w- Wmfi W5
ii} I“ ‘2': IRWI {in W

I!“
'8‘?‘ Mi
. 3.3"! "A‘lII‘ijIU . gmi?‘fi::
‘ I“! III” 13%.! .n I”; “IL: A331“: MN. 1. ”4,“; ‘
:1 A“ l is"?! It. ".I! E“ II“ ‘I‘.{ {d LI f!”
I‘II ' :4-

n “.3“ ‘ 1:". ‘4 “MAJ tat.
IMIIII IA ’PYIVW

IIm‘u -

.pn. mgl.”
AI.

A “ " “III
[’4': I: ‘.~-
'“?WfiV“‘ AAIK"IWVIIZI

.I'." '

If" I". : . LYN 411,3"!

*3‘

x .
.

J}?
:h—q——~.~
“23:”...3: 3.51;:

-w-

-vI, I‘ {ad “3 . {3‘ Imij ‘ .th NIH} 1'” .
Tflfl”w~“fi$ P$Amh AA‘ kfififi?
. .3 fimfhé‘ 1- “.1413. i3. . “1 if? '4' $1 I 1 Z. 3; I I L‘gtgghb
in 3 .. . #31:: {S}. . . .
.3 3 ..;: m. , 33333 ~.:,’ .3313} LE; 3133K. 433,334
I 3I‘ 1.;33' .
' 3923:3414.

-r::‘1::
..
"n..-
.4..-

.
”2‘12:
.- ca

.04
.44.
.

1' Hz?"

.‘J.

O

a.

no

—A- ‘
_ .1.
1.1
:"‘ .45 ?.z.:?i‘=‘:v
wz‘xé. =

.4 .

"L113

nu.

:3!

v
.——4—.v

..4. yr .o.
3. vvo-t-
_. .

-
.
v».
.-
If

I.

I IHIIJMII

{1' ‘(HIH ”It”.
“ka .EEI': “Wk: .7th
“an: 'AH. :3 33
deem H.

.o~

r-
-o

4

- -—..'i'14:-o-.

« .,:--

’7.“ ..<.- Z - ‘-

”£653 4‘3! «33%;: if}:
\Lff‘

3:95.. “13.! ”1‘

I ":I'
3%,. '
{I ‘ 2:1! “:'I§' I'. ..
I '. {3"-

.1
..
.1. . 3
:1 -“~‘:.
' 'fiz‘f‘ifis
i," 3. . “:1
w. N
mfg
g}? 1. .-.
1 ..
'5'"! or: TV“:

a..--:

$5
A“ d0}?
1:21»,

.313? #5."!
I

‘4"iA
‘: PI “531‘3,
4 :3

2'5“; oi'} "'1? 'L“

I‘wl

w. ..
..1 ..

' .--u. A":
.

2..

. .‘"'_' "1 7
,——. " .-.-.‘~ . 1.-

,“ -4.“ .. 1 < ‘
..

 

— -v‘-—1.'

I
(IA! 5‘
I'~I_‘

III ;

INIII“

 

ll!I!!!"HIIIIIIIIIIHIHIIHHIIIIIiJill!!!Will/l

301397 9525

This is to certify that the

dissertation entitled

CR ‘ “
mmic ccmaes‘iw 0%? meme 1%ka “3

BEEWRTt VES 10 C‘QMM ohfl‘i RN75 STP'EQxRLT‘l
(HEMmMé: (-s'mq Sts (WfikfieTEinAfioH Mm kkfimfi
MEQHNVu‘bH presented by

New GPEze‘iUH

has been accepted towards fulfillment
of the requirements for
.3 _~ 1., u
4i— ;k . b (I ‘V (a (Mt3t7

degree in

 

 

 

Date AmJJbLg‘l 3 O. 1‘37;

MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771

 

LIBRARY

Michigan State .
University

 

 

 

 

PLACE lN RETURN Box
to remove this checkout from your record.
TO AVOID FINE-3 return on or before date due.

 

DATE DUE

DATE DUE

 

 

JUL 1 112009

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/98 c/CIRC/DmDuaJJGS-ou

 

CATALYTIC CONVERSION OF LACTIC ACID AND

ITS DERIVATIVES TO COMMODITY AND SPECIALTY CHEMICALS:

CATALYSTS CHARACTERIZATION AND REACTION MECHANISM
By

Radu Craciun

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

1997

ABSTRACT

CATALYTIC CONVERSION OF LACTIC ACID AND
ITS DERIVATIVES
TO COMMODITY AND SPECIALTY CHEMICALS

By
Radu Craciun

Inexpensive and non-toxic alkali (Na, K, CS) and alkaline earth metal (Ba)
hydroxides supported on SiOz catalysts were successfully used to produce 2,3-
pentanedione (23P) from lactic acid with high yield and selectivity along with other
valuable chemicals (acrylic and propanoic acid). Temperature (280°-350°C), pressure (4-
6 atm), and metal hydroxide loading (0.1 - 3 mmole/g support) on the SiOZ support were
varied in order to optimize the selectivity to 23P. The yield and selectivity toward 23P
increase in the order Ba <Na < K < Cs with optimized conditions and catalysts giving
23P yields as high as 65% with selectivity up to 80%. Various analytical techniques have
been used to characterize the catalysts before and after exposure to the reaction
conditions. The reaction intermediates formed by the condensation reaction of two lactic
acid molecules to 23P have been evaluated based on product analysis, deuterium labeling
studies, and computational modeling using semi-empirical (PM3) and ab initio (STD-3G,
3-21G, and 6-3lG* levels) molecular orbital calculations. The latter theoretical
calculations have been used to evaluate variations in energy for each step of the proposed

mechanism. At optimum temperature 23P, H20, and C02 are continuously removed from

the catalyst surface, so the equilibrium steps in the mechanism are shifted toward product
formation.

Several cerium and manganese oxide supported catalysts have been successfully
applied to two catalytic oxidative dehydrogenation processes: conversion of methyl
lactate to methyl pyruvate and aromatization of 2-ethyl-3-methyl-dihydropyrazine
(obtained from condensation of 23P with ethylenediamine) to 2-ethyl-3-methylpyrazine.
Bulk and surface characterization techniques were used to characterize the unpromoted
and CeOz-promoted MnOx/SiOz catalysts. Sequential impregnation (cerium followed by
manganese), led to mixed Ce-Mn supported oxide catalysts that are active in oxidative
dehydrogenations. The Mn-Ce/Si02 catalytic activity in the above-described reactions

are reported in correlation with their structure.

TABLE OF CONTENTS

LIST OF TABLES .............................................................................................................. vi
LIST OF FIGURES .......................................................................................................... viii
LIST OF SCHEMES ........................................................................................................ xvi
INTRODUCTION ............................................................................................................... 1
CHAPTER 1

ALKALI AND ALKALINE EARTH METAL HYDROXIDE SUPPORTED ON Si02
FOR CATALYSTS USE IN LACTIC ACID CONVERSION

1.1. Introduction ...................................................................................................... 5

1.2. Experimental .................................................................................................... 8

1.3. Results and discussion ................................................................................... 34

1.4. Conclusions .................................................................................................... 60

1.5. References ...................................................................................................... 62
CHAPTER 2

THE MECHANISM OF CATALYTIC LACTIC ACID CONVERSION TO
2,3-PENTANEDIONE: CORRELATION OF EXPERIMENTAL AND
MOLECULAR MODELING DATA

2.1. Introduction .................................................................................................... 64

2.2. Experimental .................................................................................................. 65

2.3. Results and discussion ................................................................................... 69

2.4. Conclusions .................................................................................................... 94

2.5. References ...................................................................................................... 96
CHAPTER 3

DESIGN OF UNPROMOTED AND RARE EARTH PROMOTED MnO x/SiOz
CATALYSTS USED IN OXIDATIVE DEHYDROGENATION REACTIONS

3.1. Introduction .................................................................................................... 98
3.1. Rare earth promoted catalysts ...................................................................... 100
3.2. Supported MnOx catalysts used in oxidative dehydrogenation reactions....l 13
3.3. References .................................................................................................... 118

CHAPTER 4
SYNTHETIC METHODS FOR PREPARATION OF
CeOz-PROMOTED SiO2 CATALYSTS

4.1. Introduction .................................................................................................. 122

4.2. Experimental ............................................................................................... 123

4.3. Results and discussion ................................................................................. 127

4.4. Conclusions .................................................................................................. 151

4.5. References .................................................................................................... 153
CHAPTER 5

CHARACTERIZATION OF UNPROMOTED AND CeOz-PROMOTED
Mnox/Sio2 CATALYSTS

5.1. Introduction .................................................................................................. 156
5.2. Experimental ................................................................................................ 158
5.3. Results and discussion ................................................................................. 171
5.4. Conclusions .................................................................................................. 200
5.5. References .................................................................................................... 203
CONCLUDING REMARKS ........................................................................................... 205

Table 1.1

Table 1.2

Table 1.3

Table 1.4

Table 1.5

Table 1.6

Table 2.1

Table 2.2

Table 2.3

Table 2.4

Table 2.5

Table 4.1

Table 4.2

 

LIST OF TABLES
Parameters for MAS 3|P-NMR data acquisition ........................................ 14
Comparison of IR techniques for used for catalyst characterization ......... 28

Experimental conditions used in lactic acid conversion on MOH/SiOz
catalysts (M = Na, K, Cs, Ba) .................................................................... 34

Yields (selectivities) from lactic acid conversion over sodium phosphate
salts supported on Si/Al catalyst, at 300° and 350°C, respectively ........... 35

Spin-lattice relaxation time (T,) values for Si/Al supported NaH2P04
catalyst, calcined at 450°C, evaluated from the exponential dependence
in Figure1.11 .............................................................................................. 41

Yields and selectivities from lactic acid conversion over Ba(OH)2 and

NaOH supported on SiOz, at 280°C, with 1 mmole MOH/g Si02
loading ........................................................................................................ 54

Experimental and calculated enthalpies of formation (AH°f) for various
carboxylic acids ......................................................................................... 66

PM3 and ab initio data for reactants, intermediates and products involved
in the mechanism of lactic acid conversion to 2,3-pentanedione ............. 72

Relative energies calculated in gas phase and solvated models from PM3
and ab initio methods for C6HI3O7' species at each step of the proposed
mechanism, normalized to the energy of the starting materials ................. 73
a-Deuterium incorporation into lactic acid (GC-MS data) ........................ 79
Calculated vibrational frequencies for OH and C-D stretch bands in
(at-position of isopropanol from PM3 and ab initio 6-31G*

data in comparison with the experimental values ...................................... 82

BET, semi-quantitative XRD, and XPS analysis data for Ce-promoted
Si02 catalysts ........................................................................................... 128

Tentative IR band assignments for cerium treated Si02 catalysts ........... 130

vi

Table 4. 3

Tabel 5.1

Table 5.2

Table 5.3

Tabel 5.4

Table 5.5

29Si NMR peak intensity ratios and XPS ILa3d/15i2p ratios for silica in B, C,
and D configurations relative to A (see Scheme 4. l) ............................. 137

The list of the unpromoted and CeOz-promoted MnOx/SiOZ catalysts
together with the designated symbols used in this study ......................... 159

XRD and XPS MnOx dispersion data and MnOZ/Mn203 ratios evaluated
from diffraction patterns for Mnyn500 catalysts (y = 05-14 = Mn/ Si
atomic ratio) ............................................................................................. 173

XRD particle size evaluated from line broadening calculations for B-MnOz
and a-Mn203 species on Mn/Si = 0.10 (atomic ratio) catalyst calcined at
various temperatures (125°-750°C temperature domain) ........................ 180

MnOx particle size evaluated from XRD line broadening calculations, XPS
surface concentration, and Mn dispersion data for unpromoted and
CeOz-promoted MnOx/SiOz catalysts ...................................................... 184

Conversions and selectivities for methyl lactate and 2-ethyl-3-methyl-
dihydropyrazine catalytic oxidative dehydrogenation on unpromoted and
CeOz-promoted catalysts ......................................................................... 193

vii

Figure 1.1

Figure 1.2

Figure 1.3

Figure 1.4

Figure 1.5

Figure 1.6
Figure 1.7

Figure 1.8

Figure 1.9

Figure 1.10

Figure 1.11

Figure 1.12

LIST OF FIGURES

Lactic acid conversion pathways toward various chemicals ........................ 6

Schematical representation of magnetic field (H0) interaction with the
dipolar internuclear vector Rij’ and the equation which describes this
process [22] .................................................................................................. 9

Spin-lattice relation time (T1) evaluation from a double-pulse MAS -
NMR experiment: a) pulse configuration; b) NMR signal at various delay
time array, d2 .............................................................................................. 13

Schematic representation of IR beam interaction with the sample ............ 17

Optical attachment (collector): a) on axis (reprinted with permission of
Spectra-Tech 1nc.; b) off axis (reprinted with permission of Harrick
Scientific Co. for diffuse reflectance spectra collection ............................. 23

In-situ diffuse reflectance IR cell (Harrick Scientific Co.) ........................ 25
Schematic representation of the parameters from the Bragg equation ...... 30

Schematic representation of the laboratory setup used for lactic acid
conversion studies and catalysts testings ................................................... 33

MAS-NMR spectra for the NaH2P04 salt, a) calcined at 450°C; and b)
aged ............................................................................................................ 37

31P-NMR spectrum of Si/Al supported NaH2P04 calcined at 450°C ........ 38

31P-NMR intensity versus time [s], for the peaks specific to the Si/Al
supported NaH2P04 catalyst, calcined at 450°C, where #1 = 13.259 ppm,
#2 = -11.851 ppm, #3 = ~32.586 ppm, #4 = -37.186 ppm, and

#5 = -66.518 ppm ....................................................................................... 40

3 lP-NMR of Na3PO4 salt and Na3PO4 supported on Si/Al catalysts:

a) Na3PO4x12H20; b) Na3PO4 calcined at 450°C; c) Na3PO4/Si/Al catalyst
dried at 100°C; (1) Na3PO4/Si/Al catalyst dried at 450°C; e) used
Na3PO4/Si/Al catalyst in lactic acid conversion; I) regenerated
Na3PO4/Si/Al catalyst by calcination at 450°C .......................................... 42

viii

Figure 1.13

Figure 1.14

Figure 1.15

Figure 1.16

Figure 1.17

Figure 1.18

Figure 1.19

Figure 1.20

Figure 1.21

Figure 2.1

Figure 2.2

31P-NMR of Na3PO4 supported on Si/Al catalysts: a) regenerated at 6 kHz;
b) regenerated at 4 kHz; c) used at 6 kHz; (1) used at 4 kHz ...................... 44

DRIF TS spectra for Si/Al supported Na3PO4catalysts: a) following lactic
acid exposure at 320°C and b) fresh, compared with c) pure Na3PO4
salt .............................................................................................................. 46

DRIF TS Spectra for Ba/Na/hSiOz catalyst acquired at different times
during preparation: a) pure hSiOz catalyst ; b) hSi02 treated with
CH3-ONa ; c) Ba(NO3)2/NaOH on hSi02 catalyst dried at 125°C ;

d) Ba/Na/hSi02 catalyst calcined at 400°C ................................................ 49

DRIF TS spectra for Ba/Na/lSiOz catalyst acquired at different times
during preparation: a) pure lSi02 catalyst ; b) lSi02 treated with CH3-ONa;
c) Ba(NO3)2/NaO-lSi02 catalyst dried at 125°C; (1) Ba/Na/lSi02 catalyst
calcined at 400°C ....................................................................................... 50

XRD spectra for hSi02 supported Ba/N a and NaOH catalysts: a) Ba/N a
calcined at 400°C; b) Ba/Na dried at 100°C; c) NaOH calcined at 400°C;
d) pure hSi02 support ................................................................................. 52

XRD spectra for lSiOzsupported Ba/Na and NaOH catalysts: a) Ba/Na
calcined at 400°C; b) Ba/Na dried at 100°C; c) NaOH calcined at 400°C;
d) pure ISi02 support ................................................................................. 53

Conversion of lactic acid to 23P as a function of temperature, for hSiOz
supported MOH (M = Na, K, Cs) catalysts: a) yields; b) selectivities ...... 57

Temperature dependence of lactic acid conversion to acetaldehyde for
hSi02 supported MOH (M = Na, K, Cs) catalysts: a) yields,
b) selectivities ............................................................................................ 58

Activity data for lactic acid to 2,3-pentanedione conversion on hSiOz
supported MOH (M = Na, K, Cs) catalysts, as a function of MOH loading:
a) yields, b) selectivities ............................................................................. 59

Experimental setup for variable-temperature-mass spectrometry (VT-MS)
experiments: 1) temperature regulator; 2) mass spectrometer; 3) capillary
sample holder ............................................................................................. 68

DRIFTS spectra of KOH/Si02 catalysts after exposure to lactic acid vapor
at various temperature: a) 350°C; b) 300°C; c) 250°C; (1) 200°C; e) 150°C;
and before exposure at t) 25°C .................................................................. 75

Figure 2.3

Figure 2.4

Figure 2.5

Figure 2.6

Figure 2.7

Figure 2.8

Figure 2.9

Figure 2.10

Figure 2.11

Figure 3.1

Figure 4.1

Figure 4.2

Figure 2.3 Mass spectra of standards: a) lactic acid and b) dilactide ........ 77

IR spectra of the lactic acid a) after and b) before passing over a
KOH/SiOz catalyst, in comparison with c) the dried lactic acid ................ 81

2H-NMR spectra of dried lactic acid after H/D exchange over a KOH/Si02
catalyst at a) 250°C and b) 200°C in comparison with c) the lactic acid just
diluted in D20 and dried ............................................................................ 83

Mass spectra of products obtained on a 2 mole KOH/g SiOz catalyst by
VT-MS analysis after: a) 1.5 min.; b) 2.5 min.; c) 5.2 min of heating ...... 86

VT-MS data (16°C/min ramp rate) using a 2 moles KOH/g Si02
catalyst impregnated with lactic acid: a) abundance variations versus time
for ions of interest; b) mass spectrum obtained after 2.13 minutes ........... 89

VT-MS data (8°C/min ramp rate) using a 2 moles KOH/g SiOz
catalyst impregnated with lactic acid: a) abundance variations versus time
for ions of interest; b) mass spectrum obtained after 2.09 minutes ........... 90

VT-MS data (8°C/min ramp rate) using 8102 support impregnated with
lactic acid: a) abundance variations versus time for ions of interest;
b) mass spectrum obtained after 2.48 minutes ........................................... 91

VT-MS data (8°C/min ramp rate) using a 2 moles KOH/g SiOz
catalyst impregnated with dilactide: a) abundance variations versus time
for ions of interest; b) mass spectrum obtained after 1.43 minutes ........... 92

VT-MS data (8°C/min ramp rate) using a 2 moles KOH/g SiOz
catalyst impregnated with sodium lactate: a) abundance variations versus
time for ions of interest; b) mass spectrum obtained after 0.5 minutes ..... 93

Schematic representation of the catalytic cycle for a selective oxidation
process which occurs on a mixed metal oxide catalyst surface; M, and M2
are two different transition metals (x,y = number of the electrons
exchanges in the redox process by the M, and M2, and n and m are the
oxidation numbers for the MI and M2) [3] ............................................... 101

DRIF TS spectra of standard materials: a) Si02 support; b) CeOz;
c) Ce(OH)4 ............................................................................................... 129

DRIFTS spectra of CeqN catalysts: a) SiOz support; b) CeqN catalyst
dried at 125°C; c) CeqN calcined at 500°C ............................................. 132

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

Figure 4.10

Figure 5.1

Figure 5.2

Figure 5.3

Figure 5.4

DRIFTS spectra of Cqu catalysts: a) Si02 support; b) Cqu catalyst
dried at 25°C; c) Cqu dried at 125°C; (1) Cqu calcined at 200°C;
e) Cqu calcined at 300°C; 0 Cqu calcined at 500°C .......................... 134

29Si NMR spectra of the catalysts dried at 125°C: a) Cqu; b) CeqN;
c) Si02 in comparison with the deconvoluted spectra for each samples
(a’, b’ and c’) ............................................................................................ 136

Semiquantitative XRD data: a) XRD pattern of Ce02/Si02 physical
mixture with different CeOz loading; b) corresponding calibration
curve ......................................................................................................... 139

XRD patterns of: a) SiOz; b) CeqN calcined at 500°C; c) CeqN calcined at
800°C; (1) Cqu calcined at 500°C; 6) Cqu calcined at 800°C ............. 141

TPR profiles of: a) Cqu and b) CeqN catalysts calcined at 500°C ........ 143

EPR spectra of: a) CeOz; b) CeqN calcined at 500°C; c) CeqN calcined at
800°C; d) Cqu calcined at 500°C; e) Cqu calcined at 800°C ............. 146

XPS spectra of: a) CeqN calcined at 500°C; b) Cqu calcined at 500°C;
C) C602 ..................................................................................................... 148

Cerium XPS photoreduction spectra for Cqu catalyst calcined at 500°C
after: a) 5 min.; b) 15 min.; c) 30 min.; (1) 4 h of scanning ...................... 150

Oxidative dehydrogenation processes considered as probe reactions for
unpromoted and CeOz-promoted MnOx/SiOz catalysts ........................... 157

Laboratory setup for a TPR experiment: 1 - thermal conductivity cell;

2.1 - reduction valve; 2.2 - M/Ale3 catalyst; 2.3 - molecular sieves;

2.4 - Dewar trap (193 K); 2.5 - gas flow switch; 2.6 - barke capillary;

3 - reactor; 4 - filmace; 5 - temperature programmer; 6 - chart recorder;

7 - thermocouple ...................................................................................... 162

The Zeeman energy levels of a free electron placed into an external
magnetic field ........................................................................................... 164

Schematic representation of the photoelectric effect ............................... 167

xi

Figure 5.5

Figure 5.6

Figure 5.7

Figure 5.8

Figure 5.9

Figure 5.10

Figure 5.11

Figure 5.12

Figure 5.13

Figure 5.14

Figure 5.15

Schematic representation of the Kerkhof-Moulijn model for a supported
catalyst surface: 1p,s = XPS intensity; D‘s”) = detector efficiency; (p/S) =
atomic ratio; op.S = photoelectron cross section; B = UK, A = escape depth
factor (t = 2/pSo, where p = density, and So surface are of the support)..l69

XRD pattern for Mn-based catalysts calcined at 500°C: a) Mnl4n500;
b) Mn10n500; c) Mn06n500; d) Mn02n500; e) Mn005n500 .................. 172

Mn monolayer dispersion (rhombus) evaluated from Kerkhof and Moulijn
model in comparison with the experimental value (squares) for:
a) Mn-based catalysts; b) Mn10n calcined at various temperatures ........ 174

anp XPS spectra for Mn-standard compounds: a) pure Mn203; b) Mn203
obtained from calcination of Mn(NO3)2 at 750°C; c) pure MnOz; d) MnOz
obtained from calcination of Mn(NO3)2 at 300°C ................................... 176

TPR profiles for the Mn-based catalysts calcined at 500°C: a) Mn14n500;
b) Mn10n500; c) Mn06n500; d) Mn02n500; e) Mn005n500;
0 pure Si02 .............................................................................................. 177

XRD patterns for Mn10n catalysts calcined at various temperatures:
a) Mn10n750; b) Mn10n650; c) Mn10n575; d) Mn10n500; e) Mn10n400;
t) Mn10n300; g) Mn10n125 .................................................................... 179

anp XPS spectra for Mn10n catalysts calcined to various temperatures:
a) Mn10n750; b) Mn10n650; c) Mn10n575; d) Mn10n500; e) Mn10n400;
f) Mn10n300; g) Mn10n125 .................................................................... 182

TPR profiles of Mn10n catalysts calcined at various temperatures during
preparation: a) Mn10n750; b) Mn10n575; c) Mn10n500;
d) Mn10n300 ............................................................................................ 183

XRD patterns of unpromoted and CeOz-promoted MnOx/SiOZ catalyst
(y = 10) calcined at 500°C: a) a) MnlOncea; b) Cqu; c) Mnl Oncen;
d) CeqN; e) Mn10n500 ............................................................................ 185

TPR profile of the unpromoted and Ce02 promoted MnOx/SiOz and SiOz
catalysts: a) Mn10n500; b) MnlOncea; c) MnlOncen; d) Cqu;
e) Cqu .................................................................................................... 188

EPR spectra for the CeOz-promoted MnOx/SiOZ catalysts: a) MnlOncea;
b) Cqu; c) MnlOncen; d) CeqN ............................................................. 190

xii

Figure 5.16

Figure 5.17

Figure 5.18

Figure 5.19

Schematic representation of unpromoted and CeOz promoted MnOx/SiOz
catalysts .................................................................................................... 192

XRD patterns of used unpromoted and CeOz-promoted catalysts:
a) Mn10n500; b) MnlOncen; c) MnlOncea ............................................. 196

TPR profiles of the fresh and used unpromoted MnOx/SiOZ catalysts:
a) used Mn10n500; b) fresh Mn10n650; c) fiesh MnlOnSOO .................. 197

anpm XPS spectra of the used unpromoted and CeOz-promoted catalysts
in comparison with standard MnOx materials: a) Mn203 (Mn3O4);
b) MnOz; c) MnlOncea; d) MnlOncen; e) Mn10n500 ............................. 199

xiii

LIST OF SCHEMES

Scheme 2.1 The proposed mechanism of lactic acid conversion to 23P ....................... 70
Scheme 2.2 Proposed mechanism for a H/D exchange from lactic acid molecule ....... 78

Scheme 4.1 Silica structures and the corresponding NMR chemical shifts ................ 135

xiv

INTRODUCTION

Among many fascinating phenomena of natural chemistry, enzymatic catalysis is
one of the most intriguing processes. Our knowledge of catalysis is far from what nature
can do in these enzymatic processes. We have tried to mimic such processes and apply
them in many fields of chemical industry, in some cases quite successfully. In recent
years, US corn production in excess of food and feed requirements has allowed scientists
to develop biocatalytical processes for conversion of biomass materials to specialty
chemicals and commodities.

Today, lactic acid is one of the natural products which can be obtained from
biofermentation processes in industrial quantities. Lactic acid had been known for many
years (first isolation in the 18th century by Scheele), being identified in various micro-
organisms as an important component in their metabolism. It was identified in fermented
rice water as well as in many other fermented plant liquors. Maybe lactic acid is best
known as the source of the sour teste in the milk or as the accumulation in muscle tissues
which produces the muscular pain after hard physical exercises. From a chemical point
of view, lactic acid, is a bifunctional (hydroxyacid) organic compound, optically active,
with a strongly acid taste, and low volatility; when pure and anhydrous, it is a white
crystalline solid with low melting point. Generally, lactic acid is available in the form of
concentrated (SO-90%), viscous, colorless and odorless liquid aqueous solution. Lactic
acid can easily undergo intermolecular esterification to form lactoyllactic and higher

chain polyesters (oligomers). It exhibits a great variety of chemical reactivity owing to

the presence of the two functional groups in the molecule. To date, several hundred
derivatives have been discovered and listed in various books. Holten and coauthors [1] in
their review have nicely synthesized a significant amount of information about lactic acid
and its derivatives, describing in details its physical properties, preparation, spectra,
chemical reactions, optical activity, lactate salts, intermolecular esters, biochemistry, etc.
This work is focused on development of catalytic chemical processes which can
add value to lactic acid obtained by biofermentation and open new ways for its industrial
utilization. Previous studies on lactic acid’s industrial utility have been focused on
conversion to acrylic and propanoic acid [2-8]. Catalytic condensation to 2,3-
pentanedione (23P) and oxidation to pyruvic acid are two additional interesting and
important processes which lactic acid can undergo. Research on catalyst design,
structure-activity correlations, structural modification of catalysts during their utilization
in the condensation process, mechanistic studies, and reaction parameter optimization for
lactic acid catalytic conversion into 23P, are the main topics described in the first two
chapters of this dissertation. The interest in the study of the 23P reaction pathway
discovered at MSU came from its novelty and unexpected chemical outcome. 23P is a
chemical used mostly in the food industry as a flavor agent and food additive in relatively
small quantities. It is an important intermediate in many organic syntheses, being able to
participate in condensation processes, due to the vicinal carbonyl groups from the
molecule. It can easily undergo self-condensation to form duroquinone or intermolecular
condensation with various diamines such as ethylenediamine or orthophenylendiamine.
Condensation products can be further oxidized to obtain various types of pyrazines.

Many food products such as coffee, roasted beef or potatoes contain pyrazines conferring

a nutty roasted flavor. Such pyrazine-based compounds have utility in a large number of
biological (herbicide) and drug synthesis applications (additive to various antibiotics).
Cerium and manganese oxides are among the most common solid catalysts used
in oxidation reactions. Cerium and manganese catalyst design, novel preparation
techniques, catalyst structure characterization using modern analytical techniques, and
catalyst activity tests in oxidation of methyl lactate to methyl pyruvate and of 2-ethyl-3-
methyl-dihydropyrazine (DHPy), obtained from condensation of 23P with
ethylenediamine, to form 2-ethyl-3-methylpyrazine (Py), will be presented in detail in

chapters 3-5.

9‘95“!"

“95”.“?

References

Holten, C. H.; Muller, A.; Rehbinder, D. Lactic Acid - Properties and Chemistry
of Lactic Acid and Derivatives, Verlag Chemie, 1971.

Nakel, G. M.; Dirkc, B. M. US Patent 3, 1971, 579, 353.

Holmen, R. E. US. Patent 1958, 2, 859, 240.

Maresca, L. M. US. Patent 4, 1986, 611, 033.

Matsumoto, T.; Yamada, E.; Nakachi, O. Komai, T. Japanese Patent 61, 1986,
243,807 (CA 106:138889k)

Paparizos, C.; Dolhyj, 8.; Shaw, W. G. US. Patent 4, 1988, 786, 756.

Sawicki, R. A. US. Patent, 1988, 729, 978.

Moffat, J. B. Catal. Rev. -Sci. Eng. 1978, 19(2), 199.

Monma, H. J. Catal. 1982, 75, 200.

Mok, W. S. L.; Antal, M. J. Jr.; Jones, M. Jr. J. Org. Chem. 1989, 54, 4596-4602.
11. Lira, C. T.; McCrackin, P. J. Ind. Eng. Chem. Res. 1993, 32. 2608.

CHAPTER 1

ALKALI AND ALKALINE EARTH METAL HYDROXIDES CATALYSTS
SUPPORTED ON SiOz FOR USE IN LACTIC ACID CONVERSION

1.1 Introduction

Lactic acid (2-hydroxy-propanoic acid) is a bifunctional, optically active molecule
traditionally used as food additive or in textile industry. Its production via efficient
starch-based fermentation processes, has raised interest in potential applications for its
further conversion into other high value Specialty and commodity chemicals, as well as
its use in production of biodegradable polylactide polymers [1-2]. The decreased cost
associated with the application of new biotechnology processes suggests that lactic acid
could become a major biomass-based chemical feedstock in the near future. Lactic acid
conversion can lead to a variety of products; transformations pertinent to this work are
given in Figure 1.1. Reduction to propanoic acid, dehydration to acrylic acid,
dehydrogenation to pyruvic acid, and condensation to 2,3-pentanedione (23P) are some of
the pathways from lactic acid to valuable chemicals. Decarbonylation/decarboxylation to
acetaldehyde is another possible process, which usually is undesired and should be
minimized. Direct dehydration of lactic acid to acrylic acid has long been of interest as a
potential route to acrylate polymers from biomass; consequently many studies have
focused on this pathway. Phosphate salts are known to be good catalysts for the alcohol
dehydration reaction [6-7]. However, the highest literature conversion over supported

sulfate and phosphate catalysts reached only 58% yield and 65% selectivity to acrylic

acid, at 350°-400°C [3-5]. Lactic acid conversion in supercritical conditions using

phosphate catalysts reached acrylic acid yields of only 23% with 58% selectivity, at 310

bars and 360°C [8-9].

CH3—CH=O + C0 + H20 (decarbonylation)
/ Acetaldehyde + CO2 + H2 (decarboxylation)

OH
(I'IH ————> CH2=CH—-COOH + H20 (dehydration)
/ \ Arylic acid

CH3 COO“ \

° ' — — + 1/2 d :-
Lactic acid CH3 CH2 C00“ 02 (re uc Ion)

\1/20: Propionic acid

CH3 /COOH
\ICI + H20 (oxidation)
P 0' 'd
ruv1c am
i” y 1
Catalyst CH2 C
2 /c3 ——A—-> CH3/ \ C/ \ CH3 + C02 + 2 H20
CH3 OOOH Id
Lactic acid 2,3-Pentanedione
Figure 1.1 Lactic acid conversion pathways toward various chemicals

Previously, our group studied lactic acid conversion using simple phosphates [1]
and other salts [10-11] as catalysts on various type of supports. These studies led to the
discovery of lactic acid catalytic conversion toward 23P at mildly elevated pressures (0.5
MPa) and temperatures (250-350°C). It was found that the yield of 23P is enhanced with
increasing basicity of the salt catalyst, and that low surface-area supports with well
defined porosity are preferred to avoid lactic acid carbonization and catalyst fouling.

Moderately basic catalysts such as sodium phosphates (disodium or trisodium) form

meaningful quantities of 23P and acrylic acid. The Si-Al support alone has almost no
activity for 23P and acrylic acid formation. The acid sites from the neat supports provide
good catalytic activity toward decarbonylation/decarboxylation to form acetaldehyde and
coking of the support as dominant reaction pathways. Simple pyrolysis of lactic acid
gives the same products (C02, H20 and actaldehyde). Addition of basic sodium
phosphate salts neutralizes these sites and reduces the extent of undesirable reactions.

Alkali and alkaline-earth metals hydroxides are strong bases and have been used
as catalysts in various types of base—catalyzed reactions such as alkene isomerization and
cumene dehydrogenation on MgO/NaOH [12-13], Claus reaction (HZS + S02) on
NaOI-I/Si02 [14], aldol condensation reaction on NaOl-I/silica-gel [15], Knoevenagel
condensation on AlPO4/alumina [l6], dehydration of methyl lactate to methyl acrylate on
CaSO4 [17], and other organic reactions [18]. A good correlation between the electron-
donor properties of silica supported alkali metal hydroxides and catalytic activity have
been found in all these reactions [12]. Metal hydroxide loading and reaction temperature
played a key role in the optimization of reaction yields, selectivity and the kinetics
observed [19].

This work describes a study aimed to elucidate: i) the chemical transformations
which occur on sodium phosphate and barium nitrate supported catalysts during
preparation and after exposure to the lactic acid feed; ii) correlation of catalyst structure
with catalytic activity; iii) the effects of the type of silica used as support for various
alkali metal hydroxides catalysts; and iv) the effect of alkali metal hydroxide loadings

and reaction temperature on yields and selectivity toward 23P formation.

1.2 Experimental

Materials and Catalyst Preparation. Lactic acid 85% in aqueous solution,
crystalline sodium lactate (98%), NaOH, KOH, CsOH, Cal0(OH)2(PO4)6, and
NaH2P04xH20, were obtained from Aldrich. Solid Na3PO4x12H20 was obtained from
EM Science. Crystalline Ba(NO3)2, and Na5P3Om were obtained from Mallinckrodt.
Various Si-based porous materials such as the Si-Al (93% SiOZ, 7% A1203) with low
surface area (5.1 mZ/g, Johnson-Matthey), controlled-pore glass SiOz (CPGO3000D, 7.2
mz/g, CPG, Inc.) and Si02 high surface area (300 mz/g) prepared from silica-gel (Davison
Chemical Co.) by grinding (240 mesh) and calcining at 500°C were used as supports for
the alkali/alkali-earth metal salts. Sodium phosphate catalysts used Si-Al whereas the
other salts used SiOz as supports (low and high surface area). Two different silica
supports, a low surface area, with well controlled pore size distribution (lSiOz, 6.2 mZ/g)
and a high surface area (hSiOz, 280 mz/g), were used for barium nitrate/sodium-
methoxide impregnation. The supports were impregnated first with CH3ONa followed by
impregnation with Ba(NO3)2 aqueous solution. After impregnation the catalysts were
dried (125°C) and calcined (400°C). The MOH catalysts were prepared in a Similar
manner, but the calcination step was necessary. The catalyst loading utilized in the lactic
acid conversion varied from 0.1 to 3 mole of MOH(M(NO3)x)/g of support (where M is
an alkali or alkaline-earth metal). The catalysts were prepared by the incipient wetness
impregnationI method using an alkali metal salt (MOH) aqueous solution as precursor.

After impregnation the catalysts were dried at 125°C and calcined at 500°C, in air. A

 

l this methods implies the impregnation with a volume of precursor solution equal with the total pore
volume of the support

catalyst deactivation study was performed by analyzing the used and regenerated catalysts
(by calcination at 450°C).

Solid state 31P-MAS-NMR spectroscopy. Magic angle spinning (MAS) - nuclear
magnetic resonance (NMR) spectroscopy has been widely used to characterize inorganic
solids. This spectroscopic technique has proven to be an excellent tool to obtain
information about the structure of various phosphate salts [20-21]. The magic angle can
be explained based on the drawing in Figure 1.2. The nucleus executes a circular motion
about the axis of rotation and if the motion is fast enough, the nucleus can be considered

at the center of the circle (on the axis).

H0 11

/\ MW

 

(1/R3,,-) (3c0320-1)

Figure 1.2 Schematic representation of magnetic field (H0) interaction with the
dipolar internuclear vector Ru and below, the equation which describes the process [22].

In liquids, isotropic and rapid tumbling average out direct dipolar coupling and
chemical shift anisotropy by motional narrowing. In solids, these interactions are usually
not completely averaged out due to the absence of molecular tumbling and diffusion and
insufficient motional narrowing. The dipolar interaction between a pair of nuclei i and j
is described by the equation in Figure 1.2, where 0 is the angle between Rij and H0

vectors. The angular dependence term reduces to zero for 0 = 54.74°, the so called magic

10

angle. At this 6 value the internuclear vectors are lined up to the maapplied magnetic
field, and the dipolar interactions vanish [22]. In most solids, the nuclear magnetic dipole
is coupled to those of its neighbors, and only some of the pairs will be near the magic
angle. However, when the sample is rotated fast enough, around an axis fixed at the
magic angle value, the time average orientation of all the internuclear vectors is parallel
to each other and to the rotation axis. When the angle between Ho and the rotation axis is
54.74°, the 3c0520 - 1 term is zero [22]. This technique, which minimizes the dipolar
interaction in a solid, has found application in pulsed NMR experiments on solid
materials [20-26].

Chemical shifts of phosphate atoms can be observed in specific regions as a
function of the type of phosphate analyzed (orthophosphates, metaphosphates, or
pyrophosphates). Previous studies have correlated factors that influence the 31P chemical
shifts from phosphate-based compounds [22-24]. Among these factors can be mentioned
the number and the electronegativity of the ligands coordinated to phosphorus atom, the
bond angles about phosphorus, the occupation of phosphorus bonding orbitals, the
nuclear charge and radius of the phosphate counter-cation. An important aspect from
previous studies is that the NMR chemical shifts can differentiate between terminal
phosphorus atoms and a chain of phosphate sites. When during analysis complete nuclear
spin relaxation for the 3 IP atom is ensured, quantitative information can be extracted from
the data and the evaluation of chain size for a polyphosphate structure is possible [25-27].
Relaxation time experiments have been performed in order to quantify the terminal versus

chain phosphorus atom content of phosphate salts present on catalyst surfaces. Several

ll

relaxation times can be defined during an MAS-NMR experiment. The spin-lattice (Tl)
relaxation is defined to be the process of growth toward equilibrium magnetization given
by Curie’s law. Placing a sample initially non-magnetized into a magnetic field, the
nuclear moments whose orientations are initially distributed equally between the two
energy states, build up a slight excess of sites aligned parallel to the field according to the
Boltzman equation and to the energy difference 2uH (u = magnetic moment, H static
magnetic field) between parallel and antiparallel states. The rate at which magnetization
builds up in a static field depends on the mechanisms available to transfer energy by
translations, rotations, and vibrations (lattice) to the surroundings. The mechanisms of
spin-lattice relaxation influence the value T. and the type of equation which describes the
process. The spin-lattice relaxation process is described often by an exponential equation
(in an NMR experiment), when the magnetization Mu) recovers from zero (Mo) after a n/2
pulse:
Mo) = M0 [1 -eXP(-t/T1)l (1-1)

Besides spin-lattice relaxation with its characteristic time constant T,, several
other relaxation processes may be encountered such as spin-spin relaxation (T2), spin-
lattice in the rotating frame time (Tlp), and dipolar spin-lattice relaxation time (Tm).
Among these processes, the spin-spin relaxation may play an important role in the total
spin relaxation value during an NMR experiment. T2 is defined as the time constant for
decay of x-y component of the magnetization following a disturbance. When molecular
motions are very fast (non-viscous liquids), T. = T2, resulting in no additional

information about the system. For solids, Tl >> T2, so T2 may offer more information

about the molecular structure under analysis (see detailed information in reference [25]).
For our purposes, the evaluation of T, has proven to be enough to obtain quantitative data
about the phosphate-based catalysts. In general, the evaluation of T, is time consuming.
Among the various strategies to evaluate T,, the most common method of T,
determination is to measure Ma) at various values of t and fit them to an exponential
decay curve. The error for T, evaluation is given by the error to which the value for M0 is
measured (usually not well determined) [25]. Another approach to evaluate T,, specific
to pulse NMR experiments, is the double pulse sequence (inversion and saturation
recovery). In this method two pulses are applied, the first pulse (inversion recovery)
prepares the nuclear spins in some non-equilibrium configuration and then, after a
waiting period during which the spins are allowed to relax, a pulse probes the state of the
spins (d2 from Figure 1.3 a is variable). The inverted Spin population recovery is
monitored as a function of waiting time and goes from -Mo to M0, where MO is the
thermal equilibrium magnetization attainable only after waiting for a time much longer
than T,. A typical NMR signal after a variable waiting time (variable delay time = d2)
during an experiment is presented in Figure 1.3 b. In this experiment, d2 has values
varying from 3.75 S to 900 s. T, can be evaluated by fitting the exponential equation
which follows the nuclear spin relaxation (dotted line from Figure 1.3 b). Besides the
information about how long an NMR experiment should be run in order to ensure
complete nuclear spin relation, the T, value can be used to identify and assign NMR
peaks with very close chemical shifts originating from atoms in different sites, and to

provide additional structural information about a given sample (kind and type of atom

13

close to the analyzed one and type of interaction between them) [26-27]. The MAS
technique produces an isotropic band accompanied by Spinning sidebands of various
intensities Spaced at integer multiples of the spinning frequency on both sides of the main
group of peaks. The symmetry of the spinning bands can provide additional information
about the structure of the phosphorus compound. [28]. Variation in spinning frequency
may help to differentiate between a real peak from the isotropic band and a spinning

sideband [29].

a) Double pulse sequence

 

pl pw
600 s
d1 d2
3.8 s

b) NMR signal at various second delay values (relaxation), d2

Figure 1.3 Spin-lattice relaxation time (T,) evaluation from a double-pulse MAS
- NMR experiment: a) pulse configuration; b) NMR signal at various delay time array, d2.

In this study, solid state 3|P MAS-NMR spectra were obtained at room

temperature, using a Varian VXR-4OOS spectrometer with an Oxford cryomagnet

l4

generating a magnetic field of 9.395 T and operated at a Sun workstation network with the
V-NMR operating system Version 4.1. The Fourier transform of free induction decay
(FID) was observed at 161.903 MHz and at spinning rates of 4000 and 6000 i15 rot/min
(Hz), respectively. A Varian 400VT CP/MAS sample probe of 7 mm diameter and a top
Speed of about 8 kHz was used during analysis. The acquisition time was long enough to
ensure full 3”P nuclear spin relaxation for accurate quantitative data analysis. Additional
relaxation time experiments were performed in order to assign and evaluate the signals

obtained from analysis.

Table 1.1 Parameters for MAS 31P-NMR Data Acquisition

 

Instrument Parameters Numerical values

 

Transmitter frequency [MHz] 161.903
Spectral width [kHz] 100
Number of data points 16,000
Acquisition time [s] 0.08
Number of scans > 64

First delay time, d, [s] 10.0 or 600.0
Delay time array, d2 [S] 3.75 - 960
Spinning frequency [kHz] 4 or 6
Pulse width [as] 4.0

 

Table 1.1 indicates the operating parameters for data acquisition. About 0.5 g
catalyst powder was introduced into the spectrometer holder for analysis. The
spectrometer was tuned at the start of each session using the tuning band for 3 IP in the

sample probe. The chemical shifts were evaluated using calcium hydroxyapatite powder

15

(Ca,O(OH)2(PO4)6 as standard, which shows a single peak at 28:02 ppm. For other
assignments, the shifts of standard phosphate materials analyzed were typically
reproducible within :05 ppm. Bandwidths (FWHM) for the spectra covered a range of
1-18 ppm. The errors in chemical shift values increased for broader bands.

DRIFT S analysis. Diffuse reflectance infrared Fourier transform Spectroscopy
(DRIFTS) is an IR sampling technique useful for powder sample analysis. The spectra
obtained are very Similar with those obtained in normal absorption IR analysis, except
that on the (y) axis, instead of absorption they are reported as Kubelka Munk units (see
below) [30]. In the late 1970’s, Fuller and Griffiths [31-32] developed for the first time
an efficient optical system together with all the necessary instrumentation which allowed
them to perform diffuse reflectance IR spectroscopic measurements on powder materials.
After that, many researchers focused their studies on improvements and new directions to
develop this IR sampling technique [33-35]. Special in-situ DRIFTS cells were
developed for catalysis studies [36-39]. Due to its advantages, currently almost all IR
studies use the DRIF TS method to investigate heterogeneous catalysis systems. The
development of very sensitive detectors, like the mercury-cadmium-telluride (MCT)
detector, permit IR analysis even at a low throughput level (less than 5%). This type of
detector is the most suitable for DRIFTS studies. The method can be successfully used
for quantitative analysis, using the diffuse reflectance beam theory developed by Kubelka
and Munk in the early ‘303 [30]. These units are characteristic for the diffuse reflected

beam from the surface of porous or/and powder samples. The DRIFTS Spectra

l6

characterize the sample qualitatively and quantitatively. Qualitative analyses are
characterized by similar principles as normal absorption spectroscopy.

For powder samples, a DRIFTS spectrum can have a better signal-to-noise (S/N)
ratio and more clear IR bands in comparison with a classical absorption spectrum.
Quantitative DRIFTS analysis shows good linearity between the intensity of the bands
and concentrations for very dilute samples. For concentrated samples, the deviation from
linearity becomes significant making the analysis unsuitable for quantitative calculation
[37-38]. The lower limit for calibration curve linearity is given by samples with 10'5
monolayer surface coverage, whereas the highest limit for linearity is considered to be of
several molecular layers (at several monolayers the specific absorption (8) varies non-
linearly [40]). The accuracy and detection limits of the method are in the range for trace
analysis, close to electron microscopy and spectroscopy.

In 1964, Kortum and Delfs [41] used a dispersive infrared spectrophotometer
operating in a diffuse reflectance mode for surface analysis of solid catalysts. They
studied the adsorption of hydrogen cyanide and ethylene on supported metal oxide
catalysts. The results showed poor quality, the Spectra obtained being hard to interpret
due to low resolution. Niwa et a1. [42] have reported diffuse reflectance IR spectra for
simple molecules adsorbed on porous catalysts at high temperatures using the same type
of Spectrophotometers. With the development of FTIR spectrophotometers, new
experiments focused on obtaining diffuse reflectance infrared spectra for solid catalyst
materials. DRIFTS became the method of choice in heterogeneous catalysis studies,

being continuously improved and developed. One of the main advantages of DRIFTS

l7

analysis is the possibility of obtaining accurate quantitative data at low catalyst surface
coverage. The quantitative analysis is based on the Kubelka-Munk function [30] and it is
applied in a similar way as Beer’s law, which is largely used for quantitative experiments
in the UV-VIS domain. The diagram shown in Figure 1.4, illustrates the theory behind

this type of quantitative IR analysis.

 

 

Incident IR Emitted IR
Flux (I) Flux (J)
x = L
I J + dJ
‘ dx
1 - (II J
K,S
x = 0

 

 

 

Figure 1.4 Schematic representation of IR beam interaction with the sample.

In Figure 1.4, K stands for the absorption coefficient and S for the scattering
coefficient. From the diagram Shown above, two different differential equations can be

written to describe the variation of the incident and emitted fluxes when they penetrate

the surface of a sample:

fl=—(K+S)I+SJ (1-2)
dx
11—:s1 —(K+S)J (1-3)
dx

In this case, the following limiting conditions can be considered:

I = 10, when x = L and
= -RgI when x = 0, where Rg is the reflectance at the sample’s cup surface.
The reflectance is given by the emitted (J) and incident (I) flux ratio (R = M).
The expression for these two IR fluxes can be obtained by solving the differential
equation system (1-2) and (1-3), respectively; substituting the solutions for I and J yields

an expression for reflectance R, given by:

J 1 — Rg [a -— bcoth(bSL)]
R = — = (1-4)
I a — R g + bcoth(bSL)

 

where : a = l + K/S, and b = (a2 - 1)“2 respectively.

As can be observed, the expression for reflectance (R) given in relation (1-4) is
very complex, which makes it useless for practical applications. Kubelka and Munk
proposed several approximations that can be applied to simplify the relation (1-4). These
approximations lead to a function which describes the diffuse reflected beam from an
infinitely thick sample surface, called the Kubelka-Munk function.

As was mentioned before, to simplify the relation (1-4) for reflectance (R) several
approximations were considered:

1) The incident and emitted fluxes are perpendicular to the sample surface.

II) The sample is considered to be very thick, which implies that (R) will
approach zero: coth (bSL) —+ 1, when L —+ 00.

The expression for reflectance, considered as the reflectance for a sample with
infinite thickness (R00), is given by relation (1-5) presented below:

35

1/2 —I
S 1 } (1 5)

K K ,
Rs—{1+(§)-[(§) +

from which can be obtained:
K (1— R )2
— = F R = ————i"——— 1-6
S ( .0) 2R, ( )
The absorption coefficient is given by the relation (2-6):
K = c a (1-7)

where c is the sample concentration and e is the specific molar absorption.
III) The scattering coefficient is considered to be constant through the sample
thickness. In this case, the expression for the Kubelka-Munk function [F(R.,,)] can be

obtained from relation (1-6) and has the following form:
F(R..) = 5‘33 (1-8)

Equation (1-8) represents the relation between the signal collected and recorded in
DRIFTS analysis and sample concentration. The spectrum obtained has on the (y) axis
F(R.,o) instead of absorbance. The unique character of this method is given by the utility
of this function for quantitative measurements. The practical application of this function
depends on the validity of the two approximations described above. Recently, Loyalka
and Riggs [43], have reconsidered the general validity of the Kubelka-Munk function for
practical meanings due to an overestimation of the (K) cOnstant by a factor of two.
However, a redefinition of the expression for (a) from relation (1-4) based on Kubelka
and Munk calculations for scattering cross section (S), indicated a correct value. For
details regarding this new approach see reference [43].

Because the penetration of the incident beam is very small compared with the

sample thickness, and the emitted beam is diffused through the porous surface of the

20

sample, these approximations (I, II, and III) are reasonable for practical purposes.
Experimental data have shown a linear dependence between the DR-signal and sample
concentration even at very low concentrations.

Practically, there is no standard which has a perfect diffuse reflectance, so the

infinite value of the reflectance will be given by the relation (1-9), presented below:

_ R®(sample)
°° _ Rw(standard)

 

(1-9)

The standard considered must present a high diffuse reflectance and should not
absorb the incident IR beam in the frequency domain where the sample has specific
absorption bands. At high concentrations, the presence of polymolecular layers on the
surface of the analyte leads to an increase in the value of molar absorptivity (a) and
consequently to deviation from linearity of the Kubelka-Munk function. On the other
hand, when very thin samples are studied (e.g. films), the depth of penetration for the
incident beam is dependent on the scattering coefficient (S). In this case, the
approximation (L —> 00) introduces a significant error to the Kubelka-Munk function
evaluation [32]. Different IR beam penetration depths at various points of the sample
also can introduce inaccuracies to the measured relative reflectance.

Diflitse Reflectance Accessories. The advantages offered by FTIR spectrometers
were the key conditions necessary to develop and continuously improve the DRIFT
spectroscopy method. The first step in the development of diffuse reflectance as an IR
sampling technique was to design a proper attachment (collector) for the IR beam. The
attachment must be placed into the sample compartment on a support, readily adaptable

to any type of IR spectrometer. A specific optical arrangement for diffuse reflectance

21

analysis of porous materials was first successfully used in the UV-VIS region [44]. A
spherical mirror proved optimal to collect the diffuse reflectance radiation from the solid
surface. Adaptation of this type of mirror to the mid-IR domain introduces some
technical problems. In this case, the IR beam collected and focused on the detector is too
low in energy (several orders of magnitude) for accurate measurements [45]. The lower
sensitivity of detectors used for IR versus UV-VIS spectroscopy exacerbates these
difficulties. Cobletz mirrors [46], semi-ellipsoidal mirrors [47] and ellipsoidal mirrors
[40,48] were used to optimize the IR beam collection from the irradiated sample surface.
Ellipsoidal mirrors proved to be the most efficient and today are the most widely used
ones for DRIFT attachments. An optimized optical setup using ellipsoidal mirrors
appropriate and very efficient for the collection of diffuse reflected IR beam was
described for the first time by Fuller and Griffiths [31]. Another key factor to be
considered is the choice of reflective coating material for the inside of the ellipsoidal
mirrors. Materials like MgO or BaSO4 have been employed in mirrors for UV-VIS
applications; for DRIF TS collectors, MgO or aluminum were successfully used because
of their very weak absorption in the IR region (close to zero).

Diffuse reflectance collectors. Two different types of collectors, on-axis and off-
axis have been developed for DRIFTS analysis. Figure 1.5 a shows the on-axis collector
introduced on the market by Spectra-Tech Inc. [49]. The main components of the
collectors are labeled in the figure. The IR beam hits the input ellipsoidal mirror and is
dispersed on the sample, placed into a special cup. The blocker placed in the middle of
the sample holder helps to eliminate specular distortion produced by radiation reflected

without penetration into the sample, which therefore carries no spectral information. The

22

incident radiation passes into the bulk of the sample and undergoes reflection, refraction
and absorption before reemerging at the surface. The diffusely reflected IR beam is
collected by the output ellipsoidal mirror and focused onto the detector of the
Spectrometer. One of the main advantages offered by this type of collector is the high
energy output resulting in a spectrum with an excellent signal-to-noise ratio. The
collector allows a convenient sample placement from either the top or the front of the
accessory and permits operation in unfavorable sampling environments. A build-in
micrometer screw adjustment allows rapid and accurate sample positioning and
alignment. The collector can be used in three different modes of operation. For neat
samples, a pure diffuse reflectance mode allows high quality, transmission-like spectra to
be acquired. A combined diffuse/specular reflectance mode provides the highest
performance for samples dispersed in a non-absorbing matrix. Finally, the collector can
be used to acquire surface reflectance spectra in a specular mode. The disadvantage of
this type of collector is given by the high specular reflectance which can interfere and
overlap with the signal containing the Spectral information from the sample.

The 90° off-axis DRIFTS collector [50], also called the Praying MantisR
attachment is shown in Figure 1.5 b. This is the first type of collector introduced on the
market by Harrick Scientific Co. in the 1980’s and it remains widely used in the field
even today. The main components of the collector are similar with those presented in the
case of the on-axis collector, except that their arrangement is different. As can be
observed from the figure, the IR beam is reflected off-axis before it hits the input

ellipsoidal mirror. The output mirror collects and focuses the beam with the spectral

23

information toward the detector (off axis to the sample). Both ellipsoidal mirrors are
tilted forward so that the diffusely reflected radiation is collected at an azimuthal angle of
120°. This deflects the specular reflectance component behind the collection ellipsoid,

minimizing the intensity of restrain bands and of the specularly reflected light [50].

8)

OUTPUT ELLIPSOID iNPUY ELLIPSOID

 

 

 

 

b)

 

 

 

 

’r
M
1’ HF
Q. "
O O
\r\ /

 

 

 

 

Figure 1.5 Optical attachment (collector): a) on axis (reprinted with permission of
Spectra-Tech Inc. [49]; b) off axis (reprinted with permission of Harrick Scientific Co.
[50]), for diffuse reflectance spectra collection.

24

The optical geometry of the Praying Mantis attachment permits collection of up to
20% of all diffusely reflected radiation and 75% of the Open beam at the throughput with
the tilted alignment mirror. Due to the lower throughput obtained by using this type of
collector, a mercury-cadmium-telluride (MTC) detector is strongly recommended. This
type of detector has a higher sensitivity than other type of detectors such as lithium-
tantalite (LiTan) or the dimethyl-triglycine sulfate (DTGS). A unique advantage of this
type of collector is the ability to rotate the ellipsoidal mirrors (ellipsoids) in the optical
plane of the attachment. Thus, with only a minor modification, the ellipsoid can be
repositioned from their conventional downward looking mode, expanding the scope of
samples that can be analyzed (like large panels or thin layer chromatographic plates). In
addition, the collector can be used in a specular reflectance mode at 415° incidence
angle. A special sample holder is available for this purpose. The attachment can be
operated with a precision translation stage for the sample, a viewing microscope, a
sample illuminator and micro-sampling cups. The sample stage can be adjusted in all
three dimensions, which allows the analysis to be made in either diffuse reflectance mode
or specular reflectance mode, as well as a combination of both. The design of the
attachment permits a reaction chamber to be attached, for in-situ measurements.

Difluse reflectance in-situ cell. In 1984, Hamadeh et al. [36] developed an in-situ
cell for DRIFTS studies of heterogeneous catalysis processes. The cell allowed good
control of temperature and reactant flow rates over the catalyst. Further design
improvements resulted in a reliable, adaptable unit with a well controlled environment
above the catalyst surface. Figure 1.6 shows a commercially available version of an in-

situ DRIFTS cell (Harrick Scientific Co.) [50-51]. The free volume above the sample is

25

not greater than 1.75 cm3. The hemispherical windows are easily changeable, and may be
made of various infrared transparent materials depending on the reaction conditions and
spectroscopic range required. For atmospheric pressure, a KBr window can be used with
good results. For high vacuum or pressure, more chemical inert windows such as
germanium, ZnS or ZnSe windows, should be utilized. For example, with a ZnS

window, the reaction chamber can operate at pressure as high as 1.01x106 N/m2 (10 atm).

Retaining Ring

g Te flon Washer
. V”

  

‘ Sample Cup

<5? Retaining Plates (2)
‘/ ‘%
/‘ . . , \ /

 

 

 

 

Thermocouple .Q “‘\ ’L‘ a ' /:,/

I, (. Gas/Vacuum Port (2)
\‘\ e
\\

Coolant Ports (2)

Heater
Sample Cup Gas/Vacuum Port

Figure 1.6 In-situ DRIFTS cell (Harrick Scientific C o. [50]).

26

The windows of the high pressure cell are very thick, limiting the throughput to 2-5%,
leading to a relative low S/N ratio and poor resolution, but these difficulties can be
partially reduced by increasing the number of scans and lowering the resolution (8 cm!)
The cell is connected to a vacuum line which allows degassing of the sample or study of
reactions under subatmospheric pressures. Temperature is controlled by a thermocouple
and the heating of the cell is carried out by an electrical mantel. Reactant and product
flow rates are controlled and recorded before and after the gas inlet (sample cup
gas/vacuum port) and gas outlet (gas/vacuum port 2). The cell has a cooling system to
decrease the noise that might influence the quality of the IR spectra. Leyden et al. [52]
have modified the collector configuration by introducing a one-dimensional translation
stage that allows precise location of the sample at the optimum height. Thermal emission
or expansion during heating affects the intensity of the dc-signal from the sample,
decreasing the sensitivity of the MCT detector. Lowering or raising the sample height
influences the IR band intensity from a DRIFTS spectrum. For absolute quantitative
measurements using an in-situ DRIFTS cell, optimization of the sample height becomes
an important factor. Without this optimization, only relative quantitative measurements
are possible. After adjustments (sample height and position) the reproducible variation in
the IR band intensities is 3: 5-10%. Advantages like quantitative and in-situ
measurements offered by the instrumentation unique to DRIFTS analysis make this
sampling technique attractive for solid catalyst characterization and other studies in the

heterogeneous catalysis field.

27

Advantages and Disadvantages of DRIFT S Compared with other Conventional IR
Techniques. Based on information about the theory and the instrumentation used in
DRIFTS analysis provided above, it is now possible to evaluate the advantages and
disadvantages offered by this sampling technique over other IR techniques currently used
for heterogeneous catalysis studies. The discussion will be focused on conventional
transmission/absorption, and emission techniques versus diffuse reflectance. Attenuated
total reflectance (ATR) and photoacoustic IR (PAS-IR) spectroscopy are two other major
IR sampling techniques, employed in heterogeneous catalysis applications [37, 53], but
they are less suitable and/or compatible, as explained below.

Table 1.2 presents comparison data for various IR spectroscopic techniques used
for catalyst surface analysis. Transmission IR spectroscopy is a proven and simple
technique which uses samples in the form of a thin wafer. The big disadvantage is that
the sample must be semi-transparent, sample preparation is laborious, and partial sample
oxidation occurs readily [54]. Emission IR spectroscopy has proven to be an effective
technique for powder and metal samples. The method is based on the fact that any warm
body emits IR radiation proportional with its structure and temperature. Consequently,
no IR source is necessary; the sample itself plays the role of source.

Spectra obtained by emission IR include bands with good spatial resolution, but
the method has a low S/N ratio and in many cases background radiation becomes an
important problem. Sullivan et al. [53] in their review, have indicated the main

advantages and disadvantages of various IR techniques and their possible applications.

28

Table 1.2 Comparison of IR techniques for used for catalysts characterization

 

 

 

 

 

 

 

 

 

 

 

Reflection Diffuse
Transmission Emission Absorption Reflectance
’“ " g M w
Samples Thin Wafers Powders Foils Powders
Metals Single Cristal:
Simple Sample Versatility Buy Sampling
Advantages Proven Spatial Resolution Metals Improved SIN
Surface Sensitive
Semitransparent Low SIN Reproducibility
Disadvantages Materials Background Radiation Smooth Samples Dilution
Pelleting

 

Analyzing the spectra of a zeolite sample acquired using the two different methods, it was
observed that the DRIFTS spectrum shows good resolution, with well defined IR bands
in 1300-500 cm'l region as compared to the spectra obtained by emission IR. In the case
of emission IR spectroscopy, attenuation of some bands, and intense background
radiation proportional with temperature, can result in poor resolution of some bands.
Reproducibility of emission spectra constitutes another major problem in IR band
interpretation. Hence, the analysis of DRIFTS spectra is easier and more accurate than
that based on emission data.

Reflection IR Spectroscopy is used mostly for smooth metal catalyst surface
characterization, when the reflection is close to 100% and the signal is strong enough to
provide useful molecular level information about any species adsorbed on a surface [53].
The method fails to provide good quality IR spectra for samples with rough surfaces and

low reflection capabilities (not suitable for powders).

29

Based on the observations presented above and on the summary presented in
Table 1.2, the following advantages of DRIFTS analysis can be mentioned: i) easy and
non-destructive sample preparation (no pelleting necessary); ii) excellent analysis of
opaque powders; iii) avoidance of contact with other materials (like KBr, H20) which
might interfere with the sample bands or may themselves be catalysts or poisons in the
reaction considered; iv) high sensitivity and spectra with well resolved bands compared to
those obtained by using other techniques.

In the following work, DRIFTS spectra were acquired using a Perkin-Elmer
Spectrum 2000 instrument equipped with the Harrick Inc. diffuse reflectance attachment
described above. The catalyst powders (20-25 mg) were put into the sample holder and
introduced into the DRIFTS attachment. To simulate the lactic acid conversion process
the MOH/Si02 catalysts were exposed to lactic acid vapors at temperatures ranging from
25°C to 350°C, using an apparatus designed to direct vapors onto the catalyst surface
[11]. The samples were transferred to the IR spectrometer and analyzed via DRIFTS to
identify surface species present in the quenched catalyst following reaction, in
temperature range between 120°C to 350°C. The spectra are collected and reported in
Kubelka-Munk units versus wavenumber. IR Spectra were acquired with resolution of
4.0 cm'1 over a 400-4000 cm'1 wavenumber range.

X-Ray Diflraction (XRD). X-rays, with their wavelength in the Angstrom range,
can penetrate solids and are suitable to probe and provide information about the bulk,
crystalline structure. Changes in crystalline phases lead to variations in the X-ray

diffraction patterns recorded. Each diffraction pattern can provide information about the

30

particle size of the solid and the relative proportions of various isomorphous phases. X-
rays are produced by bombarding a metal (very often Cu) with high energy electrons.
The Cu KOL line (8.04 keV and A = 0.154 nm) is generated by primary electron ejection
from a K shell to leave a core hole, followed by filling of the hole with an electron from
the L shell with the emission of an X-ray quantum. This phenomenon is the basis of X-
ray sources used in XRD or in other X-ray spectroscopy techniques [X-ray photoelectron
(XPS), or fluorescence (XFS) Spectroscopies, electron diffraction microscopy, etc.]. X-
ray diffraction represents the elastic scattering of X-ray photons by the well organized
atoms placed in a crystalline lattice. The scattered monochromatic X-rays yield
interference patterns as they scatter off the evenly spaced crystal planes of the lattice.
Lattice spacing is given by the Bragg equation (1-10).
nA=2d sin0, n= 1,2, 3,.... (1-10)

where A is the X-ray wavelength, d is the distance between two lattice planes, 0 is the
angle of diffraction (X-ray versus normal to the reflecting lattice plane), and the integer n

is the reflection order [55] (see Figure 1-7).

 

Figure 1.7 Schematic representation of the parameters from the Bragg equation

31

Measuring the angles, 0, under which the X-ray constructive interference occurs when it
leaves the crystal, the Bragg equation (1-10) provides the corresponding lattice Spacing,

characteristic to a given crystalline solid material [56]. From the XRD diffraction

pattern, the mean crystallite size (3) of a given particle can be determined from line
broadening calculations [57], using the Scherrer equation (1-11):

Pr = kA/BcosG (141)
where 7,. is the X-ray wavelength, k is the particle Shape factor (1 for spherical particles,
or 0.9 for cubic particles), [3 is the full width at half maximum (F WHM) of the peak at a
given 26, in radians, and 0 is the diffraction angle. The XRD technique has some
limitations among which the most important are: i) a clear diffraction pattern can be
observed only when the solid material exhibits long range order; ii) crystalline particles
less than 2-3 nm are not observable; iii) the method is not suitable for characterization of
amorphous materials. This technique can provide clear and unequivocal information
about sufficiently large crystalline materials that are used as catalysts, revealing structural
modifications that occur during their preparation or after their use in a catalytic process.

In this study, X-ray powder diffraction patterns were obtained with a Rigaku XRD
diffractometer employing monochromatic Cu KO. radiation (A = 1.541838 A) and
operated at 45‘ kV and 100 mA. Diffraction patterns were obtained using a scan rate of 1
deg/min with 1/2 mm slits. Powdered samples were mounted on glass Slides by pressing

the powder into an indentation on one side of the slide.

32

Activity measurements. All reactions were performed in a vertical down-flow
packed bed reactor equipped with a quartz insert [1,10]. Figure 1.8 presents the
laboratory installation used for catalyst testing and lactic acid conversion studies. The
reactor is designed for pressures up to 5 MPa at temperature < 500°C. The catalyst is
supported on a quartz frit fused into a quartz tube, inserted into the reactor from the
bottom and sealed. An internal thermocouple extends from the reactor flange to the
bottom of the support frit. The reactor is electrically heated by a programmable
temperature controller. During operation, catalyst temperature is measured by the
internal thermocouple, and the reactor set point is adjusted to the desired value. The
products of the reaction exit the bottom of the reactor and pass first through a trap placed
in an ice bath. The outlet of the reactor was connected to a Riken infrared CO and C02
gas analyzer which measured the amounts of these gases produced in the reaction.
Typically, products are collected for 30 min during steady-state operation of the reactor.
The volume of the liquid product and the flow rate of the gas products are measured
during operation in order to compute a mass balance.

Product analyses were performed using a Varian 3700 gas chromatograph
equipped with FID detector and a Supelco packed column (4% Carbowax/Carbopack B-
DA packed column). The product was mixed with a solution containing 2-propanol as an
internal standard and oxalic acid as a column conditioner. Good reproducibility of the
lactic acid analysis was achieved by direct injection of 1 pl of condensed product solution
mixed with the internal standard. Major products analyzed include 2,3-pentanedione,

acrylic acid, propanoic acid, acetaldehyde, and acetol. Minor components identified from

33

product analysis include ethanol, acetone, acetic acid, pyruvic acid and other unknowns

reported as “others”.

 

 

 

 

 

 

 

 

 

 

F iltcr
lactic Acid
Feed
N2
JLJW Heater

Air

Quartz Tube
Exhaust Temperature controler
To gas analyzer
Flowmeter

 

 

Filter

 

Figure 1.8 Schematic representation of the laboratory setup used for lactic acid
conversion studies and catalyst testing.
All product yields are calculated from the ratios of product/intemal standard peaks areas
and detector response factors. Product component identification was performed by
residence time analysis in the GC, by gas chromatography/mass Spectrometry (GC/MS)
and 1H and 13C NMR spectroscopy. Product yield was reported as percentage of the

theoretical yield based on the lactic acid fed to the reactor; product selectivity is the

34

percentage of the theoretical yield based on lactic acid consumed in the reaction. The
overall carbon balance gives recoveries ranging from 85% to 105%. Each catalyst loaded
into the reactor was tested at several temperatures and residence times. The catalyst was
heated at the desired temperature under a constant lactic acid flow-rate (0.5 ml/min) and
allowed to reach steady state. Table 1.3 presents the experimental reaction conditions

used for the catalyst testing.

Table 1.3 Experimental conditions used in lactic acid conversion on MOH/SiOz
catalysts

 

 

 

 

 

 

 

 

Ord.# Reaction parameters Experimental value

1 Temperature [C] 200 - 350

2 Pressure [MPa] 0.5

3 Liquid flow rate [ml/min] 0.05 - 0.5

4 Feed composition Lactic acid: 0.08
[mole fraction] Water vapor: 0.77

Helium: 0.15

5 Catalyst weight [g] 2.0 - 6.0

6 Catalyst bed height [cm] 2.5 -7.6

7 residence time [sec] 0.3 - 5.0

 

 

 

 

 

1.3 Results and discussions

Sodium phosphates supported on Si/Al catalysts. Activity data obtained for lactic
acid conversion on NaH2P04 and Na3PO4 supported on Si/Al catalysts, in comparison
with the neat Si/Al support are presented in Table 1.4. The modifications occurring on

the sodium phosphate catalyst as a fimction of temperature were studied using MAS-

35

NMR and DRIFT Spectroscopies, and they were correlated with the differences observed
in the catalytic activity. Differences in acidity for the two types of phosphate catalysts
which have been chosen in this study, may influence the catalytic behavior. The two
protons from NaH2P04 confer on the catalyst a stronger acidity compared with the
Na3PO4. Temperature plays an important role in the catalyst structure and catalytic

activity.

Table 1.4 Yields (selectivities) from lactic acid conversion over sodium
phosphate salts supported on Si/Al catalyst, at 300° and 350°C, respectively [10-11]

 

 

 

Product Si/Al Support [%] NaHzPO4 [%] Na3PO4 [%]
300°C 350°C 300°C 350°C 300°C 350°C
L.A. conversion 5.7 17.4 2.5 17.6 19.9 39.9
23P 0.2 (3) 0.3(2) 0.2(11) 1.3(15) 4.3(31) 7.0(22)

 

Acetaldehyde 1 .5(22) 9.0(48) 0.4(22) 2.5(28) 1 .9(14) 5.2(16)

Acrylic acid 0.1(1) 0.8(4) 0.1(6) 1.5(17) 2.0(14) 9.8(31)
Propanoic acid 0.5(7) 1.2(6) 0.1(6) 1.7(20) 0.9(6) 1.5(5)
Carbon recovery 100.8 101.2 104.3 91.1 94.2 92.7

 

 

 

 

 

 

 

 

 

L.A. = lactic acid; 23P = 2,3 -pentanedione

The yield and selectivity data shown in Table 1.4 for lactic acid conversion on
sodium phosphates supported on Si/Al catalyst indicate the best activity results for
Na3PO4/Si/Al catalyst (yields of 20% with 30% selectivity, at 300°C). Elevated
temperatures favor lactic acid conversion, but selectivity toward 23P formation decreases.
Si/Al support shows good selectivity for acetaldehyde formation. This catalytic behavior

can be attributed to the presence of surface acid sites (Bronsted and Lewis acid sites) on

36

Si/Al support structure. The same trend is observed in the case of the Si/Al supported
NaH2P04 catalyst, indicating that catalyst acidity favors the acetaldehyde pathway.
Na3PO4 supported on Si/Al has no acid character. As a consequence, the selectivity for
acetaldehyde is much lower than the selectivity for 23P. From these data, it can be
concluded that the acid/base character of the catalyst plays an important role in the
reaction pathway of lactic acid conversion.

Changes in the temperature may affect the catalyst structure. It is known that the
sodium phosphate structure is very sensitive to temperature changes [22-24]. MAS-NMR
combined with DRIFT spectroscopy have been used to study the types of phosphate
species on the catalyst surface. Figure 1.9 shows the MAS-NMR spectra of pure
NaH2P04 salt after calcination at 450°C (19 a) and after aging in air for several days (1.9
b). As can be observed from Figure 1.9, simple water absorption can significantly
modify the NMR spectrum of the NaH2P04 salt calcined at 450°C. These spectra are
different from that obtained for the NaHzPO4 before calcination (not shown), which
presents a single isotropic peak at 8 = 1.8 ppm, similar with that reported in the literature
[11, 22-24]. After calcination at 450°C, five unresolved isotropic peaks, marked with
arrows, at -5.4, -15.3, -19.0, -24.0, and -26.2 ppm are observable (Figure 1.9). 31P{'H}
proton decoupling has no effect on how well the peaks are resolved in the spectrum [11].
Peaks -15.3, -19.0, -24.0, and -26.2 ppm were assigned in the literature to linear sodium

polyphosphate (NaPO3)n species, in crystalline form [23].

 

37

(a)

(b)

 

Y's!I'lrfitt‘TItIrrvrs'vlr
100 50 0 -50 -100

Chemical shifts [ppm]

Figure 1.9 MAS-NMR spectra for the NaH2P04 salt, calcined at 450°C (a) and
aged (b).
The peak at -5.4 ppm observed can be attributed to an end-chain P04 group from a
polyphosphate. This peak may also be attributed to the presence of a small quantity of
the trimeric ring (Na3P309). The decrease in the NaH2P04 peak at 1.8 ppm as
temperature increases, indicates structural modification of the sodium phosphate salt.
These observations suggest that sodium polyphosphate was formed after calcination at
450°C. When supported on Si/Al catalyst and calcined at 450°C, the NMR spectrum of
NaHzPO4 catalyst obtained (Figure 1.10) is different from that observed for the neat salt
(Figure 1.9) with chemical shifts for the isotropic peaks around 11.8 ppm. These
chemical shifts indicate that after calcination at 450°C, NaH2P04 is transformed into a
different phosphate species, probably polyphosphates (chemical shift analysis [11]). The

31P NMR spectrum for the NaHzPO4 supported on Si/Al (Figure 1.10) is less well-

 

38

resolved than that of the pure material. Following drying and calcination, peaks specific

to Na3P309 or other (NaPO3)n polymeric amorphous materials are observable [11].

-11..51

 

'32.$.‘
-3‘l.1l6

---—--13.259

 

 

 

TIIIVTITITYrIIIITIjIIII‘III'TUUTIII
150 100 50 0 -50 -100 -150

Chemical shift [ppm]

Figure 1.10 3|P-NMR spectrum of Si/Al supported NaH2P04 calcined at 450°C

The peak at 1.8 ppm specific to NaH2P04 is not easily observed in Figure 1.10, indicating
that all the NaI-IZPO4 deposited on Si/Al support was converted into other phosphate
species (as expected based on data from Figure 1.9). Heating at 450°C, a broad isotropic
peak centered at -11.85 ppm is observed. Proton decoupling NMR experiments
suggested that more than one peak is present in this chemical shift region [11]. Peaks -
15.3, -l9.0, -24.0, and -26.2 ppm were assigned in the literature to linear sodium
polyphosphate (NaPO3)n species in crystalline form [23]. Any shift in these values may

indicate the formation of amorphous sodium polyphosphate species. Peaks with chemical

39

shifts close to these values can be observed in the spectrum presented in Figure 1.10, as
well. A peak at -5.4 ppm, observed as a shoulder in Figure 1.10 can be attributed to end-
chain P04 groups, based on literature data [11]. From these observations, it can be
concluded that sodium polyphosphate was formed after calcination at 450°C, on Si/Al
support. The extent of phosphate polymerization can be calculated fi'om the intensity
ratio of the end-chain to intra-chain P04 NMR peaks.

In order to find the degree of phosphate polymerization and to identify other
possible sodium phosphate species formed, T, values were measured and full relaxation
of the 3|P nucleus was ensured during analysis. Spin-lattice relaxation time (T,) values
were evaluated for each peak from the NMR spectrum presented in Figure 1.10, based on
the exponential dependence of magnetization versus T, (see relation 1-1) shown in Figure
1.11, are reported in Table 1.5. Due to the fact that the isotropic peaks are not the most
intense peaks in the spectrum, all the peaks in the 0-100 ppm were considered in the
relaxation time experiments. Peaks indicating similar T, values correspond to the same
species, so that T, can be used to identify various phosphate species formed on Si/Al
supports. T, values presented in Table 1.5 indicate that at least three different phosphate
atoms are present on the catalyst surface. The error in the value for T, was less that 10%.
The presence of less intense peaks which overlap with some of these peaks can’t be
excluded. The peak at -32.59 ppm shows the longest T, value. A longer T, value
corresponds to a longer relaxation due to a stronger coupling between the nuclear Spin

and the neighbors Spins from the lattice [26]. As a consequence, this peak can be

40

attributed to a different phosphate species, maybe a cross-linked polyphosphate species

formed during thermal treatment of the catalyst.

 

 

 

 

 

 

 

 

 

 

fl [1

J- .L

Q

4’ D

5' V

81 +

n U

03 0

u A

is V
ITT' I[ITII'ITVFIIIIIrWYUIITUUIIUIIrIrIUIIUIIU1T1
0 100 200 300 400 500 600 100 000 900 1000

tinslsac)

Figure 1.11 3'P-NMR intensity versus time [S], for the peaks specific to the Si/Al
supported NaHzPO4 catalyst, calcined at 450°C, where #1 = 13.26 ppm, #2 = -11.85 ppm,
#3 = -32.59 ppm, #4 = -37.19 ppm, #5 = -66.52 ppm.

41

Table 1.5 Spin-lattice relaxation time (T,) values for Si/Al supported NaH2P04
catalyst, calcined at 450°C, evaluated from the exponential dependences in Figure 1.11.

 

 

Peak Peak’s chemical shift Relaxation time Error
# [ppm] [seconds] [seconds]
1 -13.26 33.5 2.4
2 -1 1.85 30.1 1.6
3 -32.59 144 18.3
4 -37. l 9 47.1 6.5
5 -66.52 45.3 5.7

 

 

 

 

 

 

Summarizing all the information from the NMR spectra (Figures 1. 9-11) and
from the T, calculations, it can be concluded that amorphous, short chain (3 <n<6) sodium
polyphosphate species are formed on Si/Al support after calcination at 450°C [11]. This
observation is reasonable considering the fact that on the support, NaH2P04 is likely to be
in a more dispersed phase that in the neat salt, so smaller condensation products such as
trimetaphosphate (trimer) are favored over polyphosphate (liner polymer). Phosphate
anions may also interact with the surface acidic sites of the Si/Al catalyst, inhibiting
formation of long polyphosphate chains [10].

Considering the low structural stability of the NaH2P04 catalyst (formation of
sodium pyrophosphate or triphosphate) and the relatively high acidity of the sodium salt
formed at elevated temperatures (competitive with lactic acid’s acidity) the observed high
selectivity to acetaldehyde can be explained (see Table 1.4). The small modifications

observed in the structure of the Na3PO4 salt and Na3PO4 supported on Si/Al catalysts with

42

temperature are well reflected by the NMR spectra obtained for these compounds and

presented in Figure 1.12 a-d.

(a)

 

 

(C)

(d)

 

(e)

(

80 40 0 .40 -80
Chemical shift [ppm]

Figure 1.12 3 lP-NMR of Na3PO4 salt and Na3PO4 supported on Si/Al catalysts: a)
Na3PO4x12H20; b) Na3PO4 calcined at 450°C; c) Na3PO4/Si/Al catalyst dried at 100°C;
(1) Na3PO4/Si/Al catalyst dried at 450°C; e) used Na3PO4/Si/Al catalyst in lactic acid
conversion; f) regenerated Na3PO4/Si/Al catalyst by calcination at 450°C.

43

However, significant changes in catalyst structures were observed after its utilization in
lactic acid conversion or regeneration by recalcination (Figure 1.12 e-f).

The NMR spectrum of Na3PO4x12HzO presented in Figure 1.12 a shows one
isotropic peak at -7.5 ppm, in good agreement with the literature [22-24]. Calcination at
450°C led to dehydration, resulting in a narrowing of the isotropic peak and a shift to -
14.2 ppm (Figure 1.12 b). Supporting the Na3PO4 salt on a Si/Al catalyst and drying at
100°C led to a phosphate catalyst with a spectrum having peaks specific to hydrated and
dehydrated Na3PO4 species (Figure 1.12 c). The spectrum of the Na3PO4/Si/Al catalyst
dried at 450°C shown in Figure 1.12 d looks similar with that of the dehydrated Na3PO4
salt spectrum (Figure 1.12 b). AS can be observed, Simple thermal treatment of the
Na3PO4 catalyst led to very minor structural changes compared to the NaHzPO4 catalyst.
Utilization of the Na3PO4 catalyst in the lactic acid conversion led to significant
modification in the catalyst structure as can be observed in Figure 1.12 e. Regeneration
of the catalyst, by recalcination at 450°C occurs only partially, as observed from the 3lP
NMR spectrum (Figure 1.12 f). To improve peak assignments, samples were analyzed at
two different spinning frequencies. Variation in the spinning rate alters the positions of
the spinning side-peaks band, leaving the chemical shift of the isotropic band unchanged.
Thus, variation of the spinning rate allows separation and identification of the isotropic
bands in the spectrum. Figure 1.13 shows the 3'P NMR spectra of the used and

regenerated Na3PO4 supported on Si/Al catalyst, at 4 kHz and 6 kHz, respectively.

44

(a)

6.2
\
N32HPO¢ 01' (N3P03)n

 

 

 

 

 

(d)
r r . , . . f. . . . ,s .
100 50 0 -50 -100

Chemical shift [ppm]

Figure 1.13 3|P-NMR of Na3PO4 supported on Si/Al catalysts: a) regenerated at 6
kHz; b) regenerated at 4 kHz; c) used at 6 kHz; (1) used at 4 kHz

The 31P NMR spectrum of the used Si/Al supported Na3PO4 catalyst presented in
Figure 1.13 c obtained at 6 kHz shows a better separation of the isotropic band from the
spinning bands, consequently it was used to identify the phosphate species formed during
its utilization in the lactic acid conversion. The spectrum shows five isotropic peaks at

3.2, 0.7, -0.6, -6.8, and -10.0 ppm. The primary species present are the anhydrous

45

Na4P20-, (3.2 ppm) and Na4PzO7x10H20 (-0.6 ppm), as indicated on the figure. The
peaks for these species are shifted slightly from the standard value but are within the
ranges reported in the literature [11, 24]. The peaks at 0.7 ppm and -6.8 ppm are assigned
to anhydrous sodium triphosphate (Na5P3O,0) and the peak at -10 ppm to a hydrated
triphosphate (Na5P3O,,,xHZO). The linear triphosphate is believed to form by
condensation of the Na3HPzO-, and NaZHPO4 (proton exchange products). The loss in
catalytic activity can be attributed to the formation of this type of polyphosphate species.
It is known that in water, the extent of sodium dissociation from phosphate anions, while
over 90% for orthophosphates, decreases with increasing chain length for condensed
phosphates [58].

As can be observed from Figure 1.13 a,b, calcination at 450°C of the used catalyst
only partially regenerates the active Na3PO4 species (reappearance of the peak at 14.3
ppm). The extra peaks from the NMR Spectrum of the regenerated catalysts, at 6.2, 0.7, -
6.8, -10 ppm, can be attributed to various types of polyphosphate species (NaPO3)n or of
NaZHPO4 which maybe formed at elevated temperature from Na4P207 and Na5P3O,0 in
the used catalyst (see Figure 1.13).

Additional information about chemical transformations that occurred on the Si/Al
supported Na3PO4 catalyst during lactic acid conversion were obtained from DRIFTS

analysis. The DRIF TS spectra for the Na3PO4 catalysts are presented in Figure 1.14.

 

46

1573 C=Ofrom

 

 

Na-Lactate
1418
1130
1294
’ 513
i l(
(a) q n
1458 1020 b
e
, q ‘ l
k
a
Dd
u
' n
k
, U
1437 'i'
85:5 m- . t
\ s
‘554 1128
1664 "’
(C)
| l l |
1800 16001) 12001) 8001) 40011

Wavenumber [cm-1]
Figure 1.14. DRIFTS spectra for Si/Al supported Na3PO4catalysts: a) following
lactic acid exposure at 320°C and b) fresh, compared with c) pure Na3PO4 salt.
The presence of IR bands specific to tetrasodium pyrophosphate (1120 cm'l) on
the supported catalyst exposed to lactic acid vapors at 320°C (see Figure 1.14 a) is in
good agreement with the results from 31P NMR. The DRIFTS spectra of Na3PO4 on

Si/Al (Figure 1.14 b) and neat Na3PO4 salt (Figure 1.14 c) confirm the lack of change in

47

the catalyst structure observed by NMR. Proton transfer from lactic acid to Na3PO4 (and
to a lower extent at NaHzPO4, Na4PzO7 or other polyphosphate species) was implied by
the observation of sodium lactate as a stable species on the catalyst surface, during
reaction [11]. The large quantity of sodium lactate on the Na3PO4 catalyst surface may
explain the higher activity relative to NaH2P04, at lower temperature (see table 1.4).
Conversion at high temperatures is affected by something other than quantity of the
sodium lactate present on the catalyst surface. This information may be useful to
understand and optimize the catalyst’s activity and to elucidate the mechanism of lactic
acid conversion to 23P (see details in Chapter 2).

I Ba/Na/SiOz catalysts; influence of Si02 support configuration on lactic acid
conversion. The interest in this study came from the fact that a simple high surface area
silica support (hSiOz) has a much lower price in comparison with a silica with low
surface area but well controlled pore size distribution (lSiOz) used in previous studies
[64-66]). Barium is known to be a useful additive to high surface area catalysts (A1203,
TiOz) conferring good thermal stability on the porous material [59]. Inexpensive and
non-toxic alkali and alkaline-earth metal hydroxides and nitrates [NaOH, Ba(NO3)2]
supported on SiOz with various surface areas, have also been successfully used to
produce 23P together with other valuable chemicals (acrylic acid or propanoic acid) with
high conversion and selectivity. Supported Ba/Na and NaOH on lSiOz, and hSi02
catalysts, prepared as described previously, were characterized by DRIFTS spectroscopy
and X-ray diffraction (XRD) and used in lactic acid conversion.

In order to understand the catalytic activity for a given solid material it is essential

to obtain information about its surface configuration and structure. Following the

48

preparation steps by IR Spectroscopy reveals the molecular level structural changes of the
catalysts during the process. Figures 1.15 and 1.16 show series of DRIFTS spectra for
hSi02 and lSi02 catalysts, respectively. Figures 1.15 a and Figure 1.16 a show Spectra of
the supports. Specific to hSiOz are the band at 3738 cm", attributed to the O-H stretching
mode of free hydroxyl groups, and at 1630 cm'l and 1291-969 cm", attributed to bending
mode of bond vibrations. The spectrum for lSi02 is very similar with that for hSi02 but
the -OH groups appear more as broad bands specific to hydrogen bonded hydroxyl
groups (see bands in the 3000-3600 cm'1 region).

Before impregnation, the Silica supports were treated with a sodium-methoxide/
methanol (CH3ONa/CH3OH) solution to annihilate any acid sites (BrOnsted acid sites)
from the surface. After drying, DRIFTS spectra were acquired. Figures 1.15 b and
Figure 1.16 b Show spectra of the silica supports impregnated with CH3ONa solution.
From these figures, it can be observed how the bands at 3738 cm'l and 3000-3600 cm",
corresponding to the surface hydroxyl groups disappeared, and a new band at 778 cm’1
(860 cm") appears. Morrow and McFarlan [60] have indicated an upward shift of bands
from the 900-750 cm'1 IR region2 where bands specific to Si-O- stretch from a Si-O-H
group are observable when hydrogen was replaced by deuterium. Consequently, the band
observed at 778 cm'1 (860 cm") from Figure 1.15 (or Figure 1.16) can be attributed to a -

Si-O- vibration when -Si-O-Na is formed [61].

 

2 firmly established in the literature as Si-O deformation and stretching modes of bond vibration [59-62].

49

1332 778
1456
1630 , = i
,. , 1 , ‘
7 . -, 1
3738 ‘8 ,0 , K
b
j e
(d) 1 ~ I
[l
, I k
a a
i M
(c) 1 1 u
i n
k
‘ 1,
' , u
i . w, , ‘ f n
1 , ‘ l , . i
(b) L 1 ‘ l 7
1 a ‘ t
s
\\\\\ X
, 3600-3000
(a) 1 ,
1291-968
- l - l > I I
4000.0 3000.0 2000.0 1000.0

Wavenumber [cm'l ]

Figure 1.15 DRIFTS spectra for Ba/Na/hSiOz catalyst acquired at different times
during preparation: (a) pure hSiOz catalyst ; (b) hSi02 treated with CH3-0Na ; (c)
Ba(NO3)2/NaOH on h8102 catalyst dried at 125°C ; (d) Ba/Na/hSiOz catalyst calcined at
400°C.

50

1448 1021

1382
3700 3000 1589 ' .

 

(b)

mar—06:7:

 

 

 

’6‘
v
van—Dc: rang

(d)

 

I ‘ I I I
4000.0 3000.0 2000.0 1000.0

Wavenumber [cm'l ]

Figure 1.16 DRIFTS spectra for Ba/Na/lSiOz catalyst acquired at different times
during preparation: (a) pure lSiOz catalyst; (b) lSi02 treated with CH3-ONa; (c)
Ba(NO3)2/NaO-lSi02 catalyst dried at 125°C; ((1) Ba/Na/lSi02 catalyst calcined at 400°C.
Aqueous Ba(NO3)2 was then used as precursor for wet impregnation of the silica

supports. After drying at 125°C and calcination at 400°C. the catalysts were

characterized using DRIFTS and XRD. Figures 1.15 c and 1.16 c Show the DRIFTS

51

spectra of the dried catalysts and Figures 1.15 d and 1.16 d the Spectra of the calcined
catalysts. In the case of dried catalyst, a band at 1382 cm'1 attributed to NO3° vibration
[62] appeared as a new feature in the spectrum, for both types of catalysts. After
calcination, this band is diminished due to the elimination of the nitrate from the system.
This process is more pronounced in the case of the Ba/Na/lSi02 catalyst due to the low
surface area which makes the elimination of nitrate gases easier.

The information obtained from the XRD spectra of the Ba/Na supported on hSiOz
(Figure 1.17 a-d) and on lSi02 (Figure 1.18 a-d), are in good agreement with those from
DRIF TS analysis. In addition, it was found that after drying at 100°C, Ba(NO3)2 starts to
decompose and form Ba(OH)2. XRD patterns Specific to Ba(OH)2 are observable in
Figure 1.17 c and 1.18 c, respectively. No patterns of crystalline Ba(NO3)2 are Observed
after calcination at 400°C. On the hSiOz support, calcination of barium nitrate favored
the formation of y-Ba(OH)2 whereas on lSiOz support, the B-Ba(OH)2 forms [63]. AS
expected, no crystalline phases are formed or observed on the NaOH catalyst supported
on Si02 and calcined at 400°C. Knowledge about catalyst structure allow us to explain
the catalytic activity observed for these solid materials (see Table 1.6).

There are differences in catalytic activity between the Ba/N a and NaOH supported
catalysts as well as between the catalysts supported on low and high surface area SiOz.
The neat Si02 supports Show little catalytic activity (lactic acid conversion is below 30
%). The hSiOz support is more active, but toward acetaldehyde formation. This fact can
be attributed to higher number of acid sites on the hSiOz than on the lSiOz support,

favoring the acetaldehyde pathway (see Figure 1.1).

52

 

 

 

(a)
Ba(N03)2
<220>
I
(b)
(C)
(d)
I I I I . I
10.0 20.0 30.0 40.0 50.0
20[degrees]

Figure 1.17 XRD spectra for hSiOz supported Ba/Na and NaOH catalysts: a)
Ba/Na calcined at 400°C; b) BafNa dried at 100°C; c) NaOH calcined at 400°C; (1) pure
hSi02 support.

53

 

B‘Ba(OH)2
Ba (N03 ‘
<210>
\ l (a)
BafN03)2
<220>
//3
(b)
(C)
(d)
rl * * l I i - f
10.0 20.0 30.0 40.0 50.0
2 0 [degrees]

Figure 1.18 XRD spectra for ISi02 supported Ba/Na and NaOH catalysts: a)
Ba/Na calcined at 400°C; b) Ba/Na dried at 100°C; c) NaOH calcined at 400°C; (1) pure
lSiOz support.

54

Table 1.6 Yields and selectivities from lactic acid conversion over Ba(OH)2 and
NaOH supported on SiOz, at 280°C, with 1 mole MOH/g SiOz loading

 

 

 

 

 

 

 

 

 

 

Low High Ba/N a Ba/Na on NaOH/ NaOI-I/
Products SiOz Si02 on lSiOz hSiOz lSiOz hSiOz
Yield in
23P 0.1 1.2 30.3 12.3 15.3 16.7
Acrylic Acid 0.4 0.5 7.8 0.0 3.7 2.9
Propanoic Acid 0.2 0.0 0.7 2.1 0.0 0.7
Acetaldehyde 1.5 18.4 9.6 37.1 2.5 5.1
Selectivity
23P 3.5 6.0 59.9 23.6 64.6 66.0
Propanoic Acid 9.8 0.0 1.4 3.9 0.0 2.6
Acetaldehyde 66.8 91.2 18.9 70.9 10.4 20.1

 

 

 

 

 

 

 

 

 

Lactic acid conversion on the Ba/Na catalysts is much higher (reaching 97% on
hSi02 support) than on the NaOH catalyst, at the same reaction conditions. This activity
is probable due to the presence of both Ba and Na hydroxide on the catalysts surface, as
observed from DRIFTS and XRD data. However, the XRD pattern shows narrow peaks
which correspond to large crystalline particles (low dispersion), specific to [3- and y-
Ba(OH)2 phases on the Si02 support. Lower Ba dispersions result in higher yield and
selectivity toward acetaldehyde, as observed in the case of the neat SiOz support. As a
consequence, on untreated supports, yield and selectivity to 23P are relatively low, and
decrease in the case of the Ba/Na catalyst supported on hSiOz. The NaOH catalysts show
high yield and selectivity in 23P with small difference between the catalyst supported on

lSi02 and hSiOz. This may suggest that NaOH is well dispersed on the SiO,. neutralizing

55

the surface acidity of the catalysts and minimizing acetaldehyde formation. Lactic acid
conversion however, reaches only 40%, much lower than in the case of Ba/Na supported
catalysts.

M0H/Si02 catalysts- Influence of MOH loading on lactic acid conversion (M =
Na, K, Cs). A basic conclusion from the work previously presented is that the yield and
selectivity toward 23P formation is favored by the basicity of the catalyst and the
configuration of the surface acidity of the support. A comparison study of several sodium
salts supported on various type of supports have indicated that the hydroxide may act as
an optimum catalyst for conversion of lactic acid to 23P [1, 10, 64].

This part of the study explores the importance of the basic character of the catalyst
for 23P formation and tries to optimize the reaction temperature and alkali metal cation
hydroxide type and loading on the hSiOz. Figure 1.19 shows the temperature dependence
of the yields and selectivities for lactic acid conversion to 23P, in the 250°-380°C range,
on various types of hSiOz supported alkali metal (M = Na, K, CS) hydroxide (1 mmole/g
support) catalysts. As observed, the optimum temperature range is around 300°C, for
CsOH and KOH supported catalyst, and slightly higher for NaOH. Yields of 40% with
selectivities of 80% in 23P were obtained for the CsOH/hSiOz catalyst. Lower
temperature favors lower conversion but higher selectivity toward 23P. Pure hSiOz
support has little catalytic activity (yields and selectivity to 23P do not exceed 10%, at
temperatures lower than 350°C). In contrast, acetaldehyde formation is favored by high
temperatures and by the neat hSiOz support, as observed in Figure 1.20. The trend of

catalytic activity is similar for all MOH supported catalysts.

56

There are significant variations in the catalytic activity toward 23P formation as a
function of MOH loading. As observed in Figure 1.21 a, the yield in 23P increases with
the MOH loading, reaching a maximum at 2 mmole MOH/g hSiOz support. The
difference between 2 mole and 1 mmole MOH is not significant. It appears that above
2 mole MOH/g support there is a saturation effect, which may be attributed to either a
limited access of the lactic acid to the MOH to form lactate or to the formation of an non-
optimum lactate/lactic acid ratio (pH value of the buffer), at the contact time used in the
experiment. At very high MOH loading (9 mmole/g, not shown) the yield and selectivity
to 23P is very low. The trend is very similar for the selectivity in 23P, as observed in
Figure 1.21 b. The best yield (42%) and selectivity (80%) was obtained in the case of the
2 mole KOH/g catalyst.

The catalysts behavior toward acetaldehyde is opposite from that observed in the
case of 23P; at low MOH loading the lactic acid conversion is higher than in the case of
high loading. This can be attributed to the catalytic effect of the support which can be
accessed by the lactic acid feed due to low MOH coverage. Compared with previous
studies on same type of catalysts, but using a lSi02 support with well controlled pore size
distribution [10, 64], the 23P yield and selectivity results are comparable (lower by z3-
5%) with those presented here, in the case of a hSiOz support. As a consequence, in this
case it appears that the support porosity plays a minor role in the 23P reaction pathway

(Figure 1.1).

b)

57

 

 

 

 

 

 

 

 

 

 

 

 

 

w I r f T r T I
50‘ I purehSiO2 .
. x Q NaOH’hSiO2
40. .. q
{a V A KOH/hSIO2
:30 x y OsOIIhSO,
KI ' A ‘
.I:
19
Q 204 . -
>- ] . g
P x . . I 3
’ O
0" '4‘- f .r I r I
260 280 300 320 340 360 380
ReactimTenperaulePC]
(X) I I r T T I f ' I
._. 703 V A ' ”“39 :
\° I e NaOl-l’hSrOz .
e. 60, A X _
a. . V A KOHIISIO2 j
E 50. . A y CsOHhSIO2 .
40L 1
O
. o .
E 30. X .J
20.‘ 9 .
. I .
éi 10. I l .
I . I . I
O

 

 

 

260'330'300'

330 ' 340360
RmimTenperaurePC]

Figure 1.19 Conversion of lactic acid to 23P as a function of temperature, for
hSi02 supported MOH (M = Na, K, Cs) catalysts: a) yields; b) selectivities.

58

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8)
704 - -
. || PURahSKJZ
F56“ . NaOIIhSio2 -
°\ 1 . 4
~— 50- A W502 . . ,
. I'r (BCXIhSK5 I
40‘ 5 -
4 II
36. 5 .
‘L ‘7
.: 20‘ V
a t - . : fl
5' 10- I 3 x .
0‘ ’ I V I ' I I r I ' l
260 280 300 320 340 360 380
ReactbnTenpetauIePC]
b)

100 1 e r 1 , I 4 e
_ 9ol I 3
§ 30: . ' L

70L ' ' L

604 A i -

50d -

40‘ F l
g 30? g . 9 . PurehSio2 I
g 20“ v A v a kOHhSIO, -

10‘ V -

65 1 v CsOIIhSIO2 -

0 I ' I ' T fi'v—I ' I r I —-I—‘

260 280 300 320 340 360 380
ReactionTerrperanrePC]

Figure 1.20 Temperature dependence of lactic acid conversion to acetaldehyde
for hSiOz supported MOH (M = Na, K, Cs) catalysts: a) yields, b) selectivities.

59

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a)
50 I I i l I I T I
i . Nili a. :
0 K01
40- a (30H ‘ 0 -
, II
E: M)- ‘. _
D.
53 I
.s
23 20- a
>- A
I
0‘! I1
. -
‘ n
,. II
0! ' If ' I ' I T I r I ' I
00 05 10 15 20 25 30
Wilminghmnle’g]
b)
I I I I T I I
a)- 1, .
A
- 1
II
60— _
E I '
53
.5
.Z‘4oa -
.5 ll
31
20- . a
- NaOH
.1. q. KOH
If .a.(30H
O I V I ' I r I ' I ' I ' I
00 05 10 15 20 25 30
Wilordingmrmle’g]

Figure 1.21 Activity data for lactic acid to 23P conversion on hSi02 supported
MOH (M = Na, K, Cs) catalysts, as a function of MOH loading: a) yields, b) selectivities.

60

Compared with previous studies on same type of catalysts, but using a lSiOz
support with well controlled pore size distribution [10, 64], the 23P yield and selectivity
results are comparable (lower by z3-5%) with those presented here, in the case of a hSiOz
support. As a consequence, in this case it appears that the support porosity plays a minor
role in the 23P reaction pathway (Figure 1.1). A high surface area favors the
acetaldehyde pathway (Figure 1.1) due to the higher number of surface acid sites.
However, this acidity is easily neutralized by the strong basicity of the alkali metal
hydroxides supported on the hSiOz, and the reaction pathway is directed more toward
23P formation, a process favored by a base catalyst. Lactic acid conversion on a 2 mole
CsOH/g hSi02 catalyst has proven to be optimum for 23? pathway. The easy access of
lactic acid to a Cs ion characterized by a large ionic radius, small hydration sphere and
mobility, low Lewis acidity allows the lactate/lactic buffer formation, with a pH optimum
for the process. The differences between KOH and CsOH are very small, so the former
may be more attractive from catalyst preparation and economic points of view.

1.4 Conclusions

From the data presented above, several sets of conclusions can be outlined: l) the
yield and selectivity of phosphate-catalyzed lactic acid conversion to 23P is a function of
the sodium phosphate species from the Si/Al support; conversion up to 20% in lactic acid
with selectivity of 30% in 23? was obtained on Na3PO4 supported on Si/Al catalyst;
lower yields and selectivity in 23P were obtained in the case of NaH2P04/Si/Al catalysts;
this fact was attributed to acidity of the catalyst, reflected by high yield and selectivity in

acetaldehyde; sodium polyphosphates are formed on the catalysts’ surface after exposure

61

to the reaction feed; 31P-NMR relaxation time experiments help to differentiate various
type of sodium phosphate species present on the Si/Al support and to evaluate the extent
of sodium phosphate polymerization; regeneration of the active sodium phosphate
species, by recalcination at 450°C can be done only partially; 2) DRIFT spectroscopy has
identified the formation of sodium lactate on the catalyst surface; this intermediate may
play an important role in the mechanism of catalytic lactic acid conversion to 23P; 3) the
presence of Ba(OH)2 on the SiOZ surface may improve thermal resistance to the NaOl—l
supported catalyst but negatively affects the selectivity of 23P formation from lactic acid;
4) a small decrease in the selectivity towards 23P with a slight increase in the selectivity
towards acetaldehyde is observed for the high surface area SiOz support in comparison to
the low surface area, SiOZ; this was attributed to a higher surface acidity for the former
support; 5) the yields and selectivities of lactic acid conversion toward 23? is a function
of the loading and type of the alkali metal salt used to prepare the MOH supported
catalysts; the yield and selectivity toward 23P increase in the order Ba < Na < K < Cs.
Optimized conditions (300°C, 0.5 MPa, 1-2 mmole MOH/g SiOz, etc) and catalysts give

23P yields as high as 42% with selectivity up to 80%.

“99°NP‘MPP’NS‘

12.

13.
14.
15.
16.

17.

18.

19.
20.

21.

22.
23.
24.
25.

26.

27.

28.
29.
30.
31.
32.

62

1.5 References

Gunter, G. C.; Miller, D. J .; Jackson, J. E. J. Catal. 1994, 148, 252-260.
Lipinsky, E. S.; Sinclair, R. G. Chem. Eng. Prog. 1986, 82(8), 25-.

Holmen, R. E. US. Patent 1958, 2, 859, 240.

Paparizos, C.; Dolhyj, 8.; Shaw, W. G. US. Patent 4, 1988, 786, 756.

Sawicki, R. A. US. Patent, 1988, 729, 978.

Moffat, J. B. Catal. Rev. -Sci. Eng. 1978, 19(2), 199-258.

Monma, H. J. Catal. 1982, 75, 200-203.

Mok, W. S. L.; Antal, M. J. Jr.; Jones, M. Jr. J. Org. Chem. 1989, 54, 4596-4602.
Lira, C. T.; McCrackin, P. J. Ind. Eng. Chem. Res. 1993, 32, 2608-.

Gunter, G. C.; Langford, R. H.; Jackson, J. E.; Miller, D. J. Ind. Eng. Chem. Res.
1995, 34, 974-980.

Gunter, G. C.; Craciun, R.; Tam, M. S.; Miller, D. J.; Jackson, J. E., J. Catal.
1996, 164, 207-219.

Matsuda, T.; Tanabe, J .; Hayashi, N.; Sasaki, Y.; Miura, H.; Sugiyama, K. Bull.
Chem. Soc. Jpn. 1982, 55, 990—994.

Kijenski, J.; Malinowski, S. J. Chem. Soc. Faraday I, 1978, 174, 250-261.
Dudzik, 2.; George, Z. M. J. Catal. 1980, 63, 72-82.

Klaassen, A. W.; Hill, Jr. C. G. J. Catal. 1981, 69, 299-311.

Holten, C. H.; Muller, A.; Rehbinder, D. Lactic Acid - Properties and Chemistry
of Lactic Acid and Derivatives, Verlag Chemie, 1971.

Clark, J. H. Catalysis of Organic Reaction by Supported Inorganic Reagents,
VCH Publishers, Inc. New York, 1994.

Walkup, P. C.; Rohrmann, C. A.; Hallen, R. T.; Eahin, D. E. PRAI U.S. Pattent
Nr. 90-468704, 1991.

Malinowski, S.; Basinski, S. J. Catal. 1963, 2, 203-207.

Andrew, E. R.; Bryant, D. J .; Cashell, E. M.; Dunell, B. A. Chem. Phys. Lett.
1981, 77(3), 614.

Tuner, G. L.; Smith, K. A.; Kirkpatrick, R. J .; Oldfield, E. J. Magn. Res. 1986, 70,
408.

Hiyashi, S.; Hayamizu, K. Bull. Chem. Soc. Jpn. 1989, 62, 3061-3068.
Prabhakar, S.; Rao, K. J.; Rao, C. N. R. Chem. Phys. Lett. 1987, 139(1), 96-102.
Duncan, T. M.; Douglass, D. C. Chem. Phys. 1984, 87, 339-349.

Diehl, P.; Fluck, E.; Gunther, H.; Kosfeld, R.; Seelig, J. NMR Basic Principles
and Progress, Springer-Verlag Berlin Heidelberg, 1994, Vol. 30, p. 3-27.
Fukushima, E.; Roeder, S. B. W. Experimental Pulse NMR - A Nuts and Bolds
Approach, Addison-Wesley Inc. 1981, p. 125-295.

Fierro, J. L. G. Spectroscopic Characterization of Heterogeneous Catalysts, Part
A: Methods of Surface Analysis, Elsevier, Amsterdam, 1990, p. A18-A28.
Herzfeld, J.; Berger, A. E. J. Chem. Phys. 1980, 73(12), 6021-6030.

Un, S.; Klein, M. P. J. Am. Chem. Soc. 1989, 111, 5119-5124.

Kubelka, P.; Munk, F. Z. Tech. Phys. 1931, 12, 593-598.

Fuller, M. P.; Griffiths, P. R. Anal. Chem. 1978, 50(13), 1906-1910.

Fuller, M. P.; Griffiths, P. R. Appl. Spectrosc. 1980, 34(5), 533-539.

33.
34.
35.
36.
37.

38.
39.
40.
41.
42.

43.
44.

45.
46.
47.
48.
49.
50.

51.
52.
53.
54.
55.
56.
57.

58.
59.

60.
61.

62.

63.
64.

63

Fraser, D. J. J.; Griffiths, P. R. App]. Spectrosc.l990, 44(2), 193-199.

Hembree, D. M. Jr.; Smyrl, H. R. Appl. Spectrosc. 1989, 43(2), 267-274.
Yeboah, S. A.; Wang, S. H.; Griffiths, P. R. Appl. Spectrosc. 1984, 38(2), 259.
Hamadeh, l. M.; King, D.; Griffiths, P. R. J. Catal. 1984, 88, 264-272.

Bell, A.T. in Vibrational Spectroscopy of Molecules on Surfaces, edit by Yate, J.
T. Wiley, New-York, 1987.

Venter, J. J.; Vamrice, M. A. Appl. Spectrosc. 1988, 42(6), 1096-1102.

Miller, M. P. App]. Spectrosc. Rev. 1987, 23, 329-345.

VanEvery, K. W.; Griffiths, P. R. Appl. Spectrosc. 1991, 45, 347-359.

Kortum, G.; Delfs, H. Spectrochim. Acta, 1964, 20, 405-413.

Niwa, M.; Hattori, T.; Takahashi, M.; Shirai, K.; Watanabe, M.; Murakami, Anal.
Chem. 1979, 51, 46-52.

Loyalka, S. K.; Riggs, C. A. Appl. Spectrosc. 1995, 49(8), 1107-1110.

Frei, R. W.; MacNeil, J. D. Difluse Reflectance Spectroscopy in Environmental
Problem-Solving, CRC Press, Cleveland, USA, 1973.

White, R. L. Appl. Spectrosc. 1976, 30, 593-596.

Coblentz, W. W.; Natl. Bur. Stand. (US) Bull. 1913, 9, 283.

Wood, B. E.; Pipes, J. G.; Smith, A. M.; Roux, J.A. Appl. Opt. 1976, 15, 940-945.
Blevin, W. R.; Brown, W. J. J. Sci. Instrum. 1965, 42, 385-392.

Spectra-Tech Inc. Product Brochure, 1995.

Harrick, N. D. Optical Spectroscopy: Sampling Techniques Manual, Harrick
Scientific Co. Ossining, New York, 1987.

Bernitez, J. J .; Carrizosa, 1.; Odriozola, J. A. Appl. Spectrosc. 1993, 47(11),
1760-1766.

Leyden, D. E.; Murthy, R. S. S. Trends in Anal. Chem. 1988, 7(5), 164-168.
Sullivan, D. H., Conner, C. W., Harold, M. P. Appl. Spectrosc. 1992, 46, 811-818.
Craciun, R.; Craciun, L. Studia, 1997, in press.

Beeston, B. E. P.; Home, R. W.; Markham, R. Electron Diflraction and Optical
Difiraction Techniques, American Elsevier Publishing Company, Inc. New York,
1973, p. 223-316

Niemantsverdriet, J. W. Spectroscopy in Catalysis, VCH, Weinheim, 1993,

p. 293-317.

H. P. Klug, and L. E. Alexander, X -ray Dijfiaction Procedures for Polycrystalline
and Amorphous Materials, I-st Ed. Wiley, New York, 1954.

Wall, F. T.; Doremus, R. H. J. Am. Chem. Soc. 1954, 76, 868-870.

Groppi, G.; Belloto, M.; Cristiani, C.; Forzatti, P.; Villa, P. L. Appl. Catal. 1993,
104,101-108.

Morrow, B. A.; McFarlan, A. J. J. Phys. Chem. 1992, 96, 1395-1400.

Kustov, L. M.; Borovkov, V. Yu.; Kazanskii, V. B.; Russ. J. Phys. Chem. (Engl.
Trans].), 1985, 59, 1314-1318.

Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination
Compounds, 3-rd Edition, Wiley & Sons Inc. New York, 1978, p. 105-401.
McClune, W. F. Powder Difiraction File: Inorganic phases, Swarthmore, 1983.
Wadley, D. C.; Tam, M. S.; Kokitkar, P. B.; Jackson, J. E.; Miller, D. J. J. Catal.
1997, 165, 162-171.

CHAPTER 2

THE MECHANISM OF CATALYTIC LACTIC ACID CONVERSION TO 2,3-
PENTANEDIONE: CORRELATION OF EXPERIMENTAL
AND MOLECULAR MODELING DATA

2.1. Introduction

Lactic acid obtained from com-based fermentation processes represents a new
resource for the chemical and food industries. Conversions of this renewable biomass-
derived feedstock to acrylic acid or propanoic acid have been explored under
heterogeneous catalysis [1,2] and supercritical conditions [3,4]. Previous studies in our
laboratory have shown that, unexpectedly, aqueous lactic acid (typically 30% in H20) can
be converted to 2,3-pentanedione (23P) along with acrylic and propanoic acid
byproducts. The reaction takes place over inexpensive and non-toxic alkali metal
phosphates, nitrates, and hydroxides supported on Si or Si/Al oxides of various surface
areas [5-7]. More recently it has become clear that the same conversion can occur in
heated aqueous lactic acid/alkali metal lactate solutions [9-10]. Optimized temperature,
pressure and alkali metal loadings have led to yields as high as 60% with 80% selectivity
toward 2,3-pentanedione formation [10-11]. Used primarily as a flavor agent and food
additive, 2,3-pentanedione has a relative small market (5,000 kg/year). Our discovery of
this new low cost route (<815/kg compared with previous methods at $120/kg) has led us

to seek new applications for this diketone and its various valuable derivatives [12]. In

64

65

previous reports, however, we have only speculated on the mechanism of 2,3-
pentanedione formation [5-11].

This paper analyzes the mechanism of lactic acid conversion to 2,3-pentanedione,
in the presence of supported base catalysts (MOH/SiOz, where M = Na, K, Cs). Possible
reaction intermediates are suggested by product analysis, variable-temperature-mass
spectrometry (VT-MS), post-reaction diffuse reflectance infrared Fourier transform
(DRIFT) spectroscopy, deuterium labeling studies, and computational modeling using
semi-empirical and ab initio molecular orbital calculations. These approaches may be
extended to the general understanding of any Claisen condensation process occurring in a
base catalytic system.

Experimental

Post-reaction DRIFT S. Diffuse reflectance infrared Fourier transform (DRIFTS)
spectra were acquired using a Perkin-Elmer Spectrum 2000 instrument equipped with a
diffuse reflectance attachment, an environmental in-situ reactor from Harrick Scientific
Co. and a mercury-cadmium-telluride (MCT) detector. The catalyst powder (20-25 mg)
was exposed to lactic acid vapor at various temperatures (25°-350°C) and analyzed by
DRIFTS. The resulting spectra are presented in Kubelka-Munk units (see Chapter 1).
For the isotopic labeling experiments, samples were analyzed using traditional FTIR
using the same IR spectrophotometer. Thin layers of liquid sample deposited on a silica
window were used to examine C-H versus C-D bands in the transmission mode utilizing
a dimethyl-triglycine sulfate (DTGS) detector. In both operational modes, spectra were

acquired with 4.0 cm'l resolution over 400-4000 cm'I range.

66

Molecular modeling. The Spartan software developed by Hehre et al. [13] and
running on a Silicon Graphics Indigo 2 computer system was used for molecular orbital
and solvation calculations. Previous evaluations of various semiempirical molecular
modeling programs such as MNDO [14], AMI [15], and PM3 [16, 17] have indicated
that the PM3 model performs the best in describing carboxylic acids and their acidities
[18]. Table 2.1 shows such AH°f values calculated using various semiempirical molecular
modeling programs in comparison in comparison with the experimental value for several

organic molecules with structures similar to that of lactic acid.

Table 2.1 Experimental and calculated enthalpies of formation (AH°f) for various
carboxylic acids [17]

 

 

 

 

 

 

 

 

 

0rd # Carboxylic Enthalpy of Formation, AH", [kcal/mol]

Add Experim. AM 1 MNDO PM3
1 Methanoic -378.7 -407 -387 -395
2 Ethanoic -432.8 -43 1 -423 -426
3 Propanoic -453.5 -456 -444 -444
4 Butanoic -475.8 -485 -464 -467
5 Pentanoic -491 .9 -513 -480 -490
6 Hexanoic -51 1.9 -542 -500 -512
7 1,6-Hexanedioic -865.1 -895 -836 -857
8 Lactic acid -l49.5 -153 -147 -l47

 

 

 

 

 

Consequently, PM3 was used as a starting point to calculate heats of formation for
each intermediate considered in the mechanism. Analogous ab initio calculations were

performed at STO-3G, 3-21G, and 6-31G* levels, models which improve the PM3 model

67

in more accurately describing hydrogen bonding and dipole-dipole interactions. This
type of mechanistic approach has been successfully applied to another unrelated catalytic
reaction mechanisms [19]. Because the reaction is thought to occur in a highly polar
condensed phase (catalyst surface or liquid film), solvation calculations were performed
using the Cramer/Truhlar SM3 model for water [20]. Ionic species of varying sizes play

key roles in the mechanism, and we expect solvation to substantially perturb the reaction.

a-H/D Isotopic labeling experiment. Lactic acid 85 % solution (Aldrich) was
dried under vacuum and by benzene-water azeotropic extraction, to remove as much of
the water as possible from the feed. Dried lactic acid was used to prepare a feed of 35%
(by weight) concentration in D20. H/D exchange reactions were performed by passing
this feed over a 0.5 mmole KOH/g SiOz catalyst, at 200°C and 250°C, temperatures
favoring the IUD exchange reaction but below the range where a significant conversion to
products occurs. The dried lactic acid, the feed and the product were analyzed by 1H and
ZH-NMR, FTIR, and GC-MS methods.

Gas chromatography-Mass spectrometry (GC-MS). GC-MS analyses were
carried out on a JEOL AX-505H double-focusing mass spectrometer coupled to a
Hewlett-Packard 5890J gas chromatograph via heated interface. GC separation employed
a DBWax fused-silica capillary column of 30m length x 0.25 mm ID. with a 0.25 pm
film coating (J&W Scientific Co). Direct (splitless) injection was used. Helium gas flow

was 1 ml/min. The GC temperature program was initiated at 50°C, with a ramp rate of

10°C/min.

68

Variable temperature-mass spectrometry (VT -MS). In order to mimic the reaction
conditions and collect in-situ data, VT-MS experiments were performed. Figure 2.1
shows the instrumental setup for this type of experiments. Samples for VT-MS were
capillary tubes containing a 2 mole KOH/g Si02 catalyst (280 mesh, 1-2 mg) and lactic
acid feed (1-2 uL); control experiments omitting the KOH catalyst or using sodium
lactate and dilactide feed were also performed These samples closely approximate the

real catalytic system and yield significant insight into the mechanism operating therein.

 

 

 

MS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 2.1 Experimental setup for VT-MS experiments: 1) temperature regulator;
2) mass spectrometer; 3) capillary holder.

69

Capillaries were placed in the temperature-controlled block heater of the mass
spectrometer. The vaporized products from a sample of catalyst mixed with lactic acid
reaction feed were analyzed via mass spectrometry as the temperature in the sample probe
was raised from 25°C to 350°C, at ramp rates of 8°C or 16°C/min. Conditions in the MS
were as follows: interface and ion source temperatures of 240°C, electron energy of 70
eV, mass spectrometer scan rate of 1 scan/s over the m/z range 0-400 amu.

Results and Discussion

Our prior optimization studies of lactic acid conversion to 2,3-pentanedione have
uncovered the following mechanistic elements: i) the reaction is catalyzed by bases,
including alkali metal lactates themselves; little or no 23P is obtained from pyrolyses of
either lactic acid or alkali metal lactates alone [21-23]; ii) kinetic analysis points to a
process that is second order in lactic acid and first order in catalyst, as expected for a
condensation process; an approximate Arrhenius activation energy (13,) of 32.3-32.7
kcal/mol is obtained for a 2 mmole KOH/g SiOZ catalyst, though its exact interpretation
is shadowed by the complexity of the continuous-flow, heterogeneous reaction conditions
[24]; iii) studies of lactic acid feeds containing ammonium lactate (as a potential impurity
found in fermentation broth) show reversible poisoning of the reaction by ammonia; full
recovery of catalytic function is seen on returning to the unadulterated feed. Formation
of 23P is favored by basic catalysts, suggesting that enolization might be a key
intermediate. Scheme 2.1 displays the proposed molecular mechanism, in which a

Claisen condensation is the critical step for C-C bond-forming reaction [25].

70

Scheme 2.1 The proposed mechanism of lactic acid conversion to 23P

nucleo hilic attack

 

 

 

 

 

 

 

 

Steal £1122
o§ /OH 0% /0' (M‘) o§ /OH H3C\ /OH
C C A {j C C
2 I + | I + II') ——>
CH\ /CH\ /CH\ / \
H3C/ OH H3C OH H3C OH HO LOH
Lactic Acid Lactate l Lactic Acid Enol
(M+ = H+, Na”, K”, Cs+) (+ Lactate 1)
Cl)“ Steal (IDH SL621
HO HO
C\ C\ H3C OH
H3C~ c/ \ o H3Cx c/ \ o \ C/
H C (I: H C A H C\ E
3 "”I ' H20 3 \ ' C02 3 /
\CH’ \ O” \CH’ \o CH \on
I OH I |
OH OH OH
Intermediate-H 1 lntennediate-H ll lntennediate-H 111
(+ Lactate I) (+ Lactate l + H20) (+ Lactate l + C0: + H20
Steal SLED—fl
| __. II
—————-> .__
_ H3C C H3C\ C CH;
H20 \ $¢ \ OH CH? \ fi/
H O
2,3-Pentanedione-enol 2,3-Pentanedione
(+ C02 + 2 H20 + Lactate l) (+ C02 + 2 H20 + Lactate 1)

Under high temperature and buffered lactic acid/alkali lactate conditions, hydrogen bonds
and ion pairing may stabilize the enolate or enol forms of lactic acid which then may
attack a molecule of lactic acid at the carbonyl. After loss of water, the six-carbon
dihydroxy-ketocarboxylate (Intermediate 1) loses C02 to form Intermediate 11. Loss of
hydroxide leads via Intermediate 111 and subsequent ketonization, to 23P, in the same way

that dihydroxyacetone isomerizes under enolizing conditions to pyruvaldehyde [25]. In

71

the following discussion, we present new evidence for this reaction sequence and show
how it explains the early findings describes above.

Molecular orbital calculations. Data from PM3, PM3-SM3, and ab initio
calculations at the STO-3G, 3-21G, and 6-31G* levels performed for the reactants,
possible intermediates, and products are presented in Table 2.2. In the proposed pathway,
most of the intermediates are stabilized by intramolecular hydrogen bonding. By
exploring various states of protonation/deprotonation relative to lactic acid/lactate, we
considered the most “relaxed” forms (with respect to proton exchange with the medium)
at each reaction stage. Though the condensation step (step 2) is entropically unfavorable,
the process is aided by the low enthalpies of formation of the resulting intermediates,
while subsequent fragmentations releasing volatile H20 and C02 are favored entropically.

The differences between PM3 and PM3-SM3 calculated energies for each
component involved in the proposed mechanism are presented in a separate column of
Table 2.2 as well. These “hydration enthalpy” values were combined with the ab initio
results to further explore the effects of solvation. Based on these calculations, the energy
variations for each step of the mechanism has been evaluated and are presented in Table
2.3. The energies vary over 20 kcal/mol range both in the gas phase and aquated models.

At the process temperatures, such energy variations are easily accessible.

Table 2.2 PM3 and ab initio data for reactants, intermediates and products
involved in the mechanism of lactic acid conversion to 2,3-pentanedione

 

 

 

 

 

 

 

 

 

 

 

 

Interm. of Enthalpy-PM3 Ab initio
the 23P [kcal/mol] [hartrees]
mechanism Gas phase SM3) A“ STO-3G 3-21G 6-31G*
Lactic Acid -147.3" -159.1 -11.8 -337.21908 -339.79087 -34l.69662
Lactatel -174.8 -246.9 -72.1 -336.49629 -339.21895 -341.l34l9
Lactate II ~157.4 -226.9 -69.6 -336.43235 -339.14697 -341.05096
Lactic Acid -132.3 -l43.2 -10.9 -337.18294 -339.75112 -341.64003
Enol
Interm.-HI -284.5 -298.3 -13.8 -674.46588 -679.59049 -683.36488
Interm.-HII -223.5 -240.6 -17.1 -599.43273 -603.96437 -607.34283
Interm.-H 111 -138.6 -149.7 -11.1 41433918 -4l7.38976 -4l9.70962
Enol form -80.5 -88.4 -7.9 -339.32205 -34l.79054 -343.70703
23P
23P -87.7 -95.2c -7.6 -339.34396 -34l.80219 -343.72497
C02 -85.1 -91.9° -6.8 -185.06839 -186.56126 -187.63418
H20 -53.4 -59.7e -6.l -74.96590 -75.58596 -76.01075

 

 

 

 

 

A = energy difference between the PM3 and PM3-SM3 calculations; experimental
enthalpies of formation [kcal/mol]: b = -l49.5i1.5; c = -83.5i0.5; d = -94.1; e = -57.8.

2 CH3CHOHCOOH —> CH3C(O)C(O)CH2CH3 + C02 + 2 H20
Lactic acid 23P

(2-1)

The experimental gas phase enthalpy of reaction for lactic acid to 2,3-
pentanedione (reaction 2-1, given above) was evaluated from literature data to be AH, =
+5.4 kcal/mole [26]. With the PM3 model, the overall reaction enthalpy was also
calculated to be endothermic, with AH, = +15.0 kcal/mol (gas phase) and AH, = +11.7
kcal/mol (aquated) in qualitative agreement with the experimental value.

The excess

endothermicity is largely due to PM3’s 9 kcal/mol underestimation of the stability of

C02. The ab initio results can be interpreted in a similar manner, as shown in Tables 2.2
and 2.3. The energies of each proposed mechanistic stage, calculated from ab initio data

and the solvation model, are presented in Table 2.3 as well.

Table 2.3 Relative energies calculated in gas phase and solvated models from PM3
and ab initio methods for C6HI3O7' speciesa at each step of the proposed mechanism,
normalized to the energy of the starting materials

 

 

 

 

 

 

 

 

Steps PM3 Ab initio STO-3G Ab initio 3-21G Ab initio 6-31G*
' [kcal/mol] [kcal/mol] [kcal/mol] [kcal/mol]
Gas ph. SM3 Gas solvatedb Gas solvatedb Gas solvated”
feed 0 0 0 0 0 0 0 0
1 +150 +15.9 +22.8 +23.6 +24.9 +259 +355 +365
2 +101 +19.9 -17.4 -7.5 -5.5 +4.4 +17.8 +27.6
3 +17.7 +17.9 +248 +16.3 +19.7 +20.2 +24.9 +253
4 +175 +16.9 +40.6 +40.3 +28.0 +27.7 +24.3 +239
5 +222 +18.5 +72.8 +69.5 +36.3 +33.1 +192 +159
6 +150 +11.7 +590 +56.3 +29.l +261 +7.9 +4.9

 

 

 

 

 

 

 

a - the components considered in our calculations for each step are presented in
Scheme 2.1; b - calculated values were corrected with the solvation energies obtained
from PM3-SM3 calculations
As with PM3, there is a leveling effect when the SM3 solvation model is used to correct
the ab initio gas phase values. The 6-31G*-SM3 data gave values for the energies
following the reaction pathway closer to the PM3 model (not greater than 25 kcal/mol
variation) and AE, = 7.9 kcal/mole in the gas phase (AH, = 4.9 kcal/mol solvated) in

reasonable agreement with the experimental AH, value. Considering the crudeness of the

methods empployed, the highest energy points found along the reaction path (19.9

74

kcal/mol and 36.5 kcal/mol at PM3-SM3 and HF/6-3lG*-PM3 levels) are in rather good
agreement with the experimental measured 32-33 kcal/mol activation energy (13,) value.
Additional experiments using DRIF TS analysis, H/D exchange and VT-MS have
been performed to probe further the 23P formation mechanism outlined in Scheme 2.1.
DRIFT S data. Figure 2.2 displays DRIFTS spectra of a 2 mmole KOH/g SiOz
catalyst before and after exposure to lactic acid vapors at various temperatures (in the
range ISO-350°C) in an in-situ DRIFTS reactor. At room temperature, only the bands
attributed to Si-O vibrations from Sio, at around 800 cm" and woo-1200 cm" are
observable [27]. After exposure to lactic acid vapors, the strong carbonyl stretching
bands characteristic of -CO0H and -C00' at 1721 cm’l and 1597 cm", respectively [6-
8], together with the less intense bands specific to these species, start to appear and
become well defined at temperatures higher than 150°C. At temperatures above 300°C,
the band at 1597 cm’l is shifted to lower frequency, indicating the formation of another
type of carboxylate, identified to be potassium propanoate; this species can remain on the
catalyst surface at very high temperature (350°C) in the absence of a continuous flux of
lactic acid [28]. The low S/N ratio observed for the DRIFTS spectra of Figure 2.2 is due
to the low light throughput of the in-situ cell. In these conditions, even the very sensitive
MCT detector does not yield very good IR spectra, but it does provide information
directly about the catalyst surface modifications under conditions closely approximating

the continuous flow reactor.

75

 

(b)

(C)

 

(d)

 

| » - l I 1
2000.0 1500.0 1000.0 500.0

Wavenumber [cm ' I]

Figure 2.2 DRIFTS spectra of KOH/SiOz catalysts after exposure to lactic acid

vapor at various temperature: a) 350°C; b) 300°C; c) 250°C; d) 200°C; e) 150°C; and
before exposure at f) 25°C.
We have previously reported the observation of lactic acid and lactate bands on silica
surfaces using transmission IR, a method that is more sensitive, but less directly
comparable to our reactor conditions [6].

Potassium lactate observed on the catalyst surface after exposure to lactic acid

vapor is one of the intermediates in the mechanism of Scheme 2.1. Indeed, an effective

catalyst can be formed by simply loading potassium lactate on silica or silica/alumina.

76

However, pyrolysis of potassium lactate alone does not form 23P; addition of fresh lactic

acid to the system is necessary for the reaction to occur [7-10, 28].

a-H/D exchange in lactic acid. Based on the mechanism presented in Scheme
2.2, base-catalyzed H/D exchange at the (It-hydrogen site is expected; the enol
intermediate or enolate of lactic acid then provides the nucleophilic center to attack
another lactic acid molecule in the condensation process. The H/D exchange process
could also be acid catalyzed, but such a path would presumably require stronger acids
than are available in our catalytic system and it is established that acid catalysts cleave
lactic acid to acetaldehyde, C0, and H20 [29]. The closely related mandelic acid system
in which the CH3 of the lactic acid is replaced by a phenyl group, undergoes analogous
proton exchange with racemization; the details of this enzyme-catalyzed process are a
topic of recent investigations [30].

The gas chromatography-mass spectrometry (GC-MS) method has been used to
observe the IUD exchange process, by analyzing the lactic acid GC peak in a lactic
acid/D20 feed mixture, before and after passage through the catalyst at various
temperatures. Standard MS spectra for lactic acid and dilactide are presented in Figure
2.3. Table 2.4 presents the abundance of the main ion at m/z = 45 amu for four samples:
the plain lactic acid in H20; lactic acid dried and rediluted in D20; and samples of the
lactic acid in D20 after passing over the KOH catalyst at 200°C and at 250°C. The peak
at m/z = 45 amu was attributed to either -+CO0H or CH3-CH+-OH ions formed during

the fragmentation of lactic acid in the MS chamber [3]-32].

77

 

 

 

 

 

 

 

 

 

 

 

a)
m)
I“, 45.0 muse
..I
1m-
29.0 43.0
19.1
o -l 'l . .I-ntfi- . j (. . l - 1 - .w". H ‘. mw-h. . . - - --
[LL/£1 a; 49. 19 an an 72 an 4L m.
b)

 

 

 

 

 

EL 1'9; r the ' 132—"' ' m ' L m

Figure 2.3 Mass spectra of standards: a) lactic acid and b) dilactide

For both ions, one D-incorporation would lead to formation of a m/z = 46 amu ion, but
only the CH3-CH+-0H has the ability to exchange more protons, including the one from
the a-position of lactic acid, leading to the formation of a m/z = 47 amu ion. There is no

molecular peak observed or reported in the mass spectrum of the lactic acid when the

78

electron impact MS technique was used for analysis [31, 35]. However, peaks with the
masses of the lactic acid molecular ion (M) m/z = 90 and m/z = 91 (M+1) were observed
in the dilactide GC peak that appeared when a concentrated solution of lactic acid was
injected into the GC or in previously reported work when direct probe/chemical
ionization (C1) or field desorption (FD) techniques were used. In all these techniques, the
mass spectrum of lactic acid reveals the formation of the [M+l]+ ion followed by
[2M+l]+ and [M-1]+ ions. These large ions must be formed in the MS chamber at high
concentration and pressure due to gas phase reactions [32-3 5]. The presence of ions that
simulate the molecular ions in these spectra caused some initial confusion in our work
which was clarified by reference to the literature and GC-MS experiments with sample
dilution and injection size. MS-MS experiment would provide more inside about the

structure of these molecular-type of ions observed ion these experiments.

Scheme 2.2 Proposed mechanism for a-H/D exchange from lactic acid molecule

a) Trivial H-D exchange on carboxyl and hydroxyl hydrogen

CH3-CH-C00H + 21320 2 CH3-CH'C00D

+
0H 0D ZHDO
b) or H-D exchange in lactic acid - base catalysis
B-
k. 1111 I8 - BH CH3\ ,0 + BD 11) go
CH3—(I: C\ .—_’ ’C=C\ ‘1’ CH3—Cl—C\

on on D0 0D '3 0D 0D

79

As mentioned above, the main fragment ion from lactic acid corresponds to m/z =
45 amu [CH3-CH+(OH) and CO0H+, though the latter may contribute little or nothing]
[32]. For the D20 treated material, the peak at m/z = 46 amu, attributable to either CH3-
CH+-0D and COOD+ ions, is enhanced compared with the undeuterated case. The
relative intensity of the peak at m/z = 47 amu, assigned to CH3-CD+-0D ion, has
significantly increased in the case of the converted lactic acid (see Table 2.4 given below)
and , as expected, is more enhanced in the sample run at higher temperature. In addition,
from the GC peak corresponding to the small amount of acrylic acid obtained in these
low conversion runs via lactic acid dehydration, the mass spectra contain the molecular
ion peak at m/z = 72 amu. Increases in the relative intensities of the molecular peaks at
m/z = 73 amu and most importantly the peak at m/z = 74 amu, corresponding to one
(CHZCHCOOD) or two (CHZCDCOOD) deuterium incorporated into acrylic acid in
comparison with the peak at m/z = 72 amu offer additional evidence of the a-H/D

exchange process.

Table 2.4 or-Deuterium incorporation into lactic acid (GC-MS data)

 

 

 

 

 

Sample Lactic acid Lactic acid in D20 Lactic acid after HID
feed exchange at
200°C 250°C

m/z Ion relative intensity Ion relative intensity Ion relative intensity
[%] 1%] 1%]

45 93.3 16.6 14.2 11.3

46 6.7 83.8 76.7 50.4

47 --- --- 9.1 38.3

 

 

 

 

 

 

80

To further confirm the a-H/D-exchange, additional information was obtained by
analyzing the DZO/lactic acid system, before and after passage through a l mmole KOH/g
Si02 catalyst, by F TIR and 2H-NMR spectroscopy. Figure 2.4 shows the F TIR spectra of
the dried lactic acid (c) from a 35% feed in D20, before (b) and after (a) passing over a
KOH/SiOz catalyst at 250°C. The IR spectra (3700-200 cm'l region) of the dried lactic
acid before H/D exchange or dissolved in D20 show only bands characteristic of -OH at
3600 cm", the -OD at 2150 cm'l, methyl and methine C-H stretches at 2994 cm", 2944
cm'l, and 2886 cm'l for lactic acid [33].

The IR spectrum of the lactic acid after passing over the catalyst bed at high
temperature shows a new band in the -CD stretch region, at 2317 cm", supporting our
assertion that or-H/D exchange has occurred. Some acrylic acid is formed at the reaction
temperature (250°C) and was identified in the product mixture by MS. An additional C-
D band observed around 2370 cm'l might be due to trace amounts of acrylic acid with D
incorporated in the (It-position, as well3. As observed from Table 2.5, calculated
vibrational frequencies for the C-H and C-D stretch bands do not agree with the

experimental value.

 

3 Frequency calculations based on PM3 data predict an or C-D stretch at 2140 cm". in relatively poor
agreement with the observations

81

2994 2944

 

(b)

(c)

I I I I
3500.0 3000.0 2500.0 2000.0

Wavenumber [cm’ 1]

Figure 2.4 IR spectra of the lactic acid after (a) and before (b) passing over a
KOH/SiOz catalyst, in comparison with the dried lactic acid (c).

82

Table 2.5 Calculated vibrational frequencies for C-H and C-D stretch bands in on-
position of isopropanol molecules from PM3 and ab initio 6-31G* data in comparison
with the experimental values

 

 

 

 

MOICCIIIBS a-C-H stretch frequency [cm'I a-C-D stretch frequency lcm'I]
experimental PM3 ab initio experimental PM3 ab initio
Lactic acid 2886 2914 3238 2318 2144 2388
K-lactate 2895 2890 3152 a 2126 2324
isopropanol 2878 2917 3239 a 2146 2400

 

 

 

 

 

a - value not available; b - these values could be corrected by dividing with 1.12

2H-NMR analysis of product obtained from lactic acid in D20 solution before and
after passing over a K0H/Si02 catalyst are presented in Figure 2.5 a-c. For the converted
lactic acid at 200°C (2.5 a) and 250°C (2.5 b) compared with the spectrum of lactic acid
just mixed with D20 (2.5 c), 2H-NMR spectra indicate the presence of a new peak with

chemical shifi corresponding to the a-methine site.

83

3.84

4.67

(a)

 

 

 

 

 

 

3.95

 

(b)

(c) Xv

[ppm]

 

 

--I
4
‘
d
-I
d
.1

Figure 2.5 2H-NMR spectra of dried lactic acid afier H/D exchange process
(passing over a K0H/Si02 catalyst) at a) 250°C and b) 200°C in comparison with c) the
lactic acid just diluted in D20 and dried

84

Considered together with the GC-MS results, these additional spectroscopic data point
clearly to incorporation of deuterium at the a-site of lactic acid. Most importantly, no
2H-NMR peaks appear at the chemical shift of the B-methyl group.

VT-MS data. An attempt was made to analyze the reaction of lactic acid
deposited on K0H/Si02 catalyst by VT-MS. These types of experiments may yield in
situ chemical and structural information about intermediates in a catalytic reaction.
However, the method can be used to analyze mostly pure compounds; analysis of
mixtures generates mass spectra that are hard to interpret, unless the compounds under
study have clearly distinguishable fragmentation patterns [36]. Application of the
method to a reacting system is even more complex, as the suite of compounds of varying
masses and volatilities, is itself changing with time over the heated catalyst. Finally, for a
condensation reaction, the relatively volatile starting material (lowest molecular weight
species of interest) must be available in large enough quantity to undergo bimolecular
reaction on the catalyst once an adequate temperature is reached. Despite these
challenges, the promise of observing the sequence of reaction intermediates and the
simplicity of the experiments led us to search for mechanistic insight using this tool.

Figure 2.6 shows the mass spectra at several temperatures of the products evolved
during heating of lactic acid on a 2 mole KOH/g Si02 catalyst, at a 16°C/min
temperature ramp-rate, with the initial temperature set at 30°C. At low temperatures
(<100°C), the mass spectra indicate only the presence of ions with m/z = 91 amu, similar
with the M+1 ion of lactic acid and the corresponding secondary ions formed in the MS

chamber observed in the case of ion source saturation from a GC-MS experiment. The

85

presence of the [M+1]+ ion together with the [2M+1]+ and [2M-H20+1]+ ions, indicate
gas phase reactions in the ion source from the mass spectrometer, due to the high gas
phase concentration of lactic acid. After 1.5 minutes (Figure 2.6 a), new, new peaks
begin to appear at m/z = 181, m/z =163, m/z = 145, m/z = 144, m/z =135, m/z = 117
(118), and m/z = 100 amu. Of these, only the resonances at m/z = 100 amu (2,3-
pentanedione) and m/z= 144 amu (dilactide) are seen in the product mixture from a
regular catalytic conversion experiment. The presence of the [M+1]+ ion together with
the [2M+1]+, [2M-COOH]+, and [2M-HZO]+ ions, points to gas phase reactions in the
MS analysis chamber. The other peaks in Figures 2.6 a-c can be attributed to fragments
formed in the MS chamber from the molecular ion corresponding to the [2M+1]+ with
m/z = 181 amu [37-38]. It appears that after a short time corresponding to a relatively
low temperature (z60°C), lactic acid evolves from the catalyst surface in large quantities
and the ions abundantly formed in the MS chamber start to react and form various type of
product ions (see Figure 2.6 a-c). Due to these gas phase reactions from the analysis
chamber, it was not possible to obtain unambiguous information about lactic acid/lactate
catalytic system from this experiment. Although the ions seen have m/z values
appropriate to the intermediates from the proposed mechanism (see Scheme 2.1), the gas
phase dimerization from m/z = 91 amu ([M+l]+) to m/z = 181 amu ([2M+1]+) and the
subsequent fragmentations via losses of H20 and C02 closely mirror the proposed

solution pathway.

86

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10° 74 91 I
(a)
181
1 I
‘ 135
217 235
,7 a, 1,. L, L
163 010.0 1
1‘7 145
0 , $1 in. AL 1L, 1.
100
74 91 135 m
(b)
163 I
a 235 f
I 217 J1
“ l
u 1
g 71* IL .!I 1L
é ‘ 145 no.0
j 117
100 l
0] ALL] [VJ 7 Y *— '1‘ A ' L J

 

 

 

 

 

 

 

 

 

 

 

 

100 }
145 (c)
74 I
1 F
100 i
117
135 163 181
89 ro.o +A j 4 i j
.1
I
0- ¢ , u.-4 —J=+__ f . .
50 100 150 200 “#2

Figure 2.6 Mass spectra of products obtained on a 2 mole KOH/g Si02 catalyst
by VT-MS analysis afier: a) 1.5 min.; b) 2.5 min.; c) 5.2 min of heating

At high desorption temperature, the only peaks observable are at m/z = 144 amu

(dilactide) or at low m/z values such as m/z = 56, and 45 amu, etc. which correspond to

87

the fragment ions from a standard MS spectrum for dilactide (see Figure 2.3 b). The
persistent presence of the peak at m/z = 144 indicates that dilactide is formed but it was
found that plain compound does not participate in further chemical transformations on the
catalyst nor play a significant role in the pathway to 2,3-pentanedione.

Figure 2.7 presents the total ion current (TIC) and VT profiles, respectively, for
several m/z ions (2.7 a) as well as a mass spectrum recorded after 2.13 minutes (2.7 b)
during the VT experiment when a 2 mole KOH/g SiOz catalyst impregnated with lactic
acid, at 16°C/min temperature ramp rate with the initial temperature set at 30°C was used.
Most of the lactic acid is desorbed in less than 2 minutes from the catalyst surface. As
explained above, due to a saturation effect it is hard to differentiate between peaks
specific to species formed on the catalyst surface or in the MS analysis chamber. Similar
VT-MS experiments were performed at a slower temperature ramp-rate (8°C/min), in an
attempt to ensure lower gas pressure in the ion source of the mass spectrometer and to
allow time to reach higher temperature before all the lactic acid has been desorbed from
the catalyst surface. Figure 2.8 shows the VT profiles of several ions observed during the
experiment (2.8 a) and the mass spectrum after 2.09 minutes (2.8 b). In this case the
lactic acid desorbtion is slower than in the previous case (see Figure 2.7 a), however most
of the lactic acid was removed after less than 3 minutes. The mass spectra from Figure
2.8 b looks very similar with that from Figure 2.7 b, but even peaks such as m/z = 117,
m/z = 163 and m/z = 181 amu, which were less abundant, are now completely saturated.

Consequently, the only different observed is in the rate of lactic acid desorption.

88

In order to try to differentiate between reaction on the catalyst and in the gas
phase from the ion source, reference experiments using pure dilactide, sodium lactate on
2 mole KOH/g SiOz catalyst and lactic acid on plain support have been performed, with
a temperature ramp rate of 8°C/min and initial temperature of 30°C. Figure 2.9 shows the
VT profiles for several ions (2.9 a) and a mass spectrum collected after 1.43 minutes (2.9
b), when the dilactide was impregnated on the catalyst. As observed in Figure 2.9 a, after
several minutes, most of dilactide was desorbed from the catalyst. The mass spectrum
presented in Figure 2.9 b looks similar with that of the standard dilactide, presented in
Figure 2.1 b. Figure 2.10 shows the VT profiles for the same ions considered in the
previous experiments (2.10 a) and the mass spectrum (2.10 b) collected afier 0.5 minutes,
when the catalyst was impregnated with sodium lactate. As observed from Figure 2.10 a,
little material is desorbed slowly from the catalyst surface, most of the signal being in the
noise level. The mass spectrum of the desorbed material presented in Figure 2.10 b is
similar with that obtained for dilactide. It seems that some dilactide is formed during
heating and desorbs slowly from the catalyst surface. Figure 2.1] shows the F ID (2.11 a)
and a mass spectrum collected after 3 minutes in the case of a control experiment with
lactic acid but no catalyst (just SiOZ support). The VT profiles look similar to that
obtained when catalyst was used, but the abundances of the main ions (m/z = 100, m/z =
117, m/z = 135, m/z = 145, m/z = 163, and m/z 181 amu) are significantly reduced. As
observed in Figure 2.11 b, the mass spectrum shows the presence of all main peaks
observed previously when the impregnated catalyst with lactic acid was used in the VT-

MS (see Figure 2.7 b).

89

 

 

3)
time [min]
a zoo coo 600 300 1000
number ofscans
b)
100
. 53 at 1I|5 ,
I
.0. r.
60‘
74
2|

 

O
0
vavafi'---

 

 

 

 

 

 

 

 

 

 

 

100 200 300 400 ', 500
I‘ll

Figure 2.7 VT-MS data (16°C/min ramp rate) using a 2 moles KOH/g Si02
catalyst impregnated with lactic acid: a) abundance variations versus time for ions of
interest; b) mass spectrum obtained after 2.13 minutes.

90

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8)
time [min]
1 1 m
0
i
i
Q
t
i
Q
t
O
i
i
o zoo ooo ooo 000 1000
number ofscans
b)
00
2545 1!’ 153 ,
00"
1 1‘1
60‘
19
40‘
1 7
20" '
, 215
q L_fl_!.llvllv ‘ 1%L4‘ YL' _ I v v V v v v
100 200 7 300 400 _ 500

I‘ll

Figure 2.8 VT-MS data (8°C/min ramp rate) using a 2 moles KOH/g Si02
catalyst impregnated with lactic acid: a) abundance variations versus time for ions of
interest; b) mass spectrum obtained after 2.09 minutes.

91

 

 

 

 

 

 

 

 

 

 

 

 

 

3)
time[min]
1 r
o 200 400 600 300 1000
numberofscans
b)
00
119 If so 74 as ,
aoi .
J b
g
. 63 ’
ooi
4
J
‘ 13
‘ Tl ll 1 l
I
0‘ W1,
0 100 200 300 , ooo

III/Z

Figure 2.9 VT-MS data (8°C/min ramp rate) using SiOz support impregnated
with lactic acid: a) abundance variations versus time for ions of interest; b) mass
spectrum obtained after 2.48 minutes.

92

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3)
time[min]
n r g r r 51 . . A . 1.0 . . . . 115 . 4
TIC
181
A 145
W
”In I“ .1. I“ I 118 ’
___j\ MM Mt? - A L A '117
A 100
-JLA. 91
f \ 56
./ \ 45
° . ‘fizfioro' . ruio' Vfisdo' . 'adof' ‘1o'oo
numbcrofscans
b)
100‘ ‘ T 5k
00:
“71 15 r
i +
‘0‘
‘ 1 5
2.3 P
I 13 i
a"? v.‘|"!‘ . v 1* v 1 r , - - f -
0 100 200 300 ‘. 460

Ill

Figure 2.10 VT-MS data (8°C/min ramp rate) using a 2 moles KOH/g Si02
catalyst impregnated with dilactide: a) abundance variations versus time for ions of
interest; b) mass spectrum obtained after 1.43 minutes.

93

 

2!)
time [min]
1 o 1
0 200 400 600 900 1000
number of scans
b)

 

 

 

 

 

 

 

300 ‘00
HI!

Figure 2.11 VT-MS data (8°C/min ramp rate) using a 2 moles KOH/g SiOz
catalyst impregnated with sodium lactate: a) abundance variations versus time for ions of
interest; b) mass spectrum obtained after 0.5 minutes.

94

From all these data, it is still not possible to conclude that the main peaks
observed in the mass spectra are correlated with the catalysis process from the catalyst
surface. The fast lactic acid desorption from the catalyst surface before the temperature
reaches 100°C, as well as the clear presence of characteristic ions from well-known gas-
phase processes, preclude a clear assignment of any differences seen among these
experiments to chemistry on the catalyst surface, even though many of the peaks
observed correlate well with the masses of the intermediates proposed and presented in
Scheme 2.1. The high masses of the proposed intermediates and their low volatility may
be an important barrier for electron impact-MS data acquisition. However, because the
highest barrier observed from theoretical calculations was related to the lactic acid
enolization (step 2, Scheme 2.1) and its downhill trend afterward, together with all
chemical reasoning presented before, support the proposed mechanism of lactic acid
conversion to 2,3-pentanedione presented in Scheme 2.1.

Conclusions

Based on experimental results, analyzed in the light of theoretical calculations, we
have outlined and explored a detailed pathway for lactic acid condensation to 2,3-
pentanedione. Lactate salts have been identified via DRIFTS on catalyst surfaces that
have been exposed to lactic acid vapor. H/D exchange in the a-position of the lactic acid
molecule has been observed by FTIR, D-NMR, and GC-MS, confirming the possibility of
lactate enolization under these relative mild (buffered) conditions. This process leads to
the key carbon nucleophile required for the condensation mechanism. PM3 and ab initio

calculations, augmented by the SM3 solvation model, have been used to evaluate energy

95

variations and protonation state along the reaction sequence; for the condensed-phase
situation, all energy changes were within accessible range at the reaction temperatures;
the maximum barrier is also in good agreement with the experimentally observed
activation energy for this process. Since most reaction steps involve simple proton
transfer or other common processes (dehydration, decarboxylation), the activation
barriers between intermediates are also unlikely to be abnormally high. The actual bond
making and breaking steps in this sequence will be examined in a separate ab initio study.
Thus, the proposed pathway is energetically plausible and consistent with all
observations. Its one surprising feature - a Claisen condensation of a free carboxylic acid
under buffered conditions - appears to be energetically feasible based on the calculated
energetics of lactate and its enolized form. Though specifically focused on the multistep
mechanism of lactic acid condensation to 2,3-pentanedione, this work more generally
illustrates the way in which theory and experiment can begin to dissect complex
processes such as catalysis and condensed phase reactivity. Via such detailed
mechanistic analyses of catalytic conversions, we hope to evolve toward a day when

rational design will vastly accelerate discovery of novel catalysts and reaction paths.

9999!"?

10.

11.
12.
13.
14.
15.
16.
17.

18.
19.

20.

21.

22.
23.

24.
25.

96

References

Paparizos, C.; Dolhyj, 8.; Shaw, W. G. US. Patent 4, 1988, 786, 756.

Sawicki, R. A. US. Patent, 1988, 729, 978.

Mok, W. S. L.; Antal, M. J. Jr.; Jones, M. Jr. J. Org. Chem. 1989, 54, 4596-4602.
Lira, C. T.; McCrackin, P. J. Ind. Eng. Chem. Res. 1993, 32, 2608.

Gunter, G. C.; Miller, D. J.; Jackson, J. E. J. Catal. 1994, 148, 252-260.

Gunter, G. C.; Langford, R. H.; Jackson, J. E.; Miller, D. J. Ind. Eng. Chem. Res.
1995, 34, 974-980.

Gunter, G. C.; Craciun, R.; Tam, M. S.; Miller, D. J .; Jackson, J. E., J. Catal.
1996, 164, 207-219.

Craciun, R.; Tam, M. S.; Perry, S. M.;Thiel, M. A.; Jackson, J. E.; Miller, D. J. in
Book of Abstracts at 8th SHHC Meeting, 1995, Balatonfured, Hungary, pl 04.
Craciun, R.; Tarn, M. S.; Miller, D. J .; Jackson, J. E. in Book of Abstracts at 4th
HCFC Symposium, 1996, Basel, Switzerland, p 268-269.

Craciun, R.; Tam, M. S.; Kokitkar, P. B.; Perry, S. M.; O’Donnell, M. A.;
Jackson, J. E.; Miller, D. J. in Book of A bstracts at 11th Congress on Catalysis,
1996, Baltimore, USA, p 010.

Wadley, D. C.; Tarn, M. S.; Kokitkar, P. B.; Miller, D. J.; Jackson, J. E.

J. Catal. 1997, 165, 162-171.

Evaluation made by discussions with experts from various companies involved in
flavor and food industry.

Hehre, W. J.; Burke, L. D.; Shusterman A. J .; Pietro, W.J. Experiments in
Computational Organic Chemistry: Spartan User Guide, Wavefunction, Inc.
Irvine, USA, 1994 p 44-77.

Dewar, M. J. S.; Thiel, W. J. Am. Chem. Soc. 1977, 99, 4899-4906.

Dewar, M. J. S.; Zeobisch, E. G. Healy, E. F.; Stewart, J. J. P. J. Am. Chem. Soc.
1985, 107, 3902-3909.

Stewart, J. J. P. J. Comput. Chem. 1989, 10(2), 209-220.

Stewart, J. J. P. J. Comput. Chem. 1989, 10(2), 221-264.

Aakeroy, C. B. J. Mol. Struct. (T heochem), 1993, 281, 259-267.

(a) Jackson, J. E.; Bertsch, F. M. J. Am. Chem. Soc. 1990, 112, 9085-9092;

(b) Mohamad, Master thesis, Michigan State University, East Lansing, USA,
1995.

(a) Cramer, C. J.; Truhlar, D. G. J. Am. Chem. Soc. 1991, 113, 8305; (b) Cramer
C. J.; Truhlar, D. G.; Science, 1992, 256, 213; Cramer, C. J.; Truhlar, D. G.

J. Comp. Aid. M01. Des. 1992, 6, 69.

Chuchani, G.; Martin, I.; Rotinov, A.; Dominguez, R. A. J. Phys. Org. Chem.
1993, 6, 54-58.

Chuchani, G.; Martin, I.; Rotinov, A. Int. J. Chem. Kinet. 1995, 27, 849-853.
Chuchani, G.; Rotinov, A.; Dominguez, R. A.; Martin, I. .1. Phys. Org. Chem.
1996, 9, 348-354.

Tam, M. W.; Craciun, R.; Jackson, J. E.; Miller, D. J. 1997, in preparation.
March, J. Advanced Organic C hemistry-Reactions, Mechanisms and Structures.
4th ed.; Wiley&Sons: New York, 1992, p 491-493.

26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.

38.

97

Keeffe, J. R.; Kresge, A. J. J. Phys. Org. Chem. 1992, 5, 575-580.

White, R. L.; Nair, A. Appl. Spectrosc. 1990, 44(10), 69-75.

Tam, M. S.; Craciun, R.; Miller, D. J.; Jackson, J. E.1nd. Eng. Chem. Res. 1997,
in press.

Thomas, A. F. Deuterium Labeling in Organic Chemistry, Appleton Century
Crofts: New York, 1991.

Schafer, S. L.; Barrett, W. C.; Kallarakal, A. T.; Mitra, B.; Kozarich, J. W.;
Gerlt, J. A. Biochemistry, 1996, 35, 5662-5669.

Stenhagen, E.; Abrahamsson, S.; McLafferty, F. W. Registry of Mass Spectral
Data, Vol. 1, John Woley & Sons, 19, p. 62.

Wood, G. W.; Oldenburg, E. J.; Lau, P. ; Wade, D. L. Can. J. Chem. 1978, 56,
1372-1377.

Li, Z.; Kruger, T. L.; Cooks, R. G. Org. Mass Spectrom.l982, 1 7(10), 519-523.
Harrison, A. G.; Kallury, R. K. M. R. J. Org. Chem. 1984, 49, 4993-4995.
Konopski, L. Org. Mass Spectrom. 1990, 25, 537-539.

Watson, J. T. Introduction to Mass Spectrometry, Lippin-Raven, Philadelphia.
1997, p. 37-41. -

McLafferty, F. W.; Turecek, F. Interpretation of Mass Spectra, 4th Ed. University
Science Books, Mill Valey, 1990.

Goulden, J. D. S.; Millard, B. J. Org. Mass Spectrom. 1969, 2, 893-900.

CHAPTER 3

DESIGN OF UNPROMOTED AND RARE EARTH PROMOTED MnOx/SiOz

CATALYSTS USED IN OXIDATIVE DEHYDROGENATION REACTIONS

The design of cerium and manganese oxide catalysts supported on SiOz catalysts
for use in the oxidative dehydrogenation of methyl lactate to methyl pyruvate and of 2-
ethyl-3-methyl-dihydropyrazine (obtained from 23P and ethylenediamine) to 2-ethyl-3-
methyl-pyrazine was studied. The Ce (Mn) precursor solution used for support
impregnation and the conditions during catalyst preparation determine the bulk and
surface structure of the catalysts. Catalyst structure configuration strongly influences the
catalytic activity. The great interest for the study of a MnOx-based catalyst is an
indication of their importance in chemical industry, particularly selective oxidation or
oxidative dehydrogenation of fine chemicals and more recently in emission control
processes (environmental protection technologies). Despite the considerable amount of
effort devoted to the study of Mn catalysts, the active sites of these catalysts, the role of
promoters, the mixed effect of single oxides, and the nature of any surface phases are still
the subject of intense investigations. In an effort to better understand the relation
between the structure of a solid material and its catalytic activity, the unpromoted and Ce-
promoted MnOx/SiOZ catalytic system have been investigated. The purpose of this
research is to design a new class of Mn/Ce/8102 catalysts with high catalytic activity and

selectivity in oxidative dehydrogenation processes applicable in the fine chemical

98

 

 

99

industry. The Mn/Ce/SiOz catalysts under study are characterized using various analysis
techniques. XRD is used to determine the chemical and physical nature of the species on
catalyst surfaces, identifying the crystalline phases in the catalysts by means of lattice
structural parameters and obtaining data regarding the particle size. Additional
information about the structure morphology and crystallography of the supported species
can be provided by electron microscopy. Analysis by transmission electron microscopy
(TEM) equipped with energy dispersive X-ray (EDX) analysis reveals information about
the composition and internal structure of the particle. Transmission electron microscopy
(TEM) is used in this work to provide a measure for crystalline phase(s) and to
characterize the amorphous or the highly dispersed surface phases that can not be
detected by XRD. Manganese or cerium are elements very labile to oxidation state
changes. X-ray photoelectron spectroscopy (XPS) is used to obtain information about the
chemical state of supported phases on the catalysts surface. Temperature programmed
reduction (TPR) gives additional information about the oxidation state changes on
catalyst surface and structure. The presence of paramagnetic species such as Ce3+ or
Mn” oxides detectable by EPR allow us to further probe the catalysts structure. Two
probe-reactions were used in this study: oxidative dehydrogenation of methyl-lactate to
methyl-pyruvate, and 2-ethyl-3-methyl-dihydropyrazine to 2-ethyl-3-methyl-pyrazine.
Catalyst structure and reaction conditions were optimized in order to obtain high yield

and selectivity in the desired product.

100

3.1. Rare earth promoted catalysts

The demand for various chemicals required by industry and the great technical
developments in analysis techniques and apparatus have allowed the introduction of a
wide variety of catalysts for total or selective oxidation and combustion reactions. The
main advantages offered by catalytic total oxidation are the wide range of fuel
concentration used in the process and the low temperature at which the reactions can
occur compared with the uncatalyzed ones (incineration). Among the advantages offered
by selective catalytic oxidation processes are high conversion into useful (desirable)
products and reduced waste generation. Catalytic exhaust gas treatments, especially for
automobiles, operate in a temperature range of 600-800°C. At high temperature nitrogen
oxides (NOX) are formed from nitrogen and oxygen in the air. The presence of NOx in
exhaust gases is undesirable from a pollution control point of view. Control of the
reaction at these high temperatures is difficult due to the limited flexibility in the fuel
concentration. The use of total oxidation catalysts for high-temperature processes
(>1000°C) is difficult because catalysts such as supported noble metals or transition
metal oxides are not resistant to such conditions. Therefore, is necessary to develop new
materials which may be used as catalysts or supports for these types of processes.

The development of advanced heterogeneous catalysts involves several important
activities. These include the development of sophisticated preparation procedures that
can be used to design specific surface structures, guided by examination of the active
sites as well as promoter/support and promoter/transition metal oxide interactions in the

active catalytic phases. Mixed metal oxides are important catalysts that are used widely

101

for selective oxidation of fine chemicals. The surface of the mixed metal oxide exhibits
two metal ions in addition to 02' ions or -OH groups present at the surface. The
oxidation process which takes place on the active site of the catalyst surface is
schematically presented in Figure 3.1 [3].

When the metal oxide catalysts consist of only one metal species, the oxidation
process is represented as one side of the cycle represented in Figure 3-1. When two metal
oxide species are present on the catalyst surface, the process is represented by the entire
cycle from the figure. The oxidation process occurs at the surface where the oxygen

atoms have limited reactivity (related to metal M,).

0.

H20 . e'
Partial oxidized
product

Figure 3.1 Schematic representation of the catalytic cycle for a selective
oxidation process which occurs on a mixed metal oxide catalyst surface; M, and M2 are
two different transition metals (x,y = number of the electrons exchanges in the redox
process by the M, and M2, and n and m are the oxidation numbers for the M, and M2) [3].

 

Activated by the metal ions, the oxygen ions react with the reactants to give water and the
partially oxidized product rather than the total oxidation process. The surface sites

involved in the process are reoxidized indirectly and not by 02 from the gas phase. The

102

02 reacts instead with the surface site associated with the second metal, M2, and thus.
oxygen is transported as ions through the bulk of the catalyst from the second site to the
first to deoxidize the surface, with the electron compensation to complete the cycle [3].

At present, the catalyst systems used in these processes are based on noble metals
which are expensive and susceptible to poisoning. Transition metal oxide catalysts have
been suggested as suitable substitutes for noble metals even though they are not as active
and are also subject to poisoning. However, some industrial catalytic processes have
been designed and exploited successfully using transition metal oxides as catalysts rather
than noble metal (e.g. Wacker oxidation using Pt/V205 catalyst [3]). The advantages of
transition metal oxides are their low cost and better resistance to chlorine poisoning. By
improving their stability, selectivity and activity using suitable promoters, transition
metal oxides may become good replacements for noble metal-based catalysts [1 -3].

Noble Metals. Noble metals typically catalyze total oxidation processes.
Catalytic incineration over noble metal catalysts has been widely used for decades. The
advantages of these catalysts are given by the high catalytic activity and selectivity.
Platinum, palladium and rhodium are the most commonly used active phases for catalytic
combustion applications [4-6]. A desirable characteristic of the noble metal catalysts is
their high resistance to sulfur poisoning compared with the transition metal oxide
catalysts [7]. Highly dispersed Pt and Pd catalysts are easily prepared using a number of
support materials like silica or alumina.

Oxidation over noble metals is considered to be a structure sensitive reaction in
which the surface configuration of the catalyst strongly influences the activity, selectivity,

reaction pathway and rate [8-12]. Briot et al. [8-9] have attributed the activity of

103

supported platinum and palladium catalysts to the high specificity of surface oxygen on
large crystallites. For these types of catalysts, an activity enhancement was observed at
low temperature, close to the catalyst ignition temperature. Catalysts with high noble
metal dispersion (small crystallites) exhibit high catalytic activity [10]. No oxygen
pressure dependence on the reaction rate was found. In the case of methane oxidation,
the reaction order was found to be around one in methane with little influence of oxygen
concentration on the reaction rate. The dissociation of the methane or oxygen-methane
complexes at the catalyst surface was proposed to be the rate-limiting step in the
mechanism [12].

Transition Metal Oxides. Oxidation over solid metal oxide catalysts has been
widely studied during the last years. The main advantage over noble metal catalysts is
the low cost of the raw material. By choosing a well defined composition of the metal
oxide in the catalyst, an oxidation activity close to that of the noble metal catalysts can be
obtained. It was found that the formation of nitrogen oxides can be avoided using metal
oxide catalysts [13].

Oxygen can be activated by interaction with the catalyst surface. Coordination or
incorporation of oxygen into the oxide lattice may be an important step in catalyst
activation. Lattice incorporation leads to a more strongly bound oxygen than in the case
of simple physical adsorption. The lattice oxygen was found to be more active in the case
of selective oxidation processes whereas the surface oxygen was more important in
complete oxidation reactions [14]. Participation of lattice oxygen in complete oxidation
reactions increases with temperature. Wise et al. [15] observed that the ratio of the rate

constants corresponding to the reactions of lattice oxygen and adsorbed oxygen increases

with temperature from 0 at 500°C to 0.3 at 800°C for methane oxidation on a LaFeO3
catalyst. At higher temperatures lattice oxygen interactions dominate the oxidation
process. Many studies have been focused on the suitability of different types of metal
oxides for catalytic incineration, trying to rationalize their catalytic activity. Comparison
studies of the catalytic activities in terms of conversion for a variety of oxides such as
CrzO3, C0203 [16], Mn203 [17] V205 [18], or CuO [19] led to the conclusion that many
of these oxides are potentially good catalysts for total oxidations.

Mixed oxides with perovskite structures XYO3, (where X and Y are different
transition metals), have revealed high activity in total oxidation reactions [20]. Their use
as catalysts was first reported in 1970, by Meadowcroft [21]. Their well defined and
stable structures make them suitable for oxidation catalysts. Perovskite electronic
structures are comparable with those of the transition metals. Seiyama [20] summarized
in a review article the important aspects of total catalytic oxidation reactions on various
perovskite oxides, like sorption and desorption of oxygen, H2 oxidation and total
oxidation of hydrocarbons. Many of the pervoskite structures incorporated lanthanides
(La, Ce, Pr) and/or alkaline earth metals (Ba, Sr, Ca) in combination with transition or
noble metals. Structures like La,.xerBO3, La,_xerMnO3, and La,_xerCoO3 [22-24] are
some of the most recent perovskite-type catalysts prepared and used successfully in total
oxidation reactions.

From the material presented above we can conclude that a great number of high-
temperature-stable, complex oxide materials are promising catalysts for catalytic

oxidation processes applicable in the large scale chemical industry.

105

Supports. Another important element in the structure and composition of a
catalyst is the support. The chemical nature of supports may play an important role in
catalyst’s activity. Supports are mostly used to improve thermal and chemical stability
while decreasing the amount of the expensive active component required to make a
catalyst. The surface area and chemical nature of the support play major roles in the
behavior of the catalysts. The most commonly used supports are based on silicates and
aluminates, but often less resistant, low surface area materials such as MgO or Ti02 are
used to improve the selectivity of a catalyst or as a catalyst site itself. The geometry and
the dimension of the particle used as support are very important factors in the catalyst
activity. The characteristics needed for a support to be used in high-temperature reactions
such as combustion can be summarized as follows: 1) high thermal stability and
mechanical strength; ii) no chemical reaction between support and the catalytic
component; iii) high volume throughput [25]. It is hard to find an ideal material that
meets all the requirements for an excellent support. Many times, additives or promoters
are used to improve the properties of the supports used for catalyst preparation. Silica is
one of the most commonly used support materials; at high temperature and in the
presence of oxygen, silica exhibits sintering phenomena. y-Ale3 is another support that
shows resistance to the sintering process compared to silica. However, at very high
temperature (1000°C) the y-A1203 phase is transformed into a-A1203 which has a lower
surface area [26]. Many studies on alumina based catalysts show loss in surface area due
to sintering of particles by solid-state diffusion. Consequently, the thermal treatment

used for the support material also affects a catalyst’s properties [27].

106

Addition of a promoter may stabilize the support by inhibiting the sintering
process. Promoters can be divided into two different classes: structural and textural.
Structural promoters lead to a transformation of the catalysts surface configuration which
affects catalyst activity. Textural promoters have less influence on the structure of the
catalyst but improve the textural properties of the catalyst like surface area, pore volume
size, and mechanical resistance of the catalyst surface. Alkaline earth metals (Ba, Ca, Sr)
are well known as additives with consequences on the textural properties as well as
mechanical resistance of the catalysts. Lanthanide oxides are known to be used as both
textural and structural promoters, influencing the thermal resistance and the surface
configuration of the catalyst. Another way to promote a catalyst is to introduce lattice
flaws into the structure. These kinds of promoters are called electronic promoters,
substances with semiconductor character. These materials will accept or lose electrons
from conduction band. Impurities in the lattice of the catalyst can thus increase its
catalytic activity.

Dispersion of the active component, loading and surface structure are important
parameters in designing a proper catalyst. The main chemical effect of a catalyst in total
or selective oxidation reactions is to force them to completion, yielding water, carbon
dioxide and nitrogen as final products (assuming only C, O, H, N are in the feed) and the
main product, respectively. Metal oxide based catalysts are of particular interest for these
types of processes. Procedures such as temperature treatment, or additives such as
promoters or activators may strongly influence the surface structure of the catalyst and its
catalytic activity. Manipulation of these factors during catalyst preparation offers control

over activity, selectivity and resistance of the catalyst to various poisons or inhibitors.

107

Tailoring a catalyst surface by controlling the dispersion and/or particle size of the active
phase, crystalline structure and oxidation state of the metals on the surface, ultimately
gives control over the catalytic process.

Promoter eflect. Rare earth additives have been employed extensively as textural
and structural promoters in supported metal catalysts [28-30]. Cerium oxide in particular
has been widely used for these purposes [31-39]; its promoter effect is given by the
nonstoichiometric crystal structure [40-41], oxygen storage capabilities [42-45], redox
properties [46] and very good thermal resistance [47-53]. Cerium promotion enhances
the catalytic activity of transition metals and/or transition metal oxide-supported catalysts
for CO oxidation and NOx reduction because of its ability to form oxygen-deficient
oxides under reducing conditions [49,40]. Due to their importance in a variety of
catalytic applications, the structures of supported cerium catalysts have been the focus of
many investigations [29-59]. Miki et al. [42] have used XRD to examine the structure of
a 20 wt.% CeOz/AIZO3 catalyst oxidized and reduced at high temperature (900°C). High
temperature reduction of the catalysts led to CeAlO3. Reoxidation of the reduced
catalysts had no effect on LaAlO3 whereas CeAlO3 was reoxidized to CeOz. Shyu et al.
[57] reported XPS, Raman spectroscopy, and TPR results for a series of Ce/Ale3
catalysts. They concluded that Ce existed as a CeAlO, precursor, small Ce02 crystallites,
and large C602 particles on Ce/A1203 catalysts. Graham et al. [35] reported that
lanthanum promotion increases the dispersion and the range of reversible reducibility of
almnina supported CeOz. High temperature led to a loss of support surface area. The
presence of promoter such as CeOz or mixed Ce-Zr oxides considerably reduces this

effect, but it can’t be totally avoided. During this process, noble metals deposited on

108

these supports are encapsulated by support materials leading to loss in catalytic activity.
Recently, it was shown using X-ray diffraction analysis that a substantial fraction (25%)
of Pd (Pt) becomes unavailable for catalysis as a result of encapsulation [60].

Spectroscopic measurements combined with ab initio calculations have shown
that in the case of supported rare earth oxides on alumina, modification of the electronic
structure of the lanthanide cation occurs, affecting the catalytic properties of the catalyst.
Charge redistribution over an alumina surface implies a modification of the acid-base
characteristics of the active sites [61]. Lanthanide cations are known to have strong
interactions with the support which results in an electronic density redistribution. Metal-
support interactions influence both the electronic state of the metal and the surface
morphology. The A1203 support provides a weak ligand field, favoring the high-spin
state of the metal and a change in the oxidation state [62].

The crystallographic and solid-state properties of the lanthanide oxides are largely
limited to the so-called sesquioxide stoichiometry (M203), the only stable composition
ordinarily observed for these materials. Exceptions from this structure have been
observed in the cases of Equ and SmOx. Dioxides having a face-centered cubic
structure (fluorite type structure) are known in the case of Ce, Pm and Th but only the
cerium stoichiometry is more stable than the corresponding sesquioxide structure, mostly
due to the stability of the Ce4+ oxidation state. CezO3 can be prepared only with
difficulty, by extended reduction of Ce02 at high temperature or by vacuum
decomposition of certain salts. Ce203 oxidizes easily in air to the dioxide. In addition,
the oxide system of Ce and Tb contain LnOx with 1.5 < x < 2.0. In all three systems

mentioned above as characteristic for lanthanide oxides, the stable phases show variable

109

Ln3+/Ln4+ ratios together with the appropriate number of 02' ions needed to achieve
electrical neutrality. Rare earth sesquioxides exhibit characteristic polymorphism and can
exist in one or more distinct crystalline modifications, depending on the radius of the
tetravalent metal ion and the temperature for oxide preparation and quenching. Pertinent
parameters of the rare earth cations and stable rare earth oxide phases are available in the
literature [63]. X-ray diffraction patterns of most of these materials are also available for
comparative purposes [64]. Various oxide modifications may be further distinguished by
analysis of IR bands from 200-700 cm'1 range. All anhydrous rare earth oxides are z750/o
ionic having a strong basic character (important property for catalytic purposes).
Lanthanide oxides undergo dehydration/rehydration processes. This cyclic
process was used to prepare anO3 having highly reproducible catalytic activities for
alcohol dehydrogenation/dehydration and for double bond isomerizations of olefins [65].
For some reactions, catalytic activities are independent of electronic or magnetic
properties of the oxides and are determined by relative acid-base properties or by the
surface structure of the catalysts whereas for other types of reactions the paramagnetic
nature, lattice oxygen mobility and cation variable valence play important roles,
governing the catalytic behavior. For instance, paramagnetic rare earth oxides have been
shown in several studies to be effective catalysts for ortho-para hydrogen conversion [66].
The rate is strongly influenced by temperature which affects the paramagnetic properties
of the catalyst materials. Diamagnetic oxides such as YZO3 or Lu203 can be activated by

thermal treatment which apparently generates the paramagnetic ions Y2+ and Lu2+ [67].

110

Preparation methods. The precursors and/or methods used for preparation can
strongly influence the surface structure of a catalyst and consequently its catalytic activity
and selectivity. The most widely used precursor for cerium impregnation is the Cey-
nitrate solution utilized often for preparation of three-way catalysts for automotive
exhaust systems [68]. After calcination, a well dispersed cerium phase is observed on the
A1203 support, at low cerium loading [69-70]. Miki et a1. [42] prepared a cerium catalyst
by incipient wetness impregnation and compared its surface structure with a cerium
catalyst prepared by the sol-gel technique, which uses Ce3+'nitrate dissolved in ethylene
glycol and aluminum tri-isopropoxide. Different Cer surface structures were obtained
which led to modified catalytic behavior.

Ion exchange is an alternative method used for preparation of catalysts [70-71].
The method is based on the electrostatic adsorption of the ions at a pH lower than the
isoelectric point (IEP) characteristic for each support when the surface is positively
charged and anion adsorption can occur. Positive and negative charged support surfaces
can adsorb oppositely charged ions from the solution. The exact chemistry depends on
support, pH and species in solution. The pH value of the solution will act as a surface
charge selection switch, favoring the deposition of a given ion. Using this method,
molybdenum was successfully well dispersed on Ti02 [72] and A1203 [73] supports.
Difficulties in controlling the equilibrium of ionic species present in the precursor
solutions at various pH values preclude the general utilization of this technique.

For metal impregnation, Kurnmer et al. [74] proposed a method which involves
thermal transport from the bulk phase of a metal foil to a dispersed phase on the support

surface. The metal concentration on such catalysts is low but highly dispersed. In this

[11

way, cerium was deposited on the support by incipient wetness impregnation using Cey-
nitrate solution, before the transition metal impregnation. Such catalysts used Ce02 as an
additive favoring the metal transport (in this case, Pt) and dispersion, leading to a very
complex CeOz-metal surface structure.

Dufour et al. [75] used organometallic compounds for rhodium impregnation of a
silica support. The interaction mode of the Rh(n3-C3H5) with the oxide support varies
according to the nature and the degree of hydroxylation of the support. Different grafted
rhodium species were found on the support surface after impregnation. Didillon et a1.
[76] extended the use of organometallics for other transition metals (group VIII-metals)
and found that highly dispersed metal phases can be successfully anchored onto low-
surface area A1203. Usmen et a1. [77] used a La3+ pretreated alumina as a support to form
a well dispersed cerium phase during impregnation. They found that well dispersed Lay'-
phases interact with CeOz leading to a better cerium and platinum dispersion. Higher
cerium dispersion leads to the retention of more reducible oxides such as CeAlO3 with
consequences on the oxygen storage capacity of the catalyst [5 7].

Van Hengstum et al. [78] and later Baiker et al. [18] have developed a synthetic
method for catalyst preparation that involves alkoxide grafting using a metal-organic
compound. Generally, an alkoxide grafiing reaction may be written as follows:

Support-OH + M - (OR),( —> Support - O - M (OR),,,, + R-OH (3-1)

The chemistry of this method involves the formation of a covalent bond between
the metal center and the support surface, followed by the elimination of an alkoxide

ligand from the metal center (M) as an alcohol. Excellent dispersion was obtained for

112

vanadium impregnation on different supports (SiOz, A1203 or TiOz), when vanadium
tetra-isobutoxide solution was used as a precursor.

For example, silica-supported cerium oxide catalysts of various contents (less than
11% Ce) were prepared by anchoring cerium acetylacetonate in organic media and
studying their surface structure. The resulting CeOz particles were 1-3 nm in diameter
after calcination in oxygen at 673 K. The highly dispersed cerium oxide obtained by
using the cerium acetylacetonate precursor shows a shift to higher XPS binding energies
for Ce3d and 0,5, increased CO stretching in CO-Ce4+ terminal groups, and a blue shift of
the band gap measured by diffuse reflectance in comparison with catalysts prepared using
the classical preparation by incipient wetness using cerium nitrate [79].

The use of a “hot grafting” technique refers to a grafiing process using a metal-
alkoxide precursor under reflux condition and elevated temperature. This technique can
be an alternative way to improve the interaction between the precursor and support
surface [80]. The high cost as well as the sensitivity to air and water of these metal-
organic compounds [81], confer to the grafting method a disadvantage for large scale
catalyst impregnation. It was found that in some cases, the presence of ethanol in the
system as an exchange ligand influences the reactivity of cerium alkoxide toward grafting
with the hydroxyl groups from the support surface [82].

Recently, lanthanide alkoxide catalysts were successfully used in such organic
reactions as synthesis of vitamins and steroids [83-84], being promising catalysts for
heterogeneous catalysis processes applied in organic synthesis. If these alkoxides are
prepared from chiral alcohols, the catalysts obtained may be used to perform asymmetric

synthesis, largely applied in the drug industry [85]. A monolayer dispersion of the

113

impregnated lanthanide-alkoxide, fixed on a solid support, may confer on the catalysts
increased selectivity and activity in these types of organic synthesis.

3.2 Supported MnOx catalysts used in oxidative dehydrogenation reactions

One of the goals in this study has been to develop new synthetic methods that
allow good control over the structure of the catalytically active element, in this case a
metal oxide material. As mentioned above, variation in the unpromoted or promoted
metal oxide bulk and surface structure would be reflected in the catalysts’ properties such
as stability, activity and selectivity. A heterogeneous catalytic process takes place by
adsorption of the reactants on the active site of catalyst surface, where the bonds in the
molecule are weakened and broken followed by the reorganization of the molecule by
forming a different bonding scheme, leading thus to product. The function of the solid
catalyst is to provide an energetically favorable pathway, with low activation energy, for
the formation of the desired product. Consequently, a catalyst should consist of small
particles acting as active sites, with high dispersion on a high surface, porous material. In
many cases, metal oxides play the role of the active site in a catalyst, being able to
disperse well on inert supports and act selectively in the catalytic process.

Manganese oxides have been successfully used as the active catalyst phase mostly
in oxidation processes. This catalytic activity was attributed to manganese’s capability of
forming oxides having different oxidation states (MnOz, Mn203, Mn3O4, or MnO) and
storing oxygen in its crystalline lattice. Unsupported or supported MnOx was found to be
an active catalyst in several oxidation processes such as CO [86-89], ethylene [90],
methanol [91] oxidation, NOx [92-93], H202 [94], O3 [95] decomposition, oxidative

coupling of methane [96-98], selective catalytic oxidation of NH3 [99-101], ethylbenzene

114

[102], hydropyrazine to pyrazine oxidative dehydrogenation [103], selective catalytic
reduction of NO with NH, [99-101, 104-105], CO [106] and C2H4 [91] hydrogenation,
HZS/Hz sulfidation [107], and Hg waste removal [108]. As observed fi'om the literature
search, a wide variety of catalytic reactions have been tested on MnOx-based catalysts.
Due to its labile oxidation state, manganese is capable of performing either as an
oxidation agent being reduced (relation 3-2) or as a reducing agent being oxidized
(relation 3-3) acting as the active component in the redox process.
Mn4+ + e' —) Mn3+ + e' —> Mn2+ (3-2)
Mn2+ - e' —> Mn3+ — e- —> Mn4+ (33)

Structural characterization of unsupported or supported MnOx catalysts identify
the presence of MnOz or mixed MnOz/Mn203 phases as active components [105, 107,
109]. The ratio between the Mn02 and Mn203 phases is a function of manganese loading
and calcination temperature of the catalyst during preparation [107, 109]. Manganese
interaction with the support or other components present on the catalyst significantly
influences its oxidation state. Thus, the anp XPS binding energy showed a 0.57 $0.04
eV shift when manganese interacted with copper and formed a mixed spinel structure of
Cu,+an2_xO4 [101]. The presence of promoter such as lanthanum influences the MnOx
structure and dispersion. It was found that lanthanum interacts with manganese during
thermal treatment of the catalyst to form mixed Mn-La oxides, leading to a better MnOx
dispersion than in the case of the unpromoted catalyst. Lanthanum promotion led to a
more robust MnOx catalyst, which minimized the formation of catalytically less active

Mn3O4 species [102]. In some cases, the unsupported MnO, catalyst was more active

115

than the supported one, due to the absence of manganese interaction with the support and
availability to the reaction [87, 103]. For unsupported MnOx catalyst it was observed that
for NO catalytic decomposition, high reaction temperatures are required in order to get
significant reaction rates (> 500°C). At high calcination temperatures (above 600°C),
anO3 releases more oxygen than Mn3O4 phase, being more catalytically active.
Because these reactions occur at high temperatures and the reaction pathway is
temperature dependent (NO can be converted either to N20 or to N02), the structural
modification of MnOx should be carefully examined. MnOx loading and dispersion of
supported catalysts influence the catalytic activity as well. Previous data for CO
oxidation have shown that the specific activity increases with decreasing concentration of
the MnOx, except for very low concentration (loadings) in which activity decreases
anyhow. Increased dispersion of the oxide on the support surface led to a significant
increase in the activity, a fact attributed to the increased quantity exposed to the reaction
feed [86, 105]. MnOx dispersion was found to be a function of manganese precursor and
loading, preparation method, and post-preparation thermal treatment. It was found that
compared with Mn-nitrate, a Mn-acetate precursor led to a highly dispersed anO3 phase
on a y-Ale3 support [109]. This was attributed to ionic interactions with acidic and basic
surface -OH groups from alumina. Mn-nitrate yields comparable dispersion with that
obtained from Mn-acetate precursor, for loading below 1 wt% Mn. Large manganese
oxide crystals are formed at higher loadings. Multiple step treatment with low
concentration Mn-nitrate solution does lead to a better MnOx dispersion. In the case of a

supported MnOx catalyst prepared with Mn-nitrate, higher loading of Mn is

116

recommended, in order to observed the influence of the calcination temperature during
preparation process. Structural transformations of MnOx were observed as a function of
calcination temperature. Thus, for a series of MnOx supported on y-A1203 catalysts with
various Mn-loadings, the B-MnOz phase was predominant for low Mn-loading and
calcination temperature during preparation. At high loadings and calcination
temperatures, the a-Mn203 phase is dominant [105, 107].

The ability of manganese to form MnO or Mn(OH)2 species makes it a useful
catalyst in hydrogenation reactions as well [91, 106]. MnO can be formed only after
calcination at very high temperatures (above 900°C), when XRD patterns start to be
observable [110]. Its catalytic activity is a function of the support and dispersion. The
MnO phase was found to act as a promoter, prohibiting the hydrogenation of C2H4 and
C3H6 to CZHG and C3H8, respectively, in CO hydrogenation processes [106].

As suggested by many previous experiments, MnOx catalysts can be used in
combination with other transition metals which can form either mixed oxides (spinel type
or pervoskite) or oxides acting as promoters for manganese (such as lanthanum, yttrium,
cerium, etc.). From previous studies, it is known that manganese can interact to form
spinel-type structures with copper [101], or with W03 [98], TiOz, A1203, [111], CeOz
[111, 112], or pervoskite-type structures with La, Sm and Ba [113-114]. Manganese has
an excellent oxygen storage capability, so if it is combined with cerium, the unit cell
parameters increase with the presence of mixed MnCer phase, forming a catalyst with
“superstoichiometric” oxygen concentration [115]. EPR characterization of these MnOx

catalysts supported on CeOZ and A1203 have shown a high degree of ionicity of the well-

117

dispersed MnOx layer, with well defined sites on CeOz and isolated sites on A1203 (poor
MnOx dispersion). Integration of the EPR signals corresponding to paramagnetic 02'
lattice species from mixed MnCer oxide allowed an evaluation of its oxygen storage
capacity [116]. Another important aspect in catalysts structure characterization of
materials with catalytic activity in oxidation processes is the mobility of lattice oxygen.
Migration of oxygen on the surface is important due to the repeated oxidation-reduction
cycles of the catalyst. Combination of two element interacting with each other, deposited
on a given support and used as catalyst in catalytic oxidation processes exhibits different
oxygen mobility in comparison with a single component catalyst. Manganese oxides are
typical berthollide compounds with very labile lattice oxygen. It was found that placed in
the proximity of CeOz, the oxygen mobility from the MnOx structure is strongly affected.
In a Mn-Ce mixed oxide, Ce provides oxygen to Mn at low temperature and, contrarily,
withdraws oxygen at elevated temperatures (> 500°C). Thus, cerium improves the
activity of MnOx in oxidation processes at lower temperature and decreases its activity at
high temperature, the active component in this case being Ce02 which takes oxygen from
MnOx [117]. Having all this information in mind, the following research objectives have

been suggested.

19.
20.
21.

22.
23.
24.
25.

26.
27.

28.

29.

30.
31.

118

3.3 References

Moffat, J. B. Theoretical aspects of heterogeneous catalysis, Van Nostrand
Reinhold, New York, 1990, p. 256-426

Thomas, J. M. Angew. Chem. Int. Ed. Engl. 1994, 33, 913-937.

Gates, B. C. Catalytic Chemistry, John Wiley & Sons, Inc. New York, 1992,
p. 3 10-421 .

Hicks, R. F. Chem. Eng. Sci. 1990, 45(3), 2647-2651.

Worley, S. D.; Rice, C. A.; Mattson, G. A.; Curtis, C. W.; Guin, J. A. J. Chem.
Phys. 1982, 76, 20-24.

Taylor, K. C. in Catalysis and Automotive Pollution Control by Crucq, A.;

F ressnet, A. 1987, Elsevier, Amsterdam.

Johnson, D. W. Jr. J. Catal. 1977, 48, 87-97.

Briot, P. Appl. Catal. 1990, 59, 141-152.

Briot, P.; Primet, M. Appl. Catal. 1991, 68, 301-314.

Hick, R. F. J. Catal. 1990, 122, 295-306.

Baldwin, T. R.; Burch, R. Appl. Catal. 1990, 66, 359-381.

Otto, K. Langmuir, 1989, 5, 1364-1369.

Ismagilov, Z. R.; Kerzhentsev, M. A. Catal. Rev. -Sci. Eng. 1990, 32(1,2), 51-103.
Arakawa, T.; Tsuchi, Y. S.; Shiokawa, J. J. Catal. 1982, 74, 317-322.

Wise, 11.; McCarty, J. G. Catal. Today, 1990, 8, 231-248.

Prasad, R.; Kennedy, L. A.; Ruckenstein, E. Combust. Sci. Technol. 1980, 22,
271-275.

Dwyer, F. G. Catal. Rev. 1972, 6(2), 261-291.

Kifenski, J .; Baiker, A.; Glinski, M.; Dollenmeier, P.; Wokaun, A. J. C atal. 1986,
101, 1-11.

Stubenrauch, J. ; Vohs, J. M. J. Catal. 1996, 159, 50-57.

Seiyama, T. Catal. Rev. - Sci. Eng. 1992, 34(4), 281-300.

F omasiero, P.; Balducci, G.; DiMonte, R.; Kaspar, J .; Sergo, V.; Gubitosa, G.;
Ferrero, A.; Graziani, M. J. Catal. 1996, 164, 173-183..

Jaecnicke, S.; Chuah, G. K. Ber. Bunsenges. Phys. Chem. 1992, 96(1), 1-9.
Arai, H. Appl. Catal. 1986, 26, 265-276.

Zhang, H. M.; Teraoka, Y.; Yamazoe, N. Appl. Catal. 1989, 6(1,2), 155-162.
Zwinkels, M. F. M.; Jaras, S. G.; Menon, P. G.; Griffin, T. A. Catal. Rev. - Sci.
Eng. 1993, 35(3), 319-358

Beguin, B.; Garbowski, E.; Primet, M. J. Catal. 1991, 127, 595-604.

Tijburg, I. I. M.; Geus, J. W.; Zandbergen, H. W. J. Mater. Sci. 1991, 26,
6479-6486.

Alvero, R.; Bemal, A.; Carrizosa, I.; Odriozola, J. A. Inorg. Chim. Acta, 1987,
140, 45-48.

Xie, Y. C.; Qian, M. X.; Tang, Y. Q. China-Japan-US. Symposium on
Heterogenous Catalysis Related to Energy Problem, 1982, B. 10C,China.
Yang, J. Y.; Swartz, W. E. Spectrosc. Lett. 1984, 17, 331-332.

Baiker, B. G.; Clark, N. J. in Studies in Surface Science and Catalysis, Delmon,
B. Grange, P. Jacobs, P. A. Poncelet, G. Elsevier, Amsterdam, 1987, 31, 455-62.

32.

33.

34.

35.

36.

37.
38.

39

40.
41.

42.

43.
44.
45.
46.

47.
48.

49.
50.
51.
52.
53.
54.
55.
56.
57.
58.

59.
60.

61.

62.
63.

119

Kieffer, R.; Kiennemann, A.; Rodriquez, M.; Bemal, S.; Rodriquez-Izquierdo, J.
M. Appl. Catal. 1988, 42, 77-86.

Shyu, J. 2.; Otto, K.; Watkins, W. L. H.; Graham, G. W.; Belitz, R. K.; Gandhi,
H. S. J. Catal., 1988, 114, 23-28.

Hay, C. M.; Jennings, J. R.; Lambert, R. M.; Nix, R. M.; Owen, G.; Rayment, T.
Appl. Catal. 1988, 37, 291-304.

Graham, G. W.; Schmitz, P. J.; Usmen, R. K.; McCabe, R. W. Cat. Lett. 1993, 17,
175-184.

Masuda, K.; Kawai, M.; Kuna, K.; Kachi, N.; Mizukami, F. in Preparation of
Catalysts VEd. Poncelet, G.; Jacobs, P. A.; Grange, P.; Delmon, B.; Elsevier,
Amsterdam, 1991, 229-238.

Sugiyama, S.; Matsumura, Y.; Moffat, J. B. J. Catal. 1993, 139, 338-350.
Bunluesin, T.; Cordatos, H.; Gorte, R. J. J. Catal. 1995, 15 7, 222-226.
Dulamita, N. Rev. Rom. Chim. 1985, 36(11), 1050-1051.

Steinberg, M. Anal. Appl. Rare Earth Mater. Nato Adv. Study Inst. 1972, 263-270.
Perrichon, V.; Laachir, A.; Bergeret, G.; Frety, R.; Toumayan, L.; Touret, 0.; J.
Chem. Soc. Faraday Trans. 1994, 90(5), 773-781.

Miki, T.; Ogawa, T.; Haneda, M.; Kakuta, N.; Ueno, A.; Tateishi, S.; Matsuura,
S.; Sato, M. J. Chem. Phys. 1990, 94, 6464-6467.

Yao, H. C.; Yao yu, Y. F. J. Catal. 1984, 86, 254-265.

Su, E. C.; Montreuil, C. N.; Rothschild, W. G. Appl. Catal. 1985, 17, 75-86.
Loof, P.; Kasemo, B.; Keck, K. E. J. Catal. 1989, 118, 339-348.

Haneda, M.; Mizushima, T.; Kakuta, N.; Ueno, A.; Sato, Y.; Matsuura, S.;
Kasahara, K.; Sato, M. Bull. Chem. Soc. Jpn. 1993, 66, 1279-1288.

Schaper, H.; Doesburg, E. B. M.; VanReijen, L. L. Appl. Catal. 1983, 7, 211-220.
Cordatos, H.; Bunluesin, T.; Stubenrauch, J. ; Vohs, J. M.; Gorte, R. J. J. Phys.
Chem. 1996, 100, 785-789.

Yu-Yao, Y.F.; Kummer, J.T. J. Catal. 1987, 106, 307-312.

Chojnacki, T.; Krause, K.; Schmidt, L.D. J. Catal. 1991, 128(1), 161-185.
Oudet, F .; Courtine, P.; Vejux, A.J. J. Catal. 1988, 114, 112-120.

Yang, J.K.; Swartz, W.E. Spectrosc. Lett. 1984, 17, 331-335.

Cordatos, H.; Ford, D.; Gorte, R. J. J. Phys. Chem. 1996, 100, 18128-18132.
Zafiris, G. S.; Gorte, R. J. J. Catal. 1993, 143, 86-91.

Yamaguchi, T.; Ikeda, N.; Hattony, H.; Tanabe, K. J. Catal. 1981, 67, 324-330.
Zafiris, G. S.; Gorte, R. J. J. Catal. 1993, 139, 561-567.

Shyu, J. 2.; Weber, W. H.; Gandhi, H. S. J. Phys. Chem. 1988, 92, 4964-4970.
Ledford, J. S.; Houalla, M.; Proctor, A.; Hercules, D. M. J. Phys. Chem. 1989, 93,
6770-6777.

Dulamita, N.; Pop, A.; Corbeanu, V. Studio, 1984, 29, 3-11.

Graham, G. W.; Jen, H.-W.; Chun, W.; McCabe, R. W. Catal. Lett. 1997, 44,
185-187.

Capitan, M. J.; Centeno, M. A.; Malet, P.; Carrizosa, 1.; Odriozola, J. A. J. Phys.
Chem. 1995, 99, 4655-4660.

Tauster, S. J.; Fung, S. C.; Garten, R. L. .I. Am. Chem. Soc. 1978, 100, 170-175.
Komer, R.; Ricken, M.; Nolting, J. J. Solid State Chem. 1989, 78, 136-147.

64.
65.
66.
67.
68.

69.

70.

71.
72.
73.

74.
75.

76.

77.

78.

79.

80.
81.
82.

83.
84.
85.

86.
87.

88.
89.
90.
91.
92.

120

Powder Difl'raction File, Inorganic Phases, JCPDS, 1983, p. 18, 91, 346, 534
Rosynek, M. P.; Cat. Rev. - Sci. Eng. 1977, 16(1), Ill-154.

Baron, K.; Selwood, P. W. J. Catal. 1973, 28, 422-426.

Aries, J. A.; Selwood, P. W. J. Catal. 1974, 33, 284-288.

Kubsh, J. E.; Rieck, J. S.; Spencer, N. D. in Catalysis and Automotive Pollution
Control 11, ed. Crucq, A. Elsevier, Amsterdam, 1991, 125-138.

Diwell, A. F.; Rajararn, R. R.; Shaw, H. A.; Truex, T. J. in Catalysis and
Automotive Pollution Control 11, ed. Crucq, A. Elsevier, Amsterdam, 1991,
139-152.

Burch, R.; F lambard, A. R. in Preparation of Catalysts III by Poncelet, G.;
Grange, P.; Jacobs, P. A. Els. Sci. Publ. Amsterdam, 1983, 311-320.

Brunelle, J. P. Pure Appl. Chem, 1978, 50(9,10), 1211-1229.

Luthra, N. P.; Cheng, W. C. J. Catal. 1987, 107, 154-160.

VanVeen, J. A. R.; deWitt, H.; Emeis, C. A.; Hendriks, P. A. J. M. J. Catal. 1987,
107, 579-582.

Kummer, J. T.; Yu-Yao, Y. F. .1. Catal. 1987, 106, 307-312.

Dufour, P.; Houtman, C.; Santini, C. C.; Basset, J. M.; Hsu, L. Y.; Shore, S. G.
J. Am. Chem. Soc. 1992, 114, 4248-4257.

Didillon, B.; El Mansour, A.; Candy, J. P.; Basset, J. M.; Le Peltier, F.;
Boumonville, J. P. in Preparation of Catalysts Ved. Poncelet, G.; Jacobs, P. A.;
Grange, P.; Delmon, B. Elsevier, Amsterdam, 1991, 717-723.

Usmen, R. K.; Graham, G. W.; Watkins. W. L. H.; McCabe, R. W. Catal. Lett.
1995, 30, 53-63.

Van Hengstum, A. J .; Van Ommen, J. G.; Bosch, H.; Gellings, P. J. Appl. Catal.
1983, 5, 207-217.

Bensalem, A.; Bozon-Verduraz, F.; Delamar, M.; Bugli, G. Appl. Catal. 1995,
121, 81-93.

Smith, P. D.; Klendworth, D. D.; McDaniel, M. P. J. Catal. 1987, 105, 187-193.
Namy, J. L.; Souppe, J.; Kagan, H. B. J. Org. Chem. 1984, 49, 2045-2049.
Mehrotra, R. C .; Kapoor, P. N.; Botsoara, J. M. Coord. Chem. Rev. 1980, 31,
67-91.

Kirk, D. N.; Slade, C. J. J. Chem. Soc. Perkin Trans, 1980, 1, 2591-2569.
Evans, D. A.; Hoveyda, A. H, J. Am. Chem. Soc. 1990, 112, 6447-6449.
LeBrun, A.; Namy, J. L.; Kagan, H. B. Tetrahedron Letters, 1991, 32(21),
2355-2358.

Mooi, J.; Selwood, P. W. J. Am. Chem. Soc. 1952, 74, 2461-2464.

Nohman, A. K. H.; Duprez, D.; Kappenstein, C.; Mansour, S. A. A.; Zaki, M. I. in
Preparation of catalyts V, ed. Delmon, B.; Jacobs, P. A.; Poncelet, G. Elsevier,
Amsterdam, 1991, p. 617-626.

Katz, K. Adv. Catal. 1953, 5, 177-183.

Boreskov, G. K. Adv. Catal. 1964, 15, 285-294.

Dmuchovsky, B.; Freeks, M. C.; Zienty, F. B. J. Catal. 1965, 4, 577-.

Baltanas, M. A.; Stiles, A. B.; Katzer, J. R. Appl. Catal. 1986, 28, 13-33.
Karlsson, H. T.; Rosenberg, H. S. Ind. Eng. Chem. Process Des. Dev. 1984, 23,
808-815.

93.
94.
95.
96.
97.

98.
99

100.
101.

102.
103.

104.
105.

106.
107.
108.
109.
110.

111.
112.

113.
114.
115.
116.

117.

121

Yamashita, T.; Vannice, A. J. Catal. 1996, 163, 158-168.

Kanungo, S. B. J. Catal. 1979, 58, 419-435.

Nishino, A. Catal. Today, 1991, 10, 107-112.

Moggridge, G. D.; Rayment, T.; Lambert, R. M. J. Catal. 1992, 134, 242-252.
Burch, R.; Chalker, S.; Squire, G. D.; Tsang, S. C. J. Chem. Soc. Faraday Trans.
1990, 86, 1607-1614.

Wu, J.; Li, 8.; Niu, J.; Fang, X. Appl. Catal. A, 1995, 124, 9-18.

Il’Chenko, N. 1.; Golodets, G. I. J Catal. 1975, 39, 57-72.

Singoredjo, L. S.; Korver, R. B.; Kapteijn, F. Moulijn, J. A. Appl. Catal. B, 1992,
1, 297-302.

Wollner, A.; Lange, F.; Schmetz, H.; Knozinger, H. Appl. Catal. A, 1993, 94,
181-203.

Craciun, R.; Dulamita, N. Catal. Lett. 1997, 46(3, 4), 229-234.

Akiyama, T.; Enomoto, Y.; Shibamoto, T. J. A gric. Food Chem. 1978, 26(5).
1176-1179.

Markvart, M.; Pour, V. Int. Chem. Eng. 1975, 15, 546-551.

Kapteijn, F.; Singoredjo, L.; vanDriel, M.; Andreini, A.; Moulijn, J. A.; Rarnis,
G.; Busca, G. J. Catal. 1994, 150, 105-116.

Longya, X.; Quigxia, W.; Yide, X.; Jiashng, H. Catal. Lett. 1995, 31, 253-266.
Strohmeier, B. R.; Hercules, J. .1. Phys. Chem. 1984, 88, 4922-4929.

Cavallaro, S.; Bertuccio, N.; Antonucci, P.; Giordano, N. J. Catal. 1982, 73,
337-348.

Kapteijn, F .; van Langeveld, D.; Moulijn, J. A.; Andreini, A.; Vuurrnan, M. A.;
Turek, A. M.; Jehng, J. M.; Wachs, I. E. J. Catal. 1994, 150, 94-104.
McClune, W. F. Powder difi'ractionfile, Inorganic phases, International center
for diflraction data, Swarthmore, 1983, p. 279, 599, 730, 731.

Baltanas, M. A.; Stiles, A. B. Katzer, J. R. J. Catal. 1984, 88, 362-373.
Abi-Aad, E.; Bennani, A.; Bonnelle, J. P.; Aboukais, A. J. Chem. Soc. Faraday
Trans. 1995, 91, 99-104.

M. Shelef, L. P. Haack, R. E. Soltis, J. E. DeVries and E. M. Logothetis,

J. Catal. 1992, 137, 114-126.

Ledford, J. S.; Kim, Y. M.; Houalla, M.; Proctor, A. Hercules. D. M. Analyst,
1992, 117, 323-327.

Ivanova, A. S.; Dzisco, V. A.; Moroz, E. M.; Nosckova, S. P. Kinet. Katal. 1986,
27(2), 428-434.

Baltanas, M. A.; DeCaria, S. J.; Katzel, J. R.; Dybowski, C. Acta Chim. Hung.
(Eng) 1985, 118(4), 285-295.

Imamura, S.; Shono, M.; Okamoto, A.; Ishida, S. Appl. Catal A, 1996, 142,
279-288.

CHAPTER 4

SYNTHETIC METHODS FOR PREPARATION OF
CeOz-PROMOTED S102 CATALYSTS

4.1 Introduction

Cerium oxide has been employed extensively as a catalyst or as a textural and
structural promoter for supported metal or metal oxide catalysts [1-28]. Recently, in a
review article Trovarelli [29] summarized many of the catalytic properties and
characterization studies of Ce02 reported during the last decades. Generally, the
structural promotion effect was attributed to cerium’s ability to form crystalline oxides
with lattice defects which may act as catalytic active sites [3-4], whereas its textural
promotion effect is given by the excellent thermal and mechanical resistance which Ce02
confers to the catalysts [2-7]. Ce-based catalysts are commonly utilized in oxidation
reactions; their use in automotive catalytic converters, for example, render them valuable
catalysts [29]. Detailed information about studies on Ce-based catalysts was given in
Chapter 3 (see 3.1 paragraph).

This part of the study is focused on the structural transformations of S102-
supported Cer species when various cerium precursors and thermal treatments were
used for catalyst preparation. DRIF TS and cross polarization (CP) 29Si MAS-NMR were
used to obtain information about support - cerium precursor interactions. The crystalline
nature and dispersion of CeOz on the Si02 support were investigated by transmission

electron microscopy - energy dispersive X-ray analysis (TEM-EDX), XRD, and

122

123

temperature programmed reduction (TPR). Additional information about the bulk and
surface structure of supported CeOz was provided by electron paramagnetic resonance
(EPR) and X-ray photoelectron spectroscopy (XPS) analysis.

4.2 Experimental

Catalysts preparation. Catalysts were prepared by the impregnation of a SiOz
support (surface area = 300m2/g) with Ce4+-ammonium nitrate in water (Mallinckrodt Co)
and Ce4+-methoxyethoxide 18-20% in methoxyethanol (Gelest Co.) precursors. The
support consists of finely ground (<23O mesh) silica-gel (Davison Chemical Co.), which
was calcined in air at 500°C (800°C) for 24 h, prior to impregnation. Catalysts that
originated from the Ce-nitrate precursor are designated as CeqN and those from Ce-
alkoxide as Cqu, where q is the Ce/Si atomic ratio (q = 8x10'2). The actual cerium
concentrations in the catalysts were confirmed by inductively coupled plasma (ICP)
analysis. The results showed a i 10% error when compared to cerium concentration
calculated from the amount of precursor added during preparation. The impregnation
techniques employed for catalysts preparation were: i) incipient wetness, using Ce“-
nitrate (CeqN) and ii) “hot grafting” using the Ce4+-methoxyethoxide solution (Cqu)
monitored under inert (Ar) atmosphere with continuous mixing for 48 h, under reflux at
80-85°C. After impregnation, all catalysts samples were dried at 125°C for 16 h followed
by calcination in air at 500°C or 800°C.

BET Surface Area. Surface area (Sa) measurements were performed using a
Quanta-Chrome Quantasorb Jr. Sorption System. Approximately 0.1 g of catalyst was

outgassed in N2 at 165°C for 12h, prior to adsorption measurements. The measurements

124

were made using N2 partial pressures in He of 0.05, 0.08, and 0.15 (N2 surface area =
0.162 nm2) at liquid N2 temperature (77 K). The estimated error for S2, results is i 5%.

Solid state ”Si MAS-NMR. Solid state 29Si MAS-NMR spectra were obtained at
room temperature using a Varian VXR-400S spectrometer with an Oxford cryomagnet
generating a magnetic field of 9.395 T and operated at a Sun workstation network with a
V-NMR operating system Version 4.1. The Fourier transform of free induction decay
(FID) was observed using a quadrature detection at 79.4559 MHz at a spinning rate of
4000 rot/min. The acquisition time was long enough to ensure full 29Si nucleus
relaxation for accurate quantitative data analysis. The cross polarization technique can be
used to obtain a better resolution for the silica peaks. The technique was developed and
used successfully to characterize a silica gel surface by Maciel and Sindorf [30].
Following their work, numerous studies have shown the reliability of this technique in
studies on various type of silica [31-34]. The 29Si-‘H cross polarization restricts detection
to silicon nuclei that are positioned close to protons or near the surface [30]. Typical 29Si
resonance occurs when the magnetic moment and the spin of silica nucleus are
antiparallel, resulting in a (-g) parameter. About 0.5 g catalyst powder was introduced
into the spectrometer holder for analysis. The spectrometer was tuned at the minimum
signal possible for both silica-29 and proton. Silica peaks were deconvoluted using the
program from the V-NMR spectrometer workstation (for details about NMR analysis,
please refer to Chapter 1).

DRIFT S analysis. Diffuse reflectance spectra were acquired using a Mattson-
Galaxy FTIR-3020 instrument with a diffuse reflectance attachment from Spectra-Tech

Inc. The catalyst powders (20-25 mg) were put into the sample holder and introduced

125

into the DRIFTS attachment. The spectra collected are in Kubelka-Munk units versus
wavenumber. IR spectra were acquired with resolution of 4.0 cm'1 over a 400-4000 cm'1
wavenumber range. For details about DRIF TS analysis please refer to Chapter 1.

Transmission electron microscopy (T Ell/0. The TEM bright field images, EDX,
and electron diffraction patterns images were recorded with a Jeol IOOCX II instrument
equipped with an X-ray analyzer for EDX analysis, at accelerating voltages of 120V.
Catalyst samples were deposited on a carbon-dusted copper grid coated with a holey film,
and introduced into the analysis chamber on a beryllium grid holder. The catalyst above
a hole of the holey film was chosen for analysis, in order to avoid interference from the
film structure. EDX spectra were collected from the same spot where the TEM electron
diffraction patterns have been recorded.

X-Ray Difiraction (XRD). X-ray powder diffraction patterns were obtained with a
Rigaku XRD diffractometer employing Cu K01 radiation (2» = 1.541838 A) and operated
at 45 kV and 100 mA. Diffraction patterns were obtained using a scan rate of 0.5

deg/min with 1/2 mm slits. Powdered samples were mounted on glass slides by pressing

the powder into an indentation on one side of the slide. The mean crystallite size ((1) of
CeOz particles was determined from XRD line broadening measurements using the
Scherrer equation (1-3) [3 5]. Semi-quantitative X-ray diffraction data were obtained by
comparing Ce02<111>/SiOZ peak ratios measured for catalyst samples with those of
Ce02 and SiOz physical mixtures of known concentration. This method assumes that
cerium addition does not disrupt the Si02 structure, and consequently, does not affect the

intensity of the SiOz line. The error in this method was estimated to be i 20%.

126

Temperature-programmed reduction (T PR). TPR profiles for the CeqN and
Cqu catalysts were obtained using a classical U-shaped quartz reactor connected to a
thermal conductivity detector (TCD) for H2 consumption analysis. The temperatures
were measured with a Ni-Cr-Ni therrnoelement placed directly in the catalyst bed. A
reducing agent mixture of 10% H2 in N2 was passed over 0.1 g catalyst powder, at 30
mL/min flow rate. The temperature interval considered for analysis is from 25°C to
950°C (10°C/min heating rate). The TPR profiles are presented as H2 consumption
(arbitrary units) versus temperature (°C).

Electron Paramagnetic Resonance (EPR). Powder catalyst samples were
analyzed in quartz sample tubes at -l 60°C using a Varian E4 spectrometer. The magnetic
field was set at 3200 Gauss with a modulation frequency of 100 kHz. The g-factor (gs)
for cerium samples (B5) was determined by comparison with the field value at resonance
for a reference (BM), in this case the diphenyl-picryl-hydrazyl radical (DPPH, gm, =
2.0036). The reference was placed together with the catalyst samples into the quartz tube
for accurate g-value determinations. Accordingly, the value for gS was evaluated based
on the relation (2) given below [36]:

g. = g...- IBM/B. <44)

X-ray photoelectron spectroscopy (XPS). Surface analyses of the catalyst samples
were carried out using a Perkin-Elmer Physical Electronics XPS 5400 spectrometer
equipped with a hemispherical analyzer. Instrument control, data collection and
manipulation were performed with an Apollo 3500 workstation, running a PHI-XPS

version 3.0 software. The standard Mg Ka X-ray source was used for all samples

127

analysis and was operated at 15 kV, 20 mA, and 300 W. The catalysts were mounted on
sample stubs using double sided tape and placed into the analysis chamber for XPS
analysis (base chamber pressure was 2x10'9 torr). The binding energies for catalyst
samples were referenced to the Sizp peak (103.4 eV). XPS binding energies were
measured with a precision of i 0.2 eV, or better. The reported XPS ICe3d/1812p values are
the average of three consecutive analyses.

4.3 Results and Discussion

Catalyst texture. BET measurements for the neat support and cerium catalysts
(see in Table 4.1) indicate a decrease of surface area for the S102 support from 300 mZ/g
(precalcined at 500°C) to 210 mZ/g after calcination at 800°C. This process is attributed
to the collapse of the silica porous structure at elevated temperature, process well known
from the literature [37-3 8]. After cerium impregnation, the surface area of the CeqN and
Cqu catalysts calcined at 500°C, remains unchanged, since CeOz is known to have no
significant porosity which may influence the texture of the promoted catalyst unless a
special preparation technique is employed [39]. For catalysts calcined at 800°C, a
decrease in the surface area was observed; however, the magnitude of the decrease is very

much attenuated due to the textural promoter effect of the CeOz.

128

Table 4.1 BET, semi-quantitative XRD, and XPS analysis data for Ce-promoted
SiOz catalysts

 

 

 

 

 

 

 

catalySts BEZT C902 Size 0/0 C802 [cad/[sup X
[m /g] [nm] Crystalline Intensity
$0? 300 - - -
SiOzC 210 - - -
CeqN" 303 9.9 89 0.4
Cech 268 10.0 97 0.2
Cqu° 300 2.5 29 6.7
Cqu° 272 2.8 43 4.6

 

 

 

 

 

 

 

a - the error in this method was estimated to be i 20%; b - samples calcined at
500°C; c - samples calcined at 800°C.

Cerium precursor/SiOz support interaction. Differences in the chemical nature of
the interaction between cerium precursors and SiOz support were used to explain why
more Ce02 crystalline phase was formed in the case of Ce8N catalysts than in the case of
the Ce8A catalysts. For Ce8N catalysts, the weak physical interactions between cerium
nitrate and silica support led after calcination at 500°C to an easy formation of large CeOz
particles with a high degree of crystallinity. For C e8A catalysts, covalent bonds between
cerium alkoxide and surface hydroxyl groups from silica support led after calcination at
500°C to small CeOZ crystallites and a low degree of crystallinity. To support these
statements, diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) and
29Si solid state nuclear magnetic resonance analysis were performed in order to

understand the interaction between cerium precursors and 8102 support.

129

(c)

(b)

| ‘ ' l | |
4000.0 3000.0 2000.0 1000.0

Wavenumber [cm'l]

Figure 4.1 DRIFTS spectra of standard materials: a) Si02 support; b) CeOz;
c) Ce(OH)4.

DRIFTS spectra of 8102 (Figure 4.1 a), Ce02 (Figure 4.1 b) and Ce(OH)4 (Figure
4.1 c) are presented as reference for interpretation of the DRIFTS spectra obtained for the

CeOz/SiOz catalysts prepared by grafting and incipient wetness impregnation.

130

Table 4.2. Tentative infrared bands assignments for cerium treated SiOz catalysts

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0rd Wavenumber IR Band Assignments References
# [cm"]
1 3740 free -OH stretch on 8102 42, 43
2 3700-3000 -OH stretch hydrogen bonded 40, 42, 43
3 3540 -OH stretch hydrogen bonded from 41

Ce(OH)4

4 2986 C-H stretch from -CH2- (R) 40, 41
5 2928, 2830 C-H stretch from CH3- (R) 40, 41
6 1638 H20 adsorbed on SiOz surface 40, 41
7 1514 -OH bend (in plane) 44
8 1473 -OH bend 44
9 1456 C-H bend from -R 40, 41
10 1382 O-C-O stretch 44
11 1292, 1076 C-O stretch 40, 44
12 1110-1000 Si-O-Si stretch 42, 43
13 970-920 Si-O-R stretch 42, 43
14 910-830 Si-OH stretch 42, 43
15 853 Si-O-Si bend 42-43
16 811 -NO stretch from Ce (NH4)2(NO3)4 40, 41
17 746 -NO bend from Ce(NH4)2(NO3)4 40, 41

 

 

 

 

 

The tentative bands assignments are listed in Table 4.1. Characteristic for the
silica spectrum (a) is the presence of the 3740 cm'l band, which is attributed to the free
hydroxyl groups present on the surface [42-43]. The other bands observed in the

spectrum are due to the residual molecules of C02 and H20 (1638 cm") adsorbed on the

131

silica surface, and the bands around 1200-1000 cm'1 are attributed to Si-O stretch and
bend from silica [44]. Due to the sensitivity of the DRIF TS analysis, several well defined
bands for Ce02 (Figure 4.1 b) were observed which are otherwise hard to see using
simple FTIR analysis. As in the case of the silica support (a), the free and coupled
hydroxyl groups from the surface, can be easily distinguished. The bands observed at
1556 cm", 1473 cm", 1299 cm", and 1034 cm'l were attributed in a recent study made
by Verduraz et al. [44], to the carbonate and carboxylate species present on the CeOz
surface. The band at 1556 cm"l corresponds to a carboxylate species, 1473 cm'I and 1299
cm'1 to a bidentate carbonate and 1034 cm'l to a monodentate carbonate surface species.
The Ce(OH)4 spectrum shown in Figure 4.1 0 includes a broad band characteristic of the
OH stretch (3500-3000 cm") involved in intra- or intermolecular hydrogen bonds, more
intense bands characteristic of C02 and H20, bands characteristic of carboxyl or carbonyl
surface species and bands characteristic of deformation of O-H bond (the broad band
between 1500 - 1200 cm") [44-47].

Figure 4.2 shows the DRIF TS spectra of the CeqN catalysts, respectively, dried at
120°C (b) and calcined at 500°C (c) for comparison with the spectrum of neat SiOz (a),
respectively. The spectrum from Figure 4.2 b reveals the formation of Ce(OH)4 species
on SiOz after drying the catalyst at 120°C. The process may be favored by the weak
acidity of the silica surface, which favors the formation of hydrated particles from Ce“-
nitrate solution. Simple water solvation and drying of cerium nitrate compounds led to
identical DRIFTS spectra as the starting compounds. Together with the formation of the

Ce(OH)4 on the support, the silica band at 3690 cm'1 corresponding to the free hydroxyl

132

groups disappears and a broad band around 3500-3000 cm'1 is observed, corresponding to
hydrogen bonded hydroxyl groups. This fact suggests that the interaction between the
cerium precursor and silica support has only a physical nature, e.g. dipole-dipole

interactions and hydrogen bonds.

(C)

(b)

(a)

A «t- ~ 1 l 1
4000.0 3000.0 2000.0 1000.0

Wavenumber [cm'l ]

Figure 4.2 DRIFTS spectra of CeqN catalysts: a) S102 support; b) CeqN catalyst
dried at 125°C; c) CeqN calcined at 500°C.

133

After calcination at 500°C (Figures 4.2 c), the IR spectrum of the catalyst became similar
to the spectrum of the pure silica support. The IR band at 3740 cm'1 corresponding to
free surface hydroxyl groups is almost completely regenerated. Some remaining bands
from bending -OH groups are still observable at 1520 cm'I and 1100 cm".

Figure 4.3 shows the DRIFTS spectra for the Cqu catalysts acquired during
impregnation and thermal treatment. The grafting process involves a chemical interaction
(covalent bond) between the precursor and support. The interaction of the cerium
alkoxide with the support may take place either through a reaction with the hydroxyl
groups involved in intramolecular hydrogen bonds or reaction with the free hydroxyl
groups from the surface leading to the formation of Si-O-Ce-Alkoxide bonds [41, 48]. If
the Ce-OH phase is formed in the solution, the precursor-support interaction could be an
intermolecular interaction through hydrogen bonds. After grafting impregnation DRIFTS
spectra shown in Figure 4.3 b-d indicate the disappearance of the band at 3740 cm'1
characteristic of the free hydroxyl groups from the Si02 surface. The presence of the IR
bands at 2837 cm’l characteristic of C-H stretching from cerium alkoxide are clearly
observed for the grafted (Figure 4.3 b), dried (Figure 4.3 c) and calcined at 200°C (Figure
4.3 (1), when all the solvent, water or other physisorbed compounds are removed from the
silica surface. This observation supports the idea of a covalent bond interaction between
the cerium alkoxide and silica surface. In addition, no Ce(OH)4 phase formation was
observed during grafting, as the DRIF TS spectrum of the dried catalyst indicates.

After calcination, the DRIFTS spectra of the catalyst indicate the regeneration of

the band corresponding to the free hydroxyl group from the support surface (1740 cm'l

134

band) and the band which can be assigned to a surface carboxylate species formed on

Ceo2 (1523 cm") [44, 47].

M
(d)
(C)
(b)

(a)

 

l I l 1

4000.0 3000.0 2000.0 1000.0

Wavenumber [cm'l ]

Figure 4.3 DRIFTS spectra of Cqu catalysts: a) SiOz support; b) Cqu catalyst
dried at 25°C; c) Cqu dried at 125°C; d) Cqu calcined at 200°C; e) Cqu calcined at
300°C; 1) Cqu calcined at 500°C.

Cerium-oxygen bands, which appear in the 500-250 cm'I region, are hard to be observed

due to the strong absorption bands of the Si02 support in this region [41].

I35

2981 solid state NMR spectra and the deconvolution of the peaks into the main
components for the neat SiOz, CeqN, and Cqu catalysts after impregnation with the
different cerium precursors and dried at 125°C, are presented in Figure 4.4 a-c. Figure
4a’, 4b’ and 4c’ show the deconvolution result spectra for the Cqu, CeqN, and SiOz
catalysts. Various types of silica from the catalyst samples were identified based on
chemical shift [in ppm] and assignments from previous work [31, 49-51]. The chemical
shifts at which silicon atoms show resonance are in the range from -85 to -110 ppm.
Sheme 4.1 presents the silicon configuration and the corresponding 29Si NMR chemical

shift observed, reported in the literature [31-33, 51].

Scheme 4.1 Silica structures and the corresponding 29Si NMR shifts [31-33]

(A) (B) (C)
-—-O\ ,0— —O\ ’O—H H —O\ ,0—H
/S'. /S‘\ /S'\
‘1’ ‘1’ ‘1’ ‘1’ ‘1’ ‘1’
= 403,7 ppm 8 = -100.3 ppm = -95.6 ppm
Trifunctional branching Middle branching
(D) (E)
H --O O—H H —O O—H
\S.’ \S.’
/ I\ / l\
1 0 <3 <1
H I H H
8 = -88.5 ppm 5 = ‘76-0 PPm

End Branching

I36

In each spectrum shown in Figure 4.4, silica shows characteristic peaks at -108.3
ppm (A), -100.7 ppm (B), -95.6 ppm (C), and -88.5 ppm (D) corresponding to silicon

atoms with the configurations presented in Scheme 4.].

 

 

(b)
(b’)

_'—

 

 

 

 

——

,30 -9'0' 7 400 ' - 4'16 -120
[ppm]

Figure 4.4 29Si NMR spectra of the catalysts dried at 125°C: a) Cqu; b) CeqN;
c) SiOz in comparison with the deconvoluted spectra for each samples (a’, b’ and c’)

137

Table 4.3 presented below, shows the ratio values of silica peaks from B, C, and
D, relative to peak A, calculated on the basis of deconvoluted peak area of each silica

peak.

Table 4.3 29Si NMR peak intensity ratios and XPS Iliad/Imp ratio for silica in B,
C, and D configuration relative to A (see Scheme 4. 1)

 

 

 

 

Catalysts [cad/Isa.) XPS ”Si NMR Peak Ratio
Intensity BIA C/A D/A
SiOz - 2.43 0.07 0.05
CeqN dried at 125°C 7.9 2.27 0.87 0.27
Cqu dried at 125°C 1.2 1.84 0.57 0.04

 

 

 

 

 

For the CeqN catalyst dried at 125°C, the B/A, C/A and D/A have very close
values with those obtained in the case of the neat Si02 support (ex. 2.27 compared with
2.43). Consequently, no chemical interaction occurred between the CeqN precursor and
the support during the impregnation and drying processes. In the case of the Cqu dried
catalyst, the B/A value (1.84) is very different from that obtained for SiOz. There is less
change in the case of the C/A and no significant change in the D/A values. This
observation can be explained by considering that after impregnation with Ce-Alk solution
and drying, cerium alkoxide is grafted on silica and many of the surface hydroxyl groups
(trifunctional silica groups, Scheme 1B) are bonded to the precursor, as shown in the
reaction (4-3):

0
80-85 (7

Support-OH + Ce-(OR)4 —> Support - O - Ce-(OR)3 + R-OH (4-3)

reflux

138

where R = -O-CH2-CH2-O-CH3.

After calcination in air at 500°C, the alkoxide part is removed from the Cqu
dried catalyst, while in the case of CeqN dried catalyst only the water has been removed
with formation of CeOz. The S/N ratio of the NMR spectra and the resolution decreases
as the proton abundance near silica atoms decreases. As a consequence, for catalysts
calcined at 500°C and 800°C the quality of the spectra is not sufficient for a proper
interpretation. The low S/N ratio observed for the spectrum of SiOz calcined at 800°C is
due to the lower concentration hydroxyl groups (mostly free hydroxyl groups, so low
proton concentration) from dehydration process which lead to a lower 29Si signal
enhancement by proton decoupling. The 29Si spectra for CeqN and Cqu calcined at
500°C and 800°C (not shown) look similar with those obtained for SiOz support,
indicating no clear chemical interaction between cerium oxide and silica. The appearance
of peak E at ~76.0 ppm for both types of catalysts indicates a reorganization in the silica
structure probably with participation of the cerium oxide formed on the Si02 surface.
The formation of Ce-Si can be excluded because the chemical shifts observed is these
type of interaction are very large (> 100 ppm) [51]. For both type of catalysts, calcination
at 500°C regenerates the surface free hydroxyl groups from SiOz, phenomena observed
by DRIFTS as well.

Ce02 crystallinity and dispersion. Information about the crystallinity of CeOz

deposited on SiOz support was obtained from TEM and XRD analyses.

I39

    
 

 

(‘0‘) 2
«200.»

I l I
20.0 40.0 (10.0

 

 

 

 

Degrees 2 8
b)
fi I l I ' l r l I
8 -i _,////—1
P '
:5“
§ 6 _ _
5" .
B—
O ,
.Q 1’
E 4 4 x. a
a I
8
N
x 1
«3
3 ,
a 2 -I 1 ’ —1
x C
/’.l
0 6‘] v I l V T ' T_ f I ' I
0 5 [) L5 20 25 30

Catalyst concentration, [wt%] CeO2 on SK):

Figure 4.5 Semiquantitative XRD data: a) XRD pattern of CeOZ/SiOz physical
mixture with different Ce02 loading; b) corresponding calibration curve.

140

The TEM electron diffraction patterns for CeqN and Cqu catalysts were
obtained at a M = 83,000 (M = magnification). In the case of CeqN catalyst, the electron
diffraction micrograph shows the presence of crystalline material in the sample,
containing diffraction patterns which can be attributed to crystalline CeOz formed on
Si02 support after calcination. X-ray analysis performed on the same spot confirms the
presence of Ce and Si as the main elements in the sample. The electron diffraction image
obtained in the case of Cqu catalysts shows only diffuse rings with no clear diffraction
patterns, presumably due to amorphous materials. The amorphous phase observed in this
electron diffraction pattern corresponds to both SiOz and Ce02 material. EDX analysis
confirmed the presence of cerium from amorphous CeOz formed on the Cqu catalysts.

The results from semiquantitative analysis were presented in Table 4.1. These
data were evaluated based on results from XRD analysis of CeOz/Sioz physical mixture,
with various cerium loading. The XRD patterns for the physical mixture samples are
presented in Figure 4.5 a. The Ce<11 l> (20 = 28.5°) have been chosen for our
calculations. The intensity for the SiOz peak was estimated by deconvolution of the XRD
pattern. The calibration curve obtained is presented in Figure 4.5 b (standard deviation
for each point was evaluated at 5%). A good linearity was obtained for the cerium
loading (CeOz crystalline %) versus Ce<111>/Si02 XRD intensity ratio. As a
consequence, the evaluation of Ce02 crystalline phase for the CeqN and Cqu catalysts
can be considered accurate. Figure 4.6 presents the X-ray diffraction patterns of the SiOz
support (Figure 4.6 a) in comparison with those obtained for the CeqN (Figure 4.6 b and

4.6 c) and Cqu catalysts (Figure 4.6 d and 4.6 e). Diffraction patterns specific to CeOz

l4]

crystalline phase [52] can be identified in both Cqu and CeqN catalyst samples. Particle
size evaluation and semiquantitative XRD analysis for the CeqN and Cqu catalysts are

reported in Table 4.1 (columns 3 and 4, respectively).

Ce02

<lll>

(e)

 

(d)

(C)

 

(b)

 

 

l | I I
20.0 30.0 40.0 50.0

Degrees 2 9

Figure 4.6 XRD patterns of: a) SiOz; b) CeqN calcined at 500°C; c) CeqN
calcined at 800°C; d) Cqu calcined at 500°C; e) Cqu calcined at 800°C.

142

The CeOz particle size for cerium supported catalysts, evaluated from line
broadening calculations, were found to be larger for CeqN than for Cqu catalysts. This
fact was attributed to the difference in the catalyst impregnation method. As observed
from the impregnation study (DRIFTS and 29Si NMR), the grafting method allows a
stronger interaction between precursor and support, as compared to the incipient wetness
method, where the interaction has a physical nature (hydrogen bonding or dipol
interaction) [37-38]. A crystalline CeOz phase is formed during calcination at high
temperatures (>400°C), so this process is slowed down in the case of Cqu catalyst by
the strong interaction between the Ce-methoxyethoxide with Si02 support, affecting the
grows of Ce02 crystalline particles.

The data from semiquantitative XRD analysis indicate the percentage of cerium
present as crystalline CeOz in the catalyst. Crystalline Ce02 accounts for most of the
cerium (>90%) in the CeqN catalysts, but for less than 30% in the case of Cqu catalysts;
consequently the rest of the cerium is present either as an amorphous phase or as small
crystallites (see Table 4.1). Relating these results to those obtained from TEM electron
diffraction, it can be stated that the rest of the cerium from Cqu catalysts is present as
amorphous cerium oxide. Furthermore, simple calcination of cerium nitrate is known to
lead to crystalline CeOz, in agreement with the results observed for CeqN catalysts [21,
53]. For Cqu, the amorphous CeOz formation can be attributed to cerium alkoxide
polymerization during the grafting impregnation process, which favors formation of
amorphous and/or small Ce02 particles after calcination. Previous studies have shown

that only heating at temperatures higher than 600°C, under H2 atmosphere, can lead to

143

formation of amorphous Cer/SiOZ catalysts by partial reduction of Ce4+ from small

Ce02 crystallites to Ce3+ [37, ,58, 59]. TPR profiles obtained for CeqN and Cqu

catalysts are presented in Figure 4.7.

879

 

H2 consumption [a.u.]

 

(a)

l l l A , n

230 450 600 350 950
Temperature [°C]

Figure 4.7 TPR profiles of: (a) Cqu and (b) CeqN catalysts calcined at 500°C

In the case of Cqu catalysts (Figure 4.7 a), the TPR profile indicates two major
peaks around 485°C and 610°C. It is known that for a low surface area neat Ce02

sample, the TRP profile contains two major peaks, one at lower temperature (around

I44

500°C), attributed to the reduction of surface oxygen from Ce02 and one at higher
temperature (around 800°C), attributed to the removal of bulk oxygen from CeOz
structure [12]. The shape of the TPR profile is a function of temperature treatment
(removing surface carbonates), CeOz surface area and particle size [39]. A decrease in
the particle size of Ce02 will result in a shift of the reduction peak maximum to lower
temperatures. Based on the TPR profile observed in Figure 4.7 a which shows a
maximum around 610°C, it can be concluded that Cqu catalysts contains small CeOz
particles. In this case, the shifiing of the reduction peak may be attributed to the presence
of amorphous CeOz, as well. The TPR of CeqN catalysts (Figure 4.7 b) shows a typical
profile for CeOz similar with those reported in the literature [12, 39, 55]. The intense
peak around 879°C from the TPR profile of CeqN catalysts indicates that most of the
Ce02 deposited on the Si02 support is in the form of large crystalline particles. The
small changes in surface area observed afier cerium impregnation, and the calcination to
500°C prior analysis (low surface carbonates content), led to negligible modifications in
the TPR profiles of CeqN and Cqu catalysts.

XRD line broadening calculations showed formation of smaller CeOz particles on
Cqu catalysts as compared with CeqN, indicating a better cerium dispersion on the SiOz
support. The XPS ICe3d/ISi2p intensity ratio can provide an additional rough estimation of
the cerium dispersion, a large value being indicative of high dispersion [56]. For the
dried (Table 4.3, column 2) and calcined (Table 4.1 column 5) Cqu catalysts, ICe3d/15i2p
has a value around 7, lower in the case of the calcined to 800°C. The corresponding

load/13,2p values for the CeqN catalysts are much lower (around 1). These XPS results are

145

in agreement with the dispersion data from XRD and TPR analyses, indicating a better
cerium dispersion in the case of Cqu catalysts in comparison with the CeqN catalysts.

XRD particle size, semiquantitative XRD and XPS analyses indicate that
calcination at 800°C led to lower cerium dispersion and higher Ce02 crystallinity (see
Table 4.1). This fact suggests that the small crystallites formed on the silica surface are
mobile enough to get together and form large crystalline particles.

Ce02 bulk and surface structure. EPR and XPS analysis were performed in order
to obtain additional information about the structure (cerium oxidation state) of cerium
deposited on the Si02 support after calcination at various temperatures. Figure 4.8 shows
the EPR spectrum of a CeOz standard material (Figure 4.8 a) in comparison with the EPR
spectra of CeqN (Figure 4.8 b and 4.8 c) and Cqu (Figure 4.8 d and 4.8 e) catalysts.
The EPR spectrum of CeOz (Figure 4.8 a) shows a peak at g = 1.952 (g i), attributed in
the literature to Ce3+-type defects present in a CeOz crystalline structure [57-61]. Non-
stoichiometric cerium oxide crystals formed during the calcination step, are most
probably responsible for the appearance of defects in the crystalline structure. Previously
obtained EPR spectra of CeOz showed two types of g,, signals corresponding to two
different Ce3+ sites, form A (g,, = 1.932), present only at low temperature, and form D (g,,
= 1.935), dominant after calcination at temperatures higher than 300°C [57-59].
Apparently, in the case of CeqN and Cqu catalysts calcined at 800°C, only cerium in
form A is present, due to the high temperature treatment (_>.SOO°C) applied during the

preparation process.

146

g =2.025 g, =10“
' g =1.952 (:3+ in 0:02

 

 

g #1932 Ce form A

 

 

superoxide 02
(a) on surface vacancies .

—w—_ '“VV-V‘V

| | l I l
2800.0 3000.0 3200.0 3400.0 3600.0

Magnetic Field [Gauss]

Figure 4.8 EPR spectra of: a) CeOz; b) CeqN calcined at 500°C; c) CeqN
calcined at 800°C; d) Cqu calcined at 500°C; e) Cqu calcined at 800°C.

The presence of EPR signals specific to bulk Ce3+ defects is associated with the signal

specific to 02' species (g = 2.011 and g = 2.036) from the CeOz crystals, however, the

I47

intensity of both signals are reduced relative to the standard signals (Figure 4.8 a). The
signals at g r = 2.011 and g” = 2.036 were attributed to 02’ species bonded to surface Ce4+
ions from non-stoichiometric ceria [57-60]. At these g values, oxygen nuclei are located
in the CeOz structure, equidistant from the surface. With higher temperature, other 02'
species (g,, = 2.025) are formed in relation to the appearance of new vacancies or other
defects into the cerium oxide structure [58]. The large percentage of crystalline CeOz
phase formed in CeqN catalysts results in observation of more 02' species (Figure 4.8 b).
A similar effect was observed for catalysts calcined at high temperature (Figure 4.8 c and
4.8 e). These types of oxygen species provide a catalyst with oxygen storage properties,
capable to generate molecular oxygen involved into an oxidation processes. Another
observation from Figure 4.8 concerns the hyperfine signals clearly observed for Cqu
catalysts (Figure 4.8 d and 4e, AH = 78.65 Gauss). These types of signals were attributed
in the literature to a nuclear spin - electron spin interaction of a paramagnetic metal
species, the signal magnitude being proportional with the metal loading and structure
distortion in the metal structure. At this moment, no clear explanation can be offered to
understand the appearance of these signals, further studies being necessary.

Figure 4.9 shows the XPS spectra for CeqN (Figure 4.9 a) and Cqu (Figure 4.9
b) catalysts, both calcined at 500°C, in comparison with the spectra for standard CeOz
powder (Figure 4.9 c). The XPS Ce3d spectrum is complicated due to hybridization of

the Ce" with ligand orbitals and fractional occupancy of the valence 4f orbital [62-65].

148

um u V

 

(b)

(a)

920.0 900.0 880.0
Binding Energy [eV]

Figure 4.9 XPS spectra of: a) CeqN calcined at 500°C; b) Cqu calcined at
500°C; c) CeOz.

The Ce3d spectrum measured for Ce02 (Figure 4.9 c) contains three main 3d5,2 features at

883.2 eV (v), 889.2 eV (v”), and 899.4 eV (v”’). The three main Ce3d3,2 features appear

149

at 901.1 eV (u), 907.7 eV (u”), and 917.3 eV (u”’). The high binding energy doublet

9”

v’”(u ) has been assigned to cerium with 4f) orbital configuration. The v” and v (u” and
u) doublets are assigned to cerium with mixed 4fl and 4f2 orbital configurations [62-64].
These states appear due to the core hole potential in the final state and the 4f
hybridization in the initial state [64-67]. The XPS Ce3d spectra acquired for CeqN
(Figure 4.9 a) and Cqu (Figure 4.9 b) catalysts are similar with that of CeOz,
corresponding to cerium species with Ce4+ oxidation state. No surface Ce3+ was observed
by XPS analysis. The low S/N ratio of the XPS spectra for CeqN and Cqu catalysts are
due to the short scanning time needed to avoid cerium X-ray photoreduction, especially
of the amorphous Cer phase [36, 68-71].

A photoreduction study on the Cqu catalysts by XPS was performed in order to
observed how easily Ce4+ is reduced to Ce3+ by X-ray during analysis (Figure 4.10). It is
known from the literature that cerium can be photoreduced during XPS analysis in high
vacuum due to intense heating of the sample surface, the presence of free electrons into
the chamber and variation in the crystallinity of the exposed particles [65-71]. For short
scanning time (Figure 4.10 a) the surface cerium phase is present as Ce“. As the
scanning time increases (Figure 4.10 b-d), the v’-peak intensity increases corresponding
to the appearance of surface Ce3+ species. Previous studies have shown that the
amorphous cerium phase is reduced more than the crystalline one. The results in Figure
4.9 indicate that cerium supported on Si02 (Cqu which contained more amorphous

phase) is photoreduced readily.

150

 

 

(C)
(b)
1 . (a)
L I J
920.0 900.0 880.0
Binding Energy (eV)

Figure 4.10 Cerium XPS photoreduction spectra for Cqu catalyst calcined at
500°C after: a) 5 min.; b) 15 min.; 0) 30 min.; d) 4 h of scanning.

The homogeneity of the catalysts was verified by comparing the XPS lam/13,2p
values from the ground and unground samples. XPS spectra acquired for CeqN and
Cqu catalysts calcined at 800°C indicate that cerium is present as Ce4+ species (in
CeOz) and the X—ray photoreduction effect is very reduced due to the higher crystallinity

of the cerium oxide phase.

151

4.4 Conclusions

Based on data obtained from DRIF TS analysis in correlation with those from 29Si
NMR, it can be concluded that Ce-nitrate precursor interacts with the support through
intermolecular hydrogen bonds between surface hydroxyl groups from silica support and
Ce(OH)4 formed during the impregnation and drying of catalysts. Using the Ce-
methoxyethoxide precursor, the silica support impregnation takes place through a strong
grafting interaction between the surface hydroxyl groups from SiOz and Ce-Alk, with
elimination of methoxyethanol (see reaction 4-3). After calcination of the catalysts at
500°C, structural modification of the silica support occurs with formation of more surface
hydroxyl groups and lower surface area, as evidenced by BET measurements and 29Si
NMR. TEM-EDX analysis has provided information about the composition and
crystallinity of the CeOz/SiOz catalyst samples, calcined at 500°C. Electron diffraction
patterns obtained by TEM have identified a crystalline Ce02 phase in the CeqN catalysts
and only amorphous phase in the Cqu catalysts. Cerium was positively identified by
EDX analysis on the catalysts surface regardless of its crystalline structure. These
observations are consistent with semiquantitative XRD analysis, which indicates that
cerium is present in the CeqN catalysts mostly as crystalline C602 (>90%), and in Cqu
catalysts as mixed amorphous/crystalline Ce02 (crystalline CeOz is less than 30%).
Particle size evaluation from line broadening calculations indicate formation of larger
Ce02 crystallites on the SiO,_ support in the case of CeqN than for Cqu catalysts. Air
calcination at 500°C of the grafted cerium from the silica support, led to an amorphous

Ce4+ phase, while calcination at 800°C led to a higher degree of crystallinity and larger

152

CeOz particles. EPR spectra showed the presence of paramagnetic Ce3+ as defects in the
CeOz crystalline structure of CeqN catalysts. No signals specific to Ce3+-type species
were found for Cqu catalysts. The large signal corresponding to 02' shows the presence
of electrons capable to form 02 molecules that firrther can be involved into a catalytic
oxidation process. The small CeOz particle size shown by XRD, low temperature TPR
profile, and relative high XPS lam/Imp intensity ratio indicate a better cerium dispersion
on the SiOz support in the case of Cqu catalysts as compared with CeqN catalysts.
Calcination at 800°C of Cqu catalysts led to formation Ce3+-type defects in the CeOz
crystalline structure. Finally, XPS analysis indicated only the presence of Ce4+-species

on the surface of CeqN and Cqu catalysts.

9°

10.
11.
12.
13.
14.
15.
l6.

17.

18.

19.

20.

21.

22.
23.

24.

25.

26.

153

4.5 References

Barrault, J.; Guilleminot, A.; Achard, J. C.; Paul-Boncour, V.; Percheron, G. A.;
Appl. Catal. 1986, 21, 307-312.

Fleisch, T. H.; Hicks, R. F .; Bell, A. T. J. Catal. 1984, 87, 398-413.

Reick, J. 8.; Bell, A. T. J. Catal. 1985, 96, 88-105.

Miki, T.; Ogawa, T.; Haneda, M.; Kakuta, N.; Ueno, A.; Tateishi, S.; Matsuura,
S.; Sato, M. J. Chem. Phys. 1990, 94, 6464-6467.

Diwel, A. F.; Rajaram, R. R.; Truex, T. J. Catalysis and Automotive Pollution
Control 11, Edit by Crucq, A. 1991, Elsevier, Amsterdam, 139-152.

Pembo, J. M.; Lenzi, M.; Lenzi, J .; Lebugle, A. Surf Inter. Anal. 1990, 15(11),
663-668.

Yu-Yao, Y. F.; Kummer, J. T. J. Catal. 1987, 106, 307-312.

Chojnacki, T.; Krause, K.; Schmidt, L. D. J. Catal. 1991, 128(1), 161-165.
Krause, K. R.; Schabes-Retchkiman, P.; Schmidt, L. D. J. Catal. 1992, 134,
204-219.

Schaper, H.; Doesburg, E. B. M.; VanReijen, L. L. Appl. Catal. 1983, 7, 211-220.
Oudet, F.; Courtine, P.; Vejux, A. J. J. Catal. 1988, 114, 112-118.

Yao, H. C.; Yu-Yao, Y. F. J. Catal. 1984, 86, 254-265.

Yang, J. K.; Swartz, W. E. Spectrosc. Lett. 1984, 1 7(6& 7), 331-337.

Gaugin, R.; Graulier, M.; Papee, D. Adv. Chem. Ser. 1975, 143, 147-149.
Yamaguchi, T.; Ikeda, N.; Hattony, H.; Tanabe, K. J. Catal. 1981, 6 7, 324-330.
Raen, S.; Braten, N. A.; Grepstad, J. K.; Qui, S. L. Phys. Scr. 1990, 41(6),
1001-1002.

Nunan, J.; G.; Robota, H. J.; Cohn, M. J.; Bradley, S. A. in Catalysis and
Automotive Pollution Control 11, Edit by Crucq, A. Elsevier, Amsterdam, 1991.
21-230.

Rodriguez, R. 1.; Guerrero, R. A.; F ierro, J. L. G. Appl. Surf Sci. 1989, 40(3),
239-241.

Silver, R. G.; Summers, J. C.; Williamson, W. B. in Catalysis and Automotive
Pollution Control 11, Edit by Crucq, A. Elsevier, Amsterdam, 1991, 167-180.
LeNormand, F.; Hilaire, L.; Kili, K.; Krill, G.; Maire, G. J. Phys. Chem. 1988, 92,
2561-2568.

Shyu, J. 2.; Otto, K.; Watkins, W. L. H.; Graham, G. W.; Belitz, R. K.; Gandhi,
H. S. J. Catal. 1988, 114, 23-33.

Graham; G. W.; Jen, H.-W.; McCabe, R. W. Catal. Lett. 1997, 44, 185.
Haneda, M.; Mizushima, T.; Kakuta, N.; Ueno, A.; Sato, Y.; Matsuura, S.;
Kasahara, K.; Sato, M. Bull. Chem. Soc. Jpn. 1993, 66, 1279-1288.

Mizukami, F.; Maeda, K.; Watanabe, M.; Masuda, K.; Sano, T.; Kuno, K. in
Catalysis and Automotive Pollution Control 11, Edit by Crucq, A. 1991, Elsevier,
Amsterdam, 557-568.

Leclercq, G.; Danthy, C.; Mabilon, G.; Leclercq, L. in Catalysis and Automotive
Pollution Control 11, Edit by Crucq, A. 1991, Elsevier, Amsterdam, 181-194.
Weibel, M.; Garin, F.; Bernhardt, P.; Maire, G.; Prigent, M. in Catalysis and
Automotive Pollution Control 11, Edit by Crucq, A. 1991, Elsevier, Amsterdam,

27.
28.
29.
30.
31.
32.
33.
34.
35.
36.

37.
38.

39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.

54.

154

95-205.

Munuera, G.; Fernandez, A.; Gonzalez-Elipe, A. R. in Catalysis and Automotive
Pollution Control 11, Edit by Crucq, A. 1991, Elsevier, Amsterdam, 207-219.
Zhou, Y.; Nakashima, M.; White, J. M. J. Phys. Chem. 1988, 92, 812-818.
Trovarelli, A. Catal Rev. 1996, 38, 439-520.

Maciel, G. E.; Sindorf, D. W. J. Am. Chem. Soc. 1980, 102, 7607-7608.

Rocha, J.; Klinowski, J. Phys. Chem. Minerals, 1990, 17, 179-186.

Harris, R. K.; Leach, M. J .; Thompson, D. P. Chem. Mat. 1989, I, 336-338
Sato, S.; Maciel, G. E. J. Molec. Catal: A, 1995, 101, 153-161.

Chuang, I. S.; Kinney, D. R.; Bronnimann, C. E.; Zeigler, R. C.; Maciel, G. E.
J. Phys. Chem. 1992, 96, 4027-4034.

Klug, H.P.; Alexander, L.E. X -ray Difiraction Procedures for Polycrystalline
and Amorphous Materials, l-st Ed. Wiley, New York, 1954.

Che, M.; Giamello, E. in Studies in Surface Science and Catalysis by Delmon, B.
and Yates, J. T. Vol. 57, Elsevier, 1990, p. B265-BZ83.

Kepinski, L.; Wolcyrz, M. Catal. Lett. 1992, 15, 329-337.

Bensalem, A.; Bozon-Verduraz, F.; Delamar, M.; Bugli, G. Appl. Catal. 1995,
121, 81-93.

Bruce, L. A.; Hoang, M.; Hughes, A. E.; Turney, W. Appl. Catal. 1996, 134,
151-362.

Craciun, R. Master Thesis Dissertation, Michigan State University, East Lansing,
MI, USA, 1996.

Nakamoto, K. in Infiared and Raman Spectra of Inorganic and Coordination
Compounds, 4-th Ed. Wiley, New York, 1986, p231, p384.

Wovchko, E. A.; Camp, J. C.; Glass, J. A.; Jr.; Yates, J. T. Jr. Langmuir, 1995,
11, 2592-2599.

Matsumura, Y.; Moffat, J. B. J. Chem. Soc. Faraday Trans. 1994, 90(8),
1177-1182.

Verduraz, F. B.; Bensalem, A. J. Chem. Soc. Faraday Trans. 1994, 90(4),
653-657.

Waltenburg, H. H.; Yates, Jr. J. T. Chem, Rev. 1995, 95, 1589-1673.

Morrow, B. A.; McFarlan, A. J. J. Phys. Chem. 1992, 96, 1395-1400.

Binet, C.; Badri, A.; Kizling, M. B.; Lavalley, J. C. J. Chem. Soc. Faraday Trans.
1994, 90, 1023-1028.

Nedez, C.; Choplin, A.; Basset, J. M. Inorg. Chem. 1994, 33, 1094-1098.
Reven, L. J. Molec. Catal., 1994, 86, 447-477.

Sanz, J.; Madani, A.; Serratosa, J. M.; Moya, J. S.; Aza, S. J. Am. Ceram. Soc.
1988, 71 (10), C-418-C421.

Diehl, P.; Fluck, E.; Kostfeld, R. in NMR Oxygen-17 and Silicon-29, Springer-
Verlag Inc. 1981, p 65-150.

McClune, W. F. Powder difii'actionfile, Inorganic phases, International center
for difiiaction data, Swarthmore, 1983, 730, 731.

Kifenski, J .; Baiker, A.; Glinski, M.; Dollenmeier, P.; Wokaun, A. J. Catal. 1986,
101, 1-11.

Nabavi, M.; Spalla, 0.; Cabane, B. J. Colloid Interface Sci. 1993, 160. 459-471.

55.

56.

57.

58.
59.

60.

61.

62.

63.
64.

65.
66.

67.
68.
69.

70.
71.

155

Laachir, A.; Perrichon, V.; Badri, A.; Lamotte, J .; Catherine, E.; Lavalley, J. C .;
ElFallah, J.; Hilaire, L.; le Normand, F .; Quemere, E.; Sauvion, G. N.; Touret, O.
J. Chem. Soc. Faraday Trans. 1991, 87(10), 1601-1609.

Niemantsverdriet, J. W. Spectroscopy in Catalysis, 1993, VCH Weinheim,

p 51-55.

Che, M.; Kibblewhite, J. F. J.; Tench, A. J. J. Chem. Soc. Faraday Trans. 1973,
69, 857-863.

Zhang, X.; Klabunde, K. J. Inorg. Chem. 1992, 31, 1706-1709.

Abi-Aad, E.; Bechara, R.; Grimblot, J .; Aboukais, A.; Chem. Mater. 1993, 5 ,
793-797.

Abi-Aad, E.; Bennani, A.; Bonnelle, J. P.; Aboukais, A. J. Chem. Soc. Faraday
Trans. 1995, 91, 99-104.

Kaufherr, N.; Mendelovici, L.; Steinberg, M.; .1. Less Common Metals, 1985, 107,
281-289.

Wuilloud, E.; Delley, B.; Schneider, W. D.; Baer, Y. Phys. Rev. Lett. 1984, 53(2),
202-205.

Fujimori, A. Phys. Rev. B. 1984, 28, 2281-2283.

Wuilloud, E.; Delley, B.; Schnieder, W. D.; Baer, Y. Phys. Rev. Lett. 1984,
53(26), 2519.

Jo, J.; Kotani, A. Phys. Scrp. 1987, 35, 570-575.

Praline, G.; Koel, B. E.; Hance, R. L.; Lee, H. 1.; White, J. M. J. Electron Spec.
Relat. Phenom. 1980, 21, 17-46.

E] Fallah, J .; Hilaire, L.; Romeo, M.; LeNormand, F.; J. Electron Spectrosc. Rel.
Phenom. 1995, 73, 89-94.

Daucher, A.; Hilaire, L.; LeNormand, F.; Muller, W.; Maire, G.; Vasquez, A.;
Surf Interface Anal. 1990, 16, 341-345.

Paparazzo, E. Surf Sci. 1990, 234, L253.

Paparazzo, E.; Ingo, G. M.; Zacchetti, N. J. J. Vac. Sci. Technol. A, 1991, 9, 1416.
Park, W. P.; Ledford, L. S. Langmuir, 1996, 12, 1794-1799.

CHAPTER 5

CHARACTERIZATION OF UNPROMOTED AND Ce02 PROMOTED
MnOx/SiOz CATALYSTS

5.1 Introduction

Supported MnO, and rare earth oxide catalysts have been successfully utilized in
oxidative dehydrogenation reactions [1-4]. Previous studies on structural characterization
and activity measurements of MnOx have emphasized catalysts in which y-A1203 was
used as support [5—9]. The activity of transition metal oxide catalysts can be modified by
addition of lanthanide oxides which can act as textural and/or structural promoters [10-
15]. Previous studies have shown that rare earth oxide promoted MnOx/y-A1203 catalysts
behave as materials with high oxygen storage capabilities, showing excellent catalytic
activity in oxidation reactions [8, 13-19]. Cerium oxide (CeOz) is known to be effective
as a promoter by increasing thermal resistance (inhibits the loss of the support surface
area), dispersion, and the stability of the transition metals used as active catalyst
components [1 1-15]. The structure of unpromoted and CeOz-promoted MnOx/SiOZ
catalysts is strongly dependent on preparation method, the chemical nature of precursors
used, and Mn/Si and Ce/Si ratio and loading [7-17, 20].

The purpose of this research was to examine the influence of Mn-loading,

calcination temperature, and the presence of Ce02 promoter on the SiO;,_ supported MnO,

156

157

catalysts. Catalytic activity and selectivity at different temperature, will be summarized

and presented in correlation with the catalysts’ structures.

a) Methyl lactate catalytic oxidative dehydrogenation reaction

O\\C/OC H3 OQC/OCH3
C'H + “2 02 cut
Methyl—Lactate h’lcthyl—Pyruvate

b) 2,3-Pentanedione condensation reaction with ethylenediamine

H3C O HzN HC

\ / 3

C/ \CHz \C” \an
\\ , H2 -2H20 \\ / H2

HSCZ/ 0 HM 059/ N

2.3-Pcntancdione Ethylene-diamine 2-Fthyl-3-Methyl

Dihydro-Pyrazine

c) Dihydro-pyrazine oxidative dehydrogenation to pyrazine

H3C N H3C\ /
\C// Ysz A C/ \pn
/ ‘ “‘7’
H5C2/ N -H20 H5C2 N
2-Ethvl-3-Mcthyl 3-l‘ithrl-3-Metlwl
Dihydro-Pyrazine Pyrazme

Figure 5.1 Oxidative dehydrogenation processes considered as probe reactions
for unpromoted and CeOz-promoted MnOx/SiOz catalysts

X-ray diffraction (XRD), temperature programmed reduction (TPR), electron
paramagnetic spectroscopy (EPR) and X-ray photoelectron spectroscopy (XPS) analyses

were used to obtain information about the bulk and surface structure, and to characterize

158

the fresh and used, unpromoted and CeOz-promoted MnOx/SiOz catalysts. Reactions
used for catalyst testing are presented in Figure 5.1. Catalytic activity data for the
catalysts used in oxidative dehydrogenation of methyl lactate (ML) to methyl pyruvate
(MP, Figure 5.1 a), and 2-ethyl-3-methyl-dihydropyrazine (23HPy, obtained from 23P
and ethylenediamine, Figure 5.1 b) to 2-ethyl-3-methyl-pyrazine (23Py, Figure 5.1 c) are
reported in correlation with their bulk and surface structure. Previously, MnOz was
successfully used in dihydropyrazine dehydrogenation reactions (with high yield and
selectivity) but in a discontinuous (batch) process [21]. Oxidative dehydrogenations of
lactic acid to pyruvic acid [22] or of ethyl lactate to ethyl pyruvate [23-25] were
performed with high yields and selectivity in the presence of Fe/P based catalysts.
Providing the activity of MnOx-based catalysts in oxidation reaction, ML to MP has been
used as a probe reaction in this study, as well.

Experimental

Catalysts preparation. The pure SiOz support was prepared from silica-gel
(Davison Chemical Co.) which was finely grounded (< 230 mesh) and calcined in air at
500°C for 24h, prior to impregnation (300 mz/g). Pure Si02 and CeOz-promoted Si02
(Cqu and CeqN, see Chapter 4) were used as supports for Mn-based catalysts. Aqueous
Mn(NO3)2x6H20 (Aldrich) was used to prepare MnOx/SiOZ (with Mn/Si atomic ratio y =
0.5-14x10'2) and MnOx/CeOz/SiOz (y = 10x10'2) catalysts. Catalysts with various Mn/Si
atomic ratios (y) were designated as MnynT, followed by the calcination temperature (T)

used during preparation.

159

Table 5.1 The list of the unpromoted and CeOZ-promoted MnOx/SiOz catalysts
together with the designated symbols used in this study.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0rd. Catalyst Description of the catalysts:
# Symbols Mn/Si and Ce/Si atomic ratio, calcination temperature,
presence of CeOz promoter

1 SiOz Pure SiOz support from silicagel, calcined at 500°C

2 Mn005n500 MnOx/SiOz catalyst, with Mn/Si = 0.005, calcined at 500°C

3 Mn02n500 MnOx/SiOZ catalyst, with Mn/Si = 0.02, calcined at 500°C

4 Mn06n500 MnOx/SiOz catalyst, with Mn/Si = 0.06, calcined at 500°C

5 Mn10n300 MnOx/SiOz catalyst, with Mn/Si = 0.10, calcined at 300°C

6 Mn10n400 MnOx/SiOz catalyst, with Mn/Si = 0.10, calcined at 400°C

7 Mn10n500 MnOx/SiOz catalyst, with Mn/Si = 0.10, calcined at 500°C

8 Mn10n575 MnOx/SiOz catalyst, with Mn/Si = 0.10, calcined at 575°C

9 Mn10n650 MnOx/SiOz catalyst, with Mn/Si = 0.10, calcined at 650°C

10 Mn10n750 MnOx/SiOz catalyst, with Mn/Si = 0.10, calcined at 750°C

11 Mn10n500 MnOx/SiOz catalyst, with Mn/Si = 0.10, calcined at 500°C

12 Cqu CeOz-promoted SiOz support with Ce/Si = 0.08, calcined at
500°C, using a Ce-alkoxide promoter

13 MnlOSicea Mixed Mn-Ce/Si02 catalyst with Mn/Si = 0.10, on Cqu
support, calcined at 500°C

14 CeqN CeOz-promoted Si02 support with Ce/Si = 0.08, calcined at
500°C, using a Ce-nitrate promoter

15 MnlOSicen Mixed Mn—Ce/Si02 catalyst with Mn/Si = 0.10, on CeqN
support, calcined at 500°C

 

The Mn-based catalysts prepared in this study were obtained by incipient wetness
impregnation with aqueous precursor solutions. The CeOz-promoted MnOx/SiOZ
catalysts were prepared by sequential impregnation (first cerium followed by manganese).

All samples were dried at 125°C and calcined for 16 h in air at 500°C after each

I60

impregnation step, prior their use in the oxidative dehydrogenation processes of ML to
MP and of 23HPy to 23Py.

X-Ray Difiiaction (XRD). X-ray powder diffraction patterns were obtained with a
Rigaku XRD diffractometer employing Cu Ka radiation (A = 1.541838 A) and operated
at 45 kV and 100 mA. Diffraction patterns were obtained using a scan rate of 0.5

deg/min with 1/2 mm slits. Powdered samples Were mounted on glass slides by pressing

the powder into an indentation on one side of the slide. The mean crystallite size (3) of
MnOx particles was determined from XRD line broadening measurements using the
Scherrer equation (1-11, Chapter 1). Semi-quantitative X-ray diffraction data for
evaluation of the MnOZ/Mn203 ratio were obtained by comparing Mn02 <110> and
Mn203 <222> peak ratios calculated for each catalyst sample (see details in Chapter 1)
Temperature-programmed reduction (T PR). A common technique used today to
characterize the redox properties of an active phase fi'om a catalysts is temperature
programmed reduction. The technique uses a nitrogen-hydrogen reducing mixture which
flows at various temperatures over the catalyst sample, connected to a thermal
conductivity detector (TCD) recording the amount of H2 evolving from the sample and
compared with the initial feed. Due to reducing processes, hydrogen is consumed by the
sample. The total uptake during a TPR analysis and the position of the reduction maxima
as a function of temperature represent a real fingerprint of the catalyst sample [26]. The
reduction of the supported metal oxides may be slowed down or promoted by their
dispersion on inert porous materials such as A1203, SiOz or TiOZ and is a function of the

metal oxide-support interaction. The surface of the metal oxide is reduced much easier

161

than the bulk phase. Characteristics such as catalyst sensitivity to reduction, autocatalytic
effect, structure, porosity, and dispersion influence significantly the position of the
maxima from a TPR profile. Any factors which limit or favor the H2 adsorption on the
catalyst sample surface should be accounted in order to correctly interpret the TPR
profile. The presence of any impurities or small molecules adsorbed on the surface
capable to consume H2 such as CO, C, various carbonate-like species, hydrocarbon
residue, nitrate-like species etc., will influence the results and interfere with the real
reduction activity of the sample [2 8].

Figure 5.2 show a laboratory setup for a classical TPR experiment. At the
beginning of the TPR experiment, the gas flows over a catalyst (accurately weighted) at a
temperature low enough to prevent reduction. The temperature on the catalyst is then
increased with a linear rate and the rate of the reaction is monitored by measuring
concentration or pressure changes in the gas phase or weight of the catalyst. The
reducing gas used in the TPR experiments is usually a mixture of Nz/Hz. The
experimental setup can be used for temperature programmed oxidation (TPO) as well, by
replacing the reducing gas with air [27].

Generally, TPR is considered to be a very sensitive technique which is
independent of any specific catalyst properties. The method has been applied to the study
of many supported catalyst systems. It was therefor considered to be a proper and useful

technique for the study of the MnOx/CeOz/SiOZ system.

162

N, H, AIR

2.5 7

a/’/"\"\
rain . . ‘ 132-1
‘__. ‘1__a

l 8
l ”ll/Ill III «I

O O
’ :VA'flflUZL' . ‘% I5
0 I

\-- "

-- P-
b--'.-

’-
I
‘0
I
|
a \

 

 

 

 

[ 2.8
0 l

 

_=-2 .24..
‘ —“— H

 

Figure 5.2 Laboratory setup for a TPR experiment: 1 - thermal conductivity cell;
2.1 - reduction valve; 2.2 - M/Ale3 catalyst; 2.3 - molecular sieves; 2.4 - Dewar trap
(193 K); 2.5 - gas flow switch; 2.6 - brake capillary; 3 - reactor; 4 - furnace; 5 -
temperature programmer; 6 - chart recorder; 7 - thermocouple [27].

TPR profiles for the CeqN and Cqu catalysts were obtained using a classical U-
shaped quartz reactor connected to a thermal conductivity detector (TCD) for H2
consumption analysis. The temperatures were measured with a Ni-Cr-Ni thermoelement
placed directly in the catalyst bed. A reducing agent mixture of 10% H2 in N2 was passed
over 0.1 g catalyst powder, at 30 mL/min flow rate. The temperature interval considered
for analysis is from 25°C to 950°C (10°C/min heating rate). The TPR profiles are
reported as H2 consumption (arbitrary units) versus temperature [°C].

Electron Paramagnetic Resonance (EPR). EPR is a spectroscopic technique well

suited to study paramagnetic solid materials as well as paramagnetic species adsorbed on

163

solid surfaces. Surface defects, paramagnetic metal oxides supported on high surface
area materials, adsorbed atoms, molecules or ions which, in many cases, may be
intermediates of a catalytic reaction, are several classes of species observable by EPR
spectroscopy, and may provide useful insights about a heterogeneous catalysis process.
The sensitivity of the technique allows the study of very low concentrations of active
species. A limitation as well as an advantage of EPR is that the technique does not detect
diamagnetic species. The oxidation state of a transition metal, crystalline defects, and
coordination with neighboring atoms are just some of the data which can be obtained
from EPR spectra of solid catalyst materials.

The principle of EPR spectroscopy is based on the Zeeman effect observed for
paramagnetic species (the presence of a unpaired electron with 1/2 spin). Placed into a
magnetic field, the energy level of the free electron splits in two levels, as shown in
Figure 5.3. At thermal equilibrium, the spin population of the level is given by the
Maxwell-Boltzmann law (5-1):

nl/nz = exp (-AE/kT) (5-1)
where n, and n2 are the spin populations characterized by the MS values of =l/2 and -1/2,
respectively. At 77K, in a field of about 300 Gauss, n, and n2 differ by less than 0.005
[29]. The transition between the two Zeeman levels can be induced by suitable
electromagnetic radiation which is able to provide the required energy gap to fulfill the

resonance condition (5-2):

I“) = ch'BB (5'2)

164

where h = Plank constant, gc = giromagnetic factor of an electron; H3 = 9.27x10'2'

erg/gauss = Bohr magneton, B = magnetic flux (in tesla), and o = flux frequency.

+ E: 1/2gI’I'BBres
MS = +1/2

  
   
 

Energy

 

bl) = g HBBres

MS = '1/2
E = '1/2 g HBBres

A Integral Spectrum

—/\/— Derivative Spectrum

 

Figure 5.3 The Zeeman energy levels of a free electron placed into an external
magnetic field [29].

At about 2.8 MHz/Gauss value for the magnetic field applied on a paramagnetic
system, resonant transitions can take place. The typical radiation, employed in EPR
spectroscopy is in the microwave region. The energy absorption necessary to promote
electrons from a lower to a upper energy level is called a resonance signal, and at this
moment n, and n2 equalizes if the absorption reaches saturation. When the electron from
the upper level emits the ho energy, it can return to the lower level and satisfy the
Maxwell-Bolzmann law. The energy can be dissipated to the lattice (T,, spin-lattice
relaxation) or it can be exchange with other neighbor spins (T2, spin-spin relaxation).
EPR experiments are typically performed under liquid N2 (77K) or He (4.2K) due to the

temperature dependence of relaxation time (T,). If TI is too long, electrons do not have

165

time to return to the initial energy level, so the intensity of the signal decreases and is no
longer proportional to the spin number present in the sample (saturation effect). The
typical shape of an EPR signal is Gaussian or Lorentzian, and it can be represented either
as an integral or as an derivative signal (see Figure 5.3) [29]. The g value from a real
sample is usually slightly different from the ge, but to extract chemical information this
value should be very accurately determined. The absolute value of g can be evaluated
either by independent B and 0 measurements or computer simulation. Equation 5-3
shows the relationship between g, B and u.

g = hU/uBB (5-3)
where h = Plank constant, [13 = 9.27x10'21 erg/gauss = Bohr magneton, B = magnetic flux
(in tesla), and u = flux frequency. In practice, the g value is often determined by
comparing the field values at resonance for the sample with that from a reference such as
diphenyl-picryl-hydrazyl (DPPH), g = 2.0036, or Cr3+ in MgO matrix, g = 1.9797.
Relation 4-1 (see Chapter 4) gives the equation for the evaluation of g. The surface of a
catalyst containing a paramagnetic species can be studied by analyzing the g value. This
value may provide the structural information for the supported metal oxides (lattice
parameters, octahedral distortion, etc.) [30]. Information about the structure of a catalyst
using EPR can be obtained by indirect study of paramagnetic gas-probe molecules such
NO [31], ”‘02 [32], or C'80 [33] adsorbed on the surface. Data derived from EPR spectra
can provide information about the surface crystal field, the redox properties of the
surface, the identification of catalytically active sites, the surface morphology, the

mobility of adsorbed species and the coordination chemistry of surface metal ions [29].

166

In this study, the MnOx/CeOz/SiOz powder catalyst samples were analyzed in
quartz sample tubes at -160°C using a Varian E4 spectrometer. The magnetic field was
set at 3200 Gauss with a modulation frequency of 100 kHz. The g-factor (g5) for cerium
samples (8,) was determined by comparison with the field value at resonance for a
reference (BM), in this case the DPPH radical (grcf = 2.0036). The reference was placed
together with the catalyst samples into the quartz tube for accurate g-value
determinations. Accordingly, the value for gS was evaluated based on relation 4-1.

X-ray photoelectron spectroscopy (XPS). The principle of XPS analysis is based
on the photoelectric effect which implies emission of electrons with a given kinetic
energy from a surface (sample) when irradiated with an electromagnetic flux (for ex. light
or X-ray). The number of photoelectrons depends on the intensity and energy of the
light. Figure 5.4 shows schematically the photoelectric process together with the Auger
process which follows [34].

The ejected photoelectron (core electrons) contains information about the surface
configuration due to the fact that its kinetic energy is specific to the electronic level and
the type of element from which originated and is proportional with the energy (ho) of the
irradiating light.

Only the surface electrons (first 10-50 A), unaltered by inelastic collisions with
other particles during ejection are information carriers. Relation 5-4 defines the kinetic
energy of a core or valence electron ejected from given sample:

KE = ho - BE - (p (5-4)

l67

where KE = kinetic energy; BE = binding energy of the photoelectron; h = Plank

constant; I) = frequency of the exciting radiation; (0 = spectrometer work function.

 

nggmb
”H—ltmh

,0 moromcmou
/ .

«on:

 

 

; MICE! ELECTION

I
/
/

WO- tuna

-—O-'l—O——Lron:
. I

l

l
——.—+——K0lts

 

 

 

Figure 5.4 Schematically representation of the photoelectric effect [34]

The sources most common used for photoexcitation are X—ray sources, either Mg K0t
(1253.6 eV) or Al K0t (1486.3 eV). The technique was developed in the 1950’s by
Siegbahn, but the first commercial instrument was available only after 1970. XPS
analysis provides information about the surface structure (elemental composition, metal’s

oxidation state) and in some cases can evaluate the dispersion of one phase over another

 

I68

[35]. Mn and Ce surface concentrations were evaluated from the intensity of the Ce3d and
anp peaks relative to the Sizp peak, after area correction with the atomic sensitivity
(ASF, [34]) factors for each atom detected at the catalyst surface.

The model proposed by Kerkhof and Moulijn [36] has been used to evaluate the
theoretical monolayer dispersion of MnOx from the XPS IMan/ISIZp intensity ratio. In this
model, a supported catalyst is considered to be a stack of sheets of supported material,
with cubic crystals representing supported particles. Figure 5.2 illustrates schematically
the proposed stratified model. The electrons ejected from the surface are considered to
have a perpendicular trajectory to the surface (support sheets) and that a Lambert-Beer
type law is valid. The XPS intensity ratio, I /IS [p = promoter (Mn); 3 = support (Si)],
assumes an exponential attenuation of the electrons escape depth in the particle and
support and is described by the relation presented in Figure 5.2 , below the catalyst model
[36]. The photoelectron cross sections for Mn, Ce, and Si were taken from tables
reported in the literature [37]. The escape depths of the anp electrons were calculated
using the relations develop by Penn for quantitative XPS analysis [38]. The surface
coverage factor was calculated by considering that manganese is deposited on a high

surface area SiOz (So =300m2/g and p = 2400 kg/m3).

I69

 

 

 

 

 

 

 

 

 

 

 

Figure 5.5 Schematically representation of the Kerkhof-Moulijn model for a
supported catalyst surface: Ips — —XPS intensity; D(8p s)— —detector efficiency; (p/s)= atomic
ratio; 0,, =photoelectron cross section; B= t/lt, Je= escape depth factor (t— — 2/pSo, where
p= density, and S0 surface are of the support) [36].

XPS data were obtained using a Perkin-Elmer Surface Science instrument
equipped with a magnesium anode (1253.6 eV) operated at 300W (15 kV, 20 mA) and a
hemispherical analyzer operated with a pass energy of 50 eV. Spectra were collected
using a PC137 board interfaced to a Zeos 386SX computer. The instrument typically
operates at pressures below I x 10'8 torr in the analysis chamber. Samples were analyzed
as powders dusted onto double-sided sticky tape. Binding energies for the catalyst

samples were referenced to the Sizp peak ( 103.4 eV) and Cl, (285.4 eV). XPS binding

energies were measured with a precision of i 0.2 eV or better.

I70

Activity measurements. As mentioned above, oxidative dehydrogenation of ML
to MP (Figure 5.1 a), and 23HPy to 23Py were used as probe reactions for catalysts
testing. The reactions were performed in a vertical, fixed-bed Pyrex reactor equipped
with a heating system capable of controlling the temperature with i1°C accuracy. The
reactions were run at various temperatures (200°C for 23HPy to 23Py and 260°C for ML
to MP) maintaining (as much as possible) all other conditions constant: ML (23HPy)
flow rate = 0.13 mL/min, air flow rate = 42 ml/min, atmospheric pressure, residence time
= 15 5. Typically, products are collected for 20 min during steady-state operation of the
reactor. The reaction products exit the bottom of the reactor and are condensed in a
cooling flask using an ice-bath. The volume of the liquid product and the flow rate of the
gas product are measured during operation in order to compute a mass balance. The .
outlet of the reactor was connected to a Riken infrared C02 gas analyzer which measured
the amount produced in the reaction. Product analyses in ML oxidation was performed
using a Varian 3700 gas chromatograph equipped with FID detector and a Supelco
packed column (4% carbovax/carbopack B-DA). Good reproducibility of reactant and
product analysis was achieved by direct on-column injection into the column of 1 ul
condensed product solution mixed with the internal standard. A response factor was
determined for each component observed in the product mixture using isopropanol as
internal standard. For 23HPy oxidative dehydrogenation reaction, the reactants and
products were analyzed using a Varian 3300 gas chromatograph equipped with a TCD

detector and a capillary column (Supelco SPB-l). The internal standard used in this case

171

was hexanol. ML and 23HPy conversions and selectivities in MP and 23Py, respectively,
were reported in wt% relative to the theoretical yields based on converted reactants.

Results and Discussions

Unpromoted fresh Mn0,/Si02 Catalysts. In the case of unpromoted MnOx/SiOz
catalysts, calcined at 500 °C and with y varying from 0.5 to 14, different XRD patterns
were observed. Figure 5.6 shows the XRD pattern observed for the Mnyn catalysts.

For low manganese loading (y = 005, Figure 5.6 e) no pattern specific to a
crystalline MnOx phase was observed; for 02 S y S 10, (Figure 5.6 c-d) only the MnOz
crystalline phase was identified; when y 2 10 (Figure 5.6 a-b) mixed XRD patterns
indicative of Mn02 and anO3 crystalline phases was identified [40]. Based on the
diffraction patterns observed and presented in Figure 5.6, the crystalline particle sizes of
the MnOz and Mn203 were evaluated from line broadening calculations. The results are
presented in Table 5.2 together with BET and XPS surface concentration data.

As observed from Table 5.2, column 2, the surface area of the Mnyn500 catalysts
does not change significantly after manganese impregnation due to the fact that MnOx
formed after calcination is not a porous material which may affect the catalyst texture.
The small differences observed are due to the variation in the catalyst weight considered

for calculation as a function of Mn loading.

 

I72

M1102

< >

._ .5
<2] 1> <222> MnOa
<101>~ 3111?;
M11203
<440>

   
  

Nan 2

 
  

 

 

(b)

(c)

 

 

(e)

l I ~ ~ I
20.0 40.0 60.0
20 [degrees]

Figure 5.6 XRD pattern for Mnyn catalysts calcined at 500°C: a) Mn14n500; b)
Mn10n500; c) Mn06n500; d) Mn02n500; e) Mn005n500.

 

173

In Table 5.2, columns 3 and 4 present the crystalline particle size values (3) for
the Mn02 and anO3 species, and column 4 shows the MnOz/MnZO3 ratio. As observed,

at low loading (y < 10) the dominant species on the catalyst is MnOz.

Table 5.2 XRD and XPS MnOx dispersion data and MnOz/Mn203 ratio evaluated
from diffraction patterns for Mnyn500 catalysts (y = 0.5-14 = Mn/Si atomic ratio)

 

 

 

 

 

 

Catalyst Surface 3 [nm] d [nm] Ra cXPS Ianp/ISin
Area [mzlg] B-MnOz 131—an03 corrected
Mn005n 282 b b --- 0.15
Mn02n 275 11.6 b 1 0.35
Mn06n 257 15.5 b 1 0.98
Mn10n 255 16.5 23.7 0.89 1.08
Mn14n 255 16.8 25.5 0.59 1.09

 

 

 

 

 

 

 

 

a - R = A(<110>MnOz)/[A(<110>Mn02) + A(<222>anO3)]; b - no XRD pattern
observed; 0 - evaluation of XPS Imp/15,2p ratio based on corrected area using the atomic
sensitivity factor (ASF) of Mn and Si specific to the instrument
The particle size of MnOz increases with the increase of the Mn loading (11.6 nm to 16.8
nm). At high Mn loading (y 2 0.1), the presence of anO3 species becomes significant,
and at y = 14, the ratio between the two manganese oxide species approaches 1/2. The
particle size for Mn203 crystallites is around 24 nm, much larger than that for MnOz.
This observation can be attributed to the fact that at low loading, after impregnation with
Mn(NO3)2 precursor, manganese as Mn2+ is well dispersed on the SiOz support and

accessible to the oxygen from air. Thus, during calcination Mn2+ is oxidized easily to

Mn4+ and forms small crystalline MnOz particles.

174

 

 

 

 

 

 

 

 

(a)
fi I 1 ' l m T ' I a T ' I
25 a -
. Md nunlayrrdispasim -
1 . Emmi-mi dsrnsron
2.0 .1 .
.9 ,
,5 I
a 1.5 - -
:g 1.0 .. . -
05 ., -
- l
’ O O O 0
00¢ - . a . . a - .
. 0.02 0.04 0.06 0.118 0.10 0.12 0'14 0.16
Ml/Sl Atomic Ratio
0))
0‘4 ' l ' 1 ' 1
I l
03 - -

Ratio
I

15‘ 02_ -

xpsrm/t
S
I

 

 

 

0 ' 200 ' 400 ' 600 ' 300
Calcination temperature of the WSi=0.l catalyst

Figure 5.7 Mn monolayer dispersion (squares) evaluated from Kerkhof and
Moulijn model in comparison with the experimental value (rhombus) for: a) Mnyn
catalysts; b) Mn10n calcined at various temperatures

175

At high Mn loadings, after impregnation, manganese is probably forming large clusters
which are less permeable to oxygen and as a consequence the oxidation process during air
calcination occurs only to anO3 which is present on SiOz surface as large crystallites.

XPS Imp/15,2p ratio based the corrected areas for anp and Sizp using the ASF
values specific to Mn and Si atoms (Table 5.2, column 5) indicates, an expected increase
with the Mn loadings leveling out at y values above 0.06. Based on this observation, it
can be concluded that Mn dispresion is decreasing with increasing the MnOx
concentration. A more accurate estimation of the Mn dispersion is possible considering
the Kerkhof and Moulijn model [36], described in the experimental section.

Figure 5.7 a shows the Mn monolayer dispersion used as a reference in
comparison with the dispersion values expressed by the XPS 1,,4,,2p/IS,2p ratio. The
experimental values follow the theoretical monolayer dispersion only for the Mn005n500
and Mn02n500 catalysts (low Mn loadings). For catalysts with higher Mn loading (y 2
06), the experimental value is situated far from the Mn monolayer dispersion. The XPS
dispersion results are consistent with the data obtained from XRD line broadening
calculations which indicated MnOz (anO3) crystalline particles larger than 15 nm for
catalyst with Mn loading greater than y = 06, which clearly indicates a low manganese
dispersion. The value of binding energy of the szm XPS peak lies around 642.2 eV,
which is attributed to Mn4+02 [9, 34]. The peak at 642.2 becomes more and more
asymmetric as the Mn loading increases, corresponding to the appearance of a peak
specific to Mn3+ and Mn2+ species, which have a binding energy for anpm at around

640.8 eV [9, 34]. For a better correlation between the binding energies specific to various

 

176

type of MnOx species, XPS analysis of standard compounds were performed and used for

comparison (see Figure 5.8 a-d).

642.2 640.8

anps/z

anp‘”2 A = 11.6 ev

 

(d)

I I I I
655.0 650.0 645.0 640.0

Binding Energy [eV]

Figure 5.8 anp XPS spectra for Mn—standard compounds: a) pure anO3; b)
Mn203 obtained from calcination of Mn(NO3)2 at 750°C; c) pure MnOz; d) MnOz
obtained from calcination of Mn(NO3)2 at 300°C.

177

XPS and XRD results can be well correlated with the TPR results obtained for the

Mnyn catalysts and presented in Figure 5.9.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

441
19
* (a)
2?
5i
. 429 97
.S , _- (b)
g __ “— V 3 wife :
g 490 ._ V w— (E?
L) r—' 419
:- 476 m _ (d)
'— (e)
f - % ‘__:_ (f)
0' - rzbo' ' ' 400' ' '600' ' '80'0' '95‘0

Temperature [°C]
Figure 5.9 TPR profiles for the Mnyn catalysts calcined at 500°C: a) Mn14n500;
b) Mn10n500; c) Mn06n500; d) Mn02n500; e) Mn005n500; 1) pure SiOz.
The TPR profiles of the Mnyn500 catalysts are similar with that observed for the
unsupported MnOz, exhibiting two major peaks at 344°C and at 445°C. This indicates

that MnOz is physically deposited on the 8102 support without a strong chemical

178

interaction. In the literature, the two peaks were attributed to the two-steps reduction
process of the Mn02 or MnOz/Mn203 mixture, first to Mn3O4 and then to MnO species
[9]. The small shifts in the peak position observed as the Mn/Si ratio increases can be
attributed to the difference in the particles size of the supported Mn02. It is known that a
higher reduction temperature corresponds to bulk phase reduction whereas a lower
reduction temperature is required for dispersed phase [27-28]. As a consequence of the
increased MnOx particle size with Mn loading, the reduction temperature varies from
419°C for the Mn02n catalyst (small MnOx particles) to 445°C for the Mn14n catalyst
(large MnOx particles). The relative intensities of TPR peaks vary with the Mn-loading,
too, the H2 consumption being proportional with the manganese available for reduction
on the catalysts.

The effect of the calcination temperature on the structure of the unpromoted
MnOx/SiOz catalysts used during the preparation process was studied as well. The XRD
spectra obtained from the analysis of Mn10n catalysts calcined at various temperature (in
the 125°-750°C range) are presented in Figure 5.10. These data indicate a variation in the
MnOsznzO3 ratios, with MnOz dominant at low temperature and anO3 at temperature
above 600°C, consistent with previous findings [7,9]. In the 400°-600°C domain, mixed
XRD patterns specific to bath MnOz and anO3 species are observable. As observed in
Table 5.3, column 4, the Ra value is close to l for catalyst dried at 125°C or calcined at
temperatures below 400°C, decreases as calcination temperature is raised and reach 0 at

temperatures above 600°C.

 

179

Mn 20 3
<222>
Mn20 3

< >
211 Mn203

<400>
(a)

(b)

 

 

_ _ l _ l | _ _ 1
10.0 20.0 30.0 40.0
20 [degrees]

Figure 5.10 XRD patterns for Mn10n catalysts calcined at various temperatures:
a) Mn10n750; b) Mn10n650; c) Mn10n575; d) Mn10n500; e) Mn10n400; f) Mn10n300;
Mn10n125.

 

180

As observed from the data presented in Table 5.3 column 2 and 3, the particle size

(d) evaluated from XRD line broadening using MnOz <110> and Mn203 <222>

diffraction peaks increased with calcination temperature.

Table 5.3 XRD particle size evaluated from line broadening calculations for B-
Mn02 and 01-an03 species on Mn/Si = 0.10 (atomic ratio) catalyst calcined at various
temperatures (125°-750°C temperature domain)

 

 

 

 

 

 

 

 

Catalyst d [nm] 3 [nm] R. CXPS IMan/ISin
B-MnOz a—Mn203 corrected
Mnlsil25 15.1 b l 2.08
Mnlsi300 15.4 b l 1.67
Mnlsi400 16.2 12.1 0.97 1.37
MnlsiSOO 16.5 23.7 0.89 1.08
Mnlsi575 17.5 25.3 0.33 0.9
Mnlsi650 b 26.0 0 0.98
Mnlsi750 b 29.4 0 0.75

 

 

 

 

 

 

 

a - R = A(<110>MnOz)/[A(<110>Mn02) + A(<222>Mn203)]; b - no XRD pattern
observed; c - evaluation of XPS Imp/15,2p ratio based on corrected area using the atomic
sensitivity factor of Mn and Si specific to the instrument
This fact can be attributed to the mobility of small MnOZ crystallites formed and well
dispersed on the silica surface during calcination allowing them to get together and form
large crystalline particles, at high temperatures. Again, the anO3 crystalline particles
are larger that those observed for Mn02. The Mn dispersion evaluated from XPS surface

concentration indicates a decrease in Mn surface coverage with increasing calcination

temperature. This observation is consistent with that from the XRD line broadening

181

calculations which indicate an increase in MnOx particle size with increased calcination
temperature. As observed from Figure 5.7 b, the XPS Imp/Imp ratio shows a steady
decrease with the increase in the calcination temperature of the catalyst.

An analysis of the Mn oxidation state using XPS data, is presented in Figure 5.11.
The S/N decreases with increasing calcination temperature. The XPS spectra from Figure
5.11 a—b show a symmetric peak at 640.8 eV indicating the presence of manganese as
Mn3+ or Mn2+ species whereas the spectra from Figure 5.11 e-g show a symmetric peak at
642.2 eV corresponding to manganese as Mn4+- type of species. These data are
consistent with the XRD data which pointed to Mn203 formation at high temperatures
and MnOz formation at low temperatures. The XPS spectra from Figure 5.11 c-d, show
the presence of mixed Mn‘H/Mn3+ type of species (with the maximum at 640.8 eV and a
shoulder at 642.2 eV) on the surface of the catalysts. Additional information was
obtained from TPR analysis of the Mn10n catalyst calcined at various temperatures
(Figure 5.12 a-d). As observed from Figure 5.12 a, calcination to high temperature
(750°C) led to an attenuation of the reduction peaks at 344°C and 445°C and the
appearance of a strong peak at 519°C. The new peak can be attributed to the reduction of
the Mn3O4 phase to MnO. Thus, high temperature calcination of the Mn10n catalysts led
to formation of a stable Mn3O4 phase. Some of the MnZO3 phase is still present. The
presence of three clear peaks in the TPR profile of the sample calcined at 575°C is

indicative of the existence of all three MnOx phases on the catalyst surface.

 

182

A = 11.6 eV 642.2 640.8

(a)

(b)

(C)

(d)

(e)

(0

(g)

I | | I
655.0 650.0 645.0 640.0

Binding Energy [eV]

Figure 5.11 anp XPS spectra for Mn10n catalysts calcined to various
temperatures: a) Mn10n750; b) Mn10n650; c) Mn10n575; d) Mn10n500; e) Mn10n400;
f) Mn10n300; g) Mn10n125.

 

183

519
445
344
(a)
ML -
441
‘ 497
(c)
(d)

0" 200‘ ' '400' ‘600' ' '800' '950
Temperature [°C]

H2 comsumption [a.u.]

 

Figure 5.12 TPR profiles of Mn10n catalysts calcined at various temperature
during preparation: a) Mn10n750; b) Mn10n575; c) Mn10n500; d) Mn10n300.

The two clear peaks in the TPR profiles of samples calcined at low temperature,
indicate the presence of a Mn02 phase on the catalyst. The two peaks correspond to the

two steps reduction of MnOz (Mn02 to anO3, and MnZO3 to MnO, respectively).

184

Ce02 promoted Mn0/Si02 catalysts. Information about the crystalline phases
formed on the CeOz-promoted catalysts was obtained from the XRD spectra presented in
Figure 5.13. For the two types of CeOz-supported MnOx/SiOZ catalysts (MnlOncea and
MnlOncen), only diffuse XRD patterns specific to a MnOx crystalline phase were
observed. This fact can be attributed either to the formation of small MnO,‘ particles or to
an amorphous manganese phase from the CeOz-promoted SiOz catalysts calcined at
500°C. It seems that the presence of cerium favors high MnOx dispersion on the SiOz
support. Analyzing the positions of the diffuse XRD peaks other than those specific to
CeOz, the most probable manganese species formed is MnOz. The particle sizes of MnO,‘
evaluated from line broadening calculations are presented in Table 5.4, together with

BET surface area and XPS dispersion data for supported MnOx.

Table 5.4 MnOx particle size evaluated from XRD line broadening calculations,
XPS surface concentration, and Mn dispersion data for unpromoted and CeOz-promoted
MnOx/SiOz catalysts

 

 

 

 

 

 

 

 

 

 

Catalyst Surf.2Area a [m] anpm BE“ 15018 Imp/18,2, 1,,.,,2,/1Sizpc
[m lg] Mn02 [eV] corrected Experrm.
Mn10n500 255 16.5 642.5(640.8) 1.08 0.176
Mnl Osicea 225 6.7 642.3 1.77 0.286
Mnl Osicen 230 13.8 642.2 1.13 0.192

 

 

a - BE = binding energy; b - evaluation of XPS Imp/15,2p ratio based on corrected
area using the atomic sensitivity factor of Mn and Si specific to the instrument; c - for y =
10, theoretical monolayer XPS Imp/15,2p = 1.67, (based on Kerkhof-Moulijn model)

185

CeC)2
<111>

CeO7
<200“>

(a)
MnOZ 1:138: CeO
<l()l> CCO 2 °

<400>

<422“>

 

 

(d)
(e) -" — -- -
MDOZ Mn02
<110> (2'1)
1 Mn203
<222>
1 | | |
20.0 40.0 60.0
20 [degrees]

Figure 5.13 XRD patterns of unpromoted and CeOz-promoted MnOx/SiOZ
catalyst (y = 10) calcined at 500°C: a) MnlOncea; b) Cqu; c) MnlOncen; d) CeqN; e)
Mn10n500.

 

186

As mentioned above, calcination of the catalyst to 500°C in air led to the
formation of a mixed MnOz-Mn203 phase on the SiOz support, with crystalline particles
around 16 nm for MnOz and 25 nm for anO3. For the fresh unpromoted catalyst (Table
5.4, column 4), the anp XPS spectrum shows that Mn is present as mixed Mn3+/Mn4+
with an asymmetric profile which can be related to a mixed anp spectrum from MnOz
(Figure 5.8 c) and Mn203 (Figure 5.8 a). However, the peak is shifted more toward the
BE of Mn“ species (642.5 eV), indicating a higher Mn02 content. This observation is in
good agreement with the XRD data for Mn10n500 catalysts (see Figure 5.10 and Table

5.2) which show diffraction patterns specific to MnOz and Mn203 species, with R = 0.89.

Evaluation of the MnOZ crystalline particle size (3) from the patterns observed in
the XRD spectra of the CeOz-promoted catalyst led to values of 6.7 nm for MnlOncea
catalyst and 13.8 nm for the MnlOncen catalyst, significantly lower than in the case of
unpromoted Mn10n500 catalyst. This fact may be attributed to the interaction of the
Ce02 promoter with Mn during impregnation which precludes formation of large MnOz
crystallites. The fact that Mn is present on the CeOz promoted catalysts just as Mn02 can
be explained by the better Mn dispersion and consequently to the better contact with the
oxygen from air during calcination which allowed the oxidation process to occur to the
highest oxidation state of the Mn (4+). For the cerium promoted catalyst, the XPS
spectra of the anpm shows symmetric peaks with binding energy located at 642.2 eV
(see Table 5.4, column 4), characteristic of Mn“. XPS dispersion data presented in Table
5.4, columns 5 and 6, confirms the better Mn dispersion on the promoted catalysts, the

IMan/ISin = 0.286 for Mn10ncea catalyst, double value in comparison with that for

187

Mn10n500 catalyst. From XPS data, the difference in Mn dispersion between the
MnlOncen and Mn10n500 samples is not significant.

Additional information about the cerium promoted catalysts were obtained from
TPR data. Figure 5.14 presents the TPR profiles for the unpromoted (5.14 a) and CeOz-
promoted catalyst (5.14 b-c) in comparison with the profiles for the Mn-free CeOz-
promoted Si02 (5.14 d-e). The TPR profile of the Mn10n500 contains two peaks and
corresponds to the two step reduction of Mn02 (see explanations above). The profiles for
the CeOz-promoted SiOZ catalysts were explained elsewhere (see Chapter 4). These TPR
data are presented in Figure 5.14 as references. The TPR profiles of the MnlOncea
(Figure 5.14 b) and MnlOncen (Figure 5.14 c) are very different from the reference
profiles, though they both show peaks for the reduction of the two metal oxides (Mn and
Ce) present on the catalysts. One of the main differences is the position of the peak
corresponding to the MnOz reduction to Mn203, which is shifted to 314°C, lower than
that observed for the unpromoted catalyst. Shifts in the peaks of the TPR profiles from
the value observed for pure materials have been reported in the literature for MnOx
species and attributed either to the formation of mixed oxides when more than one metal
was present on the catalyst surface [42] or to a catalytic effect of the second metal (Pt or
Pd) over the oxidation of Mn [43]. In other cases, these types of shifis in the TPR profile
were related to a chemical interaction with the support or to smaller particle size [27-28].
Consequently, the shift of the peak at 314°C may be due to smaller MnOz particles and

probably to an interaction between Ce and Mn. There is no shift in the second peak at

 

188

441°C. A new clear peak can be observed at 497°C for MnlOncen catalyst. This peak

overlaps with that from cerium, in the case of the TPR profile for the MnlOncea catalyst.

 

 

 

 

 

 

 

 

 

344
441
314
(a) . . - - a - -
497 72.
(b)
(C) _‘ - M.
8 9
“1).- --. a: 610
" 475
(e) - __ f:
0 200 400 600 800 - 930
Temperature [°C]

Figure 5.14 TPR profile of the unpromoted and CeOz promoted MnOx/SiOZ and
SiOz catalysts: a) Mn10n500; b) MnlOncea; c) MnlOncen; d) CeqN; e) Cqu.

 

189

This peak overlaps with that from cerium, in the case of the TPR profile for the
MnlOncea catalyst. It was found that for catalysts calcined at high temperature (25 75°C),
a peak at 519°C is related to the reduction of Mn3O4 to MnO. Thus, the assignment of
the peak at 497°C for the MnlOncea and MnlOncen catalysts should be the same, but
considering the shift due to Ce-Mn interaction. A similar shift was observed for the bulk
cerium reduction peak at 879°C which appeared in the MnlOncen catalyst at 726°C.
There is no clear shift in the case of the cerium peak at around 610°C for the MnlOncea
catalyst.

More information about the possible Mn-Ce interaction from the CeOz promoted
MnOx/SiOz catalyst was obtained from EPR data. EPR spectra of the cerium-promoted
catalysts are presented in Figure 5.15 a-d. Cerium-promoted SiOz catalysts prepared
from Ce4+-nitrate (CeqN, Figure 5.15 d) showed the presence of Ce3+ ions as well as
paramagnetic oxygen species (02' or O') stabilized in the CeOz lattice formed on the SiOz
surface. In contrast, when Ce4+-methoxyethoxide (Cqu, Figure 5.15 b) was used as
precursor, only the surface superoxide oxygen resonances were observed. Detailed
explanations about the EPR spectra interpretation for the Cqu(N) catalysts were given in
Chapter 4. The presence of manganese on the catalyst surface (MnlOncea, Figure 5.15 3)
led to a signal containing six hyperfine peaks (noted on in Figure 5.15 a, from 1 to 6).
Similar EPR spectra were reported in the literature for unsupported Mn/Ce mixed oxides
and attributed to the manganese nucleus (1 = 5/2 for 55Mn) interaction with the
Mn2+(Mn4+) spin, from ions located in sites with distorted octahedral symmetry or to Mn-

dimmers from the mixed oxide [44-48]. Considering the fact that XRD data show small

 

190

Mn02 crystallites on the MnlOncea catalyst and the anpm XPS binding energy is
situated at 642.3 eV (specific to a Mn4+ species), the mixed MnCer species likely

contain Mn4+ rather that Mn2+ ions incorporated into the mixed crystals.

g =2.025 3 =2-011

g =1.952 Cr?“ in C802

"\1

_o\

o
n-iq

 

 

g =1 .932 Ce form A

(C)

 

twm‘“

 

superoxide 0'2

on surface vacancies

l I | | I

2800.0 3000.0 3200.0 3400.0 3600.0
Magnetic field [Gauss]

Figure 5.15 EPR spectra for the CeOz-promoted MnOX/SiOz catalysts: a)
MnlOncea; b) Cqu; c) MnlOncen; d) CeqN.

191

If Mn2+ species are present in the catalyst structure, they would be located in the bulk part
of the crystal lattice. The EPR spectrum of the MnlOncea catalyst indicates an
attenuation of the signals characteristic to oxygen from surface superoxides (02’). No
Ce3+ defects were observed in the MnlOncea catalyst, consistent with XPS data which
indicated the presence of cerium on the surface as Ce“. It appears that the amorphous
Ce02 formed on the Cqu promoted catalyst is more reactive in solid state reactions with
Mn, favoring the formation of the Mn/Ce mixed crystal. The formation of the mixed
Mn/Ce oxide is not observed in the case of the Mnl Oncen catalyst. As observed in Figure
5.15 c, the EPR spectrum of the MnlOncen sample looks like that of the CeqN support.
The EPR spectrum of a physical mixture of the CeqN and MnOz sample looks similar to
the spectrum obtained for Mnl Oncen catalyst. From these data it seems that in the case of
MnlOncen catalyst, Mn02 and Ce02 crystals exist mostly separately on the SiOz support.
The formation process of mixed Mn/Ce oxide phase is significantly reduced. The low
CeOz dispersion with large crystalline particles precludes a good interaction with MnOz
formed on the Si02 surface during MnlOncen catalyst preparation, leading to the
formation of separate MnOz and CeOz crystalline phases. The EPR spectrum of
MnlOncen catalyst (Figure 5.14 c) still shows the presence of Ce3+ (g 1 = 1.952) lattice
defects from the Ce02 crystalline structure as well as the surface superoxide oxygen
species 02' (g,, = 2.025 and g 1 = 2.011). These data are consistent with that obtain from

XRD analysis for the MnlOncen catalyst which indicated Mn02 crystalline particles with

d = 13.8 nm (low MnOx dispersion, see Table 5.3).

 

192

Based on all the bulk and surface characterization data presented in Chapter 4
about the preparation of the CeOz-promoted S102 supports and in this chapter about the
unpromoted and CeOz-promoted MnOx/SiOz catalysts, the variation in their structures

can be schematically summarized as follow (see Figure 5.16).

(a) Ce8N Catalyt

   

+MnOx
H Calcined impregnation
0H ‘0 at 500"C
5102 o H _"Ce(OH)x
O
H

. 3102 " ~ '1 Crystalline CeOz . Crystalline M110x

(h) Ce8A Catalyst
Ce (0 4.11))“l _‘ g- + MnOx

Calcined
at 500°C

impregnation 4:13. ”

  

$02 0 O-AR

 

12:13 Amorphous C eO2 9"? Amorphous CeMnO2

Figure 5.16 Schematic representation of unpromoted and Ce02 promoted
MnOx/SiOz catalysts.

From Figure 5.16 it can be concluded that the impregnation method used to
prepare the CeOz-promoted catalysts strongly affects the MnOx/SiOz catalysts’ structures.
The interaction of the support with the cerium precursor determined a significant change
in CeOz promoter structure with consequences on the next step of the catalyst
preparation. All the results obtained in this study are easily reproducible. The

modifications in the catalysts’ structures observed during preparation can be accurately

193

monitored by controlling the preparation parameters (precursor concentrations,
temperature, impregnation technique, etc).

Catalyst activity. Table 5.5 present the catalytic activity data in terms of reactant
conversion and selectivity, for the unpromoted and CeOz-promoted MnOx/SiOz catalysts

for the oxidative dehydrogenation of ML at 200°C and 23HPy at 260°C, respectively.

Table 5.5 Conversions and selectivities for ML and 23HPy catalytic oxidative
dehydrogenation on unpromoted and CeOz-promoted catalysts

 

 

 

 

 

 

 

 

 

 

0rd. Catalyst Dihydro-Pyrazine Methyl Lactate
# at 200°C [%] at 260°C [%]
conversion selectivity conversion selectivity
1 SiOz 100 35.5 37.7 12.6
2 Mn10n400 97.9 61.9 42.1 10.3
3 ‘Mn10n500 98.1 51.3 33.5 16.7
4 Mn10n750 82.8(100) 66.9(55.4) 31.4 15.6
5 Mn14n500 98.9 65.4 36.8 13.6
6 Cqu 100 2.4 48.1 7.3
7 Mnlsicea 85.5(100) 48.7(41 .6) 43.8 5.4
8 CeqN 100 5.6 25.4(35) 29.9(21.7)
9 Mnlsicen 93(100) 52(48.4) 35.2 21.9

 

 

 

 

 

 

As observed from Table 5.5, column 2, the conversion at 200°C of 23HPy by
oxidative dehydrogenation to 23Py reaches 98 % with selectivity as high as 62 %.
Temperature plays an important role in the conversion value. Tests reactions performed
at low temperatures (130°C and 150°C) indicate conversions of only 3-5%. No 23Py

product was detected at these low temperatures. It seems that the catalysts start to

194

function only at temperatures above 160°C. At temperatures above 200°C, most of the
reactant or formed products are burned or converted to C0, C02 and H20. To properly
compare selectivities for various catalysts, the values considered should correspond to a
same conversion. In parenthesis are presented the calculated selectivity values at full
conversion (100%) for the tested catalysts. Catalysts with higher concentrations of Mn
(Mn14n500) and the MnOz (Mn10n400) proved to be more active catalysts than those
with anO3 phase for the oxidative dehydrogenation of 23HPy to 23Py. Consequently,
the Mn10n400 with higher MnOz content was found to be the most active in 23Py
formation (conversion of 98% with 62% selectivity). Catalysts with higher Mn loading
show higher catalytic activity. The catalysts calcined to higher temperatures or those
with a higher anO3 concentration show lower conversions and selectivities. It was
found that feeding only the product (23Py) at 200°C, around 16% of the amount passing
through the reactor have been burned. Consequently, the selectivity in 23Py might be
improved by optimizing the reaction temperature and thus, minimizing the 23Py burning
process. In addition, it was observed that lowering the concentration of the 23HPy in the
ethanol from the reaction feed decreases the burning or side reaction processes. These
observations are in good correlation with other studies implying large organic molecules
oxidative dehydrogenation which indicated that low concentrated feed (maximum 10%)
led to high conversion and selectivity in the desired product [22-24]. On the other hand,
it appears that the Ce-promoted catalysts are more active toward total oxidation than

selective oxidation, a large quantity of carbon deposition being observed.

195

In the case of ML to MP oxidative dehydrogenation, the MnlOncea catalyst (in
which MnCer mixed oxide species formation was observed, Mn making less accessible
for the catalysis process) shows low activity. In the case of MnlOncen catalyst in which
Mn02 deposits are observed to be separated from CeOz (like a physical mixture) higher
activity is found due to the doubled oxidation effect of the two active oxide species,
MnOz and CeOz, respectively, and to the presence of more surface superoxide species
(Of). Despite th relative good catalytic activity data for the CeOz-promoted MnOx/SiOz
catalysts, the structural promotion effect is little observable in these test reactions except
in the case of MnlOncen catalyst used in ML oxidative dehydrogenation. Analysis of the
used unpromoted and CeOz promoted MnOX/SiOz catalyst may reveal if cerium can act as
a textural promoter.

Used unpromoted and CeOz-promoted catalysts. The used catalysts from ML
oxidative dehydrogenation reaction were analyzed by XRD, TPR, and XPS. Diffraction
patterns obtained for these catalysts are presented Figure 5. 17.

In the case of the used Mn10n500 catalyst, new patterns are observable (Figure
5.17 a) in comparison with the fresh catalyst (see Figure 5.6 b). These patterns were
identified based on literature data and analysis of reference compounds, and thus,
attributed to Mn3O4 (MnOanzO3) species formed during the catalytic process [40]. In
the oxidation reaction, manganese oxide species act as oxygen donors, as a consequence
Mn02 acts as an oxidation agent, but is reduced to Mn3O4. On the used catalyst, MnOz
can be regenerated by recalcination at 500°C in air. To further confirm the Mn

transformation during the catalytic process, TPR analysis of the used catalyst was

 

196

undertaken and the result was compared with the fresh catalyst or standard material.
Figure 5.18 a shows the TPR profile of the used catalysts in comparison with a fresh

catalyst calcined at 650°C (Figure 5.18 b) and calcined at 500°C (Figure 5.18 c).

 

 

 

 

 

Mn 30 4
>
Mn3 O4 <112 .
<101> Mn3 O4 Mn} 04
(c) .
(b)
|
(a)
I I l I I
20.0 60.0

40.0
20 [degrees]

Figure 5. l7 XRD patterns of used unpromoted and CeOz-promoted catalysts: a)
Mn10n500; b) Mnl Oncen; c) MnlOncea.

 

197

It is known from the previous data (see Figure 5.11 ) that the TPR profiles of the
Mnyn500 catalysts are modified with calcination temperature due to the change in the
oxidation state of manganese. Supported MnOx show a TPR profile similar with that of
the unsupported MnOx, reported in the literature [9]. During the reduction process,
Mn304 and MnO phases are formed. As mentioned above, the peaks at 344°C and 441°C
were attributed to the reduction of the mixed MnOZanzO3 phase to Mn304 in a first step

followed by the complete reduction of Mn304 to MnO (Figure 5.11 b).

 

 

519
445
(a)
'5
a
c 344
.2
E“
a (b)
E
O
U
I"
441
497
(C)
0 200' ' '400' i '600' ' 800' 950

Temperature [°C]

Figure 5.18 TPR profiles of the fresh and used unpromoted MnOx/SiOZ catalysts:
a) used Mn10n500; b) fresh Mn10n650; c) fresh Mn10n500.

198

When the sample is calcined at 650°C, most of the manganese is in the form of Mn203
with highly crystalline particles. Consequently. the main peak in the TPR profile is
situated at 519°C, higher than in the case of smaller anO3 particles (441°C). The small
shoulders at 344°C and 445°C correspond to traces of small crystalline particles of Mn02
and Mn203. The TPR profile for the used Mn10n500 catalyst (Figure 5.18 a) is similar
with that for the Mn10n650 catalyst. In this case, the peak from around 519°C can be
attributed to the reduction of Mn3O4 to MnO. The TPR profile for the used Mn10n500
catalyst does not show the peak at 344°C corresponding to MnOz butretains that at 445°C
corresponding to traces amount of Mn203. These data are consistent with those observed
in the XRD spectrum presented in Figure 5.13 e. for used Mn10n500 catalyst. which
indicates patterns specific to Mn203 and Mn3O4 phases.

The XRD spectra of the used CeOz-promoted MnOX/SiOz catalysts are presented
in Figure 5.17 b and 5.17 c. Compared with the fresh catalysts (see Figure 5.13 a, c), the
XRD patterns of the used catalysts remain essentially unchanged. Low intensity XRD
patterns specific to Mn3O4 species are observed in the case of MnlOncen catalyst (Figure
5.17 b) but they are not found in the XRD patterns of the MnlOncea catalyst (Figure 5.17
c). It appears that the presence of the C602 promoter affects the resistance of MnOz to
reduction during the catalytic process, providing oxygen to manganese and maintaining
its oxidation state unchanged. On the other hand, the CeOZ promoter can also play an
important role in oxygen mobility on oxidation catalysts, by participating into an oxygen

exchange process [27, 41]

199

The observations from TPR and XRD data are also in good agreement with the

XPS data presented in Figure 5.19 c-e.

640.8
642.5

 

(C)

(d)

(e)

l | 1
644.0 640.0 636.0

Binding Energy [ev]

Figure 5.19 anpm XPS spectra of the used unpromoted and CeOZ-promoted
catalysts in comparison with standard MnOx materials: a) anO3 (Mn304); b) Mn02; c)
MnlOncea; d) MnlOncen; e) Mn10n500.

 

200

The anpm XPS spectrum for used Mn10n500 catalyst clearly shows a shift in the
binding energy of Mn2p3,2 peak at 642.5 eV observed in the fresh catalyst (see Figure 5.19
e) characteristic to Mn4+ species toward a binding energy of 640.8 eV characteristic to
Mn3+an2+ [34]. It is hard to identify a signal specific to the Mn2+ species due to the
overlap in binding energy value with Mn3+ species. Literature data show binding energy
values for anp for Mn3O4 species in the 641.3 -641.4 eV range, and for Mn203 species
in the 641.3-641.9 eV range [9, 34-35]. For the Ce02 promoted catalysts, both
MnlOncen and Mn10acea, the binding energy of the anpm peak is situated at 642.5 eV,
specific to Mn“ and similar with the fresh catalyst. The anpm peak for the MnlOncen
catalyst shows a degree of asymmetry which might be an indication of the presence of
Mn3+an2+ species. Due to the low S/N ratio, a clear deconvolution of the data was hard
to obtain. Despite this inconvenience, this observation is consistent with that from XRD,
which shows traces of XRD patterns specific to the Mn3O4 phase. The anp XPS spectra
for used CeOz-promoted MnOx/SiOz catalyst, presented in Figure 5.19 c-d, shows a weak
signal with the major peak located at BB = 642.5 eV corresponding to Mn4+ [22-25].

Conclusions

1) For the fresh catalyst, XRD and XPS analysis of the unpromoted catalyst
calcined at 500°C have indicated that manganese was present mostly as crystalline Mn02
and anO3. The ratio between the two oxides and MnOx particle size is a function of Mn
loading and calcination temperature used during catalyst preparation.

2) The presence of Ce02 favors the dispersion of manganese oxides deposited on

SiOz support during calcination by forming mixed Mn-Ce oxides (no diffraction pattern

 

201

specific to MnOx observed). The Cqu catalyst impregnated with Mn-nitrate clearly
shows the formation of mixed MnCer species, a phenomenon which was not observed
in the case of the CeqN catalyst. Manganese is present mostly as Mn4+-species on the
fresh Mn/Ce/SiOz catalyst.

3) Supported MnOz species proved to be more active catalysts than Mn203 for
23HPy oxidative dehydrogenation to 23Py. Consequently, the Mn14n500 with higher
Mn loading and Mn10n400 with higher MnOz content was found to be the most active in
23Py formation. Catalysts with higher Mn loading show higher catalytic activity. The
Ce-promoted catalysts are more active toward total oxidation than for selective oxidation.
In the case of MnlOncen catalyst in which MnOx deposits are observed to be separated
from Ce02 (physical mixture) higher activity is seen due to the doubled oxidation effect
of the two active oxide species, Mn02 and CeOz, respectively. The formation of mixed
MnCer phase in the Mnl Oncea catalyst, probably determined a poor catalytic activity in
ML and 23HPy oxidative dehydrogenations.

4) Analysis of the used catalysts showed that the manganese oxidation state on
unpromoted catalysts degenerates from Mn4+ to Mn2+, leading to the formation of Mn3O4
species. Formation of Mn3O4 observed on the used catalyst might be responsible for
decreasing the catalytic activity of the unpromoted MnOx catalysts. The presence of
CeOz on the MnOx/SiOz catalysts precludes the formation of Mn3O4 during catalyst
utilization. This effect is more evident in the case of Cqu-promoted catalyst. On the
CeOz-promoted MnOx/SiOz catalyst, manganese remains dispersed with a structure
Similar to that in the fresh catalyst. However. in the case of the CeqN—promoted

MnOx/SiOZ catalyst, small amounts of Mn3O4 are still observable. This difference was

 

 

 

 

202

attributed to the formation of “islands” of MnOx and CeOz, without a chemical
interaction between then, which allows the formation Mn3O4 from MnOz. Thus, CeOz
acts as a textural rather than a structural promoter, helping to preserve the MnOz structure

during an catalytic oxidative dehydrogenation process.

 

 

 

PS”

91”.“.0‘1"

ll.

12.
l3.

14.

15.

16.
l7.

l8.

19.

20.
21.

22.
23.
24.
25.
26.

27.

203

References

Cavani F.; Trifino, F. Appl. Catal. 1995, I33, 219.

Akiyama, T.; Enomoto, Y.; Shibamoto, T. J. Agric. Food Chem. 1978, 26(5),
1176-1179.

Craciun, R. Masters Thesis, Michigan State University, East Lansing, USA, 1996.
Sugiyama, S.; Sogabe, K.; Miyamoto, T.; Hayashi H.; Moffat, J. B. Catal. Lett.
1996, 42, 127-133.

Kim, J. J.; Weller, S. W. Appl. Catal. 1987, 33, 15.

Mooi, J .; Selwood, P. W.; J. Am. Chem. Soc. 1952, 74, 2461.

Strohmeier, B. R.; Hercules, D, M. J. Phys. Chem. 1984, 88, 4922.

Baltanas, M. A.; Stiles, A. B.; Katzer, J. R. Appl. Catal. 1986, 28, 13.

Kapteijn, F.; van Langeveld, D.; Moulijn, J. A.; Andreini, A.; Vuurrnan, M. A.;
Turek, A. M.; Jehng, J. M.; Wachs, I. E. J. Catal. 1994, 150, 94-104.

Kieffer, R.; Kiennemann, A.; Rodriguez, M.; Bemal, S.; Rodriguez-Izquirdo, J.
M. Appl. Catal. 1988, 42, 77-89.

Usmen, R. K.; Graham, G. W.; Watkins, W. L. H.; McCabe, R. W. Catal. Lett.
1995, 30, 53.

Jun-Ying, Y.; Swartz, W. E. Jr. Spectrsoc. Lett. 1984, 17, 331.

Ledford, J. S.; Houalla, M.; Proctor A.; Hercules, D. M. J. Phys. Chem. 1989, 93
6770.

Shelef, M.; Haack, L. P.; Soltis, R. E.; DeVries, J. E.; Logothetis, E. M. J. Catal.
1992, 137, 114.

Ledford, J.S.; Kim, Y. M.; Houalla, M.; Proctor, A; Hercules. D. M. Analyst,
1992, 11 7, 323.

Yamashita, T.; Vannice, A.; J. Catal. 1996, I63, 158.

Imamura, S.; Shono, M.; Okamoto, N.; Hamada, A.; Ishida, S. Appl. Catal. A,
1996, 142, 279.

Ivanova, A. S.; Dzisco, V. A.; Moroz, E. M.; Nosckova, S. P. Kinet. Kata]. 1986,
2760,285.

Baltanas, M. A.; DeCanio, S. J .; Katzer, J. R.; Dybowski, C. Acta Chim. Hung.
1986, 118(4), 285.

Craciun, R.; Dulamita, N. Catal. Lett. 1997, 46(3,4), 229-234.

Akiyama, T.; Enomoto Y.; Shibamoto, T. J. A gric. Food Chem. 1978, 26(5),
1176-1179.

Ai, M.; Ohdan, K. Appl. Catal. A, 1997, 150, 13-20.

Ai, M.; Ohdan, K. Chem. Lett, 1995, 405.

Sugiyama, S.; Shigemoto, N.; Masaoka, N.; Suetoh, S.; Kawami, H.; Miyaura, K.;
Hayashi, H.; Bull. Chem. Soc. Jpn. 1993, 66(5), 1542-1547.

Pestman, R.; Koster, R. M.; van Duijne, A.; Pieterse, J. A. Z.: Ponec, V. J. Catal.
1997, 168, 265-272.

Zotin, F. M. Z.; Toumayan, L.; Varloud, J .; Perrichon, V.; Frety. R. Appl. Catal.

A, 1993, 98, 99-114.

Hurst, N. W.; Gentry, S. J .; Jones, A. McNicol, B. D. Catal. Rev. -Sci. Eng. 1982,
24(2), 233-309.

 

28.

29.

30.

31.

32.

33.
34.

35.

36.
37.
38.
39.
40.

41.
42.
43.
44.
45.
46.
47.

48.

204

Bruce, L. A.; Hoang, M.; Hughes. A. E.; Tumey, W. Appl. Catal.A, 1996, 134.
151-362.

Che, M.; Giamello, E. in Spectroscopic Characterization of Heterogeneous
Catalysts, Part B: C hemisorption of probe molecules, Ed. Fierro, J. L. G.
Elsevier, Amsterdam, 1990, p. BZ65-B283.

Soria, J .; Conesa, J. C.; Martinez-Arias, A.; Coronado, J. M. Solid State Ionics,
1993, 63-65, 755-761.

Martinez-Arias, A.; Soria, J .; Conesa, J. C.; Seoane, X. L.; Arcoya, A.; Cataluna,
R. J. Chem. Soc. Faraday Trans. 1995, 91(11), 1679-1687.

Soria, J.; Martinez-Arias, A.; Conesa, J. C. J. Chem. Soc. Faraday Trans. 1995,
91(11), 1669-1678.

Martinez-Arias, A.; Soria, J .; Conesa, J. C. J. Catal. 1997, 168, 364-373.
Moulder, J. K.; Stickle, W. F.; Sobol P. E.; Bomben, K. D. Handbook of X-ray
photoelectron spectroscopy, Perkin-Elmer Co. 1992, p. 40-87.

Barr, T. L. Modern ESCA: the principles and practice of X-ray photoelectron
spectroscopy, CRC Press, Boca Raton, Florida, 1995.

Kerkhof, F. P. J. M.; Moulijn, J. A. J. Phys. Chem. 1979, 83, 1612-1619.
Scofield, J. H. J. Electron Spectrosc. Relat. Phenom. 1976, 8, 129-137.

Penn, D. R. J. Electron Spectrosc. Relat. Phenom. 1976, 9, 29-40.

Gates, B. C. Catalytic Chemistry, Wiley&Sons, Inc. New York, 1992, p 276-348.
McClune, W. F. Powder diflractionfile, Inorganic phases, International center
for diffraction data, Swarthmore 1983, p. 279, 599, 730. 731.

Imamura, S.; Shono, M.; Okamoto, A.; Ishida, S. Appl. Catal A, 1996, 142,
279-288.

Wollner, A.; Lange, F .; Schmetz, H.; Knozinger, H. Appl. Catal. A , 1993, 94.
181-203.

Ruchenstein, E.; Pulverrnacher, B.; J. Catal. 1973, 29, 224-245.

Chan, S. 1.; Fung, B. M.; Lutje, H. J. Chem. Phys. 1967, 47, 2121-2130.
Breivagel, F. W.; Sarkissian, V. J. Chem. Phys. 1968, 48, 2442-2445.

Schreurs, J. W. H. J. Chem. Phys. 1978, 69, 2152-2155.

Rubio, O. J. Munoz, P.; Boldu, O. J .; Chen, Y.; Abraham, M. M. J. Chem. Phys.
1979, 70, 633-638.

Abi-Aad, E.; Bennani, A.; Bonnelle, J. P.; Aboukais, A. J. Chem. Soc. Faraday
Trans. 1995, 91, 99-104.

CONCLUDING REMARKS

A combination of data regarding catalyst preparation, structural characterization,
catalytic activity and mechanistic study related to catalytic conversion of lactic acid to
various commodity and specialty chemicals have been presented in the above material.
From these results, a set of concluding remarks can be defined:

1) The yield and selectivity of phosphate-catalyzed lactic acid conversion to 23P
is a function of the sodium phosphate species from the Si/Al support; conversion up to
20% in lactic acid with selectivity of 30% in 23P was obtained on Na3PO4 supported on
Si/Al catalyst; lower yields and selectivity in 23P were obtained in the case of
NaHzPO4/Si/Al catalysts; this fact was attributed to acidity of the catalyst, reflected by
high yield and selectivity in acetaldehyde; sodium polyphosphates species are formed on
the catalysts surface often exposure to the reaction feed; 3 IP-NMR relaxation time
experiments help to differentiate various type of sodium phosphate species present on the
Si/Al support and to evaluate the extent of sodium phosphate polymerization;
regeneration of the active sodium phosphate species, by recalcination at 450°C can be
done only partially;

2) DRIFT spectroscopy has identified the formation of sodium lactate on the
catalyst surface; this intermediate may play an important role in the mechanism of

catalytic lactic acid conversion to 23F;

205

h

 

206

3) The presence of Ba(OH)2 on the SiOz surface may improve thermal resistance
to the NaOH supported catalyst but affects negatively the selectivity of 23P formation
from lactic acid;

4) A small decrease in the selectivity towards 23P with a corresponding slightly
increase in the selectivity towards acetaldehyde is observed for the high surface area 8102
support in comparison to the low surface area, SiOz; this was attributed to a higher
surface acidity for the former support;

5) the yields and selectivities of lactic acid conversion toward 23P is a function of
the loading and type of the alkali metal salt used to prepare the MOH supported catalysts;
the yield and selectivity toward 23P increase in the order Ba < Na < K < Cs; optimized
conditions (300°C, 0.5 MPa, 1-2 mmole MOH/g 8102, etc) and catalysts give 23P yields
as high as 42% with selectivity up to 80%.

6) Based on experimental results such as D labeling experiments exchange, FTIR,
D-NMR, GC-MS and VT-MS analyzed in the light of theoretical calculations, we have
outlined and explored a detailed pathway for lactic acid condensation to 2,3-
pentanedione. Lactate salts have been identified via DRIFTS on catalyst surfaces that
have been exposed to lactic acid vapor. H/D exchange in the opposition of the lactic acid
molecule has been observed, confirming the possibility of lactate enolization under these
relative mild (buffered) conditions. This process leads to the key carbon nucleophile
required for the condensation mechanism. Intermediate species with masses consistent
with the proposed dimerization followed by loss of H20 and C02 have been observed by

VT-MS. PM3 and ab initio calculations, augmented by the SM3 solvation model. have

 

 

 

207

been used to evaluate energy variations and protonation state along the reaction sequence;
for the condensed-phase situation, all energy changes were within accessible range at the
reaction temperatures. The proposed pathway is energetically plausible and consistent
with all observations.

7) Though specifically focused on the multistep mechanism of lactic acid
condensation to 2,3-pentanedione, this work more generally illustrates the way in which
theory and experiment can begin to dissect complex processes such as catalysis and
condensed phase reactivity. Via such detailed mechanistic analyses of catalytic
conversions, we hope to evolve toward a day when rational design will vastly accelerate
discovery of novel catalysts and reaction paths.

8) The chemical reactivity of the precursor and impregnation technique used for
preparation of a Ce-based S102 promoted catalysts plays an important role to the its
structure and catalytic activity. It was found that a Ce(NO3)4(NH4)2 precursor interacts
with the support through intermolecular hydrogen bonds between surface hydroxyl
groups from silica support and Ce(OH)4 formed during the impregnation and drying of
catalysts. Using the Ce-methoxyethoxide (Ce-alkoxide) precursor, the silica support
impregnation takes place through a strong grafting interaction between the surface
hydroxyl groups from Si02 and Ce-alkoxide, with elimination of methoxyethanol. After
calcination of the catalysts at 500°C, diffraction methods (XRD, TEM) indicated that
cerium is present in the Ce-nitrate derived catalysts mostly as large crystalline CeOz
particles (>90%), and in Ce-alkoxide derived catalysts as mixed amorphous/crystalline,

smaller size CeOz particles (crystalline CeOz is less than 30%). Air calcination at 500°C

208

of the grafted cerium from the silica support, led to an amorphous Ce4+ phase, while
calcination at 800°C led to a higher degree of crystallinity and larger CeOz particles.

9) Unpromoted and CeOz-promoted MnOx/SiOz catalysts were prepared and
characterized before and after their utilization in oxidative dehydrogenation processes.
On the fresh unpromoted MnOx/SiOz catalysts, manganese was present mostly as
crystalline MnOz and Mn203. The ratio between the two oxides and MnOx particle size is
a firnction of Mn loading and calcination temperature used during catalyst preparation.

10) The presence of CeOz favors the dispersion of manganese oxides deposited on
SiOz support during calcination by forming mixed Mn-Ce oxides (no diffraction pattern
specific to MnOx observed). The Ce-alkoxide derived SiOz support impregnated with
Mn—nitrate shows formation of mixed MnCer species, a phenomenon which was not
observed in the case of the Ce-nitrate derived silica. Manganese was present mostly as
Mn4+-species on the fresh Mn/Ce/Si02 catalyst.

11) Supported MnOz species proved to be more active catalysts than anO3 for 2-
ethyl-3-methyl-dihydropyrazine to 2-ethyl-3-methylpyrazine oxidative dehydrogenation.
Catalysts with higher Mn loading show higher catalytic activity. The Ce-promoted
catalysts are more active toward total oxidation than for selective oxidation.
Unfortunately, in these test reactions the CeOz promotion effect was little observable, so
Ce02 has mostly a textural promoter effect.

12) Analysis of the used catalysts showed that the manganese oxidation state on
unpromoted catalysts degenerates from Mn4+ to Mn2+, leading to the formation of Mn3O4

species. Formation of Mn3O4 observed on the used catalyst might be responsible for

209

decreasing the catalytic activity of the unpromoted MnOx catalysts. The presence of
CeOz on the MnOx/SiOz catalysts precludes the formation of Mn3O4 during catalyst
utilization. This effect is more evident in the case of Ce-alkoxide derived catalyst. On
the CeOz-promoted MnOx/SiOz catalyst, manganese remains dispersed with a structure
similar to that in the fresh catalyst. However, in the case of the Ce-nitrate derived

MnOx/SiOz catalyst, small amounts of Mn3O4 are still observable.

J.;-w

‘ 'b

v
a
.311.
.. ..

r19}?
”15:47 -
- 391;? .
LEX’
~11."
1.

r
..'

~ 4611

:24

1

 

, >. _ . ,1;.
1 , ._. _ ,
~ J'I‘iiiliughusk!
' 4* "I~"' r,“
. -Y11 Miul‘jl‘g
‘ -;:[I.\‘t(.

-- 13‘
r

11
if

.. -,1 I‘ifi'il'fi'€ . .
. ._ a"; llfiyfli

L?

I. d .A
1 I 1
SN} Jli
‘ l
1

,x, f,

411 .

1.;
‘ 3 7'»
1"

 

r,‘
A

ll
.1

I._o
v P.‘ ';' 411-5; "'1
i11'5’3."‘€" 373319163;
' {tr} stub-IV;
.

2119a -

f3; . .
rt

1.11

l f”

4 -_
. .
'l‘. rt

10.

’1' 1._’ 7'
‘ 3“?) 532111 ‘
. . 1 '.

£153

a

" 2’31 . ..
, 11"111611; 411m ;
111.331.3191 5ft“ (

4:421

1111..
Eric's
. 115,11

.,;".-.-’- ~-_ :4 .,...
. _ '1 ’ .. ._ [,rllhl'fll-
«.1 . .. ,.
134113: . 'r , 4; . 1 l. ‘I‘

 

1

at..-‘

. . T "t!

o.

cvo<o<<

-..

: 1‘ nu,
.. 11 .-.
-- '1. -1]-,.‘~
1 'II' '1':

' II 1‘1
”1" 2.1111

., ”‘1' r.

" FIE;

1

 

11':

' C .- t,
l.-.1 1"I{ I"
1~~ It .
‘1111'1 '3‘"
1-"

'5! 1115
_.rL.‘I"." '
. "1

 

,. ‘ .11
. 1,1-:..-r .. .
.‘n’l-N ' . '
lll‘l .30“ ‘ ‘ ‘

51:13”.

h_.: .' -,
15115611

1.1‘ :1
.,.

..|
[.1
r

(I

.' I‘l‘l
I‘H‘I‘l'l

' 1111
11 11‘ 1
1

} 1‘11].

1‘ 11” I,‘l‘11
‘ 1‘1‘1‘511'Illlllll IIIIU’

 

«'3'»

 

.q,~_‘ ‘ ‘2

.q...