1..
z... .1

 

 

 

:lrll (1-.

I...‘

 

 

 

 

5.3.2.52

 

A:
4‘

#51:
. .
.5wa

. 43...}. 4F
.Jndunfi. .. . v:
WWLEy1Em} .3

n {

"m3. ..

Amman”... 7

5+1 "irfi...
.

Ju
rum n»);
P.»

i

' ha"

how

u... an? ‘

s.
1 s1 :2:
.H . . Lulu... gknqh:
241 fl... ahnwbrn..»vlruMfia¢
. .90.. 9.9.3”... .inr... lulu
» .s: 2.3.519.

\
..- ,Iv’ Its...

a” fix.§.§flifluu .7... IL. .

MICHIGAN STATE UNIV TY LIBRARIES

I ll)!!! ll llllllllllull/Ill

3 1293 01564 3335

ll

 

 

 

 

 

 

 

 

 

 

 

 

Hi

This is to certify that the
dissertation entitled

THE FORMATION, STRUCTURAL DYNAMICS , AND SURFACE
PROPERTIES OF THIOL/CDLD SELF ASSEMBLED NDNOIAYERS

presented by
David Stephen Karpovich

has been accepted towards fulfillment
of the requirements for

PhoD. degree in meniStI—L

Major professor

Date October 7, 1996

MSUL! an Affirmative Action/Equal Opportunity Institution 0-12771

" J' .- q' - ' w J ‘5'

v f v v ' H‘H "

1"

 

LIBRARY
Mlchlgan State
University

 

 

 

Puce In RETURN

BOX to
TO AVOID FINES Mum on $213333” M VOW record.

    

fi

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

if“?
l
j
l
L_
r!
j

 

 

THE FORMATION, STRUCTURAL DYNAMICS, AND SURFACE PROPERTIES
OF THIOL/GOLD SELF AS SEMBLED MONOLAYERS

By

David Stephen Karpovich

A DISSERTATION
Submitted to
Michigan State University

in partial fiilfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

1 996

ABSTRACT

THE FORMATION, STRUCTURAL DYNAMICS, AND SURFACE
PROPERTIES OF THIOL/GOLD SELF ASSEMBLED MONOLAYERS

By

David Stephen Karpovich

This thesis reports on the molecular-level organization and dynamical behavior of
alkanethiol/gold self-assembled monolayers. The fimdamental understanding developed
here will allow greater preparative control and predictability in designing and building
interfaces. The mass uptake associated with the formation of alkanethiol/gold
monolayers was measured in-situ and in real time using a quartz crystal microbalance.
By modeling the data using the Langmuir isotherm, thermodynamic parameters for
monolayer formation were obtained. Kinetic data show that mass equilibrium is achieved
in less than one minute. The formation process is not complete with initial adsorption,
but includes annealing of the aliphatic chains to an ordered, all-trans conformation.
External reflectance FTIR studies show that the time-course of this annealing process
takes hours to days, and the initial adsorption conditions greatly affect the extent of the
annealing.

The fluorescence response of pyrene was studied for possible use as a
spectroscopic probe of alkanethiol/gold monolayer properties. The relative emission
band intensities of pyrene were found to be governed by the extent of vibronic coupling

between the weakly allowed first excited singlet state and the strongly allowed second

excited singlet state. This was the first fundamental photophysical explanation for the
pyrene “py” polarity scale.

Probe molecule reorientation experiments were performed on tethered pyrene in
an alkanethiol/gold monolayer. These experiments indicated that the local environments
near the aliphatic chains are rigid, and the tilt angle of the aliphatic chains is 26° with
respect to the surface normal. These results are fully consistent with known properties
of alkanethiol/gold monolayers lending credibility to this new measurement scheme.

The mass and thickness of adsorbed layers of organic molecules on gold and
thiol-modified gold were measured in-situ. These measurements showed no chemical
specific interactions between the adsorbates and substrates. Instead it was found that the
interactions of the adsorbates with the substrates are dominated by their polarizabilities.
The mass versus thickness data yielded the densities of adsorbed layers which were
higher than corresponding bulk liquid densities. The packing densities of the adsorbed
layers are consistent with site specific adsorption indicating the importance of substrate

organization on the packing of adsorbed molecules.

To My Family

iv

ACKNOWLEDGMENTS

I wish to thank Professor Gary Blanchard, my thesis advisor and fiiend, for his
encouragement, motivation, and advice. I am also grateful to my other committee
members: Professors Kris Berglund, Marcos Dantus, and James Jackson for the
invaluable professional and intellectual advice they have given me. Thanks to the
Blanchard group members who were always willing to discuss ideas, listen to practice
talks, and lend a hand. Heather Schessler and Saean Longiaru deserve special thanks for
the great job they did on their research projects which contributed to this thesis. In
addition, I wish to thank the guys in the electronics shop, the glass shop, and the
machine shop. They were always willing to work with me to design and build the many
unique items critical for my experiments. I especially want to thank my parents and my
wife Tonya for their unconditional love and support.

Finally, I would like to thank the National Science Foundation for support of this

research through grants CHE-9241237 and CHE-95-O8763.

TABLE OF CONTENTS

Page
List of Tables .............................................................................................................. viii
List of Figures ............................................................................................................... x
Chapter 1: Introduction ................................................................................................ 1
1.1. Literature Cited ...................................................................................... 5
Chapter 2: Direct Measurement of the Adsorption Kinetics of Alkanethiol Self-
Assembled Monolayers on a Microcrystalline Gold Surface ......................... 9
2. 1 . Introduction .......................................................................................... 10
2.2. Experimental ........................................................................................ 12
2.3. Results and Discussion .......................................................................... 15
2.4. Conclusion ........................................................................................... 43
2.5. Literature Cited .................................................................................... 43
Chapter 3: Quantitating the Balance Between Enthalpic and Entropic Forces in
Alkanethiol/Gold Monolayer Self-Assembly .............................................. 47
3. 1 . Introduction .......................................................................................... 48
3 .2. Experimental ........................................................................................ 5 1
3.4. Results and Discussion .......................................................................... 53
3 .5. Conclusions .......................................................................................... 71
3.6. Literature Cited .................................................................................... 72
Chapter 4: Relating the Polarity-Dependent Fluorescence Response of Pyrene to
Vibronic Coupling. Achieving a Fundamental Understanding of the py
Polarity Scale ............................................................................................ 77
4. 1. Introduction .......................................................................................... 78
4.2. Experimental ........................................................................................ 80
4.3. Results and Discussion .......................................................................... 83
4.4. Conclusions ......................................................................................... 106
4.5. Literature Cited ................................................................................... 106

vi

Chapter 5: Dynamics of a Tethered Chromophore Imbedded in a Self-Assembled

Monolayer ............................................................................................. 109
5. 1. Introduction ....................................................................................... 1 10
5.2. Experimental ..................................................................................... 1 12
5.3. Results and Discussion ....................................................................... 113
5 .4. Conclusions ....................................................................................... l 19
5.5. Literature Cited ................................................................................. 119

Chapter 6: Adsorption of Organic Molecules onto Metal and Modified Interfaces.

Evidence for Site-Specific Adsorption .................................................... 122

6. 1. Introduction ....................................................................................... 123

6.2. Experimental ................................................................... , .................. l 25

6.3. Results and Discussion ....................................................................... 132

6.4. Conclusions ....................................................................................... 147

6.5. Literature Cited ................................................................................. 147
Chapter 7: Conclusions ............................................................................................ 151

vii

Table

Table 2.1.

Table 2.2.

Table 2.3.

Table 2.4.

Table 2.5

Table 3.1.

Table 3.2.

Table 3.3.

Table 4.1.

Table 4.2.

Table 6.1.

Table 6.2.

LIST OFTABLES

Page
kob, values determined from raw data as a fiinction of thiol concentration.
............................................................................................................. 27
k, and k; determined from concentration dependence of k0,”. ................. 28

Fractional surface coverage as a fiinction of thiol concentration, calculated
using Eq. 2.6, for the I-octadecanethiol/n-hexane system. The uncertainty
is determined by uncertainty in k, and kd ................................................ 30

K“, and AG“. determined from the experimental data ............................ 3O

Limiting initial rates of monolayer formation for high thiol concentration.
............................................................................................................. 35

km values determined from raw data as a function of thiol concentration.
............................................................................................................. 62

Adsorption and desorption rate constants (k, and k4) from fits of Eq. 3.4
and kg,” = kaC + k, to the experimental data .......................................... 64

Temperature Dependence of the Equilibrium Constant (K2,) and the

Gibbs Free Energy of Adsorption (AGW) .............................................. 64
Properties of the optical response of pyrene as a fiinction of solvent.

Uncertanties are $16 ........................................................................... 100
Polar properties of the solvents used in this work ................................ 101

Saturation concentrations of organic vapors in the “wet” nitrogen gas
stream ................................................................................................. 130

Maximum mass and thickness uptake of adsorbates onto substrates ..... 133

viii

Table 6.3 Mass vs. thickness increase of adsorbates on substrates ....................... 139
Table 6.4. Adsorbed layer densities calculated from mass/thickness values ........... 142

Table 6.5. Densities of adsorbates (g/cm3) ........................................................... 143

ix

Figure

Figure 2.1.

Figure 2.2.

Figure 2.3.

Figure 2.4.

Figure 2.5.

Figure 2.6.

Figure 2.7.

Figure 3.1.

Figure 3.2.

LIST OF FIGURES

Page

Representative raw data of QCM oscillation frequency vs. time
subsequent to thiol injection .................................................................. 17

Thiol concentration dependence of QCM frequency change for
l-octadecanethiol in n-hexane. The top panel is data for a 3x1045 M
thiol concentration, the center panel is for 3x10'5 M thiol concentration
and the bottom panel is for a 3x10'4 M thiol concentration .................... 18

Fit of Langmuir adsorption isotherm (Eq. 2.4) to experimental
data ...................................................................................................... 25

Concentration dependence of km for l-octadecanethiol in n-hexane. The
best fit line through the data has a slope of k, and an intercept of kd. See
Table 2.2 for the results of the fits to the data ....................................... 26

Concentration dependence of A for ka = 2000 M‘ls'l and kd = 0.2 s'1 ...... 32

Concentration dependence for the initial rate of monolayer formation for
l-octanethiol adsorbed from n-hexane. Note the high concentration
region where the rate becomes concentration independent ..................... 33

FTIR spectra of a l-octadecanethingold monolayer formed on a QCM
surface, taken with 2 cm'1 resolution. (a) Spectrum recorded immediately
after adsorption. (b) Spectrum recorded two weeks after adsorption....38

Concentration dependence of kobs at 293 K. From these data we obtain
best fit values ofk, = 2278 i 92 M‘s" and k, = 0.08 i 0.02 s'1 .............. 61

Temperature dependence of (a) AG“. and (b) Keg. The regressed line
shown in (a) yields a slope of -AS.¢. = 48 :t 1 cal/mole-K and an
intercept of AH“. = -l9.9 3: 0.4 kcal/mol ............................................... 63

Figure 3.3.

Figure 3.4.

Figure 4.1.

Figure 4.2.

Figure 4.3.

Figure 4.4.

Figure 4.5.

Figure 4.6.

Figure 4.7.

Figure 4.8.

Figure 5.1.

Van’t Hoff plot of In K“, vs. T". From these data, we determined AH“. =
-20 :i: I kcal/mol .................................................................................... 66

Temperature dependence of the observed rate constants, km. These data
are essentially temperature independent, to within the uncertainty of the
measurements. No linear dependence of km on T is detectable from
these data ............................................................................................. 68

Time correlated single photon counting spectrometer used for
fluorescence lifetime measurements ....................................................... 82

Fluorescence decay data for pyrene in cyclohexane showing both the
instrument response function (solid line at t ~ 30 ns) and the spontaneous
emission decay (dots) ............................................................................ 84

Absorption and spontaneous emission spectra of pyrene in cyclohexane.
We report the S0 —> 81 absorption spectrum with the intensity of the 0-0
band normalized to the 0-0 emission transition ...................................... 91

(a) Radiative transition rate constant, kg, for pyrene in the solvents used
in this work plotted versus emperical solvent polarity parameter, py. (b)
Dependence of fluorescence quantum yield, (bu, on solvent polarity.
Uncertainties reported in these figures are j: 10 ..................................... 94

Radiative transition rate constant for the 0-0 transition of pyrene, plotted
versus solvent py value. The reported uncertainties are i 10 ................ 96

Radiative transition rate constant for the family of 0-v transitions that
comprise the dominant portion of the pyrene emission band, plotted
versus solvent py value. The reported uncertainties are i lo ................ 97

Nonradiative transition rate constant for the pyrene emission band,
plotted versus solvent py value. The reported uncertianties are i 10' ..... 98

Representations of nuclear motions associated with the four b 38
vibrational modes in closest energetic proximity to the “111” band of
pyrene. The frequencies reported for these modes were taken from
references 38 and 39 ........................................................................... 104

Qualitative schematic diagram of the alkanethiol/gold interface with the
tethered 4-(l-pyrenyl)-l-butanethiol Chromophore present .................. 1 11

Figure 5.2.

Figure 5.3.

Figure 5.4.

Figure 6.1.

Figure 6.2.

Figure 6.3.

Schematic of the hindered rotor model used in interpretation of the R(t)
data. The cone semi-angle, 200, and the average Chromophore tilt angle,
9, are indicated ................................................................................... 114

Experimental time resolved emission intensity data for a monolayer of
5% 4-(1-pyrenyl)-1-butanethiol and 95% l-octadecanethiol on Au. The
data (solid circles) are plotted on a logo scale to underscore the non
exponential nature of the decay. The instrumentresponse fiinction is
shown as open circles .......................................................................... 116

Top panel: Experimental emission intensity data for a monolayer
comprised of 5% 4-(l-pyrenyl)-l-butanethiol and 95% l-octadecanethiol
on Au. Note that I 1 is more intense than I" as is expected for pyrene.
Bottom panel: R(t) data calculated from the raw experimental data using
Eq. 5.1 ................................................................................................ 118

Top View schematic of the atmospheric chamber with incorporated

ellipsometer. GLP = Glan-laser prism, P = polarizer, A = analyzer, QWP
= quarter wave plate, PMT = photomultiplier detector and QCM = quartz
crystal microbalance ............................................................................ 128

Schematic of the vapor generation and mixing system ......................... 129

Plot of adsorbed mass vs. adlayer thickness of a gold surface while
exposed to an atmosphere of n-hexane in nitrogen gas ........................ 140

xii

Chapter 1

Introduction

Self-assembled monolayers (SAMs) were discovered in the 1980s by researchers

“’2’ The first SAMs studied were composed of n-alkanoic acids

at Bell Laboratories.
adsorbed onto aluminum oxide surfaces. These monolayers formed spontaneously upon
immersion of the aluminum oxide surface into an n-alkanoic acid in hexadecane solution.
Investigation of these SAMs showed them to be very stable, chemisorbed structures.
These systems were also found to be highly oriented and crystalline in nature and
therefore have been thought to be ideal substrates for studying basic interfacial
phenomena associated with chemical separations, molecular recognition and microdevice
patterning. The widespread applicability of SAMs spawned fiirther investigation into
this field which resulted in SAMs being formed from many different chemical systems
including carboxylic acids, thiols and disulfides as adsorbates and silica, silver oxide and
gold as substratesp'él Of these systems, one of the most widely investigated has been
thiol/gold monolayers because of the facile nature of their formation and the intrinsic
order available with metal substrates. [3'18]

Because of the different nature of the “head” and “tail” groups, the study of
organization in these systems has been effectively separated into examinations of
structure of each group. Studies of alkanethiol/gold monolayers have included
investigation of aliphatic chain structure by external reflectance FTIR spectrometrylg’lg]

The arrangement of the alkanethiols on the gold was studied by atomic scale

nricroscopy.[2°'23] Helium difi‘raction was used to study the aliphatic group arrangement

1

2
of the SAMs.”6’2°] In general, it was found that the aliphatic chains resided in a very

ordered, crystalline environment. The alkanethiols were found to exhibit a packing
arrangement that was determined by the gold structure. The degree of ordering of
alkanethiol/gold monolayers varied greatly depending on the length of their aliphatic
chainlm For example, alkanethiols with less than ten carbons in length exhibited less
aliphatic chain ordering than longer alkanethiols. From this information it is clear that
interactions between adsorbates play an important role in the organization of the
aliphatic functionalities.

Contact angle measurements of liquid drops on alkanthiol/gold monolayers were
used to study surface properties.”’” These studies indicated that alkanethiol gold
monolayers were hydrophobic when compared to clean metallic gold surfaces.
Thiol/gold monolayers formed using carboxylic acid or hydroxyl terminated aliphatic
thiols were shown to be more interactive with water than the corresponding alkane
terminated monolayer. These simple yet informative measurements indicated that the
surface pr0perties of the thiol/gold monolayers could be altered significantly using thiols
with functionalized terminal groups. Eventually, studies of the interaction of substances
with thiol/gold monolayers focused around specific chemical recognitionlzmn

While the average equilibrium properties of alkanethiol/gold monolayers have
been studied extensively, comparatively little is understood about the formation of these
SAMs. Before more elaborate SAM systems can be synthesized, knowledge of the
details of alkanethiol/gold formation are needed. The few early formation studies used

FTIR, optical ellipsometry, advancing contact angle measurements, and radio isotope

labeling to monitor (ex-situ) the formation of alkanethiol/gold monolayerslflml] The

3

conclusions of these formation studies centered around optimal thiol solution
composition and time of exposure of the gold surface to that solution for proper
monolayer formation. A limitation of these formation studies was that the time
resolution of the ex-situ techniques was on the order of several minutes at best so that
any shorter time processes could not be resolved.

When the research for this thesis began, there were certain important concerns
remaining about alkanethiol/gold monolayers despite the apparent maturity of the field.
Most of the existing information on SAMs involved their equilibrium structures. More
specific information about alkanethiol/gold formation was needed, especially due to the
demand for tailor-made monolayers. More information was also needed regarding their
structure on molecular length scales, which would play an important role in determining
the utility of these systems for molecular recognition applications. Although the average
mesoscopic structural properties of the monolayers had been well studied,[8"8"9’2"32]
information was needed regarded the local rigidity of the aliphatic chains. Finally, more
information was needed on how alkanethingold monolayers, as well as interfaces in
general, interact with vapor phase molecules, data which could be relevant to chemical
sensing and molecular recognition. These chemical issues are addressed in this thesis
using a host of measurement techniques developed during the course of this work.

In Chapter 2 the formation kinetics of alkanethiol/gold monolayers were
measured in-situ and in real time using a quartz crystal microbalance for direct solution

[33’3“ This technique allows monolayer formation to be studied

measurements.
gravimetrically with sub-second time resolution. The formation process of alkanethiol

gold monolayers was found to occur in two steps, and over two widely different time-

4
scales. The first step of formation, occurring in seconds, is the adsorption of thiols from

solution onto the gold. The second step, occurring in hours to days, is the annealing of
the aliphatic chains to ordered, crystalline-like structures. From the gravimetric data, the
equilibrium constants and the free energies of formation for alkanethiols onto gold were
found. It is concluded in this chapter that the alkanethiol/gold monolayer systems were
not “brick-like” structures as previously thought, but rather the alkanethiols exhibited
mobility across the gold surface.

Chapter 3 reports on the temperature dependence of the free energies of
formation of 1-octadecanethiol/gold monolayers?” From the temperature dependence
of the equilibrium constants, the enthalpy of formation was determined. The entropy of
formation was then found from the enthalpy and the free energies. These quantities
serve to provide a measure of the robustness of the monolayers. This chapter includes a
detailed discussion on the adsorption and desorption reactions involved in
alkanethiol/gold formation.

Chapter 4 is a fundamental study of the spectroscopy of pyrene“) The study of
pyrene was necessary for this research since pyrene was to be used as a probe molecule
to investigate the dynamic properties of the SAMs on a short timescale (described in
Chapter 5). The study described in this chapter is also useful for the general scientific
community. Pyrene has been used as a probe for local polarity because its emission
spectrum varies predictably with the polarity of the environment in which it resides.
While this behavior of pyrene and some other PAHs has been well documented, little was
known about the spectroscopic mechanism for this behavior. The mechanism is outlined

in this chapter.

5
Chapter 5 describes a study of the structure of the local environment around the

aliphatic chains on short, picosecond time-scalesml These studies were done by
performing orientational relaxation dynamics experiments on tethered pyrene
chromophores within alkanthiol/gold monolayers. The hindered rotor model was used to
describe the motion of the tethered pyrene. It was found that the local environments
formed by the aliphatic chains are very rigid. The results, while intuitive, demonstrate a
new measurement scheme for probing molecular dynamics within interfacial films and
unify the notion of stnicturally rigid films.

Chapters 2, 3, and 5 focus on answering the fundamental question regarding
formation and structural properties of these systems. Chapter 6 focuses on the
interaction of these interfaces with other moleculesm] which are important not only from
a fiindamental perspective, but also in the context of the chemical recognition interest
and possibly separations science interest. The results are consistent with firndamental
laws governing adsorbate/substrate interactions whether describing chemisorbed
alkanethiols/gold or physisorbed low molecular weight molecules onto an interface. It
was found in these studies that the interactions of molecules at interfaces cannot be

adequately described the same as interactions of bulk molecules.

1.1. Literature Cited

1. D. L. Allara, R. G. Nuzzo, Langmuir, 1985, 1, 45.
2. D. L. Allara, R. G. Nuzzo, Langmuir, 1985, 1, 52.

3. G. M. Whitesides, P. E. Laibinis Langmuir, 1990, 6, 87.

' 4. A. Ulman, Introduction to 7am Organic Films: From Langmuir-Blodgett to Self-
Assembly. Academic Press: Boston. 1991.

5. S. H. Chen, C. W. Frank, Langmuir, 1989, 5, 978.

O\

. P. Fenter, P. Eisenberger, J. Li, N. Camillone III, S. Bemasek, G. Scoles, T. A.
Ramanarayanan, K. S. Liang, Langmuir, 1991, 7, 2013.

\l

. L. H. Dubois, B. R. Zegarski, R. G. Nuzzo, J. Am. Chem. Soc, 1990, 112, 570.
8. R. G. Nuzzo, L. H. Dubois, D. L. Allara, J. Am. Chem. Soc, 1990, 112, 558

9. R. G. Nuzzo, E. M. Korenic, L. H. Dubois, J. Chem. Phys, 1990, 93, 767.

10. C. E. D. Chidsey, D. N. Loiacono, Langmuir, 1990, 6, 682.

11. C. E. D. Chidsey, C. R. Bertozzi, T. M. Putvinski, A. M. Mujscu, J. Am. Chem.
Soc., 1990, 112, 4301.

12. C. D. Bain, E. B. Troughton, Y. T. Tao, J. Evall, G. M. Whitesides, R. G. Nuzzo, J.
Am. Chem. Soc, 1989, 111, 321.

13. C. D. Bain, J. Evall, G. M. Whitesides, J. Am. Chem. Soc, 1989, 111, 7155.

14. E. B. Troughton, C. D. Bain, G. M. Whitesides, R. G. Nuzzo, D. L. Allara, M. D.
Porter, Langmuir, 1988, 4, 365.

15. M. D. Porter, T. B. Bright, D. L. Allara, C. E. D. Chidsey, J. Am. Chem. Soc,
l987,109,3559.

16. N. Camillone HI, C. E. D. Chidsey, G.-Y. Liu, T. M. Putvinski, G. C. Scoles, J.
Wang, J. Chem. Phys., 1991, 94, 8493.

17. C. E. D. Chidsey, G.-Y. Liu, G. C. Scoles, J. Wang, Langmuir, 1990, 6, 1804.

18. S. R. Wasserrnan, G. M. Whitesides, I. M. Tidswell, B. M. Ocko, P. S. Pershan, J.
D. Axe, J. Am. Chem. Soc., 1989, 111, 5852.

19. A. N. Parikh, D. L. Allara, J. Chem. Phys, 1992, 96, 927.

20. C. A. Widrig, C. A. Alves, M. D. Porter, J. Am. Chem. Soc., 1991, 113, 2805.
21. G.-Y. Liu; M. B. Salmeron; Langmuir, 1994, 10, 367.

22. R. L. McCarley, Y.-T. Kim, A. J. Bard, J. Phys. Chem, 1993, 97, 211.

23. R. L. McCarley, D. J. Dunaway, R. J. Willicut, Langmuir, 1993, 9, 2775.

24. L. Sun, L. J. Kepley, R. M. Crooks, Langmuir, 1992, 8, 2101

25. L. J. Kepley, R. M. Crooks, A. J. Ricco, Anal. Chem, 1992, 64, 3191.

26. O. Chailapakul, R. M. Crooks, Langmuir, 1993, 9, 884.

27. O. Chailapakul, R. M. Crooks, Langmuir, 1995, 11, 1329.

28. K. D. Truong, P. A. Rowntree, submitted to Langmuir.

29. C. D. Bain, E. B. Troughton, Y. T. Tao, J. Evall, G. M. Whitesides, R. G. Nuzzo, J.
Am. Chem. Soc., 1989, 111, 321.

30. E. B. Troughton, C. B. Bain, G. M. Whitesides, R. G. Nuzzo, D. L. Allara, M. D.
Porter, Langmuir, 1988, 4, 365.

31. J. B. Schlenoff, M. Li, H. Ly, J. Am. Chem. Soc., 1995, 117, 12528.
32. L. Strong, G. M. Whitesides, Langmuir, 1988, 4, 546.
33. D. S. Karpovich, G. J. Blanchard, Langmuir, 1994,10, 3315.

34. D. S'. Karpovich, G. J. Blanchard, J. Chem. Ed, 1995,72, 466.

35. H. M. Schessler, D. S. Karpovich, G. J. Blanchard, J. Am. Chem. Soc., in press.
36. D. S. Karpovich, G. J. Blanchard, J. Phys. Chem, 1995,99, 3951.
37. D. S. Karpovich, G. J. Blanchard, Langmuir, in press.

38. D. S. Karpovich, G. J. Blanchard, Langmuir, in review

Chapter 2

Direct Measurement of the Adsorption Kinetics of Alkanethiol Self-
Assembled Monolayers on a Microcrystalline Gold Surface

Summary

We report measurements on the adsorption kinetics of alkanethiol monolayers onto a
microcrystalline gold surface. Our data indicate that the formation of this monolayer is
rapid and is described by the Langmuir adsorption isotherm. The concentration
dependence of the monolayer formation rate indicates that there exists an equilibrium for
the adsorption/desorption process. We determine from these measurements AG“, = - 5.5
kcal/mol for 1-C13H37SH in n-hexane and cyclohexane adsorbing onto gold and A6,... = -

4.4 kcal/mol for l-CanSH in n-hexane adsorbing onto gold.

10
2.1. Introduction

The study of organic interfacial monolayer systems has attracted a great deal of
research attention over the past decade because of the potential utility of these interfaces
in materials science and chemical separation applications. Initial work on the structural
and chemical properties of interfacial organic monolayers supported on amorphous
substrates required statistical treatment to understand the role of their intrinsic
heterogeneity. Recent chemical advances in the formation of single molecular layers on

”"2' surfaces has allowed the structure of these

highly ordered metaln'” and dielectric
assemblies to be examined more directly by removing structural variability induced by the
surface. Much of the recent work on self-assembling organic monolayers at metal
surfaces has focused on the alkanethiol/gold system, where alkanethiols adsorb
spontaneously onto the metal surface to form a highly ordered array.“3'25l A variety of
tools have been applied in studies of these self-assembled monolayers to elucidate their
molecular-scale structural properties. FTIR spectroscopy has been used most commonly
for the examination of molecular ordering in these films, and these results have shown
that the interface is composed of densely packed, virtually all-trans aliphatic chains that
are tilted at ~30° with respect to the surface normalllsl Ellipsometric and contact angle
measurements have indicated that these films are uniform over macroscopic distances
(several um), and recent atomic scale microscopy has demonstrated the existence of

[26-29]

highly regular structure on molecular length scales. Based on this large body of
information, the alkanethiol/gold system has come to be viewed as largely immobile,

once formed.

ll

McCarley’s recent scanning probe microscopy work has called into question the
extent to which these monolayers are “fixed” in place subsequent to adsorption.” The
island structures formed by the alkanethiol monolayers on an evaporated gold substrate
exhibited macroscopic shape changes over periods as short as minutes, and the McCarley
group has concluded that these shape changes are associated with mobile defects in the
gold surface and that the Au-S bond is labileml This finding indicates that, either the
formation of the alkanethiol monolayer has a substantial efi‘ect on the mobility of the
surface gold atoms, or that the overlayer itself exhibits structural mobility.

Virtually all examinations of the alkanethiol/gold monolayer system have been
performed on monolayers that were formed prior to examination, and little effort has
been spent on understanding the elementary steps in the formation of the monolayers.
We have chosen to examine the formation of these monolayers gravimetrically. We have
measured the rate at which these monolayers form in real time by monitoring the change
in mass of a quartz crystal microbalance made with evaporated gold electrodes. These
data show that the alkanethiol/gold system exhibits a measurable equilibrium between
adsorbed thiol and the free species. From our data we have determined that the free
energy of adsorption of the alkanethiol on gold is ~ -5 kcal/mol, and this value depends
slightly on the aliphatic chain length of the alkanethiol adsorbate. We therefore expect
that, at any given time, a significant fraction of surface active gold sites will be exposed
and thus macroscopic changes in the appearance of the surface are likely to occur. Our
findings also imply that the rate of electron transfer across these monolayers is mediated

by the equilibrium dynamics of the interface.

12
2.2. Experimental

Kinetic measurements: We used a quartz crystal microbalance (QCM) to study
the adsorption kinetics of alkanethiols onto gold. A quartz crystal microbalance with a
resonance frequency of ~6 MHz is sensitive to mass changes as small as several
nanogramsml The electrodes of the QCM we use are vapor deposited gold that posses
a polycrystalline, predominantly (111), surface. There are several factors in addition to
the mass loading of the QCM that affect its oscillation frequency. The resonance
frequency of the QCM depends on mass loading according to the Sauerbrey equationl3 1’
This relationship holds well for gas phase measurements, but in situ solution phase
measurements using the QCM are, in many cases, only semi-quantitative due to
capacitive and colligative effects arising from the presence of the solvent and the
adlayer(s). Our primary focus in this work lies in determining the rate of mass change
associated with the formation of the alkanethiol monolayer, and we do not attempt to
extract absolute mass change information from our data.

The QCM was suspended in solution with isolated electrical connections for both
gold electrodes using a modified spring clip. AT -cut quartz crystal microbalances (6
MHz resonance frequency) with vapor deposited gold electrodes were obtained from
McCoy Electronics (part number 78-18—4). These QCM devices were sufficiently robust
to avoid breakage fi'om the spring clip pressure required to ensure reliable electrical
contact. The QCM was connected to a 6 MHz oscillator circuit (Maxtek part number
124200-4). A frequency counter (Phillips model PM 6673) was used to monitor the

QCM oscillation fiequency. This frequency counter has 1 Hz resolution with a gate time

of 280 ms. The analog output of this frequency counter was sent to a computer for data

l3

processing and storage. For all adsorption rate measurements the data were acquired at
a rate of ten readings per second to ensure adequate sampling of the kinetic response.
All electrical connections were made using shielded coaxial cables to minimize QCM
frequency instability arising from electrical interference.

The resonant frequency of a QCM is strongly temperature dependent. To
minimize QCM frequency drift associated with thermal fluctuations, the temperature of
the solution and the QCM were controlled to within : 0.05 K. Measurements were
performed in a 150 mL jacketed beaker connected to a flowing liquid temperature
controller (Neslab model RTE-110) set to 293.0 K. Total solution volume in the
jacketed beaker was 100.0 mL for each measurement. Stirring was accomplished with a
magnetic stirrer and a Teflon“9 coated stir bar. The stirring speed was set to be as fast as
possible without introducing fluctuations to the QCM baseline frequency.

Glassware was cleaned using a chromic acid solution, rinsed with distilled water
and subsequently the solvent of interest prior to each measurement. The QCMs were
cleaned with piranha solution (1:3 H202 (30%)/HzSO4) for 5 minutes, rinsed with
deionized water and dried in a stream of high purity N2. The cleaned QCMs were then
rinsed with tetrahydrofuran to remove any residual water, rinsed in the solvent to be used
in the experiment (n-hexane or cyclohexane) and used immediately thereafter. Prolonged
exposure to piranha solution damaged the QCM electrodes, and this damage was
manifested by an increase in frequency fluctuations for the baseline readings. A new
crystal was used approximately every third run, as determined by the stability of the
baseline frequency. Individual monolayer adsorption measurements began with pure

solvent in the temperature controlled vessel to establish a stable QCM baseline

l4
oscillation frequency. An aliquot of stock thiol solution was then introduced by syringe.

The stock solutions were of high enough thiol concentration to allow small injection
volumes, minimizing mechanical disruption of the reaction vessel contents. For 10" M
thiol final concentration, 99.0 mL of solvent was introduced to the jacketed beaker and,
after thermal equilibration, with stirring, data were collected for 10 seconds to establish a
QCM baseline frequency. Injection of 1.0 mL stock 10'2 M alkanethiol yielded a 10'4 M
final concentration in the reaction vessel. The stock solutions were maintained at the
same temperature as the solvent in the reaction vessel to avoid thermal disruptions from
injection.

External Reflection F TIR measurements: We obtained infrared spectra of the
adsorbed films via external reflection infrared spectroscopy using a Nicolet 520 Fourier
transform infrared spectrometer. The spectrometer was equipped with a cooled MCT
detector. The external reflection sample mount and optics used were obtained from
Harrick Scientific Corp. and the polarization of the IR light was controlled using a
Cambridge Physical Sciences broad band IR polarizer, The light was p-polarized and
was incident at 75° with respect to the surface normal of the QCM. Achieving an 86.5°
incidence angle to maximize the signal was not possible because of the geometric
constraints imposed by the QCM electrode dimensions. Spectral resolution was 2 cm’1
for the scans we report here.

Chemicals. The alkanethiols 1-octadecanethiol and l-octanethiol were
purchased from Aldrich Chemical Co. and were used as received. These thiols were
tested by gas chromatography and mass spectrometry to determine the presence of the

corresponding dialkyl disulfide, and none was detected. The solvents n-hexane,

15

cyclohexane, and tetrahydrofiiran were purchased from Baxter (Burdick and Jackson,

UV spec. grade) and were used without fiirther purification.
2.3. Results and Discussion

Despite extensive investigation of equilibrated alkanethiol monolayers, there is
little fundamental information on the kinetics or energetics of their formation. Methods
used commonly for the examination of alkanethiol-gold monolayers, such as external
reflection infrared spectroscopy, contact angle measurement and optical ellipsometry, are
sensitive to structural properties on a length scale many times longer than the aliphatic
chains that comprise the monolayer, but, more importantly, these methods have been
used to measure the response of an equilibrated chemical system. In the work we report
here, we determine the rate of mass change of the alkanethiol-gold interface resulting
from adsorption of alkanethiol fi'om solution onto gold. This kinetic response, arising
from systems that are initially far from equilibrium, provides fundamental information on
the initial steps of the alkanethiol monolayer self-assembly process.

To provide a frame of reference, we recap briefly what is known about
alkanethiol monolayers. Diffraction and scanning microscopic measurements have
shown that these systems form a hexagonal close packed monolayer structure on
Au(111) surfaces,[23’27’32’33l with infrared measurements showing that the all-trans
aliphatic chains of the alkanethiol are oriented at ~30° with respect to the gold surface
normal, and are rotated ~55° with respect to the plane formed by the surface normal and

“’1 These monolayers are packed densely, with an

the orientation vector of the chain.
interchain spacing of ~5A, established by He diffraction measurements and corroborated

by STM data.[23’26’32] While these monolayers were thought to be comparatively rigid

16
until recently, McCarley’s STM measurements have provided compelling evidence for

[29] In addition, several recent studies of

structural mobility within these monolayers.
electron transfer across these monolayers have suggested that the rate of transfer is
mediated by the density of defect sites. The data that we present in this paper provide an
explanation for these recent observations, and demonstrate that the apparently labile
nature of these monolayers is a natural consequence of the strength of the bond
responsible for the initial formation of the monolayer.

We have examined the formation kinetics for the alkanethiol-gold monolayer
from two different solvents and for two different alkanethiol aliphatic chain lengths. 1-
octadecanethiol was adsorbed onto the QCM electrodes from both n-hexane and
cyclohexane and 1-octanethiol was adsorbed from n-hexane. For each of these chemical
systems, data were collected over an alkanethiol concentration range large enough to
establish the region in which the Langmuir isotherm predictions hold. Raw data (QCM
resonant frequency vs. time after thiol introduction) for the adsorption of 1-
octadecanethiol from a 10" M solution in n-hexane are shown in Figure 2.1. It is
immediately apparent from these data that a chemical equilibrium is achieved within a
very short time (<1 minute) following thiol introduction. Further, the rate at which this
equilibrium is achieved depends on the concentration of the thiol (Figure 2.2), and it is

information from this concentration dependence that yields the free energy of adsorption

for these systems.

The adsorption reaction we detect is believed to bem]

Au(s) + RS — H(solv.) c> RS — Au(s) + §H2(solv.) 2.i

l7

 

 

 

 

 

80 —
70 l- :-.--.u..
60 -
>5
0
c:
9:3 .
g 40 ~ .
ct: .
30 — i.
20‘ ""'-.-.--‘.-.'-.-""-‘:--:.‘.-.:'2'
10 l l l r I 1 I 1 fl 1 ll
0 10 20 30 4O 50
time (s)

Figure 2.1. Representative raw data of QCM oscillation frequency vs. time subsequent
to thiol injection.

18

20_ M

 

 

 

‘V
10 -
- 3x10'6M
0 v
1 . 1 . 1 1 1 . 1
0 10 20 30 40
30 P

20— f

 

 

 

 

 

 

A l
a 10 - I
g" I 3x10'5M
4 - i
0 .9!
l r l r l 1 l 1 J
O 10 20 3O 40
3O '-
_ McMVWMWn
20 — f
_ I
10 - ;
0 v
I J 1 l 1 I 1 I
0 10 20 3O 40

time (seconds)

Figure 2.2. Thiol concentration dependence of QCM frequency change for 1-
octadecanethiol in n-hexane. The top panel is data for a 3x1045 M final thiol
concentration, the center panel is for a 3x10’5 M final thiol concentration and the bottom
panel is for a 3x10'4 M thiol concentration.

19
This reaction proceeds at a gold surface where the thiols react with the metal to form a

gold thiol monolayer (Au-SR) and hydrogen. Our experimental data show that the
Langmuir isotherm can be used to describe the kinetic response of this reaction over a
substantial concentration range (vide infra). The Langmuir isotherm is based on the
assumptions that adsorption is limited to one monolayer, all surface sites are equivalent,
and adsorption to one site is independent of the occupancy condition of the adjacent
sites. The first assumption is clearly valid based on the chemical identity of the
adsorbate. An equivalent expression of the second assumption is that the surface is
defect-flee. Clearly this assumption is not physically realistic. While there are a variety
of defects at the gold surface associated with grain boundaries, step edges, adatoms and
vacancies, the resultant modulation of surface site energies is apparently small enough
that the Langmuir approximation holds. The physical and chemical bases for the third
assumption are also open to significant question. The adsorbate alkyl chains interact
with each other, but this interaction is likely only significant for nearly complete surface
coverage because of the inherently short range nature of dispersive forces. In addition, it
is likely that the time scale over which these interchain interactions proceed is
significantly longer than the time scale associated with the formation of the gold-thiol
bond (vide infra). We note also that the Langmuir isotherm was derived initially for
physisorption, not chemisorption, but the bond strength of the alkanethiol-gold bond is

apparently small enough that this condition is not violated seriously (vide infra).

The effects of exposing the QCM electrodes to liquids have been studied by
Thompson’s group using network analysis as well as conventional series frequency

measurementsl.35’36] Network analysis involves measurement by standard circuit analysis

20

of four impedances in an equivalent circuit. These impedances are the static capacitance
Co, the motional resistance R..., the motional inductance Lu, and the motional capacitance
C... The static capacitance describes the QCM response at frequencies far from the
resonant frequency where it is a simple parallel-plate capacitor. The motional resistance
is related to the liquid density and viscosity such that Rm increases with increasing
viscosity and density. The motional inductance increases with the increase in plate
thickness associated with external mass loading. The motional capacitance represents
the elasticity of the QCM. Ideally, the above parameters, except the motional
inductance, will remain constant for over the time course of our measurements.

Given the motional resistance, and the measured frequency shift (also called the
series frequency Af,), the individual components of the frequency shift can be
estimatedl3 5’36]

M3=Md+4fm+Afa+Afx [20]
Where At], is the portion of the frequency shift related to viscous damping, Afm is caused
by adsorbed mass, Af. surface stress, and Afx is caused by energy dissipation by non-
shear couplings. We need to consider how the frequency shifts we report are affected by
each of these components.

According to Thompson’s experiments on a QCM in air with both bare gold and
thiol-modified gold surfaces, the viscoelastic properties of the QCM did not change with
adsorption and thus the motional capacitance and Afd did not change with the presence
of the thiol monolayer.[35'36] In other words, the thiol monolayer can be considered a
rigid film for purposes of the QCM measurements. In a comparison of the thiol modified

QCM and the bare gold QCM, Thompson noted that upon immersion into polar solvents

21
such as water, the frequency shifted more for the bare gold QCM than for the thiol-

modified QCM. This was attributed with the interfacial properties exhibited with water
at the hydrophilic and hydrophobic surfaces which would affect Afm and Afx.

Thompson’s work also involved a study of the effects of roughness of the
QCM’s surface on its behavior in liquids.l”‘36] He found that frequency shifts of polished
surfaces upon immersion into liquids were less than those of unpolished surfaces in the
same liquid. Electron microscopy showed that the polished and unpolished surfaces
differed greatly in roughness, with the rough, unpolished surfaces possessing a higher
number of adsorptive sites. The variability in surface roughness between individual
QCM’s introduces uncertainty in the number of active gold sites available for thiol
adsorption. There is also a mechanical stress on our QCMs due to the spring clip used
to hold the QCM. Although Af. would vary from one measurement to the next, this
variation should not cause a frequency shift (Afl) error since Af, would remain constant
through a single measurement. Changes in mechanical stress caused by the spring clip
are the result of variability in the positioning of the spring clip, which would affect the
amount of exposed gold surface available for the reaction.

The frequency changes we report for the formation of alkanethiol/gold
monolayers are not simply due to the mass uptake of the surface. As a consequence of
the adsorption, the interfacial monolayer is changing from hydrophilic to hydrophobic.
The energy of interaction of the surface with solvent, which does affect the frequency, is
therefore changing throughout the measurement in addition to the mass of the QCM.
For our measurements, the frequency shift (Afl) is affected by changing Afm and Afx (the

adsorbed mass and the energy dissipation by nonshear coupling) during a single

22
measurement. The net result of this complex process is that the frequency shifts from

our measurements are not representative of only the mass changes, and thus extraction
of fractional coverage information from these data is problematic at best. The time
dependence of this response is however, a direct result of the formation of the
monolayer, which affects both the mass uptake and changing surface/solvent
interactions. Our time domain measurements are sensitive to the kinetics of formation of
the monolayer. The run-to-run variation in Af. for our experiments results from variable
gold surface roughness of the QCMs (which affect Afm and Afx) and variability in
positioning the spring clip onto the QCM resulting in variability in the area of the
exposed gold.

We have a limited gauge of the roughness of the gold surface at the atomic
scale?” and while this information is important, it is not critical to our experiment for
the reasons we have discussed above. We discuss this point in more detail in Chaper 6,
where surface roughness is important to the interpretation of the data. The coverage of
the QCM electrode surface is expressed as a unitless quantity 9, the fraction of available
sites that have reacted, or equivalently, the fraction of a monolayer (0 .<_ 0 .<. 1). The
Langmuir isotherm dictates that the rate of surface reaction is given by

$441—$0de [2.1]

where 9 is the fraction of surface covered, (1-0) is the fraction of surface exposed, C is
the alkanethiol concentration. and k, and k; are the association and dissociation constants,

respectively. Integration of Eq. 2.1 yields the time course of the monolayer formation

23

e(t)=———— 1— ex (atkCi-kd) [2.2]
will—7; v(-( )1

This result can be simplified by the substitutions kob, = kaC + kd and K’ = C/(C+(kd /k,,)),
giving

9(1) = K'[l — exp(—ko,,,:)] [2.3]
Notice from the raw data, shown in Figure 2.2, that the maximum initial rate of
monolayer formation is not reached immediately upon injection. Depending on the
volume and concentration of thiol aliquot injected, it may take up to several seconds to
achieve the maximum adsorption rate because of limitations inherent to mixing of the
solvent and the thiol aliquot. We have determined that the aliquot mixing time is
substantially shorter than the monolayer formation time by measurement of the mixing
time using a colored dye solution. The time required for thorough mixing is related to
both the concentration and the volume of the injected aliquot; as either of these
quantities decreases, the mixing time increases correspondingly. The onset of the
maximum initial rate is caused by the concentration gradient introduced by mixing. We
account for this mixing effect by modifying the variable “t” in Eq. 2.3 to “t-to” where to

is the time required to achieve a homogeneous solution.

6(t — to) = K '[1 — exp(—kobs(t — t0))] [2.4]

Fitting the portion of the curve from the onset of the maximum rate through a significant
portion of equilibrium yields to, the predicted time of injection if mixing were
instantaneous, or equivalently, the time required for the thiol reactant to achieve a

homogeneous distribution in the reaction vessel.

24
The raw data were fit to Eq. 2.4, and the agreement between fit and data is

shown for a representative scan in Figure 2.3. The results of these fits are presented in
Table 2.1. For the purpose of fitting the data, kob, is independent of K’ because the
absolute frequency changes vary from one run to the next due to the sensitive impedance
matching conditions required by the electronics and the variation of the gold surface
morphology from one QCM to another. Because 9 is unitless, the absolute mass change
is not important to the determination of kinetic quantities k, and kd. The QCM
oscillation frequency in solution prior to thiol introduction represents a baseline and
plotting the (negative) QCM frequency change vs. time subsequent to thiol introduction
yields data of the for z0(t) vs. t where z is an arbitrary scaling factor.

Fitting the experimental data to Eq. 2.4 gives kobs. Since k0,, = kaC+kd, a plot of
kobs vs. C for a series of thiol concentrations gives a line with a slope of k, and an
intercept of kd, as shown in Figure 2.4. The values of k, and kd obtained from these plots
are presented in Table 2.2. The uncertainty in the values of k, and kd we obtain is
significant, but the data show unambiguously that desorption plays an important role in
the formation of the alkanethiol monolayers, i. e. kd > 0. This is an important finding in
and of itself, because the existence of a non-zero kd for alkanethiol-gold self assembled
monolayers demonstrates that these are equilibrium systems, and therefore at any given
instant in time, some fraction of the adsorbate sites on the surface are unoccupied.

The data we present in Table 2.2 allow the determination of the equilibrium

constant, Keg, for the monolayer system

k0 [2.5]

Keq=rla

60

50

4O

3O

-Af (Hz)

10

-10

 

25

 

 

 

,
L L I I l I J I l I
O 10 20 30 4O 50
time (s)

Figure 2.3. Fit of Langmuir adsorption isotherm (Eq. 2.4) to experimental data.

26

 

 

 

 

 

 

1.0 ~
. l
A .1—
29.
~20 0.5 ~
I
0.0 . L ‘ ' ‘ 4
0.0 1.0x10-4 2.0x10-4 3.0x10-4

l1'C18H37SHl (M)

Figure 2.4. Concentration dependence of kob, for l-octadecanethiol in n—hexane. The

best fit line through the data has a slope of k, and an intercept of It... See Table 2.2 for
results of the fits to the data.

27
Table 2.1. kob, values determined from raw data as a function of thiol concentration.

 

 

 

 

adsorbate/solvent thiol kob, i 95%
concentration (M) C.I. (5'2)

1-octadecanethiol/cyclohexane 1x10’5 0.06 i 0.02
3x105 0.23 i: 0.07

1x104 0.51 i016

3x104 0.58 i 0.43

l-octadecanethiol/n-hexane 1x106 0.06 i 0.04
3x106 014 i 0.04

1x10'5 0.24 i 0.06

3x105 0.23 i 0.13

1x10" 0441-016

311104 0.75 i 0.41

l-octanethiol/n-hexane 1x10'5 0.17 i 0.03
3x105 044 i 0.06

1x104 0.52:0.11

3x104 0.70 a: 0.23

1x103 1.20 i 0.34

3,110:3 0.90 i 0.34

 

 

28

Table 2.2. k, and kd determined from concentration dependence of k0,”.

 

 

adsorbate/solvent k, (M‘ls'l) i 95% CI. k; (5") i 95% CI.
l-octadecanethiol/cyclohexane 2059 i 1394 0.19 i 0.09
l-octadecanethiol/n-hexane 2440 i 1074 0.16 i 0.03

l-octanethiol/n-hexane 811 i 334 0.42 i 0.06

 

29
We can obtain two important physical quantities from K..,. The steady state fractional

coverage of the gold surface, 0(00), is given by

[2.6]

9‘77.

We show in Table 2.3 the fractional coverage values calculated for the 1-
octadecanethiol-gold system using the experimental concentrations and equilibrium
constant. The 9(00) values approach unity only at comparatively high thiol
concentrations. The free energy of adsorption of the monolayer is also found directly
fi'om the equilibrium constant data.

Mad, = -RT -ln Keg [2.7]

We show in Table 2.4 the free energies and equilibrium constants we have determined.
Our data provide direct proof that the alkanethiol-gold system is an equilibrium system.
Recent atomic microscopy studies have shown that regions of the monolayers exhibit
structural mobilitym] and this finding is explained by the fact that the monolayer system
is an equilibrium system.

Limits of applicability for Langmuir behavior. We observe a negative deviation
from linearity for the relationship between kobs and C for both high and low thiol
concentrations. We discuss below how the observed deviations do not indicate that
these systems are ill-described by the Langmuir adsorption model, but, rather,
demonstrate the limits intrinsic to this treatment of a complex equilibrium process.

The first deviation we consider is for low thiol concentrations. At low thiol
concentration the perturbation introduced to the system by the thiol is small. In other

words, for low thiol concentrations, the system remains in close proximity to equilibrium

30

Table 2.3. Fractional surface coverage as a function of thiol concentration, calculated
using Eq. 2.6, for the 1-octadecanethiol/n-hexane system. The uncertainty is determined
by uncertainty in k, and k4.

 

 

thiol concentration (M) 9(oo)
3x10‘ 0.04 i 0.02
111105 0.13 i- 0.07
3x10‘5 0.31 i 0.16
11110“4 0.60 i 0.34
3x10‘1 0.82 i 0.51

Table 2.4. Keg and AG“, determined from the experimental data.

 

 

adsorbate/solvent Keq (M 1) AG.d. (koal/mol)
l-octadecanethiol/n-hexane 15250 i 7300 -5.6 i 0.2
1-octadecanethiol/cyclohexane 10850 i 8950 -5.5 i0.4

1-octanethiol/n-hexane 1930 i 840 -4.4 i 0.2

 

31

and the expected mass change for such conditions lies beneath our limit of detection (0
<< 1). At equilibrium, the net rate of reaction is zero, although there remain non-zero
adsorption and desorption rates for the thiol.

d6
gt—=0=/.:,,(1—e)C—che

[2.8]
ka(1—0)C = kdfl

For small C, 9 approaches zero, and the equilibrium rate of monolayer formation

becomes very close to the initial rate of monolayer formation (ka C). The perturbation

fi'om equilibrium introduced into the system by a low concentration of thiol is negligibly
small, and the difference between the initial rate and the equilibrium monolayer formation
rate tends to zero before C=0. We define the quantity A to describe the displacement
from equilibrium for low thiol concentration. The quantity A is the difference between

initial rate of adsorption and the equilibrium rate of adsorption.

(kaC)2

(kaC + kd) [2.9]

A =kaC —k,(1—6)C =

A plot of A vs. C yields a line with a slope >> 1, indicating that A becomes negligibly
small as C—>0 (Figure 2.5). Our data are explained by this prediction, and we expect a
negative deviation from Langmuir behavior at low thiol concentrations.

For high thiol concentrations we also observe a negative deviation from
Langmuir isotherm predictions. At high thiol concentration the initial rate of adsorption
approaches zeroth order behavior, i. e. the initial rate of adsorption becomes
concentration independent for high thiol concentrations (Figure 2.6). This change in

behavior is indicative of the mechanism of monolayer formation. It is important to note

32

10°

10'1 9

10-2 0

10‘3

I ITFU

lO’4

 

105

10‘

'_+MI W I rrnflt
D

- O

I

-

—

lllll] 1 I lllllll j 111]].

10-7 1 l
10'6 10‘5 104 10‘3

1—1
‘?
‘3

concentration, C

Figure 2.5. Concentration dependence of A for k, = 2000 M'ls'l and kd = 0.2 s”.

33

 

 

1.0 b

initial rate (8'2)

 

0.4 '-

 

 

 

00 I r I r I r I r I r I
-5.0 -4.5 -4.0 -3.5 -3.0 -2.5

logloC

Figure 2.6. Concentration dependence for initial rate of monolayer formation for 1-
octanethiol adsorbed from n-hexane. Note the high concentration region where the rate
becomes concentration independent.

34

that the initial rates in the zeroth order regime are not the same for each system studied
(see Table 2.5), suggesting that the onset of pseudo zeroth order behavior is not a
stoichiometric effect. If stoichiometry were limiting, then the initial rates of adsorption
would be independent of thiol and/or solvent identity. If an intermediate state plays a
role in the adsorption reaction, then we would expect a solvent and alkanethiol
dependence, as is seen. The solvent-dependence of the initial rates in the zeroth order
regime implies that the rate of monolayer formation is limited by the time required to
form some intermediate state. Our data do not shed light on the structure of the
intermediate state, but does suggest its presence.

Comparison to electrochemical data. An important question relating to these
monolayer assemblies is the actual identity of the adsorbing and desorbing species.
Specifically, is the labile hydrogen from the thiol eliminated as a proton or as neutral
hydrogen? Recent electrochemical work by the Porter group indicates that neutral
hydrogen is the product of this reaction, or H202 if oxygen participates in the
reaction.[34’38] Using data from their work, we can estimate AG“, for the adsorption of
alkanethiols onto gold for both possible chemical reactions. Porter et. al. measured the
potential corresponding to the desorption of alkanethiols from gold.

RS—Au+le’ —>RS’+Au [2.ii]
The potential reported was -1.25 V vs. saturated Ag/AgCl, or -1.05 V vs. SHE for 10‘2
M thiol in 0.5 M ethanolic KOH, and the variation of this potential with aliphatic chain

length is small.” For adsorption to occur, the thiol must dissociate.

RSH eRs- +H+ [2.111]

35

Table 2.5. Limiting initial rates of monolayer formation for high thiol concentration.

 

 

adsorbate/solvent initial rate (8'2) 1r 95% CI.
l-octadecanethiol/n-hexane 31 i 12
l-octadecanethiol/cyclohexane 12 i 4

l-octanethiol/n-hexane 19 i 9

 

36
There is little information in the literature on the pKa of alkanethiols. Based on the

limited information available, pKa values in the range of 9 to 12 are expected, depending

[39] For pKa values of 9 and 12, we calculate the potential for thiol

on the solvent.
dissociation to be -0.53 V and -0.71 V, respectively, vs. SHE. The third reaction we

must consider is the reduction of protons to hydrogen.

2H+(solv.)+2e‘ ©H2(g) [2.iv]
We can estimate the half cell potential for this reaction by taking into account the proton
concentration contributed from thiol dissociation and the partial pressure of atmospheric
hydrogen in the air-saturated solution. For pKa values of 9 and 12 we calculate half cell
potentials of -0.23V and -0.31V, respectively, vs. SHE. From the above estimates we
approximate the free energies for alkanethiol-gold adsorption for conditions of either the
generation of a proton or of hydrogen. For the generation of a proton,

RSH(solv.) + Au(s) <=> RS — Au(s) + H+(solv.) + 1c. [2.v]

we calculate a potential of +0.43V vs. SHE, corresponding to a AG“, = -9.9 kcal/mol.

[3438] the formation of the gold thiol

If instead, as indicated by the Porter group’s data,
monolayer produces neutral hydrogen (scheme 2.i), we calculate a potential of +0.16V
vs. SHE, giving AG“, = -3.7 kcal/mol. In addition to the ambiguity of the locations of
the excess electron and proton generated in scheme 2.v, the calculated AG“. for scheme
2.i is closer to the experimentally determined values we report here. We caution that the
extent to which a comparison can be made between our data and the above estimate is

limited because the experimental conditions and sources for the electrochemical and

equilibrium data vary for parameters used in the approximation. Nonetheless, the

37

agreement between our data and the estimate is quite good, and provides further
support, albeit only qualitative, for the monolayer formation proceeding according to
scheme i and not scheme 2.ii. We return to a broader discussion of this point in Chapter
3.

Infrared spectra: External reflection infrared spectroscopy has been used
extensively to study macrosc0pic ordering in alkanethiol self-assembled monolayers.
The spectral positions of the methylene stretching modes for the alkanethiol monolayers
can be compared to those for n-alkanes which differ between liquid and crystalline
environments. FTIR studies of self assembled monolayers have shown that the alkyl
chains exist in densely packed environments, where a high degree of crystallinity was
assumed to be necessary to achieve the packing density. The band position(s) of these
vibrational modes have been reported to evolve in time subsequent to monolayer
formation, and this evolution time has been shown to be on the order of hours to daysllsl
The fact that we see chemical equilibration within one minute suggests that the
equilibrium achieved amongst the aliphatic chains can be largely separated from the

'40] In order to ensure the correspondence

formation of the alkanethiol monolayer.
between our self-assembled monolayer systems and those studied earlier, we have
examined the spectral evolution of the CH stretching region of the IR spectra of our
films.

We show in Figures 2.7 the spectra for l-octadecanethiol monolayers on the
QCM surface. The IR response of the l-octanethiol monolayers was below our detection

limit. The 1-octadecanethiol monolayers were formed by adsorption from a 31110“4 M

solution of n-hexane. Formation of the monolayer measured with the FTIR was

38

2920

2851

 

 

2840 2860 2880 2900 2920

2915

absorbance (arbitrary values)

2848 / \

 

M / ‘1.

2840 2860 2880 2900 2920

energy (cm")

Figure 2.7. FTIR spectra of a 1-octadecanethiol/gold monolayer formed on the QCM
surface, taken with 2 cm'1 resolution. (a) Spectrum recorded immediately after
adsorption. (b) Spectrum recorded two weeks after adsorption.

39
determined to be gravimetrically complete based on the time invariance of the QCM

response. The spectrum shown in Figure 2.7a was taken for a monolayer adsorbed
immediately prior to the IR measurement. The symmetric and asymmetric methylene
stretching modes occurred at 2851 and 2920 cm'l, respectively. The spectrum of the
same monolayer, stored in inert atmosphere for two weeks, is presented in Figure 2.7b.
In this spectrum the symmetric and asymmetric methylene stretching modes appear at
2848 and 2915 cm”, respectively. Because the spectral resolution used in these
measurements was 2 cm”, the red shifts we report are significant and indicate an
approach to a more densely packed, and possibly crystalline, environment for the
adsorbed alkanethiol chains. We note that the results for a monolayer that was aged for
24 hours were identical to those we report in Figure 2.7b.

Implications of the AG“, data. The magnitude of AG“, that we have measured for
alkanethiol/gold systems is important because it provides insight into the chemical
processes relevant to the formation of this interface. We consider that there are two

significant interactions that contribute to the formation of the observed monolayers.

The first interaction is that of the thiol moiety with the gold surface to form a
gold-thiol bond. The second interaction is the intermolecular attraction between
adjacent aliphatic chains that must also contribute to the stability of the system. We
consider the expected contributions from each of these interactions individually.

The first fundamental question to be addressed is the energy associated with bond
formation for gold/thiol systems. We are not aware of detailed thermodynamic
measurements on this chemical system, but note that Bockn's has reported a free energy

of formation for butanethiol on mercury of - 4.9 kcal/moll‘m The similarity of this value

40

to the AG“. we measure indicates that the dominant contribution to the formation of this
thiol-metal bond is determined by the creation of the thiol species. The strength of
interaction between aliphatic chains must also be considered. Obtaining an estimate of
this interaction energy is diflicult, but we approximate this interaction through the
viscous flow energy of the corresponding liquid n-alkanes. The comparison we make is
between a three dimensional liquid and a nominally two dimensional interface. Such a
comparison is directly relevant to the system under consideration because adsorption and
desorption at the interface must involve adsorbate molecular translation into and out of
the plane of the interface. The “viscosity” exhibited by a given liquid can be considered
as an activated process, and thus the temperature dependence will obey Arrhenius

behavior.
E .
17 = Aexp( ”'3 RT) [2.10]

The interaction energy, Em, is an energy of interaction between molecules and, while this
energy may be expected to be slightly different for a two dimensional interface than for a
three dimensional bulk liquid, we do expect this difference to be comparatively small, i. e.
within a factor of about two of the interaction energy within the monolayer.
Measurement of the viscous flow energy for a given solvent requires knowledge of the
temperature dependence of the solvent bulk viscosity, and the pre-exponential factor A is
strongly solvent-dependent. The viscous flow energy for a given liquid is typically 1/3 to
1/4 of the latent heat of vaporization for the liquidml Indeed, the qualitative reason for
the relationship of E... to AH. is not surprising, since the heat of vaporization is a

measure of the energy required to overcome the collective attractive forces within the

41
system, and those same attractive forces are responsible for viscous flow in the liquid.

For n-octane and n-octadecane we estimate the viscous flow energies to be 3 and 6
kcal/mol, respectively. This energy is essentially the same, within the uncertainty of the
estimate, to that of the measured AG.“

There are two possible explanations for the similarities of these two quantities.
The first is that we are, in fact, measuring the net energy of adsorption of the gold-thiol
system, which is dominated by interchain interactions amongst the aliphatic chains of the
alkanethiols. This explanation is not likely because, if true, it would imply that the
attractive inter-chain forces would lead to the spontaneous formation of alkanethiol
molecular assemblies in solution, and we do not observe this experimentally. The second
possibility is that the quantity we measure gravimetrically is largely unaffected by these
interchain interactions because the thiols that contribute to the observed equilibrium are
not involved in the formation of the “bulk” monolayer, and thus exhibit a AG.d,
characteristic of only the thiol-gold bond. Such a condition could occur at interfaces
between domains on the surfaces where an edge of a monolayer is at least partially
exposed to the solvent overlayer. In addition, the time scales on which the interchain
forces operate must necessarily be longer than that for the gold-thiol bond formation
because of the large number of possible conformers of the alkyl chains and the large
amplitude motions required for the chains to achieve a predominantly all-trans
conformation. We expect that it is not possible to separate the relative contributions to
the observed AG“. cleanly, and there is a contribution to the observed data from both

interchain interactions and the gold-thiol bond. Based on our estimates of the interchain

42

interaction energy, we believe that at least some diminution of this interaction is required
to measure a free energy of adsorption that is as small as we observe.

It is important to note that we measure the total free energy of the adsorption
process, and have not separated the entropic and enthalpic contributions. Previous
measurements of the heat of desorption for alkanethiol monolayers has provided a value
of ~28 kcal/mol.l43] While it might be tempting to use this value for AH“, and infer the
entropic contribution to the adsorption, this is not possible without knowing the heat of
solution of the alkanethiol in the solvents we use here. For our measurements we
determine the adsorption from solution, and for the AH,“ determination, desorption is
from the surface into vacuum. Thus the heat of solution for the alkanethiol is required in
order to determine the entropic contribution to the formation of these monolayers. We
will present in Chapter 3 a family of measurements that yield AS“, and AS“, information

directly.

43
2.4. Conclusions

We have presented kinetic measurements of the formation of alkanethiol
monolayers on microcrystalline gold. The adsorption behavior of these systems follows
the predictions of the Langmuir isotherm over a limited concentration range, and the
deviations observed fiom the Langmuir prediction are expected based on limits inherent
to the assumptions of the Langmuir isotherm. Our data demonstrate that the alkanethiol-
gold system is in equilibrium with free gold sites and alkanethiol moieties. From our data
we derive AG“. = -5.5 d: 0.4 kcal/mol for n-C18H37SH adsorption onto gold, and AGMs =
-4.4 i 0.2 kcal/mol for n-CanSH adsorption onto gold. These values are virtually the
same as our estimates for the aliphatic interchain interaction energies, indicating that the
equilibrium behavior we observe proceeds predominantly at the edges of monolayer
islands and/or at step edges and defect sites. The infrared data demonstrate that the
mass equilibrium for alkanethiol monolayers is achieved within about one minute, but the

structural equilibration of the aliphatic chains proceeds over a much longer time scale.

2.5. Literature Cited

1. D. L. Allara; R. G. Nuzzo; Langmuir, 1985, 1, 45.

2. D. L. Allara; R. G. Nuzzo; Langmuir, 1985, 1, 52.

3. A. H. M. Sondag; M. C. Raas; J. Chem. Phys, 1989, 91, 4926.

4. L. K. Chan; M. D. Porter; Chem. Phys. Lett., 1990, 167, 198.

5. H. Lee; L. J. Kepley, H. G. Hong; T. E. Mallouk; J. Am. Chem. Soc., 1988, 110, 618.

6. H. Lee; L. J. Kepley, H. G. Hong; S. Akhter; T. E. Mallouk; J. Phys. Chem, 1988,
92, 2597.

44
7. G. Cao; T. E. Mallouk; J. Solid State Chem, 1991, 94, 59.

8. S. Akhter, H. Lee, T. E. Mallouk; J. M. White; H. G. Hong; J. Vac. Sci. Tech. A,
1989, 7, 1608.

9. H. G. Hong; T. E. Mallouk; Langmuir, 1991, 7, 2362.

10. G. Cao; L. K. Rabenberg; C. M. Nunn; T. E. Mallouk; Chem. Mater., 1991, 3, 149.

11. L. J. Kepley, D. D. Sackett; C. M. Bell; T. E. Mallouk; Thin Solid Films, 1992, 208,
132.

12. H. G. Hong; D. D. Sackett; T. E. Mallouk; Chem. Mater., 1991, 3, 521.

13. L. H. Dubois; B. R. Zegarski; R. G. Nuzzo; J. Am. Chem. Soc., 1990, 112, 570.

14. R. G. Nuzzo; L. H. Dubois; D. L. Allara; J. Am. Chem. Soc., 1990, 112, 558.

15. R. G. Nuzzo; E. M. Korenic; L. H. Dubois; J. Chem. Phys, 1990, 93, 767.

16. C. E. D. Chidsey; D. N. Loiacono; Langmuir, 1990, 6, 682.

17. C. E. D. Chidsey; C. R. Bertozzi; T. M. Putvinski; A. M. Mujsce; J. Am. Chem.
Soc., 1990, 112, 4301.

18. C. D. Bain; E. B. Troughton; Y. T. Tao; J. Evall; G. M. Whitesides; R. G. Nuzzo; J.
Am. Chem. Soc., 1989, 111, 321.

19. C. D. Bain; J. Evall; G. M. Whitesides;.l. Am. Chem. Soc., 1989, 111, 7155.

20. E. B. Troughton; C. D. Bain; G. M. Whitesides; R. G. Nuzzo; D. L. Allara; M. D.
Porter; Langmuir, 1988, 4, 365.

21. M. D. Porter; T. B. Bright; D. L. Allara; C. E. D. Chidsey; J. Am. Chem. Soc.,
1987, 109, 3559.

22. G. M. Whitesides; P. E. Laibinis; Langmuir, 1990, 6, 87.

45

(I. 23. N. Camillone IH; C. E. D. Chidsey; G.-Y. Liu; T. M. Putvinski; G. C. Scoles; J.
Wang; J. Chem. Phys, 1991, 94, 8493.

24. C. E. D. Chidsey; G.-Y. Liu; G. C. Scoles; J. Wang; Langmuir, 1990, 6, 1804.

25. S. R. Wasserrnan; G. M. Whitesides; I. M. Tidswell; B. M. Ocko; P. S. Pershan; J.
D. Axe; J. Am. Chem. Soc., 1989, 111, 5852.

26. C. A. Widrig; C. A. Alves; M. D. Porter; .1. Am. Chem. Soc., 1991, 113, 2805.
\/ 27. G.-Y. Liu; M. B. Salmeron; Langmuir, 1994, 10, 367.
28. R. L. McCarley; Y.-T. Kim; A. J. Bard; J. Phys. Chem, 1993, 97, 211.

I If}
x 29. R. L. McCarley; D. J. Dunaway; R. J. Willicut; Langmuir, 1993, 9, 2775. I
30. G. G. Guilbault; Methods and Phenomena 7: Applications of Piezoelectric Quartz
Crystal Microbalances, Elsevier Science Publishing Co., New York, 1984. p. 252.

31. G. Z. Sauerbrey, Z. Phys, 1959, 155, 206.
, 32. L. Strong; G. M. Whitesides; Langmuir, 1988, 4, 546.

v’33. C. E. D. Chidsey; G.-Y. Liu; P. Rowntree; G. Scoles; J. Chem. Phys, 1989, 91,
4421.

34. C. A. Widrig, C. Chung; M. D. Porter; .1. Electroanal. Chem, 1991, 310, 335.
35. M. Yang; M. Thompson; Langmuir, 1993, 9, 802.
36. M. Yang, M. Thompson; Langmuir, 1993, 9, 1990.

37. J. B. Schlenoff, M. Li, H. Ly, J. Am. Chem. Soc., 1995, 117, 12528.

38. D. E. Weishaar; B. D. Lamp; M. D. Porter;J. Am. Chem. Soc., 1992, 114, 5860.

46

39. Shigeru an; Organic Sulflir Chemistry: Structure and Mechanism, CRC Press,
1991.

40. G. Hahner; Ch. Woll; M. Buck; M. Gnmze; Langmuir, 1993, 9, 1955.
41. E. Blomgren; J. O’M. Bockris; C. Jesch; J. Phys. Chem, 1961, 65, 2000.

42. W. J. Moore, Physical Chemistry, 3rd Edition, Prentice-Hall, Englewood Cliffs, NJ,
1963,p.724.

43. R. G. Nuzzo; B. R. Zegarski; L. H. Dubois; J. Am. Chem. Soc., 1987, 109, 733.

Chapter 3

Quantitating the Balance Between Enthalpic and Entropic Forces in
Alkanethiol/Gold Monolayer Self-Assembly

Summary

We have studied the temperature and adsorbate concentration dependence of
alkanethiol/gold monolayer formation in-situ and in real time using a quartz crystal
' microbalance to monitor the rate of reaction. The temperature dependence of the
experimental rate constants, for a given concentration, demonstrates that the formation
of these monolayers is not diffusion rate limited for the conditions we report. Our data
are modeled accurately by the Langmuir adsorption isotherm and, using this model, we
have determined the adsorption and desorption rate constants for monolayer self
assembly. We have extracted from these data the thermodynamic quantities AG,“ AH,d,
and AS“, for monolayer formation. For l-octadecanethiol adsorption from n-hexane
onto gold, AG“, is temperature dependent and is ca. -5.5 kcal/mol, AH“, = -20 i 1
kcal/mol and AS“, = -48 i 1 cal/mol-K. These data have direct bearing on the broader
field of interface chemistry because they underscore the dominant role that entropy plays
in determining the course of chemical reactions at interfaces and the robustness of the
resulting molecular assembly. We discuss the implications of these data on the stability
of the monolayers and the limits that these values place on the extent of experimental

control available over their formation.

47

48
3.1. Introduction

The self-assembly of alkanethiol monolayers onto gold has been studied
extensivelyn'm due to the potential applications of these interfaces to important
technologies such as chemical sensing, micro-device patterning and surface
lubrication.l48"3l The extensive literature that exists on these systems is aimed largely at
understanding the steady state properties of the monolayers once they have formed, and
from these studies has emerged a reasonably consistent picture. For long chain n-
alkanethiols (2 CszlSH),[91 the aliphatic chains are densely packed with a 5A inter-

d [54.55]

chain spacing on Au(lll), once formed and anneale These chains are in a

predominantly all-trans conformation, and the thiol head groups exist in a (\/3x\/3)R30°

]

configuration on the hexagonal close-packed Au( 111) surface?”9 Recently, it has

7, 8,516.57 -
[I l l wrth

become increasingly clear that these monolayers are highly dynamic,
desorption and structural rearrangement playing an important role in determining the
evolution of the macroscopic properties of the interfaces. Despite this significant level of
understanding, there remain several substantial open questions in this field relating to the
mechanism of monolayer formation and the extent to which surface defects mediate the
dynamics of the monolayers, once they have formed.

The prospect of using self-assembled monolayers for patterning and chemical
sensing applications requires that monolayers possessing functionalities more complex
than available with simple aliphatic thiols be synthesized.[4’5’27’43’50’58601 The details of the
interaction between the thiol head group and the gold surface remain, to some extent,

uncertain and in addition, it has been demonstrated that the “tail group” can alter the

organization of the formed monolayer if the organic tail contains chemically active or

49

bulky termini. Thus there is a competition between different regions of the monolayer to
determine the macroscopic organization of the system. In order to understand the
balance that exists between the thiol head group interactions with the metal surface and
organic tail group interactions with adjacent adsorbates, there needs to be a better
understanding of the initial self-assembly process, and that is the focus of this work.

We have reported in Chapter 2 on a method to monitor, in-situ, the adsorption of
alkanethiols onto gold using a quartz crystal microbalance (QCM) fabricated with
evaporated gold electrodes that are predominantly Au(lll).“8’6” We were able to
monitor the mass change of the QCM resulting from the adsorption of alkanethiols onto
the gold electrodes with sub-second time resolution, allowing access to direct
information on the formation of the monolayer. Earlier studies of alkanethiol/gold
monolayer formation, performed ex situ with techniques such as external reflection FT-
IR,“ optical ellipsometry,[6’8] radio isotope labeling“ and liquid contact angle

[6.8]

measurements, were limited to time resolution on the order of minutes to hours. Our

in situ data demonstrate a kinetic response fully consistent with that predicted by the

[64'6“ and, by modeling the experimental adsorption data

Langmuir adsorption isotherm
using this isotherm, we were able to detemrine the free energy of adsorption for the
monolayer. The values we obtained for AG“. at 293 K were -5.6 i 0.4 kcal/mol for 1-

octadecanethiol adsorption onto gold from n-hexane, -5.5 i 0.2 kcal/mol for formation

of 1-0ctadecanethiol/gold from cyclohexane, and -4.4 i 0.2 kcal/mol for formation of 1-

“” These modest free energies, on the order of the

octanethiol/gold from n-hexane.
hydrogen bond strength characteristic of liquid alcohols,“ demonstrated that these

systems are highly labile, a finding at odds with the then-prevailing thought that, once

50
formed, the monolayers exhibited few dynamics. Studies of the mobility of

alkanethiol/gold monolayers by the McCarley and Tarlov groups using scanning probe
microscopy have revealed the presence of mobile defect sites, providing a different
indication of the labile nature of these monolayers.“7’56] In that work, it was shown that
the rate at which these defects migrated across the surface was independent of
alkanethiol chain length, suggesting that the thiol head group plays the dominant role in
mediating the adsorption and desorption steps,“71 and that the first layer of gold atoms

[56' The relatively

were apparently involved in the monolayer rearrangement directly.
small free energies of adsorption we obtained in our earlier work indicated that
alkanethiols are not bound to the surface strongly, but it was not clear, based on a single
quantity (AG.d,) whether the labile nature of these films was a consequence of chemically
weak interactions between the adsorbate and the surface or the counter-balance of two
significant but opposing forces, specifically the enthalpy and entropy of adsorption. The
answer to this question was indicated by earlier work on the thermal desorption of a
monolayer of alkanethiols from gold. Nuzzo and coworkers desorbed a dimethyl
disulfide monolayer from gold thermally, recovering a heat of desorption (AHdcs) of 28
kcal/mol.[68] The implied heat of adsorption, AH“, = -28 kcal/mol was determined in the
absence of solvent and using a different adsorbate and, while leading, the utility of this
value for comparison with our experimental data on AG“. is limited due to the unknown
solvent-thiol contribution to our data. In order to determine the enthalpic and entropic
contributions to the small AG“. we measure for adsorption from solution, we have

measured these quantities directly through the temperature dependence of It, and kd, the

adsorption and desorption rate constants for monolayer assembly. Our data indicate that

51

the formation of these monolayers is spontaneous by ~-5.5 kcal/mol at room temperature
because of the offsetting effects of large enthalpic and entropic forces. In this chapter,
we focus on l-octadecanethiol/gold monolayer formation from n-hexane. We measure
the temperature dependence of the equilibrium constant K“, to determine the enthalpy of
adsorption, AH“, From AG.d.(T) and AH“. we determine the entropy of adsorption,
AS“, These thermodynamic quantities underscore the high degree of molecular
organization characteristic of self-assembled monolayers and serve to place limits on the
ultimate structural robustness and chemical “tunability” attainable with a thiol/gold
monolayer assembly approach.

3.2. Experimental

Kinetic measurements: We have reported previously on the microbalance

2‘18] and recap only

apparatus used to measure monolayer adsorption kinetics in chapter
the essential details here. The QCM was suspended in solution with isolated electrical
connections for both gold electrodes using a modified spring clip. AT-cut quartz crystal
microbalances (6 MHz resonance frequency) with vapor deposited gold electrodes were
obtained from McCoy Electronics (part number 78-18—4). The QCM was connected to a
~6 MHz oscillator tank circuit (Maxtek part number 124200-4) and a frequency counter
(Phillips model PM 6673) was used to monitor the QCM oscillation frequency. This
fi'equency counter has 1 Hz resolution with a gate time of 280 ms. The analog output of
the frequency counter was routed to a computer for acquisition, A/D conversion and

storage. For all adsorption rate measurements, the data were acquired at ten readings

per second to ensure adequate sampling of the QCM kinetic response. Electrical

52

connections between components were made using shielded coaxial cables to minimize
QCM frequency instability arising from electrical interference.

QCMs with a resonance frequency of ~ 6 MHz are sensitive to mass changes as
small as a few nanograms. Several factors, in addition to the mass loading of the device
surfaces, can affect its oscillation frequency. The resonance frequency of the QCM is
related to mass loading according to the Sauerbrey equationm This relationship holds
quantitatively for gas phase measurements, but in situ solution phase measurements
using the QCM are, in many cases, only semi-quantitative due to the complex dielectric
response of the solvent and the presence of the adlayer(s). The basis for complications
associated with QCM mass measurement in liquids is discussed in detail elsewhere,l7°’7”
and is not directly relevant to the focus of this work. Our primary objective here is to
determine the rate of mass change associated with the formation of the alkanethiol
monolayer. We do not attempt to extract absolute mass change information from our
data and thus many of the complications associated with the operation of a QCM in
solution for quantitative applications do not play a role in our measurements.

In addition to the several material factors that contribute to the experimental
response of the QCM, the resonant frequency of these devices depends sensitively on
temperature. To minimize QCM frequency drift associated with thermal fluctuations, the
temperature of the solution and QCM was controlled to within 1- 0.05 K for a given set
point. Measurements were performed in a 150 mL jacketed beaker connected to a
flowing liquid temperature controller (Neslab model RTE-110). The temperature was

varied from 288 K to 303 K in increments of 5 K. Total solution volume in the jacketed

beaker was 100.0 mL for each measurement. Stirring was accomplished with a magnetic

53

stirrer and a Teflon" coated stir bar. The stirring speed was set to be as fast as possible
without introducing fluctuations to the QCM baseline frequency.

The jacketed beakers and other glassware used for the measurements were
cleaned using the same methods reported in Chapter 2. The cleaned QCMs were then
rinsed with THF to remove any residual water, then with n-hexane and used immediately
thereafter. Individual monolayer adsorption measurements began with pure solvent in
the temperature controlled vessel to establish a stable QCM baseline oscillation
fiequency. An aliquot of stock thiol solution was then introduced by syringe. The stock
solutions were maintained at the same temperature as the solvent in the reaction vessel to
avoid thermal disruptions resulting from injection.

Conductivity measurements. The conductivity of suspensions of gold dust in
ethanol were measured before and after the addition of 1-octadecanethiol using a
conductivity meter and cell (Cole—Partner model 19100-00).

Chemicals. l-Octadecanethiol was purchased from Aldrich Chemical Co. and
used as received. The thiol was tested by gas chromatography and mass spectrometry to
determine the presence of the corresponding dioctadecyldisulfide, and none was
detected. The solvent n-hexane was purchased from Baxter (Burdick and Jackson) and
used without further purification. For conductivity measurements, 1 - 3 pm gold dust
was purchased fiom Aldrich Chemical Co. and used as received. For the conductivity

measurements, anhydrous ethanol solvent was used.
3.3. Results and Discussion

The adsorption and desorption reactions. Before presenting our experimental

data on the thermodynamics of monolayer formation, it is important to have a clear

54
understanding of the chemical reaction to which these data apply. While the initial

reactants are known and the product monolayer is comparatively well characterized, the
structural identities of monolayer constituents that desorb from the gold surface remains
open to speculation and is awaiting direct experimental verification of their identity.
Specifically, there are two points of uncertainty that exist in the literature. These are the
fate of the thiol hydrogen and its role in desorption, and the competition between thiol
desorption and disulfide desorption. We consider below the current understanding of
these reactions, the possible desorption products and the likely reaction scheme for our
experimental conditions.

The initial adsorption reaction can, in principle, proceed either by ionic

dissociation of the thiol or by formation of H,

RSH + Au —> RSAu' + 11+ [3 .1
.1

RSH +Au—’ RSAu+H. ..
[3.11]

[18,19,24]

and based on available electrochemical evidence as presented in Chapter 2 as well

[721 reaction 3.ii is favored significantly over reaction 3.i. This

as conductometric data,
portion of the monolayer formation reaction is comparatively well understood, save for

the fate of H- at the gold surface. The role of H- is different in the two most likely

desorption reactions,

k1
ZRSAU + H2
k.1 (8) [3.111]

 

2R8H(so]v) + ZAU(s)

 

55
k2

 

2RSAU(S) RSSR(so[v,) + 2Au

k.2 [3 .iv]

 

We consider first the enthalpic contributions to the reactions 3.iii and 3.iv to estimate the
course of the desorption, whether it is predominantly via k, or k2. Given the starting
materials, RSHWM) and Au“), the forward reaction, with rate constant k1, is the starting
point. In order to estimate the heat of reaction, we need first to understand the fate of

[73] and such an

H-. There is literature precedent for absorption of H- into thin gold films,
absorption would be in competition with formation of H2 from 2 H-. For the reaction
2H- —-> H2, AH = -104 kcal/mol,ml and for H- + Au“, —+ H-Au(,), AH E -2 kcal/mol.[73]
The formation of molecular hydrogen will be favored over dissolution of H- into the gold
matrix, although this is not necessarily the case for all metals, such as Pd or Pt. We must
also consider the possibility that H; is stored as H202 if 02 is available to the reaction at
the time of monolayer formation. The enthalpy of formation of H202 from H; and 02 is
~ - 45 kcal/mol, and the dissociation reaction of H202 to form H + HOz- yields -40
kcal/mol.[741 Thus the presence of 02 at the surface has some effect on the availability of
H- to the desorption but the energies are modest and the desorption to the thiol will still
proceed because of the limited amount of 02 available at the gold reaction site.

For the formation of the monolayer, where the bond dissociation energy for the
RS-H bond is taken to be ~ 87 kcal/mol,[2°' the RS-Au bond dissociation energy is ~ 40
kcal/molm and 2 H- ——> H; dominates over gold matrix absorption of H, the calculated
heat of this reaction is -5 kcal/mol. The heat of the reverse reaction, denoted by the rate

constant k.1 in scheme 3.iii, is 5 kcal/mol, if all of the reactants are available. We

compare the desorptive reaction yielding starting materials to the formation of the

56
disulfide, according to the reaction denoted by rate constant k; in scheme 3.iv. The

reaction to form RSSR involves the breakage of two RS-Au bonds (40 kcal/mol each)
and the formation of a RS-SR disulfide bond, (~74 kcal/mol for disulfide bond
cleavageml), yielding a net heat for the k; desorptive process of +6 kcal/mol. Note that
this latter reaction does not require hydrogen to proceed, where the reaction to form
thiol from the adsorbed monolayer does require a source of H2. Based on enthalpic
arguments alone, and given the presence of H; from the initial monolayer formation, we
would predict that both desorption reactions will contribute to the monolayer dynamics.
The enthalpic contributions to the formation and dynamics of these monolayers
do not provide a complete picture, however. In addition to enthalpic contributions to
the reaction, there are also entropic considerations that are non-negligible (vide infra).
As we detail below, our data indicate a large negative entropy associated with the
formation of the monolayer. In other words, there is a significant increase in the order of
the system in going from a clean metal surface and solvated thiols to an organized two-
dimensional array of gold-thiol adsorbed species. There is a significant entropic driving
force for the desorption of the monolayer, and the expected entropic contributions for
the two processes presented in schemes 3.iii and 3.iv need to be compared. We measure
experimentally that ASd... > 0, and argue that, on stoichiometric grounds, ASdesRSH ~
2ASduRSSR. Given that the enthalpic contributions to each desorption reaction are
essentially identical, to within the uncertainty of the values used in these estimates, the

difference in the free energy of desorption for the two processes will be determined by

the entropic term,

57
1165,?” — AGfiSR = —TASde,SR [3.1]
Eq. 3.1 indicates that the energetic barrier for desorption of the monolayer back to thiol

8
RS R. For our

is smaller than that for desorption of a disulfide by an amount ~TASde,
experimental conditions, which we describe below, we estimate that desorption back to
the thiol is favored over disulfide desorption by ~ 6 - 12 kcal/mol near 300 K.

As noted above, the dominance of desorption to yield the thiol requires the
availability of H2 and/or H', and if these species are not available, then desorption to
form disulfides in solution would dominate the reaction pathway. Given the presence of

H2 in the system by the formation of the monolayer, the solubility of H' in gold, and the

presence of air, we believe that the dominant reaction for our experiments is

ka

 

RSI'IQON) + AU(s) RSAU“) + 1/2‘['12(solv.)

kd [3.v]
where k, and kd are the adsorption and desorption rate constants for monolayer
formation. There are several conditions for which this could not be the dominant

[68] or thioetherslzol are used in forming the

desorption pathway, such as when disulfides
monolayer. Under these conditions, where H; is not available as a product of the initial
adsorption reaction, we expect that desorption of disulfides will be the primary reaction
pathway. With the probable chemical reaction pathway determined for our
measurements, we consider next the form and interpretation of the QCM kinetic data.
Experimental data and reaction kinetics. The monolayer formation experiments
yield information about the assembly of alkanethiol/gold monolayers based on in-situ,

real-time data (Chapter 2, Figure 2.1). Noticeable is the short time it takes to achieve a

steady state condition (seconds). Early studies reported that monolayers required

58

formation times of hours to days.l6'8’62] However, these ex-situ studies were done using
techniques that measured properties resulting from monolayer formation such as
ellipsometric thickness, IR peak position or solvent contact angle.‘6’8’62] It has since been
found that the monolayer properties mentioned above are the result of ordering brought
about by the structural annealing of the monolayer aliphatic chains. More recent studies
have shown that there are two distinct processes associated with monolayer formation.
The first process is adsorption of thiol head group to the gold surface and the second is
structural annealing of the adsorbate aliphatic chains. There is some very recent and
compelling evidence for cooperative interactions between head and tail groups in the
formation of monolayers by UHV vapor phase adsorptionml Adsorption from solution,
however, will be mediated by the solvent interactions with the thiol tail group and the
cooperative formation of highly organized islands of alkanethiols on gold for low
fractional coverages from solution remains to be demonstrated. We consider for solution
phase processes that the adsorption and annealing processes can be treated separately to
good approximation based on the significant body of experimental data indicating their
widely differing timescales.[6’8‘18‘62l An important factor in the acquisition of the raw data
is the temporal resolution of the measurement. We designed our experiment so that we
would have sufficient data for determination of the kinetic rate constants. The frequency
counter we used had a gate time of 280 ms (~ 4 readings/second), and the computer data
collection program acquired 10 readings/second. Oversampling in this manner ensures
that the frequency counter time resolution is limiting for these experiments.

We next consider the interpretation of our raw kinetic QCM responses. We have

provided a detailed explanation of the data processing and modeling in Chapter 2,” and

59

include only the essential aspects of this treatment here. The adsorption experiments
were performed in solutions with concentrations ranging fi'om 3x10‘ M to 3x10‘4 M 1-
octadecanethiol in n-hexane. The adsorption kinetics exhibit a concentration dependence
with the rates increasing in proportion to thiol solution concentration. Within the
concentration range used here, we model our Af vs. time data using the Langmuir
[64-66]

adsorption isotherm,

3‘1 = k,(1-— e)C—k,o [3.2]

where 0 is the fi'action of available sites, and the quantities k, and k, are the adsorption
and desorption rate constants, respectively. As noted previously, there are several
assumptions implicit in the use of the Langmuir adsorption isotherm for describing the
formation and desorption of these monolayers. Based on our previous measurement of
AG“, ~ -5.5 kcal/mol for these monolayersfm the assumption of non—interacting
adjacent adsorption sites is not violated seriously. In addition, the identity of the
monolayer constituents precludes multilayer formation, at least in the concentration
range used here. The Langmuir adsorption isotherm provides a framework for data
interpretation that is consistent with our knowledge of this system. Integration of Eq.
3.2 yields the time dependence of monolayer formation, indicating exponential growth of

the layer in time,

6(t) .—. ——i——-[1 - exp(— k,C+k,)z] [3.3]

(3+4

We measure, for a given thiol concentration and temperature, a build-up of the

monolayer in time that is fit best by a single rate constant, kob, = k,C + kd. We also

60
substitute k’=C/(C+[kd/k.]) into Eq. 3.3 because we are not attempting to quantitate the

mass adsorbed. The term k’ corresponds to the steady state fractional coverage of the
monolayer, 6(00). To account for finite introduction time of the thiol into the reaction
vessel, we incorporate a temporal offset (to) in our fitting equation. The simplified form

of Eq. 3.3 used to fit the data is

6(1— to) = k'[1— exp(— k0,, (t — to))] [3.4]
See Chapter 2, Figure 2.3 for an example of the fit of Eq. 3.4 to the experimental data.
Note that, as discussed above, we are not attempting to extract adsorbed mass
information because of the complications associated with solution phase measurements.
We have determined the concentration and temperature dependence of kobs (Table 3.1).
For a given temperature, the dependence of kobs on thiol concentration (Figure 3.1)
provides information on the adsorption and desorption rate constants k, and kd (Table
3.2). These data can be used to calculate the equilibrium constant K3,, and thus the Gibbs
free energy of adsorption, AG.d., for the adsorption reaction 3.v,

k
K“, = a
k, [3.5]

A6,, = —RTln K“,

 

By performing our experiments at specific temperatures in the range of 288 K to 303 K,
we have measured the temperature dependence of Keg and AG“, which we present in
Table 3.3 and Figures 3.2. The data exhibit a substantial decrease in Keq with increasing

temperature; AG“. becomes less negative with increasing temperature. Since a negative

61

1.4 -
1.2 —

1.0 -

0.3 - i/

kobs

 

0.4 - __

 

 

 

 

 

 

thiol concentration (11M)

Figure 3.1. Concentration dependence of km at 293 K. From these data we obtain best
fit values ofka = 2278 i 92 M1 s'1 and k. = 0.08 i 0.02 s;1

62
Table 3.1: kob, values detemrined fi'om raw data as a function of thiol concentration.

 

 

 

 

 

Temperature thiol k0... i 95% CI.
(K) concecirgation (8")
288 1 x 10" 0.17 i 0.09

3 x 10'5 0.28 3: 0.10
1 x 10“ 0.96 d: 0.84
2 x 10“ 1.44 i 0.86
293 1 x 10'5 0.11 i 0.04
3x10" 0.15i0.11
2 x 10" 0.51: 0.35
3 x 10“ 0.78 i 0.46
298 1 x 10" 0.16 i 0.04
3 x 10" 0.25 i 0.14
1 x 104 0.33 1. 0.22
2 x 10“ 0.53 :t 0.42
3 x104 0.84i0.54
303 1 x 10" 0.21 :t 0.08
3 ><10'5 04010.13
1 x 10“ 0.43 i 0.20
2 x 10“ 0.71 i 0.33

 

 

 

63

 

 

 

288 291 294 297 300 303

8x104 '-

6x104 -

K,q

4x104 -

2x104 *- 1

l l I l I l I l J; l L

288 291 294 297 300 303
temperature (K)

 

 

Figure 3.2. Temperature dependence of (a) AG“, and (b) Kc... The regressed line
shown in (a) yields a slope of -AS,d, = 48 i1 cal/mol-K and an intercept of ALL... = -19.9
i 0.4 kcal/mol.

64

Table 3.2: Adsorption and desorption rate constants (k, and kd) from fits of Eq. 3.4 and
kob, = kaC + k; to the experimental data.

 

Temperature kaj: 95% CI. kdi 95% CI.

 

(K) 01‘s") 6")

288 6871 i 803 0.13 i 0.09
293 2278 i 92 0.08 i 0.02
298 2201 i 198 0.14 i 0.03
303 2278 i 537 0.24 i 0.06

 

 

 

Table 3.3: Temperature Dependence of the Equilibrium Constant (K2,) and the Gibbs
Free Energy of Adsorption (AG.d,).

 

Temperature Kali: 95% CI. AG.d.i 95% CI.

 

(K) (kcal/mol)

288 52854 i 37100 -6.22 i 0.30
293 28475 i9134 -5.97 $0.14
298 15721i3653 -5.72 i010
303 9492 i 3262 -5.51 i 0.15

 

 

 

65

AG“. indicates a spontaneous condition, the trend to less spontaneous AG“. with
increasing temperature indicates that the adsorption process we measure is exothemric.
The temperature dependence of Keg allows the calculation of the enthalpy of adsorption,

AH...” using the van’t Hoff equation?“

— AH“, an,q

R - T. [3.61

 

 

We show in Figure 3.3 the linear dependence of In Keg on T'1 giving a slope of -AH.d,/R.
We obtain AH“. = -20 i l kcal/mol for the adsorption of l-octadecanethiol/gold
monolayers from n-hexane. From the AG,d,(T) and AH.d., we obtain the entropy of
adsorption,

AG“, =AHm—TASMs [3.7]
The dependence of AG“, on T is shown in Figure 3.2a. We find from a regression of
these data that AS“, = -48 i 1 cal/mol K and, as a comparison to the van’t Hoff
determination of AH.d,, we obtain AH“, = -19.9 i 0.4 kcal/mol. This alternate route to
AH“. serves as a usefiil self-consistency check on our results.

We note that, because the alkanethiol/gold system is an equilibrium system, the
fractional coverage of the Au surface depends on the concentration of the thiol solution,
and over the concentration range studied here the fraction of the surface, 6~ 0.1 (RSH =
1x10"_M) to 6~ 1 (RSH = 3x104_1\_rf).“8’ This concentration dependent surface coverage
is expressed in the prefactor of Eq. 3.3 and is an expected result based on the predictions
of the Langmuir adsorption isotherm. The phenomenon under examination in this work

is the initial formation of the gold-thiol bond, which is expected to proceed more rapidly

than rearrangement and annealing steps in the formation of the monolayer. The

 

66

11.0 r-

 

10.5

 

 

 

3
E 10.0
9.5
9.0
I r I r I r I r
3.30 3.35 3.40 3.45
T'1 (x10'3)

Figure 3.3. Van’t Hoff plot of In Kcq vs. T". From these data, we determined AH“, = -
20 2t 1 kcal/mol.

 

67

thermodynamic results on alkanethiol adsorption to gold are interesting in and of
themselves and are consistent with the other, limited thermodynamic data that exist for
these systems. We consider next the implications and chemical information content of
these results.

Before discussing the chemical implications of these data, we need to consider
whether or not the adsorption process is diffusion limited. Indeed, if the reaction under
consideration is diffusion limited, then a Langmuir treatment of our data is inappropriate.

We presume as a starting point that both adsorption and desorption occur, and this

[17] [56]

ansatz is supported by both McCarley’s and Poirier’s atomic microscopy data.
Recently, the Schlenoff group has suggested that the adsorption reaction is diffusion rate
limitedm] based on initial rate data that we had reported previouslylm The temperature
dependence of the observed rate constant for a fixed thiol concentration (Figure 3.4)
demonstrate clearly that the adsorption reaction is not diffusion rate limited. Our basis
for this statement is that a diffusion rate limited reaction should possess a rate constant

'6" and the data presented in Figure 3.5 are

that depends linearly on temperature,
independent of T, to within the uncertainty of the measurement. We note that the data
presented in Figure 3.4 are rate constants extracted directly from raw data and do not
depend on the use of the Langmuir isotherm. Regardless of the isotherm under
consideration, our data demonstrate that the adsorption of alkanethiols onto gold are not
diffusion limited for these conditions.

There are two processes intrinsic to the formation of self-assembled monolayers,

and they proceed on significantly different time scales. These processes are the initial

adsorption of the thiol head group to the gold surface, which occurs within seconds of

68

 

 

 

 

 

 

2.5 -
2.0 +—
A 1.5 [- T
3 .
1.0 - __ "'
’ e
0.5 -" 0 7
L
00 I 1 I 1 I 1 I ‘_1 I 1 J
288 291 294 297 300 303
temperature (K)

Figure 3.4. Temperature dependence of the observed rate constants, km. These data
are essentially temperature independent, to within the uncertainty of the measurements.
No linear dependence of km on T is detectable from these data.

69

exposure and, subsequent to head group adsorption, the aliphatic chains anneal from an

initially statistical distribution of conformers to a predominantly all-trans conformation
over a period of hours to days. Our QCM-based kinetic studies yield information about
thiol head group adsorption. Accordingly, we consider the processes of initial
adsorption and aliphatic annealing separately and assume that they are significantly
decoupled.

For l-octadecanethiol/gold monolayer formation from n-hexane, AG“, E -5.5
kcal/mol depending on the temperature. The magnitude of AG,d, for the 1-
octadecanethiol/gold monolayer is the same as that of a liquid alcohol hydrogen bond;
the result of a close balance between the enthalpy and entropy of adsorption, both of
which are significant for this reaction. The enthalpy of adsorption, AHW, was found to
be -20 i l kcal/mol. We note that this value does not correspond closely with the
calculated quantities discussed above for the formation and desorption of the monolayer.
We understand the differences between calculation and experiment based on our use of
standard enthalpies in the calculations and the significant difference between our
experimental conditions, which include solvation effects, and standard conditions. We
can compare our experimental value of AI—Ld, to the AH for desorption of a monolayer of
28 kcanol for desorption into air obtained by Nuzzo and coworkers.“ Our data
contain both solvation and adsorption information and their data are a direct measure of
(reversible) desorption. The difference between their data and ours indicates the
enthalpy of solvation, AH...“ ~ -8 kcal/mol for l-octadecanethiol in n-hexane because
their data represent, almost exclusively, the gold-sulfur interaction. The enthalpy of

solvation is both solvent and thiol dependent and thus -8 kcal/mol is a usefiil, but only

70

qualitative, estimate for other systems. The adsorption enthalpy we measure, AH“. (-20
kcal/mol), consists of the balance between the solvation enthalpy AH,.,]V (-8 kcanol) and
the monolayer formation enthalpy AHm. (~28 kcal/mol).
AHads = AHml — AHsolv [3-31
The quantity AH,“ is dominated by Au-S bond formation and any energy associated with
displacement of solvent from the gold surface during the formation of the monolayer.
We expect that AH,“ will be largely independent of thiol chain length because the
dominant process on these comparatively short time scales is the interaction between the
thiol head group and the metal surface. We note that one potential solvent-dependent
contribution to AH.” may be associated with the polarity of the transition state in the
formation of the gold-thiol bond and the extent to which the solvent stabilizes or
destabilizes that stated” but this effect is likely to be modest because of steric
constraints in the formation of the monolayer and the presence of the metal interface.

The entropy of adsorption, AS“, = -48 i 1 cal/mol-K. This quantity is
approximately four times larger than a typical AS,“s for a liquid-to-solid phase
transitionlm For the same reason that AH.“ is largely thiol independent, we expect the
entropy of adsorption, AS.“ to be substantially independent of thiol aliphatic chain
length. This large entropic term is dominated by the change in system order associated
with alkanethiols oriented randomly in solution going to produce a highly organized,
two-dimensional crystalline array of thiol head groups on gold.

Given the predicted thiol independence of AH... and AS.d,, it is important to

review our previous experiments indicating solvent and thiol dependent monolayer

71

formation. We account for the thiol dependence of AG“. we observed for monolayer
formation from n-hexane based on different solvation enthalpies for the two thiols. For
the adsorption of 1-C13H37SH, A6,... = -5.6 i 0.2 kcal/mol at 293 K and for l-CanSH
monolayer formation, AG“. = -4.4 i 0.2 kcal/mol at the same temperatureml As
discussed above, we believe that the enthalpy of monolayer formation, AHm], is constant
for all thiols at -28 kcal/mol. The solvation enthalpy AH”... is, of course, solvent
dependent. We calculate AH...” = -8 kcal/mol for l-octadecanethiol in n-hexane and
AHWv = —9 kcal/mol for l-octanethiol in n-hexane. Invoking the thiol independence of
AS“. yields AH“, = -19 kcal/mol for the l-octanethiol/gold monolayer. For alkanethiol
monolayer formation on gold, the enthalpy of solvation, AHsolv, is the primary quantity
under the influence of experimental control, and attempts to enhance the robustness of
alkanethiol monolayers will ultimately be limited by the extent to which AHsorv can be
adjusted.
3.4. Conclusions

We have measured the thermodynamics of l-octadecanethiol/gold monolayer
formation from n-hexane solution. Our data indicate that the comparatively small free
energy of adsorption for this system is the result of a close balance between the enthalpic
driving force for the reaction and an entropic penalty associated with the high degree of
organization that accompanies self-assembly. For our experimental conditions, the
dominant chemical reaction is the reversible adsorption of the alkanethiol, and the
temperature dependence of the QCM kinetic response demonstrates that the reaction, for

our conditions, is not diffusion rate limited. These data point collectively to the fact that

72
there are limits to the robustness of alkanethiol/gold monolayers and that the ability to

adjust the formation properties of this system is limited primarily by the solubility of the
alkanethiol in the solvent fi'om which the adsorption takes place. Because the formation
of the thiol-gold bond is fast compared to the structural annealing of the organic tail
group, these data may be of some utility in predicting the formation properties of other
thiol/gold assemblies in the limit that steric contributions from the organic tail groups do
not preclude significant monolayer formation.

3.5. Literature Cited

1. R. G. Nuzzo, L. H. Dubois, D. L. Allara, J. Am. Chem. Soc., 1990, 112, 558.

2. L. H. Dubois, B. R. Zegarski, R. G. Nuzzo, J. Am. Chem. Soc., 1990, 112, 570.

3. R. G. Nuzzo, E. M. Korenic, L. H. Dubois, J. Chem. Phys, 1990, 93, 767.

4. C. E. D. Chidsey, D. N. Loiacono, Langmuir, 1990, 6, 682.

5. C. E. D. Chidsey, C. R. Bertozzi, T. M. Putvinski, A. M. Muscje, J. Am. Chem. Soc.,
1990,112,4301.

6. C. D. Bain, E. B. Troughton, Y. T. Tao, J. Evall, G. M. Whitesides, R. G. Nuzzo, J.
Am. Chem. Soc., 1989, 111, 321.

7. C. D. Bain, G. M. Whitesides, J. Am. Chem. Soc., 1989, 111, 7155.

00

. E. B. Troughton, C. D. Bain, G. M. Whitesides, R. G. Nuzzo, D. L. Allara, M. D.
Porter, Langmuir, 1988, 4, 365.

9. M. D. Porter, T. B. Bright, D. L. Allara, C. E. D. Chidsey, J. Am. Chem. Soc., 1987,
109,3559.

10. G. M. Whitesides, P. E. Laibinis, Langmuir, 1990, 6, 87.

11. N. Camillone III, C. E. D. Chidsey, G. Y. Liu, T. M. Putvinski, G. C. Scoles, J.
Wang, J. Chem. Phys, 1991, 94, 8493.

12. C. E. D. Chidsey, G.-Y. Liu, J. Wang, Langmuir, 1990, 6, 1804.

13

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26

27

28.

73

S. R. Wasserman, G. M. Whitesides, I. M. Tidswell, B. M. Ocko, P. S. Pershan, J.

D. Axe, J. Am. Chem. Soc., 1989, 111, 5852.

C. A. Widrig, C. A. Alves, M. D. Porter, .1. Am. Chem. Soc., 1991, 113, 2805.
G.-Y. Liu, M. B. Salmeron, Langmuir, 1994, 10, 367.

R. L. McCarley, Y.-T. Kim, A J. Bard, J. Phys. Chem, 1993, 97, 211.

R. L. McCarley, D. J. Dunnaway, R. J. Willicut, Langmuir, 1993, 9, 2775.

D. S. Karpovich, G. J. Blanchard, Langmuir, 1994, 10, 3315.

C. A. Widrig, C. Chung, M. D. Porter, J. Electroanal. Chem, 1991, 310, 335.
C.-J. Zhong, M. D. Porter, J. Am. Chem. Soc., 1994, 116, 11616.

D. E. Weishaar, M. M. Walczak, M. D. Porter, Langmuir, 1993, 9, 323.

C. A. Alves, E. L. Smith, M. D. Porter, J. Am. Chem. Soc., 1992, 114, 1222.
C.-J. Zhong, M. D. Porter, Anal. Chem, 1995, 67, 709A.

D. E. Weishaar, B. D. Lamp, M. D. Porter, J. Am. Chem. Soc., 1992, 114, 5860.

E. L. Smith, C. A. Alves, J. W. Anderegg, M. D. Porter, L. M. Siperko, Langmuir,
1992,8,2707.

E. L. Smith, M. D. Porter, J. Phys. Chem, 1993, 97, 8032.
M. 0. Wolf, M. A. Fox, Langmuir, 1996, 12, 955.

J. M. Tour, L. Jones 11, D. L. Pearson, J. J. S. Lamba, T. P. Burgin, G. M.

Whitesides, D. L. Allara, A. N. Parikh, S. V. Atre, J. Am. Chem. Soc., 1995, 117,
9529.

29. N. Lewis, M. Tarlov, J. Am. Chem. Soc., 1995, 117, 9574.

30. R. J. Willicut, R. L. McCarley, Langmuir, 1995, 11, 296.

31.

K. Sinniah, J. Cheng, S. Terrataz, J. E. Reutt-Robey, C. J. Miller, J. Phys. Chem,
1995, 99, 14500.

74

32. R. L. Garrell, J. E. Chadwick, D. L. Severance, N. A. McDonald, D. C. Miles, J.
Am. Chem. Soc., 1995, 117, 11563.

33. J .-PBucher, L. Santesson, K. Kern, Langmuir, 1994, 10, 979.

34. E. Delamarche, B. Michel, H. Kang, Ch. Gerber,; langmuir, 1994, 10, 4103.
35. S.-C. Chang, I. Chao, Y.-T. Tao, J. Am. Chem. Soc., 1994, 116, 6792.

36. P. Fenter, A. Eberhardt, P. Eisenberger, Science, 1994, 266, 1216.

37 . N. Canrillone, III, P. Eisenberger, T. Y. B. Leung, P. Schwartz, G. Scoles, G. E.
Poirier, M. J. Tarlov, J. Chem. Phys, 1994, 101, 11031.

38. M. R. Anderson, M. Gatin, Langmuir, 1994, 10, 1638.
39. H. A. Biebuyck, C. D. Bain, G. M. Whitesides, Langmuir, 1994, 10, 1825.

40. H. Sellers, A. Ulman, Y. Shnidman, J. E. Eilers, J. Am. Chem. Soc., 1993, 115,
9389.

41. Y.-T. Kim, R. L. McCarley, A. J. Bard, Langmuir, 1993, 9, 743.

42. S. E. Creager, L. A. Hockett, G. K. Rowe, Langmuir, 1992, 8, 854.

43. D. M. Collard, M. A. Fox, Langmuir, 1991, 7, 1192.

44. P. Krysinski, R. V. Chamberlain, M. M. Majda, Langmuir, 1994, 10, 4286.

45. O. Chailapakul, L. Sun, C. Xu, R. M. Crooks, J. Am. Chem. Soc., 1993, 115,
12459.

46. L. Sun, R. M. Crooks, A. J. Ricco, Langmuir, 1993, 9, 1775.

47. R. C. Thomas, J. E. Houston, T. A. Michalske, R. M. Crooks, Science, 1993,
259,1883.

48. C. D. Bain, G. M. Whitesides, Science, 1988, 240, 62.

49. J. L. Wilbur, H. A. Biebuyck, J. C. MacDonald, G. M. Whitesides, Langmuir, 1995,
11, 825.

50. R. A. Drawhom, N. L. Abbott, J. Phys. Chem, 1995, 99, 16511.

75

51. Y. Xia, X.-M. Zhao, E. Kim, G. M. Whitesides, Chem. Mater., 1995, 7, 2332.
52. J. F. Ford, T. M. Vickers, C. K. Mann, J. B. Schlenoff, Langmuir, 1992, 12, 1944.

53. E. Kim, G. M. Whitesides, L. K. Lee, S. P. Smith, M. Prentiss, Adv. Mater., 1996,
8, 139.

54. N. Camillone III, C. E. D. Chidsey, G.-Y Liu, G. Scoles, J. Chem. Phys, 1993, 98,
4234.

55. L. Strong, G. M. Whitesides, Langmuir, 1988, 4, 546.

56. G. E. Poirier, M. J. Tarlov, J. Phys. Chem, 1995, 99, 10966.

57. G. E. Poirier, E. D. Pylant, Science, 1996, 272, 1145.

58. T. M. Nahir, R. A. Clark, E. F. Bowden, Anal. Chem, 1994, 66, 2595.

59. I. Taniguchi, K. Toyosawa, H. Yamaguchi, K. Yasukouchi, J. Electroanal. Chem,
1982, 140, 187.

60. T. Sagara, K. Niwa, A. Sone, C. Hinnen, K. Niki, Langmuir, 1990, 6, 254.
61. D. S. Karpovich, G. J. Blanchard, J. Chem. Ed, 1995, 72, 466.

62. K. D. Truong, P. A. Rowntree, submitted to Langmuir.

63. J. B. Schlenoff, M. Li, H. Ly, J. Am. Chem. Soc., 1995, 117, 12528.

64. S. H. Chen, C. W. Frank, Langmuir, 1989, 5, 978.

65. K. J. Laidler, Chemical Kinetics, 3'dEd, Harper Collins, New York, 1987, pp. 231-
232.

66. P. W. Atkins, Physical Chemistry, 4'” Ed, Freeman, New York, 1990, pp. 885-887.
67. G. J. Blanchard, J. Phys. Chem, 1988, 92, 6303.

68. R. G. Nuzzo, B. R. Zegarski, L. H. Dubois, J. Am. Chem. Soc., 1987, 109, 733.

69. G. Z. Sauerbrey, Z. Phys, 1959, 155, 206.

70. M. Yang, M. Thompson, Langmuir, 1993, 9, 802.

76

71. M. Yang, M. Thompson, Langmuir, 1993, 9, 1990.

72. We measured the conductivity of a suspension of 1-3 um gold in anhydrous ethanol
during the addition of ethanolic l-octadecanethiol. We detected no change in the
conductivity of the solution as a result of RSH/Au interaction (< 0.01 u-mho). If the
dissociation of IT from RS' were complete, for our experimental conditions we
would expect the change in [If] = 4.6x10‘5 M. The corresponding change in
conductivity of a 4.6x10"5 M ethanolic HCl solution relative to neat ethanol was
measured to be 0.47 u-mho.

73. L. Stobinski, R. Dus, Appl. Surf Sci, 1992, 62, 77.

74. CRC Handbook of Chemistry and Physics, 7]" Ed, Lide, D. R., Ed; CRC Press,
Boca Raton, FL, 1990.

Chapter 4

Relating the Polarity-Dependent Fluorescence Response of Pyrene t0
Vibronic Coupling. Achieving a Fundamental Understanding of the
py Polarity Scale

Summary

We have investigated the fluorescence response of pyrene, a widely used
polycyclic aromatic hydrocarbon probe molecule, in a series of solvents that span the
empirical “py” polarity scale. The relative band intensities in the pyrene fluorescence
response are determined by the extent of vibronic coupling between the weakly allowed
first excited singlet state and the strongly allowed second excited singlet state. The
solvent-dependence of vibronic coupling in pyrene can be inferred from steady state
fluorescence measurements in conjunction with fluorescence lifetime and quantum yield
measurements. These data, taken collectively, point to the presence of vibronic coupling
in pyrene arising from two different mechanisms. The solvent “polarity” sensed by the
pyrene molecule is related to dipole-induced dipole interactions between the solvent and
pyrene vibrational modes of b3g symmetry which span the nuclear coordinates of the two

excited electronic states.

77

78
4.1. Introduction

Understanding the bulk characteristics of solutions and more highly organized
systems from the perspective of individual molecules is a subject central to the study of
chemistry. Intermolecular interactions ultimately determine the bulk properties of a
chemical system, and a variety of factors need to be considered in establishing this
relationship. Among these factors are molecular structure and conformation,
electrostatic, dipolar and multipolar interactions, and short range dispersive interactions.
The “polarity” of the environment immediately surrounding a molecule often determines
basic properties such as solubility or optical response and, as such, the determination of
“polarity” on a molecular length scale has received a great deal of experimental attention.
One of the most common methods used in the determination of local polar character is
the measurement of some portion of the optical response of a molecule that is sensitive
to a particular property of the local medium. For example, solvatochromic molecules are
a class of molecules that exhibit absorption and/or emission spectral shifts that depend on
the dielectric and dipolar properties of their local environment. Pyrene and other
selected polycyclic aromatic hydrocarbons (PAH) are used extensively for probing local
polarityll'lol While these PAHs do not exhibit spectral frequency shifts similar to
solvatochromic molecules, they do exhibit a variation in the ratio of emission band
intensities that has been correlated with the “polarity” of the PAH immediate
environment. The empirical “py’ scale of solvent polarity has been established based on
the emission response of pyrene and has been catalogued for a wide variety of

experimental conditions, ranging fi'om bulk liquidsm] to interfacial monolayersm‘m

79
While the py polarity scale has enjoyed wide use, there is no general

understanding of the molecular processes responsible for the correlation between
emission band intensity ratios and solvent “polarity”.[4'9] Part of the difficulty in
elucidating the mechanism(s) underlying this correlation lies in the ill-defined nature of
the “polarity” to which the PAH emission features respond. The other significant
question that has limited the utility of this interesting photophysical effect is that there is
not presently a model that can predict, from the linear response, whether a given PAH
will exhibit this effect or not. Many PAHs have been tested and only a limited number

14-91

have exhibited behavior similar to that of pyrene. The linear optical response of

pyrene has been investigated extensively, in part because of the importance of vibronic
coupling in determining the emission and absorption responses of this molecule.“8’29]
Despite the significant understanding of pyrene’s linear response, a firm connection
between that body of information and the empirical “py” scale has not been made. We
assert that there is a direct connection between these two bodies of knowledge and it is
the purpose of this work to make that connection. We propose a mechanistic
explanation for the empirical “py” scale based on what is known about vibronic coupling
in pyrene and several experimental measurements of basic photophysical properties, such
as spontaneous emission frequency dependence, radiative quantum yield and radiative
decay rate constant. These data, taken collectively, can be used to demonstrate a direct
correlation between vibronic coupling exhibited by pyrene and the empirical py solvent
polarity scale. In this paper, following a brief description of the experimental techniques

we have applied to this problem, we detail the relationships between fundamental

quantities such as radiative decay rate constants, fluorescence quantum yield and

80
emission band intensities and then provide experimental data that demonstrate the

correlation between the empirically measured solvent “polarity” and vibronic coupling in
pyrene. We include a discussion of the molecular-scale processes that determine the
pyrene steady state fluorescence response and how the linear absorption response of
other PAHs can be used to predict the presence or absence of a solvent polarity-
dependent emission response.

4.2. Experimental

Chemicals: Pyrene was obtained from Aldrich at 99% purity and was purified by
recrystallizing twice from methanol, followed by vacuum sublimation at 130°C?“
Solvents (Burdick and Jackson Brand high purity grade) were purchased from Baxter
Diagnostics Inc., and were used as received. Prior to lifetime and quantum yield
measurements, all pyrene solutions were subjected to several (at least three)
freeze/pump/thaw cycles to remove oxygen.

Steady state absorption and emission spectra: Emission spectra were recorded
on a Hitachi F-4500 fluorimeter. Excitation was at 334 nm with the excitation slits set
for a 10 nm bandpass. Emission slit width was set to a 2.5 nm bandpass. Absorption
spectra were taken on a Hitachi U-4000 UV-Vis absorption spectrometer.

Fluorescence quantum yield measurements: The fluorescence quantum yield of
pyrene was measured in several solvents by means of comparison of the integrated
emission intensities in those solvents to the intensity measured for a solution of pyrene in
cyclohexane, where the fluorescence quantum yield has been determined before.‘3 11 The

equation used to determine the fluorescence quantum yield ism

81
2‘2: [EXLXAJ [4.1]
¢§1 n,2. Dr Ax

The designators x and r indicate sample and reference, respectively, and (fin is the
fluorescence quantum yield. The term (nxz/nf) is a correction factor for the refractive
index of the solvent and the terms D,- are the integrated intensities under the corrected
emission spectra. A,- is the absorbance of the solutions, a quantity used to correct for
concentration variation between samples. The absorbances and integrated intensities
were available from our steady state absorption and emission spectra, taken from
deoxygenated solutions. It is well known that the fluorescence spectrum of pyrene is
quenched effectively by oxygen dissolved in solution. The mechanism of this quenching
is typically taken to be collisional. Our data show that d)“ and ‘Cfl depend sensitively on
the presence of oxygen, but the M11 band intensity ratio is not sensitive to the presence
of oxygen. This finding is important in its own right because the M11 band intensity ratio
is the most commonly reported solvent polarity-dependent quantity.

Fluorescence lifetime measurements: The time-correlated single photon
counting spectrometer (Figure 4.1) we used to measure the fluorescence lifetimes of
pyrene in different solvents has been described in detail beforeml All pyrene solutions
(~10'5 M) were excited at 332 nm, and fluorescence was collected at 390 nm using a 10
nm FWHM detection bandwidth. Fluorescence was collected over all polarization angles
to ensure the absence of a contribution from rotational diffusion dynamics to the
observed data, although the rotational diffusion time of pyrene in any of the solvents
examined is negligible compared to its fluorescence lifetime. Typical instrument

response time is ~25-50 ps FWHM and the measured fluorescence lifetimes are several

82

3588888 0:50.:— 358805 com vow: :oHoEobooam w::::oo :ocosa 0&5: Baa—oboe oEF .3. 95w:—

  

 

 

 

 

 

 

 

 

 

 

HSE n52 ..ofiEoEooon :WW:
EU .
53888 ..
33588 .
§ .. ... _ob:oo
sflN—Ofi QWU ”VG—“v CO—wflNF—d—a
0’2.
02m
>38 m E: com - E: com
032333 .
88 .8: 1
:33 0% confine 538

 

83833“:
33:

 

83
hundreds of nanoseconds. A typical lifetime scan, together with the instrument response

firnction is shown in Figure 4.2. We found the measured lifetimes to be very sensitive to
the presence of oxygen. For example, we measure the fluorescence lifetime of pyrene in
deoxygenated cyclohexane to be In = 382 ns, and in air-saturated cyclohexane we
measure tn = 20 ns. Reintroduction of 02 into a deoxygenated solutions occurs very

rapidly (seconds to minutes) after exposure to air.
4.3. Results and Discussion

Previous work on the emission response of pyrene has established the empirical
relationship between solvent polarity and the pyrene spontaneous emission band ratios.”
4' In addition, there have been other reports measuring the radiative rate constant(s) of
pyrene in series of solventsm'26' The connection has not, however, been drawn directly
between vibronic coupling and solvent polarity-dependent emission response. The
correlations we report in this paper suggest that vibronic coupling can be used to explain
the empirical py scale. The relationship between vibronic coupling and solvent polarity
can explain why many polycyclic aromatic hydrocarbons do not exhibit a response
similar to that seen for pyrene, and also offers insight into the mechanism by which
solvent polarity is related to the empirical py scale.

In an effort to understand the solvent polarity-dependent emission response of
pyrene, we have measured the steady state fluorescence spectrum, the fluorescence
lifetime, and the fluorescence quantum yield in a range of solvents spanning the empirical
py scale. Each of these quantities exhibits a solvent polarity dependence, but it is

different for each measured pr0perty. As we discuss below, the spectroscopic origin

84

1000 ’

800 b

600 "

Counts

400 "

 

200 '"

 

 

0 _fi 4 1 1 1 1 1 1 1 1 1
O 100 200 300 400 500

Time (ns)

 

Figure 4.2. Fluorescence decay data for pyrene in cyclohexane showing both the
instrument response function (solid line at I ~ 30 ns) and the spontaneous emission
decay (dots).

85
responds to vibronic coupling perturbations arising only from Herzberg-Teller effects

and the emissive response according to the difference between Herzberg-Teller and so—
called Born-Oppenheimer vibronic coupling contributions. By separating these
components in the observed responses, we can determine the relationship between
solvent polarity and vibronic coupling.

The simplest model for predicting the relative intensities of electronic and
vibronic transitions in a given molecule assumes that the electronic wavefunction varies
slowly on the length scale of the nuclear motions associated with molecular vibrationsml

In the Born-Oppenheimer approximation, the wavefunctions of the states involved in the

transition of interest can be separated into their electronic and vibrational components.

1713(4). Q) = 49.01, Q)9v (Q) [42]
The electronic coordinate is designated q and the nuclear (vibrational) coordinate
by Q. This treatment includes the dependence of the electronic wavefimction on nuclear
coordinates but neglects the effect of electron motion on nuclear motion. If the
dependence of 09. on q is assumed weak, i. e. the electronic wavefunction varies slowly
over the relevant nuclear coordinate, then the transition moment for a vibronic transition

can be expressed asml

Mee',vv' = (av l<¢elz €191l¢e'>l 6v') z Ree'<6v lat”) [4-3]

where e and e’ represent the two electronic states, and v and v’ are vibrational states

within the two electronic states. R... is the average electronic transition dipole moment
for the e H e' transition. In this approximation, the vibrational overlap integral is a

deterministic term in the intensity of a particular vibronic transition. While this level of

86
approximation is useful in describing the spectral response of a variety of molecules, the

limitations of this approach are also realized in many other complex organic molecules.
It is not always possible to separate the electronic and nuclear coordinates, and we need
to consider the conditions under which these approximations are not valid. One of the
consequences of the breakdown of these approximations is that “intensity borrowing”,
the exchange of oscillator strength between electronic states, is observed, resulting in the
effective transition cross section of the weaker transition being enhanced by vibrationally
mediated coupling to the neighboring electronic statesml It is typically held that the
separation of electronic and nuclear coordinates, as described by Eq. 4.3, is the most
limiting approximation and its failure leads to so-called Herzberg-Teller (HT) vibronic
coupling. The second limitation is the breakdown of the adiabatic approximation, i. e.
non-Bom-Oppenheimer behavior, which is often termed “BO” vibronic coupling.
Vibronic coupling arises from the dependence of the transition moment, Meew, on
nuclear positions (HT coupling) and on nuclear momenta (BO coupling). It is clearly
artificial to completely separate these two mechanisms because both depend on nuclear
motion, but in different ways. In limiting cases, however, where one mechanism
dominates over the other, the two mechanisms exhibit different frequency
dependenciesllmu

Prior work on the origins of vibronic coupling in complex organic molecules has
shown that both types of coupling can, under certain circumstances, be significant.
Experimentally, the presence of both HT and BO vibronic coupling have been inferred by
the different isotope dependencies of individual absorption and emission bands and by

[19,21]

asymmetry between absorption and emission spectra. We focus here on the

87
emissive response of pyrene and how the absorption/emission asymmetry arising from

vibronic coupling can be used to understand the solvent-dependence of the emission
profile for pyrene and, by extension, other PAHs. Solvent-mediation of the efficiency of
vibronic coupling gives rise to the observed solvent polarity dependent emission
response of pyrene.

Although it is well recognized that vibronic coupling between many electronic
manifolds can contribute to the linear response of molecules like pyrene,[19'22] it is
typically assumed that the dominant vibronic coupling is between the excited states
involved in the transitions. Vibrationally mediated coupling between the electronic
manifolds can be expressed through a perturbation treatment, yielding expressions that
contain both HT and BO contributions to vibronic coupling. For a molecule with
electronic states 0, 1 and 2, where state 0 is the ground electronic state, the 0-1
transition is forbidden or weakly allowed and the 0-2 transition is more strongly allowed,

the vibronically induced absorption from the ground electronic state to a vibronic level in

the first excited state is described by“91

 

 

 

 

 

 

 

 

T hwc ] ‘
1+ x
AE0 +Ve=2hwv - hwc
M _ ._ . _ ._ : — l v=v fl v=c v=c v=0 I OVEO va
e—O,e —1,v—0,v '1 14:12 < e=2 <¢2l d]: l¢l><6e=0 lqc e=1 > e=l > R02< e—O e—2>
AEO + Ve=2h60v
~ 1
[4.4]

where the 0-2 transition is strongly allowed and therefore described adequately in the
Condon approximation. Implicit in this treatment is the assumption that a vibrational

mode spanning the appropriate coordinates of the two electronic states couples the two

88

electronic manifolds. The numerator of the second term in Eq. 4.4 describes the extent
of coupling between the electronic states 1 and 2 due to the nuclear perturbation along
the coupling coordinate q,, the vibrational coupling for mode v, between electronic states
0 and 1 and the extent of vibrational overlap between electronic manifolds 1 and 2. The
first and third terms in Eq. 4.4 will not be affected significantly by solvent polarity while
the magnitude of nuclear displacement along q, will be sensitive to the dipolar and
dielectric properties of the surrounding medium. The solvent polarity dependence of M
arises from either a change in polarizability or dipole moment associated with a nuclear
displacement along q,. The vibrational states that act as final states in the linear emission
response are Raman active modes, and because pyrene possesses a center of inversion,
none of these modes will exhibit a change in dipole moment on population. Indeed, it is
not critical that an induced dipole moment in pyrene span the entirety of qc to interact
with the solvent dipole. Local induced dipoles in pyrene that are canceled when
vibrational motions are integrated over the entire molecular framework may play an
important role in determining the strength of the solvent-solute interaction. In other
words, the solvent-solute interaction need not be symmetric with respect to the pyrene
nuclear framework. It makes intuitive sense that solvent~solute interactions need not
coincide with the solute symmetry axes, especially when the relative sizes of pyrene and
individual solvent molecules are comparedml Many solvent molecules are too small to
span the entire length of the pyrene molecule. Our previous work on T1 relaxation in
perylene/n-alkane systems shows that local solvent structure, persisting over a length
scale much shorter than the perylene molecule, plays an important part in determining the

vibrational relaxation dynamics of that molecule.[”'37' We discuss below how the solvent

89
polarity dependence of M is manifested in the emission response of pyrene. The

transition moment for the emission response (1—0) is given bym]

 

 

 

 

 

 

 

 

[[1 _ hare ]X l
AEO + 148:2th + hwc
Maura-1:11:11 = — z < (93:3... (12 lfllax 3:51, 93:10) 3:?) 1-R02(6::8 2:33)
vc=2 dfc
AE0 + ve=2hwv
[4.5]

While similar to Eq. 4.4, the transition moment for emission differs from that for
absorption by two signs. One sign change is not obvious by inspection, and arises from

>[19.21]

the anti-hermitian nature of the terms <(p2I6He/6qclcp1> and <(p1I6He/6qcltp2 while

the other sign change occurs for energetic reasons. In the case of absorption, the

frequency difference between the mode v, and the final state of the 0——)1 transition is 2110)

more than that for the corresponding (l—>0) emission process. This intrinsic asymmetry
in vibronic coupling for absorption and emission can also be considered in terms of the
B0 and HT contributions to the process. The vibronic coupling contributions are, in
general, additive for absorption and subtractive for emission. This property of vibronic
coupling is central to understanding the solvent polarity dependent fluorescence response
of pyrene.

Individual features in the emission band of pyrene change according to the
polarity of its local medium. This polarity dependence is expressed in terms of the “MU”
band ratio,[2'3] where the I band is the S {’4’ —> Sov=° (0-0) transition and III is an SI‘FO —>
So‘Fl transition. For the 111 band, the v=1 resonance is dominated by one (or more)

~1100 cm‘l b3g Raman active vibrational mode(s).[38’391 Because the S1 state in pyrene

90

(lBgu) is y axis polarized and the S; state (lBlu) is z axis polarized, the vibrational
coordinate, qc, appropriate for coupling these two states must necessarily span both y
and z axesml For a molecule belonging to the D2,. point group, the dominant coupling
mode is thus of b 38 syrmnetry. In order to understand the relationship between the M11
band intensity ratio and the solvent polarity dependence of M, we must consider the
nature of the vibronic coupling that contributes to each transition. Experimental data
presented by Hara and Wareml demonstrate that the 0-0 transition cross section of
pyrene is correlated to solvent polarity, although they did not attempt to extract from
these data a mechanistic explanation for the polarity-dependence of the MU band
intensity ratio. The strength of the 0-0 transition depends on vibronic coupling between
states S1 and 82. While the coupling between S2 and 81 involves specific vibrational
levels in each state, the second terms in Eqs. 4.4 and 4.5 couple the vibrational
wavefimctions 052‘” and 01:1‘50, and thus because the 053:0 wavefimction is affected,
the 0-0 transition moment will necessarily also be affected by vibronic coupling. For a
two level system, such as the pyrene 0-0 transition, the cross section for the absorptive

”91 The vibronic

process must be equal to the cross section for the emissive process.
coupling contribution to the absorption cross section of the 0-0 transition will scale as
the sum of the HT and B0 contributions and the vibronic coupling contribution to the
emission intensity of the 0-0 transition will scale as the difference between the HT and
B0 contributionsm'm Since the cross sections for these two transitions are necessarily
equal, as we have indicated in Figure 4.3, we are left with the conclusion that, for the 0-0

transition, the sum of the vibronic coupling contributions is equal to the difference

between them, i.e. (HT +BO)01) = (HT ~BO)01). Thus, the vibronic coupling induced

91

 

 

 

 

 

 

 

 

 

 

 

 

 

2.0 r
n / SO—-) S2 absorption

1.5 - /S0—) S1 absorption (normalized)
§ Iband
L510 _ [l y ,111 band
0
.53
. l
g S1—) S0 emission
: l /

0.5 '

l
00 1 I I 1 1
300 350 400 450

wavelength (nm)

Figure 4.3. Absorption and spontaneous emission spectra of pyrene in cyclohexane.
We report the So —-> S, absorption spectrum with the intensity of the 0-0 band
normalized to the 0-0 emission transition.

92
intensity of the 0-0 transition is determined exclusively by HT coupling between

electronic states S2 and S]. The 111 band is an (S 1".‘0 —> Sow!) transition and its intensity
will scale with the difference between the HT and B0 vibronic contributions. The ratio
of the bands will scale with the magnitude of HT coupling divided by the difference
between HT and B0 contributions. As solvent polarity increases, the M11 band ratio
should increase, as is observed experimentally.

To determine the validity of our assertion that vibronic coupling is the mechanism
of the solvent polarity dependent emission response seen in pyrene, we must relate the
quantities that we determine experimentally to the properties related directly to emission
intensities for the I and III bands of pyrene. Measurement of the spontaneous emission
spectra of pyrene in series of solvents does not, by itself, provide sufficient quantitative
information for the determination of absolute fluorescence intensities for each band.
Time domain measurements are more amenable to the detemrination of absolute
spectros00pic quantities. We separate the fluorescence spectrum of pyrene into two
portions; the 0-0 transition and all else. We can determine the fluorescence lifetime and
the fluorescence quantum yield quantitatively, and can estimate the fractional

contribution of the 0-0 transition to the total emission response. The spontaneous

emission lifetime is related to the Einstein A coefficient,m]
¢fl
AS, 480 = kR = :—
fl [4.6]
ASl—rSo : 2 ASl(v=O)—-)So(v=v) = ZAOV
”=0," v=0,n

where In is the measured fluorescence lifetime and 4)“ is the fluorescence quantum yield.

We estimate the contribution of the total emission response arising from the 0-0

93
transition by peak fitting and integration. The intensity and transition cross section of the

0-0 transition are related to the fluorescence lifetime through the Einstein A

coefficientml

12170 = Nohc Vo—vo-o
[4.7]

[0-0
00—0 = ———e"'
8MCZV3_0
Similarly, the intensity and transition cross section for the remainder of the

emission band is related to the radiative rate constants,

0'0_0 oc kg—O
[4.8]

00-v 0C kit—v
Measurement of the fluorescence quantum yield in conjunction with the fluorescence
lifetime data provides a quantitative determination of be and kNR (=(1-¢fl)/tfl). The
solvent polarity-dependence of kRo'o and kg“ provide insight into the role that vibronic
coupling plays in this process because the transition cross sections are related directly to

the transition moments bym]

87r3v _ N 2
0.4. = —3—‘;,;2—°-IM.-.I [4.91

where a is the lower energy state in the transition and N, is the population of that state.
We show in Figure 4.4a the relationship between the radiative rate constant and
pyrene I/III band ratio. At best there is a weak correlation between these two quantities,
and we endeavor to determine more precisely which contributions to the radiative decay
rate depend on solvent polarity and which ones do not. We consider, as implied above,

that there are essentially three excited state decay channels that are important to the

94

4.0x106 :- a

3.511106 —- "

3.0XIO6 "' I 0 1
$4 .

 

2.5x106 -
2.0x106 - I I
1.5x106 r- f

1.0x1060 ‘ l ‘ l ‘ l ‘ '

 

 

0.8 —

0.6 L-

~[II

¢1
1—~—1 -
1—+—4

 

 

0.0 0.5 1.0 1.5 2.0
Py

Figure 4.4: (a) Radiative transition rate constant, kg, for pyrene in the solvents used in
this work plotted versus empirical solvent polarity parameter, py. (b) Dependence of
fluorescence quantum yield, 11)“, on solvent polarity. Uncertainties reported in these
figures are i 10.

95

solvent polarity dependence of pyrene’s emission; the radiative decay rate for the 0-0
transition, with associated rate constant kg”, the radiative decay rate for the balance of
the emission spectrum, with rate constant kRa", and the non-radiative decay rate, with
rate constant kNR. Indeed, it is important to consider non-radiative processes because e97
~ 0.65, and there is no discernible correlation between fluorescence quantum yield and
the py scale (Figure 4.4b). We have determined kg” by multiplication of the total
radiative rate constant by the fraction of the emission spectrum composed of the 0-0
band. This means for the determination of kg” is valid because the measured
fluorescence lifetime is uniform across the entire emission profile, i. e. only one excited
electronic state contributes to the observed steady state emission spectrum. We show in
Figure 4.5 the correlation between kRM and the M11 band ratio (py scale). The
relationship between kRa" and the III“ band ratio is shown in Figure 4.6. While there is a
direct correlation between log” and py, kRa" is only weakly solvent-dependent. The only
other decay channel which could exhibit a solvent dependence is the non-radiative decay
channel, and we show in Figure 4.7 the relationship between km; and py. While there are
variations between solvents, it is clear that there is no systematic trend in the data. We
note that the reported 4m for acetone appears to be anomalously large. The reason for
this anomaly is that acetone absorbs light significantly at the excitation wavelengths used
for the quantum yield measurements (334 nm is 5 nm), making uncertain the amount of
energy absorbed by pyrene. Such an effect gives rise to an apparently low quantum
yield, and the value that we report is corrected according to a previously reported

method?“ We believe that this correction is limited in its accuracy because of the large

96

 

 

 

 

 

 

8x10S -
.... T
0
6x105 — .. ‘—
Q. __ I 41
9:: ’
« 0 —11—
.1.—
411105 L 1.
2x105 '- T
1 I 1 1 1 l 1 1
0.4 0.8 1.2 1.6 2.0

Py

Figure 4.5. Radiative transition rate constant for the 0-0 transition of pyrene, plotted
versus solvent py value. The reported uncertainties are i lo.

97

2.8x106 -

2.4x106- ‘ , .»

 

 

2.0x106 -

 

 

1.6x106 -

 

 

 

1.2x106 -

 

1 J 1 l 1 I 1 I

0.4 0.8 1.2 1.6 2.0
Py

 

Figure 4.6. Radiative transition rate constant for the family of 0-v transitions that
comprise the dominant portion of the pyrene emission band, plotted versus solvent py
value. The reported uncertainties are i 10.

98

4x106 - I

3x106 -
m
112 -

2x106 -
0
1x105 - I

 

 

 

 

O 1 1
0.4 0.8 1.2 1.6 2.0
py

Figure 4.7. Nonradiative transition rate constant for the pyrene emission band, plotted
versus solvent py value. The reported uncertainties are i 10.

99
concentration difference between absorbing species and the unknown extent to which

energy transfer proceeds in the dilute pyrene/acetone solution.

As detailed above, the cross section for the S1V=° (—) So":0 transition is sensitive to
vibronic coupling arising from deviations from the Condon approximation (HT
coupling), while the S {=0 —) Sov=l transition senses the difference between HT and B0
vibronic coupling contributions. Because both of these contributions depend on nuclear
motion, we expect that they will scale with one another in some manner, although the
details of this relationship are not clear and must be considered in the context of the
specific coupling vibrational mode and the dielectric and dipolar properties of the
surrounding medium. We note two important results from our data. First, kRO'o
correlates with solvent polarity, indicating that HT contributions increase with solvent
polarity, as expected. We note (Table 4.1 and Table 4.2) that the py scale and kg” both
scale with solvent dipole moment and not dielectric constant. The second significant
result is that the portion of the spontaneous response that is related to the difference
between HT and B0 terms, k120i”, depends on solvent polarity differently than kRO'O. The
correlation between solvent polarity and kRa" is in qualitative agreement with theoretical
predictions. In the classical treatment, the relationship between transition moments

[21]

arising from HT and BO coupling is described by

M30 = ——-———h“’0 [4.10]
MHT sees, IE8, “Esl

 

for a coupling mode of fi'equency has, (1100 cm‘1 for pyrene) and two non-degenerate

electronic states. For pyrene, IE3: - ESII ~ 2700 cm’1 and therefore the

100

 

8:823

 

03.3300 NN.0 18.030: 3.91.504” v0.0fl00 07102 modflog 3508::
00:38:00

02 0.3% .0 0N0 34.0.30; v0.0umnoN 8.011430 941108 8.0.135— 35086-22
$0.0Hm0m0 3.0 $.38; mm.0fl0.m 8.38.0 $00M 8.005; 0:55:08
30.03000 3.0 2 .0Hm00 2 .0.+.mm. m «0.0111000 030m 8.0002 0:88:
3.0.0330 «N0 2 .0Hmm0 2 .0H00.N 00.0HN00 Nude? N00091: 888: 350
02:25

08.32 00 3.0 2.01834” 2 .0110m.~ 8.3.000 91:2 00.3.3; 0:03:qu
000.0Hm0v0 E0 94.3.3.0 neon—Nd 3 .3210 0710?. 8.03: 3:28.:
08.03% .0 mm .0 S .011L00 2.0.12.1.0 3.0114050 0.183“ 00.?er . 0 3:083:85
000.0330 2 .0 0m.0flv._ $011021. w0.0fl0.0 SHEN 8.090; 0:023
30.033 .0 :0 N~.0flm._ 804940; 00.0%.? .0 3.03 m 8.33 .0 0:86:41
30.34080 N _ .0 8030.0 «0.04105— 30 033 8.03m .0 05:08:20.6

.... 403 ..m 40:0 7... 8.25
88$ .3320 :5 «a :6 $5 5 029, .3 2028

 

.2 H 0.8 $058.38: 4:028 00 5:88 a 3 0:23 00 8:008: Bongo 2: 00 80.8qu A... 035—.

101

Table 4.2. Polar properties of the solvents used in this work.

 

 

solvent py value dielectric dipole
i 10 constant moment
80 D
cyclohexane 0.56:0.01 2.02 0.00
n-hexane 0.56:t0.01 1.89 0.08
toluene 1.03i0.01 2.38 0.45
tetrahydrofuran 1 .27i0. 02 7. 60 1 .75
methanol 1.301.001 32.6 1.70
methylene l.33i0.02 9.08 1.60
chlon'de
ethyl acetate 1.3 5i0.02 6.02 1.81
acetone 1.59i0.01 20.7 2.88
acetonitn'le 1.70i0.01 37.5 3.92
N,N-dimethyl 1.76i0.02 36.7 3.86
formamide
dimethyl 1.9 1_+_0. 02 46.6 3 .90

sulfoxide

 

102
ratio Mao/Mm is predicted by Eq. 4.10 to be 0.41. We can estimate this same ratio from

our experimental data. The slope of the kRM polarity dependence (HT contributions) can
be compared to the slope of the logo" (HT-BO) polarity dependence, once the relative
weighting factors for kRM and kRo" are accounted for. We estimate that ~20% of the
total emission response is accounted for by the 0-0 band, based on the cross section ratio
data presented in Table 4.1. The slope of the kg” polarity dependence is ~3.5 x 10’ s",
(Figure 4.5, 20% contribution to the spectrum) and the slope of the kRa" dependence is
«6 x 105 s'1 (Figure 4.6, 80% contribution to the spectrum). Correcting for fractional

'1 if the total

contributions to the spectrum yields an effective slope of 1.75 x 106 5
emission band response is determined by HT coupling. For kg“, the slope of the polarity
dependence would be 7.5 x 105 5'1 if purely (HT-BO) contributions to vibronic coupling
determine the response. These estimates yield an experimental ratio of ago/om z 0.57
(Eq. 4.8). Because 0 at A42 (Eq. 4.9), Mao/Ml” z V0.57 = 0.75, somewhat larger than
the theoretical prediction. We consider the agreement between the estimated and
theoretical ratios to be quite good considering the uncertainty in our experimental data,
the fact that we used several estimates in determinating this ratio, and the assumptions
implicit in the derivation of Eq. 4.10.

A complicating factor that may give rise to the slight disagreement between
experiment and theory lies in the inherent complexity of the pyrene vibrational response.
Theories of vibronic coupling typically consider the coupling between electronic states to
be mediated by a single vibrational mode that spans the coordinates of the two electronic

states. Pyrene, however, has at least four b 38 modes in the frequency range of 1000 cm'1

- .1500 cm'1 that could participate in the coupling, with the dominant b 3g mode appearing

103
-1 [38,39]

near 1100 cm . We present an exaggerated schematic of the nuclear motions
associated with these vibrations in Figure 4.8. We determined these nuclear
displacements using semi-empirical molecular orbital calculationsf‘w] and take them as
instructive, rather than quantitative, representations of the relevant normal modes.
Because of the complexity associated with relating the empirical py scale to vibronic
coupling, we cannot speculate on what relative contribution each of these modes may
have to the overall process.

Not all polycyclic aromatic hydrocarbons exhibit a measurable solvent polarity

”’91 For example, perylene and pyrene are both peri-

dependent emission response.
condensed fused ring systems. Perylene contains five fused rings and does not exhibit a
solvent polarity dependent emission response, whereas pyrene, with four fused rings,
does. Both of these molecules are of D2,. symmetry, but for perylene the So —) S1
transition is allowed, (am ~ 33,000 cm'l M") and for pyrene it is nominally forbidden, or
at best only weakly allowed (emu ~ 200 cm" M“). The reason for the fimdamental
difference between these two molecules lies in the correlation and electrostatic forces
that detemline state ordering in organic systems, and a discussion of several factors that
relate to state ordering in cata- and peri-condensed systems has been reported by other
authors.[“'431 Because the allowed 80 —> S 1 transition in perylene is strong, it can be
modeled effectively by separating the vibrational and electronic coordinates, and Franck-
Condon transitions lead to nearly mirror image absorption and emission spectra.[“] For
pyrene, because the So —+ S 1 transition is nominally forbidden, but fluorescence from the

S1 is observed, a simple Franck-Condon treatment does not describe its optical response

adequately. The linear optical response of pyrene can be understood in terms of vibronic

104

1111 cm"1 1176 cm'1

 

Figure 4.8. Representations of nuclear motions associated with the four b 33 vibrational
modes in closest energetic proximity to the “III” band of pyrene. The frequencies
reported for these modes were taken from references 38 and 39.

105

coupling between the S; and S; electronic states,”

and we have argued that this

coupling is the basis for the solvent polarity dependent emission response seen for

pyrene.

It is important to test the validity of our hypothesis, that solvent polarity mediates
vibronic coupling and therefore emission band intensities, in the context of its predictive
power. We consider whether or not we can predict, based on the linear response of a
given PAH, whether or not it will exhibit a solvent polarity-dependent emission response.
If vibronic coupling gives rise to the solvent-dependent emission response seen for
certain PAHs, we predict that PAHs having a strongly allowed So <—> S; transition and
exhibiting a classical mirror image symmetry in their absorption and emission spectra will
not display a solvent polarity-dependent emission response because vibronic coupling is
too subtle an effect to contribute significantly to the intensities of strongly allowed
transitions. If vibronic coupling is the operative mechanism for solvent polarity-
dependent emission, then a PAH with a weakly allowed So <—> S; transition, and a more
strongly allowed So —) Sz transition, should exhibit a solvent polarity-dependence, as is
observed experimentally for benzene, coronene, benzo[e]pyrene, and
benzo[g,h,i]perylene.“’7] All of these molecules possess a weakly allowed So <—> S;
transition (r:max ~ 250 M’1 cm'l) and a strong So —-> Sz transition within 2000 - 4000 cm'1
of the So (—-) S;. In contrast, perylene, benzo[a]pyrene and dibenzo[a,e]pyrene all
possess a strongly allowed 8;; —> S; transition and none of these molecules exhibit a
solvent polarity dependencem While this is clearly a limited data set, largely because of

a lack of experimental data, the strong correlation between weakly allowed So <——) S; /

strongly allowed So —) S2 transitions and a solvent polarity-dependent emission response

106

indicates that, at least for the molecules listed above, vibronic coupling between excited
electronic states 8; and S2 determines the presence (or not) of a solvent polarity-

dependent emission response in polycyclic aromatic hydrocarbons.
4.4. Conclusions

We have examined the solvent polarity-dependent emission response of pyrene in
a series of solvents that span the empirical py scale. Measurement of fundamental
photophysical parameters such as radiative rate constant and fluorescence quantum yield
have provided insight into the correlation between the empirical py scale and the
efficiency of vibronic coupling between closely spaced electronic states. We assert based
on the correlations we observe that the empirical py scale for pyrene can be explained in
terms of vibronic coupling, and that the important intermolecular process that gives rise
to the py scale is solvent dipole-solute induced dipole interactions. The dipolar nature of
the solvent medium determines the extent to which of an induced dipole moment is
formed in pyrene by vibrational distortion(s) of the nuclear coordinate of b 3,, symmetry.
Vibronic coupling in pyrene can be used to describe qualitatively the trends seen in the
py scale, and can be used to predict the presence of a solvent polarity-dependent

response in a given PAH.

4.5. Literature Cited

1. R. B. Cundall; L. C. Pereira; Chem. Phys. Lett., 1973, 18, 371.
2. D. C. Dong; M. A. Winnik; Can. J. Chem, 1984, 62, 2560.
3. D. C. Dong; M. A. Winnik; Photochem. Photobiol., 1982, 35, 17.

4. W. E. Acree; S. A. Tucker; L .E. Cretella; A. I. Zvaigzne; K. W. Street; J. C. Fetzer;
K. Nakasuji; I. Murata; Appl. Spec, 1990, 44, 951.

107

5. S. A. Tucker, W. E. Acree; M. J. Tanga; Appl. Spec, 1991, 45, 57.
6. R. Waris, K. W. Street, W. E. Acree, J. C. Fetzer; Appl. Spec, 1989, 43, 845.

7. W. E. Acree; S. A. Tucker; A. I. Zvaigzne; K. W. Street; J. C. Fetzer; H.-F.
Grutzmacher, Appl. Spec, 1990, 44, 477.

8. S. A. Tucker; I.-L. Teng; W. E. Acree; J. C. Fetzer; Appl. Spec, 1991, 45, 186.

9. S. A. Tucker; A. I. Zvaigzne; W. E. Acree; J. C. Fetzer; M. Zander; Appl. Spec,
1991, 45, 424.

10. S.-H. Chen; V. L. McGuflin; Appl. Spec, 1994, 48, 596.

11. J. W. Carr; J. M. Harris; Anal. Chem, 1986, 58, 626.

12. J. W. Carr; J. M. Harris;J. Chromatog, 1989, 481, 135.

13. R. G. Bogar; J. C. Thomas; J. B. Callis; Anal. Chem, 1984, 56, 1080.

14. E. Wistus; E. Mukhtar; M. Almgren; S.-E. Lindquist; Langmuir, 1992, 8, 1366.
15. Y. S. Liu; W. R. Ware;J. Phys. Chem, 1993, 97, 5980.

16. Y. S. Liu; P. de Mayo; W. R. Ware; J. Phys. Chem, 1993, 97, 5987.

17. Y. S. Liu; P. de Mayo; W. R. Ware; J. Phys. Chem, 1993, 97, 5995.

18. A. Pellois; J. Ripoche; Chem. Phys. Lett., 1969, 3, 280.

19. P. A. Geldof; R. P. H. Rettschnick; G. J. Hoytink; Chem. Phys. Lett., 1971, 10, 549.
20. G. Orlandi; W. Siebrand; Chem. Phys. Lett., 1972, 15, 465.

21. G. Orlandi; W. Siebrand; J. Chem. Phys, 1973, 58, 4513.

22. K. Cunningham; W. C. Brand; D. F. Williams; G. Orlandi; Chem. Phys. Lett., 1973,
20,496.

23. A. Nakajima; Bull. Chem. Soc Japan, 1973, 46, 2602.

24. V. D. Tuan; U. P. Wild; M. Lamotte; A. M. Merle; Chem. Phys. Lett., 1976, 39,
118.

108

25. T. Kobayashi; J. Chem. Phys, 1978, 69, 3570.

26. K. Hara; W. R. Ware; Chem. Phys, 1980, 51, 61.

27. A. Nakajima; SpectrochimicaActa, 1982, 38A, 693.

28. F. W. Langkilde; E. W. Thulstrup; J. Mich]; J. Chem. Phys, 1983, 78, 3372.
29. G. Marconi; P. R. Salvi; Chem. Phys. Lett., 1986, 123, 254.

30. J. N. Demas, private communication.

31. J. N. Demas; G. A. Crosby; J. Phys. Chem, 1971, 75, 991.

32. L. DeWitt; G. J. Blanchard; E. LeGofl‘; M. E. Benz; J. H. Liao; M. G. Kanatzidis; J.
Am. Chem. Soc., 1993, 115, 12158.

33. G. Herzberg, Molecular Spectra and Molecular Structure, Vol. III, Krieger, 1991.
34. J. T. Edward;J. Chem. Ed, 1970, 47, 261.

35. Y. Jiang; G. J. Blanchard; J. Phys. Chem, 1994, 98, 6436.

36. Y. Jiang; G. J. Blanchard; J. Phys. Chem, 1994, 98, 9411.

37. Y. Jiang; G. J. Blanchard; J. Phys. Chem, 1994, 98, 9417.

38. A. Bree; V. V. B. Vilkos; Spectrochimica Acta, 1971, 27A, 2333.

39. E. A. Mangle; M. R. Topp; J. Phys. Chem, 1986, 90, 802.

40. Calculations were performed using Spartan version 2.0 with AMI parameterization.
41. H. B. Klevens; J. R. Platt; J. Chem. Phys, 1949, 17, 470.

42. D. E. Mann; J. R. Platt; H. B. Klevens; J. Chem. Phys, 1949, 17, 481.

43. J. R. Platt; J. Chem. Phys, 1949, 17, 484.

44. I. B. Berlman, Handbook of Fluorescence Spectra of A romatic Molecules, 2nd
Edition, Academic Press, 1971, 399-400.

Chapter 5

Dynamics of a Tethered Chromophore Imbedded in a Self-Assembled
Monolayer

Summary

We report the first measurements on the dynamical response of alkanethiol/gold
monolayers using a depth-selective tethered probe molecule technique, where 5% of the
monolayer is comprised of the probe. Our transient spectroscopic data indicate that, for
the first ~10 CH2 groups above the gold surface, the local environment is highly rigid
and, from these same data, we recover an average tilt angle of 26°, in exact agreement

with steady state measurements.

109

1 10
5.1. Introduction

Due to their numerous potential uses and nearly ideal properties as model
systems, there is wide spread interest in the organization and dynamics of self-assembled
monolayers, with particular interest on the alkanethiol/gold system because of its ease of
formation and characteristically high degree of macroscopic order.“'l3‘ Self-assembled
monolayers are used in interfacial electron transport research, micro-device patterning

114-191 In addition, there is growing interest in developing

applications, and tribology.
molecular shape-selective interfaces and in building patterned structures using self-
assembled monolayersm’z" Although much is known about the macroscopic properties
of self-assembled monolayers, little is known about their dynamics at the molecular level.
The dynamical response of these monolayers will ultimately determine their applicability
for patterning and sensor applications.

To understand these dynamics, we have performed probe molecule reorientation
experiments using tethered pyrene within self-assembled monolayers. The rotational
motion of a molecule is affected greatly by the shape and rigidity of its immediate
environment, and these measurements allow us to determine the molecular-scale
motional freedom and anisotropy characteristics of self-assembled monolayers. There
are two approaches to such measurements; physisorption of chromophores onto the
monolayerm] and chemical incorporation into the layering”. We have chosen the latter
route. By tethering the chromophores to the substrate, we minimize their translational
fi'eedom within the monolayer and sample only the rotational motion intrinsic to

molecules bound to the surface. We present here our results on a 5% 4-(1-pyrenyl)-1-

butanethiol /95% l-octadecanthiol monolayer on gold (Figure 5.1).

111

 

I
55 t-. .
vain-er}
.Ifr‘l’. .'

Figure 5.1. Qualitative schematic diagram of the alkanethiol/gold interface with the
tethered 4-(1-pyrenyl)-1-butanethiol chromophore present.

1 12
5.2. Experimental

The substrates used were formed from evaporated gold that yields predominantly
Au(l 11) microcrystalline domains'9’m' The 4-(1-pyrenyl)-1-butanethiol was synthesized
in our laboratoriesm' by the reaction of 4-(1-pyrenyl)-l-butanol with thionyl chloride to
form 4-(l-pyrenyl)-1-chlorobutane and, subsequently with thiourea to yield 4-(1-
pyrenyl)-l-butanethiol.[26l l-Octadecanethiol was purchased from Aldrich and used as
received. The monolayers were formed from n-hexane solution, with 5x10'6 M 4-(1-
pyrenyl)-1-butanethiol and 9.5x10'5 1\_/1 1-octadecanethiol concentrations. We
characterized the thiol/gold monolayers with external reflectance FTIR (Nicolet 550,
liquid nitrogen cooled MCT-A detector). The FTIR measurements indicated highly
ordered structures characteristic of long chain alkanethiol/gold monolayer assemblies.
Work on similar systems has demonstrated a linear relationship between chromophore
fractional concentration in solution and in the formed monolayer?"

Time-correlated single photon counting (Chapter 4, Figure 4.1) was used to
perform the time resolved emission measurements necessary for these studiesm’zgl We
used a mode-locked CW Nd:YAG laser (Quantronix 416) to pump a cavity dumped dye
laser (Coherent 702) operated at 630 nm using Kiton Red laser dye (Exciton). The
output of the dye laser was typically 100 mW average power at 10 MHz repetition rate,
with ~5 ps pulses. A fraction of the light is used as a reference beam, with the remaining
light being frequency doubled to provide the 315 nm excitation of the pyrene
chromophore. The polarization of the excitation beam is set to be vertical and, using a

Glan-Taylor prism, emission polarization is selected either parallel or perpendicular to

the excitation polarization. The instrument response firnction of this system is typically

113
30 ps FWHM. The dynamical information is obtained from the experimental data

 

through R(t).
1;;(0-110)
R = .
(t) I”(t)+21 1(1) [5 1]

Where 1;;(t) and I;(t) are experimental signal intensities for emission parallel and
perpendicular to excitation polarizations. For all measurements, the samples were made
just prior to measurement and the data were taken for the sample in air.

5.3. Results and Discussion

The tethered chromophore system we report here is modeled best as a hindered
rotor according to Lipari and Szabo’s treatmentml In this model, the motion of the
chromophore is described as being confined within a cone of semi-angle 200 and with 0
the average tilt angle of the chromophore within the cone (Figure 5.2). R(t) represents a
re-randomization of the chromophore orientational distribution within the confining cone
subsequent to the optical selection of a non-random subset of all chromophores within
the monolayer.

R(t) = R(oo)+ [R(0)— R(00)]exp(- t/z') [5.2]
The quantity R(O) is related to the angle between the excited and emitting transition
moments of the probe. R0») is related to 0, the average tilt angle of the chromophore
within the cone, and the quantity R(oo)/R(O) is related to 00, the maximum tilt angle
within the cone.

We detect non-aggregated pyrene emission at 400 i 30 nm. The use of a wide

detection bandwidth was necessary owing to the weak signal recovered from our sample.

114

 

Figure 5.2. Schematic of the hindered rotor model used in interpretation of the R(t)
data. The cone semi-angle 200 and the average chromophore tilt angle, 0, are indicated.

115
We sampled ~ 109 chromophore molecules in a typical experiment. The signal was

attenuated significantly by chromophore-to-metal dipolar coupling, a well known
quenching mechanism.m'32] The emission near 400 nm is characteristic of individual
pyrene molecules and not pyrene excimer (see Chapter 4). At the low chromophore
concentrations we used, the chromophores are not expected to aggregate on the surface,
but are dispersed within the alkanethiol monolayer. Even if undetectable aggregation is
occurring in these monolayers, however, we are spectrally excluding any excimer
contribution to our data because the monomer and excimer emission bands of the pyrene
chromophore are well isolated from one another. We observe a non-exponential time
decay of the fluorescence intensity (Figure 5.3) and attribute this behavior to
heterogeneity of chromophore binding sites within the monolayer assemblyml
Unfortunately, the limited signal-to-noise ratio of the data precludes the use of any
detailed fitting analysis which could provide insight into the site distribution sampled by
the chromophore. The fluorescence half-lives we recovered were very short for these

[3" We attribute the shorter

systems (Ty, ~ 500 ps) compared to pyrene in solution.
lifetimes to at least three quenching mechanisms: Dipolar coupling of the chromophore
to the metal, pyrene’s sensitivity to the surrounding environment, and the possible
presence of oxygen. We do not have the means to determine the relative importance of
each of these quenching mechanisms at present, but note that the presence of 02 may not
be an important factor based on other experiments on alkylated pyrene that show only a
limited sensitivity to this quenching agentm

We have studied the linear optical response of pyrene previouslyf‘w and have

found that the dominant absorption and emission transition moments are polarized

116
1000

N

I I IIIUT]
v-

C
'0
$

’0'“. ~fi‘.
vi
‘5.
3...

0”

Y

 

100

’1'-
if

..«i

1
0

p—
O

1 f1 rirrrv
C
0

emission intensity (counts)

1

 

l : : .-. .-...0.

 

l g l . l
0 1 2
time (ns)

Figure 5.3. Experimental time resolved emission intensity data for a monolayer of 5%
4-(1-pyrenyl)-1-butanethiol and 95% l-octadecanethiol on Au. The data (solid circles)
are plotted on a log;;; scale to underscore the non-exponential nature of the decay. The
instrument response function is shown as open circles.

117

nominally perpendicular to each other, a condition which gives rise to the negative R(O)
we observe experimentally. The dynamical response recovered from these monolayers
revealed the absence of fast chromophore motion deep within the monolayer. The R(t)
data (Figure 5.4) indicate that the intrinsic anisotropy in the system is constant over
several probe lifetimes (nanoseconds), pointing to either a highly rigid system or an
extremely free local environment, where the dynamics proceed within much less than the
30 ps response time of the spectrometer. We believe that the environment is rigid, based
on the expected stick limit free-rotor reorientation time of 10R = 220 ps for 4-(1-
pyrenyl)-1-butanethiol in n-hexadecane, a time regime that we can access easily. We
attribute the rigidity of the alkanethiol/gold system to the efficient aliphatic chain packing
of alkanethiol/gold monolayers about the tethered probe. We expect that, if we tethered
the pyrene with a longer alkane chain, we would observe rotational motion since the
pyrene would reside in a less rigid area of the alkane chain structurem This assumption
is consonant with simulations of these systems suggesting that the alkyl chains are rigid
near the surface and less rigid near the ends of the monolayer.“

Despite the absence of measurable chromophore motion, we are able to apply the

hindered rotor model to our data,
R(oo) = —0.2<I’2(c086))2 [5.3]

At t = 0, we detect the sum of the anisotropies present within the system, [R(O) + R(oo)],

and from solution phase measurements of 4-(1-pyrenyl)-l-butanethiol, R(O) = -0.05.

From these data, the contribution to R(t) fiom the structural rigidity of the

monolayer is R(oo) _—~_ -0.10, yielding the average tilt angle of the chromophore within the

118

 

 

 

 

1.0 —
.E‘
8
E emission perpendicular to excitation
'0
E 0.5 —
7::
E emission parallel to excitation
G
0.0 1 L . I
0 1 2
0.2 —
0.0 ---------------------- -- - - - -
E I. ‘ l
-0.2 — I . H ‘1
1 1‘:
-O.4 ' A: . 1 L . . .
0 1 2

time (ns)

Figure 5.4. Top panel: Experimental emission intensity data for a monolayer comprised
of 5% 4-(1-pyrenyl)-l-butanethiol and 95% 1-octadecanethiol on Au. Note that 11(t) is
more intense than 1;;(1), as is expected for pyrene. Bottom panel: R(t) data calc ulated
from the raw experimental data using Eq. 5.1.

119

cone of 0 = 26°. This value correlates well to the accepted value of 30° for the tilt angle
of alkane chains in alkanethiol/gold monolayers determined by external reflectance FTIR

spectroscopy.
5.4. Conclusions

We have demonstrated, for the first time, a viable method for probing motional
dynamics within monolayer assemblies that is sensitive to selected regions within the
layer. The results we obtained are fully consistent with known properties of
alkanethiol/gold monolayer systems. We have shown experimentally that
alkanethiol/gold monolayers are quite rigid in the intermediate region between the
substrate and the top on the interface, and measurements are underway to probe different
regions of these monolayer assemblies.

5.5. Literature Cited

1. C.-J. Zhong, M. D. Porter, Anal. Chem, 1995, 67, 709A.

2. N. Lewis, M. Tarlov, J. Am. Chem. Soc., 1995, 117, 9574.

w

. R. J. Willicut, R. L. McCarley, Langmuir, 1995, 11, 296.
4. L. H. Dubois, B. R. Zegarski, R. G. Nuzzo, J. Am. Chem. Soc., 1990, 112, 570.
5. R. G. Nuzzo, E. M. Korenic, L. H. Dubois, J. Chem. Phys, 1990, 93, 767.

6. E. B. Troughton, C. D. Bain, G. M. Whitesides, R. G. Nuzzo, D. L. Allara, M. D.
Porter, Langmuir, 1988, 4, 365.

7. M. D. Porter, T. B. Bright, D. L. Allara, C. E. D. Chidsey, J. Am. Chem. Soc., 1987,
109, 3559.

120

8. R. L. McCarley, Y.-T Kim, A. J. Bard, J. Phys. Chem, 1993, 97, 211.

9. D. S. Karpovich, G. J. Blanchard, Langmuir, 1994, 10, 3315.

10. H. M. Schessler, D. S. Karpovich, G. J. Blanchard, J. Am. Chem. Soc., in press.

11. C. A. Widrig, Chinkap Chung, M. D. Porter, J. Electroanal. Chem, 1991, 310, 335.
12. C.-J. Zhong, M. D. Porter, J. Am. Chem. Soc., 1994, 116, 11616.

13. D. E. Weishaar, M. M. Walczak, M. D. Porter, Langmuir, 1993, 9, 323.

14. C. D. Bain, G. M. Whitesides, Science, 1988, 240, 62.

15. J. L. Wilbur, H. A. Biebuyck, J. C. MacDonald, G. M. Whitesides, Langmuir, 1995,
11, 825.

16. R. A. Drawhom, N. L. Abbott, J. Phys. Chem, 1995, 99, 16511.
17. Y. Xia, X.-M. Zhao, E. Kim, G. M. Whitesides, Chem. Mater., 1995, 7, 2332.
18. J. F. Ford, '1. M. Vickers, C. K. Mann, J. B. Schlenofl‘, Langmuir, 1992, 12, 1944.

19. E. Kim, G. M. Whitesides, L. K. Lee, S. P. Smith, M. Prentiss, Adv. Mater., 1996,
8,139.

20. L. Sun, R. M. Crooks, Langmuir, 1993, 9, 1775.
21. L. Sun, L. J. Kepley, R. M. Crooks, Langmuir, 1992, 8, 2101.

22. M. J. Wirth, J. D. Burbage, Anal. Chem, 1991, 63, 1311.
23. J. W. Carr, J. M. Harris, Anal Chem, 1986, 58, 626.
24. J. W. Carr, J. M. Harris,.l Chromatog, 1989, 481, 135.

25. M. Pallmer, private communication.

121

26. G. A. Page, D. S. Tarbell, J. Am. Chem. Soc., 1953, 75, 2053.
27. J. C. Horne, G. J. Blanchard, J. Am. Chem. Soc., in review.
28. G. R. Holtom, Proc. SPIE, 1990, 1204, 2.

29. L. DeWitt, G. J. Blanchard, E. LeGoff, M. E. Benz, J .-H. Liao, M. G. Kanatzidis, J.
Am. Chem. Soc., 1993, 115, 12158.

30. G. Lipari, A. Szabo, Biophys J.; 1980, 30, 489.

31. R. R. Chance, A. Prock, R. Silbey, J. Chem. Phys, 1975, 62, 2245.
32. R. R. Chance, A. Prock, R. Silbey, Adv. Chem. Phys, 1978, 37, 1.
33. H. Wang, J. M. Harris, J. Phys. Chem, 1995, 99, 16999.

34. D. S. Karpovich, G. J. Blanchard, J. Phys. Chem, 1995, 99, 3951.

35. J. J. Tulock, G. J. Blanchard, J. Phys. Chem, in preparation.

36. S. J. Klatte, T. L. Beck, J. Phys. Chem, 1995, 99, 16024.

37. J. Hautman, M. L. Klein, J. Chem. Phys, 1989, 91, 4994.

Chapter 6

Adsorption of Organic Molecules onto Metal and Modified Interfaces.
Evidence for Site-Specific Adsorption

Summary

We have studied the adsorption of organic molecules from the vapor phase onto
clean gold and thiol-modified gold surfaces using a controlled atmosphere chamber with
an ellipsometer and a quartz crystal microbalance for simultaneous in-situ determinations
of interfacial thickness and mass. Our results on maximum adsorbate mass uptake for
each of the organic adsorbates are consistent with adsorption being dominated by dipole-
induced dipole interactions for polar molecules and induced dipole-induced dipole
interactions for non-polar molecules. For the thiol-modified surfaces, the extent of
adsorption depends on the polarizability of the substrate. These data do not indicate the
dominance of chemically selective interactions between the substrate and the adsorbate,
but the relationship between interface thickness and adsorbed mass do indicate that the
adsorbate layers are packed more densely than in their bulk liquid states. The packing
densities we measure are consistent with site-specific adsorption, at least for the bare
gold surface, where the data indicate adsorption to both types of three-fold sites present

on the Au(l 11) surface.

122

123
6.1. Introduction

There has been considerable interest in designing chemically selective interfaces
for use in sensing and separations applications. Organic-modified interfaces such as
alkanethiol/gold monolayers have received extensive attention because they form readily,
exhibit a high degree of order, and their structural and chemical properties (e. g. surface
wettability, aliphatic chain ordering, electrochemical activity) have been determined.'”81
Perhaps more important, their surface properties may be modified using thiol adsorbates

[1.10.4942]

with different (o-terminal functionalities. Early studies of alkanethiol/gold

monolayers used contact angle measurements to probe their wetting properties.“'l°]
These studies gave intuitive insight into the interactions between the monolayers and
macroscopic amounts of adsorbates (water and n-hexadecane), reinforcing the notion
that interactions between adsorbates and interfaces can be understood within the same
interpretive framework as intermolecular interactions in the liquid phase, where there are
no restrictions on intermolecular alignment or distance. As interest in these systems has
grown, monolayers with different temiinal fimctionalities have been studied using these
same approaches, and indications of chemical selectivity have been reported in certain
cases. The Crooks group has focused on carboxylic acid terminated thiol/gold
monolayers interacting with vapor-phase alkanoic acids, and they concluded from FTIR
and ellipsometry data that hydrogen bonding played a dominant role in determining inter-

1

layer attachment and organization.‘50 Their subsequent work using Cu2+-carboxylate

terminated monolayers showed chemical selectivity for organophosphonates over simple
organic solvents.“ Crooks has also explored the use of nanoporous alkanethiol/gold

monolayers where the porosity was induced by short hydroxyl terminated thiols."3"4]

124
These systems exhibited size exclusion effects with a variety of redox-active molecules.

These and other studies relate the selectivity of the monolayers to either chemically or
structurally mediated interactions. These approaches to achieving chemical selectivity
are based on the starting point that the interactions between adsorbates and interfaces
can be understood in the same terms as solution phase intermolecular interactions. In
this chapter, we present our studies on the interactions of organic vapor phase molecules
with metal and thiol-modified substrates.

We have designed an apparatus that allows us to study both the mass and
thickness change of interfacial films while adsorption of low molecular weight organic
molecules fi'om the vapor phase is proceeding. Our apparatus uses a controlled
atmosphere chamber to house the sample. An optical ellipsometer is used for in-situ
thickness measurement and a quartz crystal microbalance (QCM) is used as the
substrate. This combination of measurement capabilities allows in-situ and real-time
mass and thickness measurements to be made on the interfacial layers. We expose the
substrates to an atmosphere of organic vapor and nitrogen gas and measure in-situ the
mass and thickness of adlayers composed of the organic constituent. We used bare gold,
l-octadecanethiol/gold monolayers, ll-mercapto-l-undecanol (MUD)/gold layers and
mixed MUD and l-dodecanethiol/gold layers as substrates for adsorption of the selected
organic species. From these studies we obtained two quantities; the maximum mass
adsorbed and the relationship between adsorbed mass and adlayer thickness for each
adsorbate/surface combination. The limiting mass data are related to the strength of
adsorbate-surface interactions and the thickness versus mass data are related to the

packing density of the adsorbed molecules at the surface. We used nonpolar (n-hexane

125

and cyclohexane), polar protic (methanol and water) and polar aprotic (acetonitrile and
acetone) solvents as adsorbates. The data on these systems shed light on the dominant
factors in the adsorbate-surface interactions and on the structural constraints imposed on
the adlayers by the structure of the substrate. We find that chemically-specific
interactions, such as hydrogen bonding, do not dominate the adsorption process but,
rather, the efficiency of adsorption at these interfaces is determined primarily by the
polarizability of the substrate. In addition, we find evidence that the atomic- or
molecular-level structure intrinsic to the surface can affect the adsorbate layer packing
density significantly.
6.2. Experimental

Controlled atmosphere chamber: The chamber used in this work is a stainless
steel cube equipped with six 3.375” flange mounts (Huntington P/N VF-6-200). The
chamber is mounted to an optical breadboard by one flange. The substrates we used
were quartz crystal microbalances (QCMs) with polished, vapor deposited gold
electrodes. The QCMs were suspended in the chamber from the top flange by a
temperature controlled QCM holder (Maxtek). The temperature of the QCM was
maintained at 10°C below the ambient (chamber) temperature for all experiments using
a circulating bath. A BNC feed-through connector was used (top flange) to connect the
QCM to its associated circuitry. The oscillator circuit (Maxtek P/N 124200-4) is
powered by 5 VDC. We monitored the frequency with a universal frequency counter
(Phillips PM 6673) equipped with an analog voltage output. This voltage was collected

and stored electronically (IBM compatible PC).

126
Substrates: The QCMs (McCoy Electronics P/N 78-18-4, 6 MHz resonance

frequency), were AT cut crystalline quartz and had polished, vapor deposited gold
electrodes. The gold electrodes were cleaned using piranha solution (3:1 H202:H2SO4),
rinsed with distilled water and dried before use. The electrodes were used either as bare
gold surfaces or were modified with thiols. For these devices, the change in oscillation
frequency depends on the mass adsorbed onto the crystal surface according to the

Sauerbrey equation?”

— 2Am(fo)2n
Af : 1/2 [6- 1]
A(up)

Where it = the harmonic of the fimdamental frequency (n = 1 here), p = the density of
quartz (2.648 g/cm3), [.1 = the shear modulus of quartz (2.947x1011gcm’ls'2), A = the area
of the deposit (0.33 cmz), fo = the fiJndamental frequency of the crystal (6 MHz), and
Am is the mass change of the crystal. For the QCMs that we use, we calculate the
sensitivity to be 1 Hz decrease in frequency per 4 ng of adsorbed material, and this
relationship is quantitative when using the QCM with gas phase adsorbates.[56‘57] We
note that slight variations in the QCM fundamental oscillation frequency result in a
variation in sensitivity according the Eq. 6.1. We have determined that the slight
variation in f0 for the QCMs we use affects the sensitivity insignificantly when compared
to other sources of uncertainty in our experiment.

Chamber optical and mechanical access. In two chamber faces, at right angles
to one another, fused silica windows are mounted to pemiit the use of an optical
ellipsometer. The windows are mounted with their surfaces at 25° with respect to the

flange faces, such that the ellipsometer HeNe laser enters and exits the chamber normal

127

to the face of each window to preclude polarization changes associated with refraction.
The ellipsometric angle of incidence is 70° with respect to the substrate surface normal.
A vacuum system is connected to one flange for evacuation of the chamber. A butterfly
valve isolates the chamber from the vacuum pump when not needed. One flange
contains the inlet and outlet ports for the organic vapor/nitrogen gas stream. A
schematic of the chamber is shown in Figure 6.1.

Adsorbate concentration control. We use a dynamic flow through approach to
controlling the organic vapor atmosphere within the chamber (Figure 6.2). This method
of atmospheric composition control ensures that the atmospheric composition within the
chamber is not affected significantly by adventitious sources or sinks for organic
adsorbates within the system. The nitrogen (AGA Gas, Inc.) is dried before its
application to the gas blending system. The dry nitrogen line is split and sent to two
adjustable flow meters (Cole Parmer P/N NO32-41). The maximum flow attainable with
these meters is 148 cms/min N2. One meter controls the flow of the dry nitrogen stream
and the other meter is used to control the flow of the nitrogen stream through the gas
washing bottle (the wet nitrogen source). The two gas streams are rejoined using a “T”
fitting and sent through the chamber to a trap. The flow meters control the vapor
concentration in the chamber atmosphere. We show in Table 6.1 the saturation
concentration, expressed as mole fraction of adsorbate vapor in the wet nitrogen stream.
These concentrations are, of course, proportional to the vapor pressures of the liquids
used.

Adlayer thickness measurement: We used optical null ellipsometry for in-situ

[58-61]

thickness measurements of adlayers. The ellipsometer was constructed in-house

128

005—00800: .8005 5.000 n 200 0:0 00000000 00020800000 u 0.20 .030 0>03 00:00.0 .1. 030 50305 n <
0000200 u 0 .80.:0 0000—00—0 u 04G .0000880E0 00000000005 .23 008000 05:08:50 05 00 00080000 303 00 0. .—.0 0.53.0

 

040

030

200
E0800

 

 

8:009
02:00.020
. 200
0>_0> \ \
\
\
\
x \ 0000000080
. 0 00:00
v 0 =

 

 

129

 

.8000? max? 0:0 :000:0:0w 0000> 05 00 0008000m dd 0.:.w_m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

208
@3003 0%
®
0
O
o A 030:
.91 0:029»
O
>
‘x 10.: x N Z 003
x x x x 0 \ 0008000
\0 x x W v 00 8005
B000.~:\30_0 3m 0000::

Nz 00

130

Table 6.1. Saturation concentrations of organic vapors in the “wet” nitrogen gas

 

 

 

stream.
Adsorbate mole fraction, X vapor pressure at
20° C (Torr)

n-hexane 0.13 121
methanol 0.11 96
acetonitrile 0.09 73
cyclohexane . 0. 10 78
acetone 0.24 185
water 0.02 18

 

 

131

using discrete optical components as indicated in Figure 6.1. A linearly polarized HeNe
laser (Melles-Griot, 1 mW) was used as the light source and a Glan-laser prism (Melles-
Griot, extinction ratio 105) was used to ensure a linear polarization condition. A
polarizer (P, CVI zero order M2 plate), and a N4 plate (CV1) are used to control the
polarization condition of the light prior to its entrance into the chamber and reflection
from the QCM. The light is incident on the sample at 70° with respect to the substrate
normal. The polarization condition of the light is altered by reflection from the surface
according to the optical constants of the substrate and the thickness and dielectric
response of the adlayer(s). The reflected light passes through the analyzer (A, Melles-
Griot Glan-laser prism) to a photomultiplier detector (RCA 1P28A). With the sample in
place, the polarizer (P) and analyzer (A) are rotated until the minimum detector signal is
achieved, and this set of readings is taken to be the optical null. The P and A readings
are converted to the ellipsometric quantities A and w which are used to calculate the
thickness of the adsorbed layer, given its refractive index. We use commercial software
(Rudolph Research) for thickness calculations and a commercial automatic null
ellipsometer (Rudolph Research AutoEL II) for calibration and comparison to our
manual null ellipsometer. For all comparisons made, the data were identical. The
commercial ellipsometer was used to measure the thickness of the thiol/gold monolayers
directly after their formation due to the greater ease of sample manipulation with this
system.

Chemicals and Materials: The thiol/gold monolayers were prepared by
immersion of cleaned QCMs in 0.1 mM thiol (either l-octadecanethiol or ll-mercapto-

l-undecanol (MUD) solutions in n-hexane for 24 hours. We characterized the thiol/gold

132
monolayers with external reflectance FTIR using a Nicolet 550 FTIR with a liquid

nitrogen cooled MCT-A detector. Mixed l-dodecanethiol-MUD/gold layers were
synthesized from 0.02 mM MUD, 0.08 mM l-dodecanethiol in n-hexane solution. 1-
Octadecanethiol and l-dodecanethiol were purchased from Aldrich Chemical Co. and
used as received. MUD was synthesized according to the method described by Frey et
al.” The organic adsorbates n-hexane, methanol, acetone, acetonitrile, and
cyclohexane were purchased from either Aldrich or Burdick and Jackson and were used

as received. The water used in these experiments was in—house distilled water.
6.3. Results and Discussion

The principal foci of this work are understanding the forces that dominate
adsorbate interactions with selected surfaces and the factors that are important to the
organization of the adsorbate layers as they form. To address these two questions, we
have determined saturation mass loadings of the adsorbates onto the surfaces and the
relationship between adlayer mass and thickness up to the saturation mass loading. We
studied adsorption to bare gold to establish a point of reference for the other substrates,
and we consider the results for this surface first.

As can be seen from the data presented in Table 6.2, the maximum mass
adsorbed onto the gold substrate was essentially constant for all adsorbates except water.
The strength of interaction for the gold surface is dominated by the surface polarizability.
We expect that the mass of adsorbed water achieved at maximum vapor concentration is
substantially lower than that of the other adsorbates because of the limited vapor phase

concentration of water in nitrogen. Earlier macroscopic wetting studies of gold surfaces

133

Table 6.2. Maximum mass and thickness uptake of adsorbates onto substrates.

 

 

 

 

 

 

 

 

adsorbate substrate maximum mass maximum
molecule (ng) thickness (A)
: 95% CI. : 95% CI.
n-hexane gold 80 : 10 14.3 : 0.7
l-octadecanethiol/gold 33 : 2 8 : 3
MUD/gold 90 : 1 15 : 4
mixedMUDand l- 80:20 17:1
dodecanethiol on gold
methanol gold 90 : 20 13 : 2
l-octadecanethiol/gold 24 : 2 2 : 1
MUD/gold 110:30 16:5
mixedMUDand 1- 110:60 19:4
dodecanethiol on gold
acetonitrile gold 60 : 10 7 : 2
l-octadecanethiol/gold 20 : 10 1.0 : 0.5
MUD/gold 50 : 10 5 : 1
mixedMUDand 1- 60:20 7:2
dodecanethiol on gold
acetone gold 89 : 9 12 : 1
l-octadecanethiol/gold 50 : 10 1.3 : 0.6
cyclohexane gold 80 : 20 12 : 2
l-octadecanethiollgold 31 : 4 31 : 4
water gold 30 : 10 2 : 2

l-octadecanethiol/gold

not measurable

not measurable

 

 

134
indicated their hydrophilic nature.“'101 A generally accepted implication of that work is

that the gold surface will interact more strongly with polar adsorbates than with non-
polar adsorbates. Our data do not support this broad implication, and we offer an
explanation for this apparent inconsistency with intuition following consideration of the
adsorption behavior of the selected other surfaces.

Compared to metallic gold, the 1-octadecanthingold monolayer showed a much
lower maximum mass uptake for all adsorbates. In addition, our data do not show any
significant preference on the part of the l-octadecanthiol/gold monolayer toward any
adsorbate. We take from these data that the basic mechanism of interaction between
adsorbate and surface is the same (equally non-selective) for both substrates, and that
there is simply an absolute difference in the strength of adsorbate-surface interaction for
the two substrates. A basic physico-chemical property that satisfies the constraint of
non-selective interaction is the polarizability of the substrate. We expect that the metal
surface is more polarizable than the aliphatic surface, and this expectation is in agreement
with experimental data on strength of interaction.

While the postulation of adsorbate-surface interactions being dominated by the
polarizability of the substrate is plausible, at least intuitively, we need to test this
hypothesis using a substrate possessing a chemically selective flinctionality. Earlier
studies by the Crooks group demonstrated hydrogen bonding interactions between vapor
phase adsorbates and functionalized monolayer substratesml Their data demonstrated
adsorption of alkanoic acids a to carboxylic acid terminated thiol/gold layer, but not to
an alkanethiol/gold monolayer. We have chosen to use hydroxyl-terminated monolayers

(1 l-mercapto-l-undecanol, MUD) instead of carboxylic acid terminated monolayers as a

135
chemically selective surface. We made this choice based on the similar hydrodynamic

radii of the -OH (1.60 A) and -CH3 (1.76 A) groups,“ in the hope that steric effects at
the layer temiinus would not affect overall ordering. Our data show that the interactions
between all adsorbates and the full MUD/gold substrate are substantially the same,
suggesting the same absence of chemical selectivity for this surface that we observed for
both the clean gold and alkanethiol/gold surfaces. This is not necessarily a surprising
result because adlayers of MUD on gold should exhibit significant intermolecular
hydrogen bonding effects, precluding chemically selective interactions with adsorbates.
In an effort to evaluate this possibility, we have compared data for adsorption to a fiill
MUD/gold layer to that for a mixed MUD/l-dodecanethiol/gold interface. We observe
no significant chemical selectivity on the part of the mixed surface for any of the
adsorbates tested.

It is instructive at this point to consider the composition and structural properties
of the full and mixed MUD surfaces. We have deliberately avoided referring to the
MUD/gold systems as monolayers. Ellipsometric evidence suggests that MUD does not
form simple monolayer structures. The measured thickness of a full MUD/gold assembly
(23 A) is essentially twice that of a l-dodecanethiol/gold monolayer (12 A). It is
tempting, based on the integral relationship between these thickness values, to postulate
that the MUD forms a bilayer structure spontaneously. Such a structure, however,
would likely be accompanied by a significant level of organization within each layer.
External reflectance FTIR measurements of these assemblies do not indicate a highly
ordered structure compared to that observed for l-dodecanethiol and 1-octadecanethiol

monolayers on gold. These results leave open the possibility that the MUD adlayer

136

exists either as a disordered system in the form of a double layer or as an agglomerated
multilayer and, in either case, it is not clear that the notion of discrete layers is
appropriate. Attempts to remove overlayers of MUD by rinsing with excess methanol or
ethanol were unsuccessful (i.e. no change in ellipsometric thickness), suggesting that
interactions between MUD molecules in the assembly are chemical, not simply physical,
in nature.

It is clear based on the ellipsometric data and washing experiments that hydrogen
bonding between MUD molecules is important in determining the properties of the
MUD/gold assemblies. Liquid alcohol hydrogen bonds are substantially the same
strength as the thiol/gold interaction energy (AG... ~ -5.5 kcal/mol).“8’63] It is likely that
hydrogen bonding affects the formation and annealing of MUD/gold assemblies. The
resulting structure will be held together by hydrogen bonds and attached to the surface
by sulfiir-gold bonds. Our results demonstrate that hydrogen bonding by the MUD
assembly to adsorbates does not dominate the interactions we sense, indicating that the
free energy of adsorption of small organic molecules is substantially less than -5
kcal/mol.

Our data for the MUD/gold systems indicate fiirther the dominance of simple
dipolar interactions in the adsorption process. Gold possesses a higher polarizability
than aliphatic monolayer terminal CH3 groups. In this model, both permanent and
induced dipole moments mediate the strength of adsorbate-surface interactions. For the
gold and alkanethiol substrates, there is no permanent dipole moment in the plane of the
surface, and thus the polarizability of these media will determine their adsorption

properties. For the MUD layers, it is not possible to distinguish the presence or absence

137
of a permanent dipole moment because of the comparatively limited information

available on the mesoscopic structure of these assemblies. Thus for MUD/gold layers,
either permanent or induced dipole moments may serve to mediate its interactions with

polar and! or polarizable adsorbateswl

 

 

k
V = — ——
r6
2 2
= M (dipole - dipole interactions)
3kBT 47:60
2 a' [6.2]
k = fia—i (dipole- induced dipole interactions)
47%.},
k = 361303 [1111’] ] (induced dipole- induced dipole interactions)

The significance of Eqs. 6.2 is that all of these interactions operate over the same
length scale, with the magnitude of the interaction potential being determined by the
intrinsic properties of the substrate (s) and adsorbate (a), such as the polarizability (a’),
permanent dipole moment (u), and ionization energy (1). Because of their strong
permanent dipole moments, methanol and acetonitrile can interact strongly with the
polarizable surfaces. n-Hexane has a very small permanent dipole moment, and its
interactions with the substrates will be mediated by the polarizabilities of both, implying
the need for the interaction between several CHx groups per adsorbate molecule and the
substrate. The differences in substrate polarizabilities and/or dipole moments account for
the high mass loading of all organic vapors on bare gold and hydroxyl-containing
substrates compared to that achievable with the alkanethiol monolayers. Despite the

strong interactions between adsorbates and the MUD/gold and the mixed MUD/gold

138

systems, we found no evidence for the dominance of hydrogen-bonding interactions over
the less specific polar interactions described above.

To this point, we have considered the maximum mass loading of the various
adsorbates onto the substrates. In addition, we have monitored the thickness change of
the adlayers as a function of adsorbate mass. We determined these relationships by
controlling the adsorbate concentration in the atmospheric chamber, as described in the
Experimental section. We found a linear relationship between adlayer mass and
thickness for all systems studied (Table 6.3 and Figure 6.3). The linear relationship
between adlayer mass and thickness shows that adsorption to the surface is a statistical
process, not affected significantly by aggregation. The relationship between mass and
thickness is unique for each system, and there is chemical significance to these
relationships, as we consider below.

In order to determine the adlayer densities, we must know the surface area of the
QCM that is available for adsorption. The total surface area is the product of the
macroscopic geometric surface area and the surface roughness factor, R (R = actual
surface area/geometric surface area). Schlenoff et al. have reported surface roughness
factors for gold surfaces made and treated in several different ways.“ The values
reported range from R = 1.2 for polished, annealed gold to R = 2.0 for gold sputtered
onto glass; a perfectly smooth surface would have R = l. The gold electrodes on our
QCMs are vapor deposited and polished, and for this type of surface, Schlenoff et al.
have reported that R = 1.44.1651 The geometric active area of the surfaces we use is 0.33
cm2, and with the correction for surface roughness, the area available for adsorption onto

the QCM surface is 0.48 cmz. From the data presented in Table 6.3 and the active QCM

139
Table 6.3. Mass vs. thickness increase of adsorbates on substrates

 

 

 

adsorbate substrate slope : 95% CI.
molecule mass/thickness (ng/A)
n-hexane gold 5.4 : 0.4
l-octadecanethiol/gold 5 : 1
MUD/gold 5.4 : 0.8
mixed MUD and l- 5 : 2
dodecanethiol on gold
methanol gold 6.5 : 0.2
l-octadecanethiol/gold not measurable
MUD/gold 6.7 : 0.5
mixed MUD and 1- 5 : 2

dodecanethiol on gold

 

acetonitrile gold 8.0 : 0.5
l-octadecanethiol/gold not measurable
MUD/gold 8 : l
mixedMUDand 1- 9:2

dodecanethiol on gold

 

 

 

 

acetone gold 7.2 : 0.8
l-octadecanethiol/gold not measurable
cyclohexane gold 7.0 : 0.6
l-octadecanethiol/gold 7 : 2
water gold not measurable

l-octadecanethiol/gold not measurable

 

 

140

80!—

60

 

 

 

’53

5 40
m
(D
co
E

20

o

l l l
0 5 10 1 5
thickness (A)

Figure 6.3. Plot of adsorbed mass vs. adlayer thickness of a gold surface while exposed
to an atmosphere of n-hexane in nitrogen gas.

141
surface area, we calculate the densities of the adsorbed organic layers (Table 6.4). The

values we obtain for the adlayer densities are substantially higher than densities of the
corresponding bulk liquids, and these differences place important limits on the possible
ways in which the adsorbates can interact with the substrates.

It is fair, of course, to consider that our ability to determine adlayer density is
limited by our knowledge of the surface roughness factor, R. For the experimental
densities to match the corresponding bulk liquid densities, we would have to adjust the
surface roughness factor to be ~2.8, a quantity we believe to be unrealistically large
given the optical and physical properties of the QCM electrodes. An R value of 2.8
implies a surface much rougher than sputtered gold on glass (R = 2.0) which is unlikely
for polished, vapor deposited gold surfaces. Likewise, we recognize that our QCM
surfaces are far from perfect (R = 1), and a value of R = 1.44 is entirely consistent with
our knowledge of these surfaces. We believe the derived adlayer densities are
approximately correct and, based on this starting point, consider the limits the density
data place on adsorbate packing.

[62] and molecular weights, we

Using the calculated van der Waals volumes
calculate theoretical close-packed densities for the organic molecules (see Table 6.5)
which are similar to our experimental densities. The volumes used for calculation of
these densities do not include any free space between the molecules and therefore
represent the extreme upper limit on densities attainable for these molecules. The
theoretical close-packed densities are not, however, the preferred densities of the organic

liquids, as indicated by comparison to experimental bulk liquid densities. The bulk liquid

densities reflect the optimal intermolecular distances determined by the balance of

142
Table 6.4. Adsorbed layer densities calculated fiom mass/thickness values.

 

 

 

adsorbate substrate density : 95% CI.
molecule (g/cm3)
n-hexane gold 1.13 : 0.09
l-octadecanethiol/gold 1.0 : 0.3
MUD/gold 1.1 : 0.2
mixed MUD and 1- 1.0 : 0.4
dodecanethiol on gold
methanol gold 1.36 : 0.05
1-octadecanethiol/gold not measurable
MUD/gold 1.4 : 0.1
mixedMUDand 1- 1.1:O.4

dodecanethiol on gold

 

 

 

 

 

acetonitrile gold 1.7 : 0.1
l-octadecanethiol/gold not measurable

MUD/gold 1.7 : 0.3

mixedMUDand 1- 19:05

dodecanethiol on gold

acetone gold 1.5 : 0.2
l-octadecanethiol/ gold not measurable

cyclohexane gold 1.5 : 0.1

l-octadecanethiol/gold 1.4 : 0.4
water gold not measurable

1-octadecanethiol/gold not measurable

 

 

143
Table 6.5. Densities of adsorbates (g/cm3).

 

 

 

adsorbate bulk liquid Average density from density for site-
density experimental van der Waals specific model
density“) volumes
: 95% CI.
n-hexane 0.66 1.1 : 0.1 1.26 1.10
methanol 0.79 1.3 : 0.2 1.33 0.58
acetonitrile 0.78 1.7 : 0.2 1.45 0.70
acetone 0.79 1.5 : 0.2 1.49 0.89
cyclohexane 0.78 1.4 : 0.2 1.37 1.10

 

(”Average for the adsorbates over all substrates

 

144

attractive and repulsive forces averaged over all orientations. To achieve a density
different than that of the bulk liquid, an additional, external force must be exerted on the
system. The attractive and repulsive forces between adsorbate molecules are nominally
constant or, at worst, are altered only slightly by any molecular directionality enforced by
adsorption. The external force responsible for this change in density must result from
interactions of the adsorbates with the substrate. For the substrate to affect the inter-
adsorbate distance, the adsorbate to substrate interactions must be site specific.

For simplicity, we treat the Au surface first, because it possesses the greatest
surface order of the substrates we have used. We consider that the QCM surfaces are
comprised of predominantly Au(l 11) and, thus it is possible to estimate the geometric
constraints imposed by the adsorption sites. In the hexagonal close packed structure of
Au(l 11), the spacing between each gold atom is 5 Am} There are interstitial sites
located at voids formed by three gold atoms, 2.89 A apart from one another. There are
two types of these interstitial sites; those where a gold atom resides directly beneath the
void, and those in which there is no gold atom beneath. The nearest-neighbor spacing
between identical interstitial sites is 5 A. Based on our findings that indicate adsorbate
and substrate polarizabilities mediate the attractive interactions, we expect substrate
electron density to play an important role in determining the identities of the adsorption
sites. The most polarizable site on the Au surface will be the interstitial site with a gold
atom directly beneath the void, providing the adsorbate with at least partial access to
four gold atoms. The next most favorable site is the interstitial site without a gold atom
beneath. The least favorable site, in terms of available electron density is at the “top” of

each gold atom.

145
Adsorption to one type of site will occur at a higher rate than that for the other

types of sites and, while there may be an equilibrium between adsorption at each class of
sites, the most electron-rich sites will be occupied most frequently. With an inter-site
spacing of 5 A for a given type of site, we calculate that the available QCM surface has
~2.2 x 1014 sites. Using this site density, adsorbate molecular mass, and the adsorbate
radius extracted from its van der Waals volume, we calculate the density of adsorbate
monolayers expected from site-specific adsorption. These calculated site-specific
densities are shown in Table 6.5 along with the experimental and theoretical close-
packed densities. The experimental monolayer densities are much higher than the bulk
densities and are consistent with theoretical close packed densities. The site specific
densities in Table 6.5 are also consistent for the larger adsorbate molecules cyclohexane
and n-hexane, but are too small for the other adsorbate molecules. We have calculated
the site specific densities assuming that only the most preferred site is occupied. In the
case of methanol and acetonitrile, occupying only one type of site yields densities smaller
than those of the corresponding bulk liquid. If methanol, acetonitrile, and acetone
occupied both of the three-fold interstitial sites, the site-specific densities would be twice
those shown in Table 6.5 and consistent with the experimental adlayer densities. We
emphasize that, while our model is qualitative in nature, its predictions correspond
remarkably well to the experimental data, lending credence to the notion that variations
in substrate polarizability over molecular length scales determine the packing properties
of physisorbed adlayers.

Our data indicate that the hexanes adsorb to only one type of site while the other

adsorbates occupy both types of three—fold interstitial sites, consistent with the steric

146

constraints imposed by these comparatively large adsorbates. In addition, due to the
comparatively small polarizabilities of these aliphatic molecules, it is likely that more than
one CHx group per molecule is involved in interactions with the substrate, a condition
that would serve to obscure neighboring substrate sites.

We have applied the site-specific adsorption model to the gold substrate, and
now consider its application to 1-octadecanethiol/gold monolayers. The terminal CH3
groups of alkanethiol/gold monolayers have been shown to exhibit ordering that

“”41 Based on this structural feature, we

corresponds to the gold surface structure.
expect similar adsorbate alignment effects to be important for this non-polar monolayer,
although its influence may not be as pronounced as it is for the gold surface. We do not
know the exact structure of the MUD/gold systems, but we do know that the maximum
achievable density of the hydroxyl groups is limited by gold surface structure. The
hydroxyl groups possess lone pairs of electrons with which the adsorbates will interact.
Our results show the same behavior on the mixed MUD/l-dodecanethiol/gold systems as
the MUD/gold systems indicating similar structure. It is likely that during the formation
of the substrates that the MUD molecules adsorbed preferentially to the l-dodecanthiol

molecules thus giving essentially a MUD/gold system with less l-dodecanethiol content

than in solution.

147
6.4. Conclusions

We have studied the interactions of organic molecules in the vapor phase to gold,
polar thiol/gold monolayers and non-polar alkanethiol/gold monolayers. We find that the
interactions between the adsorbates and substrates are dominated by their polarizabilities.
We do not find evidence for chemical firnctionality-specific interactions in these systems,
and these data place limitations on the utility of thiol/gold systems for selective chemical
sensing applications. In addition, because the measurement scheme we have reported
here yields information on the density of the adsorbate layers, we have been able to
determine that the atomic- or molecular-scale organization of the substrate determines

the packing density and organization of the first adsorbate layer.

6.5. Literature Cited

1. R. G. Nuzzo, L. H. Dubois, D. L. Allara, J. Am. Chem. Soc., 1990, 112, 558.
2. L. H. Dubois, B. R. Zegarski, R. G. Nuzzo, J. Am. Chem. Soc., 1990, 112, 570.
3. R. G. Nuzzo, E. M. Korenic, L, H. Dubois, J. Chem. Phys., 1990, 93, 767.

4. C. E. D. Chidsey, D. N. Loiacono, Langmuir, 1990, 6, 682.

5. C. E. D. Chidsey, C. R. Bertozzi, T. M. Putvinski, A. M. Muscje, J. Am. Chem. Soc.,
1990, 112, 4301.

6. C. D. Bain, E. B. Troughton, Y. T. Tao, J. Evall, G. M. Whitesides, R. G. Nuzzo, J.
Am. Chem. Soc., 1989, 111, 321.

7. C. D. Bain, G. M. Whitesides, J. Am. Chem. Soc., 1989, 111, 7155.

8. E. B. Troughton, C. D. Bain, G. M. Whitesides, R. G. Nuzzo, D. L. Allara, M. D.
Porter, Langmuir, 1988, 4, 365.

\O

. M. D. Porter, T. B. Bright, D. L. Allara, C. E. D. Chidsey, J. Am. Chem. Soc., 1987,
109, 3559.

148

10. G. M. Whitesides, P. E. Laibinis, Langmuir, 1990, 6, 87.

11. N. Camillone III, C. E. D. Chidsey, G.-Y. Liu, T. M. Putvinski, G. C. Scoles, J.
Wang, J. Chem. Phys, 1991, 94, 8493.

12. C. E. D. Chidsey, G.-Y. Liu, J. Wang, Langmuir, 1990, 6, 1804.

13. S. R. Wasserman, G. M. Whitesides, I. M. Tidswell, B. M. Ocko, P. S. Pershan, J.
D. Axe, J. Am. Chem. Soc., 1989, 111, 5852.

14. C. A. Widrig, C. A. Alves, M. D. Porter, J. Am. Chem. Soc., 1991, 113, 2805.
15. G.-Y. Liu, M. B. Salmeron, Langmuir, 1994, 10, 367.

16. R. L. McCarley, Y.-T. Kim, A. J. Bard, J. Phys. Chem, 1993, 97, 211.

17. R. L. McCarley, D. J. Dunnaway, R. J. Willicut, R. J.; Langmuir, 1993, 9, 2775.
18. D. S. Karpovich, G. J. Blanchard, Langmuir, 1994, 10, 3315.

19. C. A. Widrig, Chinkap Chung, M. D. Porter, J. Electroanal. Chem, 1991, 310, 335.
20. C.-J. Zhong, M. D. Porter, J. Am. Chem. Soc., 1994, 116, 11616.

21. D. E. Weishaar, M. M. Walczak, M. D. Porter, Langmuir, 1993, 9, 323.

22. C. A. Alves, E. L. Smith, M. D. Porter, J. Am. Chem. Soc., 1992, 114, 1222.

23. DJ. Zhong, M. D. Porter, Anal. Chem, 1995, 67, 709A.

24. D. E. Weishaar, B. D. Lamp, M. D. Porter, J. Am. Chem. Soc., 1992, 114, 5860.

25. E. L. Smith, C. A. Alves, J. W. Anderegg, M. D. Porter, L. M. Siperko, Langmuir,
1992,8,2707.

26. E. L. Smith, M. D. Porter, J. Phys. Chem, 1993, 97, 8032.
27 . M. 0. Wolf, M. A. Fox, Langmuir, 1996, 12, 955.
28. J. M. Tour, L. Jones H, D. L. Pearson, J. J. S. Lamba, T. P. Burgin, G. M.

Whitesides, D. L. Allara, A. N. Parikh, S. V. Atre, J. Am. Chem. Soc., 1995, 117,
9529.

149

29. N. Lewis, M. Tarlov, J. Am. Chem. Soc., 1995, 117, 9574.
30. R. J. Willicut, R. L. McCarley, Langmuir, 1995, 11, 296.

31. K. Sinniah, J. Cheng, S. Terrataz, J. E. Reutt-Robey, C. J. Miller, J. Phys. Chem,
1995, 99, 14500.

32. R. L. Garrell, J. E. Chadwick, D. L. Severance, N. A. McDonald, D. C. Miles, J.
Am. Chem. Soc., 1995, 117, 11563.

33. J .-P. Bucher, L. Santesson, K. Kern, Langmuir, 1994, 10, 979.

34. E. Delamarche, B. Michel, H. Kang, Ch. Gerber, Langmuir, 1994, 10, 4103.
35. S.—C. Chang, I. Chao, Y.-T. Tao, J. Am. Chem. Soc., 1994, 116, 6792.

36. P. Fenter, A. Eberhardt, P. Eisenberger, Science, 1994, 266, 1216.

37. N. Camillone IH, P. Eisenberger, T. Y. B. Leung, P. Schwartz, G. Scoles, G. E.
Poirier, M. J. Tarlov, J. Chem. Phys, 1994, 101, 11031.

38. M. R. Anderson, M. Gatin, Langmuir, 1994, 10, 1638.
39. M. R. Anderson, M. N. Evaniak, M. Zhang, Langmuir, 1996, 12, 2327.
40. H. A. Biebuyck, C. D. Bain, G. M. Whitesides, Langmuir, 1994, 10, 1825.

41. H. Sellers, A. Ulman, Y. Shnidman, J. E. Eilers, J. Am. Chem. Soc., 1993, 115,
9389.

42. Y.-T. Kim, R. L. McCarley, A. J. Bard, Langmuir, 1993, 9, 743.

43. S. E. Creager, L. A. Hockett, G. K. Rowe, Langmuir, 1992, 8, 854.

44. D. M. Collard, M. A. Fox, Langmuir, 1991, 7, 1192.

45. P. Krysinski, R. V. Chamberlain, M. M. Majda, Langmuir, 1994, 10, 4286.

46. O. Chailapakul, L. Sun, C. Xu, R. M. Crooks, J. Am. Chem. Soc., 1993, 115,
12459.

47. L. Sun, R. M. Crooks, A. J. Ricco, Langmuir, 1993, 9, 1775.

150

48. R. C. Thomas, J. E. Houston, T. A. Michalske, R. M. Crooks, Science, 1993, 259,
1883.

49. B. L. Frey, D. G. Hanken, R. M. Corn, Langmuir, 1993, 9, 1815.

50. L. Sun, L. J. Kepley, R. M. Crooks, Langmuir, 1992, 8, 2101.

51. L. J. Kepley, R. M. Crooks, A. J. Ricco, Anal. Chem, 1992, 64, 3191.

52. LB. D. Green, M. T. McDermott, M. D. Porter, J. Phys. Chem; 1996, 100, 13342.
53. O. Chailapakul, R. M. Crooks, Langmuir, 1993, 9, 884.

54. O. Chailapakul, R. M. Crooks, Langmuir, 1995, 11, 1329.

55. G. Z. Sauerbrey, Z. Phys, 1959, 155, 206.

S6. C.-S. Lu, J. Vac. Sci. Tech, 1975, 12, 578.

57. J. Hlavay, G. G. Guilbault, Anal. Chem, 1977, 49, 1890.

58. F. L. McCrackin, E. Passaglia, R. R. Stromberg, H. L. Steinberg, J. Res. NBS,
1963, 67A, 363.

59. R. W. Collins, Y. T. Kim, Anal. Chem, 1990, 62, 887A.

60. H. G. Tompkins, A User’s Guide to Ellipsometry, Academic Press, 1993.

61. R. M. A. Azzam, N. M. Bashara, Ellipsometry and Polarized Light, Elsevier, 1977.
62. J. T. Edward, J. Chem. Ed, 1970, 47, 261.

63. H. M. Schessler, D. S. Karpovich, G. J. Blanchard, J. Am. Chem. Soc., in press.

64. P. W. Atkins, Physical Chemistry, 4‘“ Ed.. W. H. Freeman and Co., 1990, pp. 655-
660.

65. J. B. Schlenofl‘, M. Li, H. Ly, J. Am. Chem. Soc., 1995, 117, 12528.

66. N. Camillone III, C. E. D. Chidsey, G.-Y. Liu, G. Scoles, J. Chem. Phys, 1993, 98,
4234.

Chapter 7

Conclusions

When first discovered, self assembled monolayers appeared to posses ideal
attributes for use in chemical sensor and microdevice patterning applications. Once
made, their highly oriented structures were thought to be extremely stable and remain
unchanged. However, the QCM formation studies described in Chapter 2 indicate that
alkanethiol/gold monolayers are structurally dynamic in nature with the thiols exhibiting
mobility mediated by desorption and re-adsorption. The existence of structural mobility
limits the use of SAMs for applications such as patterning. The results reported in
Chapter 3 support fiirther the notion of structurally dynamic systems and provide a
thermodynamic perspective by determination of AH and AS for adsorption. The small
free energy for the adsorption of thiols onto gold was a result of the balance of favorable
enthalpic and countering entropic terms. These findings have a broader application,
beyond the scope of model alkanethiol/gold systems, since it is the balance of such forces
that detemiines the strength of interactions of molecules with surfaces. These studies
underscore the importance of entropy to any adsorption process

While alkanethiol mobility across gold may limit patterning applications of
thiol/gold SAMs, it has limited bearing on the average structure of the SAMs. The
average properties of model alkanethiol/gold monolayers can be used to study
fundamental interfacial phenomena. The average structure at the molecular level is
important in determining monolayer interactions with adsorbing molecules. A novel

measurement scheme presented in Chapter 5 revealed information about the dynamics of

151

152
the aliphatic chains. It was found that local environment formed by the aliphatic chains is

rigid on short, (picosecond to nanosecond) timescales. This apparent rigidity limits the
possibilities of elastic interactions with other molecules. Because of this rigidity and the
close packing of the aliphatic chains, it is also unlikely that molecules would be able to
penetrate into the monolayers, thus limiting interactions to the terminal areas of the
monolayers.

The interactions of gas phase adsorbate molecules with gold and thiol modified
gold substrates were reported in Chapter 6. It was found that methyl terminated
alkanethiol/gold monolayers did not interact as strongly with low molecular weight
organic molecules as do gold or hydroxyl terminated thiol/gold layers. The differences in
the strengths of interactions were correlated to the polarizabilities of substrates.
Chemical specificity on the part of the substrates toward interactions with the adsorbates
was not observed despite the similarity of some adsorbates to the substrates (alcohols
with -OH terminated layers and n-hexane with aliphatic monolayers). The lack of
chemical specificity is not surprising for the thiol/gold monolayers given their apparent
rigidity (see Chapter 5). Adsorbates would only be able to interact with terminal methyl
groups in the 1-octadecanethiol/gold monolayers, thus it is not surprising that n-hexane,
for instance, did not adsorb preferentially to other adsorbates. It is evident, based on
these studies, that true chemically selective substrates cannot be made without adding
more complexity to the interface. In addition, the adsorbates formed adlayers that were
more dense than their liquid counterparts. The densities of the adlayers were consistent
with site specific adsorption. Thus adlayer packing is determined by the organization of

the substrate.

153
It is evident that alkanethiol/gold monolayers are very usefiil, although for

applications different than originally thought. They are useful model systems for
studying adsorption processes and interfacial structural dynamics as demonstrated in this
thesis. The fiindamental knowledge gained from studying the formation of
alkanethiol/gold monolayers can be used to design more robust structures. The
reorientational dynamics measurements of the tethered pyrene within alkanethiol/gold
monolayers demonstrate a novel technique useful for probing motional dynamics within
interfacial systems. The studies of the interactions of gas phase organic adsorbates with
the gold and thiol modified gold served to demonstrate in—situ mass and thickness
measurements as well as provide fiindamental information about the interaction of

molecules with surfaces.

"llllllllllllll