JR. L...
A??? 2.; .
.........2....... .5... {.11
1.721451; T303: 4 oh
5 . m hi»... . 9 fin.
2“ I
.2 u :4 .- |
. . .t :2 h:r..n...5.la
I: I.~'I|“Ba4K \
~a~ 453 .3;l..‘ ‘-
5 a. 4k

1 . .1 gt, I. 2.5. ii. . .
12...“: . . . , 3...»... 33...“..... figuwcfanfiw..m
. . ., . . .. . I. ..
x ; .e , ..... 1. 1.31%.... E gs...
.- ft:
.395:

Y

‘3'" §

\

 

5"" 'V‘V'fl‘l‘lfi l

)3! i t. .3919. 5.1.3:
3-. httitavtb .531)..-
t. 035.39.. ...‘n}; x
3-. I... fut-:16“:
ii {£315.13}.

 

. t
. .ia
.muiiyraz
n. 6...... u... .a. .
.z .. a... .1.
.9... .53.... .E.
.9 2;...

.. 5‘1?!
3...). 9.13."...-
littl 2C9!

 

81..
I 7 [.1 .lttai.
. .1... it

l;..!1.!3¢i1.l. Off?!
254.15... 1L 0:. )Dltt.‘
all;

 

. f . : .. . , , . . 1...... . . .....~.....:..,
5+... , . . 3.

1|

 

”I. . i...

THE->13

1.30m

 

LIBRARY l
michigan 8:3:
Univeram l

-....-..

 

This is to certify that the

dissertation entitled

THERMAL AND PHOTOCHEMICAL REACTIONS OF FORMALDEHYDE
ADSORBED ON Cu(lOO)

presented by

TODD ROBERT BRYDEN

has been accepted towards fulfillment
of the requirements for

Ph . D 0 degree in Chemis t ry

 

 

 

Eww

QMajor professor

Date _Augna_t_9_._ZQQL_

MSU is an Affirmative Action/Equal Opportunity Institution 0-12771

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

NOVM 020322

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DateDuapes-ois

 

 

 

 

THERMAL AND PHOTOCHEMICAL REACTIONS OF FORMALDEHYDE
ADSORBED ON Cu(lOO)

By

TODD ROBERT BRYDEN

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemistry

2001

ABSTRACT

THERMAL AND PHOTOCHEMICAL REACTIONS OF
FORMALDEHYDE ADSORBED ON Cu(lOO)

By

TODD ROBERT BRYDEN

Thin films of polymers are promising candidates for a wide variety of
technological applications including corrosion protection, chemical sensing media and
photonic materials. Polymeric materials are attractive due to the potential of combining
high chemical, mechanical, and thermal stability with the ability for synthetically
tailoring electronic or optical properties. Current methods for producing polymeric thin
films (Langmuir-Blodgett (LB) and self-assembled monolayers(SAM)) have provided
few details of the polymerization reaction mechanism and thermal decomposition of the
film. Additionally, many monomer systems are amenable to neither LB and SAM
techniques nor direct adsorption due to the high reactivity of many surfaces. The use of a
co-adsorbate to block reaction on these surfaces has the potential for controlled thin film
deposition through photon and/or electron initiated polymerization.

Adsorption of formaldehyde (HZCO) on Cu(100) at 85 K formed an overlayer of
poly(oxymethylene) (POM). For coverages up to 1.06 (:0.22)x10‘5 H2CO
molecules/cm2 (9:0.69), formaldehyde polymerized to form a monolayer of disordered
POM arranged with the chain directions parallel to the surface plane. Formaldehyde
formed POM with differing chain lengths; the long (on) and short ([3) chain species. These

depolymerized to give two features in temperature—programmed desorption at

approximately 207 and 219 K, respectively. The presence of two POM species has not
been observed previously in studies of HZCO adsorption on metal surfaces.

The complex desorption kinetics observed for the CL and B POM species were
successfully modeled using equations based on the ratio of the average number of
monomers unzipped from the chain per initiation event to the length of the polymer
chain. Vibrational features were observed for the short chain (B) species at ~ 290, ~ 1020
and ~ 1120 cm'1 that can be assigned to the v(Cu-O), v(C-O) and p(CH3) modes of
oxygen and methoxy endgroups, respectively. Pre-adsorbed methanol increases the
proportion of short chain POM species by increasing the probability of termination for
the B species. Lower thermal stability for adsorbed POM was observed compared to bulk
POM and is believed to be related to the stability of the surface bound oxygen endgroup.

Saturating the Cu(100) surface with carbon monoxide (CO) inhibited the thermal
polymerization of HZCO on Cu(100). Formaldehyde weakly adsorbed on CO/Cu(100),
desorbing at 104 K, corresponding to a desorption energy of 18.2 (:08) kJ/mol.
Irradiation of the H2CO/CO/Cu(100) surface caused the molecularly adsorbed HZCO to
polymerize forming (POM). Irradiation also caused the formation of ethylene glycol.
Vibrational features observed after irradiation at 870 and 3365 cm'1 were assigned to
v(CC) and v(OH) modes, respectively, of ethylene glycol indicating that it was formed
promptly upon irradiation. The presence of ethylene glycol was confirmed by studying
its adsorption on clean and oxygen—covered Cu(100). The formation of ethylene glycol

was likely governed by geometric constraints present with the formaldehyde overlayer.

To my barber...

iv

ACKNOWLEDGEMENTS

There are so many people that deserve my thanks during the completion of this
thesis. Of course, none of the work described here could have been done without the help
and guidance of my advisor, Simon Garrett. When starting this work, I knew almost
nothing about surface science and you provided the perfect atmosphere for daily
discovery by always having your office door open for questions and discussions. From
you I’ve also learned that “try it and see” is an appropriate reason for an experiment and
that jaunty angles must be avoided at all costs when designing chambers and gas
manifolds.

I must also thank the other members of my committee, Professors Blanchard,
Baker and Swain, for listening to my ideas, answering my numerous questions and
providing constructive criticism when needed.

Inheriting a nearly empty laboratory provides the opportunity to thank all those
involved in the construction and repair of equipment. Russ Geyer deserves an award for
all the silver soldering he did for us (and leak-tight on the first try). The rest of the
machine shop (Sam, Glenn and Tom) and electronics shop (Ron, Scott and David) were
also instrumental in the completion of this work. They always seemed to comply with
my “emergency” requests.

Drs. Per Askeland and Kathy Severin deserve a special thanks for making my
transition into graduate school go smoothly. As the only remaining Ledford students,

they had a wealth of information about the department and its inhabitants.

The members of the Garrett group (Lili, Mike, Heather and Jason) were always
open to discussions running the gamut from music, MSU sports, food, and occasionally,
even science. They were often quite lively and I really enjoyed them. I should also thank
the other members of the Garrett group (John and Mike) for keeping me abreast of the
latest in fantasy sports.

Lastly, none of the work would have been possible without the help from my
family. Mom, Dad, Sue, Bob, Ann and Glenn were all instrumental in my success even
though you may not have understood what I was doing here. Barbara was a saint during
our time in East Lansing. Between working two or more jobs and putting up with my
frustration during tough experimental sessions, she deserves most of the credit for
helping me with the completion of this thesis through her never ending support. I love

you and thank you for helping me achieve this goal.

vi

TABLE OF CONTENTS

Page

LIST OF TABLES ix

LIST OF FIGURES x

INTRODUCTION 1

1.1 Polymer Thin Films 1

1.2 Topochemical Reactions 2

1.3 Surface Topochemical Reactions 3

1.4 Influence of Co—adsorbate 6

1.5 Proposed System 7

1.6 Reaction of Formaldehyde with Metal Surfaces 10

1.7 Outline of Thesis 12

1.8 References 14

Chapter 2. Experimental 20

2.1 References 29

Chapter 3. Adsorption and Polymerization of Formaldehyde on Cu(100) 31

3.1 Introduction 32

3.2 Results and Discussion 33

3.3 Conclusions 55

3.4 References 56
Chapter 4. Evidence for Two Chain Length Distributions in the Thermal Polymerization

of Formaldehyde on Cu( 100) 60

4.1 Introduction 61

4.2 Results and Discussion 62

4.3 Conclusions 88

4.4 References 90

Chapter 5. Photochemistry of Formaldehyde Adsorbed on CO-saturated Cu(100) 93

5.1 Introduction 94

5.2 Results and Discussion 95

5.3 Conclusions ‘ 116

5.4 References 117

Chapter 6. Conclusions and Future Work 120

6.1 Thermal Reactions and Control of Stability 120

6.2 Photochemical Reactions 123

6.3 References 127

vii

APPENDICES
Appendix A. Mass Spectra 130
Appendix B. Calculation of Electron Impact Ionization Cross-Sections 135
Appendix C. Electron Energy Loss Spectrometer and Operation Voltages 140

viii

Table 3.1

Table 3.2

Table 4.1

Table 4.2

Table 5.1

Table 5.2

Table B.1

Table C.1

LIST OF TABLES

Assignments of the vibrational bands (in cm") observed by
EELS for 0.5 L and 21 L exposures of HzCO on Cu(100) at 85
K. Also shown are IR data for solid poly(oxymethylene)
(POM) and H2CO.

Assignments of the vibrational bands (in cm'l) observed by
EELS for 0.7 L exposure of trioxane on Cu(100) at 85 K. Also
shown is IR data for solid trioxane.

Results of the fits to equations (4.5) and (4.8) for the data
shown in Figures 4.2 and 4.3.

Assignments of the vibrational losses (in cm") observed by
EELS for 1 ML coverages of HzCO and D2CO on Cu(100) at
85 K; also shown are IR data for solid poly(oxymethylene-hz)
(POM-hz) and poly(oxymethylene-dz) (POM-d2).

Assignments of the vibrational bands (in cm'l) observed for a
1.1 ML coverage of HZCO on CO-saturated Cu(100) after 0 and
15 minutes UV irradiation. Also shown are IR data for solid
HZCO and POM along with EELS data for POM on Ag(111).

Assignments of the vibrational bands (in cm") for the species
observed following 15 minutes UV irradiation and annealing to
270 K of a 1.1 ML coverage of HzCO on CO-saturated
Cu(100). Also shown are IR data for liquid (CH20H)2 along
with EELS data for (CHZOH)2 on O/Ag(110) and (CHZOH)2 on
O/Cu(100).

Molecular orbital constants calculated at the Hartree-Fock level
of theory with a 6-31G* basis set. These constants were used to
evaluate the electron impact ionization cross-section.

Operating voltages for the ELS3000 electron energy loss
spectrometer.

Page

39

41

71

75

102

109

137

141

Figure 1.1

Figure 1.2

Figure 1.3

Figure 1.4

Figure 2.1

Figure 2.2

Figure 2.3

Figure 3.1

Figure 3.2

LIST OF FIGURES

Schematic representation of a topochernical reaction on a solid
lattice.

Crystal structure of orthorhombic POM (o-POM). Filled circles
are methylene (CH2) units.

Projection of the o-POM b-c plane onto Cu( 100).

Schematic representation of substrate-mediated photochemistry
where adsorption of photons, by the substrate, either creates a)
subvacuum electrons (hv<work function) or b) free
photoelectrons (hv>work function).

Sample mount.

Oxygen coverage on Cu(100), resulting from the directed
dosing of O; at 300 K, as determined by XPS.

Ultraviolet/Visible spectrum of ~ 1 M NiSO4 (aq) solution used
as a filter for the medium-pressure Hg arc lamp.

Mass 29 thermal desorption spectra for increasing doses of
HzCO on Cu(100). All exposures were made at 85 K. The
inset shows the total integrated area for the features between
190 and 230 K as a function of exposure. Exposures are: a) 0.2
L, b) 0.3 L, c) 0.5 L, d) 0.6 L, e) 0.8 L, f) 1.1 L, g) 1.8 L, h) 2.2
L, i) 2.7 L.

Thermal desorption spectra for masses 2 (Hf), 18 (H20+), 28
(CO+), 29 (HCO+), 30 (H2CO+), and 31 (H2'3CO+) for a 1.0 L
exposure of HZCO on Cu(100). Each spectrum is a separate 1.0
L exposure. All exposures were made at 85 K. Spectra are
offset for clarity.

Page

13

21

25

28

34

37

Figure 3.3

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

Figure 4.1

EEL spectra for increasing exposures of HZCO on Cu(100) at
85 K. The sample was flashed to 300 K between exposures.

EEL spectra of a) 0.5 L exposure of HZCO on Cu(100) and b)
0.7 L exposure of trioxane on Cu(100). Both spectra were
acquired at 85 K.

EEL spectra as a function of annealing temperature for a 6.3 L
exposure of H2CO on Cu(100). The “as dosed” spectrum is
shown in a) while b) and c) were annealed to the indicated
temperature. All spectra were recorded at 85 K. The shoulder
at ~ 200 cm'1 on the elastic peak is the lattice mode associated
with the spectrum shown in a). This peak is absent in b) and c).

XP spectra of a) C (ls) and b) O (ls) for increasing exposures
of H2CO on Cu(100) at 85 K. The sample was flashed to 300 K
between exposures. Exposures were 0, 0.1, 0.2, 0.4, 0.5, 0.7,
0.9, 1.2, 2.1, 5.4, L. Vertical lines are drawn at 288.8 eV and
533.8 eV for the C (Is) and O (ls) regions, respectively.

Calculated coverage as a function of exposure for the
adsorption of H2CO on Cu( 100) at 85 K based on XPS C(ls)
peak areas. Equation of linear fit is y=0.898*x-0.082.

Enlarged view of the polymer features from the TPD spectra
shown in fig. 1. A constant background was subtracted from
each spectrum. Exposures ranged from 0.1 L to 1.1 L.

Mass 29 (HCO+) TPD spectra for (a) 1 ML H2CO on Cu(100),
(b) 1 ML H2CO after annealing to 209 K and cooling to 85 K
prior to TPD and (c) 1 ML HZCO annealed to 209 K and dosed
with HzCO to re-saturate the surface. A constant linear
background was subtracted from each spectrum.

xi

38

40

46

47

49

63

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

Mass 29 (HCO+) TPD spectra for a 1 ML coverage of H2CO on
Cu(100) exposed at 85 K as a function of heating rate (B). The
heating rates were (a) 1.5 Ks", (b) 3.8 Ks", (c) 5.6 Ks’1 and (d)
8.7 Ks". The solid line (—) is the raw data, the dotted lines
(---) are the fits to equations (4.5) and (4.8) and the dashed line
(----) is the sum of the fits.

Mass 29 (HCOI) TPD spectra for increasing coverages (0) of
HzCO on Cu(100) exposed at 85 K. The coverages were (a) 0.1
ML, (b) 0.3 ML and (c) 0.6 ML. The solid line (—) is the raw

data, the dotted lines (---) are the fits to equations (4.5) and
(4.8) and the dashed line (----) is the sum of the fits.

Simulated TPD spectra showing the effect of varying KIJDp.
The following constant parameters were used for all spectra: E
= 90 kJmol'l, v = 1x10”, and [3 = 5.6 Ks".

Electron energy loss spectra of (a) 1 ML HzCO on Cu(100)
exposed at 85 K and (b) 1 ML HzCO annealed to 209 K and
dosed with HZCO to re-saturate the surface.

Electron energy loss spectra of 1 ML HzCO on Cu( 100)
exposed at 85 K and annealed to (a) 140 K, (b) 194 K, (c) 209
K and (d) 300 K. Each spectrum is a separate 1 ML H2CO
dose.

Electron energy loss spectra of (a) 1 ML HzCO and (b) 1 ML
DzCO followed by a brief anneal to 140 K.

Electron energy loss spectra of (a) 1 ML HzCO and (b) 1 ML
DZCO followed by a brief anneal to 209 K.

Mass 29 (HCOI) TPD spectra after the adsorption of (a) 0.2 L
methanol (CH3OH) on the Cu(100) surface pre-covered with
0.6 ML HzCO and (b) 0.6 ML HZCO on the Cu(100) surface
pre-covered with 0.2 L CH3OH. The “0 L CH3OH” spectrum is
the same for both (a) and (b).

xii

67

68

70

72

74

77

79

83

Figure 4.10

Figure 5.1

Figure 5.2

Figure 5.3

Figure 5.4

Figure 5.5

Electron energy loss spectra of (a) 0.2 L CH3OH, (b) 0.6 ML
HZCO dosed onto the Cu(100) surface pre-covered with 0.2 L
CH3OH and (c) the spectrum in (b) annealed to 209 K.

Temperature-programmed desorption spectra for (a) m=30
(H2CO+) and (b) m=30 (H2CO+) and m=34 (D2COD+) for a 1.1
ML coverage of HzCO (DZCO) on CO-saturated Cu(100) as a
function of UV irradiation time. D2CO was used to reduce
coincident mass fragment interferences.

Electron energy loss spectra for a 1.1 ML coverage of HzCO on
CO-saturated Cu(100) for (a) 0 minutes irradiation at 85 K, (b)
15 minutes irradiation at 85 K, (c) 15 minutes irradiation
followed by an anneal to 270 K and (d) 15 minutes irradiation
followed by an anneal to 450 K. Each spectrum constituted a
separate 1.1 ML coverage of HzCO. The data were acquired at
85 K.

Temperature-programmed desorption spectra for m=31
(CH20H+) following exposures of 0.1 L, 0.2 L and 0.5 L of
ethylene glycol (CHZOH)2 on clean Cu(100). The inset is the
TPD spectrum obtained for a 0.5 L exposure on a Cu(100) pre-
covered with 0.1 ML oxygen showing a new feature desorbing
at 347 K.

Electron energy loss spectra for (a) 0.1 L exposure of ethylene
glycol (CH20H)2 on clean Cu(100) and (b) a 0.5 L exposure of
(CH20H)2 on a Cu(100) pre-covered with 0.1 ML of oxygen
followed by annealing to 270 K.

Electron energy loss spectra for (a) 1.1 ML coverage of HzCO
on CO—saturated Cu(100) irradiated for 15 minutes followed by
an anneal to 270 K and (b) a 0.5 L exposure of (CH20H); on a
Cu(100) pre-covered with 0.1 ML of oxygen followed by an
anneal to 270 K.

xiii

85

96

100

105

107

111

Figure 5.6

Figure 6.1

Figure 6.2

Figure A.1

Figure A.2

Figure A.3

Figure A.4

Figure 13.1

Electron energy loss spectra for (a) 1.1 ML coverage of DZCO
on CO-saturated Cu(100) irradiated for 15 minutes followed by
an anneal to 270 K and (b) a 0.5 L exposure of (CD20H); on a
Cu(100) pre-covered with 0.1 ML of oxygen followed by an
anneal to 270 K.

Mass=29 (HCO+) TPD spectra showing the effect of X-ray
irradiation on 1.1 ML HzCO on CO/Cu(100). Irradiation times
were a) 0, b) 84 and c) 130 seconds. The X-ray wavelength
was the Al K0t (hv=1486.6 eV) line and operated at 300 W (20
mA, 15 kV).

Mass=30 (H2CO+) TPD spectra for the UV irradiation of 1.1
ML HzCO adsorbed on a surface that had been pre-covered
with a saturation of poly(oxymethylene) (0.69 ML) for a) 0
minutes and b) 15 minutes UV irradiation.

Mass spectra of a) background vacuum prior to introduction of
HZCO and b) after a pressure of ~ 1x10'9 Torr of H2CO was
achieved within the chamber.

Mass spectra of a) background vacuum prior to introduction of
DZCO and b) after a pressure of ~ 1x10'9 Torr of D2CO was
achieved within the chamber.

Mass spectra of a) background vacuum prior to introduction of
trioxane (C3H603) and b) after a pressure of ~ 5X10'9 Torr of
trioxane was achieved within the chamber.

Mass spectra of a) background vacuum prior to introduction of
ethylene glycol ((CHZOH)2) and b) after a pressure of ~ 1x10'9
Torr of ethylene glycol was achieved within the chamber.

Electron impact ionization cross-section (051) as a function of
incident electron energy for H2CO and H20 and a comparison
to previous work done for water showing the small error
introduced by using a smaller basis set.

xiv

112

124

126

131

132

133

134

138

Figure C.l Schematic of the ELS3000 electron energy loss spectrometer. 142

XV

INTRODUCTION

1.1 Polymer Thin Films

The desire to create unique surfaces with specific chemical or physical properties
has focussed attention on monolayers and multilayers of organic molecules or polymers.
Through innovative synthetic design, polymeric materials can combine high chemical,
mechanical and thermal stability with tailored adhesion, wettability, tribological,
electronic or optical properties. Hence, macromolecular thin films are promising
candidates for a wide range of technological applications including corrosion
protectionlv2 and chemical sensing media.3

Ordered polymer films with crystalline character are especially important for
technological applications since they generally possess superior optical and electronic
properties compared to their amorphous counterparts. Device performance is strongly
dependent on defects and grain boundaries. In particular, interest in photonic
materialsflr5 field-effect transistorsé»7 and light-emitting diodes8 has focussed a great deal
of effort on the development of techniques for producing crystalline, macromolecular thin
films on solid surfaces. The polymer thin film may be deposited onto a solid surface
either pre-formed, or as a monomer adlayer followed by subsequent in situ
polymerization. Methods investigated to date include the deposition and polymerization
of unsaturated Langmuir-Blodgett (LB) films9’10 and, more recently, self-assembled
monolayers (SAMs).“'16 In these cases, the physical structure of the polymer film is
known to be influenced by the atomic arrangement of the underlying solid substrate. For

example, deposition of poly(oxymethylene) (POM), -(CH2-O)n-—, from solution onto

cleaved alkali halide single-crystals generates extended polymer chains aligned with
specific substrate crystal directions.17 The preferred directions appear to strongly
correlate with minimal lattice mismatch between the macromolecular chain and surface
unit cells.

While polymerization reactions are believed to be important in some catalytic
carbon-carbon coupling reactions,18 there are relatively few fundamental studies of small
molecule polymerization reactions on surfaces. There is some precedent for thermal
polymerization of HZCO to POM on several metal surfaces: O/Ag(110),19 Ni(110),20
Pt(111),21 Pd(111),22 O/Rh(111),23 O/Pd(lll),24 NiO(100)25 and Cu(110).26 Thermal
polymerization of acetaldehyde on O/Ag(111) has also been observed.” These
polymerization reactions, along with catalytic coupling reactions, are generally thought to
be controlled by the relative rates of diffusion of the adsorbed monomer, initiator species

and/or growing polymer chain.28

1.2 Topochemical Reactions

In contrast to the reactions described above, where surface diffusion largely
controls the rate of the reaction, solid-state, “diffusionless” polymerization reactions have
been observed. These reactions are termed topochemical because the structure of the
products can be directly related to the geometric arrangement of the reactants.29
Topochemical reactions are selective, rapid and steroespecific.3O If the product crystal
structure can be predicted from the monomer structure, such crystalline monomer to
crystalline polymer transformations are considered to be topotactic in nature.29 For both

topochemical and topotactic reactions, a minimal amount of nuclear motion is observed

and empirical rules have been developed from experimental and theoretical

investigations: reactive centers must be S 4 A apart and unsaturated moieties must be

nearly parallel.29r3'

While the ideas of topochemical reactions may only be applicable in solid
materials, it has been shown that a solid surface can exert a high degree of control over
the polymerization and subsequent deposition of a polymer thin film. In such epitaxial
polymerizations, structural information is conferred upon the first adsorbed polymer layer
by the substrate and, in some cases, polymer morphologies are observed that are

otherwise unobtainable from solution polymerization followed by crystallization.32‘35

1.3 Surface Topochemical Reactions

We hope to use the substrate to order the monomer intentionally prior to
polymerization. If the structure of the adsorbed monolayer matches closely with that of
the polymer unit cell, the probability of a topochemical reaction occurring increases. We
have chosen a system where a favorable packing geometry exists, as will be discussed
below. Many small molecules are known to form well-ordered monolayers with high
symmetry on transition metal surfaces increasing the possibility of parallel double
bonds.18 Additionally, the nearest-neighbor distance of most of the transition metals is <
4 A,36 satisfying another of the rules for topochemical reactions. This is shown
schematically in Figure 1.1 where the monomeric units are arranged on a lattice. Upon
initiation, the monomer units are allowed free rotation and are oriented for reaction to

form the polymer lattice.

 

 

 

 

Initiation <

—->

a
K)/

 

 

 

 

 

M
%/
gme

XT\

 

 

 

 

 

 

 

 

 

 

 

 

monomer lattice polymer lattice

Figure 1.1 Schematic representation of a topochemical reaction on a solid lattice.

In terms of topochemical reactions, there are several differences between an
adsorbed monolayer and a crystalline solid that will affect the polymerization reaction.
Diffusion of adsorbates on metal surfaces is typically fast (at least 2 orders of magnitude
greater than liquid diffusion coefficients) making the overlayer dynamic and less robust
than the crystalline phase.37 The fast diffusion could disrupt the monomer order prior to
polymerization, which will lessen the chance for topochemical reaction. Also, due to
interadsorbate and adsorbate-substrate interactions, unique monolayer structures may be
formed that may increase or decrease the probability of a topochemical reaction. For
example, bulk crystalline trioxane has been observed to photopolymerize topochemically
however the structure observed for monolayer trioxane adsorbed on Cu(l 1 1) (~ 15 A
separation between molecules)38 would probably preclude such a reaction from occurring
based on the empirical rules.

Mechanistic factors will also affect the polymerization reaction on a surface.
Successful creation of initiators and chain propagation along a specific direction in
accordance with the structure of the overlayer will be crucial to formation of the polymer.

The surface polymerization could suffer similar problems to solution polymerizations:

inhibition of initiation and premature termination and/or chain transfer reactions.
However, the rapid addition of monomers in a topotactic polymerization may be much
faster than termination events. Indeed, in solid-state polymerization reactions, monomer
addition may occur on a time scale as short as 10'13 5.39’40 Even in the topochemical
polymerization of crystalline formaldehyde to crystalline poly(oxymethylene), where the
monomer moves a large distance (~ 0.5 A), the addition of monomer was found to occur
every 10'5 s at 80 K."'0 This is faster than the rate seen for solution-phase free radical
polymerization where addition occurs on the time scale of 102-104 5 at room
temperature.“

Using the surface to align reactants was first shown for ZHZS —-> H2+2HS42 and
2HX —> H2+X2 (X=Cl and Br)43 on LiF(001) where photochemical product energies and
yields were markedly different than those seen in the gas-phase. The photochemistry of
HX is thought to occur through an aligned HX dimer where the close proximity of the
two HX molecules is afforded by adsorption on the LiF surface. Also, the photooxidation
of CO on Pt(779) occurs by dissociation of co-adsorbed O; at step-edges, which are
aligned to oxidize CO adsorbed at step-edges preferentially.44 The chemistry need not be
photon driven. The observed probability of non-terminal addition of deuterium atoms to
l-butene adsorbed on Cu(100) was found to be greater (~ 5x more) than that seen in the
gas-phase and was ascribed to a steric effect based on the adsorption geometry of the
molecule.45

For most of the systems which have been investigated to date, little is known
about the adsorbed monolayer structure prior to polymerization, the polymer morphology

or the details of the reaction mechanism. Moreover, the correlation between the atomic

arrangement of the substrate and polymer has not been studied systematically. We are
interested in elucidating the polymerization initiation, propagation and termination
mechanisms operative for ordered monolayers of unsaturated small molecules, and the
influence of the surface on the polymer film order, chain length, conformation and
stability. Ultimately, control of some or all of these processes may enable the design of
high-quality crystalline polymer thin films. To our knowledge, the possibility of
controlling polymerization through the use of a surface acting as a reaction "template"

(topotactic reaction) has not been investigated.

1.4 Influence of Co-adsorbate

In addition to studying polymerization reactions on clean surfaces, co-adsorbed
species offer the potential to control both the chain length and endgroups of the resulting
polymer. In solution polymerizations, chain transfer agents are used to control the
molecular weight distribution of a polymerization and determine the endgroup.41 On a
surface, the co-adsorbate may participate in the reaction through initiation and/or
termination events; in either case, the endgroup identity may be determined by the co-
adsorbate. Endgroup stability of a polymer adsorbed on a surface is known to have a
large impact on the thermal stability of the polymer. For example, bulk polyimide made
from pyromellitic dianhydride (PMDA) and oxydianiline (ODA) is stable at temperatures

up to 500 °C.""5v47 However, polyimide films made from reactive adsorption of PMDA

and ODA on Cu(110) decompose at ~ 280 °C.48949 The thermal stability of the film is

limited by the reactivity of the carboxylate endgroups bound to the surface, but the

decomposition pathway and the influence of the substrate on the endgroup stability have
not been correlated.

A co—adsorbate may also be used to control reactivity of the substrate with respect
to the monomer. Direct adsorption of the monomer is a simple alternative to LB and
SAM methods. However some metal substrates are too reactive to form the polymer thin
film through direct adsorption. For example, formation of POM films from H2CO is not
possible through LB and SAM methods and direct adsorption of HzCO on Fe(100) causes
decomposition, and not polymerization, of the adsorbed H2CO.50

Carbon monoxide is an obvious choice for use as the co-adsorbate to control the
reactivity of the substrate because it forms a well-ordered monolayer on Cu(100) at 85
K51 and is only weakly chemisorbed, desorbing molecularly at ~ 180 K52 Co-adsorbed
carbon monoxide (CO) is known to influence the chemistry observed on surfaces. For
example, co-adsorbed CO has been found to increase the stability of ethylidyne on
Ru(001),53 decrease the stability of the methyl hydrogen atoms of toluene on Ru(001),54
perturb the decomposition pathway of methylamine on Ru(001)55 and promote the

decomposition of saturated hydrocarbons on Ni(755).56

1.5 Proposed System

The system chosen for study is the adsorption of formaldehyde on Cu(100).
Formaldehyde is an ideal molecule for studying polymerization reactions due to its low
activation energy for polymerization (~ 12 kJ/mol) and the fact the reaction can be self-
initiated without the use of an initiator species. Also, the gas, solution and solid-state

reactions of formaldehyde have been studied extensively.39r40,57r53

Gaseous H2CO has a S1 (— 80 (1t* (- n) band origin at 3.49 eV57 and can
dissociate, upon UV irradiation, via two channels: H2CO -—) H2 + CO (molecular) and
H2CO —-> H' + HCO’ (radical). The molecular channel has its threshold at 3.52 eV.
Polymerization is not expected to be initiated by the molecular photoproducts. The
radical channel, whose threshold is at 3.73 eV, may initiate the polymerization through
reaction of the radical species with molecular H2CO.

Formaldehyde can be polymerized to form POM via cationic, anionic and free
radical routes.58 Poly(oxymethylene) can also be formed through ring opening of the
cyclic trimer (l,3,5-trioxane) and tetramer ( 1,3,5,7-tetraoxane) of H2CO. Both
orthorhombic (o-POM) and trigonal (t-POM) crystal structures have been observed for
POM in which the POM chains form helices.59 In t-POM, the helical conformation is
designated D(1070’9) while the o-POM conformation is D(71’).60 The crystal structure of o-

POM is shown in Figure 1.2. For t-POM, only one chain per unit cell is present.

b=765A
\ \

 

 

 

a=4J7A

38 fl
M818 8%

Figure 1.2 Crystal structure of orthorhombic POM (o-POM). Filled circles are
methylene (CH2) units.

 

 

 

 

 

c=356A

The unit cell parameters of the Cu(100) surface match well with those of
crystalline o-POM. The c dimension of the polymer unit cell (3.56 A) is only about 1 %

smaller than the bulk copper unit cell (ao=3.61 A),36 and the b dimension (7.65 A) is
about 6 % larger than 2a“. This is shown schematically in Figure 1.3 where the b-c plane

is projected onto the Cu(100) surface plane. The close match of unit cells increases the
probability for a topocherrrical reaction.

The helical conformation observed in the solid-state may not be observed on the
Cu(100) surface due to the strong adsorbate-surface interactions. The nearest-neighbor
distance for copper (2.55 A) matches closely with that of the oxygen—oxygen repeat
distance (2.49 A) for POM in a planar zigzag conformation. This conformation has not
been observed experimentally due to dipole-dipole repulsion of the neighboring C-O-C
units. The helical conformation has been calculated to be ~ 7 kJ/mol lower in energy than

the planar zigzag structure.“

 

Figure 1.3 Projection of the o-POM b-c plane onto Cu(100).

It has been shown that in the solid-state photochemistry of formaldehyde, a high
degree of order prior to irradiation is essential for long chain growth of POM.
Formaldehyde has been observed to undergo photopolymerization ('y-rays) in the solid-
state down to temperatures as low as ~ 4 K producing high molecular weight, crystalline
polymer.39r40. In contrast, no POM was detected after irradiation of amorphous films of
H2CO at 10 and 77 K and, only after doping with C12 were oligomers formed (~ 6

monomer units)!“63

1.6 Reaction of Formaldehyde with Metal Surfaces

As was mentioned previously, the adsorption of H2CO on metal surfaces has been
investigated. Non-dissociative adsorption has been observed on clean Ag(110),19r64
Ag(111),65 Au(110)66 and Zn(0001)67 at temperatures below 100 K. Reactive adsorption
of formaldehyde was found to occur on O/Ag(110),19v64 Ni(110),20 O/Zn(0001),67
Ru(001),68 Rh(111),23 Fe(100),50 Pt(l l 1),21 and Pd(111)22 with H2, CO and POM being
formed. When POM has been observed, it has been speculated the interaction of H2CO
with surface-bound oxygen forms either a forrnyl species(HCO)68 or a dioxymethylene
(H2CO2)64 species which then participates in the polymerization process. The oxygen
probably arises due to dissociation of the H2CO upon adsorption. Oxygen has the ability
to act as a nucleophile as well as create Lewis acid sites in its immediate vicinity.64r69
Both can initiate the polymerization of formaldehyde.58 Also, partial thermal
polymerization was observed on Cu(110) at 90 K but the effect of adsorbed oxygen was

not addressed.26

10

There has been one investigation on the photopolymerization of formaldehyde
adsorbed on a metal surface. Dai et al. polymerized formaldehyde adsorbed on Ag(111)
using nanosecond pulses of 355 nm radiation.65r70’7l Formation of POM was observed at
80 K for coverages from 0.1 to l monolayer (ML) of adsorbed H2CO. In the absence of
irradiation, no polymerization was observed. Also, no polymerization was observed upon
exposure to 532 nm radiation suggesting a non-thermal initiation mechanism.70 Below
60 K, no polymerization occurred after irradiation, presumably due to reduced diffusion
of H2CO or initiator species. However, polymerization did occur after these irradiated
samples were annealed to higher temperatures?1 This is indicative of a diffusion-limited
process and not a topochemical reaction.

Recent results by Dai et al. have shown laser-induced polymerization of
formaldehyde on Ag(1 11) occurs through photoexcited electrons generated within the
substrate.72’73 Substrate sub-vacuum electrons have been observed to initiate chemistry
of adsorbates as well as stimulate desorption.74‘77 This is represented schematically in
Figure 1.4a where photons (hv<work function) are absorbed by the substrate creating hot
electrons. These electrons can tunnel to empty states of the molecular adsorbate,
inducing chemistry. In Figure 1.4a, the placement of the lowest unoccupied molecular
orbital (LUMO) of H2CO is somewhat arbitrary as it is not generally known, a priori,
how the orbitals of the adsorbate line up with the surface. If photon energies greater than
the work function are used, free photoelectrons may attach to the LUMO of H2CO and
also induce chemistry. This is shown in Figure l.4b. However, for the
photopolymerization of H2CO on Ag(111), the photon energies used were less than the

work function.

11

If sub-vacuum electrons can initiate polymerization, then electrons from the
vacuum side should initiate polymerization as well. Dai et al. have found this to be
true.78 Low energy ( 0-20 eV ) electrons were used to irradiate the sample, at 80 K, with
polymerization occurring at all electron energies. The substrate sub-vacuum and vacuum
side electrons are thought to attach to H2CO forming an unstable negative ion state. This
negative ion dissociates into CH2 and O‘ on Ag(111) which is in contrast to H2CO' gas
phase dissociation into H and HCO.78 Either CH2 or 0 may be responsible for the
polymerization. In light of the thermal polymerization observed on oxygen covered
surfaces, oxygen is more likely to initiate the polymerization.

The work performed by Dai’s group indicates polymerization of formaldehyde on
Ag(l 1 l) is strongly temperature dependent. Topocherrrical polymerization reactions are
not diffusion-limited processes and thus Dai’s work should not be considered
topochenrical in nature. Also, neither the ordering of the monolayer, with respect to the
substrate, nor the resulting polymer conformation was investigated. Information
regarding both will be important for realizing some of the goals of the work described in

this dissertation.

1.7 Outline of Thesis

The first step in the work described within this thesis was to study the adsorption
of formaldehyde on Cu(100) at cryogenic temperatures (S 85 K). This is discussed in
Chapter 3. We have shown that at 85 K, H2CO spontaneously polymerizes to form a
saturated monolayer of POM.52 The POM depolymerizes, forming molecular

formaldehyde which desorbs promptly, between 190 and 220 K. A small amount of H2 is

12

Em:
——> — LUMO(1t*)

 

 

 

 

hv
Er:
— HOMO (11)
Metal HZCO
a) Creation of subvacuum electrons
followed by tunneling to LUMO of
HZCO.
C—V — LUMO (1t*)
A
vac —
hv
— HOMO (n)
EF

 

Metal H2CO

b) Creation of vacuum electrons
followed by attachment to HCO.

Figure 1.4 Schematic representation of substrate—mediated photochemistry where
adsorption of photons, by the substrate, either creates a) subvacuum electrons (hv<work

function) or b) free photoelectrons (hv>work function).

observed to desorb from the surface and is believed to arise from H2CO dissociating on
the clean Cu(100) substrate. A clean surface is obtained after heating to 300 K.

Two different POM species are formed from the thermal polymerization of H2CO
on Cu(100) at 85 K.79 This is the subject of Chapter 4. Two separate chain lengths are
produced in the polymerization and the depolymerization kinetics have been modeled
using standard depolymerization kinetic equations. The differences between the ratio of
degree of polymerization and the average number of monomer units depolymerized at
each initiation step separate the long and short chain POM species. Also, we have used
co-adsorbed methanol to preferentially form the shorter species and confirm the identity
of the endgroups.

Chapter 5 describes the use of a co-adsorbate, carbon monoxide, to control the
reactivity of the Cu(100) with respect to formaldehyde adsorption.80 A saturation
coverage of CO inhibits the thermal polymerization of H2CO on Cu(100) at 85 K. Upon
irradiation with UV photons, molecularly adsorbed H2CO reacts to form POM and
ethylene glycol, which has not been observed previously in the photochemistry of
adsorbed formaldehyde. The CO desorbs prior to POM depolymerization offering the
possibility for forming patterned polymer films through photopolymerization, followed

by desorption of monomer (from the unirradiated areas) and the CO spacer layer.

1.8 References

(1) Laibinis, P.; Whitesides, G. J. Am. Chem. Soc. 1992, 114, 9022.

(2) Zhao, M.; Liu, Y.; Crooks, R.; Bergbreiter, D. J. Am. Chem. Soc. 1999, 121, 923.

14

(3) Harsanyi, G. Polymer Films in Sensor Applications; TECHNOMIC: Lancaster,
PA, 1995.

(4) Lemoine, V.; Pocholle, J .; Le Barny, P.; Robin, P. In Molecular Nonlinear Optics;
Zyss, 1., Ed.; Academic Press: Boston, 1994; pp 379.

(5) Lytel, R.; Lipscomb, G.; Thackara, J.; Altman, J.; Elizondo, P.; Stiller, M.;
Sullivan, B. In Nonlinear Optical and Electroactive Polymers; Prasad, P., Ullrich, D.,
Eds.; Plenum Press: New York, 1988; pp 415.

(6) Katz, H.; Bao, Z. J. Phys. Chem. B. 2000, 104, 671.

(7) Xu, 6.; Bao, Z.; Groves, J. Langmuir 2000, 16, 1834.

(8) Grell, M.; Bradley, D. Adv. Mater. 1999, 11, 895.

(9) Ejaz, M.; Yamamoto, S.; Ohno, K.; Tsujii, Y.; Fukuda, T. Macromolecules 1998,
31, 5934.

(10) Shirai, E.; Urai, Y.; Itoh, K. J. Phys. Chem. B. 1998, 19, 3765.

(l 1) Ford, J. F.; Vickers, T.; Mann, G; Schlenoff, J. Langmuir 1996, 12, 1944.

(12) Sun, F.; Castner, D.; Grainger, D. Langmuir 1993, 9, 3200.

(13) Kim, T.; Chan, K.; Crooks, R. J. Am. Chem. Soc. 1997, 119, 189.

(14) Chan, K.; Kim, T.; Schoer, 1.; Crooks, R. J. Am. Chem. Soc. 1995, 117, 5875.

(15) Balasubrarnanian, K.; Cammarata, V. Langmuir 1996, 12, 2035.

(16) Batchelder, D.; Evans, 8.; Freeman, T.; Haussling, L.; Ringsdorf, H.; Wolf, H. J.
Am. Chem. Soc. 1994, 116, 1050.

(17) Mauritz, K.; Baer, E.; Hopfinger, A. J. Polym. Sci. : Macromol. Rev. 1978, 13, l.
(18) Somorjai, G. Introduction to Surface Chemistry and Catalysis; Wiley and Sons:
New York, 1994.

(19) Stuve, E.; Madix, R.; Sexton, B. Surf. Sci. 1982, 119, 279.

15

(20) Richter, L.; Ho, W. J. Chem. Phys. 1985, 83, 2165.

(21) Henderson, M.; Mitchell, G.; White, J. Surf. Sci. 1987, 188, 206.

(22) Davis, J.; Barteau, M. J. Am. Chem. Soc. 1989, 111, 1782.

(23) Houtrnan, C.; Barteau, M. Surf. Sci. 1991, 248, 57.

(24) Davis, J.; Barteau, M. Surf. Sci. 1992, 268, 11.

(25) Truong, C.; Wu, M.; Goodman, D. W. J. Am. Chem. Soc. 1993, 115, 3647.

(26) Sexton, B.; Hughes, A.; Avery, N. Surf. Sci. 1985, I55, 366.

(27) Sim, W.; Gardner, P.; King, D. J. Am. Chem. Soc. 1996, 118, 9953.

(28) Bent, B. E. Chem. Rev. 1996, 96, 1361.

(29) Thakur, M. In Encyclopedia of Polymer Science and Engineering; Wiley and
Sons: New York, 1989; Vol. 15; pp 362.

(30) Wright, J. Molecular Crystals; Cambridge University Press: Cambridge, 1987.
(31) Salcedo, R.; Sansores, L.; Valladares, A.; Likhatchev, D.; Alecandrova, L.;
Ogawa, T. Polymer 1996, 37, 1703.

(32) Yamashita, Y.; Shimamura, K.; Kasahara, H.; Monobe, K. Syn. Met. 1987, 17,
253.

(33) Lando, J.; Baer, E.; Rickert, S.; Nae, H.; Ching, S. In Initiation of Polymerization;
212 ed.; Bailey, F. E., Ed.; American Chemical Society: Washington, DC, 1983; Vol.
212.

(34) Sano, M.; Lvov, Y.; Kunitaki, T. In Annual Review of Materials Science; Tirrell,
M., Kaufman, E., Giordmaine, J., Wachtman, J., Eds.; Annual Reviews: Palo Alto, 1996;
Vol. 26; PP 153.

(35) Sano, M.; Sasaki, D.; Kunitake, T. Macromolecules 1992, 25, 6961.

(36) CRC Handbook of Chemistry and Physics, 76 ed.; CRC Press: Boca Raton, 1995-
1996.

(37) Seebauer, E.; Allen, C. Prog. Surf. Sci. 1995, 49, 265.

(38) Hofmann, M.; Wegner, H.; Glenz, A.; Woll, C.; Grunze, M. J. Vac. Sci. Technol.
A 1994, 12, 2063.

(39) Gol’danskii, V.; Benderskii, V.; Trakhtenberg, L. In Advances in Chemical
Physics; Progogine, 1., Rice, S., Eds.; Wiley and Sons: New York, 1989; Vol. 75; pp 349.
(40) Gol’danskii, V. Ann. Rev. Phys. Chem. 1976, 27, 85.

(41) Odian, G. Principles of Polymerization, 3 ed.; Wiley and Sons: New York, 1991.
(42) Bourdon, E.; Das, P.; Harrison, 1.; Polanyi, J.; Segner, J.; Stanners, C.; Williams,
R.; Young, P. Faraday Discuss. Chem. Soc. 1986, 82, 343.

(43) Cho, C.; Polanyi, J.; Stanners, C. J. Chem. Phys. 1989, 90, 598.

(44) Tripa, C.; Yates, J. J. Chem. Phys. 2000, 112, 2463.

(45) Yang, M.; Teplyakov, A.; Bent, B. J. Phys. Chem. B. 1998, 102, 2985.

(46) Karnbe, H. In Aspects of Degradation and Stabilization of Polymers; Jellinek, H.,
Ed.; Elsevier: Amsterdam, 1978; pp 393.

(47) Feger, C.; Franke, H. In Polyimides: Fundamentals and Applications; Ghosh, M.,
Mittal, K., Eds.; Marcel Dekker: New York, 1996; pp 759.

(48) Haq, S.; Richardson, N. J. Phys. Chem. B. 1999, 103, 5256.

(49) Plank, R.; DiNardo, J .; Vohs, J. Phys. Rev. B. 1997, 55, 10241.

(50) Hung, W.; Bemasek, 8. Surf. Sci. 1996, 346, 165-188.

(51) Tracy, J. J. Chem. Phys. 1972, 56, 2748.

(52) Bryden, T.; Garrett, S. J. Phys. Chem. B. 1999, 103, 10481.

(53) Sasaki, T.; Kawada, F.; Aruga, T.; Iwasawa, Y. Surf. Sci. 1992, 278, 291.

17

(54) Rauscher, H.; Menzel, D. Surf. Sci. 1995, 342, 155.

(55) Sasaki, T.; Aruga, T.; Kuroda, H.; Iwasawa, Y. Surf. Sci. 1992, 276, 69.

(56) Orita, H.; Kondoh, H.; Nozoye, H. Chem. Phys. Lett. 1994, 228, 385.

(57) Moore, C.; Weisshar, J. Ann. Rev. Phys. Chem. 1983, 34, 525.

(58) Walker, J. F. Formaldehyde, 3rd. ed.; Reinhold Publishing Corp.: New York,
1964.

(59) Tadokoro, H. In Macromolecular Reviews; Peterlin, A., Goodman, M., Okamura,
S., Zimm, B., Mark, H., Eds.; Interscience: New York, 1967; Vol. 1; pp 119.

(60) Helical symmetry is specified by the nomenclature X(2m1r/n) where X is the point
group (C or D), m is the number of turns of the chain per unit cell and n is the number of
monomer units per chain per unit cell).

(61) Tadokoro, H.; Kobayashi, M.; Mori, K. Rep. Progr. Polymer Phys. Japan 1965, 8,
45.

(62) Mansueto, E.; Ju, C.; Wight, C. J. Phys. Chem. 1989, 93, 2143-2147.

(63) Manseuto, E.; Wight, C. J. Photochem. Photobiol. A:Chem. 1991, 60, 251.

(64) Barteau, M.; Bowker, M.; Madix, B. Surf. Sci. 1980, 94, 303-322.

(65) Fleck, L.; Ying, Z.; Feehery, M.; Dai, H—L. Surf. Sci. 1993, 296, 400.

(66) Outka, D.; Madix, R. Surf. Sci. 1987, I 79, 361.

(67) Sen, P.; Rao, C. Surf. Sci. 1986, 172, 269-280.

(68) Anton, A.; Parrneter, J .; Weinberg, W. J. Am. Chem. Soc. 1986, 108, 1823.

(69) Madix, R. Science 1986, 233, 1159-1166.

(70) Fleck, L.; Feehery, W.; Plummer, E.; Ying, Z.; Dai, H. J. Phys. Chem. 1991, 95,

8428.

18

(71) Fleck, L.; Ying, Z.; Dai, H.-L. J. Vac. Sci. Technol. A 1993, 11, 1942.

(72) Fleck, L.; Howe, P.; Kim, J .; Dai, H. J. Phys. Chem. 1996, 100, 8011.

(73) Ying, Z.; Fleck, L.; Dai, H. In Laser Spectroscopy and Photochemistry on Metal
Surfaces; Dai, H., Ho, W., Eds.; World Scientific Publishing Company: Singapore, 1995.
(74) Mieher, W.; Ho, W. J. Chem. Phys. 1989, 91, 2755.

(75) Zhou, X.; Coon, S.; White, J. J. Chem. Phys. 1990, 92, 1498.

(76) Ying, 2.; Ho, W. Phys. Rev. Lett. 1990, 65, 741.

(77) Hoffman, A.; Guo, X.; Yates, J.; Gadzuk, J.; Clark, C. J. Chem. Phys. 1989, 90,
5793.

(78) Fleck, L.; Kim, J.; Dai, H.-L. Surf. Sci. 1996, 356, L417.

(79) Bryden, T.; Garrett, S. Langmuir , in press.

(80) Bryden, T.; Garrett, S. J. Phys. Chem. B. , in press.

19

Chapter 2 Experimental

Ultrahigh Vacuum Chamber (UHV). The experiments to be described were
conducted in two connected stainless steel ultrahigh vacuum chambers. Each chamber
was separately pumped by a 270 Us ion pump. The first chamber, spherical in shape,
was equipped with a 1-200 amu quadrupole mass spectrometer (QMS) (VG Masstorr 200
DX), an ion gun for noble gas ion sputtering (Phi 04-161 gun, 20-005F controller) and a
molecular leak valve (Varian 9515106). This chamber was also equipped with a dual
anode X-ray source (VG 3EXR2) and hemispherical electron energy analyzer (VG
CLAM2) for X-ray photoelectron spectroscopy (XPS). The pressure was measured by a
nude ion gauge. After being baked for ~ 48 hours at ~ 120 °C, the base pressure was
typically 2x10"o Torr. The other chamber was double 1.1-metal shielded and housed a
high resolution electron energy loss spectrometer (EELS) (LK Technologies LK3000).
The 1.1-metal shielding is necessary to reduce the stray magnetic fields present within the
chamber to < 35 mG.1

The sample was mounted at the end of a long-travel (500 mm) manipulator
capable of x, y, z, and 0 motion (Thermionics 910438NW) which was connected to the
spherical chamber. The sample mount was constructed in-house and is shown in Figure
2.1. The mount was connected in vacuum to the threaded end of the manipulator through
a Macor block. Macor is a machinable ceramic material that provided both thermal and
electrical isolation for the sample mount. Additional thermal and electrical isolation was
provided by the cryobreaks (ISI 9611004). The sample was held to the molybdenum

sample block by two tungsten clips which provided good thermal and

20

 

 

 

Top view

1 j W sample clips

/ \

TM bl k
Macor oc Cu LN2
reservoir CUHOO)
/ sample
Mo sample

Side view mount
__ m—m /
4:93:33 ee °e°\r

\
Cryobreak (ISI 9611004)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Heater filament
holes

Figure 2.1 Sample mount.

mechanical contact to the mount. The M0 mount was secured to the copper liquid
nitrogen (LN2) reservoir via a threaded molybdenum rod and nut. The sample
temperature was monitored using an E-type thermocouple placed between one of the
clips and the front face of the sample. To ensure fast response and low thermal load on
the sample, the thermocouple junction was kept as small as possible (OMEGA
Engineering, 0.13 mm diameter). The sample could be cooled to 85 K by drawing LN2
through the reservoir using a small diaphragm pump. The sample was heated indirectly
by using a tungsten filament (Alfa Aesar, 0.25 mm diameter). The heater filament was
wound (25 turns each side) around a small diameter rod and inserted into alumina tubes

(Kimball Physics Al2O3-TU-C-500) prior to being inserted into the heater filament holes

in the mount. In this fashion, the sample could be heated to > 800 K. All cryogen and
electrical lines in vacuum were covered in braided silica sheathing and connections were
made via a 2.75 in. multiport flange (MDC MMF275-5-133) attached to the manipulator.
Also, the entire mount could be electrically biased via a copper wire attached to the
mount.

Sample Preparation. The Cu(100) sample (Monocrystal, Inc. > 99.99995 %
purity) was cleaned by cycles of argon ion sputtering and annealing. The sample was
heated to 680 K, at which time argon was admitted into the chamber to a pressure of
5x10’5 torr. The ion pumps were turned off and a gate valve opened to pump the
chamber via a turbomolecular pump (Varian V-80) connected to a gas manifold system.
The sputtering was begun with the following conditions: 1000 eV, 15 mA emission, ~ 25
uA current onto the sample mount. The sample mount geometry prevented a measure of
ion current directly onto the sample. Sputtering continued for 15 minutes, at which time
the ion gun was turned off and the sample was allowed to anneal at 680 K for 10 minutes
prior to cooling to ~ 300 K for analysis. During the post-sputter anneal, the ion pumps
were turned on when the pressure was < 1><10'6 torr. The sample was considered clean
when XPS and EELS showed no contamination present.

Generation of Formaldehyde and Dosing. Gas-phase H2CO (D2CO) was
prepared by pyrolysis of paraformaldehyde (Aldrich, 95%) or paraformaldehyde-d2
(Aldrich, 99% atom) using a previously described method.2 Two Nupro stainless steel
in-line gas filters (SS-4F-7), with the sintered filter elements removed, were packed with
~ 2 g paraformaldehyde and ~ 3 g of dried MgSO4. Both materials were constrained

within the filters with glass wool that had been deactivated with trimethychlorosilane.

22

The filter assembly was attached to the gas line between the molecular leak valve and the
gas manifold by a T-connector. The leak valve, gas lines, and the manifold were heated
to ~ 100 °C to prevent condensation of the H2CO which was generated, along with water,
by heating the filter assembly to ~ 60 °C, via heating tape. The water was trapped by the
MgSO4 prior to introduction into the vacuum chamber. Residual gas analysis of the
vapor introduced into the chamber indicated ~ 6 % H2O contamination. Removal of the
drying agent, MgSO4, reduced the water content of the gas to ~ 3%. No effect on the
chemistry of H2CO on Cu(100) by the water was observed.

The majority of experiments were performed by cooling the sample to T S 85 K
and backfilling the UHV chamber with H2CO for a predetermined time (background
dosing). Exposures of H2CO are quoted in langmuirs (l langmuir = l L = 106 Torr . s)
and are uncorrected for ion gauge sensitivity. Directed dosing was accomplished by
dosing through a 1/8‘h in. OD. stainless steel tube connected to the molecular leak valve
(the gasket assembly within the leak valve was threaded for 8-32 and the tube was
soldered to a 8-32 bolt that had been drilled through). The sample was moved away from
the end of the closer and dosing was performed by opening the leak valve, while
monitoring the most abundant fragment by the QMS, until a constant pressure was
achieved. The base pressure, as monitored by the ion gauge, rose by only ~ 2x10"° Torr.
Once the QMS signal stabilized, the sample was moved under the closer for a
predetermined time. The doser-sample distance was ~ 4 mm. Based on background
dosing experiments involving H2CO, enhancement factors of ~ 300 were calculated for

directed dosing using the stainless steel tube.

23

Preparation of 0/Cu(100). Oxygen-covered Cu(100) surfaces were prepared by
closing dioxygen (AGA, UHP grade) through the 1/8th in. OD. stainless steel tube onto
the Cu(100) at 300 K. The oxygen coverage as a function of dosing time is shown in
Figure 2.2. A dose time of 300 s was used to prepare a coverage of 0.1 ML as indicated
by XPS analysis. At this coverage, oxygen forms a disordered overlayer with the oxygen
atoms in 4-fold hollow sites as determined by photoelectron diffraction.3 A single loss
was observed in EEL spectra at 335 cm'1 (corresponding to v(Cu-O)) which matches
closely with previous investigations of O/Cu(100) where a loss was observed at 340 cm'1
for a coverage of 0.11 ML.4

X-ray photoelectron spectroscopy (XPS). The surface chemical composition was
determined by XPS. All XP spectra were collected using the Al Kat X-ray line
(hv=1486.6 eV) operated at 300 W (15 kV, 20 mA) and an analyzer pass energy of 100
eV. Photoelectrons were collected at a take-off angle of 75° from the surface normal to

maximize surface sensitivity. Spectra were referenced to the Cu (2133/2) peak from clean

Cu(100) at 932.7 eV.5 Typically, photoemission data (Cu (2p), 0 (ls) and C (Is)
regions) were acquired in less than 2 minutes. In order to calculate the C(15) and O(ls)
XPS atomic sensitivity factors for our instrument, single point calibration experiments
were performed using a saturation exposure of CO (Matheson, 99.99 %) on the clean
Cu(100) surface. This system is known to form a saturated monolayer coverage of 0.57
ML (0.57 CO molecules per Cu atom) at 85 K.6 Measurement of the C (ls):Cu (2pm)
XPS peak intensity ratio for a CO-saturated monolayer 'allowed us to calculate an

absolute carbon atom concentration for any measured C (ls):Cu (2p3/2)

24

0.12

 

 

 

I
0.10 - .
0.08 -
I
A 0.06 -
_l
g XPS Analysis
G O / Cu(100)
0.04 - Directed dosing of 02
0.02 -
0.00 T

 

I"'TI""I"' 'r'
0 100 200 300 400 500
Exposure time (s)

Trrfiiilrri

Figure 2.2 Oxygen coverage on Cu(100), resulting from the directed dosing of 02 at 300
, as determined by XPS.

25

or O(ls):Cu (2p3/2) XPS ratio. This method was used to determine the absolute number
of H2CO molecules for a given exposure.

Electron Energy Loss Spectroscopy (EELS). The clean and adsorbate-covered

surface was characterized by high resolution electron energy loss spectroscopy (EELS).
In a typical EELS experiment, a monoenergetic beam of electrons is scattered off a
surface. The majority of electrons scatter elastically, but, a small fraction interact with
the surface-adsorbate system and lose quanta of energy corresponding to vibrations of the
system},7 For the experiments described here, the dominant interaction was assumed to
be dipolar. In this scattering mechanism, the electron scatters inelastically from the long-
range dipole field produced by the adsorbate. This interaction changes the momentum
minimally and, as such, the electrons that have lost energy are detected very close to the
elastic peak (specular direction).

Electron energy loss spectra were acquired in the specular scattering geometry
(9j=03=55°) with a primary electron beam energy of 6.1 eV. The elastically scattered
beam from the clean Cu(100) surface was typically 24-32 cm"1 (3-4 meV) full width at
half maximum (FWHM). Under these conditions, currents onto the Cu sample were

approximately 300 pA and count rates from the clean Cu(100) surface were >106 Hz.
From the adsorbate-covered surface, the FWHM ranged from 32-64 cm'1 (4-8 meV) with
c=0unt rates of 103-105 Hz.

Temperature-Programmed Desorption (TPD). The adsorbate-covered surface was

characterized by TPD. During TPD experiments, the adsorbate-covered surface was
heated at a linear rate (4.2 or 5.7 K-s'l) and the desorbing molecules characterized by

mass Spectrometry. The surface was positioned such that the surface normal was in

26

direct line-of-sight of the QMS. Molecules were transmitted to the mass spectrometer
through a 2 mm diameter aperture in a stainless steel shroud enclosing the ionizer. The
shroud ensured the majority of desorbing species originated from the surface and not the
sample mount. While heating, a —70 V potential was applied to the sample to prevent
electrons from the mass spectrometer ionizer inducing adsorbate chemistry.

The rate of desorption for an adsorbate is equal to the rate of change of coverage

(0) with respect to temperature (T) and can described by the Polyani-Wigner equation

Rate of desorption = :19— : V6 exp - Ed (2.1)
dT is kBT

 

 

where v is the preexponential, n is the order of desorption, B is the heating rate and Ed is
the energy of desorption. For zero-order desorption (n=0), Arrhenius plots of ln(rate)
versus ’I“1 yield values for Ed and v. Redhead8 showed that for a first-order process Ed

can be estimated from the peak desorption temperature (Tm) using equation (2.2)
13d = er[1n[v:;m ]— 3.46] (2.2)

and assuming a value for v (usually 1013 5"). Equation (2.2) has been found to vary by

 

only ~ 2 % when v falls between 108 and 10‘3 s".9 These two analyses represent only a

few of the procedures cited in the literature.9'”

Photochemistry. The adsorbate-covered surface was exposed to unpolarized
ultraviolet (UV) radiation from a medium-pressure Hg arc lamp (Oriel 6286) operated at
350 W. The lamp was equipped with an aqueous visible/infrared filter (~ 1 M NiSO4)

which transmitted photon energies in the range 5.4-3.9 eV (230-320 nm). The UVN is

27

6.19

l 1

4.96

l r

Energy (eV)
4.13
r l r r 1

3.54

l I l I

3.10

 

0.7 -

0.6 -

0.5 - '

0.4 - I

0.3 -

Transmittance

0.2 -

0.1- I

0.0 -|—'

 

1 I

200

I I

.,.
250

.....
300

. Ta
350

 

"l
400

Wavelength (nm)

Figure 2.3 Ultraviolet/Visible spectrum of ~ 1 M NiSO4 (aq) solution used as a filter for
the medium-pressure Hg arc lamp.

28

spectrum of a ~ 1 M NiSO4 solution is shown in Figure 2.3. The UV light was
introduced into the UHV chamber through a fused quartz window. The angle of
incidence of light was 45° with respect to the surface normal. Irradiation of the chamber
and sample mount was minimized through the use of an aperture affixed to the window
and irradiation of the sample with a power of ~ 9 chrn‘2 resulted in a temperature rise

of ~ 2 K. All irradiations were performed at T S 85 K.

2.1 References

(1) Ho, W. In Investigations of Surfaces and Interfaces - Part A; Rossiter, B.,
Baetzold, R., Eds.; Wiley and Sons: New York, 1993; Vol. 9A.

(2) Terentis, A.; Waugh, 8.; Metha, G.; Kable, S. J. Chem. Phys. 1998, 108, 3187.

(3) Kittel, M.; Polcik, M.; Terborg, R.; Hoeft, J.; Baumgartel, P.; Bradshaw, A.;
Toomes, R.; Kang, J.; Woodruff, D.; Pascal, M.; Lamont, C.; Rotenberg, E. Surf. Sci.
2001, 470, 311.

(4) Sueyoshi, T.; Sasaki, T.; Iwasawa, Y. J. Phys. Chem. B. 1997, 101, 4648-4655.
(5) Seah, M.; Smith, G. In Practical Surface Analysis; Briggs, D., Seah, M., Eds.;
Wiley and Sons: New York, 1990; Vol. 1; pp 531.

(6) Tracy, J. J. Chem. Phys. 1972, 56, 2748.

(7) Ibach, H.; Mills, D. L. Electron Energy Loss Spectroscopy and Surface
Vibrations; Academic Press: San Diego, 1982.

(8) Redhead, P. A. Vacuum 1962, 12, 203.

(9) de long, A.; Niemantsverdriet, J. Surf. Sci. 1990, 233, 355.

(10) Koch, K.; Hunger, B.; Klepel, O.; Heuchel, M. J. Catal. 1997, 172, 187-193.

29

(11)

King, D. Surf. Sci. 1975, 47, 384.

30

Chapter 3 Adsorption and Polymerization of Formaldehyde on Cu(100)

Abstract

The adsorption of formaldehyde (H2CO) on clean Cu( 100) at 85 K has been studied using
electron energy loss spectroscopy (EELS), X-ray photoelectron spectroscopy (XPS), and
temperature programmed desorption (TPD). For coverages up to 1.06 (i0.22)x1015
H2CO molecules/cmz, formaldehyde spontaneously polymerized to form a monolayer of
disordered poly(oxymethylene) (POM), arranged with the chain directions parallel to the
surface plane. Thermal decomposition/desorption of the polymer monolayer occurred by
two routes, producing peaks in temperature programmed desorption (TPD) at
approximately 200 K and 215 K. These features were produced by molecular H2CO
generated via depolymerization of the polymer. The 200 K and 215 K features displayed
apparent zero- and first-order desorption kinetics, corresponding to estimated activation
energies for depolymerization of 75 (i 10) and 53.9 (i- 0.5) kJ/mol, respectively. The
presence of two polymer desorption peaks is attributed to chain conformational
differences present within the monolayer, and has not been previously observed in studies
of formaldehyde adsorption on metal surfaces. Large exposures of H2CO on this surface
formed multilayers of molecular formaldehyde on top of the first polymer layer. The

second layer desorbed at 105 K and subsequent layers at ~100 K.

31

3.1 Introduction

Compared to Langmuir-Blodgett (LB)1»2 and self-assembled monolayer (SAM)3'8
techniques, direct adsorption of monomer followed by polymerization, either thermally or
through initiation with photons or electrons, offers a simple, rapid and solvent-less route
to the formation of polymer thin films. However, for most of the systems which have
been investigated to date, little is known about the adsorbed monolayer structure prior to
polymerization, the polymer morphology or the details of the reaction mechanism.
Elucidation of the polymerization initiation, propagation and termination mechanisms
operative for ordered monolayers of unsaturated small molecules, and the influence of
surface electronic and crystallographic structure on the polymer film order, chain length,
conformation and direction(s) in relation to specific crystal directions may ultimately
enable the design of high-quality crystalline polymer thin films.

There have been relatively few fundamental studies of small molecule
polymerization reactions on surfaces even though this type of reaction plays a large role
in Fischer-Tropsch catalysis}10 While the adsorption of formaldehyde on Cu(100) has
not been investigated (the focus of this chapter), there is some precedent for thermal
polymerization of H2CO to poly(oxymethylene) (POM) on several metal surfaces:
O/Ag(110),ll Ni(110),12 Pt(lll),13 Pd(111),14 O/Rh(111),15 O/Pd(lll),16 NiO(100)17
and Cu(110).18 Thermal polymerization of acetaldehyde on O/Ag(l 1 1) has also been
observed.19 In addition to thermal polymerization, ultraviolet photons or low-energy
electrons have been shown to initiate polymerization of other small molecules adsorbed at

metal substrates: TCNQ,20 dinitrobenzene,21 thiophene,22v23 formaldehydez‘iv25 and

styrene.26

32

3.2 Results and Discussion

Temperature Programmed Desorption (TPD). The adsorption of formaldehyde

on Cu(100) was studied by TPD. Figure 3.1 shows a series of m=29 (HCOI) TPD spectra
for increasing exposures of H2CO (for clarity, not all exposures shown). At 0.2 L H2CO
exposure, desorption peaks of approximately equal intensity were observed at 197 K and
212 K. With increasing exposure of H2CO, both peaks increased in intensity. However,
the 197 K peak shifted to higher peak desorption temperatures, but the 212 K peak
remained at approximately the same peak desorption temperature. The total area of these
two peaks, as a function of exposure, is shown in the inset of Figure 3.1. The area
increases linearly up to an exposure of approximately 0.9 L and then reaches a constant
value. Separate fits of the increasing and constant regions intersected at 0.87 (10.14) L,
which can be related to coverage, as discussed below.

For a 1.1 L H2CO exposure, the two desorption features merged into a broad peak
at approximately 205 K, and a new peak was observed at 105 K. For exposures >2.2 L,
an additional peak at 100 K grew in that did not saturate with increasing exposure. Based
on previous investigations,18 the 105 K and 100 K features are associated with desorption
of molecular H2CO from the second and subsequent monolayers, respectively. The
feature observed at 84 K is attributed to desorption from the heating filament and the
large, broad background observed between 100 K and 180 K is attributed to desorption
from the sample mount.27

Figure 3.2 shows TPD spectra for a 1.0 L H2CO exposure on Cu( 100) measured at

masses corresponding to hydrogen (m=2, Hf), water (m=l8, H2O+), carbon monoxide

33

 

 

 

 

 

 

80 - _
(U I
r- 9..) . I I
< I
60 _ g '
m I
r- 5 I
93
C I
40 - _ .'
(ht _ 0.0 0.5 1.0 1.5 2.0 2.5 3.0
E 20 H2CO Dose (L)
i
E? . h
a)
c 9
e 0 - f
s 8 r
c
2 b

N
I l 3
macaw-'-

 

 

OLIIIIJIIIILJJIIILILIII

1 00 1 50 200 250 300
Temperature (K)

 

Figure 3.1 Mass 29 thermal desorption spectra for increasing doses of H2CO on Cu(100).
All exposures were made at 85 K. The inset shows the total integrated area for the
features between 190 and 230 K as a function of exposure. Exposures are: a) 0.2 L, b)
0.3 L, c) 0.5 L, d) 0.6 L, e) 0.8 L, t) 1.1 L, g) 1.8 L, h) 2.2 L, i) 2.7 L.

(m=28, co”), forrnyl (m=29, Hco“), formaldehyde (m=30, H2CO+) and 13c-
forrnaldehyde (m=31, H2'3CO+). Each spectrum constituted a separate 1.0 L exposure.

For masses 28,29, 30 and 31, the relative abundances for each ion (20:100:48:l.3)
between 190 K and 230 K were similar to those obtained from the mass spectrum of
H2CO(g) (24:100:58:l.l).28 This confirms that the identity of the species desorbing
between 190 K and 230 K is molecular formaldehyde. We believe molecular H2CO is
generated as a result of depolymerization of a surface-bound polymer as will be discussed
in detail below.

For m=28, three desorption peaks were observed. The two highest temperature
peaks (approximately 200 K and 215 K) can be identified as fragment ions of molecular
H2CO, as described above. However, the 174 K peak was not observed at m=29, 30 or
31 and thus, is inconsistent with fragmentation of H2CO. The similarity of the desorption
temperature of this peak with that of a CO-exposed Cu(100) surface (data not shown)
leads us to conclude that this feature is due to desorption of molecular CO. The m=2
spectrum displays a peak at 178 K and a shoulder at ~200 K. The shoulder likely
corresponds with cracking of the H2CO and is associated with comparable peaks
observed at m=29, 30 and 31. However, the peak at 178 K cannot be attributed to H2CO
cracking. This feature was likely due to recombinative desorption of adsorbed H atoms,
which are known to adsorb on copper surfaces at 85 K2931

For the m=18 TPD spectrum, peaks are observed at 197 K and 224 K, which are
associated with desorption of water coadsorbed with the formaldehyde during the gas

exposure. A scan of 1-100 amu during desorption showed no other desorbing species.

35

 

 

 

 

t 1.0LHZCO/Cu(100)
_ m=2
f m=18(x10)
g- .
2 ' m=28
9 .-
5 .
C
2 _
. m=29
M
' m=31(x50)
'..r.L..r.L..r....r.r..
100 150 200 250 300

Temperature (K)

Figure 3.2 Thermal desorption spectra for masses 2 (Hf), 18 (H2O+), 28 (CO+), 29
(HCO+), 30 (H2CO+), and 31 (H213COI) for a 1.0 L exposure of H2CO on Cu(100). Each
spectrum is a separate 1.0 L exposure. All exposures were made at 85 K. Spectra are
offset for clarity.

Electron Energy Loss Spectroscopy (EELS). Figure 3.3 shows a series of EEL
spectra for increasing exposures of H2CO on the 85 K Cu(100) surface. Immediately
upon adsorption of H2CO on this surface, the EELS elastic beam intensity falls by more
than two orders of magnitude (from >106 Hz to ~104 Hz), indicative of a significantly
disordered adlayer. Table 3.1 summarizes the EELS results together with IR data
obtained from gas-phase H2CO32, solid H2CO33 and the polymer, POM.34 Excellent
agreement is obtained between our EELS data for the 0.5 L H2CO-dosed Cu(100) surface
and the IR data for solid poly(oxymethylene). The losses at 338 cm'1 and 2060 cm'1 can
be assigned to v(Cu-CO) and v(CEO) for adsorbed CO. Irnportantly, the carbonyl loss
due to v(C=O) for H2CO is not observed, suggesting no molecular H2CO is present on the
surface up to about 1.2 L. However, the possibility of H2CO adsorbed with the molecular
axis parallel with the surface cannot be excluded. The only mode that would be active for
dipole scattering in this geometry would be the out-of—plane deformation (B2 character)
which has an associated dipole moment oriented perpendicular to the surface. This mode
should appear at ~1167 cm'l, which is an area of the spectrum that is dominated by
polymer loss features.

The presence of 1,3,5-trioxane, the cyclic trimer of H2CO, was eliminated based
on EELS data taken for the adsorption of trioxane on Cu(100) at 85 K. Figure 3.4 is the
comparison of EEL spectra of a 0.5 L exposure of H2CO (Figure 3.4a) with that of a 0.7 L

exposure of trioxane (Figure 3.4b). Trioxane has been found to have complex adsorption

37

 

,5 _ H2CO / Cu(100) 85 K

10 -
7 d 21 L
,9 x10 )
X
7,; X150
0.
8 :-
.~§‘ c) 2.4 L
2
9 x250
E 5-

b) 1.2 L
x250

 

a) 0.5 L
x150

 

 

 

IIIlllIlllJlIllIllllllllllllllllllllllLll

0 1000 2000 3000 4000
Energy loss (cm'1)

 

Figure 3.3 EEL spectra for increasing exposures of H2CO on Cu(100) at 85 K. The
sample was flashed to 300 K between exposures.

38

 

03m comm A; .m>+m>v 8:253. Eton

39

 

Sam 3% 58.;
mam 2% Eu?
«Sm owwm Aeamvwm £83,
cowm Swm :3me EU?
88 22 5mm:
0 _ t _ E A53: Suu;
NE E: mm: 83 A382 58%
82 em: 35;
SS ewmtmmfl mmfl ES 3332 :29: 35:. \ 36: .Aoummvem
o: _ E _ 33$: eon
a: 52 8: 599g;
o2:

tum mmo mg 5099
Se 08 500cm
mmv ocoto>o
wmm A092;
me 2008 852

M mm Ace—V50 v— mw Ace—EU
25$ 4 no 3.2—om 20m 255 A 3. 3.26m DUN: 280 GUN: Beeswaé

 

.OUNI can 220% $5.ansz338 26m .8 Sun 5 2m :39? 82 .v_
mm 3 50:30 so GUN: ho 35.898 4 _m can A md .8 3mm .3 @3830 3-80 Ev mucus Econ—«.53 05 .8 mucoEchm/x _.m 2an

 

Cu(100) 85 K

 

 

 

 

 

 

15-
8(OCO)
b) 0.7 L trioxane
2:
E 10 _ ‘ X150
(0
Q.
3
E?
U) 1-
C
.92
E
0-5 " a) 0.5 L H200
x150

 

 

 

 

 

0.0-Jk

r I uuuuuuuu I rrrrrrrr l 11111111 I rrrrr
0 1000 2000 3000
Energy loss (cm'1)

 

 

 

Figure 3.4 EEL spectra of a) 0.5 L exposure of H2CO on Cu(100) and b) 0.7 L exposure
of trioxane on Cu(100). Both spectra were acquired at 85 K.

behavior on Cu(1 11) with monolayer coverage occurring at ~ 0.02 ML and multilayers
forming after saturation of the monolayer.35 The EEL spectrum shown in Figure 3.4b
likely corresponds to a multilayer coverage of trioxane on Cu(100) although absolute
coverage was not calculated. The losses observed match well to those found for bulk

crystalline trioxane as seen in Table 3.2 Also, the observation of losses corresponding

Table 3.2 Assignments of the vibrational bands (in cm") observed by EELS for 0.7 L
exposure of trioxane on Cu( 100) at 85 K. Also shown is IR data for solid trioxane.

 

 

Assignment Symmetrya Trioxane 0.7 L Trioxane / Cu(100)
Solid36
v(M-CO) 321
5(COC) A1 469 470
5(OCO) E 521 526
VS(COC) / r(CH2) A1 / E 951 / 918 946
VS(COC) E 1067 1049
Va(COC) E l 152 l 165
r(CH2) A1 1222 1215
t(CH2) E 1313 1288
w(CH2) E 1419 1392
8(HCH) E 1483 1472
v(CEO) 2059
v,(CH) E 2877 2853
_v,_,(CH) E 3031 3003

 

a) bulk symmetry, C 3,.

to E modes in bulk trioxane (C 3,, symmetry) rules out an overlayer structure with the C3
axis parallel with the surface normal. Irnportantly, the loss observed at 645 cm'1 in Figure
3.4a, assigned to 8(OCO), most closely matches the reported value for POM at 630 cm".
In the IR spectrum of solid 1,3,5-trioxane, this mode is split into two bands at 744 cm'l
and 521 cm]. For the EEL spectrum shown in Figure 3.4b, the 744 cm’1 loss was

undetected due to its inherently low intensity.36

41

On increasing the H2CO exposure to 2.4 L (Figure 3.3c), some recovery of the
elastic beam intensity was noted (approximately an order of magnitude), indicating that

the second layer is somewhat more ordered than the first (polymer) layer. The appearance

of an intense loss at 141 cm'1 has been previously observed for H2CO on a number of

surfaces and has been attributed to lattice (phonon) modes of crystalline H2CO(s).“‘
1348’37’33 At 2.4 L exposure, the loss at 1483 cm'1 significantly increased in intensity
and a new loss at 1719 cm“1 appeared. These losses are attributed to molecular H2CO and
agree with IR data taken on crystalline H2CO films at 80 K.32 The losses at 1184 cm'1

and 1240 cm'1 also gained intensity and are assigned to the out-of-plane and in-plane

bending modes of H2CO respectively. The v(C-H) region showed three distinct losses
which can be assigned to vs(C-H) and vas(C-H) of H2CO and to a Fermi resonance

between v2+v5 and V.;. This Fermi resonance has previously been observed in the IR

spectrum of crystalline H2CO”,33 The loss due to adsorbed CO, v(CE-O), has red-

shifted to 2020 cm'1 and gained intensity relative to 1.2 L, but the loss due to v(Cu-CO)
was still not detected at ~340 cm". We assume the modest red-shifting of the v(CEO)

mode is due to intermolecular coupling between adsorbed H2CO polymer and CO.
At the largest H2CO exposure studied, 21 L, the losses due to molecularly adsorbed

multilayer H2CO were observed clearly. The lattice mode has gained intensity and blue-
shifted by 87 cm'1 to 228 cm"1 relative to the 2.4 L exposure. A loss at 423 cm’1 appeared
and is tentatively assigned to an overtone of the 228 cm'1 lattice mode. The peak at 943

cm", for which there is no corresponding mode in molecular H2CO, is

42

assigned to the vs(OCO) mode of polymer formed during acquisition of the EELS
spectrum. Previous studies of H2CO adsorption on Ag(111) have shown that
polymerization of molecular formaldehyde can be initiated by low-energy electrons and
that even exposure to the low fluxes of electrons encountered in EELS can initiate
polymerization of multilayers of H2CO.25’39

Figure 3.5 shows a series of EELS spectra taken as a function of annealing
temperature for a 6.3 L exposure (multilayer) of H2CO on Cu(100). The surface was
dosed at 85 K, momentarily heated to the indicated temperature and immediately cooled to

85 K prior to data acquisition. Figure 3.5a shows the “as exposed” spectrum taken

immediately after 6.3 L of H2CO on the 85 K Cu(100) surface. As expected, it displayed

losses characteristic of both molecular H2CO and POM. In particular, the v(C=O) of

H2CO at 1720 cm'1 and the vs(OCO) of POM at 950 cm'1 are observed. The v(CEO) loss

due to adsorbed CO at 2011 cm], is observed but it is significantly red-shifted relative to
CO adsorption on the bare Cu(100) surface.‘40 However, at this exposure, the loss due to v
(Cu-CO) at ~340 cm'1 was not observed.

Figure 3.5b shows the EELS spectrum obtained after annealing the 6.3 L exposed

surface to 120 K. In contrast to the unannealed (85 K) spectrum (Figure 3.5a), the loss

due to v(C=O) of multilayer H2CO was not observed and the losses due to POM were

narrower and more intense (in particular the losses assigned to v(OCO) of POM), even
though the full width at half-maximum (FWHM) of the elastic peaks for the two spectra

were similar. We attribute this observation to increased ordering of the POM induced by

43

 

 

 

 

 

60 - 6.3 L (5.6 ML) H2CO/Cu(100)
c) 300K
"BO 40- x50
3?
’0?
o.
‘0’ . b) 120K
32‘
(D
E
E
20‘ x50
a) 85K
x50
0
rIr...rrrrrlrrrrrrrrrJJJJrrerrlrrrrrrrr.
0 1000 2000 3000 4000

Energy loss (cm'1)

Figure 3.5 EEL spectra as a function of annealing temperature for a 6.3 L exposure of
H2CO on Cu(100). The “as dosed” spectrum is shown in a) while b) and c) were annealed
to the indicated temperature. All spectra were recorded at 85 K. The shoulder at ~ 200
cm"I on the elastic peak is the lattice mode associated with the spectrum shown in a). This
peak is absent in b) and c).

44

heating to 120 K. The v(CEO) loss at 2007 cm'1 has lost intensity and the loss due to

v(Cu-CO) was not observed.

Annealing the sample to 300 K, as shown in Figure 3.5c, resulted in complete
decomposition/desorption of the polymer as evidenced by the absence of any losses due to
POM. The only losses observed are due to CO readsorbed during cooling of the sample
following annealing.

X-ray Photoelectron Spectroscopy (XPS). The adsorption of H2CO was also
studied using XPS. Figure 3.6 shows the C (Is) and 0 (Is) regions for increasing
exposures of H2CO on Cu(100) surface at 85 K. At low exposures (<1 L), a peak is
observed in the C (ls) spectrum at a binding energy (BE) of 288.2 eV. With increasing
exposure, this peak shifted to higher binding energy (288.8 eV for 5.4 L H2CO). Indeed,
our measured C (1s) binding energy for the adsorbed polymer is similar to that measured
by Pireaux et al.41 for bulk poly(oxymethylene) of 287.8 eV, suggesting that the polymer-
surface interaction is weak. Similar behavior was observed for the 0 (Is) region; a peak
was observed initially at 533.1 eV at 0.1 L which shifted to 533.8 eV at 5.4 L. These
XPS binding energies are attributed to POM at <1 L exposure and, at higher exposures, to
a mixture of POM and molecular H2CO.

As mentioned above, absolute coverages of formaldehyde on Cu(100) were
obtained using XPS by standardization with a known adsorbate system. Figure 3.7
displays the calculated coverage as a function of exposure for the adsorption of H2CO on
Cu(100) at 85 K. From the inset of Figure 3.1, TPD data indicate saturation of the
polymer desorption features occurred at 0.87 L. Using the standardized XPS intensity

ratios calculated above, this exposure corresponds to an absolute coverage of 0.69 (i014)

45

538 536 534 532 530 528 526

I I I I l I I I I r I I l I I I l I I I l I T I

b) 0 (1s)

 

 

 

 

 

 

 

 

 

 

 

 

 

a) C (1s)

Intensity (arb. units)

1x10‘ cps

 

55%

M
J

 

W

 

 

 

 

 

 

 

'1': : rI'Tn 1r1+1[L. :1 r 1 nTr'r lr—ll 1
294 292 290 288 286 284 282 280
Binding energy (eV)

Figul’e 3.6 XP spectra of a) C (Is) and b) 0 (Is) for increasing exposures of H2CO on
Cu( 100) at 85 K. The sample was flashed to 300 K between exposures. Exposures were
0, 0- 1 . 0.2, 0.4, 0.5, 0.7, 0.9, 1.2, 2.1, 5.4, L. Vertical lines are drawn at 288.8 eV and
533-8 eV for the C (Is) and 0 (Is) regions, respectively.

46

 

5 ..
H2CO / Cu(100) 85 K

Coverage (ML)

 

 

 

0 1 2 3 4 5
H200 Exposure (L)

Figure 3.7 Calculated coverage as a function of exposure for the adsorption of H2CO on
Cu(100) at 85 K based on XPS C(ls) peak areas. Equation of linear fit is y=0.898*x-
0.082.

47

ML (0.69 formaldehyde molecules per surface atom) or equivalent to 1.06 (3:0.22) x1015

formaldehyde molecules/cmz.

Depolymerization Energetics. The adsorption and reactions of formaldehyde on
the Cu(100) surface described here can be compared with previous studies of
formaldehyde on clean group IB metals. On Au(110)42 and Ag(111),24 formaldehyde
adsorbs non-reactively and desorbs as molecular H2CO at approximately 160 K and 110
K, respectively. In the case of Ag(111), UV or low energy electron irradiation induces
polymerization, with depolymerization occurring at 210 K producing molecular H2CO.

In contrast, on the 90 K Cu(110) surface, H2CO was found to adsorb reactively to
form POM.18 Desorption peaks were observed at 110 K and 225 K and attributed to
multilayer H2CO desorption and decomposition of the polymer, respectively.
Interestingly, the depolymerization temperatures noted in our work for Cu(100) are
similar to the polymer formed spontaneously on the copper (110) surface and through '
photopolymerization on the silver (111) surface (~200-230 K), implying that the polymer
produced is similar in these cases.

In contrast to the previous studies of H2CO adsorption on Cu(110) and Ag(111),
in which a single peak due to depolymerization was observed in TPD spectra, we observe
two distinct peaks. These features originate from POM. The most likely explanation for
the appearance of two depolymerization features in our TPD data for Cu(100) is a
difference in depolymerization kinetics and/or energetics. Expansion of the temperature
region in which the depolymerization features are observed (180-240 K) (shown in Figure
3.8) reveals that the lower temperature peaks share a common leading edge and a shift to

a higher peak desorption temperature with increasing H2CO exposure. For a true

48

 

- H2CO / Cu(100)

N
l

  

Ion Intensity (m=29)

Exposure

 

 

 

Temperature (K)

Figure 3.8 Enlarged view of the polymer features from the TPD spectra shown in fig. 1.
A constant background was subtracted from each spectrum. Exposures ranged from 0.1 L
to 1.1 L.

49

desorption process, these are the characteristics of zero-order desorption kinetics.43
However, for the experiments presented here the measured desorption rate of
formaldehyde is not reflective of the desorption kinetics associated with molecular H2CO,
but rather with decomposition of the polymer to produce molecular H2CO which
promptly desorbs. Hence, the zero-order desorption kinetics observed are representative
of the poly(oxymethylene) depolymerization mechanism. Thus, standard TPD data
analyses yield activation energies that are relevant to the rate of depolymerization. The
higher temperature TPD feature shows a constant peak desorption temperature, which is
indicative of a first-order desorption process.

We can estimate the activation energy and pre-exponential factors for
depolymerization by subjecting the zero-order desorption feature to a leading edge
analysis43 (plots of ln(rate) vs. T") while the activation energy for the first-order feature
was calculated by the Redhead method assuming a pre-exponential of 10'3 8".44 These
analyses yielded activation energies of 75 (i10) and 53.9 (10.5) kJ/mol for the zero- and
first-order desorption features, respectively. These depolymerization energies fall within
the broad range of 42-113 kJ/mol quoted for POM from various bulk studies.“5 The
difference in activation energy is less important than the observation of two different
kinetic orders for depolymerization and will be discussed in detail in Chapter 4.

The most reasonable explanation for the appearance of multiple desorption peaks
for polymer decomposition is the presence of different polymer species that depolymerize
by a similar pathway. The possibility of a single polymer species decomposing via two

competing pathways can be immediately discounted because such a situation would be

50

expected to produce a single desorption feature; the fastest and/or lowest barrier process
will always dominate.

Possible Polymer Species. An obvious choice for two different polymer species
would be polymer adsorbed at distinct sites such as terrace sites versus. defect or step-
edge sites. The appearance of two desorption peaks in the monolayer TPD data (Figure
3.1) of approximately equal intensity implies that the surface concentrations of defects
and/or edge sites would be approximately equal to the concentration of terrace sites. We
estimate the density of step-edges for a Cu(100) surface cut to within 0.5° (supplier’s
specification) at ~l %. Furthermore, excellent surface order is suggested by the high
EELS elastic peak intensity (>106 Hz) of the clean Cu(100) surface, and <20 % loss in
instrumental resolution relative to the "straight-through" geometry. For these reasons, we
believe defect sites and/or step-edges are not responsible for the two polymer desorption
peaks.

It is possible conformational differences could give rise to the two polymer TPD
peaks. In the crystalline state POM forms helices belonging to either the D( 10709) or
D(n') point group in the trigonal (t-POM) and orthorhombic (o-POM) crystal structures,

respectively.46r47 Both point groups are isomorphous with D" and have an associated

dynamic dipole oriented parallel (A2) or perpendicular (E) to the helical chain axis.34
This is shown in Figure 3.9a for orthorhombic POM. Adsorption of these helices with
the primary rotation axis (z-axis in Figure 3.9a) parallel to a surface would reduce the
symmetry to C1. As a consequence of the metal-surface selection rule, the original A2
modes, polarized within the surface plane, would be strictly screened and thus

unobservable for dipole scattering in EELS. The E modes would still have some

51

 

a) DUI?) A2
I: i——- y E

   

 

E
Helix viewed Helix viewed
along z-axis. along x-axis.
b) C 2v
vas(OCO), B1 VS(OCO), Al

O O O z
\//\\/\// 8v
0 O_ o o

 

 

 

 

Figure 3.9 Comparison of symmetry for a) orthorhombic (helical) and b) planar POM.
The Cartessian coordinates (x,y,z) and representations (A2, E, B1, A1) for a) correspond to
the helix in free space while for b) the substrate is considered. The filled circles are
methylene units (CH2).

52

component of their dipoles oriented perpendicular to the surface depending on rotational
orientation. Thus, for dipole scattering, the observed features in our EELS spectra must
originate from what were E modes of the polymer. Similar arguments have been made
for the thermal polymerization of acetaldehyde adsorbed on O/Ag(111).19

Another chain conformation the polymer could adopt is a planar zigzag bound to
the surface through the O atoms (the planar conformation has never been observed
experimentally in the solid) as shown in Figure 3%. Assuming the C2,, symmetry of the
planar conformation is retained upon adsorption, the vas(OCO) mode should be (strictly)
screened in this conformation but observable for helical conformations. While the
coexistence of the helical and planar forms cannot be ruled out on the basis of our EELS
data, the observation of vas(OCO) does eliminate the possibility that only the zigzag form
was present on the surface. We expect the vibrational frequencies for the planar and
helical conformations of the surface-bound POM to be very similar and therefore
indistinguishable by EELS. Small morphology-dependent frequency shifts have been
observed between orthorhombic POM and trigonal POM.48 The largest difference
observed by this study was 37 cm'1 between the o-POM (596 cm'l) and t-POM (633 cm"
I) 8(OCO) mode. The vibrational frequency of the 5(OCO) mode observed in our EELS
data (631 cm‘l) implies that the crystal habit present on the Cu(100) surface is trigonal.
Interestingly, we have preliminary evidence that this mode red shifts to approximately
599 cm“1 upon annealing the polymer monolayer to 190 K, perhaps suggesting a
transformation to the orthorhombic form (data not shown).

The surface density of formaldehyde molecules in the polymer monolayer,

provides additional insight into the conformation of POM on the Cu( 100) surface. It will

53

be recalled that the calculated density, based on standardized XPS measurements, was
1.06x1015 cm'z. For crystalline orthorhombic poly(oxymethylene), there are two possible
planes that can be projected onto the surface to produce chain directions parallel to the
surface plane (saturation of the polymer TPD features discounts polymer chain growth
perpendicular to the surface plane). The bc plane has a calculated average monomer
density of 7.35x10l4 cm'z, while the ac plane has a calculated average monomer density
of 1.18x1015 cm'z. For trigonal poly(oxymethylene), the ac plane has an average
monomer density of 1.16x1015 cm’z. Thus, it can be seen the ac plane of both o-POM
and t-POM produce surface monomer densities close to that calculated by the
standardized XPS data for saturation of the polymer layer. Although agreement between
these surface densities is good, and suggests that the polymer monolayer covers the
surface completely, it should be stressed that we have already inferred that the layer is
probably highly disordered and should not be considered crystalline.

The existence of a decomposition route that does not produce molecular H2CO is
evident by the desorption of CO and H2 prior to the depolymerization of POM. A
common origin for these two species is indicated by the coincident desorption
temperatures. Leading edge analysis of these second order desorption peaks yielded an
activation energy of 49 (i8) kJ/mol and a pre-exponential factor of 1x1015‘“ molecules

l-cm2-s". The ratio of desorption in the H2+CO versus H2CO channels showed a

maximum at ~ 0.04 at 0.66 ML (95 % of saturation of the polymer monolayer). The
intensity ratio for the m=28 peak at 178 K and the m=2 peak at 177 K, corrected for
ionization probability and fragmentation, is approximately 1.15, very close to the

expected decomposition ratio for a species containing equal proportions of H2 and CO.

54

We have already mentioned the possibility of the existence of H2CO species bound to the
surface with the molecular axis parallel to the surface plane; this adsorption geometry
may give rise to the H2 and CO desorption products observed at 178 K in TPD.
Therefore, the observation of H2 and CO desorption at 177 K must be due to
decomposition of a third, as yet unidentified, species. This species may arise at step edge
or defect site as these would be expected to more reactive than terrace sites on the

Cu(100) surface.

3.3 Conclusions

The adsorption of formaldehyde (H2CO) on the Cu(100) surface has been studied.
At 85 K, H2CO polymerizes spontaneously to form a monolayer of poly(oxymethylene)
(POM) up to a saturation coverage of 0.69 ML (1.06x10ls cm'z). This surface density
suggests that the POM chain directions are parallel to the surface plane. No molecularly
adsorbed H2CO was observed in EEL spectra prior to saturation. Further exposure of the
polymer-covered Cu(100) surface resulted in multilayers of H2CO, with EELS loss
features characteristic of crystalline H2CO. These multilayers are susceptible to
polymerization by the low-energy electron beam during EELS data acquisition.

Temperature-programmed desorption spectra indicate there is one decomposition
route available to the adsorbed polymer species: decomposition to molecular H2CO. The
route producing H2CO is observed as two desorption features at approximately 200 and
215 K which show zero- and first-order depolymerization kinetics respectively. The
observation of two orders can be possibly explained by the presence of two types of

polymer existing in different conformations. We believe differences due to adsorption at

55

defect and/or step edges do not account for the appearance of two desorption features for
route (ii). The observation of CO and H2 desorption indicates a third species is present
that is probably due to dissociation of H2CO at step edges and defect sites. Electron
energy loss spectroscopy data indicate no other species besides an adsorbed polymer first
layer and molecular H2CO multilayers on top of the polymer layer, are present on the 85

K Cu(100) surface.

3.4 References

(1) Ejaz, M.; Yamamoto, S.; Ohno, K.; Tsujii, Y.; Fukuda, T. Macromolecules 1998,
31, 5934.

(2) Shirai, E.; Urai, Y.; Itoh, K. J. Phys. Chem. B. 1998, 19, 3765.

(3) Ford, J. F.; Vickers, T.; Mann, G; Schlenoff, J. Langmuir 1996, 12, 1944.

(4) Sun, F.; Castner, D.; Grainger, D. Langmuir 1993, 9, 3200.

(5) Kim, T.; Chan, K.; Crooks, R. J. Am. Chem. Soc. 1997, 119, 189.

(6) Chan, K.; Kim, T.; Schoer, J.; Crooks, R. J. Am. Chem. Soc. 1995, 117, 5875.

(7) Balasubramanian, K.; Cammarata, V. Langmuir 1996, 12, 2035.

(8) Batchelder, D.; Evans, 8.; Freeman, T.; Haussling, L.; Ringsdorf, H.; Wolf, H. J.
Am. Chem. Soc. 1994, 116, 1050.

(9) Bent, B. E. Chem. Rev. 1996, 96, 1361.

(10) Somorjai, G. Introduction to Surface Chemistry and Catalysis; Wiley and Sons:
New York, 1994.

(11) Stuve, E.; Madix, R.; Sexton, B. Surf. Sci. 1982, 119, 279.

(12) Richter, L.; Ho, W. J. Chem. Phys. 1985, 83, 2165.

(13) Henderson, M.; Mitchell, G.; White, J. Surf. Sci. 1987, 188, 206.

56

(14) Davis, J.; Barteau, M. J. Am. Chem. Soc. 1989, 111, 1782.

(15) Houtman, C.; Barteau, M. Surf. Sci. 1991, 248, 57.

(16) Davis, J.; Barteau, M. Surf. Sci. 1992, 268, 11.

(17) Truong, C.; Wu, M.; Goodman, D. W. J. Am. Chem. Soc. 1993, 115, 3647.

(18) Sexton, B.; Hughes, A.; Avery, N. Surf. Sci. 1985, 155, 366.

(19) Sim, W.; Gardner, P.; King, D. J. Am. Chem. Soc. 1996, 118, 9953.

(20) Wells, S.; Giergel, J .; Land, T.; Linquist, J .; Hemminger, J. Surf. Sci. 1991, 257,
129.

(21) Tsai, W.; Boeri, F.; Clarson, S.; Montaudo, G. J. Roman Spectrosc. 1990, 21, 311.
(22) Cheng, L.; Bocarsly, A.; Bemasek, S.; Ramanarayanan, T. Surf. Sci. 1997, 374,
357.

(23) Land, T.; Hemrrringer, J. Surf. Sci. 1992, 268, 179.

(24) Fleck, L.; Feehery, W.; Plummer, E.; Ying, Z.; Dai, H. J. Phys. Chem. 1991, 95,
8428.

(25) Fleck, L.; Kim, J .; Dai, H.-L. Surf. Sci. 1996, 356, M17.

(26) Carlo, S.; Grassian, V. Langmuir 1997, 13, 2307.

(27) The large, broad desorption feature between 100 and 180 K was observed during
evaluation of our experimental arrangement using a known system (CO/Cu(100)). This
system generally shows no broad features in this temperature range and the appearance of
features in this temperature range in our experiments was thus attributed to desorption
from the sample mount. In subsequent experiments perfomed using directed dosing,

these features were absent.

57

(28) NIST Chemistry WebBook, NIST Standard Reference Database Number 69;
Mallard, W., Linstrom, P., Eds.; National Institutes of Standards and Technology:
Gaithersburg, MD, 1998; pp (http://webbook.nist.gov).

(29) Chorkendorff, 1.; Rasmussen, P. Surf. Sci. 1991, 248, 35.

(30) Kammler, T.; Kuppers, J. J. Chem. Phys. 1999, 111, 8115.

(31) Adsorbed hydrogen was assumed to arise from dissociation of molecular
hydrogen and/or formaldehyde at the filaments of the ion gauge and/or QMS.

(32) Khoshkoo, H.; Hemple, S.; Nixon, E. Spectrochim. Acta A 1974, 30, 863.

(33) Weng, S.; Anderson, A.; Torrie, B. J. Raman Spectrosc. 1989, 20, 789.

(34) Tadokoro, H.; Kobayashi, M.; Kawaguchi, Y.; Koybayashi, A.; Murahashi, S. J.
Chem. Phys. 1963, 38, 703.

(35) Hofmann, M.; Wegner, H.; Glenz, A.; Woll, C.; Grunze, M. J. Vac. Sci. Technol.
A 1994, 12, 2063.

(36) Kobayashi, M.; Iwamoto, R.; Tadokoro, H. J. Chem. Phys. 1966, 44, 922.

(37) Anton, A.; Parmeter, J.; Weinberg, W. J. Am. Chem. Soc. 1986, 108, 1823.

(38) Fleck, L.; Ying, Z.; Feehery, M.; Dai, H.-L. Surf. Sci. 1993, 296, 400.

(39) Fleck, L. E. Ph.D. Thesis, University of Pennsylvania, 1994.

(40) Sexton, B. Chem. Phys. Lett. 1979, 63, 451.

(41) Pireaux, J.; Riga, J.; Boulanger, P.; Snauwaert, P.; Novis, Y.; Chtaib, M.;
Gregoire, C.; Fally, F.; Beelen, E.; Caudano, R.; Verbist, J. J. Electron Spectrosc. Relat.
Phenom. 1990, 52, 423.

(42) Outka, D.; Madix, R. Surf. Sci. 1987, 179, 361.

(43) de Jong, A.; Niemantsverdriet, J. Surf. Sci. 1990, 233, 355.

(44) Redhead, P. A. Vacuum 1962, 12, 203.

58

(45) Muck, K. In Polymer Handbook; Brandrup, J., Irnmergut, E., Grulke, E., Abe, A.,
Bloch, D., Eds.; Wiley & Sons: New York, 1999; Vol. 4; pp V/97.

(46) Helical symmetry is specified by the nomenclature X(2m1r/n) where X is the point
group (C or D), m is the number of turns of the chain per unit cell and n is the number of
monomer units per chain per unit cell).

(47) Tadokoro, H. In Macromolecular Reviews; Peterlin, A., Goodman, M., Okamura,
S., Zimm, 3., Mark, H., Eds.; Interscience: New York, 1967; Vol. 1; pp 119.

(48) Kobayashi, M.; Sakashita, M. J. Chem. Phys. 1992, 96, 748.

59

Chapter 4 Evidence for Two Chain Length Distributions in the Thermal
Polymerization of Formaldehyde on Cu(100)

Abstract

The polymer species formed from the spontaneous polymerization of formaldehyde
(H2CO) on clean Cu(100) at 85 K were studied using electron energy loss spectroscopy
(EELS) and temperature-programmed desorption (TPD). Formaldehyde forms
poly(oxymethylene) (POM) with differing chain lengths; the long (or) and short (B) chain
species depolymerize to give two features in TPD at approximately 207 and 219 K,
respectively. The complex desorption kinetics observed for the (It-POM species were
successfully modeled using equations based on the ratio of the average number of
monomers unzipped from the chain per initiation event to the length of the polymer
chain. Losses were observed in EEL spectra of the short chain species at ~ 290, ~ 1020
and ~ 1120 cm'1 that can be assigned to the v(Cu-O), v(C-O) and p(CH3) modes of
oxygen and methoxy endgroups, respectively. Pre-adsorbed methanol increases the
proportion of short chain POM species by increasing the probability of termination for
the B species. Lower thermal stability for adsorbed POM was observed compared to bulk

POM and is believed to be related to the stability of the surface bound oxygen endgroup.

60

4.1 Introduction

In Chapter 3, investigations of the spontaneous polymerization of formaldehyde
(H2CO) on clean Cu(100) at 85 K were reported.1 The formation of two types of
poly(oxymethylene) (POM, -(CH2O)n-) was observed by temperature-programmed
desorption (T PD) as two separate features corresponding to the depolymerization of the
POM to produce H2CO(g). The lower temperature feature (~ 209 K) exhibited apparent
zero-order desorption/depolymerization kinetics while the higher temperature feature (~
220 K) exhibited first-order desorption/depolymerization kinetics.2 We ascribed the two
depolymerization routes to conformational differences between the two adsorbed species.
In contrast, bulk POM is known to be stable up to at least 350 K depending on the degree
of polymerization and type of endgroups.3

The lower stability of the film is believed to be related to the endgroups of the
polymer and has been observed previously. For example, bulk polyimide made from
pyromellitic dianhydride (PMDA) and oxydianiline (ODA) is stable at temperatures up to
500 °C.4’5 However, polyimide films made from reactive adsorption of PMDA and ODA
on Cu(110) decompose at ~ 280 °C.67 The thermal stability of the film was shown to be
limited by the reactivity of the carboxylate endgroups bound to the surface, but the
decomposition pathway and the influence of the substrate on the endgroup stability has
not been thoroughly studied.

In other studies on the thermal polymerization of H2CO on Cu(110), a single
depolymerization feature was observed at 225 K.8 Additionally, POM formed through

the photopolymerization of H2CO adsorbed on Ag(lll) depolymerized at 210 K.9

61

Unfortunately, no studies were performed to determine the kinetics of depolymerization
on either surface. Regardless, the similarity of depolymerization temperatures observed
for these two surfaces and the Cu(100) surface suggests the POM species formed on each
surface are similar.

While Chapter 3 dealt with identifying the products formed from the adsorption of
H2CO on Cu(100) at 85 K, the present chapter addresses the origin of the observed
depolymerization processes and the nature of the endgroups of the POM polymer. The
reduced thermal stability of the POM film formed is also considered. The reaction of
adsorbed methanol (CH3OH) with H2CO on Cu(100) at 85 K was used to probe the
reaction mechanisms of the polymerization process and help in the identification of the

POM endgroups.

4.2 Results and Discussion

Isolation of POM species. The adsorption of H2CO on clean Cu(100) at 85 K
resulted in the formation of two POM species. Temperature-programmed desorption was
used to investigate whether the two types of POM observed to form on the Cu(100)
surface could be isolated through annealing. Figure 4.1 shows m=29 (HCOI) TPD
spectra over the temperature range where depolymerization of POM occurs. Figure 4.1a

shows TPD data obtained from a POM-saturated Cu(100) surface (1 ML, equivalent to a
fractional coverage, 0, of 0.69 H2CO molecules/Cu atom).l Two features were visible:
the a-POM species with a peak desorption temperature of ~ 207 K and the B-POM

species at ~ 219 K.2 Figure 4. lb shows the TPD data obtained after one monolayer of

62

 

HCO+ (m=29) intensity (arb. units)

 

 

 

I I I

 

T l
1 80 200 220
Temperature (K)

Figure 4.1 Mass 29 (HCO+) TPD spectra for (a) 1 ML H2CO on Cu(100), (b) 1 ML
H2CO after annealing to 209 K and cooling to 85 K prior to TPD and (c) 1 ML H2CO
annealed to 209 K and dosed with H2CO to re-saturate the surface. A constant linear
background was subtracted from each spectrum.

63

POM was briefly annealed (2-3 seconds) to 209 K and then cooled to 85 K prior to the
TPD experiment. Annealing to this temperature, the approximate peak desorption
temperature for the at species, resulted in the complete loss of the or species. Due to the
sirrrilarity of peak desorption temperatures for the or and B POM, annealing also resulted
in the loss of ~ 11% of the B species (based on fits discussed below). Additionally, a
TPD spectrum identical to Figure 4. lb was produced if the B species-covered surface was
allowed to remain at 85 K for up to 10 minutes.

Figure 4.1c shows the TPD data obtained after a monolayer of POM was annealed
to 209 K, cooled to 85 K, and then exposed to H2CO to re-saturate the surface with POM.
The data appear qualitatively similar to that shown in Figure 4.1a. The TPD results of
Figure 4.1 suggest immediately that the a species can be depopulated and repopulated by
an annealing and re-dosing procedure and the B species does not convert to the or species
during the TPD experiment. Irnportantly, the higher temperature B species can be
isolated through annealing.

Depolymerization kinetics. Equations can be derived to model the kinetics of the
depolymerization process for a monodisperse polymer. For bulk POM, there are two
limiting cases for depolymerization initiated at chain ends.3’10rll In the first case, once
initiated, the POM chain fully depolymerizes to monomer and is first-order with respect
to the mass of the polymer, W (proportional to coverage in our experiments). The rate of
monomer (M) evolution, with respect to time (t), is equal to the rate of radical (R)
production

M =sz

—— 4.1
(it dt ( )

where Z is multiple of the degree of polymerization, Dp (kap where b=1 for a
monodisperse sample). The rate of radical formation is

ER— : ki [11] (4.2)
dt

where k, is the rate constant for terminal initiation and [n] is the concentration of polymer

chains. Combining equations (4.1) and (4.2), and using equation (4.3)

 

 

W
[n] = (4.3)
MM)
where Mm is the molecular weight of the monomeric unit, results in equation (4.4)
EM 2 ki W (4.4)
dt M

The equation can be transformed into the rate of monomer evolution with respect to
temperature (T) by use of the heating rate B (dT/dt), resulting in equation (4.5).

M... W
dT ‘BMm

 

(4.5)

The rate of monomer evolution is independent of Dp if the kinetic chain length KL
(number of monomer units “unzipped” per initiation event) is equal to the degree of
polymerization Dp (chain length).

In the second case, following initiation, unimolecular termination competes with
depolymerization and only a small number of monomers are unzipped prior to
termination (KL < Dp). The rate of monomer evolution, as a function of time, can be
related to the radical concentration by

dM
—dt_ — k d[R] (4.6)

65

where kd is the rate constant of depolymerization. The rate of radical production is
dependent on the rate of initiation, k,, and the rate of termination, k, such that

95 = ki[n]- k,[R]. (4.7)
(It

Assuming steady state conditions for equation (4.7), using equation (4.3) and the heating
rate, equation (4.6) becomes

dM_ kide
dT Bk,MmDp'

 

(4.8)

If the number of chains remains constant throughout the depolymerization
(number of chains = W/Dp), this process is zero-order with respect to the mass of
polymer. This condition would be met only during the initial stages of depolymerization
as short sections of each chain are removed and both W and Dp decrease at the same rate.
However, following prolonged depolymerization, the kinetic chain length (KL: kd/k,)
eventually becomes comparable to Dp. At this stage, the total number of chains on the
surface begins to decrease, the ratio W/Dp is no longer constant, equation (4.8) becomes
first-order with respect to the mass of polymer and thus is identical to equation (4.5).
Consequently, depolymerization should initially exhibit zero-order behavior (as described
by equation (4.8)) but become first-order (as described by equation (4.5)) during the later
stages of depolymerization. The temperature at which the apparent order changes will
depend on the kinetic chain length and initial degree of polymerization and occurs when
KL=Dp.

Equations (4.5) and (4.8) were used to fit TPD data both as a function of heating

rate and increasing coverage. Both coverage and heating rate data were fit to increase the

66

 

  
 

I'ITI'I'II

 

.
o
O o.
O .9
::::

 

IITIIIIIIIIlfijII'IIIIlIII

I'IrIlIljl‘r

 

 

I l I I I I I I I I I l I I I I l I I I I l I I I I

I'I'IrI'Ilr.I

   

.O
‘‘‘‘‘
O
....

 

HCO+ (m=29) intensity (arb. units)

 

I I I I I I I I I I

  

o
o u
o .0
..

 

 

 

fI'IIII'IIII'I III I 'jjI II I I I r

1 90 200 21 0 220 230 240
Temperature (K)

Figure 4.2 Mass 29 (HCO+) TPD spectra for a 1 ML coverage of H2CO on Cu( 100)
exposed at 85 K as a function of heating rate (B). The heating rates were (a) 1.5 Ks'l, (b)
3.8 Ks", (c) 5.6 Ks" and (d) 8.7 Ks". The solid line (—) is the raw data, the dotted lines

(...) are the fits to equations (4.5) and (4.8) and the dashed line (----) is the sum of the
fits.

67

 

 

 

I I I I I I I l I I I I l I I I I l I I I I l I I I I

L 0) 0:0.3 ML

 

 

I l' I I I I I I I I I lfi I I I I I I I I I I I I I

HCO+ (m=29) intensity (arb. units)

 

‘ . 6:0.1 ML
3) .-.
' ’1 K x1.7

 

o
o
O
O
'-
........
C.-

uuuuuuuuuuuuuuuuuuuuuuuuuuu

 

I I l I I I I l I I I ITrffiI I I fifil I I I I II

1 90 200 21 0 220 230 240
Temperature (K)

Figure 4.3 Mass 29 (HCO+) TPD spectra for increasing coverages (0) of H2CO on
Cu(100) exposed at 85 K. The coverages were (a) 0.1 ML, (b) 0.3 ML and (c) 0.6 ML.

The solid line (—) is the raw data, the dotted lines (- - -) are the fits to equations (4.5) and
(4.8) and the dashed line (----) is the sum of the fits.

68

reliability of the fit parameters. . Equation (4.5) alone produced a satisfactory fit to the
higher temperature B-POM peak as shown in Figures 4.2 and 4.3. Since common leading
edges were observed for the a—POM species as a function of coverage, equation (4.8) was
used to fit the TPD data for (it-POM. However, such an approach failed to accurately
reproduce the majority of the or peak shape, particularly near the peak maximum. As
such, a combination of equation (4.5) and (4.8) was used to fit the or-POM species and
values for k, and KJJDp determined. The rate constant kg was assumed to be of an
Arrhenius form and preexponential and energetic terms extracted from the coverage and
heating rate fits. Similar values cannot be determined uniquely for kd and h, only the
ratio KUDp, assumed to be temperature independent in our model. The effect of varying
KIJDp is illustrated in the simulated TPD spectra of Figure 4.4. When KIJDle the rate
of monomer evolution follows first-order kinetics and when KIJDp<<l the kinetics

approach zero-order. At intermediate values (approximately 0.1<KUDp<1), the shapes

of the leading and trailing edges are sensitive to the ratio KUDp.

Figure 4.2 shows TPD data for a saturation coverage of POM on Cu(100)
collected at different heating rates along with the individual fits to the 0t- and B-POM
TPD peaks and the sum of the fits. For both species, the maximum depolymerization rate
shifted to higher temperature and the separation between the features decreased as the
heating rate increased. Figure 4.3 shows TPD data as a function of increasing initial
coverage along with the individual fits as described above. For both Figures 4.2 and 4.3,
the TPD data can be modeled reasonably accurately. Discrepancies between the fit

and experimental data could arise from the effects of polydispersity present in the or and

69

 

    

 

P 1
A . ‘r
.22 KL/DP - :‘a
'E h .1 r:
3 —1.0 ll :1
. ' _-_. ii -'i
e L 0'7 ii 1':
£9, - """ 0-5 ii i!
c - ----- 0.2 ii;§
«9 . - ------ 01 !!i l
‘_5 . I !,-' i
g ‘ , 1: i
<1) ' .' 1’! g i
8 ' .r'~ -'- i
E ' l' ‘1, 'li ;
O t . g
g 1 s
, .
E i 'i
“5 i =
I
.9 ‘- =
C0 I
a:

 

 

Temperature (K)

Figure 4.4 Simulated TPD spectra showing the effect of varying KUDp. The following
constant parameters were used for all spectra: E = 90 kJmol'l, v = 1x10”, and B = 5.6 Ks'
r

70

B species, variations in KL with temperature or non-infinite pumping efficiency for
H2CO. The fit parameters are summarized in Table 4.1.

The preexponential and energetic terms for the or-POM species represent values
for the ratio kixkd/kt whereas the terms for B-POM are for k, only. The ratio KL/Dp was
varied to improve the accuracy of the fit, especially in the leading edge of the (it-peak.
For the data shown in Figure 4.2, a constant value of KLle=0.7 was found to best fit the
data for the Ot-peak. In contrast, KL/Dp for the coverage dependant data shown in Figure
4.3 decreased from 1.0 (strict first order) at 0.1 ML to 0.8 at 0.6 ML. This immediately
implies the degree of polymerization for the a-peak increased as a function of coverage
and the number of chains (W/Dp) does not change up to the saturation coverage. A fixed
number of chains could arise during polymerization initiated at a small number of surface

sites such as defects or step edges.

Table 4.1 Results of the fits to equations (4.5) and
(4.8) for the data shown in Figures 4.2 and 4.3.

 

 

a-POM B-POM
E (kJ/mol) 91 (2r. 2):: 60.1 (:1: 0.7)
v 9 (i 1) x102| 9 (i 1) x1012

 

a) Average values from Figures 4.2 and 4.3.
Errors represent 16.

The results of the fits also suggest that, if the same depolymerization mechanisms
are operative for both 01- and B-POM, the (at-species must be associated with long chain

POM molecules (since KL<Dp) while the B-species is associated with short chain POM

71

 

 

  
 
  

 

 

 

1.0
H2CO / Cu(100) 85 K
0 8 _ Annealed (209 K) &
' redosed (85 K)
,3 0.6 -
(D
.8
.E
“O
Q)
.N
E; GA,-
0
2 As prepared (85 K)
0.2 -
a)
’ k 3 x50
0.0 J

 

 

 

II IIIIIIIIIIIIIIIIIII IIrI'IIIj'IIIIIII

I ' l
0 1 000 2000 3000

Energy loss (cm'1)

Figure 4.5 Electron energy loss spectra of (a) 1 ML H2CO on Cu(100) exposed at 85 K
and (b) 1 ML H2CO annealed to 209 K and dosed with H2CO to re-saturate the surface.

72

molecules (KL/Dle).

Vibrational analysis. The EEL spectra of the POM layer formed from the
exposure of H2CO at 85 K and that formed through the annealing (209 K) and re-dosing
procedure were compared. Figure 4.5a shows the spectrum of the POM formed through
exposure of H2CO at 85 K. All features can be attributed to POM and closely match with
both IR data from solid POM‘Z’13 and EELS data from POM formed on Cu(110)8 and
Ag(111).9 Adventitious carbon monoxide (CO) was responsible for the loss observed at
2050 cm“1 (v(CEO)). Figure 4.5b shows the EEL spectrum formed through the process
described for Figure 4.1c: saturation of the surface with POM, annealing to 209 K to
remove the 01 species, and re-dosing H2CO to 1 ML total coverage. Within our
experimental resolution and signal-to-noise ratio (S/N), the spectrum appears to be
identical to that shown in Figure 4.5a. The EELS data are consistent with the conclusion
that the or and B layers formed through the two different procedures (dosing H2CO at 85
K versus annealing and re-dosing) are identical.

Electron energy loss spectroscopy was used to investigate the B species further.
Figure 4.6 shows EEL spectra for separate monolayer coverages of POM prepared at 85
K followed by brief annealing to the temperature indicated. Figure 4.6a shows the EEL
spectrum for a POM layer that has been annealed to 140 K. All loss features, except
those at 800, 1020, and 1120 cm”, can be assigned to POM or adsorbed CO. Results for
Figure 4.6a are summarized in Table 4.2, along with IR data for solid POM. All EEL
spectra for monolayer POM acquired for annealing temperatures between 85-190 K were

similar to that shown in Figure 4.6a. In contrast, Figure 4.6b shows the spectrum

73

 

 

  
 

 

  

 

1.2
HZCO / Cu(100) 85 K d) 300 K
- a»... W “2.31.... .1 -- . $529.
1.0 -
c) 209 K
0.8 - x50
i?
(D
a, .
E
.E
8 05 _ b) 194 K
'75,, x50
E I u) C
2 a =
0.4 -
0.2 —
a 8 a) 140 K
. x50
0.0

 

 

 

IIIIIIIIIII'IIIIIIIIIIIIIIIIIIrlII—I

0 1000 2000 3000
Energy loss (cm'1)

Figure 4.6 Electron energy loss spectra of 1 ML H2CO on Cu(100) exposed at 85 K and
annealed to (a) 140 K, (b) 194 K, (c) 209 K and (d) 300 K. Each spectrum is a separate 1
ML H2CO dose.

74

2333an .355 Q.

 

 

75

2% 3% $3 8% mmmm 20m 88 \ £81;

8% no: 88 2 am 58 \ EB;

$8 88 $8 88 6mm:

509

was we: 8: $9 :3 \ £0me

82 8: ME: we: 38 \ 2:83

25 32 25 £2 38 \ 35:

3.: ca _ 82 02 _ as 38 \ 36:
8: cm:

8: as. o2 _ ms 1 52 60on
owo ONO. mvo o~o1

8w as 5 m3 wmw N8 809;
o3 8w OE 8m

80 08 o8 mmc ms 9% 690%

m; 8m m? 609m

9% an 02 80.2;

E owm o3 5-6:

am am 1m 8m 1x 9: b. o: 2.22.8 2.22.8
85 42 _ 0%: 42 _ 85 42 _ cum: d2 _ $22 3-20m Seams?

 

.33205 Amn-o:o_>:uoE>xo;_om can 3:205 ASbESSofixon—om 23m “8 8% M: 2m :32? om? ”M
mm 3 60— V50 :0 OUNQ Ea CONE ho 39238 1:2 _ e8 3mm E @3530 3.50 as momma: 232353 2: mo £=oE=m8m< NV 035‘

obtained after annealing to 194 K. The losses attributable to POM between 600 and 1500
cm’l decreased slightly in intensity while the losses at 800 and 1020 cm'1 remained
approximately unchanged. Also, the S/N has increased in Figure 4.6b suggesting a
modest increase in order for the annealed surface.

After heating to 209 K to fully remove the or species, the spectrum shown in
Figure 4.6c was obtained. Based on TPD data shown Figure 4.1, this spectrum should be
entirely due to the [3 species at a fractional coverage of ~ 0.4. A new loss appeared at 280
cm", which was tentatively assigned to v(Cu-O) based on EELS studies of O/Cu(100).14
The detection of this mode was enhanced by the large SIN increase observed after
annealing to 209 K, implying the [3 species was more ordered than the mixture the two
species. While the losses were less intense, all modes except the strong losses at 800,
1020 and 1120 cm", were attributable to POM. The v(CH) region showed two resolved
losses at approximately 2850 and 2965 cm".

Annealing above 209 K resulted in the decrease of all mode intensities. Figure
4.6d shows the EEL spectrum obtained after annealing to 300 K. Complete
depolymerization of both adsorbed polymer species occurred and a clean surface was
recovered. The observed losses at 345 and 2087 cm'1 were due to CO re-adsorbed during
sample cooling to 85 K.

Electron energy loss spectroscopy, combined with the annealing studies, was also
used to investigate the adsorption of formaldehyde-dz (DZCO) to determine the origin of
the losses at 800, 1020, and 1120 cm". Figure 4.7 shows the adsorption of 1 ML of

H2CO (DzCO) at 85 K (mixture of a and [3 species). All observed losses, except those

76

 

1'0 ' 1 ML H2CO / DZCO annealed to 140 K

          

0.8 -
92‘
2 0.6 _ b) DZCO
.9
.9 X100
P O ,
E e
a ‘-
§ 0.4 - I3.
0 ‘—
z ' |

  

a) H2CO

0.2 b g

ML

......... ,fi.fi....................
0 1000 2000 3000

 

x50

   

 

 

 

 

 

Energy loss (cm")

Figure 4.7 Electron energy loss spectra of (a) 1 ML H2CO and (b) 1 ML D2CO followed
by a brief anneal to 140 K.

77

previously mentioned, can be assigned to POM (POM-d2). The losses for Figure 4.7 are
summarized in Table 4.2. The vibrational frequencies obtained following DzCO
adsorption (shown in Figure 4.7b) match well with IR data for solid POM- d; and the
expected isotope shifts were observed”,13 Importantly, the losses at 800 and 1020 cm"1
(Figure 4.7a) were observed to shift to 760 and 945 cm'1 upon isotopic substitution,
corresponding to isotopic shifts (v(POM-h2)/ v(POM-d2)) of 1.07 and 1.08, respectively.
These ratios are indicative of normal modes involving primarily carbon-oxygen motion in
POM, although unambiguous assignment is difficult due to the large intramolecular
coupling of modes observed for solid POM.

The identification of the loss for POM-h; observed at 1120 cm'1 is hampered by
the coincident loss observed for POM-d2 at 1150 cm'l, assigned to vas(OCO). If the loss
at 1120 cm'1 for POM-h; involved pure carbon-hydrogen motion, isotopic substitution
would shift this mode, assuming an isotopic ratio of 1.4, to ~ 816 cm]. For POM-d2, this
region is dominated by a strong loss at 825 cm’1 (vs(OCO)) thus obscuring the loss if
present.

Figure 4.8 shows EEL spectra of 1 ML of H2CO (DzCO) annealed to 209 K to
fully remove the or species and isolate the B species. The results for Figure 4.8 are
summarized in Table 4.2. Figure 4.8a is similar to that shown in Figure 4.6c. Upon
annealing to 209 K, POM-d2 shows similar behavior to that seen for POM-hz. Namely,
the losses attributable to POM-d2 decreased in intensity while the losses at 750 and 980
cm'1 (800 and 1020 cm‘1 in POM-hz) remained relatively constant. As with POM-hz, the

S/N increased upon annealing and the expected isotope shifts were observed. The broad

78

 

 

 

 

 

 

 

 

 

 

 

 

1.0 - 1 ML HZCO / DZCO annealed to 209 K
F
0.8 -
L
g; b) DZCO
9:3 0'6 _ X100
.9
U
G)
.u
“a W
E 0.4 -
0
2
0.2 . I
a) H200
L x100
on J., ......... . ........ ., .. ........ . .......
0 1000 2000 3000

Energy loss (cm'1)

Figure 4.8 Electron energy loss spectra of (a) 1 ML H2CO and (b) 1 ML DZCO followed
by a brief anneal to 209 K.

79

loss observed at 860 cm'1 probably arises from overlap of two losses: vs(OCO) of POM-
d; and the loss at ~ 816 cm'1 (red-shifted from 1120 cm’1 for POM-hz), as was previously
discussed. A loss at 271 cm"1 was tentatively assigned to v(Cu-O). The isotope shift of
1.08 observed upon deuteration suggests the loss is neither primarily derived from
hydrogen (deuterium)-carbon motion nor strictly v(Cu-O) for an isolated oxygen atom. It
probably arises from isotopic substitution on an atom adjacent to an oxygen atom bound
to the surface. The origin of this loss as well as the other unassigned losses at 800, 1020
and 1120 cm“ will be discussed below.

A possible conformation of the B-POM species can be proposed based on the
EELS data in Figure 4.8. The observation of the vas(OCO) mode for both POM-h; and
POM-d2, coupled with the surface selection rule, eliminates the planar zigzag
conformation for the B-POM species. The most probable conformation is helical, as
observed in bulk POM, and as proposed for poly(acetaldehyde) on O/Ag(111).‘5’16

Endgroup identification. The EEL spectrum shown in Figure 4.8a corresponds to
the B species. If the chains are short enough, as suggested by the TPD fits, the
concentration of endgroups for the B species may be sufficiently large to be visible in
EELS spectra. Endgroups might be responsible for losses at 290, 800, 1020, and 1120
cm'1 for the B species which cannot be ascribed to bulk POM.

Likely endgroups for POM on Cu(100) include -OH, -OCH3, and -O-Cu. The
identity of the endgroups can be elucidated by comparison of the vibrational spectra of
small molecules containing these functionalities adsorbed on Cu(100). For example, for

monolayer methanol adsorbed on clean Cu(100), losses are observed at 610-810, 1020,

80

1130, 1450, 2850, 2945 and 3290 cm'1 that can be assigned to 8(OH), v(CO), p(CH3),
5(CH3), v(CH) and v(OH), respectively.17v18 Methoxide shows a similar spectrum,
except for the lack of modes due to the hydroxyl group, and an additional loss at 295 cm’1
that can be attributed to v(Cu-O).17'19 The losses at 280, 1020 and 1120 cm", for our
POM-h; (0t and B), closely match with v(Cu-O), v(CO) and p(CH3), respectively of
adsorbed methanol or methoxide. Unfortunately, while all expected modes for
methanol/methoxide are observed in our EELS data, they are either poorly resolved or
overlap with features due to the POM backbone modes. Definitive assignment of the
endgroups is also hampered by the poor S/N seen for the mixed POM overlayer.

Upon annealing to 209 K to remove the or species, the losses due to the POM
backbone lose intensity. This observation is consistent with the loss of longer POM
chains while the shorter chains, having fewer monomer units per chain, show less intense
POM losses relative to those of the endgroups. The losses at 1020 and 1120 cm", for the
adsorbed POM-h; shown in Figure 4.8a, do not shift with annealing. While consistent
with the v(Cu-O) of an isolated methoxide species, the mode at 280 cm’1 (B species) is,
instead, assigned to a POM species bound to the surface through a terminal oxygen atom.
Deuteration shifts this loss to 271 cm'1 indicating it is not an isolated oxygen atom but is
likely due to the attachment of an O atom to an adjacent CH2 (CD2) group within the
POM chain.

For POM-d2 shown in Figure 4.8b, the loss assigned to vs(OCO) at 860 cm’1 is
probably comprised of two losses: the vs(OCO) of POM- d; and the p(CD3), accounting

for the apparent broadening. Based on the observed isotope shifts and the close match of

81

 

the losses with those of a methoxide species, we conclude the B species is short chain
length POM with one end bound to the surface through an oxygen atom and the other end
terminated with a methoxide group, written as -Oa-(CHzO)n-CH3, where O3 is the oxygen
atom of the POM chain bound to the surface. The number of monomer units involved in
the or and B species is unknown and under further investigation. The mechanism of
formation of the B species will be discussed below.

The loss observed at 800 cm", for POM-h; shown in Figures 4.7a and 4.8a, is
more difficult to assign. Comparing the frequency with reasonable candidates, it matches
best with the 5(OCO) mode of adsorbed formate observed at 758 and 780 cm‘1 on
Cu(100)”,21 and Cu(110),22’23 respectively. Two characteristic losses for adsorbed
formate/formic acid have been observed at ~ 1360 and ~ 1070 cm'1 that can be assigned
to \I,(OCO)20v21 and 1t(CH),23 respectively. Although we observed losses at similar
frequencies, the shifts upon deuteration do not match those expected for formate. The
origin of 800 cm'1 loss could arise either from the bending motion, 5(OaCO), or a 150 cm'
1 red-shifted v(OaCO) mode of solid POM-h; caused by anchoring one of the oxygen
atoms to the surface.

The presence of isolated formate, methoxide and methanol on the Cu(100) surface
can be eliminated by comparing the adsorption and desorption behavior of formic acid
and methanol. Methanol exposed to clean Cu(100) has been found to adsorb molecularly
and desorb at 179 K.17,18 Methoxide is formed following methanol exposure to
preoxidized Cu(100) at ~ 120 K and is stable up to 400 K.17 Likewise, formic acid

adsorbed on Cu(100) deprotonates to form formate. This formate species decomposes to

82

 

  
  

 

 

 

 

- b) CHSOH dosed first
,_ o L CH30H
\

l-

" 0.2 L CH30H
’3‘: ..
(U
E T
8
9 _
.5 I F .
fi ' a) H2CO dosed first
+15; - O L CH30H \
o ' 0.2 L CH30H
0 _
I ‘ “WWW

. W.,,

. I

I
L "
,l
,
I I ' a I f r I ' i fi
1 80 200 220 240

Temperature (K)

Figure 4.9 Mass 29 (HCO+) TPD spectra after the adsorption of (a) 0.2 L methanol
(CH3OH) on the Cu(100) surface pre-covered with 0.6 ML H2CO and (b) 0.6 ML H2CO
on the Cu(100) surface pre-covered with 0.2 L CH30H. The “0 L CH3OH” spectrum is

the same for both (a) and (b).

83

CO2 and H2 at 420 K.2031 In the present work, a clean surface was generated upon
annealing to 300 K indicating no isolated methoxide or formate species were formed.
Thus, isolated methoxide, methanol and formate can be eliminated as the cause of the
losses observed at 290, 800, 1020 and 1120 cm'1 for the data shown in Figure 4.8a.
Reaction of CH ,0H and H2CO. Since formaldehyde is known to react with

methanol to form compounds of the type, CH3O-(CH2O)n-H,3 and our EELS data is
generally consistent with such a species, the reaction of H2CO with CH3OH adsorbed on
Cu(100) was investigated. Figure 4.9 shows TPD data obtained following the reaction of
H2CO with CH3OH on Cu(100) at 85 K. Figure 4.9a displays the TPD data obtained
before and after exposing a POM-covered surface to CH3OH. Upon exposure to CH3OH,
little change in the TPD was observed. In contrast, when H2CO was exposed to a surface
pre-covered with methanol, a dramatic change was observed in the TPD data as shown in
Figure 4.9b. The proportion of the or species decreased while the B species increased. In
both cases, neither a change in peak position nor width was evident. These data suggest
that the POM-covered surface is non-reactive towards CH3OH but the presence of pre-
adsorbed CH3OH during H2CO exposure appears to decrease the formation of long chain
POM species (or). It is possible that CH3OH is controlling the preferential formation of
the B species indirectly, for example by site blocking. However, since no CH3QH was
observed to desorb from the surface after the exposure sequence shown in Figure 4.9b,
we discount this indirect role. Most likely, CH3OH terminates the polymerization to
form increased quantities of short chain POM species of the type -Oa-(CH2O)n-CH3, as

previously discussed.

84

 

 

 

 

 

 

 

 

 

 

 

 

LCH3OH+HZCO/Cu(100)
1.0-
0.8— C)
x150
g, L i
(I)
g .
g 0.6-
8 b)
N .
‘3 x75
E
g 0.4-
0.2-
. 6)
JL i I x150
0.0
..n.......,..............m...].n....
0 1000 2000 3000

Energy loss (cm‘)

Figure 4.10 Electron energy loss spectra of (a) 0.2 L CH3OH, (b) 0.6 ML H2CO dosed
onto the Cu( 100) surface pre-covered with 0.2 L CH3OH and (c) the spectrum in (b)
annealed to 209 K.

85

Electron energy loss spectroscopy was used to confirm that CH3OH was incorporated
into the B-POM species. Figure 4.10a shows the EELS spectrum of 0.2 L CH3OH (~
0.15 of saturation) adsorbed on Cu(100) at 85 K. The position and intensities of the
observed losses match well with previous investigations for molecularly adsorbed
methanol.“18 Exposing this CH3OH-covered surface to 0.4 ML H2CO results in the
spectrum shown in Figure 4.10b. New losses are evident that can be assigned to POM
and the resulting spectrum is similar to that shown in Figure 4.5a. Importantly, losses

due to 5(OH) are absent indicating CH3OH has been deprotonated, although the detection

of v(OH) is hampered by its small dipole scattering cross section.24 Also, the losses at
290, 1020 and 1120 cm", discussed previously, are evident in Figure 4.10b confirming
these losses arise from the methoxide endgroup of the POM species.

Annealing the overlayer shown in Figure 4.10b to 209 K results in the EEL
spectrum in Figure 4.10c. Again, the data appear similar to that seen for POM formed in
the absence of CH3Ol-I (Figure 4.5a). The loss at 800 cm'1 was still observed, along with
those due to the methoxide endgroup and POM. Interestingly, the losses due to p(CH3)
(from the endgroup) and r(CH2) (from POM) are more intense than those seen in Figure
8a. Additionally, the loss due to v(Cu-Oa) at the chain end, observed at ~ 300 cm", is
also more intense. This is consistent with an increased number of chains, and thus chain
ends, due to CH30H reacting with the growing chain and effectively terminating the
polymerization.

A reaction scheme for the formation of the B species can be proposed based on

the similarities of the species formed with and without CH3OH during the adsorption and

86

polymerization of H2CO. The most likely initiation reaction is that the copper surface
acts as a weak Lewis acid towards the oxygen of H2CO upon adsorption.25 Solution
phase H2CO is known to polymerize in the presence of Lewis acids,3 and many oxygen
containing molecules (alcohols,l7 ethers,26 acetaldehyde,27 and formic acid20»22v23) are
known to interact with copper surfaces through the oxygen lone pairs. We speculate that
for H2CO on Cu(100), propagation proceeds by reaction of H2CO molecules with an
activated carbon atom adjacent to the surface bound oxygen. The growing chain is
expected to have a lower mobility on the surface than the monomer: H2CO has been
calculated to have a small binding energy of ~0.1 eV on Cu(111) and hence will be
highly mobile at 85 K.23 The polymerization likely terminates by abstracting a hydrogen
atom from the surface, known to adsorb on copper surfaces at 85 K.2931 This series of

reactions can be written as

H2CO(g) + Cu —+ Cu-O-C*H2 (4.9)
CU-O-C*H2 + DCH20 —) Cll-O-CH2(OCH2)D.1-OC*H2 (4.10)
Cu-O-CH2(OCH2)n-l-OC*H2 + Ha —) Cu-O—CH2(OCH2)n-1—OCH3 (4.11)

where C* is the growing chain end. The initiation and termination reactions are
consistent with the EELS losses observed at 290, 1020 and 1120 cm’1 that are assigned to
v(Cu-O), v(CO) and p(CH3), respectively.

The reactions occurring when CH30H is present during polymerization are

similar to those shown above. The initiation and propagation steps shown in (4.9) and

87

(4.10) are identical but, instead of termination by a hydrogen atom, nucleophilic attack by
CH3OH on the growing chain end occurs, followed by loss of a proton to the surface.
This terminates the polymerization and forms the methoxy endgroup. This reaction can

be written as

CU-O-CH2(OCH2)n-1-OC*H2 + CH3OH3 -) CU-O-CH2(OCH2)n-OCH3 + Ha (4.12)

The reactions that lead to the longer chain species (or species) are unknown at this
time. A full description is limited by the poor S/N observed when both species are
present and the inability to isolate the on species from the B. While the or species contains
~ 1/3 of the H2CO molecules present within the POM overlayer (at 1 ML total coverage),
it is possible few chains of the longer species exist, thus lowering the likelihood that the

losses from the endgroups would be observed in EELS.

4.3 Conclusions

Formaldehyde polymerizes spontaneously on the 85 K Cu(100) surface forming
long chain (or) and shorter chain (B) poly(oxymethylene) (POM) species. Equations
describing the kinetics of depolymerization suggest the number of chains of the or species
is constant for increasing coverage and determined at low coverages by a fixed number of
surface initiation sites. However, the equations alone cannot explain the presence of two
species based solely on differences in Dp; for equal number of monomer units, the longer

chain species should depolymerize at a higher temperature. As discussed in Chapter 3,

88

conformational differences are most likely the cause of the two depolymerization features
observed in TPD.

Upon annealing to 209 K to fully depolymerize and remove the or species, losses
are observed attributable to the endgroups of the helical B species. Losses observed at
1020 and 1120 cm'1 can be assigned to v(CO) and p(CH3), respectively, of a methoxy
endgroup while a loss at 290 cm‘1 is indicative‘of the end of the POM chain bound
directly to the surface through an oxygen atom. This species can be written as -Oa-
(CH2O)n-CH3. A loss at 800 cm'1 is believed to arise from a mode associated with the
surface-bound oxygen. Compared to bulk POM, the lower thermal stability observed for
monolayer B-POM on Cu(100) most likely results from the chain being bound to the
surface through the oxygen atom, resulting in a lower barrier to initiation of
depolymerization. Methoxide endgroups are known to increase thermal stability in

POM.3 The weakening of the POM backbone due to interaction with the surface and its
influence on the thermal depolymerization is minimal, based on the similarities of the
vibrational frequencies for the adsorbed and bulk POM.

Pre-adsorbed methanol was found to terminate the polymerization and favor the
formation of the B—POM species. The thermal stability and EEL spectra of the POM
overlayer formed upon reaction with CH3OH was similar to that of B-POM formed
through the direct adsorption of H2CO. However, the losses due to the endgroups were
more intense than the POM layer formed by H2CO adsorption alone, indicating the B
species formed were shorter than those formed through the direct adsorption process.

The ability to preferentially form a specific chain length polymer with a specific

89

endgroup may open the possibility to control both thin film order and morphology and

tailor thermal stability based on the interaction of the endgroup with the surface.

4.4 References

(1) Bryden, T.; Garrett, S. J. Phys. Chem. B. 1999, 103, 10481.

(2) The 2-3 K difference in absolute peak desorption maxima observed and reported
in Chapter 4 and the previous chapter was ascribed to heating rate and thermocouple
position variations.

(3) Walker, J. F. Formaldehyde, 3rd. ed.; Reinhold Publishing Corp.: New York,
1964.

(4) Kambe, H. In Aspects of Degradation and Stabilization of Polymers; Jellinek, H.,
Ed.; Elsevier: Amsterdam, 1978; pp 393.

(5) Feger, C.; Franke, H. In Polyimides: Fundamentals and Applications; Ghosh, M.,
Mittal, K., Eds.; Marcel Dekker: New York, 1996; pp 759.

(6) Haq, S.; Richardson, N. J. Phys. Chem. B. 1999, 103, 5256.

(7) Plank, R.; DiNardo, J .; Vohs, J. Phys. Rev. B. 1997, 55, 10241.

(8) Sexton, B.; Hughes, A.; Avery, N. Surf. Sci. 1985, 155, 366.

(9) Fleck, L.; Feehery, W.; Plummer, E.; Ying, Z.; Dai, H. J. Phys. Chem. 1991, 95,
8428.

(10) Reich, L.; Stivala, S. Elements of Polymer Degradation; McGraw-Hill: New
York, 1971.

(11) Jellinek, H. H. G. In Aspects of Degradation and Stabilization of Polymers;

Jellinek, H. H. 6., Ed.; Elsevier: Amsterdam, 1978; pp 1.

90

(12) Tadokoro, H.; Kobayashi, M.; Kawaguchi, Y.; Koybayashi, A.; Murahashi, S. J.
Chem. Phys. 1963, 38, 703.

(13) Matsuura, H.; Murata, H. Bull. Chem. Soc. Jpn. 1982, 55, 2835.

(14) Sueyoshi, T.; Sasaki, T.; Iwasawa, Y. J. Phys. Chem. B. 1997, 101, 4648-4655.
(15) Tadokoro, H. In Macromolecular Reviews; Peterlin, A., Goodman, M., Okamura,
S., Zimm, B., Mark, H., Eds.; Interscience: New York, 1967; Vol. 1; pp 119.

(16) Sim, W.; Gardner, P.; King, D. J. Am. Chem. Soc. 1996, 118, 9953.

(17) Dai, Q.; Gellman, A. J. Phys. Chem. 1993, 97, 10783.

(18) Ellis, T.; Wang, H. Langmuir 1994, 10, 4083.

(19) Camplin, J. P.; McCash, E. M. Surf Sci. 1996, 360, 229-241.

(20) Dubois, L.; Ellis, T.; Zegarski, B.; Kevan, S. Surf. Sci. 1986, 172, 385.

(21) Taylor, R; Rasmussen, P.; Ovesen, C.; Stoltze, P.; Chorkendorff, I. Surf. Sci.
1992, 261, 191.

(22) Bowker, M.; Haq, S.; Holroyd, R.; Parlett, P.; Poulston, S.; Richardson, N. J.
Chem. Soc., Faraday Trans. 1996, 92, 4683-4686.

(23) Hayden, B.; Prince, K.; Woodruff, D.; Bradshaw, A. Surf. Sci. 1983, 133, 589.
(24) Ibach, H.; Mills, D. L. Electron Energy Loss Spectroscopy and Surface
Vibrations; Academic Press: San Diego, 1982.

(25) Fleck, L.; Howe, P.; Kim, J.; Dai, H. J. Phys. Chem. 1996, 100, 8011.

(26) Meyers, J .; Street, 8.; Thompson, S.; Gellman, A. Langmuir 1996, 12, 1511.

(27) Lamont, C.; Stenzel, W.; Conrad, H.; Bradshaw, A. J. Electron Spectrosc. Relat.
Phenom. 1993, 64/65, 287-296.

(28) J. Greeley and M. Mavrikakis, to be published.

91

‘\.
a

(29) Adsorbed hydrogen was assumed to arise from dissociation of molecular
hydrogen and/or formaldehyde at the filaments of the ion gauge and/or QMS.
(30) Chorkendorff, I.; Rasmussen, P. Surf Sci. 1991, 248, 35.

(31) Kammler, T.; Kuppers, J. J. Chem. Phys. 1999, 111, 8115.

92

 

Chapter 5 Photochemistry of Formaldehyde Adsorbed on CO-saturated
Cu( 100)

Abstract
The photochemistry of formaldehyde (H2CO) adsorbed on CO-saturated Cu(100) at 85 K
was studied using electron energy loss spectroscopy (EELS) and temperature-
programmed desorption (TPD). Formaldehyde was weakly adsorbed on CO/Cu(100) and
desorbed at 104 K, corresponding to a desorption energy of 18.2 (i0.8) kJ/mol.
Irradiation of the H2CO/CO/Cu(100) surface caused the molecularly adsorbed H2CO to
polymerize, forming poly(oxymethylene) (POM). Irradiation also caused the formation
of ethylene glycol (CH2OH)2. Losses observed at 870 and 3365 cm", after UV
irradiation, were assigned to v(CC) and v(OH) modes, respectively, of (CH2OH)2
indicating ethylene glycol was formed promptly upon irradiation. The presence of
(CH2OH)2 was confirmed by studying the adsorption of (CH2OH)2 on clean and oxygen-
covered Cu(100). The formation of ethylene glycol was likely governed by geometric

constraints present within the formaldehyde overlayer.

93

5.1 Introduction

Formaldehyde polymerizes spontaneously to form poly(oxymethylene) (POM), -
(H2CO),.-, upon adsorption on a variety of clean and oxygenated surfaces: O/Ag(110),l
Ni(110),2 Pt(lll),3 Pd(111),4 O/Rh(111),S O/Pd(111),6 NiO(100),7 Cu(110)8 and
Cu(100).9 Although the polymer has been positively identified through a variety of
techniques, in most cases little is known about the precise initiation, propagation or
termination mechanisms that lead to the polymer. The morphology, crystallinity,
conformation and chain length of the polymer produced are similarly poorly
characterized. Recently, formaldehyde molecularly adsorbed on Ag(111) has been
polymerized to POM using photons and electrons, offering the potential for precise
temporal and spatial control of creation and deposition of polymer monolayers.10~ll
Such control also increases the prospects for understanding surface polymerization
reactions at a more fundamental level.

We recently reported the facile thermal polymerization of formaldehyde on
Cu(100) at 85 K.9 The film appeared to be composed of two different types of polymer
chain with different conformation and/or chain length as inferred from vibrational
spectroscopy and thermal desorption/depolymerization measurements.12 We turn our
attention here to the photopolymerization of formaldehyde and the nature of the
polymeric film produced. In this case we use carbon monoxide as a spacer layer to
prevent direct interaction of the formaldehyde with the Cu(100) surface and thereby
inhibit spontaneous polymerization.

Co-adsorbed carbon monoxide (CO) is known to influence the chemistry

observed on surfaces. For example, cosadsorbed CO has been found to increase the

94

stability of ethylidyne on Ru(001),13 decrease the stability of the methyl hydrogen atoms
of toluene on Ru(001),l4 perturb the decomposition pathway of methylamine on
Ru(001)15 and promote the decomposition of saturated hydrocarbons on Ni(755).16
Carbon monoxide is weakly chemisorbed on Cu(100), desorbing molecularly at ~ 180
K.9 This desorption temperature is some 20-40 K lower than the decomposition
temperature of the POM polymer but about 80 K higher than the desorption temperature
of the formaldehyde monomer on CO/Cu( 100). Such a situation offers the possibility for
forming patterned polymer films through photopolymerization, followed by desorption of
monomer (from the unirradiated areas) and the CO spacer layer. In this way, it may be
possible to selectively deposit intact polymer onto specific regions of the Cu surface in a

controlled fashion.

5.2 Results and Discussion

Photochemistry of H2CO/C0/Cu(100). The adsorption and photochemistry of
formaldehyde (H2CO) on CO-saturated Cu(100) (CO/Cu(100)) was investigated using
temperature-programmed desorption. At 85 K, H2CO adsorbs molecularly and desorbs
as a single peak at 104 K as shown in Figure 5.1a (0 min.). This feature does not saturate
with increasing coverage and spectra for increasing coverages have coincident leading
edges (data not shown), suggestive of zero-order desorption. An Arrhenius plot of
ln(rate) vs. T1 indicated a desorption energy of 18.2 (:08) lemol, confirming H2CO is

physisorbed on CO/Cu(100). This desorption energy compares favorably to that seen

95

 

 

=30

x1

' m=30

=34
x400

Irradiation
time
30 min.

  

Irradiation .
15 min.

 

 

Ion mtensrty (arb untls)
. , .

' l
>§

 

- time .
5 min. A “I 5 min.
15 min.
0 min.
30 min.

 

 

¥ .
i

 

 

 

I I I I I I I j l I I

80 100 120 140 200 300 400
Temperature (K)

'rjIIlIIIIITi

Figure 5.1 Temperature-programmed desorption spectra for (a) m=30 (H2CO+) and (b)
m=30 (H2CO+) and m=34 (D2COD+) for a 1.1 ML coverage of H2CO (D2CO) on C0-
saturated Cu(100) as a function of UV irradiation time. D2CO was used to reduce
coincident mass fragment interferences.

96

for H2CO physisorbed on Ag(111) of 25 lemol.10 Importantly, no desorption features
were observed between 200-240 K where the polymer, formed through the thermal
polymerization on the clean Cu(100) surface, was found to depolymerize.9 This
immediately suggests the CO-saturated surface is inert towards thermal polymerization at
85 K and the CO inhibits the polymerization possibly through a site—blocking mechanism.
No other species were observed to desorb from the unirradiated surface except for the CO
monolayer, which desorbs at ~ 180 K for a saturation coverage.

Figure 5.1 also shows TPD spectra for a 1.1 ML H2CO on CO/Cu(100) as a
function of irradiation time. Irradiating the overlayer for 5 minutes caused a decrease in
the molecularly adsorbed H2CO intensity as observed in Figure 5.1a for m=30 (H2CO+).
Continued irradiation caused this feature to decrease further. The peak desorption
temperature was constant and the peak broadened slightly for increasing irradiation times.
New features were observed for m=30 after 5 minutes irradiation at ~220 and ~ 235 K as
shown in Figure 5.1b. These two features increase in intensity, as a function of
irradiation time, and merge into a large peak centered at ~ 230 K with a small shoulder at
~ 240 K. Measurement of other fragments associated with H2CO (HCO+ and H213CO+)
resulted in identical TPD spectra for the features at 104 K and between 220-250 K and
whose intensity ratios reflected that of the mass spectral fragmentation pattern for gas-
phase H2CO.l7 This confirms the features observed at 104 and between 220-250 K are
due to molecular H2CO. Importantly, comparison of pre- and post-irradiation TPD
spectra for CO (m=28), indicated no change in the desorption temperature or the

coverage of the saturated CO layer.

97

 

Molecular formaldehyde has been observed to desorb ~ 205 and ~ 220 K due to
the depolymerization of poly(oxymethylene) (POM) formed from the thermal reaction on
clean Cu(100).9 The desorption temperatures observed following irradiation in the
current work were similar to those observed for the depolymerization of POM formed
through the photopolymerization of H2CO on Ag(111).10 This suggest the features
observed between 220-250 K for m=30, as shown in Figure 5.1b, are due to the
depolymerization of POM formed upon photopolymerization of H2CO adsorbed on
CO/Cu(100). The presence of POM will be confirmed using EELS data as discussed
below. The slight shift to higher depolymerization temperatures, observed in Figure 5.1b,
could be related to changes in chain length or endgroup stability and is currently under
investigation.

A desorption feature was observed for m=31 at 350 K that did not show any
corresponding intensity at m=29 or 30 (data not shown). This feature is inconsistent with
molecular H2CO. To remove interference from coincident mass fragments and help in
the identification of this peak, the adsorption and photochemistry of D2CO on
CO/Cu(100) was investigated. Identical TPD data to H2CO were obtained for D2CO for
the features at 104 K and between 220-250 K after irradiation. The peak observed at 350
K for m=31 (H2CO) was detected at m=34 for D2CO, as shown in Figure 5.1b, indicating
the fragment contains at least 3 protons.

A number of small molecules can be eliminated as the source of the 350 K
desorption feature. The lack of intensity for m=29 excludes adsorbed alkoxides as the

source of the 350 K feature. These species are known to desorb at temperatures 2 350 K

on Cu(100),18 however, at these temperatures, alkoxides undergo B-hydride elimination

98

to produce the corresponding aldehydes, all of which show m=29 as the most abundant
mass fragment in their mass spectrum.17 Also, the lack of intensity at m=60, coupled
with lack of intensity at m=29, eliminates the simplest dialdehyde, ethanedial (CDO)2, as
the source of the feature at 350 K.

The most likely species that gives rise to the feature at 350 K is ethylene glycol.
This is consistent with previous experiments on the adsorption of ethylene glycol on
O/Cu(110)19 and O/Ag(110)20 where desorption features were observed at 390 and 365
K, respectively. In contrast to the present work, ethylene glycol was observed to undergo
B-hydride elimination to produce ethanedial on both O/Cu(110) and O/Ag(110). It is
possible, without surface oxygen present, molecular desorption is favored over ethanedial
generation for the ethylene glycol produced in the irradiation of H2CO adsorbed on
CO/Cu(100). The presence of ethylene glycol will be confirmed by the EELS data
discussed below.

The amount of ethylene glycol formed saturates after 15 minutes of irradiation, as
shown in Figure 5.1b, and was calculated using both TPD and XPS data. Initially, 1.1
ML of H2CO was adsorbed on CO/Cu(100). After 15 minutes of irradiation, ~ 80 % of
the original 1.1 ML H2CO desorbed as monomer (0.45 ML) or formed polymer (0.43
ML), leaving ~ 20 % unaccounted for H2CO. The lack of knowledge regarding the
electron impact ionization cross-section of ethylene glycol prevented quantification using
TPD. However, XPS analysis of a sample that had been irradiated for 15 minutes and
then annealed to 270 K to remove monomer, polymer and CO, revealed a coverage of

0.17 ML of carbon, consistent with the remaining 20 %.

99

 

 

 

 

 

 

 

1.1 ML H2CO + 0.57 ML CO/Cu(100)
' d)15 min.
- “1 A 450K
' x100
r
E .
'E
3 —
9'
g .
.4?
m ..
5 x20 x300
.E '
E - b)15 min.
E WW
0
z ”20 x100
a)0min.
85K
x4
x20 x150
l""l""l""l""
0 1 000 2000 3000 4000

Energy loss (cm'1)

Figure 5.2 Electron energy loss spectra for a 1.1 ML coverage of H2CO on CO-saturated
Cu(100) for (a) 0 minutes irradiation at 85 K, (b) 15 minutes irradiation at 85 K, (c) 15
minutes irradiation followed by an anneal to 270 K and (d) 15 minutes irradiation
followed by an anneal to 450 K. Each spectrum constituted a separate 1.1 ML coverage
of H2CO. The data were acquired at 85 K.

The adsorption and photochemistry of H2CO on CO/Cu(100) was investigated
using EELS. Figures 5.2a and 5.2b show EEL spectra for 1.1 ML H2CO adsorbed on
CO/Cu(100) after 0 and 15 minute irradiation. Figure 5.2a shows 1.1 ML H2CO on
CO/Cu(100) prior to irradiation. The losses observed at 350 and 2080 cm'1 are due to
v(Cu-CO) and v(CO), respectively, of the adsorbed CO and are observed at these
approximate energies for all the spectra shown in Figure 5.2. The other losses were
assigned to molecular H2CO and match well with IR data for crystalline formaldehyde.21
The frequencies are only slightly shifted from the solid-phase data confirming H2CO was
weakly adsorbed on the CO/Cu(100) surface. The observation of lattice modes at 200
and 230 cm’1 suggests the adsorbed H2CO was well ordered. Also, the observation of all
the modes of molecular H2CO indicates the molecule was adsorbed with the C-0 bond
axis tilted from the surface normal. This geometry would render the 13. (v4, v5) and B2
(v6) modes within the C2,. point group, to which H2CO belongs, visible according to the
surface selection rule.22 The data from Figure 5.2a is summarized in Table 5.1.

After 15 minutes of UV irradiation, the spectrum shown. in Figure 5.2b was
obtained. New modes appeared at 600, 950, 1120, 1480 and 2930 cm'1 that can be
assigned to the polymer, POM, and match well with both the IR data for solid POM23
and POM formed through the photopolymerization of H2CO adsorbed on Ag(111).lo
The data from Figure 5.2b is summarized in Table 5.1. The observation of the mode at
600 cm'1 is characteristic of the 5(OCO) mode of POM and eliminates trioxane (cyclic
trimer of H2CO) as a photoproduct. The corresponding mode for trioxane is split into

two bands which appear at 744 and 521 cm".24

101

 

 

@0388 8: n E
DUNE 82883390an E0: 20m 3

 

102

 

23 59>

cam 83 3 3+3

28% Emma 33 £83,

22% Emma 28. EB;

8% 3823 58...;

8% e352 £0?

$8 88 6mm;

8: 3,: E Gnu:

8.2 82 cm: :2 3582 50:2

22 35?

owe e332 .6

o: _ 3&2 _ saw

8. _ o. _ 2 52 609...;

oma 9a N8 509;
ohm

coo coo one 609m

9% cm 60-22

CMN DUOE 033w—

ONN OUOE 033m—

.E .56 2 .E .55 o 22.8 52.8
82 55598”: 82 EQOQOUN: .2 A _ 2 33:22 22 Comm anaemia.

 

.2 _ Sw< :0 20m 5m 8% 3mm 5:5 mac—w 20m can 00$.— 2.8 .5“ Sat Mn 2a :32? 82 .cocfithm >3 8558 2 can
o Sam 53 V30 93823-00 co CONE mo omab>8 1:2 2 a c8 notomno A780 .5 855 352223 05 Co mucoficwEm/x _.m 035.

Interestingly, the losses due to molecular H2CO, in particular the loss at ~1720
cm'1 for v(C=O), were not observed even though, based on TPD measurements, it makes
up ~ 40 % of the overlayer after 15 minutes of irradiation. The most probable
explanation for the absence of these modes is a change of orientation, rendering the
modes inactive according to the surface selection rule, coupled with the reduced signal-
to—noise (SIN) observed in Figure 5.2b. The absence of the strong phonon losses seen in
Figure 5.2a indicates the H2CO present after irradiation was no longer well ordered.

The spectrum in Figure 5.2b appears qualitatively different than that seen for the
thermal polymerization of H2CO on clean Cu(100).9 The losses observed in the thermal
polymerization at 1220 (r(CH2)), 1390 (w(CH2)), and 1470 (8(HCH)) cm'1 are much
more intense than those seen for the present work. As seen in Figure 5.2b, losses at 1220
and 1390 cm'1 were not detected. Additionally, the loss observed at 1030 cm'1 for the
thermally polymerized POM was also absent. This loss has previously been identified as
a mode (v(CO)) due to an endgroup of the POM chain.12 The differences in intensity
could be due to orientation effects and/or chain length dependence (the absence of the
endgroup mode in the present work implies longer chains).

There are two losses in Figure 5.2b that cannot be assigned to POM. The small
shoulder observed at 870 cm'1 is in the correct range for a mode due to v(C-C) while the
broad loss centered at 3365 cm'1 is probably due to v(OH). Adsorbed water could give
rise to the loss at 3365 cm", however, this loss was not detected prior to irradiation as
shown in Figure 5.2a. More likely, these two loses were due to the feature observed to

desorb at 350 K in the TPD data that was tentatively assigned to ethylene glycol.

103

 

"I; '

Annealing the overlayer shown in Figure 5.2b to 270 K resulted in the spectrum
shown in Figure 5.2c. This temperature was high enough to desorb the monomeric H2CO
(104 K), CO (180 K) and the POM (220-250 K) with the remaining losses due to the
feature observed to desorb at 350 K. The broad loss at 890 cm“1 matches well to that seen
in Figure 5.2b suggesting the species desorbing at 350 K was formed upon irradiation and
not during the temperature ramp for the TPD experiment. Another loss was observed at
1070 cm'1 which most likely corresponded to a mode involving carbon-oxygen motion.
The v(OH) mode at 3365 cm'1 was lost upon annealing and no v(CH) modes were
detected. The spectrum in Figure 5.2c matches well with that observed for deprotonated
ethylene glycol on O/Ag(110) where losses were observed at 890 and 1090 cm’I
corresponding to v(CC) and v(CO), respectively.20 Therefore, based on TPD and EELS
data, the species observed to desorb at 350 K was identified as ethylene glycol.

Annealing the sample to 450 K resulted in the spectrum shown in Figure 5.2d.
All losses due to molecular H2CO, POM and ethylene glycol were absent; the species
formed during irradiation had desorbed. The observed losses were due to CO re-
adsorbed during sample cooling.

Ethylene glycol on Cu( 100) and 0/Cu( 100). The adsorption of ethylene glycol
(CH2OH)2 on clean and oxygen-covered Cu(100) was studied using TPD. Figure 5.3
shows m=31 (CH2OH+) spectra for three exposures of (CH2OH)2 on clean Cu(100). A
single desorption feature was observed at 237 K for a 0.1 L exposure. For increasing
CXposures, this peak grew in intensity and the peak desorption temperature remained
unchanged. Above 0.1 L, a new feature was observed at 220 K that neither saturated nor

shifted with increasing exposures. Measurement of other mass fragments of ethylene

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ (CHZOH)2/Cu(100)
70‘ _ ,é‘ _ 0.5L(CHZOH)2+
:55 g ? O/Cu(100)
:3 .93
' - .E
E _ .5
3 1-
.? (III)
a) " N
E E
g ' ..............
:1 300 350 400 450
F"): Temperature (K)
N n
E
E: L
ON
I .
0
Increasing
' exposure
I ' ' . ' I T . ' ' I ' 1 ' r I ' '
100 200 300 400

Temperature (K)

Figure 5.3 Temperature-programmed desorption spectra for m=31 (CH2OH+) following
exposures of 0.1 L, 0.2 L and 0.5 L of ethylene glycol (CH2OH)2 on clean Cu(100). The
inset is the TPD spectrum obtained for a 0.5 L exposure on a Cu(100) pre-covered with
0.1 ML oxygen showing a new feature desorbing at 347 K.

105

glycol confirmed (CHZOH)2 adsorbed and desorbed molecularly. The features at 220 and
237 K are assigned to multi- and monolayer states, respectively. These temperatures
match well with those seen for (CH20H); adsorption on clean Ag(l 10) where multi- and
monolayer states were found to desorb at 205 and 225 K, respectively.20 Molecularly
adsorbed (CHZOH)2 was found to desorb at ~ 220 K from the Cu(110) surface,19
consistent with the current results. Assuming first-order kinetics for the state at 237 K on
Cu(100) and a preexponential factor of 1013 s", a desorption energy of 59 kJ/mol was
estimated. This value compares favorably to the 60 kJ/mol desorption energy found for
the monolayer state on Ag(110).20

The inset of Figure 5.3 shows the m=31 TPD spectrum of a 0.5 L exposure of
(CH20H)2 on a Cu(100) surface pre-covered with 0.1 ML oxygen. The desorption of the
multi- and monolayer states were unaffected by the oxygen. A new feature was observed
at 347 K which corresponds to desorption of (CH20H)2 that had reacted with the
adsorbed oxygen forming a dialkoxide species. It is known that alcohols, including
ethylene glycol, adsorb molecularly on oxygen-covered copper and silver surfaces at
temperatures < 120 K.18'20 Heating the sample above ~ 200 K causes the alcohols to
deprotonate by reaction with the adsorbed oxygen, desorbing as water and forming a
surface-bound alkoxide species. These species are stable to temperatures 2 350 K where
they desorb molecularly in competition with production of the corresponding aldehyde.

The result shown in the inset of Figure 5.3, showing desorption at 347 K, is
consistent with previous experiments on O/Cu(110)19 and O/Ag(110)20 where this
dialkoxide state was found to desorb at 395 and 365 K, respectively. Importantly, the

temperature observed for the desorbing dialkoxide species matches closely with that for

106

 

la!

 

 

 

 

 

 

1-4 " b) (CHZOH)2 + 0.1 ML O/Cu(100)
1.2 -
E
.E .
3
' 1.0 -
E, 0.5 I.
g ' Annealed 270 K
515 0.8 — X100
.5
'O ..
.‘é’
a 0.6 -
g a) (CHZOH)2 / Cu(100)
2 I
0.4 h
. 0.1 L
0.2 - , 85 K
. x100
0.0 ] L

 

 

 

'IIIIIIIIf'IIfIIIIIUrII

0 1 000 2000 3000 4000

Energy loss (cm'1)

Figure 5.4 Electron energy loss spectra for (a) 0.1 L exposure of ethylene glycol
(CH20H)2 on clean Cu(100) and (b) a 0.5 L exposure of (CH20H)2 on a Cu(100) pre-
covered with 0.1 ML of oxygen followed by annealing to 270 K.

107

the feature observed to desorb at 350 K after 15 minutes of UV irradiation of 1.1 ML on
CO/Cu(100), supporting the identification of this species as ethylene glycol.

The adsorption of (CH20H); on clean and O/Cu(100) was also studied using
EELS. Figure 5.4a shows the EEL spectrum of a 0.1 L exposure (CH20H)2 on clean
Cu(100) at 85 K. This exposure only populated the monolayer state as seen in Figure 5.3.
The losses observed match well with liquid-phase IR data25 for (CH20H); and that
observed for (CH20H); adsorbed on clean Ag(1 10).20 The loss observed at 2060 cm’1
was due to adsorbed CO. The results of Figure 5.4 are summarized in Table 5.2.
Noticeably absent from the monolayer spectrum was the loss due to vas(CO). This
indicates the (CH20H)2 was most likely bound to the surface via both oxygen atoms in a
bidentate configuration with the O-H bonds parallel with the surface and the C-0 bonds
perpendicular to the surface. This geometry would have the effect of rendering inactive
the vas(CO) mode as well as increasing the intensity of the mode due to r(OH). Similar
arguments have been made for the monolayer spectrum of (CH20H)2 adsorbed on clean
Ag(l 10).20

Annealing a 0.5 L dose of (CH20H)2 on a Cu(100) surface pre-covered with 0.1
ML of oxygen to 270 K resulted in the spectrum shown in Figure 5.4b. At this
temperature the multi- and monolayer states have desorbed and a fraction of the total
adsorbed (CH20H)2 has been deprotonated through reaction with the oxygen. X-ray
photoelectron spectroscopic analysis indicates a coverage of 0.1 ML of (CHZOH)2
remaining. The losses due to “C(OH) and v(OH) at 700 and 3320 cm", respectively were

absent in Figure 5.4b confirming deprotonation via reaction with the adsorbed oxygen.

108

 

"T

 

33588 5: n E

M 05m @2355 £53 :5on G
36on oExoxEQ 3

x53 30on Am

 

 

8mm 88 gm 85:
28$ 23% 283 288m 38 58...;
288 28% 28% 28% EN So?
28 88 omom 88 28 amp;
2: _ 82 _ ow: 83; am: E38
3: 85 m8: 255%
88 88 mm: 35?
NE 35:
sou: 2882 82: 82: 882 28. 52 So?
222 :82: mm: 68.;
28$ 28wa 28mm 28% eww 353
com 82 zoom 28% 283 .88 $8 65:
o: o: 88 £0:
8% ME. 6588
88m 65:
new

0.5.582 .8355 0.5.552 .8355 8:535 .835 32%:

83555556 +2558 A8355555£ +3523 :25qu 55:3 35:3 22:57?

 

AGO—EEO co NAIONEUV ES 2: Cw<\O co NAIONEUV 5% 83
mqmm 5:3 mac? 2:083an 2:5: .8 83 M: 8m :32? 82 .8230 58833-00 co OUNI 50 $3025 42 _._ a be M Ohm 8
321353 new 253335 >3 883:: m. wage—.8 33035 86on 2t 88 ATEo a: 3:3 3:285; 05 no mason—383‘ N.m 2an

109

The losses remaining at 1090 and 2880 cm'1 can be assigned to vs(CO) and v(CH) modes,
respectively, of the adsorbed alkoxide. The loss at 880 cm'1 was comprised of two
modes, v(CC) and p(CH2). This assignment matches well with that for the alkoxide on
O/Ag(110) where the corresponding modes were observed at 890, 1090 and 2860 cm",
respectively.20 In contrast to the work on O/Ag(110), the losses due to w(CH2) (1340
cm") and 8(HCH) (1450 cm") were not positively identified on the O/Cu(100) surface
although there appears to be a weak, broad loss in that range. This is probably due to the
lower surface coverage used in the present work, which resulted in a lower S/N. Again,
the absence of the vas(CO) mode indicates the alkoxide species was bound to the surface
through the oxygen atoms in an upright configuration.

The results of Figure 5.4b can be compared with the EEL spectrum obtained for
the species observed to desorb at 350 K after 15 minutes of UV irradiation of 1.1 ML
H2CO on CO/Cu(100). Figure 5.5a shows the spectrum of the species formed upon
irradiation and is identical to that shown in Figure 5.2c. Figure 5.5b shows the EEL
spectrum of the alkoxide species formed through the adsorption of (CH20H)2 on
O/Cu(100). Excellent agreement was observed between the spectra. The losses due to
v(CC)+p(CH2) and vs(CO) were observed at 880 and 1090 cm", respectively for the
adsorbed alkoxide and at 890 and 1070 cm‘1 for the species produced during the
irradiation as shown in Figure 5.5a. These spectra confirm the species observed in Figure
5.5a as ethylene glycol.

The same experiments shown in Figure 5.5 were repeated with the deuterated
species: DzCO on CO/Cu(100) and (CD20H)2 on O/Cu(100). The spectra obtained are

shown in Figure 5 .6. For the irradiation and subsequent annealing of D2CO on

110

 

 

 

 

 

 

 

_ b) 0.5 L (CH20H)2 + 0.1 ML O/Cu(100)
O
' 8
E
S ’ Anneal 270 K
g " o 8
7.; 8 a
2 ' x50
'93 .
.E
8 ..
g a) 1.1 ML HZCO + 0.57 ML CO/Cu(100)
m n
E E
O o
z — i "
8
- 0° 15 min. irr.
Anneal 270 K
- o
8
. L ‘ . “‘ x600
x20
l""lfi"fil""lrfi
0 1000 2000 3000

Energy loss (cm")

Figure 5.5 Electron energy loss spectra for (a) 1.1 ML coverage of H2CO on C0-
saturated Cu(100) irradiated for 15 minutes followed by an anneal to 270 K and (b) a 0.5
L exposure of (CH20H)2 on a Cu(100) pre-covered with 0.1 ML of oxygen followed by
an anneal to 270 K.

111

 

 

 

 

 

 

 

 

 

b) 0.5 L (CDZOH)2 + 0.1 ML O/Cu(100)
' 8
$2
— O
8 8 g
.2 2 a
S - 8 Annea|270K
Q l\
3 — 8
g 5
‘3 ' x200
£3
.E
8 ..
g _ a) 1.1 ML 02cc + 0.57 ML CO/Cu(100)
E 2
o 0
Z - 2"
°’ 8
. 8 :2
°° 15 min. irr.
_ E Anneal27OK
O
N
. L x600
x20
I""I"'rI'*'FI"
0 1000 2000 3000

Energy loss (cm")

Figure 5.6 Electron energy loss spectra for (a) 1.1 ML coverage of DzCO on C0-

saturated Cu(100) irradiated for 15 rninut
L exposure of (CD20H)2 on a Cu(100) p
an anneal to 270 K.

es followed by an anneal to 270 K and (b) a 0.5
re-covered with 0.1 ML of oxygen followed by

112

CO/Cu(100), as shown in Figure 5.6a, losses were observed at 800, 970, 1070, and 1180
cm”1 that can be assigned, through comparison with Figure 5.6b, to v(CC), w(CH2),
vs,a,(CO) and 5(HCH), respectively. Excellent agreement was again observed between
the spectra. The results of Figure 5.5 and 5.6 are summarized in Table 5.2 along with
liquid IR data of (CH20H)2 and (CH20H)2 adsorbed on O/Ag(110).

From the above TPD and EELS data, the results of the adsorption and
photochemistry of H2CO adsorbed on CO/Cu(100) can be now summarized.
Formaldehyde physisorbed on the CO-saturated surface at 85 K desorbs molecularly at
104 K. Upon UV irradiation, POM was formed that was observed to depolymerize, and
desorb as H2CO, between 220 and 250 K. A second photoproduct was observed to
desorb at 350 K that was identified as ethylene glycol. The formation of both POM and
(CHZOH)2 will be discussed below.

Formation of POM. The polymerization of H2CO could proceed via three
photoexcitation mechanisms. A thermal mechanism is discounted based on the
observation of an insignificant 2 K temperature rise upon irradiation. Gaseous H2CO has
a S1 (— 80 (1t* 6— n) band origin at 3.49 eV26 which is unlikely to be perturbed upon
physisorption on CO/Cu(100). Ultraviolet irradiation, in the wavelength range used here,

can dissociate H2CO via two channels: H2CO —> H2 + CO (molecular) and H2CO —> H +

HCO (radical). The molecular channel has its threshold at 3.52 eV.26 However,
polymerization is not expected to be initiated by the molecular photoproducts and no H2,

which is not known to adsorb on Cu(100) at 85 K, was observed to evolve upon

irradiation. The radical channel, whose threshold is at 3.73 eV,26 may initiate the

polymerization through reaction of the radical species with molecular H2CO. However,

113

 

for experiments performed on amorphous H2CO films deposited on C5], no
polymerization occurred upon irradiation with 4.0 eV photons.27 For both channels,
dissociation is expected to compete with electronic relaxation which is known to occur on
metal surfaces on a time scale of ~ 10'13 — 10'[4 5.23 The effect of the CO spacer layer
will be to decrease the relaxation rate.

The most probable excitation mechanism leading to POM is that proposed for
H2CO photopolymerization on Ag(111).10’29v3O In this mechanism, absorption of the UV
irradiation by the substrate produces hot electrons, which then populate the 1t* state of
H2CO, forming H2CO'. This radical anion then dissociates into products that can initiate
the polymerization. In the gas-phase, H2CO' dissociates into H’ + HCO31 while on
Ag(l 1 1), H2CO” was observed to dissociate into CH2 and 0.29 The dissociation products
on CO/Cu(100) are not yet known. In contrast to the work on Ag(111), where
subvacuum electrons were produced upon irradiation, the photon energies used in the
current work will produce both subvacuum and free photoelectrons. The work function
of a CO—saturated Cu(100) surface was found to be 4.39 eV32 and the majority of
intensity from the lamp used in this work (~ 80 %) occurs between 5.5 and 4.9 eV,
indicating most of the electrons produced will be free, low energy photoelectrons.

Formation of Ethylene Glycol. The formation of ethylene glycol can be explained
based on the well known photochemistry of carbonyl compounds.”34 These molecules
are known to be excellent hydrogen atom abstractors in their singlet 1(n1t"‘) and triplet
3(IlTl:"‘) electronically excited states, with the majority of chemistry occurring on the triplet
surface. Formaldehyde, in its triplet state, can be thought of a biradical which abstracts a

Proton from another molecule to form a hydroxymethylene radical 'CHZOH.

114

 

F
I

I.

For the UV irradiation of H2CO on CO/Cu(100), excited formaldehyde, CHzOa
abstracts a proton from a neighboring H2CO forming 'CHZOH and 'HCO. The formyl
radical can then initiate the polymerization of H2CO to form POM. The
hydroxymethylene radical can react in two ways: 'CHzOH could couple to another
'CHZOH, forming ethylene glycol (CHZOH)2 directly or it could react with another CHzO
to form 'OCH2CHZOH. This radical species could then abstract a hydrogen from another
H2CO to form (CH20H)2 and another formyl radical.

The saturation of the formation of ethylene glycol after 15 minutes of irradiation
may be due to geometric constraints. Ab initio results on hydrogen abstraction from
methane by formaldehyde indicates only approach of the hydrogen within the molecular
plane of H2CO results in reaction and an upper limit on the barrier was estimated to be ~

77 k.I/mol.35»36 This barrier was strongly dependent on Cuzco-Corr, separation with the

distance equal to 2.53 A at the transition state.35 Once ethylene glycol is formed and
POM polymerization initiated by the formyl radicals, the remaining electronically excited
H2CO may be in an unfavorable geometric position within the overlayer to abstract
further hydrogens from either H2CO, POM or ethylene glycol. Unfortunately, the
structure of the physisorbed H2CO on CO/Cu(100) is unknown at this time. However,
the structure of crystalline formaldehyde does have pairs of H2CO molecules that are
oriented correctly for H-atom abstraction to occur.37

The formation of ethylene glycol through the reaction of electronically excited
formaldehyde has not been observed in studies of solid phase H2CO”,38 nor in the

photopolymerization of H2CO on Ag(111).10 The presence of the CO spacer layer could

influence the photochemistry by lengthening the lifetime of the excited formaldehyde

115

 

In";

'Jo
.
a),

(H2CO‘) by moving it away from the metallic surface or through the formation of a

unique formaldehyde overlayer geometry that is favorable to the hydrogen abstraction

reaction.

5.3 Conclusions
The presence of a saturation coverage of CO on Cu(100) inhibited the thermal
polymerization of H2CO which formed a weakly adsorbed overlayer on CO/Cu(100).
Upon 30 minutes of UV irradiation, H2CO formed POM (4O %) and ethylene glycol (20
%). The photopolymerized POM depolymerized at higher temperatures than those
observed for POM formed from the thermal polymerization on clean Cu(100), suggesting
a longer chain length or more thermally stable endgroup. Additionally, no losses due to
endgroups were observed in EEL spectra taken after irradiation. The initiation of
polymerization likely occurs via the radical anion H2CO' formed from electron capture of
photoelectrons generated from irradiation of the CO/Cu(100) surface. The use of CO to
block the thermal polymerization of H2CO on Cu(100) and subsequent
Photopolymerization shows the potential for controlling polymer thin film formation for
1‘ eactive monomer-substrate combinations by use of a co-adsorbate. The polymer film
can be formed in a “controlled” fashion using photons and can then be deposited on the
SUbstrate after annealing to remove the co-adsorbate.
The formation of ethylene glycol from the irradiation of H2CO adsorbed on
CO/CuUOO) has not been observed previously in studies of solid phase H2CO nor in the
photopolymerization of H2CO on surfaces. The formation likely occurs via electronically

e"(Cited H2CO which abstracts a proton from a neighboring H2CO molecule forming a

116

hydroxymethylene radical °CHzOH which then reacts with other °CHzOH radicals or
H2CO ultimately forming (CHZOH)2. The saturation of formation of ethylene glycol

likely results from geometric constraints present within the overlayer.

5.4 References
(l) Stuve, E.; Madix, R.; Sexton, B. Surf. Sci. 1982, 119, 279.

(2) Richter, L.; Ho, W. J. Chem. Phys. 1985, 83, 2165.

 

( 3) Henderson, M.; Mitchell, G.; White, J. Surf Sci. 1987, I88, 206.

(4) Davis, J.; Barteau, M. J. Am. Chem. Soc. 1989, 111, 1782.

 

(S) Houtman, C.; Barteau, M. Surf. Sci. 1991, 248, 57.

(6) Davis, J.; Barteau, M. Surf. Sci. 1992, 268, 11.

(7) Truong, C.; Wu, M.; Goodman, D. W. J. Am. Chem. Soc. 1993, 115, 3647.
(8) Sexton, 3.; Hughes, A.; Avery, N. Surf. Sci. 1985, 155, 366.

( 9) Bryden, T.; Garrett, S. J. Phys. Chem. B. 1999, 103, 10481.

( l O) Fleck, L.; Feehery, W.; Plummer, E.; Ying, Z.; Dai, H. J. Phys. Chem. 1991, 95,
8428.

( 1 1 ) Fleck, L.; Kim, J.; Dai, H.-L. Surf. Sci. 1996, 356, L417.

( l 2) Bryden, T.; Garrett, S. Langmuir , in press.

( l 3 ) Sasaki, T.; Kawada, F.; Aruga, T.; Iwasawa, Y. Surf. Sci. 1992, 278, 291.
( 1 4) Rauscher, H.; Menzel, D. Surf. Sci. 1995, 342, 155.
( l 5) Sasaki, T.; Aruga, T.; Kuroda, H.; Iwasawa, Y. Surf. Sci. 1992, 276, 69.

(1 6) Orita, H.; Kondoh, H.; Nozoye, H. Chem. Phys. Lett. 1994, 228, 385.

117

"L

'JJ

 

'J.)

(17) NIST Chemistry WebBook, NIST Standard Reference Database Number 69;
Mallard, W., Linstrom, P., Eds.; National Institutes of Standards and Technology:
Gaithersburg, MD, 1998; pp (http://webbook.nist.gov).
(18) Dai, Q.; Gellman, A. J. Phys. Chem. 1993, 97, 10783.
( 19) Bowker, M.; Madix, R. Surf. Sci. 1982, 116, 549.

(20) Capote, A.; Madix, R. J. Am. Chem. Soc. 1989, 111, 3570.

(21) Khoshkoo, H.; Hemple, S.; Nixon, E. Spectrochim. Acta A 1974, 30, 863.

(22) Ibach, H.; Mills, D. L. Electron Energy Loss Spectroscopy and Surface

Vibrations; Academic Press: San Diego, 1982.

 

(23) Tadokoro, H.; Kobayashi, M.; Kawaguchi, Y.; Koybayashi, A.; Murahashi, S. J.
Chem. Phys. 1963, 38, 703.

(24) Kobayashi, M.; Iwamoto, R.; Tadokoro, H. J. Chem. Phys. 1966, 44, 922.
(25) Sawodny, W.; Niedenzu, K.; Dawson, J. Spectrochim. Acta A 1967, 23A, 799.
(26) Moore, C.; Weisshar, J. Ann. Rev. Phys. Chem. 1983, 34, 525.

(27) Mansueto, E.; Ju, C.; Wight, C. J. Phys. Chem. 1989, 93, 2143-2147.

(28) Zhou, X.; Zhu, X.; White, J. Surf. Sci. Rep. 1991, I3, 73.

(29) Fleck, L.; Howe, P.; Kim, J .; Dai, H. J. Phys. Chem. 1996, 100, 8011.
(30) Fleck, L. E. Ph.D. Thesis, University of Pennsylvania, 1994.

(3 1 ) Azria, R. Ph.D. Thesis, University of Orsay, 1972.
(32) Dubois, L.; Zegarski, B. Chem. Phys. Lett. 1985, 120, 537.

(33) Turro, N. In Molecular Photochemistry; W. A. Benjamin, Inc.: New York, 1965;

PD 1 37,

118

(34) Klessinger, M.; Michl, J. In Excited States and Photochemistry of Organic
Molecules; VCH Publishers: New York, 1995; pp 395.

(35) Severance, D.; Pandey, B.; Morrison, H. J. Am. Chem. Soc. 1987, 109, 3231.

(36) Sumathi, K.; Chandra, A. J. Photochem. Photobiol. A:Chem. 1988, 43, 313.

(37) Weng, S.; Torrie, B.; Powell, B. Mol. Phys. 1989, 68, 25.

(38) Gol’danskii, V. Ann. Rev. Phys. Chem. 1976, 27, 85.

119

 

p33}:

tfllfl

”1".“
5L; at

let‘s
ml:

iii:-

a:—
75’

100

011 it

Chapter 6 Conclusions and Future Work

6.1 Thermal Reactions and Control of Stability
The work described in Chapter 3 has shown that at 85 K, H2CO spontaneously
polymerizes to form a monolayer of poly(oxymethylene) (POM) up to a saturation
coverage of 0.69 ML (1.06x1015 cm'z). This surface density suggests that the POM chain
directions are parallel to the surface plane and is consistent with the known crystal
structures of POM.1 However, the POM overlayer is probably not highly ordered.
Temperature-programmed desorption spectra indicate there is only one
decomposition route available to the adsorbed polymer species: depolymerization to
molecular H2CO. The route producing H2CO is observed as two desorption features at
approximately 200 and 215 K which show apparent zero- and first-order
depolymerization kinetics, respectively. This behavior has neither been observed for
H2CO adsorption Cu(110)2 nor for the POM produced in the photopolymerization of
H2CO on Ag(111).3'6 We believe differences due to adsorption at defect and/or step
edges do not account for the appearance of two depolymerization features. The
Observation of CO and H2 desorption indicates H2CO is probably dissociating at step
edgfits and defect sites upon adsorption. Electron energy loss spectroscopy data indicate
"0 Other species besides an adsorbed polymer first layer and molecular H2CO multilayers
on top of the polymer layer, are present on the 85 K Cu(100) surface.
The observation of two features for the depolymerization of POM on Cu(100) can
be aSCribed to the formation of long chain (on) and short chain ([3) poly(oxymethylene)

(POM) species as described in Chapter 4. Equations describing the kinetics of

120

f-..

F b 6 1 I) s 1.. n. I
a.“ E . W. .E )1 “is

5P

depolymerization suggest the number of chains of the CL species is constant for increasing
coverage and determined at low coverages by a fixed number of surface initiation sites.
Future experiments should include increasing the number of defect sites, which may
increase the proportion of a—POM species produced upon adsorption of H2CO if initiated
at these sites. Chapter 4 has demonstrated that equations describing the bulk
depolymerization process can be used to model the TPD spectra which should be
applicable to future surface polymerization studies where multiple depolymerization

processes are observed

Upon removal of the (it species by annealing, losses are observed attributable to

 

the endgroups of the helical B species. Losses can be assigned to v(Cu-O), v(CO) and
p(CH3) of the end of the POM chain bound directly to the surface through an oxygen
atom and to a methoxy endgroup. This species can be written as -Oa-(CHzO)n-CH3.
Compared to bulk POM, the lower thermal stability observed for monolayer B-POM on
Cu( 100) most likely results from the chain being bound to the surface through the oxygen
atom, resulting in a lower barrier to initiation of depolymerization. We believe, due to
Similarities in depolymerization temperature and product (molecular H2CO), the a—POM
SPeCies likely has similar endgoups to B-POM.

Pre-adsorbed methanol was found to terminate the polymerization and favor the
fol‘Ination of the B-POM species. The thermal stability and EEL spectra of the POM
ovePlayer formed upon reaction with CH3OH were similar to that of POM formed
throllgh the direct adsorption of H2CO. However, the losses due to the endgroups were

mot e intense than in the POM layer formed by H2CO adsorption alone, indicating the B

s - . .
p ecles formed were shorter than those formed through the direct adsorption process.

121

The experiments with the co-adsorbed methanol and DzCO also lend support to
the directly bound oxygen controlling the thermal stability of the POM species. For bulk
POM, methoxy termination is known to be more stable than the hydroxy endgroup.7
This indicates the oxygen bound end is likely responsible for the ~ 100 K lower stability
observed for POM on Cu( 100) as compared to bulk POM.

The ability to preferentially form a specific chain length polymer with a specific
endgroup may open the possibility to control both thin film order and morphology and
tailor thermal stability based on the interaction of the endgroup with the surface.

Transforming the oxygen bound end into a methoxy terminus should increase the thermal

 

stability dramatically. Poly(oxymethylene) dimethyl ethers are known to be stable to
temperatures > 470 K, some 100 K higher than the dihydroxy terminated
paraformaldehyde.7 A potentially simple way to form a methoxy endgroup from the
surface bound oxygen is through the reaction with methyl radicals (CH3°). The methyl
radicals can be generated photochemically from co-adsorbed CH3Br8 or dosed from the
gas phase as CH3', produced from the pyrolysis of azomethane.9 Methyl radicals are
known to be highly mobile, even at 85 K, on Cu(111)10 and thus could react with the
OXYgen bound to the surface forming a methoxy endgroup. However, the CH3' may also
deStroy the POM backbone chain through hydrogen abstraction reactions. As in the
methanol experiments, the order of adsorption may influence the products greatly.

Acetoxy—terminated bulk POM is also known to be more thermally stable than the
hydr(My-terminated form.7 On the Cu(100) surface, acetoxy endgroups may be formed
by Dre-adsorbing acetic acid prior to adsorption of H2CO. In a similar fashion to the

forlhation of methoxy endgroups with pre-adsorbed methanol, described in Chapter 4,

122

on C
iii}

1
"P
131:...
L

acetoxy endgroups would thus be formed. In this case, as acetic acid is more acidic than
methanol, the initiation and termination mechanisms for polymerization may be different
and both endgroups may now be acetoxy. Further experiments with co-adsorbates may

allow for a full elucidation of the propagation and termination events during the surface

polymerization process.

6.2 Photochemical Reactions
UV and X-ray Photons. The presence of a saturation coverage of CO on Cu(100)

inhibited the thermal polymerization of H2CO which formed a weakly adsorbed overlayer

 

on CO/Cu(100). Upon irradiation for 30 minutes with UV photons, H2CO formed POM
(40 %) and ethylene glycol (20 %). The photopolymerized POM depolymerized at
higher temperatures than those observed for POM formed from the thermal
polymerization on clean Cu(100). Additionally, no losses due to endgroups were
observed in EEL spectra taken after irradiation, consistent with longer chains. The
initiation of polymerization likely occurs via the radical anion H2CO" formed from
electron capture of photoelectrons generated from irradiation of the CO/Cu(100) surface.
X-ray irradiation also caused polymerization of H2CO adsorbed on CO/Cu(lOO),
Izlll‘ther supporting a substrate mediated process. Figure 6.1 shows TPD spectra
displaying the effect of irradiating 1.1 ML of H2CO on CO/Cu(100) with X-ray photons
(hv=1486.6 eV, re=os nm). Figure 6.1a shows the m=29 (HCO+) TPD prior to
irradiation and is identical to Figure 5.1a-b (0 min.). A single feature at 104 K is

observed corresponding to molecularly adsorbed H2CO. Figure 6.1b shows irradiation of

the oVerlayer for 84 s

123

 

1.1' ML HzCO / CO-sat / Cu(100)

f I c)1303

 

 

" , b)84s

 

 

 

HCO+ (m=29) ion intensity (arb. units)

 

 

 

 
 

 

 

 

. , . . ,
1 00 200 300 400
Temperature (K)

Figul‘e 6.1 Mass=29 (HCO+) TPD spectra showing the effect of X- -ray irradiation on 1.1
I--I‘12CO on CO/Cu(100). Irradiation times were a) 0, b) 84 and c) 130 seconds. The

5933)’ wavelength was the Al KOt (hv=1486. 6 eV) line and operated at 300 W (20 mA,15

124

 

SEC

the

is?

10

causes the feature at 104 K to decrease while a new feature is observed at 228 K that can
be identified as H2CO arising from the depolymerization of POM formed upon
irradiation. Further irradiation causes a further decrease in the feature at 104 K, an
increase in the H2CO due to POM depolymerization and a shift to higher peak
depolymerization maximum (233 K), as shown in Figure 6.1c. The large number of
secondary electrons produced from the substrate upon X-ray irradiationll likely initiates
the polymerization via the radical anion channel, although more work is necessary to
elucidate the mechanisms of polymerization. The ability to form a polymer thin film
using X-ray photons is advantageous in the field of lithography where feature sizes of <
l ()0 nm are needed.”14

New Co-adsorbates. While the use of CO is convenient, its desorption
temperature (~ 180 K) prevents a determination of the “true” thermal stability of the
POM formed in the irradiation of H2CO adsorbed on CO/Cu(100) because, after the CO
has desorbed, the POM then can interact with the substrate, lowering the
depolymerization temperature as was discussed in Chapter 4. A more suitable co-
adsorbate would be one that is stable on Cu(100) above the bulk POM depolymerization
temperature of ~ 370 K. An excellent candidate for this is CH3O(ad) formed from the
reaction of methanol on a oxygen covered Cu(100) surface. This species forms a c(2x2)

overlayer (GCH30=O.5) similar to the CO/Cu(100) overlayer and is stable up to 400 K at

Which temperature B-hydride elimination occurs and H2CO desorbs”,16 In this fashion,
the stability of the POM may be deconvoluted from its interaction with the substrate. The

C1‘130 covered Cu(100) surface may also be prepared from methyl nitrite (CH3ONO),

e1ilhinating the need for the oxygen precoverage.17

125

 

 

. 1.1 ML H2CO/Cu(100)

-

b) 15 min. UV irradiation

GB

 

 

 

 

a) 0 min. UV irradiation

H200+ (m=30) Ion mtensnty (arb units)
. >1 T l ' l ' 1

I.

i

i

 

 

 

 

 

 

 

I I r I I I I I

250 300

 

. . . . ,
150 200
Temperature (K)

Figure 6.2 Mass=30 (H2CO+) TPD spectra for the UV irradiation of 1.1 ML H2CO
adsorbed on a surface that had been pre-covered with a saturation of poly(oxymethylene)
(0.69 ML) for a) 0 minutes and b) 15 nrinutes UV irradiation.

126

Preliminary work has been done using another co—adsorbate system.
Formaldehyde molecularly adsorbed on a POM saturated Cu(100) surface does
polymerize upon UV irradiation. Figure 6.2 displays TPD spectra (a) prior to and (b)
after 15 minutes of UV irradiation of a 1.1 ML coverage of molecularly adsorbed H2CO
on a POM saturated Cu(100) surface. There are three features observed prior to
irradiation which can ascribed to desorption of physisorbed H2CO (105 K) and H2CO
arising from depolymerization of the or and B-POM species at ~ 210 and ~ 215 K,
respectively. Fifteen minutes of UV irradiation causes both the physisorbed H2CO and
oc-POM species to decrease and a new feature to appear at 234 K. The feature at 234 K is
consistent with a new POM species and is similar to that observed for the
photopolymerization of H2CO on CO/Cu(100), shown in Figure 5. lb, where a feature is
observed at 235 K. The explanation for the decrease in (it-POM species, while little
Change in the B-POM species was observed, is still under investigation. However, the
results shown in Figure 6.2b do suggest other co-adsorbate systems can be used to

produce a more thermally stable POM thin film.

6.3 References

( 1) Tadokoro, H. In Macromolecular Reviews; Peterlin, A., Goodman, M., Okamura,
3., Zimm, B., Mark, H., Eds.; Interscience: New York, 1967; Vol. 1; pp 119.

(2) Sexton, B.; Hughes, A.; Avery, N. Surf. Sci. 1985, 155, 366.

(3) Fleck, L.; Feehery, W.; Plummer, 13.; Ying, 2.; Dai, H. J. Phys. Chem. 1991, 95,
8428.

(4) Fleck, L.; Ying, z.; Dai, H.-L. J. Vac. Sci. Technol. A 1993, 11, 1942.

127

 

(5) Fleck, L.; Kim, J .; Dai, H.-L. Surf. Sci. 1996, 356, L417.
(6) Fleck, L.; Howe, P.; Kim, J .; Dai, H. J. Phys. Chem. 1996, 100, 8011.
(7) Walker, J. F. Formaldehyde, 3rd. ed.; Reinhold Publishing Corp.: New York,
1964.
(8) Lamont, C.; Conrad, H.; Bradshaw, A. Surf. Sci. 1993, 280, 79.
(9) Bent, B. E. Chem. Rev. 1996, 96, 1361.
(10) Chan, Y.; Chuang, P.; Chuang, T. J. Vac. Sci. Technol. A 1998, 16, 1023.
(l 1) Briggs, D.; Riviere, J. In Practical Surface Analysis; Briggs, D., Seah, M., Eds.;
Wiley and Sons: New York, 1990; Vol. l-Auger and X-ray Photoelectron Spectroscopy;
pp85
(12) Lawes, R. Appl. Surf. Sci. 2000, 154-155, 519.
(13) Wallraff, G.; Hinsberg, W. Chem. Rev. 1999, 99, 1801.
(14) Cerrina, F. J. Phys. D: Appl. Phys. 2000, 33, R103.
(1 5) Dai, Q.; Gellman, A. J. Phys. Chem. 1993, 97, 10783.
(16) Andersson, S.; Persson, M. Phys. Rev. B. 1981, 24, 3659.

(17) Ihm, H.; Scheer, K.; Celio, H.; White, J. Langmuir 2001, I 7, 786.

128

 

1"“

 

APPENDICES

129

 

 

"'"A

D?

Ax“

 

Appendix A Mass Spectra

Prior to use in experiments, the molecules studied in this thesis were first
analyzed for purity. This was accomplished by opening the molecular leak valve to
introduce some of the gas-phase molecules into the UHV chamber. The mass spectrum
was acquired after a steady state pressure (as measured by the ion gauge) was achieved
within the chamber. Typical operating condition of the mass spectrometer were 1-50

amu mass range, 10 amu/s scan rate, and 1.7 kV electron multiplier voltage.

130

 

3.0

 

h

2.5 _ b) H200
- -9

2-0 '- Pchamber ~ 1X10 TOIT
~i

1.5 —

 

1.0 1

 

 

 

 

 

 

 

 

 

 

 

a) Background

Partial Pressure (Torr) x10'10
(0 C

 

 

 

 

 

2.5 _ .10
_ Pcmmr ~ 2x10 Torr
2.0 -
1.5 —
1.0 -
“L h. .1, . .fi .
0 1O 20 30 40 50

m/z (amu)

Figure A.1 Mass spectra of a) background vacuum prior to introduction of H2CO and b)
after a pressure of ~ 1x10'9 Torr of H2CO was achieved within the chamber.

131

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1.4

1.2 -

1.0 -

b) D200
-9

o 0.8 — P chambep 1x10 Torr
E 0.6 -
X
5 0.4 -
t W
s 02 J.
[fl/11.4%
931.4
a.
%1-2 - a) Background
0- 1,0 _ Pchamber ~ 2x10'1o Torr

0.8 -

0.6 -

0.4

0.2

I I I I jfi I I I I I I f I I I I I I I I IiI r
0 10 20 30 40 50

m/z (amu)

 

Figure A.2 Mass spectra of a) background vacuum prior to introduction of DzCO and b)
after a pressure of ~ 1x10'9 Torr of D2CO was achieved within the chamber.

132

100

 

80 -
_ b) Trioxane
-9
60 - me3r ~ 5x10 Torr

 

 

8

N
O

 

 

   

IL

IIIIIIII'IIII'IIIIIIIII'IIII[III

 

 

Partial Pressure (Torr) x10'9

 

 

 

 

 

 

 

 

1o
8 a) Background
_ ~10

. Pchamw ~ 2x10 Torr
61-1-
4.
2 .

JUM'W WWW

IIIIIIIII'ITII'IIIIIIIII'IIIll—rIIIITTTIIIllI'lIII

O 10 20 30 40 50 60 7O 80 90 100
m/z(amu)

Figure A.3 Mass spectra of a) background vacuum prior to introduction of trioxane
(C3H603) and b) after a pressure of ~ 5x 10'9 Torr of trioxane was achreved wrthrn the
chamber.

133

1.0

0.8

0.6

.-‘ .° 9
o N 1:

Partial pressure (Torr) x10'10
9
CD

0.6

0.4

0.2

 

b) Ethylene Glycol

 

 

 

 

 

 

 

 

 

 

 

 

 

b

' -9
. Pcmber ~ 1x10 Torr
I I I r I r I r I l I I I I ‘r I I r I I I fi I I
" a) Background
P -10
POWnber ~ 2x10 Torr

 

 

 

 

 

 

r I I I I I I I I I I I I I I l I I I

20 30 4O
m/z (amu)

Figure A.4 Mass spectra of a) background vacuum prior to introduction of ethylene
glycol ((CH20H)2) and b) after a pressure of ~ 1x10'9 Torr of ethylene glycol was
achieved within the chamber.

134

 

 

 

 

Appendix B Calculation of Electron Impact Ionization Cross-Sections

To calculate the purity of a compound accurately, based on the mass spectra
presented in Appendix A, and quantitate surface species from TPD spectra, the electron
impact (EI) ionization cross-section (GE!) needs to be known. While the experimental 0'51
is known for a few of the compounds of interest (H2O, H2 and CO),1v2 the 0’51 for H2CO
is unknown to this author. However, binary-encounter—Bethe (BEB) theory has been
shown to predict (within 10-15%) the CE] for small molecules using molecular orbital
constants easily obtained from standard ab initio electronic structure programs such as
GAMESSJ'3

The BBB theory combines the Mott description for hard electron-electron
collisions with that of the dipole interaction theory that accounts for electron interaction
at high incident electron energies. The theory provide a simple, analytic expression for
the 0'51 per molecular orbital (MO) and the total om is then a sum over all MOs. Three
orbital constants are needed: the binding energy B, the orbital kinetic energy U and the

electron occupation number N. The expression, as a function of incident electron energy

 

T,is:
GE]: 5 1n(t)1——13-+1—-1———l“(t) (3.1)
t+u+1 2 t 1 11+]
where
T
t=—- B.2
B ( )
U
u=— B.3
B ( )

135

 

 

 

 

_ 47ra(2,NR2

S B2

(B4)

with a0=0.5292 A and R=13.61 ev.

The orbital constants were calculated at the minimized geometry at the Hartree-
Fock level with a 6-31G* basis set using the GAMESS4 code. The orbital constants for
H20 and H2CO are summarized in Table BI and the 051 as a function of T is shown in
Figure B.1. The GE. and orbital constants for H2O have been previously calculated and
are compared to those calculated for this work as a measure of the error. The variation
arises due to differences in basis set used (6—311G vs. 6-31G*) and the use of the
experimental ionization energy for the HOMO of water in the previous work.2 All
quantitation is done using calculated 0'51 to eliminate large discrepancies that may arise

between experimental and calculated values.

136

 

 

 

 

Table B.1 Molecular orbital constants calculated at the Hartree-Fock level of theory with
a 6-31G* basis set. These constants were used to evaluate the electron impact ionization

cross-section.

 

 

 

 

 

 

H2O“
Molecular Orbital Binding Energy (eV) Kinetic Energy (eV)
2a] 36.46 70.14
1b2 19.12 47.78
3a] 15.47 58.73
1b. 13.52 61.89
H20b
Molecular Orbital Binding Energy (eV) Kinetic Energy (eV)
2a. 36.88 70.71
1b2 19.83 48.36
3a] 15.57 59.52
1b] 12.61 61.91
H2CO
Molecular Orbital Binding Energy (eV) Kinetic Energy (eV)
Bay 38.56 73.68
4a1 23.59 44.02
1b2 18.95 36.77
5a. 17.74 61.17
1b; 14.69 47.46
2b2 11.83 54.58
a) this work
b) Reference 2

137

 

 

 

 

 

 

 

 

 

4
. III-III...
* I. 'II
I III...
1 I I.
. . "i
3- I
1 I
g . ‘gt‘
‘0 6
2’2L‘ I I
s . e
m A
b . '
A
.l I
- ‘c‘ I HZCO
1- A 0 H20 (this work)
' A H20 (reference 2)
:
fi..1.......rs,r...
50 100 150 200

Electron Energy, T (eV)

Figure B] Electron impact ionization cross-section (051) as a function of incident
electron energy for H2CO and H20 and a comparison to previous work done for water
showing the small error introduced by using a smaller basis set.

138

 

 

B.1 References

(l) Beran, J .; Kevan, L. J. Phys. Chem. 1969, 73, 3866.

(2) Hwang, W.; Kim, Y.; Rudd, M. J. Chem. Phys. 1996, 104, 2956.

(3) Kim, Y.; Rudd, M. Phys. Rev. A. 1994, 50, 3954.

(4) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.; Gordon, M. S.;
Jensen, J. H.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.; Su, 8. J.; Windus, T. L.;

Dupuis, M.; Montgomery, J. A. J. Comput. Chem. 1993, 14, 1347.

 

 

139

 

Appendix C Electron Energy Loss Spectrometer and Operating Voltages

Shown in Figure C1 is a schematic of the ELS3000 electron energy loss
spectrometer manufactured by LK Technologies, Inc. The operating voltages for the
individual segments are shown in Table C.1. These voltages were used as initial
parameters prior to tuning the spectrometer in the “straight through” geometry. The
resolution obtained using these voltages was 3.2 meV (26 cm") full width half maximum

(FWHM) and provided ~ 70 pA of current detected at the channeltron cone.

140

 

 

 

 

Table C.l Operating voltages for the ELS3000 electron energy loss spectrometer.

 

 

 

 

 

Setting Setting
Cathode Monochromator
Filament, V 3.22 Ml 2.639
Filament, A 1.90 M l -slit -0.31
Beam, E -6.09 Ml-cover -3.56
Emission Optics AMI-cover 0.16
R -2.7 1 M2 0.638
A1 46.22 M2-slit 0. 12
AA] -3.54 M2-cover -0. 16
A2 0.70 AM2-cover 0.03
AA2 l .60 M2-exit 0.49
A3 0.10
AA3 0.56
Lenses Analyzers
B1 0.78 AN 1 0.572
AB 1 0.04 AN l-slit 0.46
B2 3.05 AN 1 -cover -0.43
B3 2.98 AANl-cover 0.14
B4 0.67 AN 2 0.576
AB4 0.1 1 AN 2-slit 0.35
AN 2-cover -0.43
Miscelaneous AANZ-cover -0.01
Shield 0.0
Sample 0.003
Ch-slit 1.00
Ch-cone 0.5 1
Scan 0.001
Scan-step 3.41
Offset 0.001
B3 ramp 1.0
B4 ramp 1.0
OS. ramp 1.0

 

 

 

141

 

 

 

Electron
multiplier (Ch) Lenses (B l-B 2)

  
 
    

Monochromator

Pre-monochromator (M l) /

Emission optics
(R, Al-A3)

 

 

 

 

Figure C.1 Schematic of the ELS3000 electron energy loss spectrometer.

142

     

llllllllllllllllllllllllzlllll

31293 02177