas hourly space velocity (GHSV) of 100,000 h-1. The feed composition was made up of 1000 ppm

NO, 10% O2 and balance He.

153

11.2.3.2 Effect of Reduction-to-Oxidation Catalyst Ratio

All the NOx reduction studies using hydrocarbon mixtures were carried out at a

GHSV 32,000 h-1 and under the following simulated engine exhaust composition 180 ppm NO2, 1737 ppm CH4, 208 ppm C2H6, 104 ppm C3H8, 650 ppm CO, 6.5% CO2, 10%

O2, (0-10%) water vapor and balance He, unless otherwise stated. The effect of the ratio of reduction to oxidation catalyst in dual-catalyst bed was evaluated in the temperature

o range of 300-500 C for NOx reduction performance. The ratio of reduction (0.3% Pd/SZ) to oxidation (10% CoOx/CeO2) catalyst was varied from 2:1, 4:1 and 8:1 (by wt) in the dual-catalyst bed. In order to keep the bed volume constant during reaction studies, quartz powder was mixed with the catalyst bed. The findings of these experiments

(section 3.2) revealed that lower amounts of oxidation catalyst in the dual-bed assisted the overall NOx conversion. Therefore, the amount of oxidation catalyst in dual-catalyst bed was further decreased using a lower cobalt-loading sample (2% CoOx/CeO2) and a similar experiment was performed under simulated engine exhaust conditions to evaluate its NOx reduction performance.

11.2.3.3 Hydrothermal Stability: Effect of Co Loading

The hydrothermal stability of the dual-catalyst bed was evaluated in the presence of 7% water vapor at 450 oC under simulated engine exhaust conditions as outlined above. Oxidation catalysts with different cobalt loadings (10% CoOx/CeO2 or 2%

CoOx/CeO2) were mixed with 0.3% Pd/SZ giving a reduction to oxidation catalyst ratio

154 of 8:1 (by wt). The cyclic experiments were conducted in order to understand the reversibility of the water effect on the dual catalyst bed.

11.2.3.4 Kinetics of Hydrocarbon Oxidation

The oxidation kinetics of the hydrocarbon mixture (methane, ethane and propane) over 0.3% Pd/SZ, 2% CoOx/CeO2 and the dual catalyst bed (Pd/SZ: 2% CoOx/CeO2=

o 8:1) was studied in the typical temperature range of 300-500 C with10% O2. Methane, ethane and propane concentrations were varied in the ranges of 889-2583 ppm, 107-311 ppm and 53-155 ppm, respectively. The reactor was treated either as a differential or an integral reactor depending upon the fractional conversion of hydrocarbons.

155

+,-."/# 0#+#(# +,-."/# +,-."/# 0#+#(# 0#+#(# +,-."/# 0#+#(#

+,-."/# 0#+#(#

34# J"7K.;/#

?7."/# 2(# @>AA-"/#

1#1#1#1# 1#1#1#1# 1#1#1#1# 2(# +,-."/# 1#1#1#1#

(;<=>."/# 5,/#$>.# @7--#:7-B"# 0"/"#I7>I"# !"# $%# ($)# *$)# (!)# (!)# :"6.# 0+(# 07EE#+-;H#(;6./;--"/# Figure 11.1: Schematic of Steady State Reaction System for HC-SCR

156

11.3 Results and Discussion

11.3.1 NO Oxidation over CoOx/CeO2

It is known that NO2 participates in the activation of hydrocarbons [230, 231], therefore the efficiency of the oxidation catalyst to perform oxidation of NO to NO2 is fundamental to the selective reduction of NO to N2. NO may exist in the engine exhaust, but it can also form through the partial reduction of NO2 during NO2-SCR, and it is necessary to convert it back into NO2. Therefore NO-to-NO2 oxidation over CoOx/CeO2 with different Co loadings was evaluated under steady state conditions, as illustrated in

Figure 11.2. As a reference, the thermodynamic equilibrium limitation (dashed line) for the conversion of NO to NO2 is also shown for the chosen feed conditions. An increase in