. '3 .c ! ¢ 'uf
.i ‘VVVV VVWV ,EVV A? 3
9:61.413. 4.. 21‘.
'VV' V’I‘YIVV [1... 23;; 1
5:333"... 3 .t: .3
intuVV 33;: -- .
3.’ 'i :!r -
9 «3:. 1

‘ I v
.
.
..
-- .-m
..‘..._ -4.
‘-0 c 4-"- ‘3.
.— ..

     

  

  

 

 
  

     
 
     
   
   
 
  
    
 
    
  

    

13 V
V33 3;." ‘3'. p "3
N : J 3"": ;,J” V‘ |
“I I13“ H VI t ' ~
‘ I 3V3 [V1 ”V‘VIVRV‘P VL ”\‘V‘VVVJV' l W 1 h {I
3 e; t,I35.{VIVVVE«.ijf'VVthguéqkzr3VVVV"iV-.32V. - ‘3 I Vii; - Vi!

‘ ' .3 "V :§~"-' '- WM?“ "..’-M ‘ "NV. “- ' :‘
3V3 } 3: 4.3V ., ..z313433293.-.gfi’fiuinwzfzfigs‘.‘€735; fiVFVVVVVEEVVVV .‘tVVVV' VV '
.33: .r. V“ *3: .HV. *"-.VV"“.'?LV 3331323313,. VJVVVE'aVéVV‘Z’JJé3'i'V. “I“

)‘i' o 1.::;V.V.. V‘H’AVVTVJ “WV5VVWVVZVIV13VVVV'V(VV[VV'VV 4&8???" V 'M "V
.353“. 'i;,.;1'3§?‘:'.:‘3 53V,}:;.'V:.=.33.132339 tV‘flué i'EVVVVVV'sst *ViVVV‘ vii. 3 ~' 'L"
‘ j'1'.VVV§2,.'3';.V‘.*.‘[1!.T"}5“'.V::«13'. ‘Vfl VVVVVVV‘“VVV‘IV'MVVV'V'VVS5 ”11"! £th-
'.3 . .'.VV ; «IVV'V‘ V‘mV‘VV-U VWVV3‘1'VV'V"«"~'3V‘VJ-g£l ("LV’WV'VVVLCVV 5‘ h
'VVV’iiV :3, y1%.;{433319'313Mu: V.I'n=~;1}t:3f:rix "ti 9’5.- ‘z .Vi..:.3.]V't ‘ ¢
.'.. 3 l ‘u 33.43.34 .3.35;;VV3 u .31.? by“? V 4:» “HIV! 5: v. - - '
. - - 3 V 3V -- ' .3335? 'VVVVVU‘V"? 4:”

' .V ' V “ 1‘. n". t“ J." ' '1 - .‘M' Aggy?” V! k
' V ':;5‘~13i"!13:.5“'1
931:}:Vgti?1:2;:¥
..3f‘];iV1V§?:VSi
.zVV-isms. gm
MVP-:Vfrs” '3
133's: 5 I

   

'31 H"
1" ..‘:".i

tQ-Ivvhgv

  
  
  

 
 
  

V}? ..V.

   
         
  

  

 
 
     

.‘17
3' .1zi51fif"‘l.w .
,;. l V ‘ 1"HIIL H '3'. L
In. 6 c.3.3 ‘3”... .33“, .33. if. :.V|° { ..
.3 v .- .', 1’31, l fl 1 ‘ ‘ .. Vt'lb
:H' ,1 1V: ’ I V a '1'.‘3 3'? L ”Q" Q.
‘ " ~ «13:23:; ' WV" - 3.3 g; .g . .. = if... , ,.

I“ V! I .r'n'w 3":’3:j i.:;. '“PM1,?i.§3.‘3‘._'.'.3.,:3'
“V. ' VP; '1‘; HV‘ “'3 v I ' ' 'I'ETL‘IE‘F}.
LV‘T .hi', ': :Plh'lt VI, . .‘l W" ‘ L,
:V VI .A .‘q‘I:
VI." .VV ' . 5.3“.~l)“t
.31., “V "HI V' IN ’ "my

... V

V} [I‘Vfll‘ v: :I ”Jr'i
V ,.:V' ':tiVV LL? . V‘ 3|. l‘ ;{:;;i !{'Vlt 2.1.1., a‘I‘l H:’V1IV‘Q’I
V "'v-"~'¥""VV“~'"' z {V‘VV' '1' $.33. “4’ ”V If“; V9135!

y ‘HIV3V1V‘;VIIV1VV,‘ V f “IV-IV
. V V..IVVV: “II. ‘VVV-Vi' {33' )HV
V ..’; 31;“: ""(VVVVVVVMVV“3.13:".

. '7I..V1‘I;V iVl. V."V’3*'IVV
.'. ‘ . ..' VI 3. .3 _ F'. Vfiglzg' ‘.'.\‘ NH
VV'V ’ . w . V 3. 11333.33,” ..IVVVVVVHVfiQ. ‘1' W3?My].V!V{.3.3.V~|14453V.V “If?” . 1
In: VVVan. "' VVJ'MTIVI: .VVVVVV‘ ‘VVVVVV‘

V i‘ V' ' V"'| VMVVVV'V V. V V" 'I'V'V ‘V uV:VI'IV.'"’)V‘IV:":-'f~V‘V IVI V.
V ' 'n .. 'mu....VV....;.‘.~...-.i}.....a".V.~LI‘..V.‘m..'w 113V:3.1:.3.;;.33m.g.3.3.£3l3{-‘r}}3

 

TVS?“ (i ‘0

 

 

This is to certify that the

thesis entitled

Pkcfi'ot\&.¢\'€on\\em}cc_\ ) E\Q<-’rr0c\/\em{¢<\.\g
and. ng‘t-gcfl A“(\\\'g;5 Sfi-oches \33 -

- C of,
Rhodocm'tnq B Ev‘xkvggth) and (U:
DAVV‘E‘ 01‘5 “~3- Tlvx Glide E.\e.c.¥~rc Q.

S u r 3;- mv. 9A presented by

BuxfiA Dean “5‘9"

 

has been accepted towards fulfillment
of the requirements for

 

IV\ S . degree in Cherfix 5351

Major professor

~.

Date \\" 3 " '11

0-7 539

 

 

 

PHOTOELECTROCHEMICAL, ELECTROCHEMICAL AND
SURFACE ANALYSIS STUDIES OF
RHODAMINE B, ERYTHROSIN, AND VARIOUS OTHER DYES

AT TIN OXIDE ELECTRODE SURFACES

by

David Dean Hawn

A THESIS

Submitted to

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

MASTER OF SCIENCE

Department of Chemistry

1977

10.2»

ABSTRACT
PHOTOELECTROCHEMICAL, ELECTROCHEMICAL AND
SURFACE ANALYSIS STUDIES OF
RHODAMINE B, ERYTHROSIN, AND VARIOUS OTHER DYES
AT TIN OXIDE ELECTRODE SURFACES
BY

David Dean Hawn

The modification of semiconductive electrodes in order
to enhance their photocurrent-producing abilities has been
of recent interest. Because of the relatively large band-
gaps of these semiconductors (3.0-3.5 volts), their current
producing abilities are rather poor under illumination with
visible light. However, modification of these materials
with suitable dye materials, which absorb light in the
visible regions and have high molar absorptions, can greatly
increase the semiconductor's photoefficiency.

In order to investigate this phenomenon, we have
examined both multilayer dye configurations constructed
electrochemically and monolayer dye assemblies constructed
covalently.

For characterization of these surfaces using surface-
sensitive Spectros00pic techniques, such as ESCA, we have
resorted to the use of various tagged dye molecules.
Because most dye molecules used to date have contained only

carbon, nitrogen and oxygen atoms, ESCA analysis of these

David Dean Hawn
materials is of little interest, since these atoms are also
present as contaminants on the surface. However, the use
of halogenated dyes has allowed us to unambiguously quanti-
tate the dye concentrations on the modified electrode.
Results from this surface spectroscopic technique have led
us to believe that electrochemical adsorption of erythrosin
dye at silane—modified surfaces produces more than one form
of surface bound dye.

For covalent attachment, two procedures have been in-
vestigated. One is an amidization reaction, between the
carboxylic acid group of the dye, and an amine group on the
electrode surface, in the presence of dicyclohexylcarbodiimide.
The other involves the formation of a thiol linkage between
a mercapto group on the surface of the electrode and the
iodine substituents of the dye erythrosin.

Various electrochemical techniques have been employed
to characterize electrodes modified in this manner, as well
as the photocurrent producing abilities of both monolayer
and multilayer assemblies. Final results have led us to
believe that the covalent linkage plays more than a passive
role in photocurrent sensitization. Covalent attachment
does more than immobilize the dye on the surface. Photo-
currents at electrodes with low concentrations of covalent
attached dye consistently exceed photocurrents at electrodes

with several layers of adsorbed dye.

ACKNOWLEDGMENTS

I thank my wife, Patti, for her help and patience in
preparing this manuscript. I also thank my Mom and Dad for
their encouragement through the years.

I also acknowledge Dr. Neal Armstrong, Dr. Lynn Sousa,
and Dr. Michael Weaver for their many hours of helpful
discussion.

For her technical assistance in preparing this manu-

script, I wish to thank Dianne Contos.

ii

TABLE OF CONTENTS

SURFACES.

Chapter
I. HISTORY. . . . . . . . . . . . . .
A. INTRODUCTION . . . . . . . . .
B. ELECTRODE PREPARATION AND
CHARACTERIZATION . . . . . .
C. SILANE MODIFICATION. . . . . .
D. COVALENT ATTACHMENT TO VARIOUS
E. DYE MATERIALS. . . . . . . .
F. HISTORICAL CONCLUSION. . . . .
II. ESCA ANALYSIS. . . . . . . . . . .
III. EXPERIMENTAL SECTION . . . . . .
A. ELECTROCHEMISTRY . . . . . . .
B. PHOTOELECTROCHEMICAL MEASUREMENTS. . . .
C. ELECTROCHEMICAL CELLS. . .~. .
D. MATERIALS AND REAGENTS . . . .
E. AQUEOUS SOLUTIONS. . . . . . .
F. COVALENT ATTACHMENT OF DYES. .
l. Thiol Attachment . . . . .
2. Amide Attachment . . . . .
G. ESCA DATA. . . . . . . . . . .
H. DECONVOLUTION. . . . . . . . .
IV. RESULTS AND DISCUSSION . . . . . .

A.

INTRODUCTION . . . . . . . . .

iii

Page

ll
l4
19
20
22
26
27
30
31
34
35
35
37
37
38
39

40

B. ELECTROCHEMICAL AND ESCA ANALYSIS
OF MULTILAYER DYE ASSEMBLIES. . . .

C. ELECTROCHEMICAL CHARACTERIZATION OF
DYE ADSORBED SURFACES USING THE
ELECTROCHEMICALLY REVERSIBLE
FERRO/FERRICYANIDE COUPLE . . . .

D. COVALENT ATTACHMENT . . . . . . .

E. PHOTOCURRENT MEASUREMENT. . . . . .

F. CONCLUSION -- SUGGESTION FOR
FUTURE WORK . . . . . . . . . . . .

REFERENCES 0 I O O O O O O O O O O O O 0

iv

43

65
75

89

94

95

LIST OF TABLES

Table Page
1 Conditions for Silanization. . . . . . . . . . . 36
2 Iodine to Tin Ratios and Binding
Energies for Electrochemically and

Mechanically Adsorbed Dyes . . . . . . . . . . . 62

3 Electrochemical Behavior of Tin
Oxide Surfaces With Adsorbed Dye Layers. . . . . 71

4 Chronoamperometric Data for
Bonded Electrodes. . . . . . . . . . . . . . . . 84

Figure

LIST OF FIGURES

Photocurrent sensitization process for

a. metal, b. n-type, c. p-type semiconductor.

Cyclic Voltammagram of pH 4 buffer at clean

tin oxide electrode; scan rate = 11.5 mv/sec.

Covalent attachment to surfaces using
cyanuric chloride . . . . . . . . . . . . .
S-operational amplifier potentiostat. . . .
4-operationa1 amplifier potentiostat. . . .
Circular electrochemical cell

constructed of Lucite . . . . . . . . . . .
Electrochemical cell constructed

of Teflon or Lucite . . . . . . . . . . . .
Structures of various dye molecules;

a. Rhodamine B, b. Fluorescein,

c. Erythrosin . . . . . . . . . . . . . . .
UV-visible spectrum of dyes in pH’? buffer;

a. Fluorescein, b. Erythrosin . . . . . . .

vi

Page

10

16

28

29

32

33

41

42

Figure

10

11

12

13

14

15

16

Page
Cyclic voltammetric behavior of Erythrosin

dye on clean SnO2 at various concentrations;

a = 5.0 x 10‘4 M, b = 2.5 x 10'4 g,
c = 1.0 x 10‘4 g, d = 5.0 x 10‘5 g,
e = 5.0 x 10"4 M. . . . . . . . . . . . . . . . 44

Cyclic voltammetry of dyes at 2.5 x 10"4 M in

pH 7 buffer; a. Rhodamine B, b. Erythrosin. . . 46
Cyclic voltammetry of dye at pr-silane and

en—silane modified electrodes at various
concentrations. a. clean Sn02, 2.5 x 10‘4 M;

b. 5.0 x 10-5 g; c. 1.0 x 10’4 g;

a. 2.5 x 10"4 g; e. 5.0 x 10‘4 M;
f. 2.5 x 10"4 M on en-silane modified Sn02. . . 48
ESCA spectrum of adsorbed dyes on tin

oxide surfaces. . . . . . . . . . . . . . . . . 50
ESCA spectrum of adsorbed dyes on tin

oxide surfaces. . . . . . . . . . . . . . . . . 51
Cyclic voltammetry of Erythrosin at SH-

silane modified tin oxide; a = 5.0 x 10’5 M,

b = 1.0 x 10'4 g, c = 2.5 x 10-4 u,

d

5.0 x 10'4 M; scan rate = 11.5 mv/sec . . . 56
Cyclic voltammetry of various dyes on pr-
silane modified tin oxide. A11 dyes 5.0 x 10‘4 M

in pH 7 buffer; a. Fluorescein, b. Rhodamine B,

c. Eosin, d. Erythrosin. Scan rate = 11.5 mv/sec S8

vii

Figure

17

18

19

20

21

Page
Observed behavior for dye electrochemically
adsorbed on clean and silane-modified
SnOZ surfaces. . . . . . . . . . . . . . . . . . 60
I/Sn ratios for electrochemically
adsorbed dyes. . . . . . . . . . . . . . . . . . 64
Cyclic voltammagram of ferri/ferrocyanide on
unmodified Sn02 after electrochemical adsorption
of dye at various concentrations;
a=4.88x10"4 g, b= .98x10‘4 M,
c = 4.88 x 10'5 g. . . . . . . . . . . . . . . . 66
Cyclic voltammagrams of Ferrocyanide solution
at a-aminOprOpyl silane modified tin oxide
electrodes after electrochemical adsorption of
dyes at various concentrations; a = 4.88 x 10"4 M,
b=.98x10’4M,c=4.88x10"5M. . . . . .. 68
Cyclic voltammagrams of 1.0 mM Ferrocyanide
solution at mercaptopropylsilane modified tin
oxide electrodes before and after electro-
chemical adsorption of dyes at various
concentrations. a. SH-Sn02 electrode, b =
4.88 x 10'4 M, c = .98 x 10‘4 M. Scan rate =

11.5 mv/sec. . . . . . . . . . . . . . . . . . . 69

viii

Figure

22

23

24

25

26

27

28

29

Page
Chronoamperometric plot for SH-silane
modified tin oxide before and after electro-
chemical dye adsorption at various concentra-
tions; a = blank, b = 4.88 x 10-5 M,
c = .98 x 10"4 g, a = 2.44 x 10"4 g,
e = 4.88 x 10"4 g. . . . . . . . . . . . . . . . 74
Reaction steps for Thiol attachment. . . . . . . 76
UV-visible spectrum of dye attached to glass
by Thiol. a. dye—attached glass, b. glass
after stripping. . . . . . . . . . . . . . . . . 78
Reaction steps for Amide attachment. . . . . . . 79
Mechanism for Amide attachment using
dicyclohexylcarbodiimide . . . . . . . . . . . . 80
UV-visible spectrum of Erythrosin attached
to amidization on en-silane modified tin
oxide; a. dye attached, b. after stripping . . . 82
ESCA analysis of 3d3/2 - 3d5/2 bands for
mechanically adsorbed and covalently bound
dyes; a. standard, b. Amide attached, c. Thiol
attached dye . . . . . . . . . . . . . . . . . . 83
Cyclic voltammagrams of 1.0 mM ferrocyanide
in pH 4 buffer after various stages of
modification; a. clean SnOz, b. mercapto—
silane modified SnOz, c. dye attached to SnOZ

by Thiol linkage. Scan rate = 11.5 mv/sec . . . 85

ix

Figure

30

31

32

33

Page
Cyclic voltammagrams of 1.0 mM ferro-
cyanide in pH 4 buffer after various stages
of modification; a. en-silane modified SnOz,
b. dye attached to SnOz by amidization.
Scan rate = 11.5 mv/sec . . . . . . . . . . . . 86
Mott-Schottky plot for dye attached to
tin oxide by Thiol linkage. . . L . . . . . . . 88
Mott-Schottky plot for Amide attached dye
on en-silane modified SnOz. . . . . . . . . . . 90
Photocurrent vs. monolayers for electro-
chemically adsorbed and covalently attached

dyes (at 1.0 volts anodic). . . . . . . . . . . 92

CHAPTER I

HISTORY

A. Introduction

 

There has recently been great interest in the investi-
gation of nonconventional forms of energy, the most important
of which are nuclear and solar energy. Unfortunately, public
awareness of the degradation of the environment has branded
nuclear power a bad risk.

The sun, on the other hand, is one of the most promising
alternative energy sources. It offers considerable amounts
of power in the form of heat and light at an extremely low
cost. Unlike nuclear power, there are no problems with solid
waste disposal. There is no risk to health from radiation,
and no fission processes to control. Solar energy is not
without its problems, however. The biggest of these is the
conversion of radiation into some more useable form, such as
electricity. Green plants, for example, take light and carbon
diOxide and, with the aid of cholorphyll, convert light to
food and water. Maybe we can borrow some technology from
the plants to overcome this problem.

One such method of conversion is the solid-state silicon
solar cell, which utilizes the energy of the sun to produce
an electrical current. Another involves the generation of
electrical currents using a semiconductor electrode, at an
electrode-electrolyte interface. Such photovoltaic processes,
commonly called Becquerel effects after their discoverer,
offer an added advantage over strictly solid-state devices,
however. When photocurrents are produced by the action of

light at these electrode surfaces, it must be at the expense

2

3

of something at the electrode, or in the electrolyte, i.e.,
electrons. At the semiconductor electrode-electrolyte inter-
face, the oxidation of water to produce molecular oxygen is

a known phenomenon. This reaction is associated with anodic
processes. In a similar fashion, cathodic processes can be
used in the production of hydrogen by the reduction of water.
Therefore, photovoltaic processes can also result in the pro-
duction of combustible fuels.

Many investigators have observed photoeffects at various
semiconductor electrode materials, such as titanium dioxide
(TiOz) (1), zinc dioxide (ZnOz) (2,3) and tin dioxide (SnOz)
(4). It is generally observed that illumination of the semi-
conductors with radiation of bandgap energy produces electron-
hole pairs which can participate in electrochemical processes.
For example, at p-type semiconductors, electrons for cathodic
photocurrents result, while at n-type materials, such as
those listed above, holes, and anodic photocurrents result.
Tin oxide, which has a bandgap of about 3.5 volts, will
absorb light at about 300 nanometers or lower, generating
current flow. Titanium dioxide, on the other hand, with a
bandgap of about 3.05 volts, absorbs light around 380 nm.

The fact that both of these materials have such a large
bandgap makes them impractical as photoelectrochemical con-
version devices.

Sensitization or enhancement of photocurrents can be
accomplished if the electrode material is placed in contact

with a suitable dye material, either adsorbed to the electrode

surface or placed in the electrolyte. This observation has
generated many investigations since it was initially observed
by Memming and Kursten (5). In this instance, illumination
of the electrode-electrolyte interface with light absorbed

by the dye only is sufficient to generate current flow.

Figure 1 summarizes this dye sensitization process, for
metals, and both n-type and p-type semiconductors.

For a metal electrode, photocurrents can hardly occur,
owing to the high density of occupied states in the metal.

In this case, both the ground state SO of the dye, and the
lowest excited state 81 overlap with occupied energy bands in
the metal, resulting in a two-step process which does no more
than quench the excited dye molecule.

A completely different situation results in the case of
semiconductors, however. The existence of a bandgap, or for-
bidden region, between the conduction band and the valence
band permits a one-step process to occur, since it requires a
finite time for recombination of holes and electrons within
the semiconductor. Since an electron in the conduction hand
must traverse the forbidden region to reestablish electro-
neutrality in the semiconductor, it can only do so by passing
through an external circuit, doing work in the process. If
either the ground state 50' or the excited state 51 of the
dye falls within the forbidden region of the semiconductor,
no recombination step can occur after the oxidation or re-
duction of the dye, and a photocurrent should result. This

requires that the bandgap of the dye is less than the bandgap

of the semiconductor. Photochemical sensitization is indeed
the observed behavior for these types of electrodes. The
utilization of this effect to increase photocurrents has been
the entire object of this research. The use of various dye
materials, either adsorbed or covalently attached, will be of

significance in further energy research (6,7).

B. Electrode Preparation and Characterization

 

The principle substrate used in this research is chemi-
cally vapor-deposited polycrystalline tin oxide. During
recent years SnOz electrodes have been frequently employed as
optically transparent electrodes, or OTEs, for the spectro-
photometric study of chemical reactions at electrode-
electrolyte interfaces (8,9). Sn02 layers on glass or quartz
substrates show excellent optical transparency in the visible
wavelength region, as well as high electrical conductivity.

Most tin oxide surfaces are prepared, either privately
or commercially, by pyrolytic oxidation of various organo-tin
compounds. The normal preparative procedure is to mix tin
tetrachloride (SnCl4) or tetrabutyl tin with a volatile
solvent such as hexane or butylacetate. This mixture is
atomized with nitrogen or oxygen and the vapors blown onto a
substrate heated to 450 to 6000 C. Since pure tin oxide is
an insulator, the conductivity can be widely varied by adding
a dopant material, most commonly antimony or fluoride ion,
rendering the material an n-type semiconductor. Layers
ranging from 1000 to 5000 Angstroms thick are normally de-

posited on the substrate.

7

The physical and chemical properties of tin oxide have
been studied in detail (10,11), as well as the surface
characteristics (12).

Tin oxide is normally characterized by two parameters,
its flat band potential and its carrier density. The flat
band potential is the applied potential at which no banding
occurs in the space charge region (62). The carrier density,
which is related to the number of dOpant molecules present,
represents the number of majority carriers. In the case of
tin oxide, the majority carriers are electrons, so that this
material is an n-type semiconductor. Both of the above
parameters can be determined by experimental measurement of
the electrode capacitance as a function of applied potential,
as outlined by Gileadi and others (13,14,15). The electrode
capacitance, as measured in contact with an electrolyte,
results from two major contributions, the Helmholtz capaci-
tance (CH) from the solution and the space charge capacitance
(CSC) from the semiconductor. These two capacitances act in
series, so that the total capacitance, C, can be expressed as

l/C = l/CH + l/CSC
Investigations with other semiconductors have shown that
differences in electrode potential lead primarily to changes
in the potential drOp across the space charge region. There-
fore, in most cases, the total capacitance is essentially

determined by the Space charge value only, so that

l/C = l/Csc.

Calculation of the carrier density, no, using Poisson's

equation (14) gives rise to an equation of the form
1/c2 = 2/esono)(U - Ufb - kT/e)

Here, 6 and e are dielectric constants of the tin oxide and

o
of vacuum, e is the charge on an electron, and k is Boltzmann
constant. U is the applied potential. The unknown parameters

in this equation, n and the flat band potential Ube can be

0
calculated from a plot of l/C2 against the applied bias U,
commonly called a Mott-Schottky plot. Inspection of the
above equation reveals that the 310pe of the plot is in-
versely proportional to the carrier density, while at the x-
intercept, where l/C2 = 0, U equals Ube with a small correc-
tion term. The flat band potential Ufb is a particularly
important parameter, since it is the applied potential at
which no band bending occurs within the space charge region
of the semiconductor. Provided the carrier density of the
semiconductor is high, it is reasonably safe to assume that
the fermi level and the conduction band of the semiconductor
are within 0.1 volts of each other (62). Carrier densities
for polycrystalline tin oxide are normally in the range of
1019 to 1021 carriers/cm3, with a flat band potential of
approximately -l.0 volts as measured against a silver/silver
chloride reference (15).

The surface of SnOz is made up primarily of amphoteric

Sn-OH groups. The solution pH can influence the degree of

dissociation of these groups according to the relations

Sn-OH Sn-o‘ + 8*

A
<-

or
->

Sn-OH Sn+ + OH‘

.1
Thus, the solution pH can have a large effect on the flat
band potential, due to variations in potential drop across
the Helmholtz plane, but should have no effect on the carrier
density (16).

Linear sweep voltammetric studies of Sn02 surfaces show
three regions of electrochemical interest as shown in Figure
2 (17). In aqueous solutions, SnOZ electrodes show a large
rise in current production at potentials of 1.2-1.4 volts
anodic, as measured against a silver/silver chloride refer-
ence. In this region oxygen evolution occurs due to the
oxidation of water. At the opposite end, at potentials of
-0.2 to -0.5 volts, the reduction of water and the evolution
of hydrogen gas is observed. Although brief excursions to
and beyond the anodic limit do not disrupt the tin oxide
surface in any way, potentials beyond the cathodic limit
result in destruction of the conductive layer, which is re-
duced to metallic tin(l7).

In the intermediate region, only slight current passage
is observed, due to charging and discharging of the surface
space charge region. The quality of the tin oxide and the
solution pH strongly affect the magnitude of charging

current and the location of both the anodic and cathodic limits.

10

.umm\>e m._~ u mum; cmom mmvoeuompm muwxo cw» campo um xmwmzn a In to EmmeEEmupo> owpuzo .N whamwu

ed

.2

o4

 

_

_

 

 

20:38 Edna : In.

11

C. Silane Modification

 

The silanization of surface-active materials in order to
alter their physical or chemical properties is widely known.
Silanes are that class of compounds which consist of a silicon
atom at the center, attached to four organic groups of various
functionality. Most commonly, silanes are used to modify
surfaces of silica chromatrographic supports, such as Zipacs
or diatomaceous earths, in order to improve their behavior as
stationary phases. These silica supports, their chemical
modification, and chromatographic performance have received
great attention in the literature.

Various references in the area of both gas chromatography
and liquid chromatography have discussed the silane modifica-
tion of solid supports to improve their chromatographic
properties (18,19,20). An excellent review of these station—
ary phases, and their use as chromatographic columns has been
published by Locke (21).

The use of silanes for the chemical modification of
semiconductor surfaces, however, has only recently been of
interest. The wide variety of these silicon-based materials,
in terms of availability as various organofuctional deriva-
tives (23), have made them attractive for intermediates for
covalent attachment to these surfaces. Further, their highly
reactive nature, especially with oxide-type surfaces, and
excellent thermal and chemical stability, make them attrac-

tive choices for modification.

12

The reaction of an organosilane reagent with a surface
hydroxyl group, such as tin oxide or titanium dioxide, can be

represented by the following reaction (24):

-1 . ‘1 .
I? -— OH + XSiYZR -> :M--O——-SiYZR + HX

surface silane surface

Treatment of these derivatized surfaces with concentrated
nitric or hydrochloric acids at room temperature have little
effect on the surface. Over long periods of time, however,
hydrolysis of the surface groups is observed, with the subse-
quent removal of bonded silane material (24).

The known order of reactivity of organosilanes with
silica surfaces, such as those used in chromatography, is
X3SiR > XZSiR > XSiR3 (25). Most commonly, the reactive
leaving groups (X) are chloride (Cl), methoxy (MeO), or
ethoxy (EtO). For a given silane reagent, reactivity with a
surface increases in the series X=Cl > X=MeO > X=EtO.

Organosilanes with more than one reactive group (X)
have the possibility of binding to the surface at more than
one site. This does not necessarily mean that silanes with
two reactive groups must bind at two sites, however. The
fate of the remaining unbonded reactive groups is also of
some importance, especially in the formation of stable con-
ductive surface-bound phases. Polymeric formation can occur
at the surface in the presence of water (20,26), forming a

nonconductive surface layer of the form

13

R R
l l

-Si-O-Si-—

rag/aw

Polymerization can be avoided by the exclusion of water
during attachment procedures or by the use of silanes with
only one reactive group.

The reaction of methylchlorosilanes, either trimethyl-
chlorosilane or dimethyldichlorosilane, with tin oxide gels,
have been investigated by Harrison and Thornton, using infra-
red spectrometry (27, 28). It was observed that tin oxide
discs which had been evacuated by 320 K showed an intense
band due O-H stretching at 3600-2000 cm'l. After reaction
with a silane, however, the O-H band intensity decreased
drastically. The mechanical properties of tin oxide gels so
reacted also changed greatly, due to the reduction of inter-
particle hydrogen bonding. The most drastic changes occurred
using dimethyl dichlorosilane, where the discs became
extremely brittle.

The electrochemical behavior of SnO2 electrodes modified
with various silanes has been studied by Murray (29,35).
Replacement of surface hydroxyl groups (Sn-OH) with silane
groups showed little effect on the cyclic voltammetric behav-
ior, although the magnitude of the charging current was
reduced in the silane-modified case. Cyclic voltammetric
behavior toward the electrochemically reversible couple

ferri/ferrocyanide were also unchanged between the clean and

l4

silane-modified tin oxide surfaces. The authors did notice
a slight increase in the peak potential separation of the
(Fe2+/Fe3+) oxidation and the (Fe3+/Fe2+) reduction after
the surface had been modified, indicating a slight reduction
in the accessibility of the reactants to the electrode
surface.

Murray (64) has attempted various reaction conditions
ranging from reactions in pure organosilane, to reflux in a
two percent solution of the silane in dry solvent, in order
to optimize monolayer coverage. Where electron transfer from
a bound electrochemically active substance to the electrode
surface is desirable, the authors have found the monolayer
coverage of the silane at the surface is most desirable.

ESCA (Electron Spectroscopy for Chemical Analysis) analy-
sis showed significant variation in the tin oxide surfaces
before and after silanization. Reduction in the intensity of
the Sn 3d bands in the 500-480 eV binding energy region,
accompanied by the appearance of Si 2p transitions in the 110-
100 eV binding energy region were always observed for silane
modified SnO2 surfaces. This type of behavior would be
expected for the addition of overlayers of bonded material

(see ESCA analysis section).

D. Covalent Attachment to Various Surfaces

 

Although most electrode surfaces can be modified by ad-
sorption, it would be more desirable to permanently modify a

surface by covalent attachment. Methods for securely and

15

irreversibly anchoring molecules to surfaces could be utilized
to produce surfaces with unique and widely varying surface
properties.

The binding of proteins to highly porous surfaces was
first pioneered by H. Weetall (30). Prior to the development
of covalent attachment, proteins were immobilized on chromato-
graphic columns, where they were to be used for sequencing.
The experimental basis for sequencing, called Edman degrada-
tion, has also been carried out by attaching the protein to
a solid support (32). Solid supports used include poly-
styrene resins, and more recently, derivatized glasses.

Harper (34) accidentally discovered a method of attaching
acid-base indicators to porous glass while attempting to
activate their surfaces for the binding of proteins. In the
process of activation, a pale orange sintered glass funnel
was produced, from which the color could not be removed.
Washing with acids failed to remove the color, and only
turned the funnel reversibly red. These bound-type indi-
cators are claimed to have several advantages over soluble
indicators.

The mechanism for attachment is outlined in Figure 3.
Initially the surface is modified using a silane, most
commonly a-aminOprOpyltriethoxysilane. The alkylamine glass
produced is then coupled to cyanuric chloride by refluxing
in a chloroform solution, and to this the indicator, having
either an amino, phenolic, or alcohol functional group may

be added.

c: n I
I \’ \elc
— o — filliflz’al‘flz + N 'l'l —)
\(3/
0:
.CI
u c’
I / R—NH
- 0-Si(CH2)3Nll—c/ ‘u 2
I \~_c/ [H.011
‘m
nun
I
-e
- 0 - Si (0'12) NH-—-¢4;| Vfi '
\u-c’
‘08’

chloride.

16

+

a
5i(cuzbnnz

(EHH3Sin2)NH2 ___..

. Covalent attachment to surfaces using cyanuric

17

Harper used exclusively porous glass, since he noticed
little or no retention of the dye on smooth glass surfaces.
The fact that these porous glasses irreversibly adsorb dyes
indicates that a true covalent linkage may not always be
formed.

The bonding of an aniline azo dye to glass slides has
been achieved by first treating the surface with thionyl
chloride (65). Reaction of the treated surface with an
amine group of the dye produced a stable amide bond. The
immobilized dye absorption spectral characteristics closely
resembled those for the analogous solution forms of the dye.

The coverage of the dye, calculated as 1.1 x 10-10 2

moles/cm
(6.6 x1013molecules/cm2) is consistent with monolayer
coverage.

The covalent attachment of reversible electrode react-
ants to various electrode materials have only recently
received attention in the literature. Moses and Murray (35)
have bound 3,5-dinitrobenzoylchloride, which possesses two
electrochemically active nitro groups, to a ruthenium oxide
(RuOZ) electrode via 3-(2-aminoethylamino)prOpyltrimethoxy-
silane (en-silane). The coupling reaction also involved the
formation of an amide, with the loss of HCl. Cyclic
voltammagrams of surfaces thus modified showed behavior
similar to that for n-butyl-B, S—dinitrobenzamide in a
solution at a RuOZ electrode. Similar results have also

been achieved at platinum electrodes (36). Miller (37,38)

has produced a chiral electrode with interesting properties

18

by bonding an optically active amino acid to a carbon elec-
trode. (S)-(-)phenylalanine methyl ester was attached to
electrodes which had first been modified with thionyl
chloride, resulting in the formation of an amide bond. For
the study of the electrodes, the reduction of 4-acetyl-
pyridine was chosen, in order to verify the success of the
attachment. This ketone is reduced at low cathodic poten-
tials to an alcohol. When the reduction is carried out in
the presence of the optically active amino acid, an optically
active alcohol results. This behavior was observed at
chemically modified electrodes. These experiments demon-
strated the feasibility of achieving more selective electro-
chemical processes through electrochemical modification of
electrode surfaces.

Murray has also attached aminophenylporphyrins to
glassy carbon by a similar attachment procedure (66).
Cyclic voltammetric studies of the modified electrodes
showed electrochemical behavior consistent with that for the
porphyrin in nonaqueous solution. Metallation of the porphy-
rin ring with cobalt (II) produced cathodic reduction waves
consistent with the Co(II) - Co(I) reduction. A similar
attempt to bond iron porphyrins using dicyclohexylcarbodiimide
to tin oxide electrodes was inconclusive (67). Davis and
Murray observed that the porphyrin was irreversibly ad-
sorbed to the surface regardless of whether the surface
had been previously silanized or not. Attempts to charac-

terize these electrodes using ESCA spectroscopy were also

19

inconclusive, since the Fe 2p3/2 and Sn 3d3/2 bands occur
at almost the same binding energy.

Fujihira, Matsui, and Osa (39) have studied the double-
layer capacitance of ESCA behavior of tin oxide electrode
surfaces modified with p-nitrobenzoylchloride via an organo-
silane. ESCA spectra of the modified surfaces showed a
nitrogen is band consistent with the presence of a nitro
group at the surface. Cyclic voltammagrams verified a
cathodic redox process consistent with the known electro-
chemistry of the nitro group.

Double-layer capacitance measurements indicated little
difference in the slope of the Mott-Schottkey plots, indi-
cating little difference in the carrier density of the tin
oxide surface for the bonded or unbonded case. A significant
increase in the flatband potential toward more cathodic
values was observed. This change was attributed to an in—
crease in the potential drop across across the Helmholtz
layer, which would be expected upon the addition of a low-
conductivity organic group to the electrode surface. Recently
Fujihira (4) reported the covalent attachment of Rhodamine B
dye to a tin oxide electrode. The photocurrent action
spectrum of the dye-modified electrode mimicked the ultra-
violet-visible spectrum of the dye. Cyclic voltammetric

behavior or ESCA data were not reported for these electrodes.

E. Dye Materials

 

Fluorescein and erythrosin have been used for color

copied images in photography (74). Transfer plates for the

 

 

 

20

copied images were made of a dispersion consisting of ZnOz,
butadiene-vinyltoluene polymer, fluorescein, methylene blue,
and Rose Bengal. Harper (45) has produced insoluble indi-
cators covalently bound to glass surfaces. The dye and
carrier were bound via a silane coupling agent, using a
sulfonamide linkage. These dyes have also been investigated
for possible use as laser sources, using THF, ethanol, or
water as solvents (41). Ionov and Akimov (42) have used
erythrosin for the spectral sensitization of the photo-
electric effect in narrow-band semiconductors. Measurement
of the photoconduction of germanium, silicon, lead sulfide,
and selenium crystals coated with a thin film of the dye
showed a spectral sensitizing effect due to the dye, with a

very high efficiency.

F. Historical Conclusion

 

The study of tin oxide surfaces and their modification
has been presented. Although there has been much recent
research on these semiconductor surfaces and their silane
modification, the field of covalent attachment of dye mole-
cules is relatively new and unstudied. It is the goal of
this research to thoroughly study, by both electrochemical
and surface analysis means, dye modified surfaces. In this
way we may be able to learn more about the properties of
modified surfaces.

The photocurrent sensitizing effect of these and other
dyes either adsorbed, or more recently, covalently attached

to electrode surfaces, is also a well-studied phenomenon.

21

Little is known about the photoefficiency of such surfaces,
however. Further, we do not know how critical a covalent
linkage is to photocurrent production. The entire goal of

this research will be to answer these questions.

CHAPTER II

ESCA ANALYSIS

Among the many new instrumental techniques deve10ped in
recent years, electron spectroscopy for chemical analysis
(ESCA) is one of the most ideally suited for examination of
surfaces. It is sensitive to almost every element, and its
signal can be used as either a qualitative or quantitative
tool.

The ESCA technique was developed by Kai Siegbahn at
Uppsala University in Sweden. His early work is well docu-
mented in two books (68,69), which outline in detail the
principles involved.

The ESCA instrument is composed of five basic parts:

(1) a source, (2) high vacuum sample chamber, (3) electron
energy analyzer, (4) detector, and (5) readout systems.

Bombardment of the sample by high energy x-ray photons
causes the emission of electrons from the core and valence
bands of the bombarded sample. For ESCA analysis, these core
electrons are most important, because the energies with which
they leave the parent atom are well defined and specific to
the parent atom. The electrons emitted from the sample,
commonly called photoelectrons, leave the parent atom with
kinetic energies which are a function of the incident photon
energy, the energy required to cause their removal from the
parent atom, and the work function of the spectrophotometer.
While the energy required to remove the electron from its
parent atom varies from atom to atom, all other parameters
remain constant. Therefore, the kinetic energy which the

electron has when it reaches the analyzer is unique to a

23

24

particular atomic species. Further, these photoelectrons,
because of their relatively low kinetic energies, can escape
from only the top few monolayers of material. Normally, the
kinetic energies the photons have upon reaching the analyzer
are converted to binding energies, ranging from 0 to 1000
electron volts. ESCA analysis can be used to qualitate
atomic species present at a surface. Since the binding
energies are very specific to the environment of the atoms
in the sample, binding energies can also be used to determine
the oxidation state for various atoms at the surface.

Use of ESCA analysis to determine overlayer thickness on
a substrate is one of the more valuable pieces of information
this technique can supply. Bonding or adsorption of overlayers
of material on a surface causes a reduction of the ejected
photoelectron intensities from the underlying sample. This
decrease in intensity for a particular component can be used
to determine overlayer coverage. In this type of quantita-
tion work, however, it is necessary to know the mean free
path, or escape depth of electrons from the underlying sample.
These values vary from element to element, and also depend to
a small extent on the material composing the overlayer. These
values have been tabulated for several elements (71).

The appropriate equation for calculation of overlayer
thickness is

I/I = exp(-d/A)
clean

Both I and I represent the intensities of the native
clean

25

photoelectron before and after overlayer coverage, respec-
tively. A is the escape depth for the photoelectron through
the material, and d is the overlayer thickness. For the tin
3d5/2 photoelectrons, A is normally taken as 11 Angstroms (71).
Murray and coworker (64) have used this method with moderate
success to calculate overlayer coverages of silane bonded to
tin oxide.

Although values for the amount of material present at a
surface cannot be absolutely determined, these can be calcu-
lated relative to some element present and measureable at the
surface. For example, Armstrong and coworkers (72,75) have
determined relative amounts of oxide on tin oxide or titanium
oxide surfaces, by ratioing the oxygen to titanium or oxygen
to tin intensities. This method can then be used to determine
surface stoichiometry for a particular material.

Since the excitation source for ESCA, x-ray radiation,
does not affect all atoms the same, corrections must be made
for this occurrence. For a given number of photons striking
the surface, different numbers of photoelectrons are produced
for each element. Therefore, intensity ratios must be
corrected by multiplying the values by their relative photo-
ionization cross sections. Values for most common elements

have been tabulated by Scofield (55).

CHAPTER III

EXPERIMENTAL SECTION

..‘. L—g—-.——._.- .

A. Electrochemistry

 

All electrochemical measurements were performed on a
four or five-amplifier potentiostat of conventional design
(73). The circuit diagrams for both configurations are shown
in Figures 4 and 5. Each has external connections for
auxillary, reference and working electrodes, with the working
amplifier acting as either a voltage follower (Figure 4) or
as a current-to—voltage converter (Figure 5). In the second
case, various precision resistors could be substituted into
the feedback loop of the Operational amplifier for different
current sensitivities, ranging from lOuA/mv to l uA/mv out.
For low-frequency chopped photocurrent measurements, the
Operational amplifier could be capacitively damped to limit
noise contributions.

Double-throw switch S1 could be used to apply external

bias, as regulated by variable resistor R External inputs

1'
W1 and W2 could be used to apply various external waveforms.

A compensation circuit was also connected through single-
throw switch 82 for compensating for IR drop between working
and auxillary electrodes.

Double-layer capacitance measurements were made according
to the method of Gileadi (13). This method involves appli-
cation of a sine wave signal superimposed on an applied bias
potential. In this manner, electrode capacitance can be
investigated at various externally applied potentials. A
small amplitude sine wave was used (less than 20 mV peak to

peak) in a 300-800 Hz frequency range, in order to limit

faradaic contributions.
27

’)(’

 

 

“——

 

.2;

 

.pmumowucmpoa xmwchQEm Pocowomtmaonm

 

\N.
11

’9”

 

 

““‘

t”’

ull(<(((IIIIIJ

 

 

41:1

 

 

 

 

N; .3

 

 

 

.e mxzmem

29

.pmpmowacmuoa ngCPFaEm chomumxmao-v .m wxzmma

 

 

 

 

u \
"” l —‘

““‘l

 

>””
J‘““

 

 

 

 

 

 

 

Ir,”

 

“

 

 

30

As long as the impedance measured at the electrode is
mainly capacitive, the response to a small triangular wave is
a square wave whose amplitude is proportional to the capaci-
tance. Any IR drop through the solution is easily compensated
for. The working amplifier is acting as a differentiator,
with the electrode capacitance acting as a capacitor in a
circuit. After correct compensation had been achieved, using
a triangle wave, a sine wave was substituted. Capacitance
measurements were performed in pH 7 phosphate buffer of
approximately 0.1 ionic strength. Current output was
measured by using a phase-sensitive lock-in amplifier
(Princeton Applied Research, Model 126).

Cyclic voltammetric studies were performed with the
same potentiostat, using an externally applied ramp function
of varying slope and range. Current-voltage output was re-
corded with a Houston Series 1100 X-Y recorder. For
electrochemical adsorption of various dyes at the electrode,
the charge passed was measured using a digital coulommeter
built in this department. Chronoamperometric data were
recorded on a Heath strip chart recorder. For all measure-
ments, the potential was stepped to +1.0 volts measured

against a Ag+/AgCl reference.

B. Photoelectrochemical Measurements

 

The light source used for photocurrent generation con-
sisted of an Electropowerpacs power supply and Oriel universal

lamp housing, with a 450 or 1000 watt Hanovia Zenon arc lamp.

31

The lamp output was passed through a 6-inch water filter to
remove infrared wavelengths, followed by an infrared inter-
ference filter. A set of Oriel interference filters was used
to isolate the wavelength desired. A variable frequency
chOpper was used at 13 Hertz for these measurements. The
potentiostated photocurrent was measured using the lock-in
amplifier previously mentioned, which was triggered by the
chopper output.

Photocurrent measurements were performed at a bias of
+1.0 volts against a silver/silver chloride reference. The
electrolyte used was pH 4 buffer which had been made 0.1 M

in pottasium oxalate.

C. Electrochemical Cells

 

The new principle types of cells used are shown in
Figures 6 and 7. The cell of circular design was constructed
of lucite, with two glass stopcocks for filling and emptying.
The reference electrode consisted of a glass tube with a 8-
inch thirsty quartz frit (Corning Glass Works, Corning, New
York) sealed at one end with heat-shrinkable tubing. A
silver wire deposited with silver chloride, dipping into a
saturated potassium chloride solution, acted as an Ag+/AgCl
reference. A short length of platinum wire, coiled and
sealed in the cell, served as auxillary electrode. The
working electrode was sealed in the end of the cell by a
Viton O-ring, and held by a lucite face plate and screws.

Exposed electrode area, calculated from Chronoamperometric

32

Eu m.m

 

 

 

 

.mbwozq co umuoaxumcoo __mu quwemguoxuumpm Lm—suewu .o mxam_a

0

a";

d-

 

 

 

Eu—

 

 

 

 

*0.—

n

 

 

.L

a:

33

.muwozL to coped» do cmpuagumcoo Fpmo peevemzuoguumpm .N ox:m_m

Eu a.m

 

 

 

 

r-.. 4"1
'6’ o ..A
. _

c .

g .

o o
c u
n .

. .

o g
. -

n L
I

5:

 

 

 

 

 

 

 

 

Eum L
A .
0.-..--lo 1.... v
gas .I. “Ju "
4 .... n. .
“sznv " nu " u so A
\\\Q g . .
s u c a n
S o g u u
R . u u u
a. .
. n .r A .L. .1
a .5: .
\\.l\v
m

34

data, was 0.636 cm2. Electrical contact was made by a
circular brass contact pressed against the electrode surface
concentric with the exposed electrode area. This insured
that, for tin oxide layers on nonconducting bases, the series
resistance was as low as possible to facilitate capacitance
measurements at higher frequencies. Such a contact configura-
tion also minimized capacitance dispersion due to distributed
resistance of the thin film.

The rectangular cell shown was constructed of either
lucite or teflon. The teflon cell was used where aqueous dye
solutions were investigated. In this way, the cell could be
disassesmbled and cleaned in a suitable solvent. The auxillary
electrode compartment was separated from the working elec-
trode compartment by a fine glass frit. With this config-
uration, any products produced at the auxillary electrode

could be kept from contaminating the working electrode.

D. Materials and Reagents

 

Vapor-deposited tin oxide on glass was obtained from
Pittsburgh Plate Glass (Pittsburgh, Pennsylvania). The films,
of approximately 4-7000.A thickness, were either fluoride or
antimony dOped. The plates were cut into squares approxi-
mately 2 cm on a side prior to use.

Toluene (Drake) was dried in magnesium sulfate and
distilled from potassium metal prior to use. Tetrahydrofuran
(Mallinkrodt) was dried by the same procedure. Both were

distilled under nitrogen atmosphere.

35

3-(2-aminoethylamino)propyltriethoxysilane, mercapto-
propyltrimethoxysilane (Dow Corning), and u—aminOpropyl-
trimethoxysilane (ICN) were not purified prior to use. All
silanes were stored in a freezer when not in use. Dyes
(Eastman Kodak) were used as is. Erythrosin was dried in a
vacuum oven at 35-400 C for twenty-four hours prior to use
in the amide reaction.

Dicyclohexylcarbodiimide (DCC) (ICN) was used as ob-

tained from the manufacturer.

E. Aqueous Solutions

 

All solutions were prepared in doubly-distilled deionized
water and deoxygenated by bubbling thoroughly with nitrogen
gas.

Either pH4 or pH 7 buffers were prepared using potas-
sium hydrogenpthalate (KHP) and nitric acid, or potassium
dihydrogen phosphate and sodium hydroxide. 1.0 millimolar
ferrocyanide solutions were prepared in pH 4 buffer using

potassium ferrocylencide (K4Fe(CN);6H20).

F. Covalent Attachment of Dyes

 

The Snoz surfaces were prepared for silanization by first
washing the electrodes in a detergent solution, followed by
washings in ethanol and distilled water, using an ultrasonic
cleaner. The surface was activated prior to reaction by re-
fluxing in 0.1 N HNO3 for two hours. This step assured that
all tin oxide surface sites were protonated, allowing more

complete silane coverage. Finally, the electrodes were

36

rinsed in ethanol, followed by drying in vacuum.

The conditions for silanization are shown in Table 1.
Special care was taken to exclude air during the reactions
as much as possible. In this way, cross-polymerization of

the silane groups could be held to a minimum.

Table 1. Conditions for Silanization

 

 

 

Silane Solution Reaction Conditions
-aminopr0pyltrimeth- 5% in dry 12 hours reflux

oxysilane Toluene

3-(2-aminoethylamino)- 1% in dry % hour reflux

prOpyltriethoxysilane Toluene

mercaptoprOpyltrimeth- 2-5% in dry 1 hour, no reflux

oxysilane

 

 

After silanization, the electrodes were washed in portions

of the dry toluene solvent, and refluxed for one to two hours
to remove excess unbonded reagent. The degree of surface
coverage was extremely difficult to control, and varied
between batches of tin oxide.

1. Thiol Attachment. Tin oxide electrodes which had

 

been modified with the mercaptosilane (SH-SnOz) were placed

in a pH 7 buffer solution (0.1 ionic strength) which was

1.0 x 10'3 M in erythrosin dye. The reaction was allowed to
proceed twelve hours at room temperature, with continuous
stirring, under nitrogen atmosphere. Excess dye was removed
from the surface by extended period of extraction in a sohxlet

apparatus using either ethanol, acetone or water as solvent.

37

2. Amide Attachment. For the amidization reaction,

 

several solvents were investigated, the most satisfactory of
which was tetrahydrofuran. Carbon tetrachloride and nitro-
benzene are other widely used solvents for amidization reac-
tions, but neither worked successfully for the attachment
reaction. The relative reaction rates for these three sol-
vents are 34:18:l (CC14: Nitrobenzene: THF) (61).

A 1.0 mM solution of erythrosin dye, previously dried,
was prepared in dry THF. The THF/dye solution was cooled to
00 C in an ice bath, after which a stoichiometric amount of
DCC, which was dissolved 10% w/w in THF, was added dropwise.
The solution was allowed to stand fifteen minutes, then the
amine-modified electrodes were added. The reaction was
allowed to proceed in the dark, with continuous stirring, for
forty-eight hours. The modified electrodes were extracted
afterwards to remove excess dye. The covalently modified
electrodes could be stored for extended periods of time in
the dark, but showed almost complete bleaching of the dye

after two weeks in light conditions.

G. ESCA Data

 

ESCA (Electron Spectrosc0py for Chemical Analysis)
spectra were obtained on a Physical Electronics Model 548
instrument, using a magnesium anode source operating at 400
watts. Vacuum was maintained at 10'8 torr or lower. For
signal averaging, the instrument was interfaced to a Data
General Nova-800 computer with 32k core memory. For overall

spectra, pass energy was set at lOOeV, using a scan rate of

38

7 eV sec-1. High resolution spectra of the tin 3d5/2 and
3d3/2 bands were obtained using a 20 eV window, from 500 to
480 eV, with a pass energy of 25 eV. In the case of the

iodine 3d5/2 and 3d bands, a 50 eV window (650-600 eV)

3/2
with 50 eV pass energy was used. These pass energies were

selected to optimize the signal-to-noise ratio.

H. Deconvolution

 

ESCA spectra were deconvoluted into their individual
components by methods similar to those previously published
(72). The spectra were deconvoluted by computer simulation
of the total spectrum using the suspected components present.
The simulation program used the following parameters: (1)
the slope of a linear background, (2) the binding energy of
each component, (3) the full width at half maximum of each
component (FWHM), (4) the peak heights of each component,
and (5) the percent Gaussian and Laurentian contributions to
the total line shape. Parameter 3 was evaluated from stan-
dards made of pressed pellets of erythrosin in sodium
chloride or SnO2 powders. Parameter 5 was typically 50% for
the pass energies used. All reported binding energies were
corrected for charging effects by referencing to the Sn 3d5/2
band at 486.2 eV (12). Simulations were performed on a PDP-
11 computer coupled to a Vector General CRT display. Hard
c0pies of spectra could be obtained after simulation on a

Versatek plotter.

A. Introduction

 

The structures of rhodamine B, erythrosin and fluorescein
are shown in Figure 8. Note that fluorescein dye is simply
the unhalogenated form of erythrosin. At pH = 7, rhodamine
B is a cation, since the phenolic group has a pK of 9 (7).
Similar behavior would be expected for both fluorescein and
erythrosin at this pH.

Varying the pH only slightly affects the wavelength of
maximum absorbance in erythrosin, indicating that the
carboxylic acid group has little to do with the light-
absorbing process. Comparing the wavelengths of maximum
absorbance for erythrosin and fluorescein, however, indicates
that the substitution of iodine groups on the conjugated
ring system of the dye strongly affects the absorption pro—
cess. As Figure 9 shows, erythrosin at pH = 7 has a Amax at
536 nm, whereas fluorescein has a maximum absorbance at 489 nm.

This indicates that, for erythrosin, the ground state of
the dye is separated from the lowest excited state by 2.32
electron volts. This separation is 0.22 eV less than fluor-
escein, where 2.54 eV separates the ground and lowest excited
states. This trend is consistent with expected behavior,
since the addition of an electron-donating group such as
iodine should make it easier to excite electrons in the ring
structure from ground to the lowest excited state.

The principle motivation for using a halogenated dye
was to allow quantitation of the modified surfaces using

ESCA analysis. Although all the dyes previously mentioned

40

A. Introduction

 

The structures of rhodamine B, erythrosin and fluorescein
are shown in Figure 8. Note that fluorescein dye is simply
the unhalogenated form of erythrosin. At pH = 7, rhodamine
B is a cation, since the phenolic group has a pK of 9 (7).
Similar behavior would be expected for both fluorescein and
erythrosin at this pH.

Varying the pH only slightly affects the wavelength of
maximum absorbance in erythrosin, indicating that the
carboxylic acid group has little to do with the light-
absorbing process. Comparing the wavelengths of maximum
absorbance for erythrosin and fluorescein, however, indicates
that the substitution of iodine groups on the conjugated
ring system of the dye strongly affects the absorption pro-
cess. As Figure 9 shows, erythrosin at pH = 7 has a Amax at
536 nm, whereas fluorescein has a maximum absorbance at 489 nm.

This indicates that, for erythrosin, the ground state of
the dye is separated from the lowest excited state by 2.32
electron volts. This separation is 0.22 eV less than fluor—
escein, where 2.54 eV separates the ground and lowest excited
states. This trend is consistent with expected behavior,
since the addition of an electron—donating group such as
iodine should make it easier to excite electrons in the ring
structure from ground to the lowest excited state.

The principle motivation for using a halogenated dye
was to allow quantitation of the modified surfaces using

ESCA analysis. Although all the dyes previously mentioned

40

41

a. (02H5)2N 0 +u(c2 "5) 2 3|—

 

 

Figure 8. Structures of various dye molecules; a. Rhodamine B,
b. Fluorescein, c. Erythrosin.

42

0mm

.cwmoggpxgm .n .cwmommxoapa .m "edeman n :a c_ maze we Eagwomam mFDwmw>n>=

com

1F

omm

J

u

 

com

 

 

om:

d

co:
3 a

1111111111.

.a missed

 

BONVHHOSHV

 

42

.CWmoxgpxgm .a .cwoummxozpm .6 mimemzn N :a cw memo to Esxpumam m_nwmm>->:

Re So omm 8m

F . 1H q a

j

 

 

 

om:

u

8:

d u

l/II

.m 8:3...

 

EDNVHUDSHV

 

43

contain carbon, nitrogen and oxygenations, these are not
satisfactory for ESCA analysis, since they can also be present
at the surface as contaminants resulting from air or solution
exposure. Therefore, by providing an unambiguous tag atom,
with a favorably high molecular cross section, surface

analysis could be employed.

B. Electrochemical and ESCA Analysis of Multilayer Dye

 

Assemblies

 

The electrochemical behavior of erythrosin is interesting
and has led to a rather unique way of constructing multilayer
dye assemblies at the electrode surface. The cycle voltam-
metric behavior of the dye on a clean tin oxide electrode
surface is shown in Figure 10. The concentrations of dye in

pH = 7 buffer range from 5.0 x 10"4

M (Figure 10a) to

5.0 x 10-5 M (Figure 10d). The cyclic shows that the dye has
a peak potential on clean tin oxide of .875 volts, as measured
against a silver/silver chloride reference. Figure lOe

shows a 5.0 x 10‘4 M dye solution, also on a clean electrode
surface. The observed cathodic shift in the case is due to
the quality of the tin oxide, its doping level, and film
thickness. It is not uncommon to notice a variation between
batches of tin oxide made by the same manufacturer. In order
to make the results presented in this section more reliable,
we were careful to use the same quality of electrode material

throughout a particular experiment, or where comparisons

were to be made between separate experiments.

44

.z atop x o.m u m " muop x o.m u u 2 «top x o.~ n o z wuop x m.m u a z euop x o.m u a

mmcowpmxpcmocou mzowxm> um Noam campo co mat :wmoxcuxgm Co Low>mgma owxumssmupo> owpuxu .o_ mg:m_m
. as 2 3..
q _

 

<1 om m

 

<1 o_

 

++ ++

<1 om

 

45

The charge passed in each cyclic voltammagram
increases with increased dye concentration, as is apparent
in the Figure. There is little effect on the peak potential
with varying concentration, although at lower concentrations
much of the observed background slope is due to the proximity
of the anodic limit of Sn02 to the peak potential.

This electrochemical phenomenon can be attributed to a
simple adsorption process at the electrode (48), similar to
the behavior of chloride at SnOZ surfaces (4). There is
no precedent to indicate that the dye is undergoing an oxida-
tion process, although this may also be occurring. A chemi-
sorbed phase at the electrode surface may be formed by this
faradaic process. The observed behavior might be due to
charging and discharging of position surface states (tin
atoms), caused by a dipole-dipole or dipole-ion interaction
with the dye (56,63). This configuration would be more or
less equivalent to the charging of a capacitor. The reverse
scan does not produce a reduction wave anywhere in the full
potential region of SnOz.

Comparing the peak potentials of erythrosin and
rhodamine B at clean tin oxide, there is a cathodic shift of
190 millivolts for erythrosin, as shown in Figure 11.

Rinsing the unmodified tin oxide electrodes with either
water or ethanol after electrochemical adsorption of the dye
quickly removed any remaining visible traces, although ESCA
data indicated that a tightly bound form of the dye was still

present at the electrode surface of about 10 to 20 monolayers.

46

£325»; .a .m 82.5.8051 .6 26:3 A :1 E a -2 x m.~ pm mm? ..B 1.58532; 0288 .3 9531

v
> 2 ms 3..

 

_ H _

<1 om

 

<1 S

 

 

 

47

Silanization of the tin oxide surfaces with either
o-aminopropyltrimethoxysilane(pr-silane) or 3-(2-aminoethyl-
amino)~pr0pyltriethoxysilane(en-silane) caused the peak
potential for the erythrosin adsorption to shift approximately
50 millivolts cathodic relative to clean tin oxide. However,
examination of the shape of the anodic wave indicated that
there was a considerable broadening of the wave for the silane-
modified case. Further, the amount of charge passed was about
30% greater for the silane modified case. This indicates
that the presence of the terminal amino group at the tin oxide
surface had some catalytic effect on the adsorption or oxida-
tion process.

Varying the solution dye concentration for the pr-silane
modified electrodes also produced some interesting informa-
tion in the cyclic voltammetry, as can be seen in Figure 12.
As the dye concentration increased from 5.0 x 10'5 M to
5.0 x 10‘4 133, there was an apparent cathodic shift of the
peak potential toward more cathodic values. Further examina-
tion of these voltammagrams, however, indicated that the
anodic process appeared to be made up of two anodic waves,
which varied in their relative magnitude with varying con-
centration. At lower concentration, the more anodic of the
two processes dominated, whereas at an intermediate
(2.5 x 10"4 M) dye concentration, the magnitude of both fara-
daic waves appeared to be equal. Comparing the shapes of the
anodic waves for clean and pr-silane modified electrodes

tended to confirm this.

48

naeem 8:68 ousted .6 a You x 3 .t a me. x as .o a 11: x as .e
m: euop x 0.1 .o m: mtoP x o.m .o "2 stop x m.m . ocm :ompu .o .mcowpoxucmucou
meowxo> um mwootuompm umwmwuoe mcopwmuem use meopwmtxa be ~16 yo soquEeopo> owpuxu .N_ «gamma

> One oé

A _ _
7 L.

 

 

..DU‘QJ

 

 

 

49

Unlike clean (unmodified) tin oxide electrodes, rinsing
the pr-silane electrodes after dye adsorption failed to re-
move all visible traces of the dye, so that the electrodes
showed a distinct red area where the dye had been adsorbed.
Extended periods of soaking in either ethanol or water also
failed to remove this tightly held layer from the surface.

Figure 12f also shows a cyclic voltammagram of the dye,
at 2.5 x 10‘4 M, at a tin oxide electrode modified with the
en-silane. The peak potential, the amount of charge passed,
and the peak shape appear to be identical to that for the
pr—silane modified electrodes at the same concentration
(Figure 12d).

A series of electrodes which had been modified with pr—
silane, en-silane, and a clean electrode were carried through
the electrochemical adsorption, washed in ethanol and water,
and transferred to the sample chamber for ESCA analysis at
10’9 torr or lower. The resulting iodine high resolution
spectra, in the 650-600 eV binding energy region, are shown
in Figures 13 and 14, along with their computer-deconvoluted
components. The binding energy for the iodine 3d5/2 bands
were corrected for charge shift by referencing to the Sn
3d5/2 band at 486.2 eV (12).

Figure 13a shows the high resolution iodine bands for
the dye electrochemically adsorbed, at 2.5 x 10""4 M, on a
clean tin oxide electrode. The larger peak at 621 eV binding
energy (I 3d5/2) as well as the accompanying band at 632 eV

(I 3d3/2) are due to the iodine covalently attached to the

 

 

 

 

/

6‘0

/

29.4035.
150 SEC

 

 

 

 

 

 

counts
300 sec

 

IL

L _;—
620 610 B.E. (eV)

 

Figure l3. ESCA spectrum of adsorbed dyes on tin oxide surfaces.

5]

counts
l000 sec

 

 

0‘

counts
1000 -§EE;-

 

ME)

 

 

 

 
   

 

J l l I
640 630 620 610 B.E. (eV)

Figure 14. ESCA spectrum of adsorbed dyes on tin oxide surfaces.

52

dye rings. This was verified by running the spectrum of an
electrode to which the dye had been mechanically adsorbed.

After silanization of the surface with pr-silane, and
subsequent electrochemical dye adsorption, both I 3d bands
showed splits into doublets, the magnitude of which depended
on the concentration at which the dye was adsorbed. At the
lowest dye concentration (5.0 x 10'5 M), as shown in Figure
13b, the predominant peak for the iodine transition is loca-
ted at 618 eV binding energy, with a second band at 621 eV.
Varying the dye concentration in solution, which subsequently
increased the total charge passed, resulted in an increased
intensity for the higher binding energy 621 eV band, while
the lower binding energy peak was greatly diminished. At
the highest concentration used (5.0 x 10'4 M), as shown in
Figure 14b, the lower binding energy band is only a minor
component in the spectrum, so that its relative magnitude
can only be approximated.

Spectra shown in Figures 14a and 14c also merit further
consideration. These show high resolution spectra for the
pr-silane and en-silane modified tin oxide electrodes,
respectively. Both surfaces have undergone electrochemical
adsorption of the dye at the same concentration (2.5 x 10"4 M).
Measurement of the charge passed indicated that the total
charge in both cases was identical, at least within experi-
mental error. However, in the case of the en-silane elec-
trode, the magnitude of the 618 eV band is much larger than

for that band on the pr-silane modified electrode.

53

There are two possible explanations which can be sup-
ported by the data. The first assumes different molecular
surroundings of the dye at the electrode surface and of the
dye layers on tOp of this. The second assumes the produc-
tion of molecular iodine or iodide ion at the electrode
surface, covered with overlayers of adsorbed dye.

If it is assumed that the dye is electrochemically ad-
sorbed in a layer-on-to-of-layer fashion, results like those
mentioned above might be expected. Those dye molecules at
or close to the chemically modified electrode surface would
be expected to be in a different molecular environment than
those molecules making up the successive overlayers. There-
fore, it would be expected to see a small shift in binding
energies for the molecules in different environments. How-
ever, shifts as large as observed on the modified surfaces
(3 eV) would not be expected. Further, a similar phenomenon
is not observed for the case of unmodified (clean) tin oxide,
as might be expected based on the same above discussion.

The second explanation, the formation of either 12 or
I“ at the electrode surface, is based on several previous
observations. The reported binding energies for iodine (12)
and Iodide (I’) are 618.0 eV, exactly as observed on the pr-
SnO2 and en-SnO2 electrodes. However, the presence of I2 or
I' would require the displacement of an iodine molecule for
the dye, as might occur by neucleOphilic attack.

It is known that a thiol can displace iodine bound to

either aromatic or aliphatic molecules, forming a carbon-

54

sulfur bond, with the resulting liberation of HI (50,51).
Nitrogen, in the form of an amine, is also a good nucleo-
phile; therefore, it is conceivable that the same reaction
could occur. Further, there is some precedent for the
electrochemical displacement of iodine by nitrogen (52,53).
Miller and Hoffman (54) have observed this displacement of
iodine from alkyl iodides, at platinum, via a carbonium ion

intermediate:

-———————9
R-I R+ + 1/2 12

R+ + CH3CN ——o R-N = c+-CH3
_ _ +- =
R N — c CH3 H20 ——9

CH3-C-NH-R
O

This reaction proceeds at potentials near the anodic limit
for platinum, +1.9 to 2.1 volts, as measured against a
saturated calomel reference. Further, the reaction is not
observed for alkyl bromides or alkyl chlorides. For conju-
gated systems where resonance stabilization can occur, such
a reaction is possible. A possible mechanism might be:

(-e‘)
R - I ' R+ + 1/2 12
R+ + :NHZR'-(S) ——+ R-NHR'-(S)

(R = dye, (S) = substrate)

55

In order to test this proposal, a number of experiments
were performed, which were aimed at either proving or dis-
proving it.

To test the effect on the peak potential using a
different neucleophilic group, the same electrochemical ad-
sorption was investigated using mercaptoprOpyltrimethoxy-
silane-modified SnO2 (SH-silane). Figure 15 shows a series
of cyclic voltammagrams, done at various dye concentrations,
on mercaptosilane-modified tin oxide. At high concentrations
(5.0 x 10"4 M), the anodic peak appeared again to be made up
of 13K) distinct processes occurring at potentials near to
each other. Estimating that the peak potential for the more
anodic process is at .825 volts vs. Ag+ / AgCl, we see that
this potential is very near than for pr-SnO2 electrodes
(Vpeak==0.810 volts: Figure 12). Varying the concentrations
of these two peaks, however, does not vary the relative
intensities of these two processes, as occurred in the pr-
silane modified electrode case. However, the obvious
cathodic shift would be expected in the presence of a better
electron-doner, which the thiol is. Hence, it should be
easier to produce the carbonium ion intermediate in the
proximity of the electrode surface, because of a reduction
in the activation energy. ESCA analysis for an SH-silane
electrode with the electrochemically adsorbed erythrosin

4

(1.0 x 10' M) was inconclusive. The appearance of two peaks

for either the iodine 3d3/2 or 3d bands could not be

5/2
detected due to poor signal~to-noise ratio caused by a

z m S x e. m- a sets :5 Stress 22.5-:

> ed

.omm\>e m. FFu 3m.» :Sm mm «A: x o.m no e- x m.~ n o z 73 x o F
m pm 50 Eguztm Fo StumEEeuFo> uFFuxu

m6 0.?

.mF 95m:

 

_

56

++++

<1 om

 

 

 

 

57

rapidly deteriorating anode.

The next experiment performed involved the investigation
of the unhalogenated form of erythrosin, fluorescein (see
Figure 8), and eosin, a tetrabrominated form of fluorescein.
These dyes were investigated at pr-silane, SH-silane and clean
tin oxide electrodes. The cyclic voltammetry of these dyes
on pr-silane modified electrodes are shown in Figure 16.

On pr-silane modified electrodes, rhodamine B no longer
has an anodic peak, contrary to its behavior on clean tin
oxide. Further, fluorescein (Figure 16a), shows only a
slight increase in background current, which could be caused
by IR drop across the SnOZ and across the cell. Eosin, how-
ever, under identical conditions of linear potential scan,
does show an anodic peak at 0.895 volts as measured against
a Silver/silver chloride reference. This is a shift of 80
millivolts more anodic than erythrosin, which is shown in
Figure 14d. For eosin, the total charge passed is also 40%
less than for erythrosin at the same concentration.

The behavior is identical to that expected. Bromine,
being a poorer leaving group than iodide, would cause an
anodic shift relative to iodine. It should require more
energetic conditions to produce a carbonium ion where bromide
is a substituent. 'On the basis of the electrochemical and
ESCA data presented, as well as visual observations, the
possibility of neucleophilic attack, with the loss of an

iodine molecule, is conceivably possible.

58

.umm\>s m.Frlu out; zoom

.cFmoccuxgm .u .cFm0m .u .m mcFEmuocm .n .chommxand .m ”commas F :1 :F 2 vqu x o.m

 

maze FF< .mquo eFF omFFFooe mcmFFm-ta so most maoFto> we StumEEopFo> oFFuzu .oF mtsza
> 0.0 3 3+
4 F q
<1 S o

+ + + +

 

 

 

59

Figure 17 represents the speculated behavior for deposi-
tion of erythrosin on both clean and pr-silane modified tin
oxide electrodes. For clean tin oxide surfaces, the first
layer of deposited dye is in the form of the native dye.

ESCA analysis verify this to be the case, which is indicated
as Type I erythrosin in the associated Figure. At silane-
modified electrodes, however, the first layers of adsorbed
dye are of a form different than native erythrosin, indicated
as Type II. Successive adsorbed overlayers are considered to
be Type I. We assumed that the Type II dye was deposited at
the electrode surface, since increasing the charge passed

on the adsorption step (and hence the number of adsorbed
monolayers), caused such a drastic reduction in ESCA signal
intensity for this 618 eV binding energy component. The
above model, however, may be somewhat misleading, as it
assumes that continuous layers of either the Type I or Type
II dye are deposited. It is equally possible, due to the
surface irregularities of the vapor deposited SnO2 layer,
that the same phonomenon may be occurring in separate islands
on the electrode surface.

Assuming that each erythrosin molecule lies flat on the
surface, we calculated that each dye molecule should occupy
90 square angstroms. Therefore, there should be approxi-
mately 1.1 x 10'14 molecules/cmz. Based on a one-electron
process for the adsorption, and an electrode area of 0.636
cmz, 11.3 microcoulombs of charge should be passed for every

monolayer of adsorbed dye. Assuming further that all

60

.8845; Noam 35:89.15me EB
:83 :0 $9.83 128.528.5qu 98 to» 3.2.2.3 3238 FF 83m:

81me

    

Neem 8588::

61

adsorbed layers of dye remain on the surface (not likely),
the total number of adsorbed monolayers can be calculated
knowing the charge passed. The results of such calculations
are shown in Table 2.

For the deconvolved spectra, the iodine—to—tin ratios
for the iodine 3d5/2 and sin 3d5/2 bands are also shown.

All ratios are corrected for the differences between the
relative cross sections for both elements (55). Those
entries labelled with either (1) or (2) indicate the appro-
priate iodine-to-tin ratios for the particular binding
energy components.

The values reported in the Table are also compared to
those values for a sample of physically adsorbed erythrosin,
and an electrochemically adsorbed sample of erythrosin on
SH-silane modified SnOZ.

Accompanying Table 2 is Figure 18, which shows the _
equivalent monolayers of adsorbed dye plotted against the
particular component iodine-to-tin ratios and total iodine-
to-tin ratios. The apparent gradual increase in iodine-to-
tin ratio for the 618.0 eV binding energy band also indicated
that the relative amounts of Type II dye produced at the
electrode surface remained almost constant, even though the

number of monolayers adsorbed increased rapidly.

62

 

E New tome E 0.3
E New :5 E :6
tom mam ._,._ 112 x Z n 16E
m>ooe mo esom
E New 3% E 84
E NSH as; E 85
was 0.8. a 4-2 x 3 u 3er
o>ooo mo meow
E New. News E as
E New eée E a:
e: Z: . .1 TS x as u E:
Nocm\mcoFFm-ca co
emotomuo xFFouFemcootuumFm
Ne H come is Tom a is x RN u 3.13
Noam comFu co
omotomoe SFFooFEmguotuumFm
F>mv sucmcm cm H mtmonocoz mFQEom
o meFocFm o 2\ z mpmeFxoxaa<

mesa twosomn< FFFooFcogomz new FFquFemcootuumFu to» mmFmtmcm meFucFm use monmm cFF on mcFqu .m anmF

63

ANFV.>o N.owe ea eeea N\mem em oo oooeaeoeoea

Ammv.mmocmcmFeFo conumm-mmoto tonumFos :zocx to; omaumttouo

 

 

 

N.o.H m.ome eo.o 6:6 Noam eaopo eo
coco mmmF emotomoo zFFoomeca mxc
as H 3% Ed me .2 TS x as u EB
Noem\oeepem-=m :6
emotomuo xFFooFEmcuocpomFm
ANV N.o.u e.ome Amy a.mm
E NSH oée E 4;:
m.ae m.am .m e-o1 x m.~ u onev
Nocm\mcoFFm-em co
emotomoo xFFouFEmzootuumFm
1E EOHNFS 3v T:
11E N.o.w e.mFo 11v ee.a
e.em m.ae .1 .OF x o.m u onev
Noem\meoFquta :o
omoxomom xFquFEmcootuomFm
F>mv motmcm cm H mxmonocoz «Foamm
o mcFucFm o 2\ z mumeFxoxaa<

 

 

.1.e.oeoov N o_aeF

 

64

0 total I/Sn ratio

A 62l.0 eV binding energy I/Sn

 

 

 

 

mL I 6l8.0 eV binding energy I/Sn
/ l
701 / /‘
’ /
60" / /
a //
E
531’ en silanee / /
D //
L0- / /
e/ /
31" Blank SnOZO , /
//
20L ///
/ e
/ /
10'- / / "‘—__.L
/:Y z’.’
5%; 1" . .
10 20 3) ’40 50

Figure 18.

EQUIVALENT MOLAYERS

I/Sn ratios for electrochemically adsorbed dyes.

65

C. Electrochemical Characterization of Dye Adsorbed Surfaces
Using the Electrochemically Reversible Ferro/Ferricyanide
Couple

For the determination of effective electrode areas and
to determine how electron transfer between the electrode and
the solution is affected by overlayer formation, the ferro-
cyanide (Fe(CN)63‘ oxidation was studied. The kinetics of
the ferri/ferrocyanide oxidation at SnOZ electrodes are well
understood (56). The solutions used were pH 4 buffer, which
were 1.0 millimolar in potassium ferrocyanide. Electrode
areas were determined chronoamperometrically using the

Cottrell equation:

it”2 = nFAc’(D/1r)l/2

The diffusion coefficient for ferricyanide ion is given as

6.30 x 10'6 cmz/sec (56). Rearranging this equation to solve

for A, we arrive at the prOper form of the equation for

determining the electrode area:
A = (it1/2)/nFC - (fl/D)l/2

The symbols shown have the usual meanings:

n = number of electrons involved in the oxidation
process per molecule oxidized,

F = Faraday constant,

C = solution concentration, expressed in moles/cm3.

Figure 19 shows a series of cyclic voltammagrams on
clean tin oxide electrodes which have been previously treated
by electrochemical adsorption of erythrosin at varying con-

centrations. All electrodes were rinsed briefly with

66

.1 -2 x 8.4 u o

..a 73 x mm. 0 a .8 «L: x wmé n m mmcoFumxucmocou 30?? FM «Am ..6 5.51.183

FSEmgootpomFm .13: ~95 omEFooEc: co mchguotcmtEmw .3 5638233, 8:98

> 06 m6 QT
.

J 1

   

 

 

 

 

 

 

2 85a:

67

distilled water after the electrochemical treatment, in
order to remove any loosely held dye.

There is little effect on the peak potential separation
of the redox couple with varying concentration of adsorbed
dye, as is apparent from the Figure. There is apparently
little retention of the dye at the electrode surface since
there is almost no increase in the peak potential separation
(AEp = Epeak anodic - Epeak cathodic) with increasing number
of monolayers of adsorbed dye.

Figure 20 shows a series of cyclic voltammagrams of
ferrocyanide on pr-silane modified electrodes after under-
going dye adsorption at various concentrations. Figure 20c,
with the lowest concentration of adsorbed dye, shows little
difference in the peak potential separation from that for a
pr-silane electrode before dye adsorption. At the highest
adsorbed dye concentration, shown in Figure 20a, there is
significant reduction of the peaks for either half of the
redox couple, but a significant increase in background
charging current. This behavior indicates that the elec-
trode is largely blocked off to electron transfer.

A similar phenomenon is observed for SH-silane modified
electrodes, as shown in Figure 21. However, in this case
there is little concentration effect, unlike that observed
for pr—silane electrodes. Even at the very lowest concentra-
tion investigated (1.0 x 10’5 M), both the oxidation and
reduction waves were severely depressed, accompanied by a

significant increase in background current, presumably due

68

.m 4.2 x 8.4 u o 1:1: x ma. u a .2 wt: x was ... 6 353228.66
maoFxo> um mesa mo :on romeo FouFEwcooxp mFm Embed monoxpomFm mquo :Fu

 

 

 

 

 

 

 

 

 

 

SIZE: 2.25 anotaochuo pm 5.533. .4135??th .3 meotmEEBFg 0:98 .8 2.5 F...
1+
\ U
l; unnnnuuulllu o
<1 oe iii
a
<1 ooF

+

69

.. .u .omm\>e m.FF u out;
zoom .2 eloF x mm. u o .2 equ x mm.e u o .wootuomFm Noemuzm .o .meoFumtucwocou
maoFto> be mesa Fo :onatomoo quFEmzoocuqum tween one mcowmn mwuoguumFm mcho :Fu
umFFFooe mcoFFmanotaopaootme Fe :onzFom moFcoxuotcmm.mE o.F to maetmoeemuFo> uFFuxu

. owe ms 3..
q

.FN erased

 

 

 

 

 

70

to double-layer charging effects. However, there are two
distinct breaks in the cyclic voltammagram at all concentra-
tions, indicating that the redox process is still occurring
at the electrode surface.

Determination of the effective electrode area for each
type of modified surface also provided some interesting in-
formation. A summary of the data is given in Table 3. Also
provided in this Table are values for the charge passed on
each electrochemical dye adsorption, along with the concen-
tration of the dye in solution for which the electrochemical
adsorption was performed. AEpeak' the peak potential separ-
ations for the redox couple ferri/ferrocyanide are also
given. Values of this particular parameter are also given
before the adsorption step. Since the Chronoamperometric
data did not give the correct values for electrode area due
to cell configuration, all values were scaled to the known
geometric area, 0.636 cmz.

For unmodified tin oxide, Chronoamperometric data
showed only a 15% reduction in effective area between the
clean surface and that modified by adsorption at the lowest
concentration (5.0 x 10'5 M). After this initial reduction,
only a small decrease was observed with increased concentra-
tion. An apparent reduction in electrode area is also
observed in the case of pr-silane modified electrodes after
the dye adsorption. Similar to the behavior observed on
clean tin oxide, there is only a small decrease in electrode

area between the pr-silane electrode surface and the dye

71

 

0000 000.0 0.000 0.01 x 0.0 0000 000110-00
0041 000.0 0.000 0.01 x 0.0 0000 000110-00
001 000.0 0.001 11.01 x 0.1 0000 000110-00
mNF wwm.o m.moF mqu x o.m Noam memFFm-ca
00 010.0 .---- 1100101 0 0000 000110-10
001 000.0 0.010 .0.01 x 0.0 0000 00016
001 000.0 0.000 0.01 x 0.0 0000 00016
011 000.0 0.001 0.01 x 0.1 0000 00016
001 100.0 0.00 0-01 x 0.0 0000 00016
001 0000.0 .---. 1100101 0 0000 00016
1001 1000 10001 0000 1moeowmmwmuo1ev 10m01 010000

 

 

.000001 was concomo< zqu mmuemcsm oeFxo :FF mo soF>0=mo FaoFewsuoxaqum .m mFaoF

.00cmmssmuFo> uFFuzu 0000 0000000000 FeFuewpoa 0000 0000000000

.0sz0> 000080000 0:00: 1058 omo.ov 0000 chumEomm uomcgou czocx o» umqum 0maFe>0

 

 

72

 

0001 011.0 0.0001 0-01 x 0.0 0000 000110-00
0001 001.0 0.000 0-01 x 0.0 0000 000110-00
0001 101.0 0.000 0-01 x 0.1 0000 000110-00
0001 100.0 0.000 0-01 x 0.0 0000 000110-00
00 0000.0 .---. 1100100 0 0000 000110-00
001 010.0 0.001 0-01 x 0.0 0000 000110-00
1.001 10018 to< Fmgowmwwmuo F15 10mg 0 Fasmm

 

 

.10.00000 0 01000

73

adsorbed surface (5.0 x 10"5 M). At the highest dye concen-
tration, however, there is an almost 40% reduction in the
area, which is accompanied by an increase in the peak poten-
tial separation for the redox couple.

For SH-silane modified tin oxide, the reduction in
electrode area with increased dye concentration for the ad-
sorption step is much more dramatic. For these electrodes,
the quantity of charge passed for the dye adsorption step is
at least 80% greater than for either the clean tin oxide or
pr-silane modified surface. The AEpeak values remain almost
unaffected.

Figure 22 shows a plot of it8 against time for SH-
silane electrodes after dye adsorption at various concentra-
tions. Extrapolation of each series of points back to zero
time gives the approximate value for area calculations.

For the SH-silane electrodes with dye electrochemically ad-
sorbed at low concentrations, there is only a small reduction
in the time = 0 intercept. At higher concentrations, how-
ever, the it* values approach a lower limit.

This limiting behavior is also observed for pr-silane
electrodes. Examination of Table 3 indicates that there is
a sharp reduction in electrode area after about 250 micro-
coulombs of charge are passed. Based on earlier calcula-
tions, this represents about twenty equivalent monolayers of
adsorbed material. Doubling the charge passed, and hence
the number of adsorbed monolayers, has little effect on the
electrode area after about the first twenty monolayers of dye

have been adsorbed.

74

l .0 0-010 00.0 n 0 .
. 0 0-010 00.0 n 0 .1020 u 0 1
. ocuotpomFm 00000 0

a 0-01 x 00.0 n 0
00 000000 qux

 

 

00000000000000 0=oFt0> 00 0000100000
0 :00 00000005 000F00-xm 000 qua 600005000150ocotcu .00 001000
10.81% 00 00 . 00 00 0
1 1 0 0 0
I. Funnel.” I-l I I
0 alluuh-nu-wunuunlll
11.nu .qu.1nu1.Lnu ii. .I: .Iu
U|
0.0
“a
7x:
. VA
.- - M
II."
III. II‘
nil. .
l. I.- 1......
l-l... ul- \-
III II. III .I‘ .
.1; ul- i|-..|i o QH
‘ all!
a \- Ila-.1
‘.\ ‘

9mm

75

The apparent reduction in electrode area results in
part by blocking of the electrode surface to electron
transfer (56,57). The electrode area may be being blocked
in part by islands of dye isolated on the surface. By exami-
nation of the equation for calculation of the area, it is
apparent that a reduction in current values (i) will result
in a decrease in area. The reduction in current can be
accounted for by an increase in charging current for the
electrode double layer, or by consumption of current in a

second oxidative process at the electrode surface.

D. Covalent Attachment

 

Two methods of covalent attachment were investigated,
one of which is novel to the covalent modification of semi-
conductor surfaces. The other, more conventional method,
involves the formation of an amide bond.

The thiol attachment has been widely employed in the
biochemical sciences for protein linkage or modification (50,
51). The kinetics of the thiol reaction are widely known.
Figure 23 summarizes the reaction steps involved. The reac-
tion proceeds rapidly at room temperature, in aqueous solu-
tions at or near pH = 7. Chaiken and Smith (50) observed
the kinetics of the reactions of the sulfhydroyl group of
papain with chloroacetic acid, and found that the reaction

is second order, proceeding by the following reaction:

k
ESH —-L—9 ES" + H+ 2 : ESR

THIOL
’2

l

m

gn-O-Si (CH2)3-S 0

Figure 23.

rSn-O-Sl(CH2)3SH + I

76

n-O + (MeO)3Si(CH2)3SH ——-—-«>

 

HO I

/

Reaction steps for Thiol
attachment.

77

The rate constant for kl was found to be largest at pH = 5,
while k2 reached a maximum at pH = 6.

The dye-modified electrodes showed good retention of
the dye, even after long periods of extraction in various
solvents. Clean glass or tin oxide electrodes which had not
been silane-modified quickly lost all evidence of dye reten-
tion upon extraction, however. Figure 24 shows a UV-Visible
spectrum of the dye attached to clean glass, which was run
using only the unmodified glass as a reference. The upper
spectrum shows the dye-modified glass after being extracted
twelve hours in ethanol. Calculation of the number of mole-
cules per square centimeter, assuming a molar absorptivity
of 130,000 M'l cm'l, gives approximately 4.0 x 1014 molecules-
cm‘l. This number is about four times what would be expected
for monolayer coverage, if the molecules were assumed to be
lying flat on the surface. The second spectrum in Figure 24
shows the same piece of glass after a brief stripping with
concetrated nitric acid. Note that the dye peak has com-
pletely disappeared.

A schematic for the amide attachment is shown in
Figure 25. The reaction, commonly used in peptide synthesis
and coupling reactions (58), produces an amide linkage with
the resulting liberation of carbon dioxide. For the attach-
ment, either en-silane electrodes or pr-silane electrodes
were used. The accepted mechanism for the reaction using
DCC, is shown in Figure 26 (59,60,61).

Like the mercapto-attached dye, electrodes modified in

78

.000000000

.00 000010

 

00100 000Fm .n .000Fm 00000010-000 .0 .FoFgF an 000F0 o1 umzompum 010 we 50000000 0F010F>u>0
05:10
000 000 80 000 80
1 1 1 _ 1

 
  

00100 0000000000 1.0

 

 

79

AMIDE

(Me0)3Si(CH2)3NHCH2CH2NH2

/
\

‘Q$$§9&%§§§;

 

n-O + ————>
(Me0)3Si(CH2)3NH2
a _ R occ
;Sn-O-S1(CH2)3NHCH2CH2NH2 + HO-C-R
(dye)

‘QQSS5

a 0
ll

Figure 25. Reaction steps for Amide attachment.

80

O~=c=~-O —»

(DCC)
(dye)

O~=c=iO 4‘1
<:>—~ O ~2—

 

”-6-6%
z-z

o
R-g-NH-R'-(S) + O—y—g—g—O
II o a

Figure 26. Mechanism for Amide attachment using dicyclohexyl-
carbodiimide.

81

this manner could be extracted for extended periods of time
without extensive loss of color. However, electrodes or
glass which had not been previously silane-modified also
retained small amounts of dye after the reaction, which could
not be extracted. Therefore, it cannot conclusively be
stated that this is a true covalent attachment. Rather, it
may simply be the hydrophilic tendency of the dye to irrevers—
ibly adsorb to the electrode surface in a nonaqueous solu-
tion. There seems to be no evidence indicating the attach-
ment should occur at unmodified tin oxide surfaces. The UV-
Visible spectrum of a dye-attached tin oxide electrode is
shown in Figure 27, both before (a) and after (b) a brief
stripping with concentrated nitric acid. Note the similarity
of spectrum (a) to that shown in Figure 21 and to the solu—
tion spectra shown in Figure 9, which confirms the presence
of bound dye.

ESCA analysis of both forms of the bound dye verify the
presence of dye at the electrode surfaces. There is little
difference in the peak shapes for either the physically
adsorbed dye or either of the methods of covalent attach-
ment (Figure 28). Calculation of the I/Sn ratios for both
cases gave .594 for mercapto attachment and approximately
1.0 for amide attached. Note the absence of two components
in either iodine 3d3/2 or 3d5/2 bands, as was observed in
the case of electrochemically adsorbed dyes on pr-silane
modified electrodes. There is no apparent iodine or iodide

at the mercapto-attached tin oxide surface, as would be

82

.mcwaaweum memm .5 .umgumuum mat .m "muwxo :1“ umwwwuos
mam—1m-cm co :owumN1uwam o» nonempum cwmoggusgm mo Eagpomam m131m1>u>= .NN meamwa

es.

8m 8m 8m 8:
1 1 1 1

 

 

GHOSHV

 

3 .E:
mocoagoma< 1o.

 

 

 

83

 

 

 

 

 

a
counts
100 _§EE_—
b
counts
100 sec
’5
z
c
counts
‘00 sec
5 5th: 3% 15 3.5. (eV)
Figure 28.

ESCA analysis of I 3d3/ 2 - 3d5<2 bands for
mechanically adsorbed and cova ently bound dyes;
a. standard, b. Amide attached, c. Thiolattached
dye

84

expected due to this method of attachment, indicating that
most of the iodine liberated is easily extracted from the
electrode surface.

Ferro/ferricyanide cyclic voltammagrams of the surface
showed little effect on the electrode area, indicating that
.most of the surface was accessible to electron transfer.
Examples of cyclic voltammagrams on tin oxide electrodes
before and after dye attachment are given in Figures 29 and
30. Comparison of these results to those shown for electro-
chemically adsorbed dyes indicates little effect on the
surface. Chronoamperometric data also showed little differ-
ence in the electrode area, again verifying that the surface
area had been only slightly affected. However, the trends
observed for both types of attachment are important. These

results are summarized in Table 4.

Table 4. Chronoamperometric Data for Bonded Electrodes

 

 

 

Sample it3 (at t=0) Aeff (cm2)a
Clean SnOz 7.45 x 10-5 .545
su-Sno2 7.70 x 10’5 .564
Dye - SH-Sn02 8.00 x 10-5 .586
en-Sno2 ~ 8.50 x 10“5 .622
Dye - en-SnOz 8.29 x 10-5 .607

 

 

aUsing D = 6.50 x 10'5 cmZ/sec. (56)

85

.omm\>e m.11 u mum; zoom .mmmxcmp Powgp >9 Noam ou umcumpum use .u
.Nocm umwmwcoe mcmp1mopamucms .a .NOcm :mw1o .m “comumopmwuos to mommum

 

 

 

 

 

 

 

 

m=o1gm> cmuem commas e In :1 mupcmxoocgmm :5 o.1 do msmemmEEmupo> owposu .mm mgzmwa
> 0.0 m.o Osw+
\q q. .
rk/ “‘ u
/\/ V a
sPII.
<1 om
a

 

+

86

 

10 11A
.4.
a
10 uA
+
b
J 1 __
+0.5 0.0V

Figure 30. Cyclic voltammagrams of 1.0 mfl_ferrocyanide in pH 4
buffer after various stages of modification;
8. en-silane modified SnOz, b. dye attached to Sn02
by amidization. Scan rate = 11.5 mv/sec.

87

In each case, the electrode area increased slightly after the
electrodes were treated with the silane. After dye attach-
ment, electrode area again increased, in the case of the
thiol attachment, but was reduced for the amide. The observed
trend may be caused by an increase in charging current after
silane, with a further increase after dye attachment (thiol
case).

Measurements of double layer capacitance in order to
determine carrier density (no) and flatband potential also
indicate two separate trends for the thiol and amide attach—
ments. The lepe, as measured from a plot of (l/capacitance)2
against bias potential (Mott—Schottky plot), is inversely
proportional to carrier density.

l/C2 vs potential plots for the thiol modified SnO2
electrodes, shown in Figure 31, indicate only a small change
in carrier density between clean and either form of the
modified surface. The slight cathodic shift of the flat band
potential for either modified case cannot be attributed
entirely to changes in the space charge capacitance. In the
case of a semiconductor, the Helmholtz capacitance is assumed
to be much larger than the space charge capacitance, and also
to be independent of potential (62). However, when the
surface of these electrodes are modified, most of the major
structural activity, and hence change, occurs within the
outer Helmholtz plane. Therefore, the observed behavior
cannot be attributed to changes in space charge capacitance

alone.

88

.wmmxc11

 

16111 as 681xo :11 61 U6166118 ass toe “61a s111616m-1162 .1m 6L=m1a
1um<1m< .m> m<1m
QT m6. 11.11 md 114+
fl, / u d -
It. /
/ Io /
/
I/o //
II. /
I I /
I o
I I. l
/ I ll
/ ./
/ ./ Ia/a
1,,w/ +. .141.
z 4/
I 4
p70 14
I I I‘
> 8.11- u 11.; /./o In a
11: x 11.1 u o: z I; 114
m o/o/ I
of 816318-63 .6 o o; I,
.v
> no.7 u 91> I. /o: w
231.79. I x.
—N m 0.! I.
o=m\m§11m-=m .1. o o
> 8.0- .. e (x.
- 1.11 10/0 o
x . u o
FNOP W _. C n- ’v’.
95 1.38 .8 m1: x Nu: / /
a m

 

89

An increase in the flat band potential indicates that
the potential drop across the space charge region is larger
at any applied potential, which requires that the potential
drop across the Helmholtz layer is also greater. Addition
of a nonconducting functionality to the electrode surface
should give a greater potential drop across the Helmholtz
region, consistent with the observed data.

Mott-Shottky behavior for amide attachment using pr-
silane modified electrodes showed identical behavior to that
for a mercapto attachment. In the case of en-SnOz, however,
there was a significant effect in both carrier density and
flat band potential. This is consistent with the chemical
behavior of those electrodes modified using the 3-(2-
aminoethylamino)-propyltriethoxysilane (Figure 32). The
silanization reaction is extremely difficult to control, and
cross-polymerization is more possible. Therefore, the silane
addition acts to add a significant insulating character to
the electrode surface, causing a reduction in the carrier
density. The visibly greater degree of dye attachment in
the amide case at en-silene modified surfaces would act to
further insulate the electrode surface, resulting in a

greater potential drop across the Helmholtz region.

B. Photocurrent Measurement

 

The entire object of this research has been to observe
photocurrent generation, and how it is enhanced by modifica-

tion using either monolayer (covalent) or multilayer

9O

.Nocm um1w1coe mcm11mucm co axe nwsomuum mcws< Low 101a Axuuosom-upoz .Nm mgzm1u

1om<\a< .m1 m<1m

 

 

.3. E..- (1.9 11.11 mo 11.1..
.1 I, ‘ q u di
I I I
I III I
I I II-
I I ill
I I II,
I I " I I
II II ‘I"
I I "I
II II ‘l "I
I .1 I
I I II ‘ "l"
I I C m
I. .I
> om.1- u sc> .,_z .. .
. I o I. .I
1N1: x 8m 1 - 1. I .1
~05 11,8531?me .o II 0101
I
> 8.1- ... £11 I .17
2.xew;.uo= .1. .I. . .
_.N I I
NocmEcmZmucm .11 a! II
> #11- u £1, .1. I
o .1. n
. I I
52 x NM 11 .. c .1. III a I
95 1:165 .8 1 I.
I.
.I

91

(electrochemical) dye assemblies. A goal of this research
has been to compare the relative abilities of these two
types of dye-modified surfaces to produce photocurrent. In
this way, we may gain some insight into how critical covalent
attachment is.

Figure 33 shows a plot of number of equivalent mono-
layers against photocurrent output. The photocurrents were
measured at +1.0 volts, in pH 4 buffer oxalate solution
(0.1 g). Filters used were an IR, LP 39 and BP 52 so that
wavelengths between 425 and 625 were isolated, with a maxi-
mum transmittance occurring at 520 nm. Incident light was
chopped at thirteen Hertz. The approximate number of mono-
layers for the electrochemical adsorption case were calcula-
ted as mentioned earlier. For the covalent case, one
monolayer of dye was assumed, which may not be the case for
the amide-attached route.

The most important effect to notice is that, in any

given instance, the covalent attachment is always more
efficient than the electrochemical adsorption per monolayer
of material. Three distinct regions in the Figure are
apparent. In region I, an increase in adsorbed monolayers
has only a slight effect on the photocurrent production.
At about 25 equivalent monolayers, the photocurrent output
increases to a maximum (region II), and abruptly decreases
(region III).

The decrease in photocurrent at higher coverages is

expected. Gerisher (47) has observed that the quantum

92

.11_=11a 11.91 c.— 1~1

mmxu cmgowb1m >1pcm1m>oo vcm noncomum x11mo1smcuocpom1m sow mcmampocos .m> pcmcgzoouoca

mama); 1.2m._<>1:om

.mm 62=m11

 

 

om ma as mm om mm om m1 o1 m
1 1 q 1 u A 1 1 a
a.
a.
_\ .u. ..u snu.||..:. .ugv..._..
/. .v
// \ 1 5.11161. 116.111: 0
_\
/ \
111 Ssgfi/ xx 11 15.111611 (
// \ w
x \ w
/ 1 m
/ \ 11263111 3151218 0 18.111150

 

1A

~o=m\m:m1.1m-.111 co 113.1813 12181118581811 0

 

 

93

efficiency for this type of configuration is limited by the
fact that only monolayers of dye are active for electron
transfer to the semiconductor in the excited state. Thicker
layers only increase the resistance of the system without
adding to the current generation. We have observed a

similar phenomenon. The fact that the maximum in photo-
current occurs at such a high number of equivalent monolayers
is not particularly disturbing, since we have no way of
knowing that all the electrochemically adsorbed dye layers,

as calculated from charge passed, are remaining on the surface.

This Figure, however, does indicate the importance of
the covalent bond in electron transfer from the excited
state of the dye to the semiconductor. Murrary (64,66) has
suggested that the covalent linkage may allow the dye mole-
cule to bend over in order to obtain the most efficiency for
electron transfer. This plot indicates that the covalent
linkage plays more than a passive role in the photocurrent
process by allowing the dye more freedom to assume the most
efficient orientation for electron transfer.

The fact that the photocurrent for the electrochemically
adsorbed case reaches a maximum at such a high number of
monolayers suggests that the adsorbed dye may reach an
optimum surface concentration resulting in favorable over-
lap of the pi electrons of the dye and enhanced photocurrent

activity.

94

F. Conclusion -- Suggestion for Future Work

The work presented is a major contribution towards the
synthesis and characterization of modified electrode surfaces
for photocurrent generation. In one case, that of the thiol,
the covalent attachment plays a significant role in dye
retention at the electrode surface. In the other case,
amidization, proof of covalent attachment is insufficient,
and we cannot be certain that an adsorption phenomenon is
not occurring instead of the prOposed attachment.

The multilayer electrochemical adsorption, which may
also be producing a covalent linkage, also provides some
interesting information on electrode modification.

The comparison of these two separate pathways in terms
of their ability to effect photocurrent production at the
semiconductor indicates the importance of the covalent
linkage in photocurrent generation.

This research leaves several questions unanswered,
however. Experimental work aimed at better characterization
of the electrochemical adsorption process would be of im-
portance, since it would answer the question of whether or
not dye adsorption produces true covalent attachment. Such
research would deeply involve such new techniques as ESCA
and photoacoustic spectrosc0py.

Further, the synthesis of multilayer covalently attached
dye, not only erythrosin, would be of particular importance,
based on the observations for the photocurrent efficiency

for monolayers constructed by covalent attachment.

LI ST OF REFERENCES

10.

ll.

12.

13.

l4.

l5.

16.

LIST OF REFERENCES

H. Horisaki, M. Hariya, K. Yazawa, Appl. Phys. Lett.,
30,7 (1977).

 

H. Tributsch, H. Gerisher, Ber. Bunsenges. Phys. Chem.,
73, 850 (1969).

 

T. Watanabe, A. Fujishima, O. Tatsuoki, K. Honda, Bull.
Chem. Soc. Japan, 49, 8 (1976).

 

H. Kim, H. A. Laitinen, J. Electrochem. Soc., 122, 53
(1975).

 

R. Memming, G. Kursten, Ber. Bunsenges. Phys. Chem.,
76, 4 (1972).

 

H. Gerisher, Photochem. and Photobiol., 16, 243 (1972).

 

H. Gerisher, F. Willig, Topics Current Chem., 61, 31
(1976).

 

T. Kuwana, R. K. Darlington, D. W. Leedy, Anal. Chem.,
36, 2023 (1964).

 

W. N. Hansen, T. Kuwana, R. A. Osteryoung, ibid., 38,
1810 (1966).

D. Elliot, D. L. Zellmer, H. A. Laitinen, J. Electro-
chem. Soc., 117, 1343 (1970).

 

 

F. Mollers, R. Memming, Ber. Bunsenges, Phys. Chem.,
76, 469 (1972).

 

N. R. Armstrong, A. W. C. Lin, M. Fujihira, T. Kuwana,
Anal. Chem., 48, 741 (1976).

 

M. Babai, N. Tshernikovski, E. Gileadi, J. Electrochem.
Soc., 119, 1018 (1972).

 

R. DeGryse, W. P. Gomez, F. Cardon, J. Vennik,
J. Electrochem. Soc., 122, 711 (1975).

 

E. G. Gagnon, J. Electrochem. Soc., 121,512 (1974).

 

W. Mehl, F. Lohmann, Electrochim. Acta., 13, 1459 (1968).

 

95

17.

18.

19.
20.
21.

22.

23.

24.

25.

26.

27.

28.
29.

30.

31.
32.

33.

34.

35.

36.

96

H. A. Laitinen, C. A. Vincent, T. M. Bednarski,
J. Electrochem. Soc., 115, 1024 (1968).

 

J. J. Kirkland, J. J. DeStefano, J. Chromatog. Sci.,
8, 309 (1970).

 

J. J. Kirkland, J. J. DeStefano, ibid., 12, 337 (1974),

W. A. Aue, G. R. Hastings, J. Chromatog., 42, 312 (1969).

 

D. C. Locke, J. Chromatog. Sci., 11, 120 (1973),

 

K. G. Allum, R. D. Hancock, I. V. Howell, S. McKenzie,
R. C. Pitkethly, P. J. Robinson, J. Organomet. Chem.,
87, 203 (1975).

 

Petrarch Catalog, Petrarch Chemicals, Levittown, PA.

P. R. Moses, L. Wier, R. W. Murray, Anal. Chem., 47,
1882 (1975).

 

E. Grushka, Ed., "Bonded Stationary Phases in Chroma-
tography," Ann Arbor Science Pub., Ann Arbor, MI.

C. R. Hastings, W. A. Aue, J. M. Augl, J. Chromatog.,
53, 506 (1970).

 

P. G. Harrison, E. W. Thornton, J.C.S. Faraday I,
72, 1310 (1976).

 

P. G. Harrison, E. W. Thornton, ibid., 72, 1317 (1976).

C. M. Elliott, R. W. Murray, Anal. Chem., 48, 1247 (1976).

 

Corning Biomedical Supports, Pierce Chemical Company,
Rockford, IL., 61105.

P. Edman, Acta. Chem. Scand., 4, 283 (1950).

 

R. A. Laursen, Eur. J. Biochem., 20, 89 (1971).

 

E. Wachter, W. Machleidt, H. Hofner, J. Otto, FEBS
Letters, 35, 97 (1973).

G. B. Harper, Anal. Chem., 47, 246 (1975).

 

P. R. Moses, R. W. Murray, J. Electrochem. Soc.,
77, 393 (1977).

 

J. R. Lenhard, R. W. Murray, J. Electron. Chem.,
78, 195 (1977).

 

37.

38.

39.
40.

41.

42.

43.
44.
45.

46.

47.

48.

49.

50.

51.
52.

53.

54.
55.

56.

57.

97

B. F. Watkins, J. B. Behling, E. Kariv, L. L. Miller,
JACS, 97, 3549 (1975).

B. F. Firth, L. L. Miller, M. Mitani, T. Rogers, J.
Lennox, R. W. Murray, JACS, 98, 8271 (1976).

M. Fujihira, T. Matsui, T. Osa, Chem. Lett., 875 (1976).

 

M. Fujihira, Nature, 264, 349 (1976).

P. R. Hammond, A. N. Fletcher, D. E. Bliss, R. A. Henry,
R. L. Atkins, D. W. Moore, Appl. Phys. Lett., 9, 67
(1976).

 

L. N. Ionov, I. A. Akimov, Pis'ma Zh. Tekh. Fiz., l, 881
(1975).

 

V. Lebedev, Khim. Zhizn, 9, 85 (1973).

 

M. M. Martin, L. Lindquist, J. Luminesc., 10, 381 (1975).

 

B. G. Harper, U. S. Patent 3904373 (Oct. 25, 1973).

H. Tributsch, M. Calvin, Photochem. and Photobiol.,
14, 95 (1971).

 

H. Gerischer, Electroanal. Chem., 58, 263 (1975).

 

B. E. Conway, H. Angerstein-Kojlowska, W. B. A. Sharp,
Z. Phys. Chem., 98, 61 (1975).

 

L. Balsene, H. Berthou, C. K. Jorgensen, Chimia, 29, 64
(1975).

I. M. Chaiken, E. L. Smith, J. Biol. Chem., 244, 5096
(1969).

 

B. B. Hasinoff, Can. J. Biochem., 49, 742 (1971).

 

O. R. Brown, Electrochemistry, 5, 220 (1975).

 

A. Laurent, E. Laurent, R. Tardival, Tetrahedron Lett.,
4861 (1973).

 

L. L. Miller, A. K. Hoffman, JACS, 89, 593 (1967).

J. H. Scofield, J. Electron Spectrosc., 8, 129 (1976).

 

R. N. Adams, "Electrochemistry at Solid Electrodes,"
Marcel Dekker, New York (1969).

M. Gratzel, K. M. Bansal, A. Henglein, Ber. Bunsenges.
Phys. Chem., 77, 11 (1973),

 

 

58.
59.
60.
61.

62.

63.

64.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

75.

98

C. Birr, Justus Liebigs. Ann. Chem., 729, 213 (1969).

 

R. Paul, G. W. Anderson, JACS, 82, 4596 (1960).

N. F. Albertson, Organic Reactions, 12, 205 (1962).

 

D. F. Mironova, Uka. Khim. Zh., 35, 726 (1969).

 

A. Vijh, Electrochemistry of Metals and Semiconductors,
Marcel Dekker, New York (1973).

 

S. S. Fratoni, S. P. Perone, Anal. Chem., 48, 287 (1976).

 

F. Untereker, J. C. Lennox, L. M. Wier, P. R. Moses,
. W. Murray, J. Electroan. Chem., (in press).
T
a

 

. . Minnis, M. A. McKnight, R. W. Murray, Anal. Chim.
ct , 89, 355 (1977).

 

 

J. C. Lennox, R. W. Murray, J. Electroan. Chem., 78, 395

 

D. G. Davis, R. W. Murray, Anal. Chem., 49, 194 (1977).

 

K. Siegbahn, C. Nordling, A. Fahlman, R. Nordberg, K.
Hamrin, J. Hedman, G. Johansson, T. Bergmark, S.
Karlsson, I. Lindgren, B. J. Lindberg, "ESCA Atomic and
Molecular and Solid State Structure Studies by Means of
Electron Spectrosc0py," Almquist and Wiksells, Uppsala
(1967).

K. Siegbahn, C. Nordling, G. Johansson, J. Hedman, P. F.
Heden, K. Hamrin, U. Geluis, T. Bergmark, L. Werme,

R. Manne, Y. Baer, "ESCA Applied to Free Molecules,"
North Holland-American Elsevier, Amsterdam, New York
(1969).

T. A. Carlson, G. E. McGuin, J. Electron Spectrosc.,
l, 161 (1972/73).

 

P. W. Palmberg, Anal. Chem., 45, 549A (1973).

 

A. W. C. Lin, N. R. Armstrong, T. Kuwana, Anal. Chem.,
49, 1228 (1977), and references therein.

 

E. Gileadi, E. Kirowa-Eisner, I. Pencinea, "Interfacial
Electrochemistry," Addison-Wesley, London (1975).

A. Arai, U. S. Patent 3871879 (March 18, 1975).

N. R. Armstrong, C. N. Sayers, personal communication,
Michigan State University, East Lansing, MI.

99

76. H. Gerisher, J. Electrochem. Soc., 113, 1174 (1966).

 

77. H. Gerisher, H. Tributsch, Ber. Bunsenges Phys. Chem.,
72, 437 (1968).