An Investigation of Porphyrin Aggregation Using Spectroscopic

An Investigation of Porphyrin Aggregation Using

Spectroscopic and Microscopic Methods

BENJAMIN A. FRIESEN

A dissertation submitted in partial fulfillment of the requirements for the degree of

DOCTOR OF PHILOSOPHY

WASHINGTON STATE UNIVERSITY College of Sciences

MAY 2011

To the Faculty of Washington State University:

The members of the Committee appointed to examine the dissertation of

BENJAMIN A. FRIESEN find it satisfactory and recommend that it be accepted.

Ursula Mazur, Ph.D., Chair

Kerry Hipps, Ph.D.

James Satterlee, Ph.D.

ACKNOWLEDGMENT

I would like to express my gratitude to my advisor Dr. Ursula Mazur for the opportunity to work in her lab, for providing financial support, and her patient guidance through the completion of my doctoral degree. I would also like to convey my thanks to my committee members Dr. K.W. Hipps and Dr. James Satterlee and to Dr. Jeanne

McHale for a profitable collaboration.

I would also like to thank my wife Angela and my daughter Elizabeth. Their love and support provided much needed encouragement throughout this process.

iii

An Investigation of Porphyrin Aggregation Using

Spectroscopic and Microscopic Methods

Abstract

By Benjamin A. Friesen, Ph.D. Washington State University May 2011

Chair: Ursula Mazur

Aggregates of diacid tetrasulfonatophenylporphine (H2(H4TSPP)) exhibit light harvesting and electron transport capabilities and are therefore promising candidates as device components. Before these aggregates can be used to construct devices their structural and electronic properties must be understood. Solution UV-visible and RLS studies confirmed the formation of H2(H4TSPP) aggregates with increasing solution ionic strength. Aggregation of H2(H4TSPP) was indicated by the presence of new absorbance bands in UV-visible spectra and a marked increase in RLS concurrent with the appearance of the new UV-visible absorbance bands. Both UV-visible and Raman spectroscopy confirmed the intact deposition of H2(H4TSPP) aggregates on solid substrates.

The deposited aggregates were imaged by AFM and STM. The AFM and STM images revealed individual rods with diameters of ~ 30 nm and lengths of hundreds of

iv nanometers. We report for the first time high resolution STM images of H2(H4TSPP) on

Au(111) and HOPG. In addition to aggregates an ordered monolayer of H2(H4TSPP) monomers was found on HOPG. The well-defined monolayer islands of H2(H4TSPP) self- assembled on HOPG were studied in ultrahigh vacuum using STM, OMTS, UPS, and XPS. Unlike meso-tetrakis(4-carboxyphenyl)porphine (Hx(H4TCPP)), the carboxylate analog, H2(H4TSPP) monolayers are stable on HOPG and can be studied at room temperature without the addition of a second stabilizing compound. Protonation of the porphyrin nitrogens in the surface species is confirmed by XPS. High-resolution

STM images of single molecule layers show a well-defined deformation of the porphyrin ring, as expected with complete protonation of the central nitrogen atoms. OMTS and

UPS were used to identify the HOMO and LUMO of the H2(H4TSPP) monolayer species, and results are contrasted to those of nickel(II)tetraphenylporphyrin (NiTPP). Current vs. Voltage (I(V)) curves of single and stacked rods taken by STM are consistent with conduction in a band formed from the LUMO of H2(H4TSPP). Aggregate I(V) curves were consistent with N-type semiconductors and showed increasing current rectification with increasing aggregate thickness. These findings show that H2(H4TSPP) aggregates can be used as organic semiconductors with tunable current versus voltage characteristics.

Table of Contents

ACKNOWLEDGMENT...... iii

Abstract ...... iv

Table of Contents ...... vi

List of Figures ...... ix

List of Tables ...... xxvi

Glossary of Terms ...... xxviii

Chapter 1: Introduction ...... 1

1.1 Properties and Applications of Porphyrins...... 1

1.2 Literature Review of Porphyrin Aggregation...... 9

1.2.1 Non-Ionic Porphyrin Aggregates in Solution and at Surfaces ...... 10

1.2.2 Ionic Porphyrin Aggregates in Solution and on Surfaces ...... 11

1.2.3 Aggregates of Tetrasulfonatophenyporphine: ...... 14

1.3 The Electronic Structure of Porphyrins and Changes upon Aggregation: ...... 29

1.3.1 The Electronic Structure of Porphyrins ...... 29

1.3.2 Exciton Theory of Dimer and Aggregate Formation: ...... 35

Chapter 2: Experimental Techniques ...... 42

2.1 UV-visible and Resonance Light Scattering Spectroscopy...... 42

2.2 X-ray and Ultraviolet Photoelectron Spectroscopy ...... 48

2.3 Scanning Tunneling Microscopy (STM)...... 52

2.4 Atomic Force Microscopy (AFM) ...... 63

2.5 Raman Spectroscopy: ...... 65

2.6 Helium Ion Microscopy: ...... 71

2.7 Transmission Electron Microscopy: ...... 73

Chapter 3: Experimental Methods ...... 75

3.1 Materials, Reagent, and Instrument List ...... 75

3.2 Glassware Cleaning Procedure...... 78

3.3 Preparation of Au(111)/mica Substrates ...... 79

3.4 Preparation of STM Tips ...... 82

3.5 Preparation of Unaggregated and Aggregated Porphyrin Solutions ...... 85

3.6 Preparation of H2(H4TSPP) Nanorod Solutions Containing Chloroauric Acid ...... 86

3.7 Preparation and Analysis of STM and AFM Samples ...... 86

3.7.1 SPM Sample Preparation ...... 86

3.7.2 SPM Data Acquisition ...... 88

3.8 Preparation of Raman Samples ...... 90

3.9 Preparation of UV-visible and Resonance Light Scattering Samples ...... 91

3.10 Preparation and Measurement of XPS and UPS Samples...... 92

3.10.1 UPS and XPS Sample Preparation ...... 92

vii

3.10.2 UPS Spectral Acquisition ...... 92

3.10.3 XPS Spectral Acquisition ...... 93

3.11 Preparation of Helium Microscope Samples ...... 94

3.12 Preparation of Transmission Electron Microscopy Samples ...... 94

3.13 Optimization of H2(H4TSPP) geometry with Electron Affinity and Ionization

Potential Calculation...... 95

3.14 Fabrication of and Current vs. Voltage Measurements of H2(H4TSPP) Nanorods

Deposited on Interdigitated Electrodes ...... 95

Chapter 4: Results and Discussion ...... 105

4.1 Characterization of Tetrasulfonatophenyl Porphyrin and its Aggregate by UV-

visible and Resonance Light Scattering Spectroscopy ...... 105

4.2 Characterization of H2(H4TSPP) Aggregates by Ambient SPM Studies ...... 119

4.2.1 Characterization of H2(H4TSPP) Aggregates by Tapping Mode AFM ...... 119

4.2.2 Characterization of H2(H4TSPP) Aggregates by Ambient Scanning Tunneling

Microscopy ...... 125

4.3 Characterization of Tetrasulfonatophenyl Porphyrin and its Aggregate by Raman

and Resonance Raman Spectroscopy ...... 175

4.4 X-ray and Ultraviolet Photoelectron Spectroscopy Analysis of TSPP and its

Aggregate ...... 186

4.5 Ultra-High Vacuum STM Studies of H2(H4TSPP) Nanorods ...... 207

viii

4.5.1 UHV-STM Imaging Studies of H2(H4TSPP) Nanorods ...... 207

4.5.2 UHV-STM Imaging Studies of H2(H4TSPP) Monomers Deposited on HOPG211

4.5.3 UHV-STM Current versus Voltage Studies of H2(H4TSPP) Nanorods and

Monomers...... 224

4.6 Helium Ion Microscopy Studies ...... 232

4.7 Transmission Electron Microscopy Studies ...... 238

4.8 Nanorod Current vs. Voltage Studies via Interdigitated Electrode ...... 239

Chapter 5: Future Work ...... 244

Chapter 6: Conclusions ...... 246

List of Figures

Figure 1.1 1: Three forms of porphine, free base, diacid, and metallated along with the

nomeclature tracking the poisitions of substituent groups...... 2

Figure 1.1 2: Crystal structure of a light harvesting complex in Rhodospirillum

molischianum. Green squares are aggregated bacteriochlorophyll a molecules, blue

squares are monomeric bacteriochlorophyll a, and the yellow structures are

carotenoids. image from reference ...... 7

Figure 1.1 3: Schematic of photon absorption and electron transport among light

harvesting complexs in Rhodospirillum molischianum. image from reference ...... 8

Figure 1.2.3 1: Two forms of tetrasulfonatophenyporphine: free base and diacid...... 15

Figure 1.2.3 2: Structure of an H2(H4TSPP) dimer...... 16

Figure 1.2.3 3: Proposed model of the structure of the H2(H4TSPP) aggregate based on

cryo-electron microscopy from reference 62. Primed quantities refer to the tilting of

individual porphyrins. The sheet is rolled about the C vector to make the tube...... 21

Figure 1.2.3 4: Proposed models of the structure of the H2(H4TSPP) aggregate from

reference 103. B is the direction of the magnetic field. The small arrows are

perpendicular to the porphyrin macrocycle...... 23

Figure 1.3.1 1: Schematic of Porphine diacid with overlayed ring and the energy levels

predicted by the particle in a ring model...... 29

Figure 1.3.1 2: Orbitals and symmetries which constitute the four orbital model...... 30

Figure 1.3.1 3: States arising from configuration interactions in a porphryin...... 31

Figure 1.3.1 4: Wavefunctions associated with the electronic states of a D4h porphyrin. 32

Figure 1.3.1 5: UV-visible spectrum of H4(H4TSPP)Cl2 in HCl (pH 3.35)...... 33

Figure 1.3.2 1: Wavefunctions associated with the electronic states of a D4h porphyrin

and its corresponding dimer...... 36

Figure 1.3.2 2: Energy level diagram of a dimer with parallel transition dipole moments

with allowed (solid) and forbidden (dashed) transitions...... 37

Figure 1.3.2 3: Energy level diagram of a dimer with in line transition dipole moments

with allowed (solid) and forbidden (dashed) transitions...... 38

Figure 1.3.2 4: Energy level diagram of a dimer with co-planar transition dipole moments

with allowed (solid) and forbidden (dashed) transitions...... 39

Figure 1.3.2 5: Diagram of the H2(H4TSPP) dimer with superimposed transition dipole

moments...... 40

Figure 2.1 1: Electronic structure of a D4h porphyrin...... 43

Figure 2.1 2: Diagram of the electronic states of a D4h porphyrin and its dimer with

overlayed electronic transitions to an excited electronic state...... 44

Figure 2.1 3: Diagram of the the electronic states of a D4h porphyrin and its dimer with an

overlayed scattering process...... 45

Figure 2.1 4: RLS spectra of H2(H4TSPP) monomer (0.74 µM H2(H4TSPP) in 0.75 M

HCl, left) and aggregate (5.9 µM H2(H4TSPP) in 0.75 M HCl, right)...... 47

Figure 2.2 1: Diagrams illustrating photoemission from the valence band (left) and core

levels (right) ...... 49

Figure 2.2 2: XPS spectrum of ethyltrifluoroacetate carbon 1s spectrum illustrating the

chemical shifts of the different carbons. image from reference ...... 50

Figure 2.2 3: UPS Energy diagram of UPS illustrating the energy of the HeI photon and

the work function (Left), and UPS spectrum of HOPG (Right)...... 52

Figure 2.3 1: Diagram of a potential box with a finite barrier...... 54

Figure 2.3 2: Probability distribution of a particle in a box with a finite barrier...... 55

Figure 2.3 3: Tip-sample energy diagram where A) no tunneling is allowed, B) tunneling

from tip to sample is allowed, and C) tunneling from sample to tip is allowed...... 56

Figure 2.3 4: Diagrams illustrating the two different modes of STM operation; constant

current and constant height...... 58

Figure 2.3 5: Diagram illustrating the effect of differing conductivites on the path of an

STM tip...... 60

Figure 2.3 6: Tip-sample energy diagrams with the states of an adsorbed molecule under

a positive sample bias (A) and negative sample bias (B). The arrow indicates the

direction of electron flow...... 61

Figure 2.4 1: Diagram of the principle components of the AFM...... 64

Figure 2.4 2: Diagram depicting tapping mode AFM...... 65

Figure 2.5 1: Diagram of possible scattering events...... 67

Figure 2.5 2: Diagram the Raman experiment and definitions of S and P polarization. .. 69

Figure 2.5 3: Diagram of an adsorbed molecule and its image charges. The charges

cancel in the left-hand case and reinforce in the right-hand arrangement...... 70

Figure 2.6 1: A schematic of the Helium Ion Microscope. image from reference ...... 72

Figure 2.6 2: A schematic of the Helium Ion Microscope Tip. image from reference 134

...... 73

xii

Figure 3.3 1: Diagram of the apparatus and mask used to make Au(111)/mica substrates

(not to scale)...... 80

Figure 3.3 2: Ambient STM image of the terraced surface morphology of Au(111)...... 81

Figure 3.4 1: UHV-STM image and I(V) curve of Au(111). This curve was acquired at

(I,V)=(15 pA, 1.6 V). This I(V) curve is an average of 64 curves...... 84

Figure 3.4 2: UHV-STM image and I(V) curve of HOPG. This curve was acquired at

(I,V)=(15 pA, 1.6 V). This I(V) curve is an average of 64 curves...... 84

Figure 3.7.1 1: Photograph of a Au(111)/mica substrate mounted on the spin chuck used

for deposition...... 87

Figure 3.7.1 2: UV-visible spectra of 5 µM H2(H4TSPP)/0.75M HCl before and after

refluxing for 10 min. The black spectrum has been shifted up 0.2 absorbance units

for clarity...... 88

Figure 3.14 1: Schematic of the electrode to be used in nanorod I(V) experiments...... 96

Figure 3.14 2: AFM image of IdE and cross section...... 97

Figure 3.14 3: Cross section of IdE showing the electrode gap distance...... 98

Figure 3.14 4: Experimental setup for IdE I(V) experiments...... 100

Figure 3.14 5: Photograph of the experimental setup for interdigitated electrode I(V)

measurements...... 101

xiii

Figure 3.14 6: Photographs of the channels used to connect the breakout box to the

preamp...... 102

Figure 3.14 7: Photographs of the electronics boxes used to hold the resistor for

calibration and the electrode...... 102

Figure 3.14 8: Calibration curve for a 1 gigaohm resistor in the experimental setup

described in Figure 3.14 4...... 103

Figure 4.1 1: UV-visible specta of tetrasulfonatophenyl porphyrin: free base (Red, pH

10.36) diacid (Black, pH 3.35), and an intermediate pH with both free base and

diacid present (Blue, pH 5.27)...... 106

Figure 4.1 2: UV-visible spectra of varying concentrations of H4(H4TSPP)Cl2 in 0.75 M

HCl...... 107

Figure 4.1 3: Graph of absorbance at 490 nm vs. time during nanorod formation...... 109

Figure 4.1 4: UV-visible traces demonstrating the reversibility of aggregation: nanorod

solution at pH 0.12 (green), the same solution after rasing the pH to 10.32 (red), and

a reference free base spectrum at pH 10.00 (black)...... 110

Figure 4.1 5: RLS specta of varying concentrations of H4(H4TSPP)Cl2 in 0.75 M HCl. 111

Figure 4.1 6: RLS specta of varying concentrations of H4(H4TSPP)Cl2 in 0.75 M HCl. 112

Figure 4.1 7: Comparison of the scattering intensity of the J- and Q-bands of

H4(H4TSPP)Cl2 in 0.75 M HCl...... 113

Figure 4.1 8: Deconvoluted UV-visible spectrum of 5.9 µM H4(H4TSPP)Cl2 in 0.75 M

HCl...... 114

xiv

Figure 4.1 9: Plots of the intensities of various UV-visible peaks as a function of total

porphryin concentration. The concentration of HCl for the left graph is 0.75 M, the

right is 0.40 M...... 115

Figure 4.1 10: UV-visible spectra of free base (left) and diacid (right) TSPP solution (red)

and solid phase spectra (black). The concentration of both solution spectra is 2.618

µM. The pH‟s of the solutions in the solution phase spectra are 7.53 and 3.35 for the

free base and diacid respectivly. The solid phase spectra are of 50 µM porphyrin

solutions dried on quartz plates. The pH‟s of the solutions used for deposition are

7.65 and 3.73 for the free base and diacid respectivly...... 117

Figure 4.1 11: UV-visible spectra of 5 µM H2(H4TSPP) in 0.75M HCl: solution spectrum

(black) and deposited on a quartz plate for 90 min (red) followed by spin drying. 118

Figure 4.2.1 1: Tapping mode AFM images of H2(H4TSPP) nanorods deposited on

Au(111) (left) and HOPG (right)...... 120

Figure 4.2.1 2: Tapping mode AFM images of H2(H4TSPP) nanorods deposited on

Au(111) (left) and HOPG (right) with cross sections through single nanorods...... 121

Figure 4.2.1 3: Tapping mode AFM image of H2(H4TSPP) nanorods deposited on

Au(111). This solution was kept in the dark during preparation and deposition. .. 122

Figure 4.2.1 4: Tapping mode AFM image of H2(H4TSPP) nanorods deposited on HOPG.

Holes in the H2(H4TSPP) monolayer are marked with white circles...... 123

Figure 4.2.1 5: High resolution tapping mode AFM image of H2(H4TSPP) nanorods

deposited on HOPG...... 124

Figure 4.2.2 1: AFM (left) and STM (right) images of H2(H4TSPP) nanorods deposited

on Au(111) and accompanying cross sections...... 126

Figure 4.2.2 2: STM image of H2(H4TSPP) nanorods deposited on Au(111) and

accompanying cross section. The setpoint is 1 pA at 1.6 V sample bias...... 127

Figure 4.2.2 3: STM image of H2(H4TSPP) nanorods deposited on Au(111). The inset is

a subsequent smaller scan of the two tubes in the left of the image. The setpoint is 1

pA at 1.6 V sample bias in both images...... 128

Figure 4.2.2 4: AFM (left) and STM (right) images of H2(H4TSPP) nanorods deposited

on HOPG and accompanying cross sections...... 130

Figure 4.2.2 5: Standard “bird‟s eye view” of H2(H4TSPP) nanorods on HOPG (left) and

a three dimensional graph of the same image...... 131

Figure 4.2.2 6: High resolution images of the same H2(H4TSPP) nanorod section imaged

in succession at (A) 0.9 V and 0.015 nA, (B) 0.7 V and 0.015 nA, and (C) 0.5 V and

0.015 nA...... 132

Figure 4.2.2 7: Proposed circular model for the organizations of H2(H4TSPP) molecules

within a single disk substructure observed in high resolution images of H2(H4TSPP)

nanorods. The average disk diameter used in this model is ~6 nm. The CPK

representation of the porphyrin molecule used in the model is based on the van der

Waal radii and ~21° porphyrin ring deformation. To generate the aggregate ring

structure, the molecules were manipulated and displayed in DS Viewer Pro

(Accelrys). A 25 nm2 STM section of the high resolution image in Figure 4.2.2.6 is

xvi

inserted for reference. Below the H2(H4TSPP) model is a schematic illustration

showing a portion of a circular aggregate containing N monomers, which deviate

from planarity by R = 2π/(N + 1). Explanation of the model is provided in the text.

...... 135

Figure 4.2.2 8: UV-visible spectra of nanorod solutions with varying concentrations of

Au(III)...... 140

Figure 4.2.2 9: UV-visible spectra of nanorod solution with chloroauric acid added after

the nanorods were finished aggregating. A reference spectrum of nanorods with no

chloroauric acid is included...... 141

Figure 4.2.2 10: UV-visible spectra of monomeric H2(H4TSPP) heated in the presence of

chloroauric acid. A reference spectrum of monomeric H2(H4TSPP) with no

chloroauric acid is included...... 142

Figure 4.2.2 11: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au

foil. Reference spectra of 5 µM H2(H4TSPP)/0.75 M HCl with no Au foil are

included...... 144

Figure 4.2.2 12: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au

foil in the range of 350 nm to 200 nm. Reference spectra of 5 µM H2(H4TSPP)/0.75

M HCl with no Au foil are included...... 145

Figure 4.2.2 13: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl with Au foil and 5

µM HAuCl4/0.75 M HCl in the range of 350 nm to 200 nm. A reference spectrum of

5 µM H2(H4TSPP)/0.75 M HCl with no Au foil is included...... 146

xvii

Figure 4.2.2 14: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au

foil in the range of 750 nm to 600 nm. Reference spectra of 5 µM H2(H4TSPP)/0.75

M HCl with no Au foil are included...... 147

Figure 4.2.2 15: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au

foil in the range of 460 nm to 400 nm. Reference spectra of 5 µM H2(H4TSPP)/0.75

M HCl with no Au foil are included...... 148

Figure 4.2.2 16: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au

foil in the range of 500 nm to 460 nm. Reference spectra of 5 µM H2(H4TSPP)/0.75

M HCl with no Au foil are included...... 149

Figure 4.2.2 17: Plots of peak positions in UV-visible spectra of 5 µM H2(H4TSPP)/0.75

M HCl over time with and without Au foil...... 150

Figure 4.2.2 18: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl solution before

and after reflux and addition of 6 cm2 Au foil...... 152

Figure 4.2.2 19: UHV-STM images of peeled HOPG (left, setpoint 100 pA at -0.05 V

sample bias) and HOPG treated with 0.75 M HCl for 1 hr (right, 30 pA at -0.05 V

sample bias)...... 154

Figure 4.2.2 20: UHV-STM images of annealed Au(111) (left, setpoint 1 pA at 1.6 V

sample bias) and Au(111) treated with 0.75 M HCl for 1 hr (right, setpoint 1 pA at

1.6 V sample bias)...... 155

Figure 4.2.2 21: UHV-STM image of Au (111) treated with 0.75 M HCl for 1 hr (setpoint

1 pA at 1.6 V sample bias)...... 156

Figure 4.2.2 22: Histogram of disk widths observed on various substrates by UHV-STM.

5 µM H2(H4TSPP)/0.75 M HCl deposited for 60 min on Au(111) (Black), 5 µM

xviii

H2(H4TSPP)/0.75 M HCl deposited for 60 min on HOPG (Green), 1 µM

H2(H4TSPP)/0.75 M HCl deposited for 40 min on HOPG (Red)...... 157

Figure 4.2.2 23: Histogram of disk widths observed on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 20 min (Red), 5 µM H2(H4TSPP)/0.75 M

HCl aggregated in 5 µM HAuCl4 deposited for 20 min (Blue), 5 µM

H2(H4TSPP)/0.75 M HCl with 5 µM HAuCl4 added after aggregation deposited for

20 min on HOPG (Green), 5 µM H2(H4TSPP)/0.75 M HCl with 2.5 µM HAuCl4

added after aggregation deposited for 20 min (Black)...... 158

Figure 4.2.2 24: Histogram of disk widths observed on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 20 min (Red), 5 µM H2(H4TSPP)/0.75 M

HCl aggregated in 5 µM HAuCl4 deposited for 20 min (Blue), 5 µM

H2(H4TSPP)/0.75 M HCl with 5 µM HAuCl4 added after aggregation deposited for

20 min on HOPG (Green), 5 µM H2(H4TSPP)/0.75 M HCl with 2.5 µM HAuCl4

added after aggregation deposited for 20 min (Black), 5 µM H2(H4TSPP)/0.75 M

HCl exposed to Au foil for 1 hr (Purple)...... 159

Figure 4.2.2 25: STM image of disks on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 20 min setpoint 1 pA at 1.6 V sample bias.

No gold is present in this sample...... 160

Figure 4.2.2 26: STM image of disks on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl aggregated in 5 µM chloroauric acid deposited for 20 min

(setpoint 1 pA at 1.6 V sample bias)...... 161

xix

Figure 4.2.2 27: STM image of disks on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl. 5 µM chloroauric acid was added after aggregation was

complete. Deposition time was 20 min. setpoint 1 pA at 1.6 V sample bias...... 162

Figure 4.2.2 28: STM image of several nanorods on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl. 5 µM chloroauric acid was added after aggregation was

complete deposited for 20 min. setpoint 1 pA at 1.6 V sample bias...... 163

Figure 4.2.2 29: STM images of nanorods on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl. 5 µM chloroauric acid was added after aggregation was

complete deposited for 20 min. setpoint 1 pA at 1.6 V sample bias...... 164

Figure 4.2.2 30: STM image of several disks on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 20 min. This sample was heated with Au foil

in it (setpoint 2 pA at 1.6 V sample bias)...... 165

Figure 4.2.2 31: Histogram of disk widths observed on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 60 min (Green), 5 µM H2(H4TSPP)/0.75 M

HCl deposited for 30 min (Red), 5 µM H2(H4TSPP)/0.75 M HCl deposited for 20

min (Black)...... 167

Figure 4.2.2 32: Histogram of disk widths observed on HOPG by STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 60 min scanned in air (Green), 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 60 min scanned in UHV (Red), 1 µM

H2(H4TSPP)/0.75 M HCl deposited for 40 min scanned in UHV (Black)...... 168

Figure 4.2.2 33: Histogram of disk widths observed on Au(111) by ambient STM. 10

µM H2(H4TSPP)/0.3 M HCl deposited for 10 min (Red), 10 µM H2(H4TSPP)/0.3 M

HCl deposited for 60 min (Black), 10 µM H2(H4TSPP)/0.3 M HCl deposited for 120

min (Blue)...... 169

Figure 4.2.2 34: TEM image of 5 µM H2(H4TSPP)/0.75 M HCl deposited on a carbon

coated Ni Formvar TEM grid for 20 min followed by spin drying for 30 sec at 4000

rpm...... 170

Figure 4.2.2 35: TEM image of 5 µM H2(H4TSPP)/0.75 M HCl treated with Au foil

deposited on a carbon coated Ni Formvar TEM grid for 20 min followed by spin

drying for 30 sec at 4000 rpm...... 171

Figure 4.2.2 36: TEM image of 5 µM H2(H4TSPP)/0.75 M HCl treated with Au foil

deposited on a carbon coated Ni Formvar TEM grid for 20 min followed by spin

drying for 30 sec at 4000 rpm...... 172

Figure 4.2.2 37: TEM image of 5 µM H2(H4TSPP)/0.75 M HCl treated with Au foil

deposited on a carbon coated Ni Formvar TEM grid for 20 min followed by spin

drying for 30 sec at 4000 rpm...... 173

Figure 4.2.2 38: EDS spectrum of 5 µM H2(H4TSPP)/0.75 M HCl treated with Au foil

deposited on a carbon coated Ni Formvar TEM grid ...... 174

Figure 4.3 1: Atom labels used in the vibrational assignments of TSPP...... 176

Figure 4.3 2: SS (black) and SP (red) polarized Resonance Raman Spectra of free base

4- H2TSPP . The excitation wavelength is 413.1 nm. A Gaussian background has

been subtracted from the data. The solution pH is 9.68...... 179

xxi

Figure 4.3 3: SS (black) and SP (red) polarized Resonance Raman Spectra of diacid

2- H2TSPP . The excitation wavelength is 457 nm. A Gaussian background has been

subtracted from the data. The solution pH is 3.94...... 181

Figure 4.3 4: SS (black), SP (red) and SP x3 (dotted blue) polarized Resonance Raman

Spectra of H2(H4TSPP) nanorods in 1.5 M HCl. The excitation wavelength is 488

nm. A Gaussian background has been subtracted from the data...... 183

Figure 4.3 5: Resonance Raman Spectra of H2(H4TSPP) nanorods in solution (black) and

deposited on Au(111). The excitation wavelength is 488 nm. A Gaussian

background has been subtracted from the data...... 185

Figure 4.4 1: XPS survey spectrum of Na4(H2TSPP) powder on In...... 186

Figure 4.4 2: XPS sulfur, carbon, oxygen, and sodium spectra of Na4(H2TSPP) powder on

In...... 187

Figure 4.4 3: XPS survey spectrum of H4(H4TSPP)Cl2 powder on In...... 188

Figure 4.4 4: XPS sulfur, carbon, oxygen, and chlorine spectra of H4(H4TSPP)Cl2 powder

on In...... 189

Figure 4.4 5: Nitrogen 1s XPS spectra for free base and diacid powders on indium. .... 190

Figure 4.4 6: TSPP wireframes showing the difference in protonation state between free

base and diacid...... 190

Figure 4.4 7: XPS survey spectrum of H2(H4TSPP) aggregates deposited on Au(111). 195

Figure 4.4 8: High resolution elemental XPS scans of H2(H4TSPP) nanorods on Au(111).

...... 196

xxii

Figure 4.4 9: S 2p XPS spectra of free base (blue), nanorods (red), and sulfurous acid

(black) deposited on Au(111)...... 197

Figure 4.4 10: XPS survey spectrum of H2(H4TSPP) aggregates deposited on HOPG. 199

Figure 4.4 11: High resolution elemental XPS scans of H2(H4TSPP) nanorods on HOPG.

...... 200

Figure 4.4 12: XPS survey spectrum of H2(H4TSPP) monomers deposited on HOPG. . 202

Figure 4.4 13: High resolution elemental XPS scans of H2(H4TSPP) monomers on

HOPG...... 203

Figure 4.4 14: N 1s spectra of the H2(H4TSPP) monomer deposited on HOPG...... 204

Figure 4.4 15: UPS spectra of peeled HOPG (blue), HOPG treated with HCl (red), and a

monolayer of H2(H4TSPP) on HOPG (black)...... 206

Figure 4.5.1 1: UHV-STM images of H2(H4TSPP) nanorods on Au (111) (left, setpoint 1

pA at 1.5 V sample bias) and on HOPG (right, setpoint 1 pA at 1.6 V sample bias).

...... 208

Figure 4.5.1 2: UHV-STM images of H2(H4TSPP) nanorods deposited on Au(111) (left,

setpoint 1 pA at 1.3 V sample bias) and HOPG (right, setpoint 1 pA at 1.6 V sample

bias) with cross sections through single nanorods...... 209

Figure 4.5.1 3: UHV-STM image of H2(H4TSPP) nanorods and monomers deposited on

HOPG (setpoint 1 pA at 1.6 V sample bias)...... 211

xxiii

Figure 4.5.2 1: B3LYP 6-31+G(d,p) optimized structure of the H2(H4TSPP) diacid, top

and side...... 213

Figure 4.5.2 2: STM image of H2(H4TSPP) monolayer on HOPG. 1.6V and 1 pA. .... 216

Figure 4.5.2 3: High resolution image of H2(H4TSPP) monolayer on HOPG showing

detailed molecular packing and distortion of porphyrin due to complete macrocycle

protonation. V=1.6V, setpoint is 1 pA. Note the difference in orientation of

molecules within square and within ellipse...... 218

Figure 4.5.2 4: Unit cell for the more common H2(H4TSPP) surface structure on HOPG.

...... 220

Figure 4.5.2 5: Unit cell for the more common H2(H4TSPP) surface structure on HOPG

showing the unit cell vectors in terms of HOPG coordinates...... 221

Figure 4.5.2 6: Unit cell for the more common H2(H4TSPP) surface structure on HOPG

with overlaid ellipses to show correlation with STM images...... 222

Figure 4.5.2 7: Unit cell for the less common H2(H4TSPP) surface structure on HOPG

with overlaid ellipses to show correlation with STM images...... 223

Figure 4.5.3 1: I(V) curves of 4 nm tall H2(H4TSPP) nanorods deposited on Au(111)

(black) and HOPG (red). In both cases the intial parameters were setpoint 15 pA at

1.6 V sample bias...... 225

Figure 4.5.3 2: STM image of a stack of nanorods (setpoing 1 pA at 1.6 V sample bias)

and I(V) curves taken at various points on the stack...... 226

Figure 4.5.3 3: Plot of the natural log of current flow at -2 V vs. nanorod thickness. .. 227

xxiv

Figure 4.5.3 4: Plot of the natural log of nanorod rectification ratios vs. nanorod

thickness...... 228

Figure 4.5.3 5: dI/dV spectra of NiTPP, TSPP, and a 12 nm stack of nanorods...... 229

Figure 4.5.3 6: Diagram of the effect of aggregation on the electronic structure of

H2(H4TSPP)...... 231

Figure 4.5.3 7: I(V) curves of 4 nm nanorods taken at room temperature and 90 K...... 232

Figure 4.6 1: He ion micrograph of H2(H4TSPP) nanorods deposited on HOPG...... 233

Figure 4.6 2: He ion micrograph of H2(H4TSPP) nanorods deposited on HOPG showing

the effects of He ion sputtering on the sample...... 234

Figure 4.6 3: He ion micrograph of H2(H4TSPP) nanorods deposited on Au(111)...... 235

Figure 4.6 4: He ion micrograph of H2(H4TSPP) nanorods deposited on Au(111) and

tilted 20°...... 236

Figure 4.6 5: He ion micrograph of H2(H4TSPP) nanorods deposited on Au(111) and

tilted 20° with the dimensions of several rods...... 237

Figure 4.7 1: TEM micrograph of H2(H4TSPP) nanorods deposited on a carbon coated Ni

Formvar TEM grid...... 238

Figure 4.7 2: TEM micrograph of H2(H4TSPP) nanorods deposited on a carbon coated Ni

Formvar TEM grid...... 239

Figure 4.8 1: SEM micrographs of H2(H4TSPP) nanorods deposited on an IdE...... 241

xxv

Figure 4.8 2: I(V) curve of H2(H4TSPP) nanorods deposited on an IdE and an IdE treated

with HCl. Both measurements were taken with the samples in a closed box to

prevent photoconductivity...... 242

List of Tables

Table 1.1 1: Absorbance bands for two different Zn porphyrins...... 4

Table 1.1 2: Electrochemical and absorbance data for metalated octaethyl porphyrins.

The potentials are vs. SCE and taken in methanol or chloroform-methanol. data

from reference ...... 5

Table 1.2.3 1: Reported dimensions of H2(H4TSPP) nanorods prepared under different

conditions. (* Apparent height) ...... 25

Table 4.3 1: Vibrational mode positions (Δυ in cm-1) and assignments of TSPP...... 177

Table 4.4 1: Theoretical and experimental elemental ratios in TSPP free base and diacid

powders...... 191

Table 4.4 2: Elemental ratios of Na4(H2TSPP) by XPS, Columbia Analytical Services,

and the theoretical values...... 192

Table 4.4 3: Elemental ratios of H4(H4TSPP)Cl2 by XPS, Columbia Analytical Services,

and the theoretical values...... 193

xxvi

Table 4.4 4: Elemental ratios of H4(H4TSPP)Cl2 by XPS, Columbia Analytical Services,

and the theoretical values...... 194

Table 4.4 5: Atomic ratios by XPS of the theoretical diacid, diacid powder, and nanorods

deposited on Au(111)...... 198

Table 4.4 6: Atomic ratios by XPS of the theoretical diacid, diacid powder, and nanorods

deposited on HOPG...... 202

Table 4.4 7: Atomic ratios by XPS of the theoretical diacid, diacid powder, and nanorods

deposited on HOPG...... 205

Table 4.4 8: Peak positions in eV of the elements studied by XPS...... 205

Table 4.5.2 1: Electron Affinity values in Different Phases a) Reference 194; b)

E1/2+4.71 V; c) Reference 191; d) The work function for NiTPP/Au(111) taken as

5.2 eV and that for ...... 215

Table 4.5.2 2: Ionization Potential values in Different Phases a) Reference 194; b)

E1/2+4.71 V; c) Reference 191; d) The work function for NiTPP/Au(111) taken as

5.2 eV and that for H2(H4TSPP)/HOPG as 4.9 eV; e) Reference 195; f) Reference

196; g) Reference 192; h) UPS data from reference assuming the Ag work function

was 4.6 eV...... 215

xxvii

Glossary of Terms

AFM: Atomic Force Microscopy

BO Approximation: Born Oppenheimer Approximation

DFT: Density Functional Theory

DLS: Dynamic Light Scattering

EA: Electron Affinity

EB: Binding Energy

EDS: Electron Dispersive Spectroscopy

EF: Fermi Energy

Ekin: Kinetic Energy

ELS: elastic light scattering

EV: Vacuum Energy

HIM: Helium Ion Microscope

HOPG: Highly Ordered Pyrolytic Graphite

IdE: Interdigitated Electrode

IP: Ionization Potential

IR: Infrared Spectroscopy

xxviii

KHD: Kramers-Heisenberg-Dirac

LDOS: local density of states

Nanorod/Nanotube: Idiomatic for H2(H4TSPP) aggregate

NiTPP: Nickel Tetraphenylporphyrin

OMTS: orbital mediated tunnleing spectroscopy

Q-Band: An absorbance band to the red of the Soret band in a porphyrin

RLS: Resonance Light Scattering

SAXS: Small Angle X-ray Scattering

SCE: saturated calomel electrode

SEM: Scanning Electron Microscope

Soret (B) -band: A strong absorbance in the blue region of the UV-visible spectrum of a porphyrin

SPM: Scanning Probe Microscopy

STM: Scanning Tunneling Microscopy

TEM: Transmission Electron Microscope

UHV: Ultrahigh Vacuum

UPS: Ultraviolet Photoelectron Spectroscopy

xxix

UV-visible: Ultraviolet-Visible Spectroscopy

WKB: Wentzel-Kramers-Brillouin

XPS: X-ray Photoelectron Spectroscopy

xxx

Chapter 1: Introduction

1.1 Properties and Applications of Porphyrins

Porphyrins are a class of aromatic organic compounds composed of carbon, nitrogen, and hydrogen. Porphyrins were first described by Alfred Treibs in 1936 as a part of his work in the field of geochemistry.1 The most basic porphyrin, called porphine, is a 24-membered ring composed of four pyrroles connected by methine bridges. This

24-membered ring is referred to as the macrocycle. The electronic structure of porphyrins is of central importance to this work and will be covered in detail in a separate section. Porphyrin compounds are highly varied so it is worthwhile to briefly cover nomenclature. Figure 1.1 1 shows three common forms of porphyrins: free base, diacid, and metallated.

Figure 1.1 1: Three forms of porphine, free base, diacid, and metallated along with the nomeclature tracking the poisitions of substituent groups.

The terminology with respect to protonation of the nitrogens is straightforward.

The free base has two unprotonated nitrogens which can accept hydrogen ions, hence the term base. In this form the molecule has a neutral charge and is assigned D2h symmetry.

The diacid‟s nitrogens are fully protonated and the overall molecular charge is +2.

Protonation of the macrocycle nitrogens has important effects on the molecule‟s geometry. To first approximation the diacid belongs to the D4h point group and is commonly treated as such. In reality the macrocycle is not large enough to accommodate four hydrogens so the molecule adopts a saddled geometry.2 The saddled nature of porphyrin diacids is an important factor to consider during the study of porphyrin aggregation and its organization on substrates. Metallated porphyrins are synthesized by

2 replacing the two central hydrogens with a metal ion such as (but not limited to) Co, Ni,

Cu, Zn, and Mg.3 R groups can be added to the base porphine unit in order to affect the molecule‟s properties. For example, substituting benzoic acid groups at the meso positions on the ring would increase solubility in water and add intermolecular hydrogen bonding interactions. Pyridyl substitution at the meso position adds additional proton acceptor sites creating the potential of a highly charged ion in acidic solutions. The macrocycle carbons are named as shown in Figure 1.1 1, and the position of substituent groups is tracked by the appropriate number or Greek letter.4 For example, if phenyl groups were attached at the α and β positions the molecule would be α,β- diphenylporphine.

Porphyrins are an important family of compounds for several reasons. They can serve as components in solar cells5, non-linear optical devices6, sensors7, act as catalysts8, and serve as sensitizers in photodynamic tumor therapy9. One of the major strengths of porphyrins is flexibility. Porphine can be modified in a number of ways by adding R groups and substituting metals into the macrocycle. Modifications to the base porphine unit affect its optical and electronic properties allowing porphyrins to be tuned to a particular use. Consider the absorption spectra of Zn(II)-1,2,3,4,5,6,7,8- octaethylporphine and Zn(II)-α,β,γ,δ-tetraphenylporphine, the two species differ only in the substituents attached to the macrocycle.

Species Absorbance Band Position (nm)

Zn(II)-1,2,3,4,5,6,7,8- 404 534 572 N/A octaethylporphine10 (Strongest)

Zn(II)-α,β,γ,δ-tetraphenylporphine11 400 470 745 845 (Strongest)

Table 1.1 1: Absorbance bands for two different Zn porphyrins.

Differing substituents on the macrocycle have a clear impact on the position and number of absorption bands. The strongest peak in the phenylated porphine is 66 nm to the red of the octaethyl species. In addition, the reddest band in the phenylated porphyrin is 273 nm to the red of the lowest energy band in the octaethyl species. The data in Table 1.1 1 shows that the electronic structure of a porphyrin is strongly influenced by substituent groups. Table 1.1.2 contains absorbance and electrochemical data for a series of metallated octaethylporphines. In all cases listed in Table 1.1.2 the strongest absorbance band is within 8 nm of 400 nm. The weaker bands to the red show more scatter but are all between 510 and 580 nm. The UV-visible spectra of these porphyrins are not strongly influenced by the choice of central metal. In contrast the midpoint potential is strongly influenced by the central metal and increases with the electronegativity of the central metal ion.10

Species Midpoint Potential (mV) Absorbance Bands (nm)

Mg(II)-1,2,3,4,5,6,7,8- 427±5 404 (Strongest), 542, octaethylporphine 578

Zn(II)-1,2,3,4,5,6,7,8- 525±10 404 (Strongest), 534, octaethylporphine 572

Cu(II)-1,2,3,4,5,6,7,8- 601±10 399 (Strongest), 522, octaethylporphine 560

Ni(II)-1,2,3,4,5,6,7,8- 636±15 392 (Strongest), 517, octaethylporphine 551

Pd(II)-1,2,3,4,5,6,7,8- 726±15 392 (Strongest), 512, octaethylporphine 546

Table 1.1 2: Electrochemical and absorbance data for metalated octaethyl porphyrins.

The potentials are vs. SCE and taken in methanol or chloroform-methanol. data from reference 10

The conclusions from Table 1.1 1 and Table 1.1 2 are specific to these species; we do not mean to cast them as rules governing the whole family of porphyrins. What the data does illustrate is how the choice of substituents or metal ion can be used to synthesize a molecule with electrochemical and absorbance properties tailored to a particular purpose.

Thus far we have restricted ourselves to two different porphyrins and six different metals in order to prove a point about the properties of porphyrins. The narrow focus does not do justice to the potential porphyrins offer. The field of synthetic organic chemistry has a myriad of R-groups and coupling reactions which can be used to modify the porphyrin.

We have not considered the effects of charged R-groups, different R-groups at different

5 positions, reducing one of the double bonds in the macrocycle, or any number of other modifications. The possibilities of these molecules are as broad as our imaginations.

The vast potential of porphyrin chemistry becomes almost limitless when we consider self-assembled supramolecular aggregates of porphyrins. A molecular aggregate is an ordered array of molecular units12 and it has been shown that dimers and larger aggregates can exhibit properties which are different from the parent molecules.13

Molecular self-assembly is the spontaneous association of molecules under equilibrium conditions into stable, structurally well-defined aggregates joined by noncovalent bonds.14 Just as the characteristics of a porphyrin monomer can be influenced by things like substituent groups the characteristics of molecular aggregates are strongly influenced by their constituent molecules. Supramolecular structures could be made from a single type of porphyrin, porphyrins with different or opposing charges, metallated with non- metalated, two porphyrins with different metals, water-soluble with water-insoluble, a porphyrin with another type of molecule, and so forth. The experimental conditions may also play a role in the properties of supramolecular aggregates. A particular aggregate may form in solution, or at a solution-surface interface, or upon vapor deposition of the constituents. The extensive library of porphyrin compounds offers a wellspring of species to be used in the synthesis of molecular aggregates; with the choice of building block influencing the properties of entire aggregate.

As an example of the potential of molecular aggregation consider the light harvesting complexes in the photosynthetic bacterium Rhodospirillum molischianum. In this organism bacteriochlorophyll a molecules are employed in aggregated and

6 unaggregated forms to widen the range over which the light needed for photosynthesis can be absorbed and to improve photosynthetic efficiency.15 The light harvesting complexes are composed of a ring of bacteriochlorophyll a molecules (the aggregate) and isolated monomers.

Figure 1.1 2: Crystal structure of a light harvesting complex in Rhodospirillum molischianum. Green squares are aggregated bacteriochlorophyll a molecules, blue squares are monomeric bacteriochlorophyll a, and the yellow structures are carotenoids. image from reference 15

Monomeric bacteriochlorophyll a has an electronic absorption band at 800 nm, while the ring porphyrins absorb at 850 nm. By employing both the ring-shaped aggregate and the monomer a bacterium can absorb photons over a wider range of the electromagnetic spectrum. In addition to widening the absorption envelope the complexes are capable of inter-aggregate electron transfer. The redox chemistry of photosynthesis occurs in a set

7 of bacteriochlorophyll a molecules called the special pair. The special pair can run photosynthetic redox reactions 100 times faster than it can absorb light.15 In order to increase the efficiency of the process the special pair is surrounded by a circular aggregate of 16 bacteriochlorophyll a molecules called light harvesting complex I. Light harvesting complex I is in turn surrounded by many circular aggregates called light harvesting complex II (see Figure 1.1 3). Photons are absorbed by light harvesting complex II and transferred from complex to complex until they reach the special pair.

This arrangement of complexes increases photon absorption, providing electrons to the special pair at an accelerated rate increasing photosynthetic efficiency.15

Figure 1.1 3: Schematic of photon absorption and electron transport among light harvesting complexs in Rhodospirillum molischianum. image from reference 15

The light harvesting complexes in Rhodospirillum molischianum provide a powerful example of the importance and capabilities of molecular aggregates. A second example of the importance of molecular aggregation is nanowires formed by metallated

1,2,3,4,5,6,7,8-octaethylporphine (Co, Ni, Cu, Zn, and Mg).3 Aggregates 200 nm wide and several microns long can be formed by vapor depositing metalated

8 octaethylporphyrin on varying substrates. The nanowires were used to bridge electrical contacts and were shown to be both photoconducting and field-emissive. Molecular aggregates are also potentially useful in the field of catalysis. Co(II)-α,β,γ,δ- tetrasulfonatophenylporphine electropolymerized with aniline forms nanowires 30-50 nm across and 500 nm long.16 When deposited on a glassy carbon electrode the aggregates proved capable of reducing molecular oxygen illustrating the possible utility of molecular aggregates as components in fuel cells.16

These examples demonstrate that molecular aggregates have great potential as device components. In order to effectively utilize molecular aggregates in practical applications a more fundamental understanding of their formation, structure, electronic properties, and the interplay of these three features is needed. A firm grasp of the integral concepts of molecular aggregation will open up the possibility of constructing nanoscale wires, switches, diodes, capacitors, and other electronic components simply by selecting a suitable constituent molecule.

1.2 Literature Review of Porphyrin Aggregation

In the last decade there has been tremendous interest in inorganic,17-21 metallic,22-

26 and carbon27-30 nanostructures. Applications of these nanostructures range over a wide area, including hydrogen storage,31,32 studies of living cells,33 drug delivery,34 photonic materials,18 sensing and detecting,35-37 optics,38 catalysis,22,39 and electronic devices.40

Organic nanostructures are also known,41-46 and are especially intriguing because of the wide range of compounds from which they may be made. This advantage is greatly magnified by the demonstrated fact that the optical and electronic properties of organic

9 nanostructures clearly differ from those of the bulk materials.16,47-49 For example supramolecular nanorods of 5,15-diaryl substituted porphyrins exhibit a much broader absorption spectrum than simply deposited material.48

An interesting class of these organic nanostructures is the self-assembled structures built from porphyrins.3,49-53 The ease of synthesis and robust character of porphyrin based materials allows for the production of a novel class of nanomaterials with potential applications in catalysis, sensor, solar cells, and electronic devices.50 The shapes formed upon aggregation are highly varied ranging from spheres to flower-like structures to rods. Porphyrin aggregates range in shape and size as the porphyrin skeleton can be substituted in a number of different ways resulting in different intermolecular interactions.

1.2.1 Non-Ionic Porphyrin Aggregates in Solution and at Surfaces

Porphyrin aggregates formed in solutions take several forms. PdCl2 linked porphyrin dimers in a toluene solution act as molecular tweezers in a complexation reaction with fullerenes.54 Fullerenes complexed in the “jaws porphyrin” have fluorescence and absorbance spectra which are not a sum of the two constituent compounds indicating that complexation induces changes in electronic structure. The complex also has potential applications in photoconduction. A second example of porphyrin containing chelating agents is a pair of zinc porphyrins connected by an aromatic oligoamid spacer.55 This porphyrin has been shown to chelate crown ethers.

Examples of non-ionic porphyrin aggregates forming larger aggregates are also known. Both α-(3‟-pyridyl)-β,γ,δ-tris(4‟-carboxyphenyl)porphine and α-(2‟-quinolyl)-

β,γ,δ-tris(4‟-hydroxyphenyl)porphine form extensive sheet-like aggregates upon evaporation of the chloroform/ethanol solvent.56 The aggregate is held together by hydrogen bonding between carboxyl and pyridyl groups in the case of α-(3‟-pyridyl)-

β,γ,δ-tris(4‟-carboxyphenyl)porphine and hydroxyl and quinone groups in α-(2‟- quinolyl)-β,γ,δ-tris(4‟-hydroxyphenyl)porphine aggregates. While both porphyrins form extensive hydrogen bonded networks the structures are different indicating that the choice of R-group can influence aggregate morphology. Rings ranging from 10 nm to 10

µm in diameter and up to 200 nm tall can be formed on both glass and graphite substrates by the evaporation of bis(21H,23H-α(4-pyrydyl)-β,γ,δ-tris(4- hexadecyloxyphenyl)porphine)platinum dichloride dissolved in chloroform.57

Comparison of UV-visible spectra of the porphyrin dissolved in chloroform and on glass shows shifts in visible absorption bands upon solvent evaporation indicating electronic coupling among constituent molecules. In both cases it appears that the increase in concentration by solvent evaporation is an important driver in aggregation.

1.2.2 Ionic Porphyrin Aggregates in Solution and on Surfaces

Of particular interest to us, are the porphyrin nanostructures created by ionic self- assembly. Ionic self-assembly is the coupling of structurally different simple ionic blocks (charged tectons), or structurally similar (or identical) zwitterionic building blocks. Instead of hydrogen bonding or Van der Waals interactions the aggregate is held together by stronger electrostatic interactions. It is logical to divide our discussion of

11 ionic porphyrin aggregation in to two sections: solution and solid phase studies. Solution phase studies are important because the systems discussed in this section aggregate in solution. Spectroscopic techniques such as UV-visible spectroscopy and Resonance

Light Scattering (RLS) are useful for studying aggregation processes in solution. While important, optical spectroscopy is not well suited for studying the shapes of aggregates.

Microscopic techniques such as scanning electron microscopy (SEM), transmission electron microscopy (TEM), and scanning probe microscopy (SPM) can provide detailed structural data that optical spectroscopy can not. With the notable exception of SEM images of frozen solutions these three microscopy techniques are not capable of visualizing an aggregate in solution. SEM and TEM require samples to be under vacuum during analysis. The deposition and desiccation of aggregates necessary for SEM and

TEM may influence the structure and/or properties of the system. Solution phase SPM offers the opportunity to study aggregates in solution, but the aggregates still must be deposited on a substrate. It is important to remember that the deposition and desiccation typically required for microscopy of aggregates formed in solution may cause changes in the aggregate structure and or properties.

α,β,γ,δ-tetraphosphonatophenylporphine (H2TPPP) forms aggregates in aqueous solutions at low pH‟s, and in the presence of a G5 poly(amidoamine) dendrimer.58 The charge on H2TPPP is pH dependant, in basic solution each phosphonato group is -2 for a total charge of -8. Below pH 2 the phosphonato groups are protonated and uncharged while the macrocycle is +2 due to nitrogen protonation. Both dendrimer and acid addition cause changes to the H2TPPP absorption spectrum indicative of solution phase

12 aggregation. Weak RLS indicates that the solution phase aggregate are relatively small.

When H2TPPP in solution with G5 poly(amidoamine) is deposited on HOPG aggregates ranging from globules tens of nanometers wide to strings up to 300 nm are observed by

58 AFM. A similar molecule, α,β,γ,δ-tetracarboxyphenylporphine (H2TCPP) forms aggregates in acidic solutions as well.59 Tetracarboxyphenyl porphyrin is similar to

α,β,γ,δ-tetraphosphonatophenylporphine; the only structural difference is the carboxylate vs. phosphonate groups. Similarly to the H2TPPP case the peripheral acid groups and macrocycle nitrogens of H2TCPP protonate in acidic solutions resulting in an ion with a

+2 charge. H2TCPP is an interesting case because its aggregation is sensitive to the type of acid used. Monomeric H2TCPP‟s Soret band is located at 440 nm, upon addition of sufficient HCl to reach pH 0.9 the 440 nm absorbance dies away and a new band grows in at 417 nm. If nitric acid is used to reach the same pH the monomer band decreases in favor of two new bands at 406 and 467 nm. The appearance of different absorption bands upon addition of different acids suggests that aggregate formation is a function of acid species. RLS data of H2TCPP aggregated in both HCl and HNO3 is consistent with the formation of different aggregates with different acids. The RLS spectrum of H2TCPP in HCl shows little scattering which indicates small aggregates. H2TCPP aggregated in

HNO3 has a sharp RLS peak superimposed on a broad scattering background which indicates that the aggregates formed in nitric acid are larger than those formed in HCl.

Deposition of both solutions on silica provides further evidence of aggregation as a function of acid.60 AFM images of the HCl aggregate reveal rings with diameters of 200-

2000 nm and heights of 4-5 nm. Aggregates formed in nitric acid were rod-shaped ~4 nm tall ~20 nm wide and microns long. H2TCPP is an interesting aggregation case study

13 due to its dependence on the acid species used to induce aggregation. Ionic porphyrin aggregates are not limited to rod and ring structures. Four-leaf clover-shaped aggregates

5 µm in diameter can be made from a mixture of ZnIIT(N-EtOH-4-Py)P4+ and

SnIVtetrasulfonatophenylporphine.50

1.2.3 Aggregates of Tetrasulfonatophenyporphine:

The prototypical ionic porphyrin aggregate, and the one to be studied here, is the

4- aggregate formed from the α,β,γ,δ-tetrasulfonatophenylporphine, (H2TSPP) ion.

4- H2TSPP is probably one of the most studied synthetic porphyrin complexes. A

4- Scifinder search in January of 2011 returned 197 references for the H2TSPP molecule.

4- Understanding H2TSPP aggregation begins with a discussion of the acid base characteristics of the molecule. In solution below pH 4, the inner nitrogen system is fully protonated to give a net +2 charge to the central region of the porphyrin and a net -2

4- 61 4- charge overall, (H4TSPP) . No crystal structure of (H4TSPP) exists, but inferences can be drawn from the crystal structure of a similar molecule diacid α,β,γ,δ- tetraphenylporphine. The crystal structure of diacid α,β,γ,δ-tetraphenylporphine shows a saddled macrocycle with two nitrogens pointing up and two pointing down.2 The saddling was attributed to a combination of steric hindrance of the four hydrogens and electrostatic repulsion.2 Around pH 1, two additional protons are added to two of the

62 sulfonate groups yielding a highly zwitterionic, neutral species, H2(H4TSPP). The two forms of α,β,γ,δ-tetrasulfonatophenylporphine (free base and diacid) are shown in Figure

1.2.3 1:

Figure 1.2.3 1: Two forms of tetrasulfonatophenyporphine: free base and diacid.

It is thought that the combination of electrostatic attraction between peripheral negative sulfonate groups and +2 central regions, and the π-π interactions of adjacent porphyrins hold the aggregate together. The co-planar or staircase model is often used to describe

58,63-69 the aggregation of H2(H4TSPP). In the staircase model a negative sulfonato group interacts with the positive macrocycle forming a dimer as pictured in Figure 1.2.3 2.

More molecules can be added to the dimer to create aggregates which are very long. The models in Figure 1.2.3 2 were individually optimized in the gas phase and do not take in to account any geometry changes that may occur upon aggregation.

Figure 1.2.3 2: Structure of an H2(H4TSPP) dimer.

With an understanding of the H2(H4TSPP) dimer structure we now turn to a discussion of aggregate formation. Under acidic, aqueous conditions (usually pH <2), the

51,64,65,70-79 diacid species of the porphyrin, H2(H4TSPP), forms aggregates. Formation of these aggregates can also be induced by other cationic species including alkali metal ions,51,72,80,84 surfactants,85,86 ionic liquids,87 and cationic porphyrins.88 The aggregation

63,66,72,81, 89-92 of H2(H4TSPP) is signaled by changes in UV-visible spectra. Upon aggregation of H2(H4TSPP) the monomeric absorbance bands at 434, 595, and 644 nm decrease in favor of new transitions at 424, 490, and 707 nm. The 490 nm band of the aggregate has been assigned to a Frenkel exciton transition, and its sharpness has been interpreted in terms of motional narrowing, which averages the local inhomogeneities.93

In the absence of disorder, the line width is inversely proportional to N1/2, where N is the number of coherently coupled chromophores (or coherence number), which has been

94 estimated to be in the range 11-120 for the H2(H4TSPP) aggregate. Aggregation of

H2(H4TSPP) also results in perturbations to the bands at about 600-700 nm, which are red shifted and enhanced in intensity relative to the monomer.93,95-97

Kinetic studies of H2(H4TSPP) aggregate formation have been previously reported. One of the experiments involved mixing a porphyrin solution with acid followed by monitoring the absorbance at 490 nm.89 An initial induction period followed by a rapid increase in the aggregate absorbance at 490 nm was observed. The induction period was attributed to a rate-limiting nucleation step followed by aggregate growth. A similar study was carried out using NaCl to start the aggregation process.81 This study also reported an induction period followed by a rapid increase in the 490 nm aggregate absorbance. Concurrent with the 490 nm data trends in absorbance at 434 and 413 nm were also monitored. The monomer band at 434 nm began decreasing slowly during the induction period followed by a more rapid drop afterward consistent with a mechanism where nucleation is rate-limiting. The absorbance at 413 nm increased marginally during the induction period and began decreasing at the same time as the monomer band. The

413 nm absorbance was attributed to the formation of an intermediate species. The authors did not speculate on the identity of the intermediate.

UV-visible spectroscopy is capable of probing the kinetics of aggregation and changes in electronic structure, but not the number of molecules or size of an aggregate in solution. To answer these questions we turn to RLS, Dynamic Light Scattering (DLS), and Small Angle X-ray Scattering (SAXS). RLS is a useful spectroscopic technique for studying aggregates because it is sensitive to particle size. Light scattering increases with the square of the volume of the scatterer98 so large aggregates have strong RLS signals.

The size selectivity is very helpful when studying UV-visible spectra of solutions with monomer/aggregate equilibrium as only the aggregate peaks will scatter.99 RLS studies

17 of H2(H4TSPP) aggregates consistently show a strong scattering signal at 490 nm indicating the presence of a large aggregate in solution.58,99-101 This technique has been employed to estimate the aggregation number of H2(H4TSPP) as being quite large, on the

5 100 order of 10 . DLS studies of aggregates estimate the size of the H2(H4TSPP) aggregates as ranging from 0.6 to 1.5 µm wide.69,70 SAXS data is consistent with the large aggregate sizes indicated by RLS and DLS. SAXS scattering profiles are fit by a hollow cylinder of radius 7.0 nm, wall thickness of 2.1 nm, and length 350 nm.71 A region of “an impressive higher electron density value relative to the solvent” was assigned as the shell of the aggregate tube. The scattering signal from the interior of the tube was similar to, but not equal to the solvent. The authors did not speculate as to the makeup of the tube‟s interior.

Raman studies of both monomeric and aggregated H2(H4TSPP) have been carried out at different excitations. Excitation wavelengths of 488 nm64,72,80,102 and 413 nm102 have been employed to study aggregated H2(H4TSPP). The 488 and 413 nm laser lines were used because they fall within the excitation envelopes of the 490 and 424 nm aggregate transitions. Raman spectra at non-resonant wavelengths of 457.9 nm72,102 and

80 514.0 nm have also been reported. The Raman spectrum of aggregated H2(H4TSPP) excited at 488 nm is dominated by two low frequency modes at 242 and 316 cm-

1.64,72,80,102 The 242 and 316 cm-1 bands have been assigned as out of plane saddling and an in plane porphyrin breathing modes respectivly.102 These two modes are intimately connected with the 490 nm transition of the aggregate. Raman spectra excited at wavelengths other than 488 nm have much weaker low frequency modes.64,72,80,102

Enhanced intensity of low frequency modes has been reported for aggregates of H2TCPP

59,64 as well. Raman spectra of aggregated H2(H4TSPP) excited at 413 nm were not subjected to extensive interpretation. The author only went as far as to say that the species responsible for the 424 nm band is the same as the species responsible for the 490 nm band because the positions of the bands in the Raman spectra are very similar.102

Raman spectra of monomeric H2(H4TSPP) excited at 457.9 nm show vibrational modes of similar energies to the aggregate. Some shifts of ~ 6-10 cm-1 in macrocycle modes are observed which were attributed to increased planarity of the macrocycle upon

102 aggregation. The interpretation of H2(H4TSPP) Raman spectra were carried out in terms of the staircase model of aggregation and tended to attribute changes in the position of vibrational bands in terms of changes in the shape of the molecules upon aggregation.

Aggregates of H2(H4TSPP) exhibit interesting optical properties upon interaction with polarized light. If a solution containing H2(H4TSPP) aggregates is passed through a narrow (0.5 mm) tube the tubular aggregates will preferentially align themselves with their long axes parallel to the flow direction.72 Absorption studies of plane-polarized light oriented parallel and perpendicular to the direction of flow (and the long axis of the aggregate) indicate that the aggregate visible absorption bands are polarized in different directions. The bands at 490 nm and 707 nm are polarized down the long axis of the aggregate while the band at 424 nm is polarized perpendicular to the aggregate long axis.72 Circular dichroic spectra show a very curious interaction with polarized light.

Absorption studies of circularly polarized light can be used to determine the chirality of species in solution. Circular dichroism studies show that aggregates of H2(H4TSPP) are

19 chiral and that different isomers form based upon the preparation of the aggregate solution. If a solution of achiral H2(H4TSPP) is aggregated in the presence of chiral L- lysine circular dichroism spectra show that chiral aggregates form.101 The same effect is seen when L-tartaric acid is used.72 Interestingly, stirring the solution during aggregation also produces chiral aggregates.72 Aggregates formed while the solution was stirred clockwise are different isomers than aggregates formed during counterclockwise stirring.

No stirring yielded inconsistent spectra. These two factors; the polarization of the absorbance bands and aggregate chirality are helpful in evaluating proposed aggregation models.

A great deal of effort has been put forth over the past few years to obtain detailed structural information about the supramolecular assemblies of H2(H4TSPP) molecules.

Studies employing fluorescence microscopy,103 electron microscopy,62 atomic force microscopy (AFM),51,58,73,75,101,104-107 and scanning tunneling microscopy (STM)73 provide useful but varied information about the geometrical sizes of the aggregates but could not resolve their structure at the molecular level. Since the molecular level arrangement of monomers in an aggregate influence the electronic and optical properties of the aggregate an understanding of the interplay between structure, electronic properties, and optical properties is critical to the study of functional nanomaterials.

Complementary to the SAXS, RLS, and DLS studies is the cryo-electron

62 microscopy data from Vlaming et al. An aliquot of H2(H4TSPP) aggregate solution was frozen and scanned by electron microscopy. The micrographs show tubes several hundred nanometers long with walls less than 2.5 nm thick and radii of 9 nm. This is an

20 important study because it shows conclusively that the aggregates are hollow tubes, and any aggregate model must account for this data. The model built by the author to explain the tubular nature of the aggregate involves a modification of the staircase model into a sheet followed by rolling into a tube.

Figure 1.2.3 3: Proposed model of the structure of the H2(H4TSPP) aggregate based on cryo-electron microscopy from reference 62. Primed quantities refer to the tilting of individual porphyrins. The sheet is rolled about the C vector to make the tube.

This aggregate model predicts two absorption bands, one polarized perpendicular to the long axis of the aggregate and one polarized parallel to the long axis which were confirmed by linear dichroism spectra.62,72 The C, or rolling, vector in this model is predicted to be a chiral vector, which accounts for reported aggregate chirality.72,101 A weakness of this model is its density. There are four hydrogens associated with each porphyrin macrocyle which could be sterically unfavorable.

Fluorescence microscopy studies of H2(H4TSPP) in solution by Kitahama et al. report long, thin aggregates ranging in length from 4 – 20 µm along the long axis; the short axis was not discussed.103 The aggregates are susceptible to magnetic fields; application of a 10 T field caused the aggregates to orient themselves parallel to the magnetic field. Polarized absorption spectra of the aggregate in the field showed that the

490 and 707 nm absorbance bands are oriented along the long axis of the aggregate, while the 424 nm band is polarized perpendicular to the long axis. Based upon the fluorescence microscopy, polarized absorption spectra, and magnetic susceptibility data the author proposed the following models for the aggregate‟s structure:

Figure 1.2.3 4: Proposed models of the structure of the H2(H4TSPP) aggregate from reference 103. B is the direction of the magnetic field. The small arrows are perpendicular to the porphyrin macrocycle.

Model A was discarded because a simple stack of linear arrays would not align in the magnetic field and it does not account for the tubular geometry of the aggregate. Model

D was also discarded because rings perpendicular to the long axis would exhibit a 490 nm absorbance (J-band) polarized along the short axis in conflict with the data. Models

B and C were deemed the most likely of the four because they are consistent with the polarized absorption data, magnetic field alignment, and length of the aggregate. The issues with models B and C is that they do not account for observed aggregate chirality and the polarization of the 424 nm band is not accounted for.

With these two solution models in hand we now turn to solid phase microscopy studies of H2(H4TSPP) aggregates. Quality AFM and STM data of H2(H4TSPP) aggregates is scarce. AFM images have been reported for H2(H4TSPP) aggregates deposited on a number of substrates: mica51,73, glass51,67,73,105, HOPG51,58,73, polystyrene51, and silicon.73,105 In accordance with the cryo-electron microscopy and fluorescence microscopy data AFM and STM images show that the H2(H4TSPP) aggregates are rod- like. Due to their shape the aggregates are often referred to as “nanorods” or “nanotubes”.

Rod widths range from 20-50 nm, heights 2-4 nm, and lengths range up to several microns. Several of the dimensions reported in the literature are summarized in Table

1.2.3 1.

Rod Rod Aggregation Width Height Method Substrate Technique (nm) (nm) Ref.

mica/glass/polystyrene

HCl/NaCl /HOPG AFM up to 40 3.8 51

HCl/dendrimer template HOPG AFM 20-50 1.5-5 58

HCl Glass AFM 30 4 67

HCl HOPG/glass/mica AFM 48-56 2-2.5 73

HCl HOPG STM 40 1.5* 73

HCl Si/glass AFM 40 4.5 105

Table 1.2.3 1: Reported dimensions of H2(H4TSPP) nanorods prepared under different conditions. (* Apparent height)

The lengths of H2(H4TSPP) nanorods reported in solution phase studies and on rods deposited on substrate line up well. The heights and widths are a different matter.

Solid phase SPM data gives heights and widths shorter and wider, respectively than

Vlaming‟s solution data. The differing dimensions in solution and on surfaces can be explained by assuming that the tubes collapse either during deposition or desiccation.

With the collapsed tube assumption and using 35 nm for the width of a desiccated rod

(the middle of the 20-50 nm width range reported by SPM) we estimate a tube radius of ~

11 nm prior to collapse, well in line with Vlaming‟s 9 nm radius. The walls of

Vlaming‟s tubes were reported to be less than 2.5 nm; therefore a collapsed tube would be less than 5 nm tall; again this is consistent with the SPM data. Interestingly if the top

25 half of a collapsed tube was removed a structure ~2 nm tall would result which is close to the shorter heights reported by SPM.

Intimately coupled with the structure of the nanorods are the electronic properties of the aggregate. Porphyrin materials in general are known to be photoconductors108 as well as photovoltaics109 and capable of light-induced charging.110 There has been little reported in terms of the conducting properties of their self-assembled nanostructures.

Reduction potentials have been reported for monomeric and aggregated H2(H4TSPP) vs.

65 a normal hydrogen electrode. Cyclic voltammetry of monomeric H2(H4TSPP) in sodium acetate buffer shows a reversible reduction peak at E1/2 = -0.29 V. Aggregated

H2(H4TSPP) in sodium acetate buffer showed a reversible reduction peak at E1/2 = -

0.22V. When deposited on indium tin oxide glass under acetonitrile the aggregate exhibited two reductions; E1/2 = 0.6 V (reversible) and E1/2 = 0.3 V (irreversible).

Schwab and co-workers49,111 have reported on the photoconductivity of nanorods formed from highly acidic solutions of H2(H4TSPP). They found that the photoconductivity grows over hundreds of seconds upon light exposure and decays slowly when the light is turned off. A qualitative model was proposed where conduction occurred through the

LUMOs of the molecules. They also reported that the rods were insulating over the voltage range studied (∓0.5 V) in the dark. The same group reported similar current voltage behavior in nanorods formed from meso-Tri(4-sulfonatophenyl)

112 monophenylporphine (this molecule is the same as H2(H4TSPP) minus one sulfonato group). In contrast Otsuka et al. reported ohmic behavior for H2(H4TSPP) in the absence of illumination.113 It should be noted that the device fabrication and nanorod deposition

26 employed by the two groups was very different. Schwab deposited nanorods on the finished device while Otsuka‟s nanorod deposition was done during device fabrication.

114 Rubires et al. have investigated the effects of water vapor pressure (PH2O) on

H2(H4TSPP) aggregates deposited on a comb-shape microelectrode. Over a bias range of

±100 mV the aggregates were insulating if PH2O < 8 Torr. Above PH2O = 8 Torr a linear

I(V) response was seen, the magnitude of which increased with increasing water vapor pressure. The group reported high levels of hysteresis in their measurements which were attributed to capacitance in the aggregate.

The H2(H4TSPP) nanotube system is interesting and important because the aggregate exhibits properties which are either different from or lacking in the monomer.

For example the nanotubes produce RLS signals while the monomer does not, the tubes are chiral unlike the monomer; the tubes have different electronic properties than the monomer, etc. The potential of this aggregate system is quite broad, both as a synthetic analog of bacterial light harvesting complexes and as a model with which we can study the fundamentals of self-assembled supramolecular aggregation. Our study of the

H2(H4TSPP) aggregation system has nine goals.

1. To measure solution UV-visible and RLS spectra of monomeric H2(H4TSPP) and it‟s

aggregate to verify the aggregation of H2(H4TSPP) and to parse the electronic

spectrum into monomeric and aggregate components. We will also deposit the

aggregate on glass slides in order to investigate the effects of deposition and

desiccation on the tubes.

2. To conduct AFM studies on H2(H4TSPP) nanotubes deposited on substrates to

characterize the dimensions of the tubes.

3. To conduct STM studies on both monomeric H2(H4TSPP) and it‟s aggregate in order

to test the validity of the proposed aggregation models.

4. To carry out Raman studies of the H2(H4TSPP) monomer and nanotubes in solution

to investigate the effects of aggregation on the vibrational structure of the system.

We will also report Raman spectra of nanotubes deposited on Au(111) and compare

with solution data.

5. To conduct compositional testing on the deposited monomer and aggregate by XPS.

6. To conduct the first reported STM I(V) study of both the monomer and aggregate to

investigate the effects of aggregation on the electronic structure of the system. Save

Otsuka et al. the I(V) studies covered in the review were all constrained to bias

voltages less than one volt. We will report on the I(V) characteristics of monomeric

H2(H4TSPP) and it‟s aggregate over a range of ±2 V.

7. To deposit the nanotubes on an interdigitated electrode and measure the conductivity

of the aggregates across the long axis of the tubes over a range of ±2 V.

8. To conduct the first reported HIM studies on H2(H4TSPP) nanotubes in order to test

the validity of the proposed aggregation models.

9. To conduct TEM studies H2(H4TSPP) nanotubes in order to test the validity of the

proposed aggregation models.

Each of these nine objectives will further our understanding of H2(H4TSPP) aggregates specifically and our understanding of the fundamentals of supramolecular aggregation in

28 general. In particular we will report the first ever imaging and I(V) study at the single molecule/aggregate level.

1.3 The Electronic Structure of Porphyrins and Changes upon Aggregation:

1.3.1 The Electronic Structure of Porphyrins

The most basic description of the electronic structure of porphine diacid is to approximate the molecule as a particle in a ring problem. This model yields wavefunctions and energy levels similar to the 2-dimensional rigid rotor.115 Figure 1.3.1

1 shows a diacid porphyrin molecule with an overlayed ring of radius 0.4 nm. This radius predicts a HOMO-LUMO transition at 580 nm.115

Figure 1.3.1 1: Schematic of Porphine diacid with overlayed ring and the energy levels predicted by the particle in a ring model.

Hückel theory offers a better description of porphyrins and serves as the foundation of Gouterman‟s four orbital model. In short, Hückel theory describles π orbitals as linear combinations of p-orbitals. In the case of a D4h porphyrin the relevent

29 orbitals are degenerate LUMOs and a pair of nearly or accidentally degenerate

(depending on the specific porphyrin) HOMOs.116

Figure 1.3.1 2: Orbitals and symmetries which constitute the four orbital model.

A further step is required before using this model to understand porphyrin UV-visible spectra. If the excited configurations arising from this model are of the same symmetry

116 1 configuration interactions must be considered. The ground state is A1g, as both the a1u

2 and a2u orbitals are closed shell. Two excited configurations are possible: a1ua2u eg and

2 1 3 a1u a2ueg. The resulting terms in both cases are Eu and Eu. The electronic structure is shown in Figure 1.3.1 3.

Figure 1.3.1 3: States arising from configuration interactions in a porphryin.

A few predictions regarding electronic transitions can be made at this point. First, group

1 theory selection rules allow transitions from the ground state to both Eu excited states in

1 x and y polarization; transitions from the ground state to the upper Eu are called Soret or

1 B-bands, transitions to the lower Eu are named Q-bands. Spin selection rules forbid

3 promoting an electron from the ground state to either of the Eu states. The upcoming discussion of aggregate formation (vide infra) necessitates a few words on the wavefunctions arising from the configuration interaction. The ground state wavefunction

1 is the product of the a1u and a2u orbital wavefunctions while the Eu states are admixtures of excited configurations.

Figure 1.3.1 4: Wavefunctions associated with the electronic states of a D4h porphyrin.

The formation of dimers or larger aggregates will mix these states into new dimer or aggregate states. In the preceeding discussion we have considered the interactions of a limited number of orbitals. For more rigouous work many more orbitals would be included, but the four orbital model is valuable as a didactic tool.

The model of porphyrin electronic structure based on configuration interactions predicts weak or forbidden Q-bands when the a1u and a2u orbitals are close in energy due to cancellation of transition dipole moments.116 As can bee seen from Figure 1.3.1 5 porphyrins do exhibit Q-bands of moderate intensity; much of which arises from vibronic borrowing.117

Figure 1.3.1 5: UV-visible spectrum of H4(H4TSPP)Cl2 in HCl (pH 3.35).

Vibronic borrowing is a mixing of electronic and vibrational states which results from breakdown of the Born-Oppenheimer (BO) approximation. Within the BO approximation the electronic Schrodinger equation is solved for a fixed nuclear position; stated differently the electronic Hamiltonian is independent of normal modes. The breakdown of the BO approximation is the source of Q-band intensity in porphyrins. To explain vibronic borrowing we begin with a perturbed Hamiltonian:117

(1.3.1-1)

(1.3.1-2)

th H0 is the zeroth order Hamiltonian and ξr is the r normal mode. The perturbed

Hamiltonian allows vibronic states of the same symmetry to interact leading to intensity

33 borrowing. The operator transforms as the totally symmetric irreducible representation so the derivative and the normal mode must be of the same symmetry in order to interact.

The mixing of vibronic states yields wavefunctions of the form:

(1.3.1-3)

Capital letters denote electronic states, lower case letters denote vibrational states, and a superscripted zero indicates equilibrium geometry. K is the state to which the Q state is coupling. The sum runs over all normal coordinates and vibronic states. The degree to which Q interacts with other states is governed by a number of factors. If the energetic separation between the two states is large (high ΔE) they will not couple strongly. The operator H‟r transforms as the totally symmetric irreducible representation so the matrix

element will vanish if K and Qx belong to different irreducible representations. ξr can connect vibrational states which differ by one quantum, otherwise the integral will vanish and the states will not mix. With the wavefunction in hand we can discuss the intensity of the Q-bands.

(1.3.1-4)

Where E is the electric dipole moment operator. As was discussed earlier the transition

dipole moment for matrix element is zero due to cancelation. The terms in the summation show that Q-band gains intensity by mixing with electric dipole

1 allowed states of the same symmetry. Equation 1.3.1-4 predicts that the lower Eu state

1 (Q) will couple with the upper Eu state (Soret) and gain intensity.

1.3.2 Exciton Theory of Dimer and Aggregate Formation:

With an understanding of the electronic structure of a D4h porphyrin in hand we can discuss the formation of dimers and larger aggregates. The following treatment is taken from Kasha et al.13,118 We begin by considering the Hamiltonian and wavefunctions of a dimer. The ground state wavefunction, Hamiltonian, and ground state energy of a dimer formed from identical monomers u and v are:

(1.3.2-1)

(1.3.2-2)

(1.3.2-3)

Where Vuv is a van der Waals interaction term. Excited state wavefunctions are made from combinations of monomer ground and excited state wavefunctions. The dagger denotes an excited state.

(1.3.2-4) r and s are yet to be determined coefficients. The requisite matrix along with the relevant solutions (denoted α and β) are:

(1.3.2-5)

( (1.3.2-6)

(1.3.2-7)

( (1.3.2-8)

(1.3.2-9)

In this case r and s are unity and the excited state wavefunctions are equal admixtures of monomer states. The first two terms in the energy are the energies of the excited and ground states respectively, the third term, VvdW, is the aforementioned van der Waals interaction. The fourth term is the “exciton splitting term”. The states arising from monomer formation are split by 2V12, with one state higher in energy and one of lower energy than the monomer. At this point it is instructive to cast these dimer expressions in terms of the porphyrin states previously described. The electronic structure of a dimer formed from two of the previously described D4h porphyrins is shown in Figure 1.3.2 1.

Figure 1.3.2 1: Wavefunctions associated with the electronic states of a D4h porphyrin and its corresponding dimer.

We now turn to a discussion of UV-visible selection rules. Of the four possible transtions in the dimer none are degenerate with the monomer so we would expect the

UV-visible spectrum of the dimer to be different than the monomer. The selection rules can be determined from a semiclassical electrostatic vector model.13,118 The vector model uses the orientation of the individual molecule‟s transition dipole moments to evaluate selection rules. Three dimer cases will be considered parallel, in-line, and co planar transition dipole moments. In the following discussion molecules will be represented by ovals and their transition dipole moments by arrows. For a dimer there are two ways to orient the transition dipole moments; in phase and out of phase. In the case of parallel transition dipole moments the out of phase orientation is an attractive arrangement (of lower energy) while in phase is repulsive (and consequently higher in energy).

Figure 1.3.2 2: Energy level diagram of a dimer with parallel transition dipole moments with allowed (solid) and forbidden (dashed) transitions.

Selection rules for transitions result from the sum of the dipole vectors. In Figure 1.3.2 2 it can be seen that the lower energy state sums to zero while the higher energy in phase arrangement reinforces itself. In this case we would expect to see a new peak in the UV- visible spectrum to the blue of the monomer peak. This type of configuration is called an

H-dimer.

The situation is different for in line transition dipoles. In this case the in phase arrangement is lower in energy than the out of phase. Here we see the opposite effect on

UV-visible spectra; a new red shifted peak increases upon monomer formation. This case is called a J-dimer.

Figure 1.3.2 3: Energy level diagram of a dimer with in line transition dipole moments with allowed (solid) and forbidden (dashed) transitions.

The final arrangement we will consider, co-planar transition dipoles, is particularly relevant to aggregates of H2(H4TSPP). This configuration is more complex

38 than the previous cases because the selection rules are dependant on the angle between the monomers ( ) as defined in Figure 1.3.2 4.

Figure 1.3.2 4: Energy level diagram of a dimer with co-planar transition dipole moments with allowed (solid) and forbidden (dashed) transitions.

The two extremes of 0° and 90° are equivalent to the previously discussed parallel and in-line cases respectivly. In between these two extremes the magnitude and sign of

Vuv depends on the transition dipole moment ( ), the distance between the centers of the monomers ( ), unit vectors pointing in the direction of the molecular transition moment dipoles (uu and uv), and the angle between the monomers ( ). The expression for Vuv in this case is:

(1.3.2-10)

(1.3.2-11)

For θ‟s below 55° the state with in phase transition dipoles lies at a lower energy and a new red-shifted peak arises in the electronic spectrum. For θ‟s greater than 55° the level with in phase transition dipoles lies higher and a blue-shifted peak is seen in the electronic spectrum. The expression also states that the coupling between orbitals in a dimer is dependant on transition dipole moments. The states involved in strongly allowed transitions will couple robustly, a weaker transition less so, and a forbidden transtion not at all.

The discussion of exciton splitting so far has only included the interaction of two transition dipole moments at a time. Porphyrins like H4(H4TSPP)Cl2 have degenerate x and y transition dipole moments.59 Figure 1.3.2 5 shows the staircase dimer of

H2(H4TSPP) with the transition dipole moment vectors superimposed on the molecules.

Figure 1.3.2 5: Diagram of the H2(H4TSPP) dimer with superimposed transition dipole moments.

The model in Figure 1.3.2 5 predicts two new peaks in the UV-visible spectrum upon aggregation. The µx transition dipoles on the constituent monomers couple as in the co- planar example with a predicted red-shifted peak. The µy transition dipoles on the constituent monomers couple as in the parallel transition dipole moment example with a predicted blue-shifted peak. µx and µy are orthogonal so no coupling occurs for these two dipoles. The terms J-dimer or H-dimer do not describe the coupling of H2(H4TSPP) monomers well. The H2(H4TSPP) dimer geometery is such that both types of coupling are expected.

Chapter 2: Experimental Techniques

2.1 UV-visible and Resonance Light Scattering Spectroscopy

Ultraviolet-visible spectroscopy (UV-visible) and Resonance Light Scattering

(RLS) are valuable techniques for studying chromophores and their aggregation. UV- visible gives information on the electronic structure of a system, while RLS can be used to probe the size of the species responsible for a specific absorption peak. By using the two techniques in concert peaks in UV-visible spectra of aggregate/monomer mixtures can be assigned to either the monomer or aggregate.

UV-visible measures the absorption of ultraviolet and visible photons by molecules and atoms. Photons in this range have sufficient energy to promote electrons from occupied to unoccupied levels giving information on the electronic structure of molecules. Figure 2.1 1 shows the previously discussed electronic structure of a D4h porphyrin. Transitions from gerada to ungerada (and vice versa) are allowed (LaPorte selection rule). Singlet to triplet transitions are not allowed by spin selection rules so only two peaks corresponding to transitions from the ground state to the upper and lower

1 Eu states will be seen in the UV-visible spectrum of this molecule.

Figure 2.1 1: Electronic structure of a D4h porphyrin.

Aggregation induces changes in electronic structure as molecular orbitals merge to form aggregate orbitals. Provided the new states are of a different energy than the monomeric states and selection rules allow the transition, aggregation can be tracked by

UV-visible spectroscopy. Figure 2.1 2 shows a comparison of the electronic configurations of a monomer and its dimer. As aggregation proceeds the number of monomers will decrease and the intensity of the corresponding transition will fall.

Conversely more dimers will form and the intensity of the corresponding transition will increase. The same logic holds true for large aggregates such as H2(H4TSPP) nanorods.

Figure 2.1 2: Diagram of the electronic states of a D4h porphyrin and its dimer with overlayed electronic transitions to an excited electronic state.

RLS is a relatively new technique first described by Pasternack et. al in 1993.119

RLS measures the light scattered by a molecule within the envelope of an electronic transition. The quantum mechanical expression for scattering will be discussed in the section on the theory of Raman scattering; for now we will consider a more elementary picture. The scattering phenomenon can be expalained by a two photon process. First, an incident photon is absorbed promoting an electron to an excited electronic state. The molecule then returns to the ground state by emiting a photon of the same energy as the incident photon. This process is illustrated for a monomer and its dimer in Figure 2.1 3.

Figure 2.1 3: Diagram of the the electronic states of a D4h porphyrin and its dimer with an overlayed scattering process.

The interpritation of RLS signals can be discussed by employing two different frameworks. The scatterer can be viewed as a particle with a different index of refration than the solvent or as a manifold of polarizable molecules.119 Both treatments are important because they demonstrate the dependance of RLS on different factors. The expression for scattering cross section in terms of indicies of refraction demonstrates that particle size is a critical variable. An RLS signal is possible for any HOMO-LUMO transition, but are weak for monomers beacause RLS is sensitive to particle size.98 Both the absorption and scattering cross sections (Cabs and Cscat) are functions of particle size, but the scattering cross section increases much more rapidly with size than the absorption cross section. The expressions for the two quantities are:120

(2.1-1)

(2.1-2)

Where γ is the radius of the scatterer, nmed is the index of refraction of the medium, λ0 is the incident photon wavelength, and m is the complex refractive index. The absorbance

6 cross section increases with the cube of the scatterer while Cscat increses as γ . These two expressions show why RLS is so valuable for studying aggregation. Electronic transitions which belong to large particles like aggregates show significant scattering, while monomers will not. RLS data can be used to parse UV-visible spectra into aggregate peaks and monomer peaks.

The expression for RLS intensity which is derived by considering the aggregate as a set of polarizable molecules illustrates the technique‟s dependance on extinction coefficient:121

(2.1-3)

R is the scattering intensity, M is the concentration, N is Avogaro‟s number, λ is the wavelength of the incident light, P is the induced diplole moment, is frequency in wavenumbers, and ε is the extinction coefficient. The primes denote scattered light. This equation for scattering intensity shows that high extinction coefficients correspond to good scattering. The polarizability and refractive views of scattering can be reconciled by the use of the Kramers-Kronig transform.121 The purpose of considering both treatments is to emphasize that scattering intensity is a function of a both particle size and extinction coefficient.

As an example of the utility of RLS scattering data for the H2(H4TSPP) monomer and aggregate is presented in Figure 2.1 4. The monomer spectrum has a dip at ~ 430 nm corresponding to the Soret band of the monomer. For particles the size of monomeric

H2(H4TSPP) absorbance dominates over scattering. The opposite is true for the aggregate. The aggregate RLS spectrum shows a strong peak at ~ 490 nm corresponding to the aggregate Soret band. A dip in the spectrum is visible just to the blue of the scattering peak indicates that both scattering and absorption are occuring, but clearly scattering is the dominant process. The raw intensities of the two spectra are very different as well. The aggregate RLS spectrum is two orders of magnitude greater than the monomer spectrum. When graphed on the same axes the monomeric RLS spectrum looks almost flat.

Figure 2.1 4: RLS spectra of H2(H4TSPP) monomer (0.74 µM H2(H4TSPP) in 0.75 M

HCl, left) and aggregate (5.9 µM H2(H4TSPP) in 0.75 M HCl, right).

The last point concerns the Q-bands. Notice the peak at ~ 650 nm in the monomer RLS spectrum. This peak is fluorescence from the residual monomer Q-band not RLS. An important weakness of RLS is its inability to discern fluorescence from scattering. What is referred to as an RLS spectrum is really an emission spectrum. The fluorimeter is configured to detect photons emitted at 90° to the incident beam irrespective of their origins. Photons from RLS, fluorescence, and phosphorescence will all appear in the spectrum so careful interpritation of the spectra are required. For H2(H4TSPP) aggregates the spectra are relativly easy to interprite because the only non-RLS signal observed is monomer Q-band fluorescence.

2.2 X-ray and Ultraviolet Photoelectron Spectroscopy

Both X-ray Photoelectron Spectroscopy (XPS) and Ultraviolet Photoelectron

Spectroscopy (UPS) are powerful surface analysis techniques capable of probing the electronic structure of an analyte. XPS and UPS employ the photoelectric effect described by Einstein in 1905122 to analyze surfaces and/or adsorbate molecules. X-ray or ultraviolet photons incident on a sample will eject electrons from that sample with a range of kinetic energies because x-ray and ultraviolet photons impart sufficient energy for electrons to overcome nuclear attraction, called binding energy, and escape the sample. Electrons in the valence shells are held least tightly and are therefore easiest to remove. Because the mean escape depth of photoelectrons from a surface is ~ 1 nm in the X-ray region photoelectron spectroscopy only probes 6–10 nm into a sample,123 thus it is a surface sensitive technique. Ultraviolet Photoelectron Spectroscopy (UPS) is used to study valence shell electrons as ultraviolet photons (10-45 eV) have sufficient energy

48 to eject valence electrons. If x-ray (200-1500 eV) photons are employed electrons from core levels can be photoemitted. The process is illustrated in Figure 2.2 1.

Figure 2.2 1: Diagrams illustrating photoemission from the valence band (left) and core levels (right)

A photon with energy hν strikes a sample imparting its energy into the sample. If the energy of the incident photon is greater than the binding energy of the sample electrons will be emitted from the sample orbitals. The kinetic energy of the photoemitted electrons can be related to the binding energy by the relation:

(2.2-1)

Binding energy in is an important quantity because it is characteristic of both orbital and element; an electron ejected from a specific orbital in a specific element will have a specific binding energy.124 By reversing this logic peaks in a photoemission spectrum can be assigned to elements and orbitals based on the binding energy where they appear.

Photoemission peaks can be assigned to certain elements and the ratios of peak areas used

49 to calculate empirical formulas of deposited molecules. Photoemission spectra can also be used to gather information on the chemical environment of individual atoms in a sample. As an example consider the molecule ethyltrifluoroacetate:

Figure 2.2 2: XPS spectrum of ethyltrifluoroacetate carbon 1s spectrum illustrating the chemical shifts of the different carbons. image from reference125

The C1s signal in the XPS spectrum of this molecule is split into four separate peaks due to the differing environments of the carbons.126 Fluorine, with its high electronegativity strongly withdraws electrons from its attached carbon, making it more difficult to remove electrons from the fluorinated carbon. Stated differently, electrons in the fluorinated carbon will have the highest binding energy. Electrons with the next highest binding energy belong to the ester carbon as oxygen is not as electronegative as fluorine.

Hydrogen is a much less efficient electron withdrawing species than either fluorine or oxygen so the carbon not bound to fluorine or oxygen will have the lowest binding energy. This example illustrates how XPS can be used to gather information on the differing chemical environments of a particular element in a sample. The ratios of the differing carbons can also be deduced by comparing the peak areas. Not surprisingly in the case of ethyltrifluoroacetate the ratio is 1:1:1:1. Ethyltrifluoroacetate serves as an excellent example of the capabilities of XPS in elemental analysis and determining chemical environment.

UPS is analogous to XPS save that it probes the valence energy levels of the sample. Photons from a helium discharge lamp with an energy of 21.2 eV124 are used to irradiate the sample as opposed to X-rays. The ultraviolet photons eject valence electrons from the sample which are analyzed in a similar fashion to XPS. In addition to elucidating the energies of valence orbitals UPS can be used to find the workfunction of a sample. The 21.2 eV HeI photon impinging on the sample imparts its energy into the sample. In order for the electron to escape the sample and be measured by the detector it must have sufficient energy to overcome the sample‟s workfunction. The most deeply

51 buried electron photoemitted will have a binding energy of 21.2 eV – Φ. Anything buried deeper than this will not have sufficient energy to reach the vacuum level.

Figure 2.2 3: UPS Energy diagram of UPS illustrating the energy of the HeI photon and the work function (Left), and UPS spectrum of HOPG (Right).

Figure 2.2 3 shows the UPS spectrum of HOPG. From zero binding energy (electrons close to the Fermi level) to ~ 16.5 eV electrons are photoemitted, after this point photoemission shuts off abruptly. Below 16.5 eV electrons do not have sufficient energy to escape the sample. By subtracting the energy where photoemission ceases from the

21.2 eV of the incident photon the sample‟s workfunction can be calculated.

2.3 Scanning Tunneling Microscopy (STM)

STM can be explained solving variants of “particle in a box” problems. To solve a quantum mechanical problem the Hamiltonian and boundary conditions must be known. For a particle in a one dimensional box of length l with infinitly tall walls the

Hamilton is:

(2.3-1)

Where h is plank‟s constant, m is the mass, q is the coordinate, and is the wavefunction. In this case there is only one boundary condition; the wavefunction must vanish at the walls because there is zero probability of the particle being found outside the box. To find the energies, E, the following differential equation must be solved:

(2.3-2)

(2.3-3)

The solution to this differential equation is well known:

(2.3-4)

The energy levels of the particle are therefore:

(2.3-5)

The next step is a particle in a box with a finite wall of thickness x. The problem is more tractable if it is broken into three regions:

Figure 2.3 1: Diagram of a potential box with a finite barrier.

The Hamiltonians for regions 1 and 3 are the same as the previous problem. In region 2 a perturbation, V, equal to the height of the barrier is added. The particle is assumed to have E

Schrodinger equations to solve.

(2.3-6a)

(2.3-6b)

(2.3-6c)

As with the previous problem wavefunctions are constructed by considering boundary conditions. There are two criteria that the wavefunctions must satisfy. First, must vanish at the left and right sides of the box, second, the wavefunction and its first derivative must be continuous at the left and right side of the barrier. The solutions are:

(2.3-7a)

(2.3-7b)

(2.3-7c)

This problem exemplifies the difference between classical and quantum mechanics.

Classically there is zero probability of finding the particle inside or on the far side of the barrier. The quantum mechanical expression shows that the probability of finding the particle in the classically forbidden region is nonzero. The particle is said to be

“tunneling” through the barrier because it lacks the energy necessary to surmount it.

Figure 2.3 2: Probability distribution of a particle in a box with a finite barrier.

The trace of in Figure 2.3 2 shows that the probability of finding the particle on the far side of the barrier is much lower than in Region 1. If the energy of the particle remains the same after passing through the barrier it is called elastic tunneling.

This particle in a box model can be used to understand tunneling in the STM on a conceptual level. In the case of STM a sharp tip, commonly made of tungsten or a platinum/iridium blend, is brought close enough to a conducting sample for tunneling to

55 occur. In this example the tip and sample are the two “boxes” and the gap between is the thin barrier. For the following discussion we will assume that both the tip and sample are clean, in a vacuum, and at zero Kelvin. When the tip and sample are brought close to one another (on the order of a few nanometers) quantum tunneling becomes possible.

Possible is emphasized because the energy gain or loss during tunneling must be considered. In the absence of a bias voltage no current will flow because tunneling from the tip to the sample is energetically unfavorable. If an electron from the valence band of the tip elastically tunneled into the sample it would enter an occupied energy level which is forbidden. Tunneling can be made energetically favorable by the application of a bias voltage to one of the surfaces, typically to the sample in STM. The effect of the bias voltage is to shift the Fermi energies of the tip and sample relative to each other. If a negative voltage is applied to the sample (negative sample bias) the Fermi energy will rise relative to the tip. If a positive voltage is applied to the sample (positive sample bias) the Fermi energy will drop in relation to the tip. In the first case electrons tunnel from the tip to the sample; in the second electrons move from the sample to the tip.

Figure 2.3 3: Tip-sample energy diagram where A) no tunneling is allowed, B) tunneling from tip to sample is allowed, and C) tunneling from sample to tip is allowed.

During the STM scan the tip is rastered across the surface to probe the sample‟s properties. The microscope can be operated in two different modes: constant current and constant height. In constant current mode a feedback loop modulates the tip-sample separation to keep a constant tunneling current. In this mode the STM image is a map of the motion of the tip as it moves over the surface. One of the advantages of this method is the chance of crashing the tip is minimized, the disadvantage is slow scan speed.

Constant height mode fixes the tip at a particular tip-sample separation during scanning.

The resulting image is a map of tunneling current at each location. The advantage of this mode is greater scan speed than constant current; the weakness is potentially crashing the tip while scanning over rough surfaces.

Figure 2.3 4: Diagrams illustrating the two different modes of STM operation; constant current and constant height.

The STM experiment can be handled in a more rigorous way by examining the relevant tunneling equations. Within the Wentzel-Kramers-Brillouin (WKB) approximation tunneling current, I, can be expressed as:127

∓ (2.3-8)

Where ρs(r,E) is the density of states of the sample, ρt(r,E) is the density of states of the tip, r is the position of the tip, E is energy with respect to the Fermi level, eV is the bias

58 voltage, and T(E,eV,r) is the transmission function. When both plus and minus signs are shown the upper operator corresponds to positive bias and vice versa. If we restrict ourselves to low bias voltages and trapezoidal barriers Equation 2.3-8 reduces to:

(2.3-9)

Within this approximation tunneling current is dependant on five variables: the density of states of the tip, the density of states of the sample, the bias voltage, the work function of the sample, and the tip-sample separation (A is a constant). The potential effect of ρs on

STM images is illustrated in Figure 2.3 5. We begin by assuming the sample is biased negatively with respect to the tip, and the STM is operating in constant current mode. As the tip moves over the lower surface it follows a contour which mirrors the geometry of the surface. The situation is much different as the tip crosses over the adatoms (red and blue). At Vb the red atom has a high density of states so the tip pulls back to maintain constant current making the atom appear taller than its true height (positive contrast). At

Vb the blue atom has a low density of states so the tip drives in to maintain constant current making the atom appear shorter than its true height (negative contrast). Because of its dependence on density of states STM images are maps of conductivity and may not represent the true surface geometry.

Figure 2.3 5: Diagram illustrating the effect of differing conductivites on the path of an

STM tip.

The dependence of tunneling current on opens up the possibility of performing spectroscopy with the STM. As an example consider a semiconductor; the valence and conduction bands are separated by a band gap with zero density of states. By monitoring the tunneling current as a function of bias voltage the electronic structure of the material can be mapped. At sufficiently negative biases electrons from the conduction band will tunnel into the tip. As the bias becomes more positive the current will shut off as the top of the conduction band passes the Fermi energy of the tip. No current will flow until the

Fermi energy of the tip reaches the bottom of the conduction band where states are available for occupation by the tip‟s electrons. Once this point is reached tunneling current will begin increasing.

The next step in the discussion of STM spectroscopy is a sample composed of a monolayer of molecules on a conducting substrate. In the following discussion the energy levels of the adsorbed molecules are assumed to be pinned relative to the Fermi energy of the substrate. Figure 2.3 6 shows the energy levels of a generic molecule adsorbed on a metal surface. Under a poisitive sample bias electrons from the tip‟s

60 conduction band can tunnel into the adsorbed molecule‟s unfilled orbitals and then drain into the substrate. Electrons from the filled orbitals can not tunnel into the tip because the conduction band is occupied. A negative sample bias raises the sample and substrate‟s energy levels allowing electrons from the adsorbate‟s filled orbitals to enter the tip‟s conduction band. By modulating the bias voltage STM can be used to sudy both the filled and empty states of an adsorbed molecule.

Figure 2.3 6: Tip-sample energy diagrams with the states of an adsorbed molecule under a positive sample bias (A) and negative sample bias (B). The arrow indicates the direction of electron flow.

The clearest way to extract electronic structure data from current vs. voltage data is to look at the derivative of equation 2.3-8 with respect to voltage which can be approximated as:128

(2.3-12)

Tunneling current increases with increasing bias voltage because more and more electrons have enough energy to tunnel into unfilled states. Since tunneling is most

61 probable for electrons near the Fermi level there will be an increase in the tunneling current as the Fermi level of the tip comes in to resonance with one of the adsorbate‟s states. As the Fermi energy of the tip moves to higher potentials the tunneling channel through the adsorbate‟s state does not close, and current still flows through the adsorbate.

Because the tunneling channel does not close off resonance with an adsorbate energy level is indicated by a change in the slope of the current vs. voltage curve. The density of states term for the tip contributes a constant factor to dI/dV due to its ohmic character.

The transmission coefficient term increases monotonically with V contributing a rising background on which the orbital peaks are superimposed.

Having discussed the dependence of tunneling current on density of states we now turn to a discussion of bias voltage and tip-sample separation. Tunneling increases linearly with bias voltage and drops off exponentially with tip sample separation. The choice of bias voltage and tunneling current are important experimental parameters. For example if the bias voltage is set in the band gap of a semiconductor no current will flow and the tip will crash. Conversely, if the bias is set in either the valence or conduction bands electrons will tunnel. The exponential dependence on tunneling current is particularly germane to STM studies of H2(H4TSPP) nanorods. The nanorods are ~ 4 nm tall and several may be stacked on top of each other. The risk of crashing the tip into a rod can be reduced by operating at a low setpoint (1 pA); at the cost of signal strength.

STM experiments can be carried out under ambient conditions or in ultrahigh vacuum (UHV). Both techniques have their strengths and weaknesses. For example, ambient STM‟s are capable of atomic resolution, are less expensive to purchase and

62 maintain than UHV models, and can scan samples immersed in liquids. The drawbacks of ambient STM include higher noise levels, sensitivity to thermal drift, and oxide buildup on the tip. One of the great advantages of UHV STM is the ability to control the environment where the experiment is performed. For example, UHV STM‟s are capable of imaging at 0.3 K,129 and images taken at 580 K have been reported.130 These temperatures are well out of the range of ambient STM. Attached prep chambers and sputtering guns allow samples to be prepared without exposure to the atmosphere. The controlled environment of UHV-STM also meets the experimental conditions of clean tips and samples required for spectroscopy.129 Tips and substrates can be repeatedly sputtered and annealed to remove oxides and other contaminants. Substrates with analyte molecules deposited on them can be heated in the vacuum to remove contaminants or solvent. Both ambient and UHV-STM are valuable surface analysis tools. Ambient models offer ease of use and are capable of scanning in liquids, while UHV-STM‟s offer greater temperature ranges, stability, and spectroscopic capabilities.

2.4 Atomic Force Microscopy (AFM)

AFM is a surface analysis complimentary to STM, and can be explained by analogy to constant current STM. In both techniques a probe is brought close to a surface and rastered across it. A signal from the probed is monitored by the feedback loop which makes adjustments to the probe‟s position. The techniques differ in the probe and the signal. STM monitors tunneling current; AFM measures the deflection of a cantilever.

Unlike STM the AFM cantilever is brought into physical contact with the sample during scanning. A block diagram of an AFM is presented in Figure 2.4 1. A laser is bounced

63 off the end of the cantilever on to a position sensitive photodetector. As the cantilever moves over the surface it flexes with the contours of the sample. The motion of the cantilever moves the laser spot on the photodetector, the position of the laser on the detector is recorded and used to generate topography.

Figure 2.4 1: Diagram of the principle components of the AFM.

The mode of AFM described above is characteristic of the first AFM‟s and is

131 known as contact mode. This method is not well suited to the study of H2(H4TSPP) aggregates due to the lateral force the cantilever exerts on the sample. Stated differently, the tip may scrape the rods off the surface. The solution is to use dynamic or tapping mode AFM. In this configuration the cantilever drive at one of its resonance frequencies and a set amplitude.131 The vibrating cantilever is scanned across the surface, as the tip moves it intermittently touches the surface. The motion of the vibrating cantilever is recorded on a photodiode just like contact mode. This mode of AFM is depicted in

Figure 2.4 2.

Figure 2.4 2: Diagram depicting tapping mode AFM.

The AFM is a valuable surface analysis tool, because it measures the deflection of a cantilever it can be used on nonconducting samples ulike STM. The AFM is also better suited for taking large-scale pictures (10-50 µm) which makes it apropos for optimizing nanorod deposition. Two types of tapping mode cantilevers were used: Nanosensors

Pointprobe (29-79 N/m spring constant, 290-400 kHz, tip radius of curvature < 10 nm) and Vistaprobe T300R-25 (20-75 N/m spring constant, 200-400 kHz, tip radius of curvature < 10 nm)

2.5 Raman Spectroscopy:

Raman spectroscopy uses scattered photons to probe the vibrational character of molecules. It provides similar information to infrared spectroscopy (IR), but with different selection rules. The acivity of a vibrational mode in an IR spectrum is based on the change in a molecule‟s dipole moment as a function of displacement along a normal

65 coordinate. As a consequence homonuclear diatomic molecues such as O2 are inactive in

IR spectroscopy. Raman selection rules are based on the polarizability of a molecule‟s electron cloud. IR and Raman are complimentary techniques. A vibrational mode which is active in one technique may or may not be active in the other because of the differing selection rules.

A two photon process is often used to explain scattering. In the first step a photon strikes a molecule which is excited to a “virtual state”. From the virtual state the molecule relaxes to some final state and emits a scattered photon. If the molecule returns to its initial state (elastic scattering) the emitted photon will have the same wavelength as the incident photon. This process is the previoulsy discussed RLS and is the most probable result. If the target molecule relaxes to a vibrationally excited state the scattered photon will have a wavelegth to the red of the incident photon. The shift in wavelenght will be equal to the spacing between the intial and final vibrational energy levels. This case is called Stokes scattering. The final possibility is that an molecule in an excited vibrational state is excited to the virtual state and decays to the ground vibrational state.

This process is called anti-Stokes scattering, and results in a scattered photon with a blue shifted wavelength. Anti-Stokes scattering requires a populated excited state, hence it is the least probable type of scattering as most excited vibrational levels are not significanly thermally populated.

Figure 2.5 1: Diagram of possible scattering events.

A more complete description comes from the quantum mechanical expression for

Raman scattering. Raman selection rules are based on polarizability and the polarizability of a molecule can be expressed as a second rank tensor in the form:

(2.5-1)

Raman scattering is possible for all nine elements of the tensor so each must be considered. The quantum mechanical expression for the nine elements of Raman scattering, also called the Kramers-Heisenberg-Dirac or KHD expression is presented here without derivation:

(2.5-2)

µ is the transition dipole moment operator, i and f are initial and final states, n is an intermediate state, ω is the frequency of the incident photon, ωni is the frequency of the transition between the initial and intermediate state, ωnf is the frequency of the transition from the intermediate state to the final state, and iΓn is a damping term. The virtual state

67 in the simplistic explaination is really a sum over number of different electronic states. In the special case of Resonance Raman the energy of the incidient photon is coincident with an electronic transition and the denomenator of the first term becomes small due to the ω - ωni term. In this case normal modes which correspond to the geometry of the resonant excited state are enhanced in intensity.

Raman spectroscopy is a powerfull tool for determining the degeneracy of unoccupied states. We begin by explaining the impact of the polarization of incident an scattered light in Raman. Figure 2.5 2 shows the Raman setup used in this work. S polarized ligh has its electric field vector pointing out of the page, P polarized light‟s electric field vector is in the plane of the page. The half wave plate is used to rotate the polarization of the incident laser beam. The polarization rotator serves the same function as the half wave plate, save that it rotates the polarization of the scattered light as opposed to incident photons. There are four possible polarization combinations SS, SP,

PS, and PP. The first letter indicates the polarization of the incident beam, the second refers to polarization of the scattered beam which is passed through the polarization rotator. The depolarization ratio, ρ, is defined as the ratio of the scattered light detected in S and P polarization. For example, the intensity of a mode detected in SP divided by the intensity of the same mode detected in SS polarization. The value of the depolarization ratio indicates the degeneracy of the excited state. ρ = 1/3 indicates a singly degenerate state, ρ = 1/8 arises from a doubly degenerate state.

Figure 2.5 2: Diagram the Raman experiment and definitions of S and P polarization.

The case of Raman spectroscopy of molecules deposited on a surface requires additional discussion. Image charges can be used to discern selection rules in surface

Raman. The concept of image charges is illustrated in Figure 2.5 3. As a polar diatomic molecule approaches a surface, the molecule induces an equal and opposite image charge in the substrate. The orientation of the molecule on the surface is an important factor in surface Raman selection rules. If the molecule lies flat on the surface the transition dipole moment vanishes, while the transition dipole moment of a molecule perpendicular to the surface is reinforced. Consequently, vibrations perpendicular to the surface are augmented, those parallel, diminished.132

Figure 2.5 3: Diagram of an adsorbed molecule and its image charges. The charges cancel in the left-hand case and reinforce in the right-hand arrangement.

The concept of augmented and diminished vibrational modes as a consequence of orientation relative to a surface can be expressed in the following equations (each are proportional to intensity of the Raman spectra):132

(2.5-3a)

2.5-3b) (

2.5-3c)

(2.5-3d)

The primed quantities correspond to scattered radiation, alphas are the nine components of the polarizability tensor, r is an optical constant called a Fresnel coefficient, and phi is

70 the angle of incidence or scattering as differentiated by the prime. If the substrate is a very good reflector the four equations reduce to:132

(2.5-4a)

(2.5-4b)

2.5-4c)

(2.5-5d)

In this limit only the PP orientation will remain. By measuring surface Raman spectra in the four configurations it is possible to gain information on the elements of the Raman tensor. When interpereting surface Raman spectra it is critical to remember that the equations presented for the intensities of the various polarized spectra are derived for thin films. As the thickness of the deposited molecules increases the distinctions between the four configurations decrease.133

2.6 Helium Ion Microscopy:

The helium ion microscope (HIM) can be understood by analogy to a scanning electron microscope (SEM). In both techniques a beam of probe particles is accelerated away from a tip and focused by a series of electrostatic lenses. The beam of probe particles is scanned over the sample via a set of scanning deflectors before passing through a final lens. Just like SEM the HIM is capapble of detecting both inelastically scattered secondary electrons and elastically backscattered probe particles. A schematic of a HIM is shown in Figure 2.6 1.

Figure 2.6 1: A schematic of the Helium Ion Microscope. image from reference 134

The helium ions used as probe particles are generated by the interaction of helium atoms with a sharp tip. The construction of the HIM tip is proprietary134 so an exhaustive discussion is not possible here. What is known is that the tip is shaped like a pyramid ending in a trimer. The advantage of this geometry is that only atoms close to the apex emit ions when exposed to He gas, narrowing the beam diameter.134 The tip is biased positively, creating He+ ions which are then accelerated away from the tip through the previously described lenses. A diagram of the process is shown in Figure 2.6 2.

Figure 2.6 2: A schematic of the Helium Ion Microscope Tip. image from reference 134

The HIM has a few advantages over the SEM. First, the secondary electron yield per particle is two to eight times higher in HIM than SEM. Higher secondary electron yield means better signal strength and clearer pictures. Second, the the use of an electron flood gun in tandem with the He+ beam eliminates charging so nonconducting samples can be imaged without sputtering and consequent loss of detail.135 Third, the smaller

DeBroglie wavelength of helium allows for better resolution. As of 2010 the ORION®

PLUS manufactured by Carl Zeiss has better resolution (0.24 nm136) than the best SEM on the market; the Hitachi S-5500 (0.4 nm137). Fourth, the energy of the inelastically scattered secondary electrons is much lower in HIM than SEM allowing for more surface detail in images.138

2.7 Transmission Electron Microscopy:

Transmission Electron Microscopy (TEM) is a microscopy technique based upon the scattering of electrons by atoms in a sample. For this discussion we will consider thin samples (defined as <20 nm139). Electrons are generated in a gun by Schottky, thermionic, or field emission. The gun is held at a negative potiential which accelerates

73 the electrons away from the gun and through a series of electronic lenses which focus the beam onto the sample. Acceleration voltages vary, ranging from 200-500 kV to as high as 3 MV for high resolution work.140 Image contrast is created by the scattering of electrons by the sample (this is called scattering contrast). For this discussion we will limit ourselves to bright field imaging which detects electrons passing through the specimen unscattered. Electron scattering by atoms in the sample can be described in detail by Coulombic interactions, for the purposes of this discussion a much simpler model will suffice. The loss of electrons by scattering events (T) can be described by the following empirical equation:139

(2.7-1)

Where Z is atomic number, A is the atomic weight , b and a are empirical constants, and x is the mass thickness defined as the sample density multiplied by the thickness of the sample. This equation illustrates several important aspects of TEM. Heavier atoms (high

Z and A) will scatter more than low Z materials leading to better scattering contrast.

Second, thicker samples will scatter more than thinner samples. Third, the electron loss due to scattering is logarithmic, so thin samples are required. After interaction with the sample the unscattered electrions are imaged by a photographic plate or a CCD camera attached to a fluorescent screen.140 TEM is capable of resolution superior to both SEM and HIM. The FEI Titan G2 60-300 TEM has a resolution of 80 pm.141

Chapter 3: Experimental Methods

3.1 Materials, Reagent, and Instrument List

The following reagents and materials were used in course of this project:

Reagent grade concentrated hydrochloric acid

Reagent grade concentrated nitric acid

Reagent grade 30% hydrogen peroxide (stabilized)

Reagent grade sodium hydroxide pellets

α,β,γ,δ-tetrasulfonatophenylporphine dihydrochloride (Frontier Scientific T1239) http://www.frontiersci.com/

Tetrasodium α,β,γ,δ-tetrasulfonatophenylporphine dodecahydrate (Alfa Aesar 30538) http://www.alfa.com/

Chloroauric acid trihydrate (HAuCl4•3H2O Aldrich 520918-G) http://www.sigmaaldrich.com/sigma-aldrich/home.html

99.999% pure gold splatters (Cerac G-1065) http://www.cerac.com/

1x4 cm mica sheets (Ted Pella #54) http://www.tedpella.com/

9.9 cm diameter mica disks (Ted Pella #50) http://www.tedpella.com/

GridStick TM adhesive (Ted Pella #155-9) http://www.tedpella.com/

Highly ordered pyrolytic graphite 5x5x1, 7x7x1, 10x10x1 mm SPI-2 grade (SPI supplies

479HP-AB, 480HP-AB, 436HP-AB) http://www.2spi.com/

Tungsten wire 0.25 mm diameter 99.95% pure (Alfa Aesar 10408) http://www.alfa.com/

Platinum0.8Iridum0.2 wire 0.25 mm diameter (California Fine Wire Company 100168) http://www.calfinewire.com/

AFM tapping mode cantilevers 29-79 N/m 290-400 kHz (Nanosensors Pointprobe) http://www.nanosensors.com/

AFM tapping mode cantilevers 20-75 N/m 200-400 kHz (Vistaprobe T300R-25) http://www.vistaprobes.com/vp/

UV-visible cuvettes 1cm path length 170-2200 nm range (Spectrocell R-3010-T) http://www.spectrocell.com/Spectrocel-old/index.htm

Fluorescence cell 1 cm path length (Precision Cells Inc. 3G10) http://www.precisioncells.com/

The following instruments were used in the course of this project:

Ultraviolet-Visible Absorption Spectroscopy:

Perkin-Elmer 330 UV-visible spectrophotometer (www.perkinelmer.com)

Shimadzu UV-2501PC UV-visible spectrophotometer (www.shimadzu.com)

Resonance Light Scattering Spectroscopy:

PTI Quanta Master Fluorimeter (www.pti-nj.com)

Raman Spectroscopy:

Spex 14018 Spectrometer outfitted with a double monochromator and thermoelectrically cooled photomultiplier tube detection.

Lexel Model 95 Ar ion laser (www.lexellaser.com)

Spectra Physics Beamlock 2060 Kr ion laser (www.newport.com)

Atomic Force Microscopy:

Digital Instruments Nanoscope II AFM (www.veeco.com)

Molecular Imaging Pico Plus combination AFM/STM (www.agilent.com)

Ambient Scanning Tunneling Microscopy:

Molecular Imaging Pico 5 STM (Equipped with a 1 µm STM head) (www.agilent.com)

Molecular Imaging Pico Plus combination AFM/STM (Equipped with a 1 µm STM head with a low current preamp) (www.agilent.com)

Ultra-High Vacuum Scanning Tunneling Microscopy:

Model UHV300 Scanning Tunneling Microscope and control electronics (model

SPM100) from RHK Technology (www.RHK-tech.com)

Helium Ion Microscopy:

Carl Zeiss ORION® PLUS Helium Ion Microscope (www.zeiss.com)

Transmission Electron Microscopy:

Philips CM-200 Transmission Electron Microscope (www.fei.com)

Scanning Electron Microscopy:

FEI 200F Scanning Electron Microscope (www.fei.com)

X-ray and Ultraviolet Photoelectron Spectroscopy:

Kratos Axis-165 electron spectrometer (www.kratos.com)

3.2 Glassware Cleaning Procedure

In order to keep aggregation conditions constant all glassware was cleaned with a solution of 20 parts concentrated nitric acid, 2 parts 30% hydrogen peroxide, and 5 parts deionized water. The corrosive and oxidizing properties of the solution ensured consistently clean glass surfaces during aggregation. The cleaning solution was prepared in either a large beaker or crystallization dish with space left to prevent boil-over. If a beaker is to be used it is advisable to set the beaker with cleaning solution in a larger beaker to serve as double containment. The beaker or crystallization dish was covered with a watch glass during cleaning. The glassware was soaked in the cleaning solution for at least 2 hours. After soaking, the cleaning solution was stored in a container for reuse up to three times. The cleaned glassware was rinsed 10 times in deionized water and allowed to dry. If desired the cleaning solution can be boiled to shorten soaking time to 1 hour, we elected not to do this due to the risk of severe bumping.

3.3 Preparation of Au(111)/mica Substrates

Au(111)/mica substrates were made by vapor depositing 99.999% pure gold splatters on mica in a UHV chamber. Mica slides were repeatedly cleaved with clear tape or a razor blade, cut into pieces, and placed in a mask positioned over a tungsten boat filled with ~ 1 g of gold splatters. The chamber was pumped down to 500 mTorr and then cleaned by glow discharge with nitrogen plasma five times. After evacuation of the chamber with a cryo pump the mica was outgassed by heating at ~ 500°C for several hours via a heater inside a copper block. While the mica outgassed the tungsten boat with the gold splatter was outgassed by repeatedly heating to ~ 500°C for a period of one hour. Just prior to deposition the mica was cooled to 350°C and held there for the duration of the deposition. Deposition of the gold was carried out by resistively heating the tungsten boat until the Au began to sublime. The deposition rate was monitored by a quartz film thickness monitor until the rate stabilized. After stabilization the mica substrates were rotated over the gold boat and held there until the desired thickness was achieved. The rate of deposition was monitored periodically during deposition with a quartz film thickness monitor. Thicknesses were typically in the range of 1000 to 2000

Å, with deposition rates ranged from 0.1 Å/s to 1 Å/s. The chamber pressure during deposition was kept below 4x10-9 Torr. After deposition the copper block and gold boat were allowed to cool to room temperature before the chamber was opened and the new

Au(111)/mica samples removed. The cooling procedure took several hours.

Figure 3.3 1: Diagram of the apparatus and mask used to make Au(111)/mica substrates

(not to scale).

The Au(111) mica substrates grown in the deposition chamber show a terraced morphology expected for a Au(111) surface.142 The size of gold islands in the films is intimately connected to the deposition rate and substrate temperature. A model of gold island formation on mica which explains the dependence of gold morphology was proposed by Sobotik et al.142 The theory states that gold island size is inversely proportional to deposition rate and directly proportional to the substrate temperature.

Elevated substrate temperature increases the rate of surface diffusion for the deposited gold atoms allowing them to incorporate into existing gold grains as opposed to forming new nucleation sites. By depositing at a slow rate the mean free path of a gold atom on the mica surface is increased; making incorporation into an existing grain more likely.

Eventually the individual gold grains fill the mica surface. After the mica surface is covered further gold deposition thickens the existing grains. Figure 3.3 2 shows an STM

80 image of one of the Au(111)/mica substrates. The characteristic terraced surface is clearly visible.

Figure 3.3 2: Ambient STM image of the terraced surface morphology of Au(111).

Prior to sample deposition all Au(111)/mica substrates were hydrogen flame annealed using a homemade quartz torch. The annealing step serves to clean the gold and increase grain size. The Au(111)/mica substrate was place on a ceramic plate in a darkened room. The 0.5 to 1 inch long hydrogen flame was passed in a circle around the gold until water vapor no longer condensed on the surface. This step preheats the substrate so it does not crack during annealing. After preheating, the hydrogen flame was passed over the gold repeatedly at a rate of about 1 Hz until a dull red glow could be seen in the gold film. The substrate was then allowed to cool to room temperature.

Au(111)/mica substrates were used to check the condition of STM tips when imaging in vacuum. The Au(111)/mica substrates were hydrogen-flame annealed, loaded into the RHK, and heated to 400°C for 3 min to drive off volatile contaminants. After the initial heating the gold was Ar+ sputtered for 3 min and reannealed at 300°C for 3 min.

This process was repeated (usually 2-3 times) until a clean surface with repeatable current vs. voltage curves was achieved.

The preparation of HOPG substrates for microscopy was much simpler than

Au(111)/mica. The HOPG was repeatedly cleaved with clear tape to expose a clean surface prior to deposition. No further preparation was used.

3.4 Preparation of STM Tips

STM tips were prepared from 0.25 mm diameter Pt0.8Ir0.2 and 0.25 mm W wire.

Pt0.8Ir0.2 tips were prepared by either cutting with sharp scissors or electrochemical etching. The Pt0.8Ir0.2 wire was etched in 2 M NaCl with a Ni ribbon as the counter electrode. The wire was cleaned prior to etching by immersing it halfway in the electrolyte and applying a bias of 7 V for ~5 sec. After cleaning the wire was lowered until it touched the surface of the electrolyte and advanced 0.5 mm into the electrolyte solution. The main etch was carried out at 25 V until etching ceased. A post etch cleaning was performed by the same procedure as the pre etch cleaning. The tip was sequentially dipped in three beakers of Millipore water. The tip was dried by touching a

Kimwipe to the shaft of the tip removing water drops by capillary action. Care was taken to keep the Kimwipe away from the apex of the tip. The tips were examined under a

82 microscope to ensure sharpness. Pt0.8Ir0.2 tips were used in both UHV and ambient STM experiments.

Electrochemically etched W tips were used exclusively for UHV-STM. Prior to etching the W wire was annealed in a high vacuum bell jar at 10-6 torr. Three segments of W wire 8” long were placed in the bell jar and annealed by passing 6.7 A of current through a wire for 30 min. Each wire was annealed separately. After annealing, the tips were etched in 1 M NaOH with graphite as the counter electrode. The wire was cleaned prior to etching by lowering halfway in the electrolyte and applying a bias of 5 V for ~5 sec. After cleaning 5 mm of the wire was immersed in the electrolyte solution and etched at 5 V until etching ceased. A post etch cleaning was performed by the same procedure as the pre etch cleaning. The tip was dipped in three beakers of Millipore water sequentially and dried with a Kimwipe. The tips were examined under a microscope and discarded if unsatisfactory. The tips were deemed unsatisfactory if dull, split, or otherwise macroscopically misshapen.

When used for UHV-STM both types of tips were cleaned by Ar+ sputtering and checked on a Au(111)/mica sample which was cleaned by repeated cycles of Ar+ sputtering and annealing. The tips were evaluated for resolution by examining the

Au(111) reconstruction. Current Voltage (I(V)) curves were run on the Au(111) to ensure linearity. Figure 3.4 1 shows the Au(111) reconstruction and an I(V) curve.

Figure 3.4 1: UHV-STM image and I(V) curve of Au(111). This curve was acquired at

(I,V)=(15 pA, 1.6 V). This I(V) curve is an average of 64 curves.

Highly Ordered Pyrolytic Graphite (HOPG) was also used as an STM substrate. Figure

3.4 2 shows an atomic-scale STM image of peeled HOPG and an accompanying I(V) curve.

Figure 3.4 2: UHV-STM image and I(V) curve of HOPG. This curve was acquired at

(I,V)=(15 pA, 1.6 V). This I(V) curve is an average of 64 curves.

3.5 Preparation of Unaggregated and Aggregated Porphyrin Solutions

Both the sodium salt (Na4(H2TSPP)) and the chloride salt (H4(H4TSPP)Cl2) of tetrasulfonatophenylporphine were used in this study. The H2(H4TSPP) nanorods made from the different starting materials were indistinguishable save for the presence of sodium ions when the sodium salt was employed. Free base solutions were made by dissolving Na4(H2TSPP) in Millipore water previously degassed by boiling for 1 hr to make a 1-5 mM solution which was then diluted to make solutions ranging from 1 to 50

µM. The pH was adjusted with NaOH over a range of 7.65 to 12.84. Solutions of the unaggregated diacid were prepared by dissolving H4(H4TSPP)Cl2 in sufficient Millipore water (also degassed by boiling) to make a 1-5 mM solution which was then diluted to make solutions ranging from 1 to 50 µM. Diacid solution pH was adjusted with HCl and ranged from 2 to 3.7. Solution concentrations were checked via the intensity of the Soret

-1 -151 band in UV-visible spectra. The extinction coefficients are ε413 nm = 533000 M cm

-1 -1143 (free base) and ε434 nm = 443000 M cm (diacid).

The H2(H4TSPP) aggregates were synthesized by methods previously reported.66,105,144-146 The porphyrin salt was diluted in Millipore water previously degassed by boiling for 1 hour to make a 1-5 mM stock solution. The stock solution was further diluted to the micromolar range and then combined with an equal volume of HCl to initiate aggregation. The aggregate solution used for most SPM experiments was 5 µM

H2(H4TSPP) in 0.75 M HCl. Solutions of 50 µM H2(H4TSPP) in 0.75 M HCl were used in Raman experiments.

3.6 Preparation of H2(H4TSPP) Nanorod Solutions Containing Chloroauric Acid

To investigate the effects of gold ions on aggregate formation and morphology several samples were prepared with concentrations of HAuCl4 ranging from 0.1 to 5 µM.

Two protocols were used to prepare the gold/nanorod solutions. The first protocol involved adding chloroauric acid to the acid solution prior to mixing it with an equal volume of porphyrin. In the second protocol a few microliters of concentrated chloroauric acid was added to the already aggregated nanorod sample. Small aliquots were use to keep changes in solution volume small.

3.7 Preparation and Analysis of STM and AFM Samples

3.7.1 SPM Sample Preparation

SPM samples were prepared using freshly made solutions of TSPP having molarity in the range of 1 μM to 10 μM. HCl concentrations ranged from 0.3 M to 1.5

M. HOPG and Au(111)/mica were used as substrates. The HOPG substrates were repeatedly cleaved with clear plastic tape prior to deposition. The Au(111)/mica substrates were H2 flame annealed prior to deposition.

SPM samples were made using two different protocols. The first protocol involved placing a drop of solution on the substrate followed by spin drying for 30 seconds at 4000 rpm. Deposition times ranged from 20 min to 2 hrs. During deposition the substrates were affixed to an AFM puck by double-sided tape. The puck was affixed to the spin chuck by double-sided tape as well. This setup is illustrated in Figure 3.7.1 1.

Figure 3.7.1 1: Photograph of a Au(111)/mica substrate mounted on the spin chuck used for deposition.

The second protocol involved heating the nanorod solution. The nanorod solution was boiled under reflux for 10 min and then cooled to 90°C. At 90°C an aliquot was taken and placed on the substrate. From this point the method was the same as the first protocol. The second protocol was used in an effort to increase surface concentration of the nanorods. It has been demonstrated that nanorod aggregation is a function of temperature with higher temperature favoring monomers.51 Also, AFM images showed that, while the diameter of the H2(H4TSPP) aggregates remained unchanged, the elevated temperatures of the solution promoted growth of longer nanorods compared with those formed at room temperature.51 The operating theory was that as the solution aliquot on the substrate cooled the nanorods would begin to nucleate at the substrate surface, thereby increasing the surface concentration. In order to ensure that the heating step did not affect the aggregate a 5 µM H2(H4TSPP)/0.75 M HCl solution was checked by UV- visible spectroscopy before and after refluxing for 10 min.

Figure 3.7.1 2: UV-visible spectra of 5 µM H2(H4TSPP)/0.75M HCl before and after refluxing for 10 min. The black spectrum has been shifted up 0.2 absorbance units for clarity.

Figure 3.7.1 2 shows that refluxing the nanorod solution has a moderate effect on the nanorod UV-visible spectrum. The aggregate band at 490 nm is the strongest peak in both spectra. The process of refluxing has moved the equilibrium to the monomer as well.

3.7.2 SPM Data Acquisition

Tapping mode AFM images were acquired using a Digital Instruments Nanoscope

II AFM and a Molecular Imaging Pico Plus combination AFM/STM. Si cantilevers with a typical resonance frequency of 300 kHz and force constant of 40 N/m were used for imaging. Ambient STM images were taken on two different microscopes: a Molecular

Imaging Pico 5 STM and a Molecular Imaging Pico Plus combination AFM/STM. A 1

µm STM head with a low tunneling current preamp (0.1 nA/V) was used for ambient

STM on the Pico Plus.

Samples for UHV-STM analysis were transferred via air-lock into the STM vacuum chamber (working pressure < 1x10-10 Torr) where the aggregate samples were heated to 100C for a period of 2 min in order to remove HCl and water. Non-aggregated samples were heated to temperatures between 100 and 200C prior to image collection.

The STM and controller were purchased from RHK technology and both constant current images and I(V) data were acquired with this system. Unless otherwise stated, the images were plane-fit and low-pass filtered. Most of the data was acquired at either at

293K or 90K on HOPG, but a few images and I(V) curves were obtained from nanorods on Au(111). Both etched W and Pt0.8Ir0.2 tips were used. Generally, the tips required a cleaning step (Argon ion sputtering) in order to produce high quality I(V) curves on a clean gold surface. Spectroscopy was performed by using the RHK software to measure current as a function of sample bias voltage, I(V), at fixed tip-sample separation

(feedback off). Multiple curves were acquired at each setting and then averaged. dI/dV curves were obtained as a numerical derivative of the average I(V). Orbital mediated tunneling spectra, OMTS, (dI/dV at fixed height) were measured on well defined monolayer islands of the porphyrin on HOPG. While dI/dV was determined on nanorod structures (and is reported here), it should not be interpreted as OMTS since the primary conduction mechanism of the nanorods in positive bias is not tunneling. Rather, dI/dV is simply related only to the conductivity of the nanorods.

3.8 Preparation of Raman Samples

Solution Raman and surface Raman spectra of the various porphyrin solutions and gold-adsorbed porphyrin samples, respectively, were acquired with a Spex 14018

Spectrometer outfitted with a double monochromator and thermoelectrically cooled photomultiplier tube detection. Solution phase spectra were obtained at a 90° scattering geometry using a quartz flow cell. The analyte was pumped through the flow cell via a peristaltic pump to avoid sample degradation. Absorption spectra were recorded before and after acquisition of the Raman spectrum to confirm sample integrity. Raman spectra were excited with vertically polarized light. The vertical (polarized) and horizontal

(depolarized) components of the scattered light were detected by means of a polarization analyzer, followed by a scrambler to eliminate the polarization bias of the monochromator/detector. Three different laser wavelengths were used for excitation: 488 nm and 457.9 nm lines of the Ar ion laser were used for the aggregate and diacid respectively. The 413.1 nm line of a Kr ion laser was used to excite the spectrum of the free base. Laser powers ranged from 15 to 60 mW for solution samples. The displayed data were collected with a 2 cm-1 integration interval.

50 μM porphyrin solutions of the free base and unaggregated diacid were used to acquire monomeric solution Raman spectra. 5 and 50 µM H2(H4TSPP) solutions were used to acquire the solution Raman spectrum of the aggregate. Surface substrates were prepared by vapor depositing gold on pre-cleaned glass microscope slides and mica substrates. The gold on glass films were deposited a rate 0.2 nm/s in a high-vacuum deposition chamber at a base pressure of 5 × 10-7 Torr. The gold on mica films were

90 deposited in the same fashion as the Au(111)/mica substrates used for SPM studies. For the sample adsorbed on Au(111), Raman spectra were excited with both vertically and horizontally polarized light by using a half-wave plate to rotate the polarization of the incident light, followed by detection of both the vertical and the horizontal components of the scattered light to generate SS, SP, PS, and PP spectra. Aggregate samples on

Au(111) were prepared by deposition from a 5 μM solution of H2(H4TSPP) in 0.75 M

HCl when using 488 nm excitation and from a 50 μM solution in 0.75 M HCl for measurements at the 413 nm line. Raman samples were prepared by placing a drop of the

5 μM porphyrin solution on Au(111) substrate (same as for the STM samples) for 1 h followed by spinning for 30 s at 4000 rpm. Higher porphyrin concentration for surface

Raman samples were made by allowing a drop of the 50 μM solution to completely evaporate from the Au(111) surface. The gold substrates were mounted in a spinner rotating at 3000 rpm oriented such that the angle between the propagation direction and the surface normal was fixed at 66 and 24° for the incident and scattered beams, respectively.

3.9 Preparation of UV-visible and Resonance Light Scattering Samples

Electronic absorption spectra of the porphyrin in solution and as thin solid films along with appropriate references were collected with Perkin-Elmer 330 and a Shimadzu

UV-2501PC UV-visible spectrophotometers. Quartz cuvettes of 1 mm and 10 mm were used for the solution spectra measurements of monomer, free base, and aggregated porphyrin. Porphyrin concentrations for UV-visible studies ranged from 1 μM to 50 μM.

The pH was adjusted with either NaOH or HCl. For liquid phase spectral measurements,

91 we employed aqueous reference solutions of the same pH as the sample. Pre-cleaned 3 mm thick and 3 cm in diameter quartz disks were employed for depositing a thin film of

H2(H4TSPP) aggregate from solution. H2(H4TSPP) solid samples were prepared by placing 0.1 mL of the aggregate solution on a quartz plate for 90 min followed by spin- drying at 4000 rpm for 30 s. A matching quartz disk was treated with an acidified solution containing no porphyrin and served as a reference. RLS spectra were measured with a PTI Quanta Master Fluorimeter using a 1.00 cm quartz fluorescence cell and slit widths of 0.125 and 2.00 nm for excitation and emission, respectively. RLS intensities were recorded by scanning simultaneously the excitation and emission monochromators

(Δλ = 10 nm) from 200 to 800 nm. A porphyrin concentration range of 0.74 to 5.9 μM in

0.75 M HCl was used.

3.10 Preparation and Measurement of XPS and UPS Samples

3.10.1 UPS and XPS Sample Preparation

XPS and UPS samples were prepared by two different methods. Powder XPS samples were pressed into In shot. Aggregate UPS and XPS samples were prepared by depositing 5 µM H2(H4TSPP) in 0.75 M HCl on either Au(111) or HOPG substrates followed by spin drying at 4000 rpm for 30 seconds. The aggregate was deposited on

Au(111) for 90 min and on HOPG for 60 min. The monomer was deposited on HOPG for 40 min.

3.10.2 UPS Spectral Acquisition

UPS data were obtained with a homemade He lamp source and the He I (21.2 eV) line was used exclusively. A platinum coated concave 600 groove/mm reflection grating with a 3.5 blaze angle coupled with a gold coated spherical focusing mirror was used to produce monochromatic UV radiation. The base pressure in the monochromator chamber was 8x10-10 Torr. The UPS system is attached via a UHV valve to a Kratos Axis-165 electron spectrometer having a base pressure of 8x10-10 Torr.

First, the appropriate reference surfaces were studied. These HOPG and Au(111) samples were then used as substrates for the porphyrin structures prepared as described above. Samples were heated to 100°C in UHV prior to final measurement. The UPS data was acquired using a hybrid lens that focused the ejected electrons into the Kratos spectrometer. A bias of – 20 V was applied to the sample to shift the spectra out of the non-linear region of the analyzer (KE = 0-10 eV). The spectrometer was used in fixed analyzer transmission mode with a pass energy of 10 eV and spatial resolution of 120

µm. The photo-emitted electron energies were analyzed by a Kratos hemispherical analyzer and counted by 8 channel electron multipliers. Under these conditions the energy resolution of the spectrometer is better than 150 meV which was determined at the

Fermi edge of an Ar etched silver sample.

3.10.3 XPS Spectral Acquisition

Porphyrin samples for XPS analysis were prepared with the same procedures as for the UPS samples. 180-200 watts of achromatic radiation at energy 1253.6 eV (MgKα) was used as XPS excitation sources. The analyzer was set for a spatial resolution of 120

µm. The energy resolution was set to 1.0 eV for survey spectra, and to 0.15 eV for the

93 higher resolution acquisitions of C 1s, N1s, S 2p, and Au 4f7/2 peaks. Binding energies were calibrated against the Au 4f7/2 peak taken to be located at BE = 84.3 eV and against the C1s peaks for HOPG samples (BE=284.5 eV).

3.11 Preparation of Helium Microscope Samples

Samples to be analyzed by Helium Ion Microscopy (HIM) were prepared using freshly made solutions of 5 μM H2(H4TSPP) in 0.75 M HCl. HOPG and Au(111)/mica were used as substrates. HOPG substrates were repeatedly cleaved with clear plastic tape prior to deposition. The Au(111)/mica substrates were H2 flame annealed prior to deposition.

HIM samples were made by placing a drop of solution on the substrate followed by spin drying for 30 seconds at 4000 rpm. Each sample was deposited for 1 hr. During deposition the substrates were affixed to an AFM puck by double-sided tape. The puck was affixed to the spin chuck by double-sided tape as well. This setup is illustrated in

Figure 3.7.1 1. HIM samples were checked by AFM prior to imaging by HIM. The microscope used was an ORION® PLUS manufactured by Carl Zeiss located at Pacific

Northwest National Lab in Richland, Wa.

3.12 Preparation of Transmission Electron Microscopy Samples

TEM samples were made by placing a drop of solution on a carbon coated Ni

Formvar TEM grid attached to an AFM puck with GridStick TM adhesive. The AFM puck was affixed to the spin chuck with double sided tape. After a 20-40 min deposition

94 the substrate was spun dry for 30 seconds at 4000 rpm. The samples were analyzed on a

Phillips CM-200 TEM.

3.13 Optimization of H2(H4TSPP) geometry with Electron Affinity and Ionization

Potential Calculation.

All calculations were performed using the commercial program Gaussian03. All reported results are based on Density Functional Theory (DFT) calculations using the

B3LYP functional and the 6-31+G(d,p) basis, or the 6-31+G(2d,p) basis in the case of

Nickel Tetraphenylporphyrin (NiTPP). Ionization potentials and electron affinities were determined by computing the energy differences between initial molecule and the appropriate ionized species, usually in the computed equilibrium geometry of the gas phase parent molecule (vertical IP and EA). In a few cases, the geometry of the ionized species was also optimized. Solution phase IP and EA values were determined using the

PCM method with the Gaussian03 parameters for acetonitrile.

3.14 Fabrication of and Current vs. Voltage Measurements of H2(H4TSPP)

Nanorods Deposited on Interdigitated Electrodes

Interdigitated electrodes (IdE‟s) were fabricated by the University of California

Santa Barbara Nanofabrication Facility (http://www.nanotech.ucsb.edu/). The electrodes were modeled after a design previously reported.49 A 400 nm oxide layer was grown on a p doped silicon wafer. Interdigitated Au electrodes were deposited via photolithography on a chromium adhesion layer. A schematic of the device is presented in Figure 3.14 1.

Figure 3.14 1: Schematic of the electrode to be used in nanorod I(V) experiments.

The IdE‟s arrived from the fabrication facility covered in a protective layer of photoresist.

This layer was removed prior to nanorod deposition by immersion in acetone for 2 min followed by rinsing with ethanol and drying in a 100° C over for 10 min. The IdE‟s were characterized by AFM to check the electrode heights and gaps. A representative

AFM image and IdE cross section are shown in Figure 3.14 2.

Figure 3.14 2: AFM image of IdE and cross section.

Figure 3.7 shows a smaller section of the profile in Figure 3.14 3. The sides of the electrodes are not straight wich we attribute to the width of the AFM cantilever. The electrode gap, full width at half maximum height is 400 nm. This gap width is apropriate for nanorods as rod lengths range from a few hundred nanometers to over a micron.

Figure 3.14 3: Cross section of IdE showing the electrode gap distance.

There is a potential issue with the construction of the IdE; the electrodes are ~120 nm tall. If a nanorod is to connect two electrodes in the IdE it will either span the gap with part of the nanord unsupported or fall down into the space. If the rods fall into the gaps it will be difficult to image them by AFM. Also it is not known if the nanorods are cohesive enough to bridge the electrodes unsupported since we do not have any data regarding the stiffness of H2(H4TSPP) nanorods. Another concern is that nanorods which do span the gap may be cut by the AFM cantilever. To avoid this problem IdE‟s with nanorods deposited on them were imaged by SEM.

Two different concentrations of the H2(H4TSPP)/HCl nanorod solution was deposited on the electrode: 5 µM H2(H4TSPP)/0.75 M HCl and 10 µM H2(H4TSPP)/0.3

M HCl. 5 µL of the nanorod solution was placed in the center of the electrode and either dried under Ar or spun dry after 10 min for 30 sec at 4000 rpm. In order to maximize the number of nanorods bridging the electrode the deposition was repeated up to 20 times. It

49 has been reported in the literature that H2(H4TSPP) nanorods are photoconducting. In order to control this variable the IdE was affixed to the bottom of an electronics box with double sided tape. Leads connected to BNC‟s grounded to the box were soldered onto the Au pads with indium metal. Unless otherwise stated the box was closed during I(V) measurements. I(V) measurments were performed using a Digital Instruments

Nanoscope II SPM. The microscope was reconfigured as shown in Figure 3.14 4 using a breakout box. The breakout box was connected to a current-to-voltage preamplifier with was connected to the IdE BNC‟s. The I(V) measurements were performed by applying a bias voltage generated by the DI controller to the IdE. The voltage difference causes current to flow across the nanorods which is routed through the current-to-voltage preamp. The I(V) data is then dispayed by the DI software. The setup is not dissimilar to an STM with the tip-sample junction replaced by an electrode. The IdE was placed in a glove bag with Ar gas flowing through it for all measurements to avoid reactions with the ambient atmosphere.

Figure 3.14 4: Experimental setup for IdE I(V) experiments.

Photographs of the setup are presented below. During I(V) measurements the preamp and electrode are placed in a plastic bag with Ar flowing through it. The bag is omitted in these pictures for clarity.

100

Figure 3.14 5: Photograph of the experimental setup for interdigitated electrode I(V) measurements.

101

Figure 3.14 6: Photographs of the channels used to connect the breakout box to the preamp.

Figure 3.14 7: Photographs of the electronics boxes used to hold the resistor for calibration and the electrode.

The experimental setup for I(V) measurements was calibrated by replacing the electrode with a 1 gigaohm resistor. I(V) curves were run +/- 2 V to ensure ohmic behavior and correct current outputs. The I(V) curve for the resistor is shown in Figure

3.14 8.

102

Figure 3.14 8: Calibration curve for a 1 gigaohm resistor in the experimental setup described in Figure 3.14 4.

Ohm‟s law (I=VR) predicts that a 1 gigaohm resistor biased at 1 V will produce 1 nA of current. A 2 V bias will produce 2 nA of current. The raw I(V) curve is linear, but the current magnitude is too low. By using Ohm‟s law the raw data generated by the DI can be corrected by the following equation:

(3.14-1)

103

Where I is the corrected current output and Iraw is the raw current output from the DI.

Once this correction is applied the 1 gigaohm resistor I(V) curve (Figure 3.14 8 red trace) exhibits the correct values.

104

Chapter 4: Results and Discussion

4.1 Characterization of Tetrasulfonatophenyl Porphyrin and its Aggregate by UV- visible and Resonance Light Scattering Spectroscopy

Because of its high extinction coefficient solutions of TSPP are highly colored.

The intense color of the porphyrin solution makes UV-visible and RSL suitable techniques for studying the aggregation of H2(H4TSPP). The UV-visible spectrum of

4- monomeric (H2TSPP) (free base form) shows a strong Soret or B-band at 413 nm and four smaller Q-bands at 514 nm, 550 nm, 578 nm, and 638 nm. As its name implies free

4- base (H2TSPP) exists primarily in basic solutions. As solution pH is lowered the free base macrocycle is protonated in two steps. The first species formed is the monoacid

3- 61 (H3TSPP) (pKa= 6.0 ). Further addition of protons converts the monoacid to the diacid

2- 61 (H4TSPP) (pKa=4.5 ). The diacid UV-visible spectrum shows a Soret band red shifted relative to the free base at 434 nm and two Q-bands at 595 nm and 644 nm.

105

Figure 4.1 1: UV-visible specta of tetrasulfonatophenyl porphyrin: free base (Red, pH

10.36) diacid (Black, pH 3.35), and an intermediate pH with both free base and diacid present (Blue, pH 5.27).

Consistent with the conversion of one species to another Figure 4.1 1 shows two isosbestic points; one in the Q-bands at 579 nm and one in the Soret band at 421 nm. The collapse of the four free base Q-bands to two diacid Q-bands is expected as there is a symmetry change in the molecule upon macrocycle protonation. The Q-bands arise from a 0-0 and 0-1 vibronic progression on Qxx and Qyy, as reported in studies on

59 tetracarboxyphenylporphine (H2TCPP). In the free base form Qxx and Qyy are

4- distingushable due to (H2TSPP) „s putative twofold symmetry. After protonation the molecule is fourfold symeteric, resulting in degenerate Qxx and Qyy vibronic progressions and only two Q-bands.

106

It has been demonstrated that H2(H4TSPP) will aggregate under a number of conditions: acid addition,51,64,65,70-79 a combination of acid and salts,51,72,80-84 the addition of ionic surfactants,85,86 template molecules101, proteins88,147,148, electrical potentials149, sol-gel matricies150, rotary evaporation151, layer-by-layer films152, and the presence of other porphyrin species.88 Figure 4.1 2 follows the aggregation process as porphyrin concentration increases at constant pH. Aggregation is indicated by the appearance of new absorbance bands to the red and/or blue of the monomer absorbance bands.72 In the case of H2(H4TSPP) several new peaks are seen at 424 nm, 490 nm, and 707 nm as the porphyrin concentration rises.

Figure 4.1 2: UV-visible spectra of varying concentrations of H4(H4TSPP)Cl2 in 0.75 M

HCl.

107

The width (or lack thereof) of the aggregate‟s 490 nm band is a consequence of

118 motional narrowing. In a H2(H4TSPP) aggregate the excited states are delocalized over many monomers; UV-visible and fluorescence studies estimate that the delocalization extends over 11-22 monomers.72 By extending the excitation over many monomers the geometry of each monomer is relativly undisturbed upon exciting an electron to an unoccupied state. In other words only the 0-0 transition is allowed in the

490 nm band.

By monitoring the intensity of the 490 nm band the kinetics of H2(H4TSPP) nanorod formation can be studied. H2(H4TSPP) aggregation is , along with other variables, a function of temperature; increasing temperature shifts the equilibrium away from aggregates to monomers.51 To study the kinetics of aggregate formation 5 µM

H2(H4TSPP) in 0.75 M HCl was boiled under reflux for 10 min to force the mixture into monomer form. After boiling, the solution was placed in a room temperature water bath, cooled to 40° C, and placed in a cuvette in a UV-visible spectrophotometer set to 490 nm.

The absorbance at 490 nm was monitored as a function of time (Figure 4.1 3). Initially there is a rate determining induction period followed by a rapid increase in nanorod concentration. This can be explained by a slow nucleation step followed by rapid rod growth. Pasternack et al. observed a similar trend in the 490 nm band.89 Figure 4.1 3 also shows that aggregation (at this pH and porphyrin concentration) reaches a steady state after about 40 min. All the nanorod solutions used in this work were allowed to aggregate for at least one hour to ensure steady state.

108

Figure 4.1 3: Graph of absorbance at 490 nm vs. time during nanorod formation.

It is reasonable to ask if the aggregation process is an irreversible chemical reaction as opposed to the formation of a supramolecular aggregate where the structure of the individual monomers is intact. To investigate this question we prepared a solution of nanorods, charaterized it by UV-visible spectroscopy, then added NaOH to the solution until the pH reached 10.32. Figure 4.1 4 compares the nanorod spectrum, the spectrum of

4- the basified nanorods, and a control spectrum of free base (H2TSPP) at pH 10.00. Aside from broadening in the basified nanorods the control and basified spectra are identical indicating that aggregation is reversible and nondestructive to the monomers.

109

Figure 4.1 4: UV-visible traces demonstrating the reversibility of aggregation: nanorod solution at pH 0.12 (green), the same solution after rasing the pH to 10.32 (red), and a reference free base spectrum at pH 10.00 (black).

The principle peaks discussed in UV-visible spectra are the monomer Soret band at 434 nm and the aggregate band at 490 nm. Additionally, a band at ~ 424 nm and a prominent a Q-band at 707 nm are present in aggregate solution. Resonance light scattering (RLS) can be used to determine which peaks are associated with aggregates and which peaks are due to monomers and smaller species. As was discussed in the theory section on RLS an aggregate the size of a H2(H4TSPP) nanotube will scatter a great deal while a species closer in size to a monomer will not. RLS data of H2(H4TSPP) nanotubes is presented in Figure 4.1 5. This data was taken using the same solutions as

Figure 4.1 2.

110

Figure 4.1 5: RLS specta of varying concentrations of H4(H4TSPP)Cl2 in 0.75 M HCl.

The monomeric signal (0.74 µM, black trace) in Figure 4.1 5 is so weak as to be almost indistingushable from the x axis. A strong RLS band , the dominant feature of the spectra, quickly grows in at ~ 495 nm as the concentration of porphyrin increases.

Smaller bands can be seen between 600 and 750 nm, as is shown in Figure 4.1 6.

111

Figure 4.1 6: RLS specta of varying concentrations of H4(H4TSPP)Cl2 in 0.75 M HCl.

The small feature around 660 nm is attributed to residual monomer fluorescence. As aggregation proceeds this band diminishes until it is roughly equal in intensity with a peak growing in at ~711 nm. The 711 nm band increases with the extent of aggregation as can be seen in Figure 4.1 7. The increase in the 711 nm peak in the RLS spectrum correlates well with the scattering signal from the Soret band. Figure 4.1 7 traces the scattering signal from the 495 nm and the Q-bands as a function of porphyrin concentration at constant pH. Both traces rise up to 3 µM H4(H4TSPP)Cl2 before leveling out. The similarity in the 495 nm and Q-band traces implies that they arise from the same species. It is tempting to conclude that the Q-band signal is less intense than the

Soret band due to the small extinction coefficient of the Q-bands relative to the Soret band. There is a study in the literature on RLS of chlorophyll a aggregates which addresses Q-band scattering.153 Aggregated chlorphyll a molecules exhibited strong RLS

112 out of the Soret band and 1/114 the scattering out of the Q-band; very similar to the

H4(H4TSPP)Cl2 ratio of 1/100. The authors were unable to explain the difference in scattering intensity. The interesting difference in the two cases is that the Q- and Soret bands of chlorophyll a are roughly equal in intensity unlike H4(H4TSPP)Cl2 where the

Soret and Q-band intensities are very different. This indicates that the reduced Q-band scattering in seen in H2(H4TSPP) nanotubes is not due to a lower extinction coefficent.

Figure 4.1 7: Comparison of the scattering intensity of the J- and Q-bands of

H4(H4TSPP)Cl2 in 0.75 M HCl.

Further insight into the relationship among the 424, 490, and 707 nm absorption bands can be gained by deconvolution of the UV-visible spectra into component peaks and plotting the intensities of the peaks as a function of H2(H4TSPP) concentration at a constant HCl concentration. CASA XPS analysis software was used to deconvolute the spectra. An example of the deconvolution is presented in Figure 4.1 8.

113

424 Gaussian

707 Gaussian

Figure 4.1 8: Deconvoluted UV-visible spectrum of 5.9 µM H4(H4TSPP)Cl2 in 0.75 M

HCl.

The gaussians used are for the most part self-explanatory. We believe that the 683 nm gaussian is truly a sum of smaller bands. In order to probe the connection between the

424 nm band and the 707 nm Q-band two sets of solutions were prepared. Both sets ranged in concentration from just under 1 µM to 6 µM H4(H4TSPP)Cl2. One set was 0.4

M in HCl the other was 0.75 M in HCl. The UV-visible spectrum of each solution was taken and deconvoluted followed by graphing the intensity of the 424 nm, 490 nm, 683 nm, and 707 nm Q-bands as a function of H4(H4TSPP)Cl2 concentration.

114

424nm 424nm

707nm 707nm

Figure 4.1 9: Plots of the intensities of various UV-visible peaks as a function of total porphryin concentration. The concentration of HCl for the left graph is 0.75 M, the right is 0.40 M.

The two peaks most closely correlated in Figure 4.1 9 are the 424 nm band and the 707 nm Q-band. The simplest explaination for this data is that the 424 and 707 nm bands arised from one species while the 490 nm absorbance comes from a different species.

There is a study of the effect of pressure on the UV-visible spectra of H2(H4TSPP) nanorods which also suggests a connection between the 424 nm and Q-bands.154 Chan et al. found that the monomer, 490 nm, 424 nm, and Q-bands all red shift with increasing pressure (up to 40 kbar). The rate of red-shift with increasing pressure for the 424 nm and Q-bands were much closer to each other than the rate of red-shift for the monomer or

490 nm peaks. The data presented in Figure 4.1 7 and Figure 4.1 9 does not lend itself to an unambiguous conclusion. The RLS data shows a correlation between the scattering form the 490 nm and 707 nm bands indicating that they arise from the same species. The

424 nm band does not exhibit an observable RLS signal. It is worth noting that a

115 scattering peak of comparable intensity to the 711 nm peak could be obscured by the dip in RLS spectrum due to monomer absorbance. The RLS data seems to show that the 707 nm and 490 nm absorbance bands arise from the same species, while the data is not clear with respect to the 424 nm band. The 424 nm band‟s scattering intensity may be too low to observe or there may be no signal at all. Based upon RLS data we can not assign the

424 nm band. The peak height vs. concentration data indicates that the 424 nm and 707 nm bands grow in at the same rate with the 490 nm band growing in much faster. This data suggests that the 490 nm species is different from the species which is responsible for the 424 and 707 nm bands. Both the 424 and 490 nm bands can be connected to the

707 nm band through UV-visible and RLS data respectivly, but we have not been able to directly connect the 424 and 490 nm bands. Based upon the data presented here we can not unambiguoulsy assign the aggregate peaks in the UV-visible and RLS spectra, although some tennative conclusions can be drawn. It is certain that the 490 nm peak belongs to the nanotubes because of its strong RLS signal. The assignments of the 424 nm and 707 nm peaks are uncertain because of the conflicting RLS and UV-visible data.

Since the main focus of this work is charaterizing H2(H4TSPP) nanorods deposited on substrates UV-visible spectra of free base, diacid and aggregated

H2(H4TSPP) deposited on quartz plates were recorded to evaluate the effects of deposition on the molecule in its different forms. The free base and diacid data is displayed in Figure 4.1 10.

116

Figure 4.1 10: UV-visible spectra of free base (left) and diacid (right) TSPP solution (red) and solid phase spectra (black). The concentration of both solution spectra is 2.618 µM.

The pH‟s of the solutions in the solution phase spectra are 7.53 and 3.35 for the free base and diacid respectivly. The solid phase spectra are of 50 µM porphyrin solutions dried on quartz plates. The pH‟s of the solutions used for deposition are 7.65 and 3.73 for the free base and diacid respectivly.

The solution and solid phase free base spectra match well. There is broadening and a small hypsochromic shift in the Soret band going from solution to solid phase and a small bathochromic shift in the Q-bands. The data shows that the free base deposits on quartz intact. The diacid spectrum is more interesting, instead of broadening and small wavelength shifts the solid phase spectrum is very different from the solution spectrum.

The data shows that a non- aggregated diacid solution forms aggregates when deposited

117 to dryness on quartz. This finding makes sense because as the solution dries the porphyrin concentration increases to the point that aggregates begin to form.

A different approach was taken for the study of H2(H4TSPP) nanorods deposited on substrates. Instead of depositing to dryness as was done with the free base and diacid, the aggregate was deposited on quartz in the same fashion as a microscopy sample. This was done to mimic, as closely as possible, the condition of the aggregate in a microscopy sample. The data is presented in Figure 4.1 11:

Figure 4.1 11: UV-visible spectra of 5 µM H2(H4TSPP) in 0.75M HCl: solution spectrum

(black) and deposited on a quartz plate for 90 min (red) followed by spin drying.

The nanorods on quartz spectrum is noisey, but there is a clear peak at 490 nm which correlates well with the solution spectrum and previously reported UV-visible spectra on glass.51,66 This peak was assigned to the nanotubes by RLS studies (vide supra). The Q-

118 bands are hard to read, it looks like there is a peak in the solid phase spectrum at ~ 705 nm which is similar to the solution spectrum. The 424 nm band/monomer region of the solid phase spectrum is also difficult to interperet. The absorbance is elevated above the background, but it is almost flat so no definitive statement can be made comparing this part of the spectrum to the solution data. The critical message from Figure 4.1 11 is that, as far as UV-visible spectroscopy can discern, the H2(H4TSPP) aggregate deposits intact.

4.2 Characterization of H2(H4TSPP) Aggregates by Ambient SPM Studies

4.2.1 Characterization of H2(H4TSPP) Aggregates by Tapping Mode AFM

Almost all of the H2(H4TSPP) nanorod images reported in the literature were acquired by AFM using mica or HOPG as substrates.49,51,58,73,105 We are using both

Au(111) and HOPG to image the H2(H4TSPP) aggregates because we wish to compare nanorod morphology and electronic properties on multiple substrates. HOPG is a problematic substrate because it possesses native features that emulate rod-like formations.155,156 By carefully characterizing aggregate heights and widths on Au(111) the probability of mistaking a native graphitic feature for a nanorod can be minimized.

2 Figure 4.2.1 1 compares 10 µm tapping mode AFM images of H2(H4TSPP) nanorods deposited on both Au(111) and HOPG. The image on Au(111) clearly shows the Au terraces with straight and narrow nanorods. The HOPG surface is stepped as opposed to terraced and also shows nanorods of similar dimensions.

119

Figure 4.2.1 1: Tapping mode AFM images of H2(H4TSPP) nanorods deposited on

Au(111) (left) and HOPG (right).

2 Figure 4.2.1 2 compares two 2 µm AFM images acquired from H2(H4TSPP) aggregates deposited on an Au(111) and HOPG. Below each image is the corresponding cross-sectional profile of two adjacent nanorods. It is gratifying to observe a very close correspondence in nanorod dimensions between the two different substrates. Nanorod size does not change upon subsequent scanning, at least in the large-scale images, indicating that reasonably sturdy intermolecular interactions hold the rods together.

When considering the cross sections of nanorods deposited on the two different substrates we observe that their heights are approximately equal (~4 nm). The average width of an

H2(H4TSPP) rod seen in our AFM images is about 35 nm. The reported nanorod dimensions based on AFM data are 18-38 nm for the width and 4-10 nm for the height and agree well with our rod measurements obtained from similar size images.68

120

Figure 4.2.1 2: Tapping mode AFM images of H2(H4TSPP) nanorods deposited on

Au(111) (left) and HOPG (right) with cross sections through single nanorods.

The height of the nanorods is worth further discussion. Our AFM (and STM) analysis indicates that a single nanorod is 4 nm tall. Figure 4.2.1.2 clearly shows that the nanorods bundle and stack upon one another. What is consistently observed is that bundles of nanorods exhibit heights of multiples of two. This is easily explained by

Vlaming et al.‟s work with cryo electron microscopy.61 Vlaming‟s data clearly indicated that H2(H4TSPP) nanorods are tubes in solutions with 2 nm thick walls. When deposited on a substrate and dessicated the tubes collapse leaving structures with an average height of 4 nm. Two tubes would be 8 nm tall, as so on. We have observed structures 6 nm tall by STM (vide infra) which we attribute to one and one half tubes.

121

To eliminate the possibilty of a photoinduced reaction interfering with our microscopy studies a sample of 5 µM H2(H4TSPP) in 0.75 M HCl was prepared in a dark room lit with a 25 W red light bulb. The solution was not exposed to artificial or natural light sources save the red light bulb until deposition on HOPG was finished. The sample was then checked by AFM for rod morophology.

Figure 4.2.1 3: Tapping mode AFM image of H2(H4TSPP) nanorods deposited on

Au(111). This solution was kept in the dark during preparation and deposition.

Figure 4.2.1 3 shows an AFM image of H2(H4TSPP) nanorods deposited on Au(111) without light exposure. The cross section shows that the aggregates are ~4 nm tall and

~30 nm wide (full width at half max) the same dimensions as the samples exposed to ambient light. This finding shows that rod dimensions are not affected by exposure to light and elimniates the need to prepare samples in a darkened room.

122

While the nanorod morphology appears to be independent of substrate and light exposure there is an important difference between the Au(111) and HOPG samples.

After deposition of nanorods on HOPG the surface appears pockmarked. As will be shown in the STM section these pockmarks are not holes in the graphite, but rather holes in a monolayer of monomeric H2(H4TSPP). Several of these monolayer vacancies are marked with white circles in Figure 4.2.1 4. This is an exciting observation as it gives us the opportunity to investigate the structure and electronic properties of both monomeric

H2(H4TSPP) and its aggregate on the same sample.

Figure 4.2.1 4: Tapping mode AFM image of H2(H4TSPP) nanorods deposited on HOPG.

Holes in the H2(H4TSPP) monolayer are marked with white circles.

Figure 4.2.1 5 shows a high resolution AFM image of nanorods deposited on

HOPG. This image is the clearest small-scale image we were able to obtain by AFM.

123

The holes in the underlying monolayer can clearly be seen in the background. The AFM image reveals four nanords lying side by side (one of which appears to be cracked), but no interior detail can be seen.

Figure 4.2.1 5: High resolution tapping mode AFM image of H2(H4TSPP) nanorods deposited on HOPG.

We have used AFM to verify the deposition and morphology of H2(H4TSPP) nanorods on both Au(111) and HOPG. Our dimensions are in line with previously reported values and the aggregate density is sufficient for STM studies. The AFM was unable to resolve any internal structure of the nanorods, but a background monolayer of

H2(H4TSPP) is seen on HOPG samples.

124

4.2.2 Characterization of H2(H4TSPP) Aggregates by Ambient Scanning Tunneling

Microscopy

As was stated in section 4.2.1 almost all of the H2(H4TSPP) nanorod images reported in the literature were acquired by AFM using mica or HOPG as

49,51,58,73,105 substrates. We will use both Au(111) and HOPG to image the H2(H4TSPP) aggregates because we wish to compare nanorod morphology and electronic properties on multiple substrates. The goal of this study is to evaluate the accuracy of the staircase model of aggregation and, if necessary, propose a new model of aggregate structure.

Because the nanorods are tall structures low tunneling currents and high sample biases will be used to avoid crashing the tip into the nanorods.

Initially we will compare the results of our AFM study with STM images to ensure good correlation between the two techniques. Figure 4.2.2 1 compares 1 μm2 size

AFM and STM images acquired from H2(H4TSPP) nanorods deposited on a Au(111) substrate. Below each image is the corresponding cross-sectional profile of two adjacent nanorods. It is gratifying to observe a very close correspondence between the AFM and the STM images of H2(H4TSPP) nanorods, which appear as long straight features. The gold terraces with single atomic steps are also observed in the images. The size of the nanorods does not change considerably upon subsequent scanning, indicating that reasonably sturdy intermolecular interactions hold the rods together. We note that scanning under positive sample bias yields clear STM images. In reverse bias, it was difficult to obtain well-resolved nanorod topography. This point is of prime importance and will be discussed in detail in section 4.5.3.

125

Figure 4.2.2 1: AFM (left) and STM (right) images of H2(H4TSPP) nanorods deposited on Au(111) and accompanying cross sections.

When considering the cross sections of two adjacent nanorods in the AFM and

STM images we observe that their heights are approximately equal (~4 nm) but their widths are not. Similar rod height values obtained by the two different SPM techniques may be related to aggregates‟ high conductivity and/or the presence of water on their surface. This question will be explored in detail in section 4.5.3. The width of an average H2(H4TSPP) rod seen in our AFM images is about 35 nm while the same rods imaged by STM measured 25 nm. This discrepancy in the width values of the nanorods is

126 due to the difference in the apex size of the imaging probes employed in AFM and STM.

Taking the value of the width at half the rods‟ height in the AFM cross section (~24 nm) brings the thickness value more in line with that obtained from the STM rod profile. We note that the rod cross section in our STM image shows further evidence of twinning, a phenomenon that was observed but not fully understood by others.144 This substructure was attributed to some kind of backbone in the nanorod self-assembly.

As was discussed in the section on nanorod AFM the nanorods are flattened tubes with a wall thickness of ~ 2 nm. STM images corroborate this feature of the nanorods.

Figure 4.2.2 2 shows several nanotubes on Au(111). Several of the rods have a “stepped” appearance resulting from the tube fracturing upon either deposition or desiccation. The step is clearly visible in the cross section in Figure 4.2.2 2.

Figure 4.2.2 2: STM image of H2(H4TSPP) nanorods deposited on Au(111) and accompanying cross section. The setpoint is 1 pA at 1.6 V sample bias.

127

Figure 4.2.2 2 shows that the rods are layered structures with each layer being ~2 nm tall.

To our knowledge this feature of nanorods deposited on substrates has not previously been reported. Further examples of the tube-nature of the aggregates can be seen in

Figure 4.2.2 3. The two aggregates which almost form a right angle in the left of the figure clearly look tube-like. The inset in Figure 4.2.2 3 is a subsequent scan over the intersection of the two tubes. In this image part of the upper layer of the tube has been sheared off revealing the lower layer.

Figure 4.2.2 3: STM image of H2(H4TSPP) nanorods deposited on Au(111). The inset is a subsequent smaller scan of the two tubes in the left of the image. The setpoint is 1 pA at 1.6 V sample bias in both images.

Just as we compared AFM and STM data of nanorods on Au(111) we will compare AFM and STM images of nanorods deposited on HOPG. Figure 4.2.2 4

128

2 compares 1 μm size AFM and STM images acquired from H2(H4TSPP) nanorods deposited on HOPG. Below each image is the corresponding cross-sectional profile of two adjacent nanorods. It is gratifying, just as with nanorods on Au(111) to observe a very close correspondence between the AFM and the STM images of H2(H4TSPP) nanorods. Also the nanorods on HOPG exhibit the layered structure previously described for the rods on Au(111). Additionally, the monolayer of porphyrin monomers can be seen in the background of both the AFM and STM images of rods on HOPG unlike the case for rods on Au(111).

129

Figure 4.2.2 4: AFM (left) and STM (right) images of H2(H4TSPP) nanorods deposited on HOPG and accompanying cross sections.

It is worth discussing the heights of the nanorods relative to single molecules. In section 4.6.2 we will show that H2(H4TSPP) monomers laying flat on HOPG have an apparent height of 0.6 nm by STM. This is six and a half times smaller than the 4 nm heigth of a single nanorod. The contrast is even more stark when stacks of multiple rods are considered. Figure 4.2.2 5 shows a standard two dimensional STM image of nanorods on HOPG and a three dimensional rendering of the same image. Nanorod stacks of up to 12 nm are visible in Figure 4.2.2 5. The three dimensional rendering

130 illustrates the height differences to be navigated by the STM. Large steps are seen as the

STM tip moves from layer to layer in a stack of rods. By comparison the monolayer is barley noticable. Since tunneling current drops off exponentially with tip sample separation the setpoint must be kept low (typically 1 pA). Bias voltages come in to play as well, with high positive sample biases being best for imaging the aggregate.

Figure 4.2.2 5: Standard “bird‟s eye view” of H2(H4TSPP) nanorods on HOPG (left) and a three dimensional graph of the same image.

The high resolution STM image of a twin-nanorod in Figure 4.2.2 6 provides a potential insight into composition of the rod and its surrounding environment. First, we observe a uniform disk-like substructure that appears to constitute the morphology of the rod. Second, similarly sized disks lie scattered around the body of the rod. The disks present on the surface of the rod are irregularly arranged as though they were perturbed by the scanning process. When we collect consecutive images of the same rod section, the disk distribution becomes more disrupted, and the disks are swept away (Figure 4.2.2

6) by the STM probe. Eventually, the nanorods disintegrate completely. Interestingly,

131 we observe that the disks themselves exist as stable units and remain intact upon repeated scanning. The disks that we observe have an average diameter of 6 nm with an apparent height of 1 nm. The diameter of a single disk is much larger than the width of a single

H2(H4TSPP) molecule which measures ~2 nm across based on van der Waals radii. We note that, because of thermal drift and piezoelectric creep, the disks often appear slightly distorted.

Figure 4.2.2 6: High resolution images of the same H2(H4TSPP) nanorod section imaged in succession at (A) 0.9 V and 0.015 nA, (B) 0.7 V and 0.015 nA, and (C) 0.5 V and

0.015 nA.

Disk morphology has not been observed in any of the previously reported SPM

51,58,73,105 studies of H2(H4TSPP) nanorods deposited on either mica or HOPG. These reports promoted a linear H2(H4TSPP) aggregation and proposed models of ribbon-like building blocks for the nanorods.49,58,73,105 A linear assembly does not accurately describe the disk structure seen in the nanorods deposited on Au(111). A circular arrangement of the porphyrins, on the other hand, is quite satisfactory. Earlier X-ray and light scattering studies of H2(H4TSPP) aggregates in solution have proposed the existence

132 of ring-like structures. Small angle X-ray scattering (SAXS) analysis suggested a winding spiral or a series of rings forming a hollow tube measuring 16.0 nm across and 2 nm thick composed of ~26 TSPP monomers.71 A combination of elastic (ELS) and dynamic (DLS) light scattering techniques was used to detect the presence of hierarchical

H2(H4TSPP) structures in solution: large (1-1.5 μm), medium (100-200 nm), and small

(3-6 nm) sized aggregates were reported.63,103 An aggregation number ranging between 6 and 32 was calculated for the smaller components, whereas a range of 105 to 106 was found in the case of the large clusters.63 Ab intio calculations using optimized

H2(H4TSPP) geometry predicted a closed loop aggregate structure composed of 60-70

157 porphyrin monomers. The circular arrangement of H2(H4TSPP) molecules not only explains the structure of the disks and the nanorods they apparently makeup but also provides insight into their electronic behavior.

The model that we tentatively propose is a circular assembly of the H2(H4TSPP) monomers depicted in Figure 4.2.2 7. This model is similar to the crystal structures of circular aggregates of bacteriochlorophyll molecules (BChls B850) found in the light harvesting systems (LH1 and LH2) of photosynthetic bacteria.15,158,159 The LH2 complex in Rhodospirillum molischianum adopts a highly symmetrical hollow ring shape with a radius of about 7 nm composed of 18 overlapping chlorophyll monomers.15 The internal radius of the B850 ring is about 3.1 nm. The circular arrangement is dictated by the noncovalent interactions between the chlorophylls (<0.35 nm) and their surrounding protein environment. In our 6 nm H2(H4TSPP) disk model, we offer that the zwitterionic porphyrins are arranged in a side by-side fashion stabilized by electrostatic interactions

133 between the negative sulfonate groups and the protonated pyrrole nitrogen atoms and π-π interactions between overlapping phenyl rings. Furthermore, we do not expect the porphyrins to be planar. X-ray diffraction studies of a protonated tetraphenyl porphine analog, H2TPP, indicate a puckered macrocycle with pyrrole rings tilted by 28° to 33° to minimize free energy.160 The phenyl rings are twisted by 35° allowing for a greater conjugation with the macrocycle.160 Thus, near-planar geometry of the phenyls in the

H2(H4TSPP) monomer would favor aggregation because the charged groups on the monomers can get close enough to interact while puckering of the porphyrin macrocycle

157,161 would induce curvature in the H2(H4TSPP) aggregate. The formation of a symmetrical circular aggregate of N nonplanar molecules is illustrated in Figure 4.2.2 7, where the nonplanar porphyrin is represented by a one-dimensional shape, which deviates from linearity by the angle R. A closed ring of N molecules results when this angle equals 2π/(N + 1). To our knowledge, the molecular structure of the diacid of

H2(H4TSPP) is not available, however, the structures of the diacid of

2+ tetraphenylporphyrin, H2TPP , in the presence of different halide counterions has been explored.161 It was found there that the meso-substituted phenyl rings are rotated 21-33° from the mean porphyrin plane, and the pyrrole rings adopt a saddled configuration in which alternate rings are tilted up and down by 28-33° from the mean porphyrin plane.

Nonplanarities in this range are consistent with the formation of closed rings containing

10 to 16 porphyrins. For illustration purposes, we consider the case N = 16 (α = 21.2°) here, taking the width of the porphyrin to be 2L = 2.2 nm based on van der Waals radii.

The porphyrins are arranged around the ring of radius R such that the positively charged centers, represented by open circles in the figure, overlap with the negatively charged

134 sulfonato groups, represented by filled circles, on adjacent molecules. The separation of the parallel overlapping planes is taken to be d = 0.34 nm. The length of the polygon side is then s = √L2 + d2 = 1.2 nm. The radius of the ring R is found from sin (π/N) = s/2R which gives R = 3 nm consistent with the size of the disks imaged in STM.

Figure 4.2.2 7: Proposed circular model for the organizations of H2(H4TSPP) molecules within a single disk substructure observed in high resolution images of H2(H4TSPP) nanorods. The average disk diameter used in this model is ~6 nm. The CPK representation of the porphyrin molecule used in the model is based on the van der Waal radii and ~21° porphyrin ring deformation. To generate the aggregate ring structure, the

135 molecules were manipulated and displayed in DS Viewer Pro (Accelrys). A 25 nm2 STM section of the high resolution image in Figure 4.2.2.6 is inserted for reference. Below the

H2(H4TSPP) model is a schematic illustration showing a portion of a circular aggregate containing N monomers, which deviate from planarity by R = 2π/(N + 1). Explanation of the model is provided in the text.

We noted earlier that scanning under positive sample bias yielded clear STM images of the nanorods. With reverse bias, it was difficult to obtain well-resolved topography. This type of behavior is indicative of primarily LUMO mediated tunneling associated with transient reduction (related to electron affinity) of the adsorbate.128,162

Tightly coupled LUMOs of close-packed H2(H4TSPP) were also thought responsible for the photoconductive behavior of the nanorods upon illumination with 488 nm light source.49 We can argue that the value of +1.3 V bias at which we obtained our best images may be related to tunneling into delocalized π* system of the H2(H4TSPP) monomers forming the ring structure. By taking the work function of Au(111) in air to

163 be 4.8 eV, the electron affinity of the H2(H4TSPP) aggregates is then 4.8 V - 1.3 V =

3.5 eV. This number is very close to the value measured for the reversible reduction of

H2(H4TSPP) J-aggregates deposited as films on ITO substrates and measured in

65 acetonitrile with TMAP as the supporting electrolyte. The E1/2(SCE) peak reported at -

0.84 V can be easily converted to vacuum state reference by adding 4.71 + E1/2(SCE), resulting in a value of 3.87 V.162 The same study also reported that the reversible reduction of the monomeric porphyrin occurred at E1/2 referenced to vacuum at 4.18 V.

The observed decrease in the reduction potential of the thin films of the H2(H4TSPP) J-

136 aggregate was postulated to result from the delocalization of electrons between the monomer porphyrin units in the aggregate.85 The electrochemical oxidation of the

H2(H4TSPP) J-aggregate as well as the monomeric porphyrin in solution was found to be irreversible.65 This may account for our unsatisfactory attempts to image the

H2(H4TSPP) nanorods in the negative bias. It is gratifying that the reported electrochemical results are consistent with our STM findings. The electronic structure of the H2(H4TSPP) aggregate will be discussed in detail in section 4.5.3.

While the proposed circular porphyrin assembly accounts well for the general shape of the H2(H4TSPP) disks observed in our STM images, it does not account for the lack of a more detailed structure within the disks themselves and particularly why the disks have no cavities as our model in Figure 4.2.2 7 suggests. One possibility is that we are tunneling into one giant delocalized π* system of the noncovalently coupled

H2(H4TSPP) circular aggregate. Interestingly, STM images of molecules composed of 24 covalently linked Zn porphyrins (C24ZB) deposited on Cu(100) appear as discrete elongated doughnut-like shapes ~7 nm in diameter with a clearly visible hollow.164 Thus, we tend to discount this explanation. On the other hand, the SAXS data which suggested a winding spiral or a series of rings forming a hollow tube composed of H2(H4TSPP) monomers showed a region of electron density in the middle of the ring that could not be identified.71 It is possible that the electronic conductivity of the central region of the

H2(H4TSPP) disks is due to coordinated water molecules and counterions. TG and DSC analysis shows that H2(H4TSPP) aggregates deposited on substrates release water up to

400°C indicating strongly bound water.90 These observations are also consistent with the

137 fact that the apparent height of the nanorods is the same in AFM and STM. A significant ionic conductivity through the nanorods would explain the large conductivity necessary for the metal-like imaging. We have identified and proposed a possible structure for the building blocks that make up the H2(H4TSPP) nanorods. We speculate that the disks may organize themselves into what others have termed as “ribbon-like” stacked assemblies, which in turn group together to make up the bodies of the rods. It is also possible that the disks may self-assemble into helical superstructures or hollow nanotubes in solution.

These nanotubes become flattened and appear as nanorods on a solid surface upon dehydration. The adhesive forces generated by the thin film of surface water (surface tension) may also cause collapse of tubes to rods as the solvent layer evaporates. The absence of the stabilizing aqueous media may render the solid state rods or tubes brittle, and they disintegrate easily with repeated probing by the STM tip making the exploration of the mesoscopic structure of the H2(H4TSPP) nanorods difficult. The self-limiting dimensions of the widths (20-40 nm) and heights (4 nm) of the nanorods deposited on

Au(111) and other solid substrates are intriguing and require further investigation.

For all of its strengths there are issues with the disk model. Exciton theory at its simplest level predicts that the red-shifted 490 nm band of a circular aggregate derives from the transition to the doubly degenerate (k = 1) excitonic states, while the blue- shifted 424 nm band should derive from transitions to the k = 0 exciton.165 Our depolarization ratios for 490 nm band excitation, however, are consistent with resonance with a nondegenerate excited state (vide infra). We also note that the standard staircase model for the aggregate is incapable of accounting for the apparent Soret band splitting,

138 since, according to the Gouterman four-orbital model, an ungerade perturbation of this sort cannot lift the degeneracy of the Soret and Q-band excited states.166 A second issue with the theory is that after repeated efforts to visualize this diskotic structure in nanorods deposited on HOPG no such structures were seen. We were unable to verify the disk model with HOPG substrates. Our inability to visualize the disks on graphite brings the model into question. We wish to answer the question what are the 6 nm disks and why are they seen only on Au(111)?

Our hypothesis is that the 6 nm disks are the result of an interaction between the gold and the porphyrin. One possibility is that gold is inserting into the porphyrin. The synthesis of Au(III)TSPP has been reported in the literature.167 The procedure is simple:

H4(H4TSPP)Cl2 is refluxed for 5 min in water with a five-fold excess of KAuCl4.

Another possibility is that gold is initiating a reaction like dimerization or breaking down the porphyrin. We will investigate these scenarios in the hopes of discovering the identity of the 6 nm disks.

As a first step UV-visible experiments were run on mixtures of gold with both monomeric porphyrin and aggregated porphyrin. Au(III)TSPP‟s Soret band is at 406

167 nm as opposed to monomeric H2(H4TSPP)‟s Soret band at 434 nm. If Au(III)TSPP forms from a mixture of porphyrin and gold and HCl it may explain the disks. The source of Au in these experiments was chloroauric acid trihydrate (HAuCl4(H2O)3).

Figure 4.2.2 8 shows the UV-visible spectra of several nanorod solutions with varying concentrations of chloroauric acid. The solutions were all prepared by the same protocol:

10 mL of 10 µM H2(H4TSPP) was mixed with 10 mL 1.5 M HCl. If the solution

139 contained chloroauric acid the gold was included in the HCl solution. For these samples the nanorods aggregated in the presence of Au(III). The solutions were allowed to stand for 1 hr prior to analysis by UV-visible spectroscopy.

Figure 4.2.2 8: UV-visible spectra of nanorod solutions with varying concentrations of

Au(III).

Figure 4.2.2 8 is a negative result. There are no significant differences among the four spectra. There is no discernable peak at 406 nm which would be attributed to

Au(III)TSPP. It is possible that there is a very small Au(III) peak hidden under the nanorod absorption envelope, but that is a conjecture without proof. A similar experiment was run to investigate the effects of chloroauric acid on aggregated nanorods.

A few microliters of concentrated chloroauric acid were added to a nanorod solution which was allowed to stand for 1 hr. A UV-visible spectrum was taken 1 hr after adding the chloroauric acid.

140

Figure 4.2.2 9: UV-visible spectra of nanorod solution with chloroauric acid added after the nanorods were finished aggregating. A reference spectrum of nanorods with no chloroauric acid is included.

As with Figure 4.2.2 8 there is no visible peak at 406 nm indicative of gold insertion.

The 490 nm and Q-bands in Figure 4.2.2 9 are indistinguishable with regards to the presence of Au(III). The biggest difference between the two spectra is the monomer band is less intense in the presence of Au. The cause of the change in monomer absorbance is not clear; especially considering that no trend of decreasing monomer absorbance with increasing gold concentration was seen in Figure 4.2.2 8. To sum up,

Figure 4.2.2 8 and Figure 4.2.2 9 do not show clear trends in UV-visible spectra with the addition of chloroauric acid.

One of the procedures used to deposit nanorods on Au(111) involved heating the nanorod solution to boiling and dropping the hot solution on the Au(111) substrate. In order to investigate the effects of gold on porphyrin at elevated temperatures two

141 solutions were prepared: one with 5 µM H2(H4TSPP) and 25 µM chloroauric acid, and a control with just 5 µM H2(H4TSPP). Both of these solutions were boiled for 1 hr with the

UV-visible spectrum collected before boiling, after 30 min, and after 1 hr of boiling. The solutions were allowed to cool before UV-visible spectra were run. The control solution,

(5 µM H2(H4TSPP) only) exhibits two peaks; one at 434 nm and one at 413 nm. Two peaks are present because the pH of the solution (4.62) is close to the diacid pKa of 4.5.61

The pH of the solution with H2(H4TSPP) and chloroauric acid is low enough (4.2) that only the diacid peak is present.

Figure 4.2.2 10: UV-visible spectra of monomeric H2(H4TSPP) heated in the presence of chloroauric acid. A reference spectrum of monomeric H2(H4TSPP) with no chloroauric acid is included.

Intensity shifts between the two peaks in the control spectrum, but there is no apparent loss of total intensity. The gold/H2(H4TSPP) spectrum is very different; there is clear intensity loss from the diacid peak with concurrent growth of a new peak at 406 nm

142 as the solution is boiled. 406 nm is the literature value for the Soret band of

Au(III)TSPP. The data indicates that the H2(H4TSPP) is metallated at high temperatures.

Other issues aside from metallation are raised by the data in Figure 4.2.2 10. Using extinction coefficients the total concentration of porphyrin can be tracked throughout the

- reaction. The extinction coefficients of H2(H4TSPP) and Au(III)TSPP are 443000 M

1cm-1 @ 434 nm143 and 370000 M-1cm-1 @ 406 nm167 respectively. Initially the concentration of H2(H4TSPP) is 5.63 µM. After boiling for 1 hr the concentration falls to

1.72 µM, while the Au(III)TSPP concentration rises to 1.35 µM. The sum of these two concentrations is 3.07 µM, 2.56 µM less than the initial concentration of 5.63 µM.

Almost half of the initial concentration is unaccounted for. The rise of the 406 nm peak and the loss of half the total porphyrin concentration make the heating process a problematic method for aggregate deposition.

We have demonstrated that Au(III) has a clear effect on H2(H4TSPP) at elevated temperatures. We also wish to investigate the effect of metallic gold on H2(H4TSPP).

The distinction is important because the only form of gold the porphyrins are exposed to during sample preparation is a Au(111)/mica substrate. In this experiment a 5 µM

2 H2(H4TSPP)/0.75 M HCl nanorod solution was made by the standard protocol. 6 cm of gold foil previously cleaned by soaking in dilute nitric acid and rinsed in ethanol was added to a cuvette with 3 mL of the nanorod solution. A spin bar was added as well and the sample was continuously stirred. UV-visible spectra were acquired of the nanorod/Au foil sample and a control sample at regular intervals. The data is presented in Figure 4.2.2 11:

143

Figure 4.2.2 11: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au foil. Reference spectra of 5 µM H2(H4TSPP)/0.75 M HCl with no Au foil are included.

The data is best dealt with in sections. We will consider the region from 200 to

350 nm first. There are no porphyrin peaks in this region. HAuCl4 exhibits a ligand to metal charge transfer band at ~ 310 nm in acidic solution.168 As the pH of a solution

- containing the HAuCl4 ion increases the chloride ligands are replaced by OH and the charge transfer band fades to a rising background. Below pH 3.40 the band at 310 nm is strong and does not begin to noticeably fade until pH 4.3. Since the nanorod solution is highly acidic (0.75 M HCl) and has many available chloride ions it is reasonable to expect that the Au(III) would be complexed with chloride. No peak is visible at 310 nm in either spectrum; either there is no Au(III) in solution, it is too dilute to see, or the

Au(III) is complexed with an ion other than chloride. The last possibility can most likely be dismissed due to the high chloride ion concentration and low pH of the nanorod

144 solution. The control spectra show a dip at ~ 240 nm which is not present in the nanorod/Au foil sample. The origin of this discrepancy is not known.

Figure 4.2.2 12: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au foil in the range of 350 nm to 200 nm. Reference spectra of 5 µM H2(H4TSPP)/0.75 M

HCl with no Au foil are included.

In order to investigate the dip at 240 nm in the nanorod with no Au foil solution a

UV-visible spectrum of 5 µM HAuCl4/0.75 M HCl was run. Figure 4.2.2 13 compares the UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl without Au foil, 5 µM

H2(H4TSPP)/0.75 M HCl with Au foil, and 5 µM HAuCl4/0.75 M HCl. The 310 nm charge transfer band discussed in ref 168 is barely visible in the 5 µM HAuCl4/0.75 M

HCl spectrum, but there is a peak at ~230 nm which was not discussed in ref 168. It is difficult to make a conclusive statement about the presence of Au(III) in solution. The

310 nm peak is weak at low concentrations making it difficult to see. The much stronger

145 peak at 230 nm does not fit exactly in the dip at 240 nm in the gold free nanorod solution spectrum. In summary, based on Figure 4.2.2 12 and Figure 4.2.2 13 we cannot confirm the presence of Au(III) in solution. We cannot rule it out at small concentrations either.

Figure 4.2.2 13: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl with Au foil and 5

µM HAuCl4/0.75 M HCl in the range of 350 nm to 200 nm. A reference spectrum of 5

µM H2(H4TSPP)/0.75 M HCl with no Au foil is included.

Analysis of the Q-band region is straightforward. In all cases the spectra with Au foil are lower than the control spectra. Interestingly, the control spectra are constant while the spectra with Au foil drop markedly from the 30 min spectrum to the 60 min spectrum. The reason for the difference is not clear from the UV-visible data. The aggregate responsible for the 707 nm band may be depositing on the Au foil, or the porphyrin is being broken down in a similar fashion as Figure 4.2.2 10.

146

Figure 4.2.2 14: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au foil in the range of 750 nm to 600 nm. Reference spectra of 5 µM H2(H4TSPP)/0.75 M

HCl with no Au foil are included.

The monomer/424 nm band section of the UV-visible spectrum shows two interesting things. First, there is no peak at 406 nm which would be indicative of

Au(III)TSPP. Second, the ratio of monomer to 424 nm band is different with respect to the presence of Au foil. This is an interesting difference because both the nanorods with

Au foil and the control sample were pulled from the same nanorod solution. The data in

Figure 4.2.2 15 indicates that the presence of metallic Au increases the amount of monomer in a nanorod solution.

147

Figure 4.2.2 15: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au foil in the range of 460 nm to 400 nm. Reference spectra of 5 µM H2(H4TSPP)/0.75 M

HCl with no Au foil are included.

Analysis of the 490 nm band region is also straightforward. In all cases save one the spectra with Au foil are lower than the control spectra. The exception is the 30 min sample with Au foil. The spectra with Au foil drop markedly in intensity from 30 to 60 min.

148

Figure 4.2.2 16: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl over time with Au foil in the range of 500 nm to 460 nm. Reference spectra of 5 µM H2(H4TSPP)/0.75 M

HCl with no Au foil are included.

In order to be more quantitative about the decrease in UV-visible spectra over time peak intensities were graphed as a function of time. The 490, monomer, and 424 nm band spectra were deconvoluted using CASA software.

149

Figure 4.2.2 17: Plots of peak positions in UV-visible spectra of 5 µM H2(H4TSPP)/0.75

M HCl over time with and without Au foil.

The 490 nm band graph in Figure 4.2.2 17 shows a similar profile for the spectra with and without Au foil. Both sets of data drop quickly over the first two hours followed by a small drop overnight. The data could be explained by the aggregate depositing on the cuvette walls and Au foil, if present. The spectra with Au foil would be lower in absorbance due to the extra surface area provided for deposition by the Au foil.

In contrast the two sets of data in the Q-band graph show different shapes. The trace with

Au foil is similar to both of the 490 nm band data sets; a steep drop over two hours, and much less intensity loss overnight. Unlike the 490 nm band the Q-bands behave much differently without Au foil. Over the course of the experiment the intensity of the Q- bands without Au foil does not drop much compared to the solutions with Au foil. The deposition explanation postulated for the 490 nm band data does not explain the difference between the two Q-band traces.

150

424 nm H-band

The monomer band traces are similar to the Q-band without Au foil. Without Au there is a small decrease in intensity, while the spectra with Au foil looks constant. The simplest explanation is that the monomer is stable in solution and does not deposit on the quartz or

Au. The 424 nm band traces are closest to the 490 nm band behavior. The species responsible for the 424 nm band appears to be depositing on the cuvette walls and on the

Au foil.

We now turn to a discussion of the interaction of Au foil with H2(H4TSPP) at elevated temperatures. One of the procedures used to deposit nanorods on Au(111) involved heating the nanorod solution to boiling and dropping the hot solution on the

Au(111) substrate. In order to mimic this procedure in a way that can be analyzed by

UV-visible spectroscopy a solution of 5 µM H2(H4TSPP)/0.75 M HCl was boiled under reflux for 10 min and allowed to cool. When the solution reached 90°C 6 cm2 of Au foil was added to the solution. After 1 hour the Au foil was removed and the solution was checked by UV-visible spectroscopy.

151

Figure 4.2.2 18: UV-visible spectra of 5 µM H2(H4TSPP)/0.75 M HCl solution before and after reflux and addition of 6 cm2 Au foil.

The UV-visible spectrum of nanorods refluxed with Au foil does not show the peak at

406 nm indicative of Au(III)TSPP. The startling observation is the conspicuous loss of intensity in the spectrum after 1 hr exposure to Au foil. The two most likely interpretations to the data in Figure 4.2.2 18 are that the nanorods are depositing on the

Au foil or the process of heating and exposure to Au destroys the porphyrin. The nanorod solution in Figure 4.2.2 18 and a 0.75 M HCl solution treated with Au foil in the same fashion as the nanorods were analyzed for Au content on an Agilent 7700 ICP-MS at the WSU GeoAnalytical Lab (GeoAnalytical Lab at Washington State University).

The solution with porphyrin was 22.2 ppb (1.12x10-7 M) in Au while the HCl solution

-8 was 4.20 ppb (2.18x10 M) in Au. Using the initial concentration of H2(H4TSPP) (5 µM) there is about 1 Au molecule per 40 porphyrins. Using this data an upper limit on the concentration of Au(III)TSPP can be estimated. If every Au atom in the solution inserted

152 into a porphyrin the resulting absorbance at 406 nm would be 4x10-2 (based upon the extinction coefficient); too low to be seen in the UV-visible spectra in Figure 4.2.2 18. It is interesting that the porphyrin solution is much higher in Au than the HCl solution. It may be that the porphyrin plays a part in the dissolution of the Au foil. Another possibility is that the difference in Au concentration is indicative of the reproducibility

(or lack thereof) of the experiment. The same amount of foil was used in both experiments, but upon addition of the foil to the hot solution it settled to the bottom of the flask. Depending on how the foil settled at the bottom of the flask a different amount of surface area may have been exposed to the solution, changing the rate of Au dissolution.

ICP-MS testing has confirmed the presence of Au in the refluxed nanorod solution, but at low concentrations.

Having demonstrated the dissolution of Au into HCl solution at elevated temperatures STM images of Au(111)/mica and HOPG substrates treated with 0.75 M

HCl (only) for the 1 hr were taken in order to explore the effects of HCl on microscopy substrates. Figure 4.2.2 19 shows STM images of peeled HOPG and HOPG treated with

0.75 M HCl for 1 hr.

153

Figure 4.2.2 19: UHV-STM images of peeled HOPG (left, setpoint 100 pA at -0.05 V sample bias) and HOPG treated with 0.75 M HCl for 1 hr (right, 30 pA at -0.05 V sample bias).

Treatment of HOPG with HCl does not alter the atomic structure of the graphite. XPS studies of HOPG treated with HCl do not show the presence of chlorine. The situation is very different for Au(111) treated with HCl. XPS results show a surface reaction between HCl and Au(111) and STM images of Au(111) treated with HCl show a loss of reconstruction lines and a roughening of the Au surface. We attribute the apparent roughness to a reaction of HCl with Au over the course of the 1 h treatment. Figure 4.2.2

20 shows STM images of Au(111) and Au(111) treated with 0.75 M HCl for 1 hr.

154

Figure 4.2.2 20: UHV-STM images of annealed Au(111) (left, setpoint 1 pA at 1.6 V sample bias) and Au(111) treated with 0.75 M HCl for 1 hr (right, setpoint 1 pA at 1.6 V sample bias).

The Au(111) reconstruction can be clearly seen in the left pane of Figure 4.2.2 20. After treatment with HCl the reconstruction pattern disappears and the surface is noticeably roughened. A high resolution STM image of HCl treated Au(111) is presented in Figure

4.2.2 21.

155

Figure 4.2.2 21: UHV-STM image of Au (111) treated with 0.75 M HCl for 1 hr (setpoint

1 pA at 1.6 V sample bias).

The roughening of the Au(111) is consistent with the ICP-MS data showing dissolution of Au into HCl solution. Taken together the ICP-MS data and the STM images show that treatment of the Au(111) substrate with HCl etches the surface and greatly impacts the substrate‟s surface morphology liberating gold atoms into the solution. It is particularly important to note that the STM data indicates that heating is not necessary for modification of a gold surface.

In order to further investigate the effects of gold on the porphyrin system solutions of varying acid concentration, porphyrin concentration, and gold concentration were prepared and cast onto both HOPG and Au(111). These samples were scanned by

STM and histograms prepared of the sizes of the objects on the surface. Figure 4.2.2 22

156 shows data from three different samples scanned in UHV: 5 µM H2(H4TSPP)/0.75 M

HCl deposited for 60 min on Au(111) (Black), 5 µM H2(H4TSPP)/0.75 M HCl deposited for 60 min on HOPG (Green), 1 µM H2(H4TSPP)/0.75 M HCl deposited for 40 min on

HOPG (Red). The sizes of disks on HOPG run from ~3-4.5 nm, while the disks on

Au(111) range from ~5.5 to 7 nm. There is a clear difference in the sizes of the small structures with respect to the presence of gold. When H2(H4TSPP) is deposited on

Au(111) and scanned in UHV the average size of disks on the surface is bigger than the disks on HOPG.

Figure 4.2.2 22: Histogram of disk widths observed on various substrates by UHV-STM.

5 µM H2(H4TSPP)/0.75 M HCl deposited for 60 min on Au(111) (Black), 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 60 min on HOPG (Green), 1 µM

H2(H4TSPP)/0.75 M HCl deposited for 40 min on HOPG (Red).

157

Figure 4.2.2 23 shows data from several samples scanned by ambient STM. In all cases the substrate is HOPG. The samples are as follows: 5 µM H2(H4TSPP)/0.75 M HCl deposited for 20 min (Red), 5 µM H2(H4TSPP)/0.75 M HCl aggregated in 5 µM HAuCl4 deposited for 20 min (Blue), 5 µM H2(H4TSPP)/0.75 M HCl with 5 µM HAuCl4 added after aggregation deposited for 20 min on HOPG (Green), 5 µM H2(H4TSPP)/0.75 M

HCl with 2.5 µM HAuCl4 added after aggregation deposited for 20 min (Black). The data in red has no gold; every other data set has HAuCl4 as a component. Figure 4.2.2 23 shows the same trend as Figure 4.2.2 22. When Au is present in a solution of nanorods the average size of the disks increases.

Figure 4.2.2 23: Histogram of disk widths observed on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 20 min (Red), 5 µM H2(H4TSPP)/0.75 M HCl aggregated in 5 µM HAuCl4 deposited for 20 min (Blue), 5 µM H2(H4TSPP)/0.75 M HCl with 5 µM HAuCl4 added after aggregation deposited for 20 min on HOPG (Green), 5

158

µM H2(H4TSPP)/0.75 M HCl with 2.5 µM HAuCl4 added after aggregation deposited for

20 min (Black).

Figure 4.2.2 24 is the same as Figure 4.2.2 23 with the addition of one data set

(purple). The new data set is a 20 min deposition on HOPG of the nanorod solution which was treated with Au foil (the solution from Figure 4.2.2 18). The average disk size measured in this data is 5.1 nm. The data for the solution which interacted with Au foil is more like the samples with chloroauric acid (all data except red) than the Au free data

(red). Although the data is not as clear cut as the UHV histogram Figure 4.2.2 24 indicates that the Au foil has a similar effect on disk size as chloroauric acid.

Figure 4.2.2 24: Histogram of disk widths observed on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 20 min (Red), 5 µM H2(H4TSPP)/0.75 M HCl aggregated in 5 µM HAuCl4 deposited for 20 min (Blue), 5 µM H2(H4TSPP)/0.75 M HCl

159 with 5 µM HAuCl4 added after aggregation deposited for 20 min on HOPG (Green), 5

µM H2(H4TSPP)/0.75 M HCl with 2.5 µM HAuCl4 added after aggregation deposited for

20 min (Black), 5 µM H2(H4TSPP)/0.75 M HCl exposed to Au foil for 1 hr (Purple).

Histograms provide good insight into the size distribution of disks on surfaces, but it is important to consider the images from which the data were taken. Figure 4.2.2

25 shows a 100 nm ambient STM image of 5 µM H2(H4TSPP)/0.75 M HCl deposited for

20 min on HOPG. No gold is present in this sample. Several disks are marked with widths. The marked sizes range from 2.79 to a little over 5 nm which is in line with the data in Figure 4.2.2 24.

Figure 4.2.2 25: STM image of disks on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 20 min setpoint 1 pA at 1.6 V sample bias. No gold is present in this sample.

160

Figure 4.2.2 26 shows a 100 nm ambient STM image of 5 µM H2(H4TSPP)/0.75 M HCl with 5 µM HAuCl4 added after aggregation was completed deposited for 20 min on

HOPG. Several disks are marked with widths. The marked sizes range from 3.76 to ~ 6 nm which is in line with the data in Figure 4.2.2 24. We can see from this STM image that the average disk size increases with gold content.

Figure 4.2.2 26: STM image of disks on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl aggregated in 5 µM chloroauric acid deposited for 20 min

(setpoint 1 pA at 1.6 V sample bias).

Figure 4.2.2 27 shows a 100 nm ambient STM image of 5 µM H2(H4TSPP)/0.75

M HCl aggregated in the presence of 5 µM HAuCl4 deposited for 20 min on HOPG.

Several disks are marked with widths. The marked sizes range from 4.3 to ~ 5.4 nm

161 which is in line with the data in Figure 4.2.2 24. We see the same trend here again.

Exposure to Au increases disk size.

Figure 4.2.2 27: STM image of disks on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl. 5 µM chloroauric acid was added after aggregation was complete. Deposition time was 20 min. setpoint 1 pA at 1.6 V sample bias.

Figure 4.2.2 28 shows a 100 nm ambient STM image of 5 µM H2(H4TSPP)/0.75

M HCl with 5 µM HAuCl4 added after aggregation deposited for 20 min on HOPG. This picture shows several nanorods lying next to each other. This picture is worthy of discussion because the nanorod appears to be made up of disks. Sizes appear to be at most ~5.5 nm. This picture is interesting because efforts to visualize the disk structure seen on Au(111) on HOPG have failed in the absence of gold. This image is the closest we have come to repeating the data on Au(111), 5 µM HAuCl4 was present in the sample.

162

Figure 4.2.2 28: STM image of several nanorods on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl. 5 µM chloroauric acid was added after aggregation was complete deposited for 20 min. setpoint 1 pA at 1.6 V sample bias.

Smaller images show an interesting structure in the nanorods. Consider Figure 4.2.2 29 which is a magnification of a section of the rod in Figure 4.2.2 28.

163

Figure 4.2.2 29: STM images of nanorods on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl. 5 µM chloroauric acid was added after aggregation was complete deposited for 20 min. setpoint 1 pA at 1.6 V sample bias.

It appears that the nanorod is made of disks with varying sizes some of which are marked with distances. The largest disks approach 6 nm; the smallest are ~ 3 nm. Figure 4.2.2

29 strikes at the heart of the 16-mer model. Either the smaller structures are a part of the rod or they are not. The disks do not look like snow on top of the rod; it looks like the rod is composed of the disks. If that is the case the 16-mer model is not correct because it does not predict disks of varying sizes. The smallest disks would be difficult to explain by a circular meso aggregate because of their small size. A possible explanation would be a fracturing nanorod. I(V) data indicates that the nanorod conduction band is delocalized over many monomers. If, upon deposition and desiccation, the nanorod begins to break apart it is possible that it would divide into pieces of varying size. Some might be tetramers, or dimers, or monomers, each with a different size. If the tetramers and dimers are still electronically coupled it may not be possible to resolve the

164 constituent molecules. It may be that Figure 4.2.2 29 shows a nanorod beginning to break apart into monomers and smaller coupled moieties.

Figure 4.2.2 30 shows a 100 nm STM image of 5 µM H2(H4TSPP)/0.75 M HCl deposited for 20 min on HOPG. This is the nanorod solution which was heated with Au foil. Several disks are marked with widths. The marked sizes range from 4.6 to ~ 5.4 nm which is in line with the data in Figure 4.2.2 24. Again, in the presence of Au disk size increases.

Figure 4.2.2 30: STM image of several disks on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 20 min. This sample was heated with Au foil in it

(setpoint 2 pA at 1.6 V sample bias).

165

It is important to emphasize that, at this point, we do not know what the disks are.

We also have not confirmed that they are all the same species. We have learned that chloroauric acid and heating the nanorod solution with Au foil in it have similar effects on disk size, namely increasing it. Figure 4.2.2 29 is the closest we have come to reproducing the disk structure seen on Au(111). That sample was 5 µM in chloroauric acid.

In order to address the possibilty that the disks are a product of a ripening process on the HOPG surface 5 µM H2(H4TSPP)/0.75 M HCl was deposited on HOPG for 20, 30, and 60 min. No gold was present in this system. The samples were scanned in air and a histogram of the disk sizes was prepared. The data is presented in Figure 4.2.2 31. The data in the three samples is grouped around ~4 nm, only a very few data points can be seen around 6 nm. This data does not show a clear increase in the size of disks with deposition time in the absence of Au.

166

Figure 4.2.2 31: Histogram of disk widths observed on HOPG by ambient STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 60 min (Green), 5 µM H2(H4TSPP)/0.75 M HCl deposited for 30 min (Red), 5 µM H2(H4TSPP)/0.75 M HCl deposited for 20 min (Black).

We also wished to compare data on HOPG taken in UHV to data taken under ambient conditions. Figure 4.2.2 32 shows data from several different samples: 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 60 min scanned in air (Green), 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 60 min scanned in UHV (Red), 1 µM

H2(H4TSPP)/0.75 M HCl deposited for 40 min scanned in UHV (Black). The disk sizes from the ambient nanorod, UV nanorod, and UHV monomer samples are clustered around ~3.5nm, far short of 6 nm.

167

Figure 4.2.2 32: Histogram of disk widths observed on HOPG by STM. 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 60 min scanned in air (Green), 5 µM

H2(H4TSPP)/0.75 M HCl deposited for 60 min scanned in UHV (Red), 1 µM

H2(H4TSPP)/0.75 M HCl deposited for 40 min scanned in UHV (Black).

Figure 4.2.2 33 is similar to Figure 4.2.2 31. Nanorod solution was deposited on

Au(111) for various lengths of time and scanned under ambient conditions. The data is not clear cut. The 10 min sample (red) is skewed to small features compared to the 60

(black) and 120 min (blue) samples. As the time the nanorod solution spends in contact with the Au(111) surface the size distribution seems to narrow and the average increases.

168

Figure 4.2.2 33: Histogram of disk widths observed on Au(111) by ambient STM. 10

µM H2(H4TSPP)/0.3 M HCl deposited for 10 min (Red), 10 µM H2(H4TSPP)/0.3 M HCl deposited for 60 min (Black), 10 µM H2(H4TSPP)/0.3 M HCl deposited for 120 min

(Blue).

It would be helpful to correlate our STM data with another microscopic technique. To that end 5 µM H2(H4TSPP)/0.75 M HCl with no Au present was deposited on a carbon coated Ni Formvar TEM grid for 20 min followed by spin drying for 30 sec at 4000 rpm. These are the same deposition conditions and concentration as were used to investigate the disks on HOPG. The sample was then imaged by TEM.

169

Figure 4.2.2 34: TEM image of 5 µM H2(H4TSPP)/0.75 M HCl deposited on a carbon coated Ni Formvar TEM grid for 20 min followed by spin drying for 30 sec at 4000 rpm.

Figure 4.2.2 34 shows a representative TEM image of nanorods on the carbon coated grid. Single rods are difficult to see due to poor contrast between rod and substrate.

Better contrast can be seen in bundles of rods. The nanorod solution which was heated in the presence of Au foil was deposited on for the same length of time on an identical TEM grid and imaged. A large scale image is shown in Figure 4.2.2 35.

170

Figure 4.2.2 35: TEM image of 5 µM H2(H4TSPP)/0.75 M HCl treated with Au foil deposited on a carbon coated Ni Formvar TEM grid for 20 min followed by spin drying for 30 sec at 4000 rpm.

There is much more bundling of the nanorods in the sample treated with Au foil than without Au foil. We are not certain if the increased bundling is due to the Au foil treatment. The more interesting feature is the presence of disks in certain parts of the rod bundles. This type of structure was not observed by TEM without the Au foil treatment.

The inset of Figure 4.2.2 35 shows a magnified section of the nanorod bundle. In this section of the bundle some of the rods show a disk-like morphology while some do not.

Disks can be clearly seen in the rod running down the middle of the bundle. Disk sizes range from ~ 6-10 nm. This morphology was seen in several different sections of the

171

TEM grid and on different samples. Figure 4.2.2 36 shows a smaller image of the intersection of the two rod bundles. The same type of morphology can be seen in Figure

4.2.2 36; some of the rods have disks and some do not.

Figure 4.2.2 36: TEM image of 5 µM H2(H4TSPP)/0.75 M HCl treated with Au foil deposited on a carbon coated Ni Formvar TEM grid for 20 min followed by spin drying for 30 sec at 4000 rpm.

Figure 4.2.2 37 is the highest magnification TEM image of a nanorod bundle we have taken. In this image the “disks” look more like rectangles than true disks. The diameter of the structures ranges ~ 5-10 nm, about twice as big as the disks seen in STM. The

TEM images of H2(H4TSPP) with and without exposure to Au foil do not lend support to

172 the 16-mer theory of H2(H4TSPP) aggregation. Diskotic structures are not seen in TEM images of nanotubes without exposure to Au foil; it is only when Au foil is added to the aggregation system that disks are seen.

Figure 4.2.2 37: TEM image of 5 µM H2(H4TSPP)/0.75 M HCl treated with Au foil deposited on a carbon coated Ni Formvar TEM grid for 20 min followed by spin drying for 30 sec at 4000 rpm.

Electron Dispersive Spectroscopy (EDS) was run on nanotubes deposited from the solution treated with Au foil. This technique is capable of giving information on elemental composition. An EDS spectrum was acquired over a bundle of tubes covered with the disks. The scan shows the presence of C, Cr, Si, Fe, Ni, Cu, and Cl. No Au was

173 detected. Ni and C are not surprising as the grid is made of Ni and C. The Cr, Si, Fe, and

Cu are surprising; it may be that the EDS is detecting the alloys in the sample holder.

There is a strong Cl signal but S, O, and N are absent. Since the EDS is not detecting the component elements of the nanotubes save carbon the lack of an Au signal should not be interpreted as conclusive evidence of the absence of Au.

Figure 4.2.2 38: EDS spectrum of 5 µM H2(H4TSPP)/0.75 M HCl treated with Au foil deposited on a carbon coated Ni Formvar TEM grid

174

We have reported on a number of experiments on the relationship between the disks observed in STM images of the H2(H4TSPP) aggregate system. In the presence of either metallic Au or chloroauric acid the size of the disks increases. TEM images show that disks are present only when the aggregate solution was exposed to Au foil. We have not been able to identify the structure of the disks, but the data presented in this section indicates that the presence of Au is intimately correlated with their size; with increasing gold exposure leading to larger disks. After analysis of the data it is likely that the disk model of H2(H4TSPP) aggregation is not correct and the disks are a result of a side reaction with Au.

4.3 Characterization of Tetrasulfonatophenyl Porphyrin and its Aggregate by

Raman and Resonance Raman Spectroscopy

Before presenting Raman data on TSPP and its aggregate it is worthwhile to cover the naming convention which will be used to identify normal modes. Figure 4.3 1 shows the system which will be used to keep track of vibrational modes.

175

Figure 4.3 1: Atom labels used in the vibrational assignments of TSPP.

Diacid H2(H4TSPP) has 94 atoms and with 3N – 6 vibrational modes there are 276 normal modes to keep track of. Table 4.3 1 shows vibrational mode assignments for the three forms of TSPP: free base, diacid and aggregate.

Free Diacid Aggregate Assignment Reference Base Monomer

N/A 234 240 (soln) oop Cm-φ 72 (soln)

245 (Au)

312 316 314 (soln) pyr tilt 72

316 (Au)

N/A N/A 360 pyr tilt 169

N/A N/A 453 pyr rot 171

N/A N/A 550 γ(Cα-Cm) 171

N/A N/A 580 γ(Cα-Cm) 171

623 Weak 620 phenyl 171

176

733 703 700 δ(N-Cα- Cm)/ υ(Cα-N) 77

806 N/A 806 pyr fold 171

N/A N/A 820 δ(pyr def) 171

885 N/A 880 δ(pyr def) 171

N/A ~925 (broad) ~915 (broad) phenyl 171

965 993 984 υ(pyr breath) 170

1003 1015 1015 υ(pyr breath) 170

1084 1083 ~1082 δ(Cβ-H) 172

1124 1120 ~1120 δ(Cβ-H) 77

1234 1238 1229 (soln) υ(Cm-φ) 172

1231 (Au)

1293 N/A N/A υ(pyr half-ring) 170

N/A 1327 1320 υ(pyr quarter-ring) 171

1364 1370 1355,1380 (soln) υ(pyr half-ring) 170

1340 (Au)

1440 1428 1428 phenyl 171

N/A 1476 1477 υ(Cα-Cm) 171

1549 1539 1536,~1530 sh (soln) υ(Cβ-Cβ) 171

1538,~1530 sh (Au)

N/A 1564 1561 υ(Cα-Cm) 171

1601 1600 1591 Phenyl 172

Table 4.3 1: Vibrational mode positions (Δυ in cm-1) and assignments of TSPP.

177

Figure 4.3 2 and Figure 4.3 3 show solution phase Raman data for the free base and diacid forms, each excited at a wavelength near the maximum in the respective absorption spectrum. Polarized and depolarized Raman spectra were recorded. The

4- -1 strongest Raman mode in the spectrum of H2TSPP , at 312 cm , is an out-of-plane motion of the porphyrin core, while that of the diacid is the totally symmetric Cm-phenyl

-1 -1 stretch at 1238 cm , followed closely by the band at 1539 cm assigned to Cβ-Cβ stretching. The mode at 234 cm-1 in the spectrum of the diacid is absent in the spectrum of the free base. Frequency shifts on protonation of the pyrrole nitrogens are observed.

For example, the blue shift in the totally symmetric pyrrole breathing vibration from 965 cm-1 in the free base to 993 cm-1 in the diacid is similar to what is observed in the Raman spectrum of tetraphenylporphyrin and its diacid.171 The diacid mode at 1476 cm-1 is not observed in the free base, similar to what was found for TCPP.64 This mode is tentatively

-1 assigned to the in-plane motion of the phenyl ring (the ν13 mode found at 1480 cm in benzene) which becomes resonance enhanced in the diacid as a result of the phenyl rings being more coplanar with the porphyrin ring.173

178

Figure 4.3 2: SS (black) and SP (red) polarized Resonance Raman Spectra of free base

4- H2TSPP . The excitation wavelength is 413.1 nm. A Gaussian background has been subtracted from the data. The solution pH is 9.68.

Depolarization ratios, ρ, of porphyrins with D4h symmetry are expected to fall into three categories: 1/8 for totally symmetric (A1g) vibrations, 3/4 for non-totally symmetric

174 B1g and B2g vibrations, and infinite for anomalously polarized A2g vibrations. In the case of resonance with a single nondegenerate excited electronic state, ρ is expected to be

1/3 for totally symmetric modes. Depolarization ratios for the lower symmetry free base and diacid monomers may be complicated by vibronic coupling leading to depolarization dispersion. Both the free base and the diacid have formal symmetry that is lower than

D4h, which could lift the degeneracy of the excited electronic state, and in the case of the aggregate, magnetic circular dichroism spectra clearly indicate the degeneracy is lifted.72

The Soret band of the free base, however, is not significantly split. By using an

179 excitation wavelength very near its maximum at 413 nm, depolarization ratios typical for resonance with a doubly degenerate excited electronic state are observed. This results in values of ρ for the free base which are close to 1/8 as revealed in Figure 4.3 2, consistent with expected strong resonance Raman activity of totally symmetric modes.

In contrast, the diacid Raman modes display a range of depolarization ratios including ρ near 3/4 in the case of the 1477 and 1540 cm-1 modes. Although depolarization ratios for the diacid are less certain than those for the free base owing to weaker signal, the range of observed ρ values for the former suggests symmetry lowering as expected for out of plane distortion of the protonated pyrrole rings. Also, the excitation wavelength is intermediate between the wavelengths of the B and those of the

Q-bands resulting in depolarization ratios for some modes being perturbed by vibronic coupling of the B and Q-band excited electronic states. Thus, both the Raman and the optical absorption spectra point to lower symmetry in the diacid compared with the free base.

180

Figure 4.3 3: SS (black) and SP (red) polarized Resonance Raman Spectra of diacid

2- H2TSPP . The excitation wavelength is 457 nm. A Gaussian background has been subtracted from the data. The solution pH is 3.94.

The resonance Raman spectrum of the H2(H4TSPP) aggregate in solution is presented in Figure 4.3 4. The low-frequency out-of-plane modes are the most intense, as is generally observed in the resonance Raman spectra of porphyrin aggregates. The appearance of intense low frequency porphyrin modes is also associated with nonplanar distortions,174,175 and may account for the activity of the 245 and 316 cm-1 modes of the diacid monomer. Though the 241 and 314 cm-1 bands have been referred to as “ruffling” and “doming” modes of the porphyrin core, respectively, their precise assignment is

-1 uncertain. The reported red-shift by several cm of both bands in D2O is evidence that

N-H (N-D) motion contributes to this mode.77 Thus, it is reasonable to assume they are out-of-plane vibrations that are strongly coupled to the delocalized electronic transition

181 via perturbation of the interchromophore separation. Depolarization ratios are close to 1/3 for most modes, as expected for excitation resonant with the nondegenerate aggregate excited electronic state. However, the depolarization ratios of the low frequency modes at 240 and 314 cm-1 are significantly greater than 1/3. The reason for this is not clear.

The data reveal only subtle frequency shifts of diacid Raman modes on aggregation but large intensity changes, most notably the increase in relative intensity of the two lowest- frequency modes. Owing to the delocalization of the aggregate excitation over a number of porphyrin units, these out-of-plane vibrations are more strongly coupled to the electronic transition of the aggregate than to that of the monomer diacid. The weak mode found at 706 cm-1 in the diacid monomer is red shifted by a few wavenumbers in the aggregate spectra. In the 380-500 cm-1 range, a number of vibrations associated with the phenyl rings are observed in the aggregate spectra but absent in the free base and diacid spectra.

Theory predicts the J-band of a circular aggregate, which arises from coupling of

B-band transition moments that are parallel to the plane of the ring, to be doubly degenerate.176 This conflicts with the observed Raman depolarization ratios, which deviate from 1/8 for all the bands shown in Figure 4.3 4. On the other hand, a nondegenerate J-band polarized along the nanotube axis can not account for the values ρ

= 0.5 observed for the low-frequency modes.

182

Figure 4.3 4: SS (black), SP (red) and SP x3 (dotted blue) polarized Resonance Raman

Spectra of H2(H4TSPP) nanorods in 1.5 M HCl. The excitation wavelength is 488 nm. A

Gaussian background has been subtracted from the data.

The resonance Raman spectrum of the H2(H4TSPP) aggregate in solution is compared to the surface spectrum of the aggregate on gold in Figure 4.3 5. The spectra are quite similar confirming that deposition of the solution phase aggregates onto the gold substrate generally preserves their structural integrity. In the solution phase aggregate, the relative intensities of the higher frequency skeletal modes appear to be greater than those for the aggregate on gold, but this may be merely the result of greater attenuation of the intensities of the low-frequency modes in the spectrum of the darker solution phase

-1 aggregate. The Cm-phenyl stretch at 1237 cm in the diacid monomer is red shifted by

183 about 8-10 cm-1 in the aggregate Raman spectra, while the mode at 234 cm-1 in the diacid is blue shifted to 240 cm-1 in the solution phase aggregate and 245 cm-1 in the aggregate on gold, suggesting perturbations to the porphyrin core on aggregation which may be slightly greater on gold than in solution. For the most part, the aggregate Raman spectra are similar for the solution phase and gold-deposited samples. A notable exception to this is the mode at 1340-1380 cm-1, assigned to a totally symmetric pyrrole vibration found at 1370 cm-1 in the diacid monomer.

This band appears to split into two bands, at 1355 and 1380 cm-1 in the spectrum of the solution phase aggregate, while only one band at about 1340 cm-1 is observed for the aggregate on gold. It has been noted171 that both the frequency and intensity of this band are very sensitive to acidic species in solution, so minor differences between the aqueous and the gold-deposited aggregate might be a consequence of differences in

-1 protonation. The appearance of a shoulder on the 1540 cm ν(Cβ-Cβ) stretch in the diacid and aggregate spectra is perhaps indicative of a coexisting conformer. We are not aware of any assignment for this feature in the literature.

184

Figure 4.3 5: Resonance Raman Spectra of H2(H4TSPP) nanorods in solution (black) and deposited on Au(111). The excitation wavelength is 488 nm. A Gaussian background has been subtracted from the data.

We have established that nanotube formation causes significant changes to Raman spectra, particularly a marked enhancement in the low frequency modes of the aggregate both in solution and deposited on Au(111) are observed. Comparison of the nanotube solution phase Raman spectrum with the Raman spectrum of the tubes deposited on

Au(111) confirms the intact deposition of the aggregate on Au(111).

185

4.4 X-ray and Ultraviolet Photoelectron Spectroscopy Analysis of TSPP and its

Aggregate

XPS was used to characterize the aggregate starting material. Both the sodium and chloride salt of TSPP were pressed on indium shot and analyzed. Free base

Na4(H2TSPP) was tested for Na, C, S, O, and N. The survey spectrum is shown in Figure

4.4 1:

Figure 4.4 1: XPS survey spectrum of Na4(H2TSPP) powder on In.

The high resolution elemental scans are shown in Figure 4.4 2:

186

Figure 4.4 2: XPS sulfur, carbon, oxygen, and sodium spectra of Na4(H2TSPP) powder on

In.

The scans have good signal to noise, and the 2p1/2/2p3/2 spin orbit splitting can be seen in the sulfur 2p scan. The sulfur peak is located at 167 eV which is the expected position for a sulfonic acid group177. The small broad peak at ~ 162 eV is worth noting, the 162 eV peak may be a result of contaminants in the powder as this is the region where sulfide

178 is found. The diacid H4(H4TSPP)Cl2 powder was also analyzed, the survey is shown in

Figure 4.4 3:

187

Figure 4.4 3: XPS survey spectrum of H4(H4TSPP)Cl2 powder on In.

The high resolution elemental scans are shown in Figure 4.4 4:

188

Figure 4.4 4: XPS sulfur, carbon, oxygen, and chlorine spectra of H4(H4TSPP)Cl2 powder on In.

The scans have good signal to noise. As with the free base the sulfur peak is located at

167 eV which is the expected position for a sulfonic acid group.177 The free base and diacid nitrogen 1s scans are worth discussing separately. They are shown in Figure 4.4 5:

189

Figure 4.4 5: Nitrogen 1s XPS spectra for free base and diacid powders on indium.

The free base scan shows two peaks at 400 eV and 398 eV while the diacid spectrum shows only one peak at 400 eV. The two peaks in the free base spectrum are a result of inequivalent nitrogens in the porphyrin macrocycle.

Figure 4.4 6: TSPP wireframes showing the difference in protonation state between free base and diacid.

Free base Na4(H2TSPP) has two different types of nitrogens in its macrocycle, protonated and deprotonated. The different protonation states give rise to the two peaks seen in the

190 free base scan. The peak at 400 eV is associated with protonated nitrogen, the peak at

398 eV arises from deprotonated nitrogen.179 The disparity in peak size is due to the relative amounts of protonated and unprotonated nitrogen. The free base N 1s spectrum is the sum of two spectra, the free base and “contamination” by diacid. If the diacid form is present in the free base powder the 400 eV peak will increase in intensity compared to the 398 eV peak; which is exactly what we see in the free base scan. This mix of two spectra has been previously reported in tetrakis(p-carboxyphenyl)porphyrin.179

The elemental ratios of the free base and diacid were calculated from their respective spectra and the theoretical atomic ratios are presented in the Table 4.4 1:

Experimental Experimental Theoretical Free Theoretical Element Free Base Ratios Diacid Ratios Base Ratios Diacid Ratios

O 3.4 3.8 3.0 3.0

N 1.0 1.0 1.0 1.0

C 8.8 11.0 11.0 11.0

S 1.0 1.0 1.0 1.0

Na 0.93 N/A 1.0 N/A

Cl N/A 0.18 N/A 0.50

Table 4.4 1: Theoretical and experimental elemental ratios in TSPP free base and diacid powders.

For clarity the ratios have been normalized to sulfur. Carbon aside, the elemental ratios in the free base work out as expected. Oxygen is higher than expected, but this can be attributed to waters of hydration. The free base powder is a dodecahydrate; the extra

191 oxygen may simply be coordinated water that has not pumped away in the XPS vacuum.

The carbon content in the free base is puzzling; prima facie we might expect excess carbon due to contamination. Since the free base powder purchased from Alpha Aesar is

95% pure and was not purified before analysis the carbon discrepancy may be due to impurities.

Both the free base and diacid powders were sent to Columbia Analytical Services

(Tucson, Az.) for elemental analysis. The results were reported as mass percentages so the theoretical ratios and the XPS data have been converted to match in Table 4.4 2.

Element XPS Free Base Free Base by External Analysis Theoretical Free Base

O 21.5 22.03 18.69

N 5.54 4.94 5.45

C 41.77 48.25 51.44

S 12.67 9.45 12.49

Na 8.45 8.65 8.96

Cl N/A N/A N/A

H N/A 2.3 2.55

Table 4.4 2: Elemental ratios of Na4(H2TSPP) by XPS, Columbia Analytical Services, and the theoretical values.

Oxygen runs high in both the XPS and the external analysis. The free base powder was dried under vacuum at 125°C prior to elemental analysis at Columbia Analytical

Services. Clearly most of the 12 waters of hydration have been removed by heating but in both XPS and external analysis some of the waters of hydration appear to remain.

192

Nitrogen, sodium, and hydrogen are unremarkable, but sulfur and carbon are troubling.

The XPS data for sulfur is close to the theoretical number, but the external analysis is lower. Carbon is low by both XPS and external analysis. This is very difficult to explain with O, N, and Na being similar. Sulfur differs from XPS to external analysis but not to the degree which carbon deviates.

The diacid salt was also sent for analysis; the mass percentages are below:

Element XPS Diacid Diacid by External Analysis Theoretical Diacid

O 24.02 22.7 18.97

N 5.54 5.29 5.54

C 52.21 54.58 52.21

S 12.67 12.07 12.67

Na N/A N/A N/A

Cl 1.26 0 7

H N/A 3.81 3.19

Table 4.4 3: Elemental ratios of H4(H4TSPP)Cl2 by XPS, Columbia Analytical Services, and the theoretical values.

Just as with the free base oxygen is high but this can be attributed to water. There are no waters of hydration listed on the diacid bottle, but it would not be surprising if some water was present in the powder considering that the free base is sold as a dodecahydrate.

Nitrogen, carbon, and sulfur match nicely in all three cases and hydrogen by external analysis is reasonable. Chlorine is low in the XPS and the external analysis data; this can be accounted for by assuming that the chloride outgases as hydrochloric acid (this sample

193 was heated under vacuum as well). If we assume that two molecules of HCl are lost per porphyrin the theoretical ratios can be recalculated and compared to the experimental data:

Diacid by External Theoretical Diacid Ratios Element XPS Diacid Analysis Minus 2HCl with 2 waters

O 24.02 22.7 22.96

N 5.54 5.29 5.75

C 52.21 54.58 54.18

S 12.67 12.07 13.15

Na N/A N/A N/A

Cl 1.26 0 0

H N/A 3.81 3.51

Table 4.4 4: Elemental ratios of H4(H4TSPP)Cl2 by XPS, Columbia Analytical Services, and the theoretical values.

The addition of two waters per porphyrin brings the theoretical and experimental data closer together. The worst offender is sulfur with an error of 8% between the external analysis and theoretical. To summarize, the free base powder has a serious carbon issue and a comparably smaller sulfur problem. The diacid XPS, external analysis, and the theoretical composition line up much better. No purity is listed on the diacid powder bottle, but the data indicates it is a better starting material than the free base.

194

XPS was also performed on H2(H4TSPP) deposited on Au(111). Prior to deposition the flame annealed Au(111) substrate was checked for contamination, save carbon and oxygen the substrates were contaminant-free by XPS.

Figure 4.4 7: XPS survey spectrum of H2(H4TSPP) aggregates deposited on Au(111).

The nanorod XPS survey is dominated by gold peaks; the C 1s peak is the only non-metal element visible in the survey. The high resolution scans are shown below:

195

Figure 4.4 8: High resolution elemental XPS scans of H2(H4TSPP) nanorods on Au(111).

The carbon, oxygen, and chlorine peaks are unremarkable. The nitrogen spectrum indicates a fully protonated macrocycle as would be expected for H2(H4TSPP) deposited from an acidic solution. The sulfur 2p spectrum is puzzling as it shows two peaks; one at

168 eV and a second at 161 eV. The peak at 168 eV is the expected result for a sulfonic acid group,177 while 161 eV is typical of a sulfide.180

To investigate the shape of the S 2p peak three additional tests were run. 0.75 M

4- HCl, 0.77 M sulfurous acid, and 50 µM free base (H2TSPP) were deposited on Au(111) substrates. XPS of the HCl treated Au(111) analysis showed oxygen, chlorine, and

196 carbon but no sulfur ruling out contamination by the acid. An overlay of the aggregate, free base and sulfurous acid on Au(111) are presented in Figure 4.4 9. The peak positions in the sulfurous acid S 2p scan are similar to the nanorods and free base on

Au(111) spectra. The relative intensities of the spectra are different and the peaks do not overlay exactly but the spectra are clearly similar.

Figure 4.4 9: S 2p XPS spectra of free base (blue), nanorods (red), and sulfurous acid

(black) deposited on Au(111).

The reaction of sulfur dioxide with silver metal has been previously reported.181 Gaseous sulfur dioxide forms sulfurous acid as it dissolves in water on the surface of the silver.

The acid dissociates protons and a series of electrochemical reactions occur resulting in several different sulfur-containing products, which in turn give rise to multiple peaks in

XPS spectra. Based on Figure 4.4 9 it would appear that there is a reaction between the

197 sulfonate groups on the porphyrin and the Au(111) surface which is the cause of the extra sulfur peak in the XPS spectrum.

The XPS data of H2(H4TSPP) nanorods on Au(111) is compared to the experimental XPS data for the starting diacid powder and the theoretical atomic ratios in

Table 4.4 5; as before the data is normalized to sulfur. Sulfur and nitrogen are close to both the theoretical and powder ratios. Oxygen and carbon both run higher than expected. The extra carbon is probably contamination; the extra oxygen is likely some form of water. It is difficult to make a statement about the high chlorine content due to the use of HCl to initiate nanorod aggregation. Based on the nitrogen and sulfur numbers it appears that the composition of the nanorods is consistent with the starting material.

Nanorods on Au(111) Element Theoretical Diacid Ratios Powder Diacid Ratios Ratios O 3.0 3.8 11.6 N 1.0 1.0 1.19 C 11.0 11.0 25.9 S 1.0 1.0 1.0 Cl 0.50 0.18 0.89

Table 4.4 5: Atomic ratios by XPS of the theoretical diacid, diacid powder, and nanorods deposited on Au(111).

H2(H4TSPP) nanorod deposition on HOPG was also analyzed by XPS. The survey spectrum can be seen in Figure 4.4 10.

198

Figure 4.4 10: XPS survey spectrum of H2(H4TSPP) aggregates deposited on HOPG.

The nanorod XPS survey is dominated by the C 1s peak. Oxygen is the only other element visible in the survey. The high resolution scans are shown below:

199

Figure 4.4 11: High resolution elemental XPS scans of H2(H4TSPP) nanorods on HOPG.

The carbon and oxygen peaks are unremarkable. The extra peak in the sulfur 2p spectrum seen in the Au(111) deposited samples is absent reinforcing the idea of a reaction between the Au and sulfonic acid groups. In the case of both HOPG and

Au(111) samples, only a single N 1s peak is observed at 400 eV, clearly signaling that the protonation state that exists in solution carries through to the molecules comprising the nanorods. This reinforces the previously described concept of ionic self-assembly through interaction between the +2 core and adjacent sulfonate groups. Chlorine is absent when HOPG is used as a substrate in contrast to the case on Au(111). In order to

200 better understand this, we performed XPS studies on clean substrates, substrates treated with 0.75 M HCl, and those treated with 5 µM diacid in 0.75 M HCl. In the case of

HOPG, no chlorine was ever detected. In the case of Au(111) a clear chlorine peak of about the same intensity was seen from either HCl only or diacid in HCl treated samples.

When these Au substrate samples were heated to 100C in UHV, the chlorine signal disappeared. We interpret these results to mean that the Au is reacting with HCl to form a thin surface chloride layer that can be decomposed on heating.

Similarly to the nanorod on Au(111) case XPS data of H2(H4TSPP) nanorods on

HOPG is compared to the experimental XPS data for the starting diacid powder and the theoretical atomic ratios in Table 4.4 5; as before the data is normalized to sulfur. Sulfur and nitrogen are close to both the theoretical and powder ratios. Oxygen and carbon both run higher than expected. The extra carbon is not surprising considering the graphite substrate. The extra oxygen is likely some form of water; it is interesting that there is much less water in the HOPG sample than on Au(111). Based on the nitrogen and sulfur numbers it appears that the composition of the nanorods is consistent with the starting material.

201

Element Theoretical Diacid Powder Diacid Ratios Nanorods on HOPG Ratios Ratios

O 3.0 3.8 6.3

N 1.0 1.0 1.2

C 11.0 11.0 178

S 1.0 1.0 1.0

Cl 0.50 0.18 0

Table 4.4 6: Atomic ratios by XPS of the theoretical diacid, diacid powder, and nanorods deposited on HOPG.

Samples of H2(H4TSPP) monomers deposited on HOPG were analyzed by XPS as well. The survey is dominated by the C 1s peak. Oxygen is the only other element visible in the survey.

Figure 4.4 12: XPS survey spectrum of H2(H4TSPP) monomers deposited on HOPG.

The survey of H2(H4TSPP) monomers deposited on HOPG is similar to the nanorod on

HOPG XPS spectrum. The C, S, and O high resolution scans are shown below. Just like

202 the aggregate samples no chlorine was detected in monolayer samples.

Figure 4.4 13: High resolution elemental XPS scans of H2(H4TSPP) monomers on

HOPG.

The carbon, oxygen, and sulfur spectra of the monomer on HOPG are similar to the aggregate samples. The N 1s spectra of the monomer on HOPG are worth discussing separately. Figure 4.4 1 shows the N 1s spectrum from two different monomer samples.

The dominant peak is at 400.2 eV, in line with what was reported for the aggregate on

Au(111) and HOPG. In the red trace there is a small, poorly resolved feature at ~398 eV.

This peak was not seen in the aggregate samples and varies from sample to sample. The red trace is the biggest peak seen at 398 eV in the monomer samples; the black trace is the smallest. 398 eV is the binding energy of an unprotonated macrocycle nitrogen.

203

Unprotonated nitrogen is not likely in a solution as acidic as the monomer so we attribute this feature to a sample preparation or handling effect.

Figure 4.4 14: N 1s spectra of the H2(H4TSPP) monomer deposited on HOPG.

Similarly to the nanorod on Au(111) and nanorod on HOPG cases XPS data of

H2(H4TSPP) monomers on HOPG is compared to the experimental XPS data for the starting diacid powder and the theoretical atomic ratios in Table 4.4 7; as before the data is normalized to sulfur. Sulfur and nitrogen are close to both the theoretical and powder ratios. Oxygen and carbon both run higher than expected. The extra carbon is not surprising considering the graphite substrate. The extra oxygen is likely some form of water. Based on the nitrogen and sulfur numbers it appears that the composition of the monolayer is consistent with the starting material.

204

Element Theoretical Diacid Powder Diacid Monomer on HOPG Ratios Ratios Ratios

O 3.0 3.8 7.8

N 1.0 1.0 1.1

C 11.0 11.0 218

S 1.0 1.0 1.0

Cl 0.50 0.18 0

Table 4.4 7: Atomic ratios by XPS of the theoretical diacid, diacid powder, and nanorods deposited on HOPG.

Table 4.4 8 shows the binding energies of the aggregate on Au(111), HOPG and the monomer on HOPG. The biggest change in binding energy is oxygen with a 0.4 eV shift from the aggregate on Au(111) to the monomer on HOPG. The shift in oxygen binding energy is difficult to interpret because the oxygen signal likely comes from both the porphyrin and water. The nitrogen and sulfur peaks do not shift appreciably from substrate to substrate or upon aggregation.

Element Aggregate on Au(111) Aggregate on HOPG Monomer on HOPG

C 1s 284.8 284.5 284.5

N 1s 400.3 400.4 400.2

O 1s 531.9 531.7 532.1

S 2p 168.3 168.4 168.3

Cl 2p 199.5 N/A N/A

Table 4.4 8: Peak positions in eV of the elements studied by XPS.

205

UPS was performed on both clean graphite and a sample made with HOPG that had been treated with 1 µM diacid. STM results indicated that this treatment produced predominantly monolayer coverage. The secondary electron edge was used to determine the work function of both the clean graphite and the surface formed with a near monolayer.182,183 The monolayer samples were heated to 150ºC in UHV and cooled in

UHV prior to measurement in order to duplicate the conditions used in obtaining the

OMTS. A work function of 4.9±0.1 eV was determined for the monolayer of diacid on

HOPG after heating.

Figure 4.4 15: UPS spectra of peeled HOPG (blue), HOPG treated with HCl (red), and a monolayer of H2(H4TSPP) on HOPG (black).

Treatment of peeled HOPG with acid lowers its workfunction by ~0.3 eV. Based on

STM images and I(V) curves treatment of HOPG with HCl does not change the properties of the surface. The reduction in workfunction may simply be a result of acid- cleaning the HOPG surface. Based on the data in Figure 4.4 15 the acid treated HOPG

206 and the monolayer sample have the same workfunction (within 0.2 eV). This suggests that monolayer deposition has very little effect on workfunction.

4.5 Ultra-High Vacuum STM Studies of H2(H4TSPP) Nanorods

4.5.1 UHV-STM Imaging Studies of H2(H4TSPP) Nanorods

While there are several proposed structures for the rods that are consistent with an ionic self-assembly pattern,50,61 there is no general agreement on exactly how the molecules form the observed rods. It should be noted that there is good evidence that these “rods” are actually collapsed nanotubes. For example, as was mentioned in the literature review Vlaming and coworkers have recently shown Cryo-EM images of these nanostructures in frozen solution that are clearly tubes of 18 nm diameter with walls ~2 nm thick.61 In some of our STM images, we have also seen evidence that the 4 nm thick rods are actually collapsed tubes having 2 nm thick walls. Typical large UHV-STM images of nanorod sections on both Au(111) and HOPG are presented in Figure 4.5.1 1.

All samples were heated to 100°C prior to imaging.

207

Figure 4.5.1 1: UHV-STM images of H2(H4TSPP) nanorods on Au (111) (left, setpoint 1 pA at 1.5 V sample bias) and on HOPG (right, setpoint 1 pA at 1.6 V sample bias).

The image on Au(111) shows a single rod, while the image on HOPG shows two nanorods twinning. Both samples were prepared by depositing 5 µM H2(H4TSPP) in

0.75 M HCl on the respective substrates for 1 hr followed by spin drying. The width of the nanorods deposited on both substrates is similar, with the average nanorod width measured in UHV of 27 nm. The size and shape of the nanorods imaged by UHV-STM agrees with the data from both AFM and ambient STM studies. If one looks at regions off the nanorods on HOPG, one finds islands of monolayers with small clusters of molecules atop the monolayer. The area covered by a monolayer and the cleanliness of the monolayer can be greatly improved (at the cost of decreased nanorod deposition) by decreasing the concentration of porphyrin in the adsorption solution to 1 μM. Detailed

UHV-STM studies of the monolayer regions will be covered in section 4.5.2. This is

208 different from STM images of areas off the rod on the Au(111) samples where the surface is consistently rough (on a 0.1 nm scale) and noisy.

It is important to note, that the height, width, and length of the nanorods deposited on HOPG and on Au(111) were essentially the same. Moreover, I(V) curves (which will be discussed in detail in section 4.5.3) obtained from rods supported on either substrate were the same. We believe that any results reported here for the UHV properties of the nanorods on HOPG are directly transferable to nanorods supported on Au(111).

Figure 4.5.1 2: UHV-STM images of H2(H4TSPP) nanorods deposited on Au(111) (left, setpoint 1 pA at 1.3 V sample bias) and HOPG (right, setpoint 1 pA at 1.6 V sample bias) with cross sections through single nanorods.

209

Both cross sections in Figure 4.5.1 2 show nanorods ~ 30 nm wide and ~ 4 nm tall. As indicated earlier, the “rods” appear to be collapsed nanotubes having a wall thickness of 2 nm which is well illustrated in the cross section of the nanorod on HOPG.

The cross sectional trace and image clearly show a layered structure to the nanorod. The nanorod on HOPG is lying on the boundary of a monomer island and bare HOPG. This can be seen in the cross section as the background height is different on the two sides of the nanorod. The rod is 4 nm tall measured from the monolayer side and 4.5 nm tall measured from the bare HOPG side. The extra 0.5 nm is most likely due to the rod lying on top of the monolayer.

So far AFM and STM studies have proven futile in resolving the question of precise nanorod structure. Figure 4.5.1 3 shows a relatively small scale STM image of a nanorod and a region of monomers. While no internal molecular structure can be seen in the aggregate individual molecules are resolved as well as individual molecular vacancies are observed in the monolayer. Extra molecules and smaller aggregates, possibly dimers or trimers, can be seen on top of the monolayer as well.

210

Figure 4.5.1 3: UHV-STM image of H2(H4TSPP) nanorods and monomers deposited on

HOPG (setpoint 1 pA at 1.6 V sample bias).

In contrast, the internal structure of the nanorod is not resolved despite imaging with a tip sharp enough to distingush individual molecules. Prima facie one would expect a tip capable of resolving monomers to be capable of resolving the internal arrangement of the nanorod. Figure 4.5.1 3 clearly indicates that the this is not the case. We will see in section 4.5.3 that the lack of STM resolution is due to the electronic structure of the aggregate.

4.5.2 UHV-STM Imaging Studies of H2(H4TSPP) Monomers Deposited on HOPG

So far, SPM imaging has focused on the large H2(H4TSPP) nanorods. At the other end of the self-assembly scale, one might expect acidic porphyrins to form two dimensional self-assembled monolayers through hydrogen bonding. Most of the reports on such structures to date have centered on carboxylate substituted tetraphenyl

211 porphyrins. Species with one to four carboxylates (CmTPP, m=1-4; the m=4 species is commonly referred to as TCPP, have been investigated by scanning tunneling

184-188 189,190 microscopy (STM) on Au(111) and on HOPG. While CmTPP can be imaged on Au(111) in either UHV or in the electrochemical environment, it cannot be observed on HOPG unless a second species is used to reduce the mobility of the porphyrin on the

HOPG surface. Nicholls and coworkers used a buffer layer of 5-(octyldecyloxy) isophthalic acid (a monolayer on HOPG) to reduce the mobility of CuTCPP to the point where a structured adlayer could be observed.189 In this case, a very open square structure with a molecular lattice having a spacing of 2.5 nm results from end-on H- bonding between CuC4TPP molecules. Lei et al. coadsorbed TCPP with steric acid in order to get a dried, mixed adlayer in which islands of TCPP could be imaged by STM.

They report two different structures, one with direct-on H-bonding having a 2.3 nm lattice spacing, and a higher density square structure with a 1.8 nm lattice constant where there is a more communal H-bonding between four carboxylates. It is not clear from

Lei‟s data whether the steric acid adlayer merely compresses the TCPP, or if the TCPP adds on top of it as in the isophthalic acid case. Another oddity of most of these papers is that the protonation state of the pyrrole nitrogens is not considered. Borguet appears to be the only author who seriously includes this complication.184 The pKa values of the pyrrole nitrogens and of the carboxylates in TCPP are close to 5.59 Thus, it is extremely difficult to protonate the carboxylates without protonating all the central nitrogens, thereby producing a zwitterionic species. In this section we will report on the monolayer coverage of H2(H4TSPP) on HOPG. We will show that the structure and electronic properties of this system are significantly different than the carboxylate analog.

212

The equilibrium geometry of the diacid, H2(H4TSPP), was determined as shown in Figure 4.5.2 1. The relative positions of the OH groups within the local rotational minima have little effect on the overall energy with a total rotational barrier for one sulfonate relative the fixed geometry of the rest of the molecule being 2.7 kcal/mole.

Thus, when adsorbed on the surface, the OH orientation will be able to easily adjust to give the maximum H-bonding with adjacent sulfonates. The side view of H2(H4TSPP) shows the saddling geometry of the macrocycle resulting from full protonation of the macrocycle nitrogens. The saddled conformation should be visible in high resolution

STM images.

Figure 4.5.2 1: B3LYP 6-31+G(d,p) optimized structure of the H2(H4TSPP) diacid, top and side.

While we believe the structure is reasonably accurate, the computed energies for the isolated molecule and its ions are not appropriate for the monolayer system studied

213 here. In the past, we have demonstrated that the ionization potentials and electron affinities for thin films on metal substrates are much closer in energy to values associated with solution phase redox chemistry than they are to gas phase values.128 It has been previously noted that electron affinities (ionization potentials) and reduction (oxidation) potentials in solution and in the solid state can differ from the gas phase values by as much as 2 eV.128,191-193 Thus, a more appropriate calculation (in the absence of a full multi-adsorbate metal-slab calculation), is the ionization potential and electron affinity for the molecule in a typical non-hydroxylic solvent (we used CH3CN). Both gas phase

vac vac sln sln (IP calc , EA calc) and acetonitrile solution phase (IP calc, EA calc) calculated values for selected porphyrins are displayed in Table 4.5.2 1 and Table 4.5.2 2. Experimental oxidation and reduction potentials are given for comparison in Table 4.5.2 1 and Table

194 4.5.2 2, and the calculated solution phase values (EAsol, IPsol) were obtained by equating the reference SCE potential to 4.71 eV relative to the vacuum level.128 Values of IPg are the experimental gas phase 1st ionization potentials are also shown in Table

4.5.2 1 and Table 4.5.2 2,195,196 and compare reasonably well with the free molecule calculated values for NiTPP and H2TPP. While the calculated values of the solution phase IP and EA for NiTPP and H2TPP do not perfectly match the electrochemical values, the error is very much less than for the gas phase values. Moreover, the small errors are probably attributable to differences in ionic strength and dielectric constants in the electrochemical measurements. Also given in Table 4.5.2 1 and Table 4.5.2 2, and discussed later, are the electron affinity (EAOMTS) and ionization potential (IPOMTS) measured in this work. We note that the calculated IP and EA values for the diacid, even in CH3CN, are too large compared to the measured surface values. This is to be expected

214 since the calculation does not take into account the image charges generated in the substrate which greatly compensate for the large point charges in the zwitterionic diacid.

red b d sln vac ox a Species E1/2 EAsol EAomts EA calc EA calc EAg E 1/2

a h c H2TPP -1.05 3.66 3.80 2.89 1.53 1.69 0.95

NiTPP -1.18a 3.53 3.55 2.76 1.28 1.51c 1.10a

H2(H4TSPP) 3.60 4.15 4.82

Table 4.5.2 1: Electron Affinity values in Different Phases a) Reference 194; b)

E1/2+4.71 V; c) Reference 191; d) The work function for NiTPP/Au(111) taken as 5.2 eV and that for

H2(H4TSPP)/HOPG as 4.9 eV; e) Reference 195; f) Reference 196; g) Reference 192; h)

UPS data from reference 197 assuming the Ag work function was 4.6 eV.

b d sln vac Species IPsol IP omts IP calc IP calc Ipg

h h e,f H2TPP 6.30 6.30 5.29 6.28 6.40

NiTPP 6.50 6.50 5.29 6.46 6.29g

H2(H4TSPP) 6.64 6.64 5.95 6.87

Table 4.5.2 2: Ionization Potential values in Different Phases a) Reference 194; b)

E1/2+4.71 V; c) Reference 191; d) The work function for NiTPP/Au(111) taken as 5.2 eV and that for H2(H4TSPP)/HOPG as 4.9 eV; e) Reference 195; f) Reference 196; g)

Reference 192; h) UPS data from reference 198 assuming the Ag work function was 4.6 eV.

215

In order to more fully investigate the monolayer region, we prepared surfaces using 1 µM diacid in 0.75 M HCl as described in the experimental section. These samples showed very few nanorods and a low density of small aggregates. The HOPG substrates were partially covered with monolayer islands. A region of two such adjacent islands is shown in Figure 4.5.2 2. Also shown in Figure 4.5.2 2 is a cross section taken through a break in the islands indicating that the apparent height was about 0.6 nm at

+1.6 V sample bias. The bright features seen above the monolayer are mostly single molecules (1.5 nm across and 0.6 nm high) which are less well resolved because they are not locked in position as are the monolayer constituents.

Figure 4.5.2 2: STM image of H2(H4TSPP) monolayer on HOPG. 1.6V and 1 pA.

The fact that H2(H4TSPP) easily forms stable islands on HOPG at room temperature immediately sets it apart from the various protonated TCPP species which cannot form stable islands under the same conditions. Since the HOPG-TSPP interaction

216 should be similar to the HOPG-TCPP interaction, one immediately suspects a difference in the hydrogen bonding between TSPP molecules and that between TCPP molecules.

The only problem here is the issue of whether the protonation state of the nitrogens is the same as the diacid in the previous studies. Few of these studies of TCPP have clearly recognized the possibility of protonation of the nitrogens. While the carboxylates are significantly more basic than the sulfonates, and therefore protonate first, the inner nitrogens are all expected to be protonated at pH less than about 4.199 At pH below 4, the inner ring is therefore expected to be highly positively charged for both TCPP and TSPP.

Since the electrochemical studies on TCPP were carried out in acid solution, it is likely that they involved the inner protonated species. Borguet appears to be the only investigator to clearly understand this and his model for the electrochemical process explicitly involves the nitrogen protonation.184 One is left with two possibilities. 1) Both

TCPP and TSPP had completely protonated nitrogens and it is solely the difference in hydrogen bonding that causes TSPP to order on HOPG; or 2) In previous studies of

TCPP on HOPG the pH was such that only two of the nitrogen atoms were protonated.

In this case the role of image charges in stabilizing the zwitterionic diacid of TSPP may play an equal or even greater role than the H bonding differences.

A high resolution image of the monolayer is shown in Figure 4.5.2 3. Note that while the monolayer shows some order, it is defective. On closer examination one sees that two types of individual molecular images are present. Some of the molecules appear to be divided in half nearly vertically (see rectangle in Figure 4.5.2 3) and others are divided in half at right angles to the first set (see oval in Figure 4.5.2 3). This central

217 division is due to the highly distorted shape of the fully protonated central porphyrin core. In order to accommodate four protons, two nitrogens move up out of the plane and two move down. This causes an overall distortion as seen in the side-view molecular structure in Figure 4.5.2 1. Because this porphyrin structure does not have four-fold symmetry, molecular packing is not equivalent in the two directions. Moreover, we have identified two different structures that occur within a given large island. Thus, the presence of two similar but distinct structures, and what may be marginal stability

(Hx[H4TCPP] is unstable) of the diacid on this surface, contribute to the highly defective nature of the islands.

Figure 4.5.2 3: High resolution image of H2(H4TSPP) monolayer on HOPG showing detailed molecular packing and distortion of porphyrin due to complete macrocycle

218 protonation. V=1.6V, setpoint is 1 pA. Note the difference in orientation of molecules within square and within ellipse.

The insert in Figure 4.5.2 3 is a high resolution image in which we have superimposed appropriately scaled ball and stick models of the diacid. On average, there are two protons at each juncture made by the four sulfonate groups, each from a different diacid molecule. These protons are almost certainly hydrogen bonding across adjacent sulfonates and helping to stabilize the net -2 charge. Further, it is important to remember that there are image charges in the substrate that help stabilize both the negative and positive portions of this zwitterion and thereby makes the planar structure most energetically favorable.

Using the measured molecular spacings and orientations, we first determined that the slightly more common unit cell had lattice dimensions A=B=1.550.05 nm and

=932. The second structure had similar but slightly different parameters, A= B=

1.50.05 nm, B=1.50.05 and =942. Based only on these measurements, the two cells are identical within our measurement precision. However, an additional factor can be used to separate these into two different commensurate lattices. It has been well established that the long straight graphite edges are the chair configuration edges.200-202

Thus, the [1,1,0,0] direction (and symmetry equivalents) can be identified. The directions of unit cell edges can then be determined by measuring the angle they make with a long straight step edge. By measuring the angles formed between the ordered rows of a given ordered domain (which differed for the two unit cell types), it is possible to determine two appropriate and differing structures. We will demonstrate this process

219 using the slightly more common unit cell. The unit cell spacing, internal angle, and the relative angle of the unit cell vectors to the [1,1,0,0] direction can be determined from

STM images. Using this data we can construct the unit cell seen in Figure 4.5.2 4.

Figure 4.5.2 4: Unit cell for the more common H2(H4TSPP) surface structure on HOPG.

We wish to express the vectors for the unit cell in Figure 4.5.2 4 in terms of HOPG lattice vectors. The HOPG coordinate system uses four different vectors; three of which are coplanar with one sheet of graphene and one normal to the graphene sheet. Figure 4.5.2

5 shows the same unit cell as Figure 4.5.2 4 along with the three coplanar HOPG coordinate vectors. The porphyrin molecules have been replaced with blue dots for clarity. Since we are only working with the outermost layer of the HOPG the fourth vector, which would be coming out of the page, is not needed.

220

Figure 4.5.2 5: Unit cell for the more common H2(H4TSPP) surface structure on HOPG showing the unit cell vectors in terms of HOPG coordinates.

The unit cell vectors (A and B) can be expressed in terms of HOPG lattice vectors. All three HOPG coordinates are not needed, vectors A and B can be expressed in terms of a1 and a2 alone:

(4.6.2-1a)

(4.6.2-1b)

The most concise way to express the H2(H4TSPP) surface structure is to use matrix notation. Using this format the more common unit cell can be written as:

(4.6.2-2)

In order to illustrate the correlation of the H2(H4TSPP) surface structure surface with

STM data Figure 4.5.2 6 shows the same ball and stick molecular model with appropriately scaled ellipses to mimic the structure seen in STM. Saddled porphyrins appear as twofold structures in STM images because the pyrrole groups with the

221 nitrogens pointing toward the surface show high contrast.203

Figure 4.5.2 6: Unit cell for the more common H2(H4TSPP) surface structure on HOPG with overlaid ellipses to show correlation with STM images.

      The second lattice can be visualized as A  6a1  6a 2 and B  3a1  4a 2 . In both cases, there are several symmetry related equivalents. This would preclude the phase separation one might normally expect. The first lattice (Figure 4.6.2.6) has unit cell parameters A=B=1.54 nm and =92.2. The second proposed commensurate unit cell is slightly smaller with A=1.48 nm, B=1.50 nm, and =94.7.

222

Figure 4.5.2 7: Unit cell for the less common H2(H4TSPP) surface structure on HOPG with overlaid ellipses to show correlation with STM images.

The unit cells proposed here are much smaller than any proposed for TCPP in any environment. The cells reported for TCPP to date are square cells with lattice vector 1.8 nm long,184,190 a hexagonal array with lattice constant near 1.7 nm,184 and very large cells with lattice constants of the order of 2 nm and larger.187,189,190 Moreover, it has been suggested that the phenyl groups lay flat on the surface in order to achieve the 1.8 nm spacing,190 a situation that clearly does not occur for TSPP. The origin of the much tighter structure and the non-parallel phenyl group orientations may be due to the near tetrahedral nature of the sulfonate group. While a two carboxylates can only achieve full

H-bonding when directly aligned, the presence of three tetrahedrally coordinated oxygen ions provide much more structural flexibility.

223

4.5.3 UHV-STM Current versus Voltage Studies of H2(H4TSPP) Nanorods and

Monomers

Current vs. voltage (I(V)) studies were carried out on nanorods deposited on

HOPG and Au(111) and on monomeric H2(H4TSPP) on HOPG. All measurements were carried out with the vacuum system windows covered to eliminate photoinduced reactions. Monomeric I(V) curves were not collected on Au(111) due to the surface reaction with HCl, thus we will focus on I-V and OMTS data acquired from HOPG supported samples that have been heated to 100°C. Tip cleanliness is of prime importance when collecting I(V) so a particular protocol was followed in order to ensure tip condition. First, as was described in section 2.4 tips were sputtered with argon ions to remove the native oxide layer. After sputtering the tips were checked by scanning over a

Au(111) surface cleaned by repeated cycles of argon sputtering and annealing. A good tip resolved the well-known Au(111) reconstruction and exhibited a linear I(V) curve. A typical image of the Au(111) reconstruction and an accompanying I(V) curve are shown in Figure 3.4 1. I(V) curves were collected over nanorods by taking an initial scan of ~

100 nm with a typical setpoint of 1 pA at 1.6 V sample bias. After the initial scan the tip was positioned over the rod, the setpoint increased to 15 pA, and I(V) curves run. The setpoint was increased after the tip was positioned over the nanorod because scanning at higher setpoints disrupted the aggregates and contaminated the tips. Figure 4.5.3 1 presents I(V) curves of 4 nm tall nanorods deposited on Au(111) and on HOPG.

224

Figure 4.5.3 1: I(V) curves of 4 nm tall H2(H4TSPP) nanorods deposited on Au(111)

(black) and HOPG (red). In both cases the intial parameters were setpoint 15 pA at 1.6 V sample bias.

Figure 4.5.3 1 is a very interesting result. The current through a single nanorod shows a considerable band gap (~2.4 V). This is certainly consistent with Schwab's observation49 that the rods are insulating in the voltage range between 0.6 V. The two curves essentially overlay, this indicates that either the work functions of the two samples are the same, or that the Fermi level of the nanorod is equilibrating with that of the substrate -- a very different situation than often encountered for molecular adsorption where the vacuum levels tend to equilibrate. Since the nanorod curves on Au(111) and HOPG are identical and owing to the surface reaction between the Au(111) substrate and HCl we will concentrate our study on nanorods deposited on HOPG.

225

The shape of the nanorod I(V) curve can be explained by comparing I(V) curves taken on a stack of several nanorods. Figure 4.5.3 2 shows an STM image of a bundle of nanorods and the associated I(V) curves. I(V) data was collected on a single rod (4 nm tall), a double rod (8 nm tall), and a triple rod (12 nm tall). In addition, there is a small section of a tube half lying on top of a single intact rod (total height 6 nm). In all cases shown in Figure 4.5.3 2 the bias voltage was initially set at 1.6 V with a setpoint of 15 pA and the I(V) curves were acquired with the feed-back loop turned off. The current through a single rod shows the same band gap as Figure 4.5.3 2. For rod bundles, the

I(V) curve becomes extremely asymmetrical with almost no current flowing in negative bias out to -2 V. In contrast, the positive bias current is generally only weakly dependent on the number of rods in a stack.

Figure 4.5.3 2: STM image of a stack of nanorods (setpoing 1 pA at 1.6 V sample bias) and I(V) curves taken at various points on the stack.

226

In order to better understand this variation in negative bias current with thickness, we plotted the natural log of the normalized current at -2 V sample bias versus rod bundle thickness (Figure 4.5.3 3). What we observed is an exponential decrease in current with rod thickness with a characteristic attenuation of -0.11 per Angstrom. This type of dependence is normally associated with single or multiple barrier tunneling, or superexchange.204,205 This decay constant is small (high conductivity) in comparison to tunneling through most single molecules.205-208

Figure 4.5.3 3: Plot of the natural log of current flow at -2 V vs. nanorod thickness.

It is important to note also that simple barrier tunneling does not account for the huge asymmetry in the I(V) curve. To represent the difference in positive and negative bias conductivity in a quantitative manner, one may consider the rectification ratio, defined as

I(+2V/I(-2V), and displayed as a ln-linear plot in Figure 4.5.3 4. A stack of only three nanorods results in a rectification factor of about 10,000! Clearly there is something more complex than barrier tunneling occurring here.

227

Figure 4.5.3 4: Plot of the natural log of nanorod rectification ratios vs. nanorod thickness.

To better understand the rectifying nature of the nanorods, knowledge of the

HOMO and LUMO of the parent diacid is required. To this end, we performed orbital mediated tunneling spectroscopy (OMTS) on the monolayer. In Figure 4.5.3 5, this

OMTS is presented along with the dI/dV of the 12 nm rod stack. We want to make very clear that we do NOT interpret the dI/dV of the stack as the density of states of the stack.

This should be simply interpreted as the differential conductivity of the rods. To see this, consider the negative bias region where the amount of current flow for the 12 nm stack is very close to the noise level. Any reasonable estimate of dI/dV will be zero in this region. Further, the connection between dI/dV and the density of states is based upon the current arising entirely via tunneling – clearly not the case for the 12 nm thick rod bundles. On the other hand, the conduction mechanism for a single molecule in the monolayer region is entirely tunneling in nature and we do interpret the dI/dV spectra as related to the LDOS.

228

Figure 4.5.3 5: dI/dV spectra of NiTPP, TSPP, and a 12 nm stack of nanorods.

The OMTS of H2(H4TSPP) on HOPG (black) shows a clear peak at +1.3 V that we can associate with the LUMO of the porphyrin ring. The structure that starts up near -

1.5 V is due to one or more occupied MO‟s. The assignments can be better understood if one considers the OMTS spectrum of nickel tetraphenylporphyrin (NiTPP) previously reported on Au(111) and shown in (blue) in Figure 4.5.3 5.209 We note that NiTPP

2+ results from H2TSPP if the sulfonate groups are removed and Ni replaces the two central protons. The band at +1.6 V bias in the NiTPP spectrum is known to be due to the porphyrin ring LUMO, while that near -1.3 V is due to the HOMO. If one shifts the

NiTPP spectrum by -0.45 V, the green curve in Figure 4.5.3 5 is obtained. A shift of some magnitude is required to account for the difference in work function of the NiTPP

128,183 covered Au(111) surface and the H2(H4TSPP) covered HOPG surface. According to Scudiero et al.,209 the work function for the NiTPP case was 5.2 eV. For the nanorod covered HOPG surface, a work function of 4.9 ± 0.1 eV was measured. Thus, a shift of

229

0.3 V in the OMTS is reasonable based on the work function of differences and the additional 0.15 V shift is probably associated with a difference in binding energy. The one curious feature in the H2(H4TSPP) spectrum in comparison to that of NiTPP is the absence of a plateau in the negative bias spectrum. This may be due to the splitting and mixing of occupied orbitals that are degenerate and near degenerate in the NiTPP but not in H2(H4TSPP).

Based on the OMTS data, it is clear that the dramatic increase in conductivity of the nanorods above + 1 V is due to a band formed from the H2(H4TSPP) LUMO. This band is sufficiently conducting that, once it is occupied, it brings the potential of the entire rod to that of the substrate. The HOMO, on the other hand, provides such small conductivity that the tip-substrate potential is primarily lost across the rods themselves.

The absence of any molecular structure in the images of the surface of the nanorods also suggests that there is a delocalized conduction band in the region above +1.2 V bias.

Since the tunneling image reflects the local density of states (LDOS), e.g.; silicon or

HOPG. For highly delocalized states, as in the case of metals, it is extremely difficult to observe single atoms. Figure 4.5.3 6 is a diagram of the effects of aggregation on the electronic structure of H2(H4TSPP). In section 1.4.2 exciton theory was discussed in terms of dimer formation. Upon dimer formation the respective orbitals of the monomers interact and split creating a new electronic structure. Aggregate formation can be thought of as dimer formation, followed by trimer, followed by tetramer, ect. As the number of monomers increases the aggregate orbitals become so closely spaced that a continum of states is formed. As was previoulsy discussed I(V) data shows that the aggregate

230 conduction band is highly conducting and delocalized. The valence band is more difficult to charaterize because up to -2 V there is comparativly little conductivity so it has been depicted as a small band in Figure 4.5.3 6.

Figure 4.5.3 6: Diagram of the effect of aggregation on the electronic structure of

H2(H4TSPP).

An important issue when considering conduction mechanisms is the temperature dependence of the conductivity. We measured the I(V) curves from several single nanorods on HOPG both at room temperature and at 90 K. Figure 4.5.3 7 shows representative results. While the curvature in I(V) at positive bias may be somewhat greater at 90 K than at 298 K indicating some thermal contribution to the onset, it may also be within the range of curvature measured from different rods. What is clear is that cooling by a factor of more than three has not significantly affected the negative voltage side, further supporting our contention that the conductivity in negative bias is due to tunneling.

231

Figure 4.5.3 7: I(V) curves of 4 nm nanorods taken at room temperature and 90 K.

4.6 Helium Ion Microscopy Studies

One of the major goals of this work is to resolve the molecular structure of

H2(H4TSPP) aggregates. To this end the nanorods were analyzed by HIM. Micrographs were gathered for nanorods deposited on Au(111)/mica and peeled HOPG. The ORION®

He microscope is capable of imaging with two different detectors, an Everhart-Thornley and a Rutherford backscattering detector. The Everhart-Thornley detector measures secondary electrons ejected from the sample by the incident He ions, while the

Rutherford backscattering detector measures He ions scattered by the sample nuclei.135

Both modes were used to analyze the nanorods. The Rutherford backscattering technique failed to show contrast between the nanorods and either substrate. The lack of contrast is likely due to low backscattering yield from the light atoms in the nanorods as Rutherford scattering increases with nuclear charge (Z).135 Figure 4.6 1 shows representative micrographs of H2(H4TSPP) aggregates deposited on HOPG taken using the Everhart-

232

Thornley detector. The images show nanorods of varying lengths distributed over the surface; some of which are isolated and some of which are clumped together. The secondary electron yield from the nanorods is lower than from the substrate, hence the negative contrast. Both H2(H4TSPP) and HOPG are highly pi-conjugated structures so the much lower electron yield from the nanorods is interesting.

Figure 4.6 1: He ion micrograph of H2(H4TSPP) nanorods deposited on HOPG.

Quality nanorod micrographs were difficult to obtain. Upon scanning the nanorods were quickly sputtered off the surface by the ion beam along with the HOPG substrate. To remedy this issue the beam blank was used and images were collected as quickly as possible. The effect of sputtering can be seen in Figure 4.6 2.

233

Figure 4.6 2: He ion micrograph of H2(H4TSPP) nanorods deposited on HOPG showing the effects of He ion sputtering on the sample.

He ion micrographs of H2(H4TSPP) nanorods deposited on Au(111) were also collected. These images show the same negative contrast when using the Everhart-

Thornley detector as Figure 4.6 1 and Figure 4.6 2. Rutherford backscatter detection proved just as ineffective on Au(111) as on HOPG. The HIM images of nanorods on

Au(111) are similar to STM and AFM images taken on the same substrate; both individual and clumped rods can be seen in the micrographs.

234

Figure 4.6 3: He ion micrograph of H2(H4TSPP) nanorods deposited on Au(111).

Our best results came from the nanorods on Au(111) sample after it was tilted 20°.

Figure 4.6 4 shows two micrographs taken with the sample stage tilted. The sputtering mention on the HOPG sample is also evident in the right image in Figure 4.6 4. By tilting the sample the rounded end of one of the nanorods can be clearly seen.

235

Figure 4.6 4: He ion micrograph of H2(H4TSPP) nanorods deposited on Au(111) and tilted 20°.

Figure 4.6 5 shows the same micrograph as Figure 4.6 4 with the addition of rod widths. Nanorod heights cannot be meaningfully discussed with respect to HIM micrographs acquired with an Everhart-Thornley detector because the contrast is a result of different secondary electron yields between the substrate and nanorod, not a physical height.

236

Figure 4.6 5: He ion micrograph of H2(H4TSPP) nanorods deposited on Au(111) and tilted 20° with the dimensions of several rods.

The narrowest rod is 28.6 nm wide, well in line with the average width by UHV-STM.

The two thicker rods (54.2 and 54.0 nm) are about twice the width of the thin rod. The most likely explanation for the wider rods is two aggregates lying next to each other. It is interesting that the two individual rods are not resolved. This is not unexpected as individual nanorods in bundles are not resolved in many images. Since the microscope lacked the resolution to resolve individual nanorods in bundles it is not surprising that we were unable to resolve the molecular structure of the nanorods using the He microscope.

237

4.7 Transmission Electron Microscopy Studies

In the continuing effort to resolve the molecular structure of the nanorods 5 µM

H2(H4TSPP) in 0.75 M HCl was deposited on carbon coated Ni Formvar TEM grids for

20 min followed by spin drying for 30 sec at 4000 rpm. A typical TEM imgae is presented in Figure 4.7 1. In the TEM micrographs the same bundling tendancy is present as has been seen in the other microscopic techniques employed in the study of

H2(H4TSPP) nanorods. Unlike AFM, STM, and HIM the nanorods do not contrast well with the substrate. Clustered rods can be seen easily, but single rods are difficult to differentiate from the background.

Figure 4.7 1: TEM micrograph of H2(H4TSPP) nanorods deposited on a carbon coated Ni

Formvar TEM grid.

238

Nanorod widths in the TEM images are in line with the other techniques discussed. Figure 4.7 2 shows two single rods 29 and 30 nm wide; very close to the average width of 27 nm measured in UHV-STM images. Height cannot be discussed meaningfully in TEM micrographs since image contrast is a function of the sample‟s ability to transmit electrons. We were unable to resolve the internal structure of the nanorods via TEM.

Figure 4.7 2: TEM micrograph of H2(H4TSPP) nanorods deposited on a carbon coated Ni

Formvar TEM grid.

4.8 Nanorod Current vs. Voltage Studies via Interdigitated Electrode

We have demonstrated in section 4.5.3 that H2(H4TSPP) nanorods are highly conducting across the short axis of the rod provided the bias voltage is high enough to access the aggregate‟s conduction band. The HOMO, on the other hand, provides such

239 small conductivity via a tunneling mechanism that the tip-substrate potential is primarily lost across the rods themselves. The STM I(V) data presented relates to transverse conduction in these nanorods. The effects of scattering centers, impurities, and physical defects are minimized because of the very short path length. The role that imperfections and transport direction plays in conduction is critical for any practical applications of this intriguing material. We will address this question by running I(V) curves on nanorods deposited on interdigitated gold electrodes. The electrode gap in the device is 400 nm as measured by AFM so any rod conductivity observed will be in the longitudinal direction.

The experimental setup is covered in detail in the experimental section.

The construction of the electrode precludes AFM measurements after nanorod deposition. In order to check nanorod surface coverage the IdE was imaged by SEM.

Figure 4.8 1 shows two SEM images of nanorods deposited on the IdE. As was the case when the nanorods were imaged by HIM using secondary electron detection the rods appear dark against the bright Au electrode.

240

Figure 4.8 1: SEM micrographs of H2(H4TSPP) nanorods deposited on an IdE.

Individual rods are difficult to see in the left pane of Figure 4.8 1. There are several larger bundles which appear to bridge the electrode gaps. The right pane of Figure 4.8 1 shows a higher resolution image of one of the rod bundles. The negative nanorod contrast makes it difficult to be sure, but it appears that the rod bundle does connect the two electrodes.

I(V) curves of IdE‟s treated with HCl alone were run as blanks. The HCl treated electrodes showed no discernable current flow between +2 and -2 volts. A number of electrodes and nanorod deposition protocols were used with limited success. All of the electrodes burned out; the best sample failed after 42 I(V) curves with a marked loss of current as I(V) curves were run. Although the current fell as the device was run the shape of the curves remained constant so a few rough and qualitative statements can be made.

The positive bias side of the nanorod treated electrodes is comparable to the STM I(V) curves. Just like the STM I(V) curves no current flows between 0 and ~1.2 V with a

241 sharp rise at higher voltages indicating the presence of a conduction band. The data in negative bias does not correlate well with STM data. Instead of a small rise in conductivity below -1.25 V a significant increase can be seen at -0.75 V. Based on a series of experiments where the data was acquired at varying rates we attribute at least part of the current in negative bias to capacitive coupling. To mitigate the coupling as much as possible the sampling time for each data point was kept at 3400 µs; the maximum allowed by the DI software.

Figure 4.8 2: I(V) curve of H2(H4TSPP) nanorods deposited on an IdE and an IdE treated with HCl. Both measurements were taken with the samples in a closed box to prevent photoconductivity.

It has been demonstrated that similar devices employing the same aggregate are capable of sustaining photocurrent levels of several hundred pA for hours when the bias

242 voltage is kept between ± 0.5 V.49,111 There are a number of possible explanations for the instability of our samples. It may be that beyond ± 0.5 V the rods are unstable with respect to longitudinal conduction which would account for the irreproducibility of our results. There is one study of H2(H4TSPP) nanorod conductivity over ± 5 V which showed an ohmic behavior.113 This study was different from ours in several important ways. The device was very different in construction; nanorods were deposited on the substrate followed by deposition of patterned Au metal on top of the aggregates. The measurements were carried out in low vacuum (10-4 Torr); our data was gathered under

Ar. It was never stated whether or not the photoconductive nanorods were illuminated during measurements. The differences in construction and experimental methodology make comparison between the data problematic. A second explanation for irreproducible results is the aforementioned capacitive coupling. I(V) measurements reported on a similar device required a 1 hour wait between voltage changes for the current to stabilize;49 our curves were taken in under one minute.

In summary we do not have good enough data to make firm conclusions regarding conductivity down the long axis of H2(H4TSPP) nanorods. In order to obtain quality data several issues must be addressed. First, the device should be redesigned so the electrodes are at the same level as the substrate. Second, the issue of capacitive coupling must be dealt with, possibly by taking data at a slower rate. Third, we must ascertain if the devices are burning out because the nanorods are not stable at higher voltages or due to a design flaw in the experimental setup.

243

Chapter 5: Future Work

There remains much work to be done on the H2(H4TSPP) aggregate system. One of the major goals of this project was to conclusively determine the molecular structure of the H2(H4TSPP) aggregate. Because of the highly delocalized electronic structure of the aggregate we were unable to visualize its molecular arrangement by STM. A technique such as X-ray diffraction may be needed to solve the structure of the aggregate. The second unresolved issue is the IdE I(V) measurements. The first set of IdE‟s were not well suited to nanorod deposition due to their tall electrodes and propensity to fail after just a few measurements. Redesigned electrodes will solve the discrepancy between the heights of the electrode fingers and the substrate which will aid in nanorod deposition.

The issue of device failure will require more study. It must be determined if failure is due to experimental setup, a different flaw in the electrode design, or if it is due to degradation of the aggregates.

Future work could also include studies of aggregates formed from mixtures of porphyrins such as H2(H4TSPP) and tetrapyridylporphyrin (TPyP). Preliminary studies have been carried out in our lab on aggregates formed in acidic, aqueous solutions of

H2(H4TSPP) and TPyP. Similar to H2(H4TSPP) the bimolecular aggregates form rod- shaped structures, but unlike aggregates composed purely of H2(H4TSPP) the widths of the aggregates varies considerably as can be seen in Figure 5 1.

244

Figure 5 1: HIM images of a bimolecular aggregate formed in an acidic solution containing H2(H4TSPP) and TPyP.

Figure 5 1 gives rise to many questions. What is the ratio between the two porphyrin species? Can the morphology of the aggregate be changed based upon the relative amounts of the two porphyrin species? Is the composition of the smaller aggregates the same as the larger aggregates? What role does pH play in aggregate formation? Do these bimolecular aggregates exhibit the same delocalized band as H2(H4TSPP) aggregates? If so, is it the valence band or the conduction band? The study of bimolecular aggregation will allow us to understand how the composition of an aggregate is related to its electronic properties. With this knowledge in hand we have the potential to design nanoscale components in electrical devices with tunable properties.

245

Chapter 6: Conclusions

Revisiting the stated goals of this project outlined in the introduction:

1. To measure UV-visible and RLS spectra of monomeric H2(H4TSPP) and it‟s

aggregate to verify the aggregation of H2(H4TSPP) and to parse the electronic

spectrum into monomeric and aggregate components.

We have confirmed the aggregation of H2(H4TSPP) by UV-visible and RLS. The 490

nm band is assigned to the nanorod due to the high RLS signal at 490 nm. The

assignment of the 424 and 707 nm bands is uncertain. The scattering intensities of

the 490 nm and 707 nm ban track with each other indicating that the 707 nm and 490

nm bands arise from the same species. On the other hand UV-visible experiments

link the 424 nm and 707 nm band together. More experiments are needed to

understand the scattering and UV-visible data. The 434, 595, and 644 nm peaks in

the Uv-visible spectrum are assigned to the diacid monomer.

2. To conduct AFM studies on H2(H4TSPP) nanotubes deposited on substrates to

characterize the dimensions of the tubes.

AFM analysis indicates that H2(H4TSPP) nanotubes can be deposited on Au(111) and

HOPG. The nanotubes have consistent heights and widths from substrate to

substrate (30 nm and 4 nm respectively) and vary greatly in length from a few

hundred nm to over a micron.

3. To conduct STM studies on both monomeric H2(H4TSPP) and it‟s aggregate in order

to test the validity of the proposed aggregation models.

246

STM analysis of H2(H4TSPP) nanotubes on Au(111) and HOPG is consistent with

the dimensions from AFM. The only substructure visible by ambient STM was disk-

like which is likely the result of a side reaction with gold.

4. To carry out Raman studies of the H2(H4TSPP) monomer and nanotubes in solution

to investigate the effects of aggregation on the vibrational structure of the system.

We will also report Raman spectra of nanotubes deposited on Au(111) and compare

with solution data.

Raman spectra of the H2(H4TSPP) monomer and nanotubes show small shifts in

vibrational frequencies upon aggregation along with large increases in the intensity

of the low frequency modes. Comparison of solution spectra and spectra of the

aggregates deposited on Au(111) demonstrates intact deposition of the tubes on

Au(111).

5. To conduct compositional testing on the deposited monomer and aggregate by XPS.

XPS tests of the TSPP powders indicate that the chloride salt is the better starting

material. XPS data of deposited nanorods on Au(111), HOPG, and the monomer on

HOPG show that there is no compositional change upon aggregation and/or

deposition.

6. To conduct the first reported STM I(V) study of both the monomer and aggregate to

investigate the effects of aggregation on the electronic structure of the system. Save

Otsuka et al. the I(V) studies covered in the review were all constrained to bias

voltages less than one volt. We will report on the I(V) characteristics of monomeric

H2(H4TSPP) and it‟s aggregate over a range of ±2 V.

247

UHV-STM confirmed the presence of two different surface structures of monomeric

H2(H4TSPP) on HOPG. H2(H4TSPP) nanotubes were visualized on Au(111) and

HOPG. I(V) measurements confirmed the presence of a LUMO at 1.3 V in

monomeric H2(H4TSPP) which, upon aggregation, couples to form a highly

conducting valence band in the aggregate extending from 1.3 V to at least 2 V. No

delocalized valence band was seen in the aggregate down to -2 V.

7. To deposit the nanotubes on an interdigitated electrode and measure the conductivity

of the aggregates across the long axis of the tubes over a range of ±2 V.

Some I(V) measurements were carried out which supported the STM I(V) data. The

device suffered from design issues and high capacitance.

8. To conduct the first reported HIM studies on H2(H4TSPP) nanotubes in order to test

the validity of the proposed aggregation models.

HIM images were consistent with the widths observed by STM and AFM. The

nanorods did not scatter well and appeared dark in the HIM images. The

microscope’s resolution was insufficient to resolve aggregate substructure.

9. To conduct TEM studies on H2(H4TSPP) nanotubes in order to test the validity of the

proposed aggregation models.

TEM images were consistent with the widths observed by STM and AFM. The

nanorods suffered from poor contrast in the TEM images. The microscope’s

resolution was insufficient to resolve aggregate substructure.

248

References

1 Treibs, A. E.; Angewandte Chemie 1936, 49, 682.

2 Stone, A.; Fleisher, E. B. JACS 1968, 90, 2735-2748.

3 Hu, J. S.; Ji, H. X.; Wan, L. J. J. Phys. Chem. C 2009, 113, 16259-16265.

4 Gouterman, M. J. Chem. Phys. 1959, 30, 1139.

5 Campbell, W. M.; Jolley, K. W.; Wagner, P.; Wagner, K.; Walsh, P. J.; Gordon, K. C.;

Schmidt-Mende, L.; Nazeerudding, M. K.; Wang, Q.; Gratzel, M.; Officer, D. L. J. Phys.

Chem. C Lett. 2007, 111, 11760-11762.

6 Yao, C.; Yan, L.; Guan, L.; Liu, C.; Song, P.; Su, Z. Dalton Trans, 2010, 39, 7645-

7649.

7 Thyagariajan, S.; Leiding, T.; Arskold, S. P.; Cheprakov, A. V.; Vinogradov, S. A.

Inorganic Chemistry, 2010, DOI: 10.1021/ic100968p.

8 Riberio, S. M.; Serra, A. C.; Rocha Gonsalves, A. M. d‟A. Journal of Molecular

Catalysis A 2010, 326, 121-127.

9 Verma, A.; Facchina, S. L.; Hirsch, D. J.; Song, S.; Dillahey, L. F.; Williams, J. R.;

Synder, S. H. Molecular Medicine 1998, 4, 40-45.

10Fuhrhop, J. H.; Mauzerall, D. J.A.C.S. 1969, 91, 4174-4181.

11 Pekkarinen, L.; Linschitz, H. J.A.C.S. 1960, 82, 2407-2411.

12 McRae, E. G. Aust. J. Chem. 1961, 14, 329-348.

13 Kasha, M.; McRae, E. G. Radiation Research, 1963, 20, 55.

14 Whitesides, G. M.; Mathias, J .P.; Seto, C. T. Science, 1991, 254, 1312-1319.

15 Hu, X.; Schulten, K. Physics Today August 1997, 28-34.

249

16 Zhou, Q.; Li, C. M.; Li, J.; Lu, J. J. Phys. Chem. C 2008, 112, 18578-18583.

17 Rao, C. N. R.; Agrawal, V. V.; Biswas, K.; Gautam, U. K.; Ghosh, M.; Govindaraj, A.;

Kulkarni, G. U.; Kalyanikutty, K. P.; Sardar, K.; Vivekchand, S. R. C. Pure and Applied

Chemistry 2006, 78, 1619-1650.

18 Wang, Y.; Zhou, W.. J. Nanosci. Nanotech. 2010, 10, 1563-1583.

19 Brock, S. L. Angewan. Chemie, Int. Ed. 2009, 48, 7484-7486.

20 Tenne, R. Chemistry of Nanostructured Materials 2003, 147-182.

21 Seayad, A. M.; Antonelli, D. M. Advanced Materials (Germany) 2004, 16, 765-777.

22 Haruta, M. Catal. Today 1997, 36, 153-156.

23 Jain, P. K.; El-Sayed, M. A. Chem. Phys. Lett. 2010, 487, 153-164.

24 Le Moal, E.; Fort, E.; Leveque-Fort, S. Actualite Chimique 2009, 332, 36-44.

25 Zhang, J. Z. J. Phys. Chem. Lett. 2010, 1, 686-695.

26 Khomutov, G. B.; Koksharov, Y. A. Advances in Colloid and Interface Science 2006,

122, 119-147.

27 Stoyanov, S. R.; Titov, A. V.; Kral, P.. Coord. Chem. Rev. 2009, 253, 2852-2871.

28 Avasthi, D. K.; Kumar, A.; Singhal, R.; Tripathi, A.; Misra, D. S. J.Nanosci. Nanotech.

2010, 10, 3767-3779.

29 Wen, Z.; Li, J. J. Mat. Chem. 2009, 19, 8707-8713.

30 Kamat, P. V. J. Phys. Chem. Lett. 2010, 1, 520-527.

31 Benard, P.; Chahine, R. Solid-State Hydrogen Storage 2008, 261-287.

32 Sepehri, S.; Cao, G. Annual Review of Nano Research 2010, 3, 487-514.

33 Yum, K.; Wang, Ning; Yu, M. F.. Nanoscale 2010, 2, 363-372.

250

34 Deriu, A.; Di Bari, M. T.; Gerelli, Yuri. Zeitschrift fuer Physikalische Chemie

2010,224, 227-242.

35 Braunschweig, A. B.; Schmucker, A. L.; Wei, D.; Mirkin, C. A. Chem. Phys. Lett.

2010, 486, 89-98.

36 Berven, C. A.; Dobrokhotov, V.; McIlroy, D. N.; Chava, S.; Abdelrahaman, R.;

Heieren, A.; Dick, J.; Barredo, W. IEEE Sensors Journal 2008, 9, 930-935.

37 Avasthi, D. K.; Kumar, A.; Singhal, R.; Tripathi, A.; Misra, D. S. J. Nanosci.

Nanotech. 2010, 10, 3767-3779.

38Torres -Costa, V.; Martin-Palma, R. J. J. Materials Sci. 2010, 45, 2823-2838.

39 Moshfegh, A. Z. J. Phys. D 2009, 42, 233001/1-233001/30.

40 Banhart, F. Nanoscale 2009, 1, 201-213.

41 Zhang, X.; Wang, Q.; Wu, L,; Lv, W.; Lu, J.; Bian, Y.; Jiang, J. J. Phys. Chem. B

2010, 114, 1233-1240.

42 Djurisic, A. B.; Ng, A. M. C.; Cheung, K. Y.; Fung, M. K.; Chan, W. K. J. Mater. Sci.

& Technol. (Shenyang, China) 2008, 24, 563-568.

43 Karl, N. Proceedings International School of Physics “Enrico Fermi” 2002, 149th

(Organic Nanostrucutres: Science and Applications), 25

44 Shi N. E; Yin, G.; Han, M.; Jiang, L.; Xu, Z,; Chem. Eur. J. 2008, 14, 6255-9.

45 Kamat, P. V. J. A. C. S. 2008, 130, 14020-143019.

46 Krueger, A. Nachrichten aus der Chemie 2009, 57, 273-279.

47 Xi, H.; Wei, Z.; Duan, Z. Xu, W.; Zhu, D. J. Phys. Chem. C 2008, 112, 19934-38.

251

48 Hasobe, T.; Oki, H.; Sandanayaka, A. S. D.; Murata, H. Chem. Commun. 2008, 724-

726.

49 Schwab, A. D.; Smith, D. E.; Bond-Watts, B.; Johnson, D. E.; Hone, J.; Johnson, A. T.; de Pula, J. C.; Smith, W. F. Nano Lett. 2004, 4, 1261-1265.

50 Medforth, C. J.; Wang, Z.; Martin, K. E.; Song, Y.; Jacobsen, J. L.; Shelnutt, J. A.

Chem. Commun. 2009, 47, 7241-7277.

51 Schwab, A.; Smith, D. E.; Rich, C. S.; Young, E. R.; Smith, W .F.; de Paula, J. C. J.

Phys. Chem. B, 2003, 107, 11339-11345.

52 Drain, C. M.; Varotto, A.; Radivojevic, I. Chem. Rev. 2009, 109, 1630-1658.

53 Dini F.; Martinelli, E.; Pomarico, G.; Paolesse, R.; Monti, D.; Filippini, D.; D'Amico,

A.; Lundstrom, I.; Di Natale, C. Nanotechnology 2009, 20, 055502.

54 Sun, D.; Tham, F. S.; Reed, C. A.; Chaker, L.; Burgess, M.; Boyd, P. D. W. J. Am.

Chem. Soc. 2000, 122, 10704-10705.

55 Wu, J.; Fang, F.; Lu, W. Y.; Hou, J. L.; Li, C.; Wu, Z. Q.; Jiang, X. K.; Li, Z. T.; Yu,

Y. H.; J. Org. Chem. 2007, 72, 2897-2905.

56 Vinodu, M.; Goldberg, I. CrystEngComm 2005, 7, 133-138.

57 Hofkens, J.; Latterini, L.; Vanoppen, P.; Faes, H.; Jeuris, K.; De Feyter, S.; Kerimo, J.;

Barbara, P. F.; De Schryver, F. C.; Rowan, A. E.; Nolte, R. J. M. J. Phys. Chem. B 1997,

101, 10588-10598.

58 Kubat, P.; Lang, K.; Janda, P.; Anzenbacher, P. Langmuir 2005, 21, 9714-9720.

59 Choi, M. Y.; Pollard, J. A.; Webb, A.; McHale J. L. J. AM. CHEM. SOC. 2003, 125,

810-820.

252

60 Doan, S. C.; Shanmugham, S.; Aston, D. E.; McHale, J. L. J. Am. Chem. Soc. 2005,

127, 5885-5892.

61 Kubat, P.; Lang, K.; Mosinger, J.; Wagnerova, D. M. Zeitschrift fuer Physicalische

Chemie, Bd 1999, 210, 243-256.

62 Vlaming, S. M.; Augulis, R.; Stuart, M. C. A.; Knoester, J.; van Loosdrecht, P. H. M.

J. Phys. Chem. B 2009, 113, 2273-2283.

63 Micali, N.; Romeo, A.; Lauceri, R.; Purrello, R.; Mallamace, F.; Scolaro, L. M. J.

Phys. Chem. B 2000, 104, 9416-9420.

64 Akins, D. L.; Zhu, H. R.; Guo, C. J. Phys. Chem. 1996, 100, 5420-5425.

65 Maiti, N. C.; Mazumdar, S.; Periasamy, N. J. Porphyrins Phthalocyanines 1998, 2,

369-376.

66 Escudero, C.; El-Hatchemi, A.; Crusats, J.; Ribo, J. M. J. Porph. Phthal. 2005, 9, 852-

863.

67 Miura, A.; Shibata, Y.; Chosrowjan, H.; Mataga, N.; Tamai, N. Journal of

Photochemistry and Photobiology A 2006, 178, 192-200.

68 Ribo, J. M.; Crusats, J.; Jarrera, J. A.; Valero, M. L. J. Chem. Soc. Chem. Commun.

1994, 681-682.

69 Micali, N.; Mallamace, F.; Romeo, A.; Purrello, R.; Scolaro, L. M. J. Phys. Chem. B

2000, 104, 5897-5904.

70 Micali, N.; Villari, V.; Castriciano, M. A.; Romeo, A.; Scolaro, L. M J. Phys. Chem. B

2006, 110,

8289-8295.

253

71 Gandini, S. C. M.; Gelamo, E. L.; Itri, R.; Tabak, M. Biophys. J. 2003, 85, 1259-1268.

72 Ohno, O.; Kaizu, Y.; Kobayashi, H. J. Chem. Phys. 1993, 99, 4128-4139.

73 Snitka, V.; Rackaitis, M.; Rodaite, R. Sens. Actuators, B 2005, 109, 159-166.

74 De Luca, G.; Romeo, A.; Scolaro, L.M. J. Phys. Chem. B, 2006, 110, 7309-7315.

75 Simkiene, I.; Sabataityte, J.; Babonas, G. J.; Reza, A.; Beinoras, J. Materials Science and Engineering C, 2006, 26, 1007-1011.

76 Ren, B.; Tian, Z. Q.; Guo, C.; Akins, D. L. Chemical Physics Letters, 2000, 328, 17-

22.

77 Chen, D. M.; He, T.; Cong, D. F.; Zang, Y. H.; Liu, F. C. J. Phys. Chem. A, 2001, 105,

3981-3988.

78 Castriciano, M. A.; Romeo, A.; Scolaro, L. M. Journal of Porphyrins Phthalocyanines

2002, 6, 431-438.

79 Frolov, D.; Bagdonas, S.; Kelbauskas, L.; Dietel, W.; Streckyte, G.; Rotomiskis, R.

Lithuanian Journal of Physics, 2001, 41, 484-494.

80 Akins, D.L.; Zhu, H. R.; Guo, C. J. Phys. Chem. 1994, 98, 3612-3618.

81 Aggarwal, L. P. F.; Borissevitch, I. E. Spectrochimica Acta Part A 2006, 63, 227-233.

82 Collini, E.; Ferrante, C.; Bozio, R. J. Phys. Chem. C 2007, 111, 18636-18645.

83 Kano, H.; Kobayashi, T. Bull. Chem. Soc. Jpn. 2002, 75, 1071-1074.

84 Maiti, N. C.; Ravikanth, M.; Mazumdar, S.; Periasamy, N.; J. Phys. Chem. 1995, 99,

17192-17197.

85 Maiti, N. C.; Mazumdar, S.; N. Periasamy, N. J. Porphyrins Phthalocyanines 1998, 2,

369-376.

254

86 Gandini, S. C. M.; Yushmanov, V. E.; Borissevitch, I. E.; Tabak, M. Langmuir 1999,

15, 6233-6243.

87 Wu, J. J.; Li, N.; Li, K. E.; Liu, F. J. Phys. Chem. B 2008, 112, 8134-8138.

88 Purrello, R.; Bellacchio, E.; Gurrieri, S.; Lauceri, R.; Raudino, A.; Scolaro, L. M.;

Santoro, A. M. J. Phys. Chem. B 1998, 102, 8852-8857.

89 Pasternack, R. F.; Fleming, C.; Herring, S.; Collings, P. J.; dePaula, J.; DeCastro, G.;

Gibbs, E. J. Biophysical Journal 2000, 79, 550-560.

90 Ribo, J. M.; Rubires, R.; El-Hachemi, Z.; Farrera, J. A.; Campos, L.; Pakhomov, G. L.;

Vendrell, M. Mater. Sci. Eng. C 2000, 11, 107-115.

91 Castriciano, M. A.; Romeo, A.; Villari, V.; Micali, N.; Scolaro, L. M. J. Phys. Chem. B

2003, 107, 8765-8771.

92 Gulbinas, V.; Karpicz, R.; Augulis. R.; Rotomiskis, R. Chemical Physics 2007, 332,

255-261.

93 McRae, E. G.; Kasha, M. Physical Processes in Radiation Biology; Augenstein, L.,

Mason, R., Rosenberg, B., Eds.; Academic Press: New Yourk 1964: p 23

94 Ogawa, T.; Tolumaga, E.; Kobayashi, T. Chem. Phys. Lett. 2005, 408, 186-191.

95 Zimmermann, J.; Siggel, I.U.; Fuhrhop, J.H.; Roder, B.J.M.; Srivastava, T.S.;

Chatterjee, A. J. Phys. Chem. B 2003, 107, 6019-6021.

96 Kano, H.; Saito, T.; Kobyashi, J. J. Phys. Chem. B 2001, 105, 413-419.

97 Shelnutt, J. A. J. Chem. Phys. 1981, 74, 6644-6657.

98 Pasternack, R. F.; Collings, P. J. Science 1995, 269, 935-939.

99 Pasternack, R. F.; Schaefer, K. F. Inorg. Chem. 1994, 33, 2062-2065.

255

100 Collings, P. J.; Gibbs, E. J.; Starr, T. E.; Vafek, O.; Yee, C.; Pomerance, L. A.;

Pasternack, R. A. J. Phys. Chem. B 1999, 103, 8474-8481.

101 Koti, A. S. R.; Periasamy, N. Chemistry of Materials 2003, 15, 369-371.

102 Chen, D. M.; He, T.; Cong. D. F.; Zhang, Y. H.; Liu, F. C. J. Phys. Chem. A 2001,

105, 3981-3988.

103 Kitahama, Y.; Kimura, Y.; Takazawa, K. Langmuir 2006, 22, 7600-7604.

104 Maiti, N. C.; Mazumdar, S.; Periasamy, N. J. Phys. Chem. B 1998, 102, 1528-1538.

105 Rotomiskis, R.; Augulis, R.; Snitka, V.; Valiokas, R.; Liedberg, B. J. Phys. Chem. B

2004, 108, 2833-2838.

106 Escudero, C.; Crusats, J.; Diez-Perez, I.; El-Hachemi, Z.; Ribo, J.M. Angew. Chem.

Int. Ed., 2006, 45, 8032-8035.

107 Miura, A.; Shibata, Y.; Chosrowjan, H.; Mataga, N.; Tamai, N. Journal of

Photochemistry and Photbiology A 2006, 178, 192-200.

108 Shimidzu, T.; Segawa, H.; Wu, F.; Nakayama, N. J. Photochem. Photobiol. A 1995,

92, 121-127.

109 Liu, C.; Tang, H.; Bard, A. J. J. Phys. Chem. 1996, 100, 3587-3591.

110 Fox, M. A.; Bard, A. J. Science 1993, 261, 897-899.

111 Riley, K. C.; Muller, E. A.; Feldman, B. E.; Cross, C. M.; Van Aken, K. L.; Johnston,

D. E.; Lu, Y.; Johnson, A. T.; de Paula, J. C.; Smith, W. F. J. Phys. Chem. C 2010, 114,

19227-19233.

112 Yeats, A. L.; Schwab, A. D.; Massare, B.; Johnston, D. E.; Johnson, A. T.; de Paula, J.

C.; Smith, W. F. J. Phys. Chem. C 2008, 112, 2170-2176.

256

113 Otsuka, Y.; Naitoh, Y.; Matsumoto, T.; Mizutani, W.; Tabata, H.; Kawai, T.

Nanotechnology 2004, 15, 1639-1644.

114 Rubires, R.; Muller, C.; Campos, L.; El-Hachemi, Z.; Pakhomov, G. L.; Rbo, J. M. J.

Porph. Phth, 2002, 6, 107-113.

115 Gouterman, M.; The Porphyins Vol III 1978 Academic Press Inc.

116 Gouterman, M.; J. Chem. Phys. 1959, 30, 1139-1161.

117 Linder, R. E.; Barth, G.; Bunnenberg, E.; Djerassi, C.; Seaman, L.; Moscowitz, A.

Journal of the Chemical Society, Perkin Transactions II 1974, 14, 1712-1718

118 Kasha, M.; Rawls, H. R.; Ashraf El-Bayoumi, M. Pure and Applied Chemistry, 1965,

11, 371-392.

119 Pasternack, R. F.; Bustamante, C.; Collings, P. J.; Giannetto, A.; Gibbs, E. J.; J. Am.

Chem. Soc. 1993, 115, 5393-5399.

120 Lu, W.; Band, B. S.; Yu, Y.; Li, Q. G.; Shang, J. C.; Wang, C.; Fang, Y.; Tian, R.;

Zhou, L. P.; Sun, L. L.; Tang, Y.; Jing, S. H.; Huang, W.; Zhang, J. P.; Microchim Acta

2007, 158, 29-58.

121 Stanton, S. G.; Pecora, R. J. Chem. Phys. 1981, 75, 5615-5626.

122 Einstein, A. Ann. Physik 1905, 17, 132-148.

123 Brundle, C. R. J. Vac. Sci. Technol. 1974, 11, 212-224.

124 Briggs, D. Handbook of X-ray and Ultraviolet Photoelectron Spectroscopy 1977

Heyden & Son Ltd.

125 Alov, N.V. Journal of Analytical Chemistry 2005, 60, 297-300

126 Swartz, W. Analytical Chemistry 1973, 45, 788-795.

257

127 Hamers, R. J. Annu. Rev. Phys. Chem. 1989, 40, 531-559.

128 Hipps; K.W. in “Handbook of Applied Solid State Spectroscopy”, Ed: Vij, D.R.,

Springer Verlag (2006) ISBN: 0-387-32497-6

129 Wiebe, J.; Wachowiak, A.; Meier, F.; Haude, D.; Foster, T.; Morgenstern, M.;

Wiesendager, R. Review of Scientific Instrumentation 2004, 75, 4871-4879.

130 Kowalczyk, P. Applied Surface Science 2007, 253, 4036-4040.

131 Giessibl, F. J. Rev. Mod. Phys., 2003, 75, 949-983.

132 Moskovits, M. J. Chem. Phys. 1982, 77, 4408-4416.

133 Dowdy, J.; Hoagland, J. J.; Hipps, K. W. J. Phys. Chem. 1991, 95, 3751-5755.

134 Morgan, J.; Notte, J.; Hill, J.; Ward, B. Microscopy Today, 2006, 14, 24-31.

135 Scipioni, L.; Sanford, C. A.; Notte, J.; Thompson, B.; McVey, S. J. Vac. Sci. Technol.

B 2009, 27, 3250-3255.

136 Carl Zeiss website http://www.smt.zeiss.com/c1256e4600305472/Contents-

Frame/e87f4889b0612d9bc125734f00373b5f Accessed 12/6/10

137 Hitachi website http://www.hhtc.ca/microscopes/sem/s5500.htm accessed 12/6/10

138 Scipioni, L.; Stern, L. A.; Notte, J.; S. Sijbrandij, Griffin, B. Advanced Materials &

Processes 2008, 27-30.

139 Misell, D. L. J. Phys. D: Appl. Phys., 1977, 10, 1085-1107.

140 Reimer, L.; Transmission Electron Microscopy 4th ed Springer-Verlag (1997) ISBN:

3-540-56849-2

141 FEI website http://www.fei.com/products/transmission-electron- microscopes/titan.aspx accessed 12/6/10

258

142 Sobotik, P.; Ostadal, I. Journal of Crystal Growth, 1999, 197, 955-962

143 Mahmudur, R.; Harmon, J.H.; J. Porph. Phth. 2007, 11, 125-129.

144 Ribo, J. M.; Crusats, J.; Farrera, J. A.; Valero, M. L. J. Chem. Soc., Chem. Commun.

1994, 681-682.

145 Castriciano, M. A.; Romeao, A.; Villari, V.; Micali, N.; Scolaro, L. M. J. Phys. Chem.

B 2003, 107, 8765-8771.

146 Rubires, R.; Cursats, J.; El-Hatchemi. Z.; Jaramillo, T.; Lomez, M.; Valls. E.; Farrera,

J.A.; Ribo, J. New J. Chem. 1999, 189-198.

147 Andrade, S. M.; Costa, S. M. B. Chem. Eur. J. 2006, 12, 1046-1057.

148 Huang, C. Z.; Li, Y. F.; Li, N.; Li, K. A.; Tong, S.Y. Bull. Chem. Soc. Jpn. 1998, 71,

1791-1797.

149 Guo, C.; Ren, B.; Akins, D. L. J. Phys. Chem. B 1998, 102, 8751-8756.

150 Delmarre, D.; Meallet-Renault, R.; Bied-Charreton, C.; Pasternack, R. F. Analytica

Chimica Acta 1999, 401, 125-128.

151 Crusats, J.; Claret, J; Diez-Perez, I.; El-Hachemi, Z.; Garcia-Ortega, H.; Rubires, R.;

Sagues, F.; Ribo, J. M. Chem. Commun. 2003, 13, 1588-1589.

152 Egawa, Y.; Hayashida, R.; Anzai, J.I. Langmuir 2007, 23, 13146-13150.

153 de Paula, J. C.; Robblee, J. H.; Pasternack, R. F.; Biophysical Journal 1995, 68, 335-

341

154 Chan, I. Y.; Hallock, A. J. J. Chem. Phys. 1997, 107, 9297-9301.

155 Clemmer, C. R.; Beebe, T. P. Science 1991, 251, 640-642.

156 Pong, W.-T.; Bendall, J.; Durkan, C. Surf. Sci. 2007, 601, 498-509.

259

157 Augulis, R.; Tamuliene, J.; Tamulis, A.; Rotomiskis, R. Diffus. Defect. Data, SSP

2004, 97, 225-228.

158 McDermott, G.; Prince, S. M.; Freer, A. A.; Hawthornthwalte-Lawless, A. M.; Papiz,

M. Z.; Cogdell, R. J.; Isaacs, N. W. Nature 1995, 374, 517-521.

159 Ranck, J. -L.: Ruiz, T.; Pehau-Arnaudet, G.; Arnoux, B.; Reiss-Husson, F. Biochim.

Biophys. Acta 2001, 1506, 67-78.

160 Hamor, J. J.; Hamor, T. A.; Hoard, J. L. J. Am. Chem. Soc. 1964, 86, 1938-1942.

161 Rosa, A.; Ricciardi, G.; Baerends, E. J.; Romeo, A.; Scolaro, L. M. J. Phys. Chem. A

2003, 107, 11468-11482.

162 Mazur, U.; Hipps, K. W. J. Phys. Chem. 1999, 103, 9721-9727.

163 Hansen, W. N.; Hansen, G. Surf. Sci. 2001, 481, 172-184.

164 Hori, T.; Aratani, N.; Takagi, A.; Matsumoto, T.; Kawai, T.; Yoon, M. C.; Yoon, Z.

S.; Cho, S.; Kim, D.; Osuka, A. Chem. Eur. J. 2006, 12, 1319-1327.

165 Dempster, S. E.; Jang, S.; Silbey, R. J. J. Chem. Phys. 2001, 114, 10015-10023.

166 Shelnutt, J. A. J. Chem. Phys. 1981, 74, 6644-6657.

167 Shimidzu, T.; Iyoda, T.; Segawa, H.; Honda, K. New Journal of Chemistry 1986, 10,

213

168 Habib, A.; Tabata, M.; Wu. Y.G. J Porph. Phth. 2004, 8, 1269-1275

169 Li, X. Y.; Czernuszewicz, R. S.; Kincaid, J. R.; Spiro, T. G. J. Am. Chem. Soc. 1989,

111, 7012-7023.

170 Bell, S. E. J.; Al-Obadi, A. H. R.; Heagarty, M.; McGarvey, J. J. J. Phys. Chem. 1995,

99, 3959-3964.

260

171 Saini, G. S. S. Spectrochimica Acta Part A, 2006, 64, 981-986.

172 Bell, S. E. J.; Al-Obadi, A. H. R.; Heagarty, M.; Hester, R. E.; McGarvey, J. J. J.

Phys. Chem. 1993, 97, 11599-11602.

173 Rosa, A.; Ricciardi, G.; Baerends, E. J.; Romeo, A.; Scolaro, L. M. J. Phys. Chem. A

2003, 107, 11468-11482.

174 Lemke, C.; Dreybolt, W.; Shelnutt, J. A.; Quirke, J. M. E.; Scweitzer-Stenner, R. J.

Raman Spectrosc. 1998, 29, 945-953.

175 Oakes, R. E.; Spence, S. J.; Bell, S. E. J. J Phys Chem. A 2003, 107, 2964-2973.

176 Dempster, S. E.; Jang, S.; Silbey, R. J. J. Chem. Phys. 2001, 114, 10015-10023.

177 Gnana, K. G.; Pil, K.; Ae Rhan, K.; Kee Suk, N.; Nimma, E.R. Materials Chemistry and Physics 2009, 115, 40-46.

178 Moulder, J.; et. al. Handbook of X-ray Photoelectron Spectroscopy 1992 Perkin Elmer

Corporation p61

179 Yamashige, H.; Shuij, M.; Tsutomu, K.; Perera, R. C.; Hisanobu, W. Analytical

Science 2005, 21, 635-639.

180 Stypula, B.; Stoch, J. Corrosion Science, 1994, 36, 2159-2167

181 Watanabe, M.; Ando, H.; Handa, T.; Ichino, T.; Kuwaki, N. Zairyo-to-Kankyo 2007,

56, 10-15.

182 Sumi, H. J. Phys. Chem. B 1998, 102, 1833-1844.

183 Hipps, K. W.; Scudiero, L. J. Chem. Ed. 2005, 82, 704-711.

184 Yuan, O.; Xing, Y.; Borguet, E. J. A. C. S 2010, 132, 5054-5060.

185 Yoshimoto, S.; Sawaguchi, T. J. Amer. Chem. Soc. 2008, 130, 15944-15949.

261

186 Yoshimoto, S.; Sawaguchi, T. ECS Transactions 2009, 16, 77-82.

187 Yoshimoto, S.; Yokoo, N.; Fukuda, T.; Kobayashi, N.; Itaya, K.. Chem. Comm. 2006,

500-502.

188 Yokoyama, T.; Kamikado, T.; Yokoyama, S.; Mashiko, S. J. Chem. Phys. 2004, 121,

11993.

189 Nicholls, D.; McKinzie, W. P.; Oncel, N. J. Phys. Chem. C 2010, 114, 14983-14985.

190 Lei, S. B.; Wang, C.; Yin, S. X.; Wang, H. N.; Xi, F.; Liu, H.W.; Xu, B.; Wan, L.J.;

Bai, C. L. J. Phys. Chem. B 2001, 105, 10838-10841.

191 Ruoff, R. S.; Kadish, K. M.; Boulas, P.; Chen, E. C. M. J. Phys. Chem. 1995, 99,

8843-50.

192 Seki, K. Mol. Cryst. Liq. Cryst. 1989, 171, 255-270.

193 Nakato, Y.; Abe, K.; Tsubomura, H. Chem. Phys. Lett. 1976, 39, 358-361.

194 Davis, D. G. “The Porphyrins” Ed: Dolphin, D.; Chapter 4, 1978, 143-144.

195 Piet, D. P.; Danovich, D.; Zuilhof, H.; Sudholter, E. J. R. J. Chem. Soc., Perkin

Trans. 2, 1999, 1653–1661.

196 Gruhn, N. E.; Lichtenberger, L.; Ogura, H.; Walker, F. A. Inorg. Chem. 1999, 38,

4023-4027.

197 Lukasczyk, T.; Flechtner, K.; Merte, L. R.; Jux, N.; Maier, F.; Gottfried, J. M.;

Steinruck, H-P. J. Phys. Chem. C 2007, 111, 3090-3098

198 Lukasczyk, T.; Flechtner, K.; Merte, L. R.; Jux, N.; Maier, F.; Gottfried, J. M.;

Steinruck, H-P. J. Phys. Chem. C 2007, 111, 3090-3098

199 Kruk, M. M.; Braslavsky, S. E. J. Phys. Chem. A 2006, 110, 3414-3425.

262

200 Kobayashia, Y.; Kusakabe, Koichi; Fukui, Ken-ichi; Enoki, T. Physica E 2006, 34,

678-681.

201 Nicholls, D.; McKinzie, W. P.; Oncel, N. J. Phys. Chem. C 2010, 114, 14983-14985.

202 Giunta, P.L.; Kelty, S. P. Chem. Phys. 2001, 114, 1807-1812.

203 Bredie, J.; Linares, M.; Lensen, R.; Rowan, A. E.; Funk, M.; Broring, M.; Hoffmann,

G.; Wiesendanger, R. J. Vac. Sci. Technol. B 2009, 27, 799-804.

204 Segal, D.; Nitzan, A.; Davis, W. B.; Wasielewski, M. R.; Ratner, M. A. J. Phys.

Chem. B 2000, 104, 3817-3829.

205 Holmlin, R. E.; Haag, R.; Dhabinyc, M. L.; Ismagilov, R.; Cohen, A.; Terfort, A.;

Rampi, M.; Whietsides, G. M. J. Am Chem. Soc. 2001, 123, 5075-5085.

206 Sedghi G.; Sawada, K.; Esdaile, L. J.; Hoffmann, M.; Anderson, H. L.; Bethell, D.;

Haiss, W.; Higgins, S. J.; Nichols, R. J. J. AM. CHEM. SOC. 2008, 130, 8582-8583.

207 Wold, D. J.; Haag, R.; Rampi, M. A.; Frisbie, C. D.; J. Phys. Chem. B 2002, 106,

2813-2816.

208 Hirota, S.; Itoh, U.; Takada, K. Thin Solid Films 1988, 165, 337-345.

209 Scudiero, L.; Barlow, D. E.; Mazur, U.; Hipps, K. W. J. Am. Chem. Soc. 2001, 123,

4073-4080.

263