Group Title: synthesis and characterization of reversed phase stationary phases for high performance liquid chromatography /
Title: The Synthesis and characterization of reversed phase stationary phases for high performance liquid chromatography
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00099575/00001
 Material Information
Title: The Synthesis and characterization of reversed phase stationary phases for high performance liquid chromatography
Physical Description: xi, 162 leaves : ill. ; 28 cm.
Language: English
Creator: Barnes, Karen Wink
Publication Date: 1986
Copyright Date: 1986
 Subjects
Subject: High performance liquid chromatography   ( lcsh )
Liquid chromatography   ( lcsh )
Ultrasonic waves   ( lcsh )
Silica-alumina catalysts   ( lcsh )
Chemistry thesis Ph.D
Dissertations, Academic -- Chemistry -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
Thesis: Thesis (Ph.D.)--University of Florida, 1986.
Bibliography: Bibliography: leaves 156-161.
General Note: Typescript.
General Note: Vita.
Statement of Responsibility: by Karen Wink Barnes.
 Record Information
Bibliographic ID: UF00099575
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000940978
notis - AEQ2512
oclc - 016656138

Downloads

This item has the following downloads:

PDF ( 5 MBs ) ( PDF )


Full Text












THE SYNTHESIS AND CHARACTERIZATION OF REVERSED PHASE
STATIONARY PHASES FOR HIGH PERFORMANCE LIQUID CHROMATOGRAPHY











By

KAREN WINK BARNES
























A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA


1986

































To my family with all my love. It was because of you

that I had the courage to begin and the strength to

finish this endeavor.















ACKNOWLEDGEMENTS


There are many people to whom I am deeply indebted for

making my graduate school years so memorable. My deepest

gratitude is due to Dr. John Dorsey for his guidance,

patience, inspiration, and for his "lessons in life." During

the years that I spent as his student, I matured both person-

ally and professionally from his lessons.

Special thanks are due to the Dorsey group members, past

and present, for their friendship. I am going to miss all

of them. Special thanks are also due to Dr. Morteza Khaledi

for his help during the early days of this project.

Dr. John Novak at Alcoa provided guidance and support as

well as alumina samples and SEM shots. His help is grate-

fully acknowledged. I am especially indebted to Mel Courtney

for the elemental analysis, to Danny Coffman for his expert

drafting, and to Laura Griggs for her unwavering friendship

throughout the years and for her diligent typing efforts.

There are no words suitable to thank my husband, Rob

Barnes, for what he has invested in my degree. His willing-

ness to help me implement my Rube Goldberg systems and to

listen even when he did not understand what I was talking

about meant the world. I'd have never made it without him

by my side, and I promise that when his turn comes I will be

there for him.















TABLE OF CONTENTS



Page

ACKNOWLEDGEMENTS. . . ... . . . . . . iii

LIST OF TABLES. . .... . . . . . .. vi

LIST OF FIGURES . . . .. . . . . * vii

ABSTRACT. . . ... . . . . . .... x

CHAPTERS

I INTRODUCTION. . . . .... . . .. .. 1

Background. . . . .. . . . . .... 1
Reversed Phase Stationary Phases . . . 3
Silica . . . . . . . . .. 3
Silanol functions and surface water on
silica . . . . ... . .. 6
Surface modification . . ... . 9
Bonding Process. . .... . . . .. 14
Evaluation of Bonded Phases. .... . .. 15
Limitations of Reversed Phase Liquid Chroma-
tography. . . . .. . . . 24
Research Objective. . . ... . . . . 25

II ULTRASOUND. . . . ... . . . . 26

Introduction. . . ... . . . . . 26
Cavitation. . . ... . . . . . . 27
Reversed Phase Stationary Phases from Ultrasonic
Cavitation . . . ... . . . . 30

III ULTRASOUND REACTIONS WITH SILICA. . . . ... 32

General Experimental Conditions ... . . . 32
Reagents ...... . . . . . . 32
Instrumentation ... . . . . .. 33
Reaction Conditions. .... . . . . 34
Precision of Elemental Analysis . . . .. 36








Page

Ultrasound vs. Refluxed C18 Silica. . . .. 38
Experimental . . . . . . . . 38
Results. . . .. . . . . . . 39
Reproducibility of Column Packing . . . .. 55
Effect of Acoustic Power of C18 Reactions . . 67
Experimental . . . . . . . .. 67
Results. . . . . . . . . ... 69
Reproducibility of Ultrasonic Reactions ... . 77
Ultrasound with Heat. . . . . . . ... 85
Commercial C18 Silica Column. . . . . .. 88
Chromatographic Characterization of the Station-
ary Phase Materials. . . . . . .. 91
Conclusion. . . . . . . . . .. 98

IV ALUMINA . . . . . . . . ... . 100

Introduction. . . . . . . . . ... 100
Reversed Phase Stationary Phases on Alumina . .103

V ALUMINA REACTIONS . . . . . . ... 106

General Experimental. . . . . . . ... 106
Controlled Pore and Pseudoboehmite Alumina. . 107
Optimization of Reaction Procedure. . . .. .109
Activated Gamma Alumina Experiments . . .. 117
Effect of Water on the C12 Bonding Reaction . 120
Gamma Alumina and Ultrasonic Powers . . .. .123
Conclusions . . . . . . . . ... 150

VI CONCLUSIONS AND FUTURE WORK . . . . . 151

Ultrasound. . . . . . .. . . . . 151
Alumina Phases. . . . . . . . .. .152

APPENDIX SCHEMATIC OF THE WATER LEVEL CONTROL SYSTEM
FOR THE ULTRASONIC BATH . . . . ... 154

REFERENCES. . . . . . . . . . . 156

BIOGRAPHICAL SKETCH . . . . . . . . . 162















LIST OF TABLES


Page

1 CHN Precision. . ... . . . . . . . 37

2 Column Packing Statistics for Pre-endcapped Ultra-
sound C18 Silica . . .. . . . . . 63

3 k', a, aCH2 Values for Toluene, Benzene, Naphtha-
lene, 2-Aminonaphthalene, and a Homologous Series
of Alkylbenzenes for C18 Silica Columns. ... . 95

4 B/A and h Values for C18 Silica Columns for Naph-
thalene and 2-Aminonaphthalene . . . . . 97

5 Values for tR, k', B/A, and h for Toluene Measured
after Each of the Washing Processes with C12
Alumina . . . . . . . . . . 143















LIST OF FIGURES


Figure Page

1 Hydrogen bonding and silanols occurring on
silica . . . . . . . . .. 8

2 Term definitions for Equation 4 . . ... 19

3 Van Deemter curve . . . . . . . 22

4 Elemental analysis of C18 bonding reactions
from reflux and ultrasound . . .. 41

5 Reduced plate height vs. flow rate curve for
endcapped refluxed C18 silica . . . . 44

6 Reduced plate height vs. flow rate curve for
endcapped ultrasound C18 silica . . ... 46

7 Test mixture on refluxed C18 silica ... .48

8 Test mixture on endcapped refluxed C18 silica 50

9 Test mixture on low power ultrasound C18
silica column . . . . . . . . 52

10 Test mixture on endcapped low power ultra-
sound C18 silica. . . . . . . ... 54

11 Scanning electron micrograph of leached
silica. . . . .. . . . . . . 57

12 Scanning electron micrograph of endcapped
refluxed C18 silica . . . . . .. 59

13 Scanning electron micrograph of endcapped
ultrasound C18 Silica . . . . . . 61

14 Packing reproducibility study . . .. 65

15 Elemental analysis of C18 bonding as a func-
tion of acoustical power . . . .. 71

16 Reduced plate height vs. flow rate curve for
endcapped high power C18 silica . . ... 74








Figure Page

17 Reduced plate height vs. flow rate curve for
endcapped high power C18 silica. . . .. .76

18 Test mixture on high power ultrasound C18
silica . . . . . . . . . . 79

19 Test mixture on endcapped high power ultra-
sound C18 silica . . . . . . . 81

20 Reproducibility of ultrasonic reactions. . 84

21 Effect of heat on ultrasonic reactions . . 87

22 Reduced plate height vs. flow rate curve for
commercial C18 silica. . . . . . ... 90

23 Test mixture on commercial C18 silica. ... .93
18

24 Effect of solvents on C12 bonding of pseudo-
boehmite and controlled pore alumina ... .112

25 Scanning electron micrograph of controlled
pore alumina . . . . . . . .. 114

26 Scanning electron micrograph of pseudoboehmite
alumina. . . . . . . . . . 116

27 Elemental analysis of C12 bonding reactions on
activated y alumina . . . . . . 119

28 Effect of water on C12 bonding reaction. . 122

29 Reduced plate height vs. flow rate curve for
ultrasound C12 pseudoboehmite alumina
(10-18 pm) . . . . . . . . . 125

30 Scanning electron micrograph of 10-15 Um frac-
tion of pseudoboehmite alumina . . . .. .127

31 Elemental analysis of C12 bonding with differ-
ent reaction modes on y alumina . . .. 130

32 Elemental analysis of C12 bonding with differ-
ent reaction modes on y alumina. . . .. .132

33 Scanning electron micrograph of y alumina. . 135

34 Scanning electron micrograph of low power
ultrasound C12 y alumina . . . . .. 137


viii








Figure Page

35 Reduced plate height vs. flow rate for low
power ultrasound C12 y alumina. . . .. .139

36 Reduced plate height vs. flow rate for low
power ultrasound C12 y alumina . . . 141

37 Stability of low power ultrasound C12
y alumina . . . . . . . . 145

38 Test mixture on refluxed C12 y alumina. . 148















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy



THE SYNTHESIS AND CHARACTERIZATION OF REVERSED PHASE
STATIONARY PHASES FOR HIGH PERFORMANCE LIQUID CHROMATOGRAPHY

By

Karen Wink Barnes

December 1986


Chairman: John G. Dorsey
Major Department: Chemistry



Reversed phase liquid chromatography (RPLC) is a widely

used separation technique today. The stationary phase, com-

posed of hydrocarbon moieties chemically bound to a silica

support, is commonly prepared by refluxing the silica with a

reactive silane in an appropriate solvent. Siloxane bonds,

Si-O-Si, are formed. There are limitations to RPLC in that

nonhomogeneous surface coverages, detrimental to efficient

separations, result from the bonding process, and because

the reaction is never complete due to steric restrictions.

Also, silica is soluble at high pH values, and the Si-C bond

binding the hydrocarbon to the silica is labile at low pH

ranges. Thus, the usable pH range for silica stationary

phases is 2.5 to 7.5, and often this range is too narrow to

allow the separation of a mixture. The use of ultrasonic








cavitation to catalyze silane bonding was investigated to

determine whether the vigorous ultrasonic process would drive

reagents into the surface pores and better distribute the

hydrocarbonaceous reagent, thereby producing a more efficient

stationary phase. Tandem reactions proved that ultrasonic

bonding procedures are as effective as the refluxed, and that

the chromatographic efficiency of the ultrasonic phases was

comparable, if not superior, to the refluxed. Chromatographic

tests also indicated the ultrasonic phases were comparable

to commercially available phases. The reproducibility of

the reactions and the effect of acoustic power and heat were

also investigated.

A second set of experiments investigated substitution

of alumina for silica because the alumina crystalline struc-

ture remains intact over a pH range of 2 to 12. A trifunc-

tional modification scheme generating a thick cross-linked

hydrocarbon matte was used because the Al-0 bond is suscep-

tible to hydrolysis, which strips the bound hydrocarbons, by

polar RP solvents. Five aluminas were bonded and tested for

chromatographic utility. Reaction conditions were optimized,

and ultrasonic bonding was investigated. It was found that

alumina with a sufficiently active surface is modifiable, and

that the modified surface is remarkably stable for use with

acidic and basic buffers and in neutral, polar solvent mix-

tures. The chromatographic utility of alumina phases was

demonstrated and compared with the silica results.















CHAPTER I

INTRODUCTION



Background


A recent review article (1) presented data from a market

survey conducted by Centcom, Ltd. from Westport, CT, which

indicated that the column liquid chromatography (LC) market

for 1983 was the largest single analytical instrument market

for that year with proceeds of $568 million. The LC market,

which includes High Performance Liquid Chromatography (HPLC),

column chromatography, amino acid analysis, and ion chromat-

ography was projected to grow to $866 million in 1986. In

the above survey, HPLC was the largest category within the

LC market with worldwide sales of $365 million for 1983.

Sales of HPLC supplies alone were $52 million in 1983 and

have been projected to increase by 18% annually through

1988 (2). Of all the separation techniques available, the

reversed phase (RP) method is the most widely used by chro-

matographers today. It has been estimated that about 70%

of all LC analyses are performed by RPLC (3). The popularity

of the method is demonstrated by the vast quantity of RP

media on the market; currently more than 200 different col-

umns are available (4). In part, the popularity of RP is due

to the wide variety of applications possible with the tech-

nique.

1








In all chromatographic techniques, a mobile phase and a

stationary phase are employed. In HPLC the stationary phase

is an immobile material consisting of very fine particles

commonly ranging from 3 to 10 microns (pm) in diameter (5).

The stationary phase is typically packed into a stainless

steel column. The mobile phase, some appropriate solvent,

is forced through the column with high pressure pumps. In

RPLC the mobile phase is polar in character with water,

methanol (MeOH), acetonitrile (ACN), tetrahydrofuran (THF),

or mixtures of these being most common. The strongest mobile

phases are those which most closely match the polarity of

the stationary phase. Stronger mobile phases are defined as

those which decrease the retention of solutes. The station-

ary phase is nonpolar in character and is often composed of

hydrophobic moieties chemically bound to a polar support

material. Silica is the material traditionally used as the

core of RP phases.

Separation of a mixture of solutes is accomplished via

differential migration rates through the column (5), caused

by the differing affinities of the solutes for the mobile or

stationary phases. In the RP mode, polar solutes with the

highest affinity for the mobile phase and little attraction

for the stationary phase move rapidly through the column and

are eluted before nonpolar components which are retained on

the nonpolar stationary phase. The actual mechanism of

retention in RPLC is still under debate. It has been pro-

posed that liquid-liquid partitioning occurs between the

mobile phase and a swollen network composed of the bound









hydrophobic portion of the stationary phase and associated

molecules from the mobile phase (6). Retention may be visu-

alized as the sum of several interactions between solute and

stationary phase including coulombic forces and dipole-dipole

attractions, as well as dispersion (7). Martire and Boehm (8)

developed a lattice model to study retention and selectivity

in RPLC. They analyzed the composition and structure of the

stationary phase as a function of chain length of the bonded

moiety, the intrinsic chain stiffness, the surface coverage,

and the nature of the mobile phase. Retention behavior was

analyzed based on the above considerations and also upon the

nature of the solute and the temperature of the system.


Reversed Phase Stationary Phases

To be a viable stationary phase for a RP separation, the

packing material must be inert with respect to irreversible

adsorption of polar solutes, it must have high pressure

stability (7), and it must be stable over a wide range of

solvent compositions, of either hydrophobic or hydrophilic

nature (9). The material must also be of small size and of

narrow particle size distribution, it must have optimum sur-

face characteristics, and it must have a high concentration

of organic groups (10).


Silica

Silica is the core material which is most commonly used

for RPLC. In nature silica exists in crystalline and noncrys-

talline forms. Crystalline silica is found at room tempera-

ture in three forms: quartz, tridymite, and crystobalite (6).








Amorphous silica, the form used exclusively for HPLC, is

found in nature as opal, infusorial earth, and diatomaceous

earth. Quartz glass produced by supercooling liquid silica

is intermediate between crystalline and amorphous forms.

The discovery of silica is credited to Sir Thomas Graham who

in 1861 prepared the material by mixing a solution of sodium

silicate with hydrochloric acid (11). Today most synthetic,

noncrystalline silica is produced by Patrick's process

involving the gelation of alkali metal silicates with

acids (11). The reaction is as follows:


(HO)3-Si-OH + HO-Si(OH)3 --+ (HO)3-Si-O-Si-(HO)3 + H20


The condensation process generates an irregular, reticulated,

three-dimensional network of SiO4 tetrahedra. Electron

micrographs of these elementary particles indicate that their

structure is a coherent aggregate of roughly spherical par-

ticles with diameters of approximately 100 A (11). X-ray

data show that each silicon atom is bound to four oxygens

and that each oxygen is linked to two silicon atoms. These

silicons may be in adjacent tetrahedra within the same ele-

mentary particle or in adjacent particles. Siloxane bonds,

(Si-O-Si), are common to all forms of silica. The SiO4

tetrahedra complete their coordination with hydrogen at room

temperature to form silanol groups, (SiOH), which are respon-

sible for the polar, hydrophilic nature of silica.

The siloxane linkages between the elementary particles

of silica result in a porous surface, the chromatographic

importance of which will be discussed later. The





5

morphological properties used to characterize the surface of

the silica are pore diameter, pore volume, and surface area.

The IUPAC designates pores smaller than 2 nm micropores,

those exceeding 50 nm are macropores, and those intermediate

in diameter are mesopores (6). Since every pore represents

distinct geometric shape, the shape must be considered.

Three models have been proposed: 1) cylindrical and open

at one or both ends, 2) ink-bottle pores with dual diameters

including a narrow neck region, and 3) slit pores composed

of layers of plates. A diagram of the pore structures may

be found in Chapter I of Unger's monograph on porous

silica (6).

The specific surface area of porous silica equals the

sum of the internal (Si) and external (S ) surface areas.

The S. term represents the area of the walls of the pores

that are exposed to the surface and is several orders of

magnitude greater than S which corresponds to the geometric

surface of the particle per gram of material (6).

Specific pore volume is defined as that amount of liquid

which fills the total volume of the pores per gram of mate-

rial (6). This includes pores of all sizes, micro, meso,

and macro. Characterization of the pore properties is typic-

ally performed via the Brunauer, Emmet, and Teller (BET)

method of gas adsorption and by the technique of mercury

penetration. The first technique correlates adsorption with

capillary condensation of gases within the pores, and the

second is based on capillary depression, the phenomenon of

penetration into pores by a non-wetting liquid. An excellent






6

review of the two techniques may be found in Chapter II of

Unger (6). Berendsen et al. (12) demonstrated in some detail

how various pore and surface measurements are derived using

the BET method, and El Rassi and Gonnet (13) measured surface

properties of deactivated silica gel using the BET method.

An interesting aspect of the surface area-pore membrane

thickness relationship was demonstrated by Verzele et

al. (14), who postulated that the enormous specific surface

suggests that the wall thickness between pores is very thin,

ranging from 1-2 nm for typical chromatographic grade silica.

This implies that the walls contain only one or two silicon

atoms at any particular position along the wall.

Bonding densities of hydrocarbon chains onto silica are

directly dependent upon the surface characteristics. For RP

systems increased retention of solutes is found for materials

with increased surface area (15,16), and Berendsen et al. (12)

have proposed a model to estimate the maximum surface coverage

based on the pore structure of the silica core.


Silanol functions and surface water on silica

The surface modification of silica to produce a chromat-

ographic RP stationary phase is performed at the silanol

groups, which are attached in various ways to the silica

surface and are very much dependent upon water which is

physisorbed onto the surface of the silica. Peri and

Hensley (17) proposed that hydrogen-bonding interactions

occur between adjacent silanol functions on amorphous silica,

and they label silanols interacting in this fashion vicinal








hydroxyls. Exposure of silica to water vapor results in the

adsorption and hydrogen-bonding of the water. The hydrogen-

bonding and the types of silanols occurring on the surface

of silica are diagrammed in Figure 1. According to Unger (6),

the degree of hydration of the surface is a function of the

vapor pressure, the temperature, and the specific surface

area of the silica. This sorbed water must be removed in

order to bond the silica. There are several differing

schools of thought concerning appropriate temperatures for

thermal treatment of the silica to remove the water (13).

The objective of drying is to remove the maximum amount of

surface water without sintering the silica, which occurs

when at sufficiently high temperatures, the internal and

surface silanols migrate through the silica matrix and con-

dense into water which is then lost (6). Permanent loss of

surface area results. There are a variety of values proposed

for the number of hydroxyls that remain after thermal treat-

ment (6,11). According to Verzele et al. (14), a maximum

number of 7.8 pmole OH/m2 specific surface area is plausible

based upon stoichiometric considerations. This value in-

cludes silanol functions residing in the pores of the silica.

Fripiat and Uytterhoeven (18) maintain that the maximum sur-

face hydroxyl concentration possible after heating at 513 K

is 6.7 pmole/m'. At this temperature residual water is

negligible compared to the silanol (6). For this reason all

silica for this work was heated at 513 K.

Silanols may be monitored chemically and physically (11).

Chemical methods involve the reaction of the surface hydroxyls














Ur vn Isolated Silanol
I Functions
Si Si
/1\ /I\






0 0 -H
0V "''0 H
I I Vicinal Silanols
Si Si
/1\ /I\


H, ,H
H H\
0 0

Si Si
/\\ /1\


Water Hydrogen-
Bonding with
Silanols


Figure 1. Hydrogen bonding and silanols occurring
on silica.








with a reagent and the consequential production of an analyt-

ically measurable entity. Physisorbed water will interfere

with all of these measurements. Two of the more commonly

used reactions are


Silica + Diborane: SiOH + B2H6 -- SiO-B2H5 + H2


Silica + Methyllithium: SIOH + LiCH3 -- SiO-Li + CH4


Infrared spectroscopy (IR) in the frequency range of

2000-4000 cm-1 is the physical method typically used to mon-

itor the silanols and adsorbed water directly (6,19). Iso-

topic exchange is also used to measure the hydroxyls.

Deuterium and Tritium are used as indicated.


SiOH + D20 0 SiOD + HDO


SiOH + HTO i SiOT + H20


The exchange is monitored by shifts in the IR spectra.


Surface modification

Surface modification of silica is accomplished via the

heterogeneous reaction of organosilanes with surface silanol

groups. An interesting historical note is that Howard and

Martin (20) were the first to use silane reactions to produce

a hydrophobic support material for RPLC. The modification

involves a reactive organosilane of the type R (4n)Si-X(),

where lns!3; R is the ligand which is bound to the surface

of the silica; and X is an easy leaving group, most commonly

an alkoxy or halogen (9,21-25). Specifically, R is






10

covalently attached to the surface through siloxane formation

between the organosilane and the silica, and an acid or

aliphatic alcohol is liberated in the process. This is shown

below with a monofunctional silane:


R R
ESiOH + X-Si-R' -Si-0-Si-R' + HX
R R



The R' is an alkyl group with 1 to n carbon atoms. Values

of 2, 4, 8, and 18 are typical for n, and R is often a methyl

group. The most common bonding reaction is performed by

mixing silica and dimethyloctadecylchlorosilane and is shown:


CH3 CH3
:SiOH + Cl-Si-(CH )17CH3 -- ESi-0-Si-(CH2)17CH3 + HC1
CH3 CH3


Generally the reaction is performed by refluxing the materi-

als in a high boiling solvent such as benzene or toluene

with a basic substance such as pyridine added as a scavenger

for the HC1. The resultant surface has been described as

resembling a dense forest of tall trees (10) or as a surface

covered with bristles (26). For these models the alkyl

chains are assumed to be fully extended. The surface also

has been described (26) as covered with a blanket of alkyl

grass. In this model the chains are assumed to be folded

upon themselves. When bonding is accomplished using mono-

functional silanes, a monolayer coverage is achieved. The

uniformity of surface coverage is a matter of some debate

and will be discussed more fully later.









Polymeric reactions are also performed using a di- or

trifunctional silane. In these reactions a thick matte is

formed around the silica as shown here:


=Si-OH =Si-0
1 Cl, R R'
S + Si -- 0 Si + 2HC1
I C R / 'R
=Si-OH =Si-0


If all the chlorides do not react with the surface silanols,

and water is present, cross-linking may occur between adja-

cent bonded silanes (6).

R R
I 1
=Si-0-Si-R' =Si-O-Si-R'
=Si-H C1 R' H20 OH 0 RO
0 + 2 Si 0 Excess 0 S
S C1 R OH Silane 0 R'
=Si-OH OH I ,
=Si-0-Si-R' =Si-0-Si-R'
H H
R R

Other cross-linking combinations are possible between vicinal

hydroxyls and the reagents. Trifunctional silanes act in a

similar manner.

Other, more novel reactions have been performed. For

example, Haas and Kohler (27) found the use of partially

fluorinated ligands gave completely different chromatographic

character to stationary phases than did the use of hydrogen-

ated analogs. Schomburg and coworkers (7,28) found that

polymeric phases could be synthesized by cross-linking

oligomers of polymethyloctylsiloxane or polybutadiene onto

silica with gamma radiation, and they also discovered that

peroxides could be used to induce the cross-linking reac-

tion (28). Crown ether complexes have been covalently bound

to silica by Waddell and Leyden (29), and Hartwick et al. (30)









produced and bonded the silanes in situ to the silica support

in a microwave oven.

Regardless of the mode of bonding employed, the attain-

able coverage of hydrocarbon moieties in pmoles/m2 is sub-

stantially lower than the number of surface hydroxyls theo-

retically available for bonding. The extent of loading of

the organic onto the silica is very dependent upon surface

properties of the silica (12,15,16), as mentioned previously,

and upon steric factors of the silane (9,15,25,31). Cheng

and McCown (32) propose a mechanism to explain the steric

effect in which silanization represents a self-forming size

exclusion process, and they suggest that maximum loading

limits are attained when the density of bound chains block

the bonding of additional chains. They relate the process

to planting trees in a dense,uniformly planted forest.

Ultimately, no more trees may be planted due to lack of

space. Sander and Wise (15) explain the size exclusion con-

cept as it pertains to the pore structure of the silica;

small pores can be blocked easily by the silane reagent

thereby limiting the extent of bonding within the pores.

Kirkland (10) presents the matter as an umbrellar shielding

of residual silanol groups by tightly packed organic groups.

The bonding limit reported in the literature is about

4.0 pmoles/m2 for trimethylsilane, the smallest of the silane

used chromatographically and approximately 3.5 pmoles/m2 for

larger monomeric silanes (3,9,25). Polymeric silanes can

give higher bonding densities (15).





13

Nonreacted or residual silanol groups on the surface of

the treated silica can severely interfere with a separa-

tion (3,7,24,33). Endcapping is commonly performed in an

attempt to remove the residual silanols. Typically, a small

silane is reacted with the bonded silica. Hexamethyldisi-

lazane has been used (23), but the most common reagent used

is trimethylchlorosilane. This reagent is small enough to

slip between the bulky alkyl chains. The reaction is similar

to the C18 bonding reaction in that a siloxane bond is

formed. The endcapping of C18 silica is shown.


CH3 CH3




| 3H3 I 3
-Si-OH + C-Si-CH Si-0-Si-CH
=Si-O-Si-(CUH)17CH3 =Si-0O-Si-(CH2)17CH3
1 CH3 CH3



CH CH3
111 CR3
-Si-OH + Cl-Si-CH -+ Si-0-Si-CH


CH3 CH3
=Si_0-Si-(CH 17CH3 =Si-0-Si-(CH 17CH3
CH3 CR3


The endcapping process is never complete. Silanols

remain on the surface of the silica and within the pores even

after exhaustive silanization. These reactive hydroxyl

groups are available to solute molecules and interfere with

the separation as discussed above. This is a serious limit-

ation of RPLC today. The chemical and physical methods used

to test bare silica have also been used to determine post-

endcapping, residual silanols (25). Additionally, Cheng and

McCown (32) report residual SiOH functions may be monitored






14

by a modification of Sears' method in which silica suspended

in sodium chloride is titrated to pH 9.0 with sodium hydrox-

ide. Another common test for residual silanols involves the

adsorption of methyl red, whereby silanols are indicated by a

red staining on the surface (6). Nondek and Vyskocil (34)

proposed a nondestructive test whereby a plug of dimethylzinc-

tetrahydrofuran is injected into a dry, packed column, and the

evolved methane is measured with a katharometer. Monitoring

of residual silanols has also been performed by chromato-

graphic retention measurements of solute mixtures of nonpolar

and polar (basic) components and is discussed in works by

Schomburg et al. (9), Lochmuller et al. (35), and Sadek and

Carr (36).

The problem of nonreacted silanols is sometimes overcome

by masking, that is, via deliberate addition of interactive bases

to the mobile phase. Some of the most successful silanol

blockers have been alkylammonium compounds used in conjunc-

tion with buffers (36). Sadek and Carr (36) extended this idea

further with the use of cyclic tetraaza (cyclam) compounds.


Bonding Process

According to Snyder and Kirkland (5), the appropriate

synthetic procedure for bonded phases is as follows:


1) Acid leach the silica support at reflux tempera-
tures for 24 h. This serves to remove trace
metals in the silica matrix which have been shown
to adversely affect the chromatographic perfor-
mance (15,37,38) via chelation with certain
solutes. Leaching also serves to fully hydroxyl-
ate the surface and to open the pore structure,
thereby increasing the accessibility of the silane.






15

2) Dry the silica at temperatures above 473 K under
vacuum to remove the physisorbed water. This tends
to prevent the reactive silanes from polymerizing
into silicon oils (26). All water must be excluded
from the reaction for the same reason.

3) Reflux the silica with a large excess of chlorosi-
lane, twice stoichiometric quantities are recom-
mended, in a dry high-boiling solvent. A base
should be added to remove HC1. Kinkel and Unger (22)
suggest that the solvent and base of choice are
methylene chloride (MeC12) and 2,6-lutidine.

4) Wash the material. Kinkel and Unger (22) recommend
a sequence of MeCl MeOH, 50/50 Me0H/H20, MeOH,
and diethylether. The objective of the washing
sequence is to convert any nonreacted silanes to
silanols.

5) Dry the material under vacuum.

6) Endcap following the procedures in 3).

7,8) Wash and dry.



Evaluation of Bonded Phases

The extent of the bonding reaction is commonly calcu-

lated from elemental carbon analysis. The calculation of

bonded rates on silica modified with a monochlorosilane from

carbon analysis is given in Equation 1 (25).


%C-106
S12nC S(100 (%C/12nC)(Mwt 36.5)) (

where P = surface coverage (pmoles/m2)
%C = weight percent of C/100g bonded silica
nC = moles of carbon/mole of silane
wt = molecular weight of silane
S = surface area of unbonded silica


The loss of 36.5 g is due to evolution of HC1 during the

course of the reaction. Typical values have been mentioned






16

previously. Engelhardt and Ahr (25) report that good repro-

ducibility of bonding results if the reaction process is

standardized and the same batch of silica is used.

The material must be packed into a column before

chromatographic evaluation is possible. The best packing

process is determined by the size and nature of the bonded

material (5). For particles less than 20 um in diameter

slurry packing procedures must be used. In this process a

suitable liquid is chosen to wet, disperse, and suspend the

packing material. The slurry is then extruded into a clean

column blank at high constant pressures. Typical slurry

liquids are of density equivalent to the phase material and

of low viscosity, thereby permitting high slurry flow veloc-

ities through the column at the onset of packing which

markedly enhances column efficiency. Repetitive repressur-

izations by the pump during the packing procedure serve to

stabilize the stationary phase bed,thereby prolonging column

life (10). Conditioning and equilibration solvents are also

passed through the column in the course of the packing

process.

Chromatographic figures of merit is a term which has

been used to collectively refer to parameters used in the

characterization of chromatographic peaks and columns (39).

Perhaps the best known of these is the number of theoretical

plates of a column (N) which relates the variance of an

eluted peak to the square of the retention time. The N value

for a column is often measured according to Equations 2 or 3.








N = 16(tR/Wb)2 (2)


N = 5.54(tR/ W05)2 (3)


where tR = retention time of a peak (cm or s)
Wb = width of the peak at the base (cm or s)
W.5 = width of the peak at half height (cm or s)


These equations are based upon the assumption that the peak

is perfectly Gaussian. Kirkland et al. (40) have shown that

serious errors in plate counts result if either equation is

used if the peak is asymmetrical. Foley and Dorsey (39)

suggest usage of Equation 4 to compensate for peak skew.

This equation has been shown to be valid for perfectly

Gaussian (B/A=1.0), as well as peaks which are skewed up to

B/A values of 2.76.


41.7(tR /W 1)2
S B/A + 1.25


Figure 2 illustrates the terms presented in Equation 4.

The height of a theoretical plate (H) may be calculated:


H = L/N (5)


where L = length of column (cm)


The dispersion of a solute as it passes through the

column may be attributed to various factors (5). The con-

tributions from these factors are summed in the following

equation:




































Figure 2. Definition of terms for
Equation 4.































z
0


z


where W.1 = width at 10% peak height
0.1









H = Au0.33 + B/u + Cu (6)


where u = L/t0 (7)

and A = contribution from eddy diffusion
B = contribution from longitudinal diffusion
C = resistance to mass transfer
u = mobile phase velocity (cm/s)
L = column length
t, = retention time of nonretained peak


Knox and coworkers (41-43) have shown for an ideal porous

silica based column, Equation 6 may be reduced to the follow-

ing:


h = (2/v) + (v0.33) + 0.05v (8)


and


h = H/dp (9)


and


= u d (10)
Dm

where h = reduced plate height
dp = particle diameter (cm)
u = flow velocity from Equation 7 (cm/s)
v = reduced velocity
Dm = diffusion coefficient of a solute in liquids


An ideal Van Deemter curve, which is similar to a Knox

curve, except that H rather than h is plotted as a function

of reduced velocity, is shown in Figure 3. Reduced plate

heights are commonly used for comparison and evaluation of

columns because differences in particle diameters, column

lengths, and flow rates are standardized.










































U
-4



0





4-

C)

C-

0,

U

C-
C)












C7



cC


CCl
C),





23

In order to compare retention data from column to col-

umn, the capacity factor (k'), the expression relating the

length of time a solute is retained by the stationary phase

to the time the solute spends in the mobile phase, is the

fundamental parameter used. Equation 11 is used to calculate

k' values.


k' = (tR t0)/t0 (11)


and

a = k'2/k'1 (12)


where a = separation selectivity
tR = the retention time of the retained solute
to = the elution time of an unretained solute


Snyder and Kirkland (5) state that t0 is measured in various

ways but that the most common method is to measure the center

of the first baseline disturbance. Other methods involve

the injection of a solution of NaNO3, weighing a packed col-

umn filled with two different mobile phases, and measuring

the retention time of a homologous series of compounds (44,45),

in which a plot of the logarithm of net retention time

versus the carbon number (nC) is used to extrapolate the

value of tO.

Additional information may be gained by investigation

of retention behavior of a homologous series of compounds

differing by only one methylene (CH2) group (46). First, the

value of the slope (log a) of the line of log k' vs. the nC

characterizes the nonspecific interactions between the









solute and stationary phase for a given type of stationary

phase, i.e., C18. With the comparison of log a values for a

homologous series, differences in stationary phases such as

specific surface area, carbon loading, and residual silanols

are standardized because, while one member of the series may

interact with residual silanols, for example, this interac-

tion will be masked because all the members will interact in

a comparable manner. Any differences in values of log a

will be due to the actual nature of the bonded phase struc-

ture (chain length, etc.) (46,47). Colin et al. (46) claim

for this reason that log a is an excellent means to charac-

terize solvent strengths. Additionally, the value of the

intercept of the plot of log k' vs. nc can provide informa-

tion characterizing the mobile phase, stationary phase, and

the series itself (46).


Limitations of Reversed Phase Liquid Chromatography

There are three limitations to RPLC today. The first,

persistent residual silanol groups, has been discussed.

Inhomogeneity of alkyl chain bonding also has been mentioned.

This effect would result in decreased chromatographic effi-

ciency. Lochmuller and coworkers have presented results

that tend to support this theory (48,49). They report

fluorescence from excimers formed between pyrene probes which

had been bound to the surface. Excimer formation has a

critical interaction distance of 3-8 A. Under the given

reaction conditions, i.e., low bonding densities, excimer

formation would not have been possible if the probe








distribution were uniform. The validity of this experiment

tends to be limited because of the propensity of pyrene to

aggregate, evident by the high boiling and melting points.

However, Gilpin (26) supports Lochmuller's assertion with

spectroscopic evidence.

The final hindrance to RPLC today results from the

chemical instability of the silica support material. It is

commonly accepted that the pH of mobile phases should be

maintained within the range of 2.5 tb 7.5 (5). The lower

limits are imposed by the Si-C bond. While Si-C is the most

stable of the Si-X bonds, with the exception Si-O (6), it is

quite labile at lower pH values. The dramatic increase in

solubility of the siloxane bonds limits high pH values (6).

Often in practical LC these limits must be exceeded, with

decreased column life the result.



Research Objective


The importance of the stationary phase to HPLC was the

motivation for this research. The objective was to improve

stationary phases by removing the aforementioned hindrances

to the method. The first study involved the use of ultra-

sound to drive the reactions in an attempt to improve the

distribution of the alkyl chains and to reduce the post-

endcapping silanols. The second study involved substitution

of alumina for the silica core in the attempt to extend the

usable pH range of RP phases.















CHAPTER II

ULTRASOUND



Introduction


Recently there has been a renewal of interest in the

effect of ultrasonic waves on chemical reactions (50-62).

The use of ultrasound serves to dramatically increase both

the rates and the yields of reactions, and usually the reac-

tion conditions required are substantially milder than for

the conventional reaction procedure. There are also instan-

ces where reaction yields were very low or no reaction was

observed in the absence of ultrasound (57-62). Current

analytical applications of ultrasound primarily have involved

the nebulization of samples for analysis by atomic absorp-

tion (63,64). One European group has devised a separation

technique using ultrasonic standing waves of varying acoustic

energy. They have found that particles of different size

move through a medium at different rates as a function of

the acoustic energy of the ultrasonic standing waves; separ-

ation is achieved with cyclic repetitive changes in the

energy of the waves (65).






27

Cavitation


Ultrasound is defined as wave lengths which are above

the audible range for humans and is arbitrarily set at values

greater than 16000 Hz (66). As an ultrasonic wave is propa-

gated through the medium, usually water contained in an

ultrasonic cleaning bath, the particles of the medium are set

into oscillation, thereby producing regions of compression

and rarefaction (67). A negative pressure exists at the

region of rarefaction, and if the pressure drops below the

threshold of the strength of the liquid, a cavity will be

formed. This cavity may be a true void or may be filled

with gas or liquid vapor. The gas-filled cavities are the

most common types, unless the liquid has been degassed (66,67),

and, due to the refractance of light, are observed as shiny

silver bubbles. As the next ultrasonic wave passes through

the medium, the cavities collapse from the exerted compres-

sion forces, and a hiss is audible as the cavities implode.

This process is known as cavitation and is generally thought

to be responsible for the cleaning action of ultrasonic

baths in addition to the catalytic effect of ultrasound on

chemical reactions (67).

The induction of chemical reactions by cavitation is

thought to be due in part to the considerable mechanical

forces which are liberated with the destruction of the void,

forces which are sufficient to produce lattice defects in

crystal structures (67). These effects are manifest in

reactions with metals, for example, Grignard reactions








involving alkylhalides and magnesium or lithium are signifi-

cantly ameliorated with ultrasound (56,68). Secondary

effects of cavitation are emulsification and macroscopic

heating (61,66,67,69) due to the pulsation of resonance bub-

bles. Periodic pulsation occurs when, during the compres-

sion phase, the bubble does not completely collapse and

begins to resonate with the ultrasonic frequency. Rapid

motion and collisions between the bubbles is the result.

Ultimately, the bubble attains the optimum size for the par-

ticular ultrasonic frequency, maximum vibration occurs, and

local pressures can exceed the hydrostatic pressure by a

factor of 150,000 (67). Griffing (70,71) suggested that high

temperatures occur within the pulsating bubbles as a result

of adiabatic heating. The propagation of ultrasonic waves

is considered an adiabatic process because changes in pres-

sure and density are so rapid that thermal energy cannot be

transferred into the surrounding solution. During the com-

pression phase of the process, intense heats build within

the bubble, heats which are sufficient to catalyze reactions

in the vapor that has penetrated into the cavity (67). The

larger resonating bubbles attain even greater temperatures.

A third effect of ultrasound on reactions is accomplished

via electrochemical effects which occur in the cavitation

void in the initial stage of its formation. Frenkel (72)

has proposed that when cavitation occurs, ions are distrib-

uted on the faces of the voids, with one high probability

distribution being the case where ions of one charge predom-

inate on one wall and ions of the other charge congregate on








the other. In essence, a capacitor is formed in solution.

An electrical discharge in a cavitation bubble, which can

occur while the distance between the walls of the cavity is

not great and the vapor pressure is low, will generate free

radicals, solvated electrons, molecules, and ions.

Weissler (73) established the presence of hydroxyl radicals

when water was exposed to ultrasound, and Prakash and

Prandey (74) suggested that ultrasonic cleavage of aliphatic

halogens and aromatic rings in aqueous solutions was initiated

by the hydroxyls.

There are several factors which limit the cavitation

process (66). In general any action which increases the

strength of the solution will decrease the extent of cavita-

tion which occurs, including pressure and viscosity effects.

Viscosity is self-explanatory in that changing to a more

viscous solution with high cohesive forces will in effect

increase the solvent strength, thereby requiring more acous-

tical power before cavitation is initiated. The effect of

increasing atmospheric pressure on the solution is less

obvious. It has been found that the production of cavitation

in liquids is very much dependent upon the undissolved gas

content of the liquid because microbubbles of these gases

act to seed the formation of the cavities (66). Increasing

the pressure on the solution tends to dissolve the gas,

thereby removing the seed nuclei. Increasing the temperature

of the solution above a certain point has the same overall

effect of decreasing cavitation. Up to a particular temper-

ature, the concentration of nuclei is actually increased, and






30

this combined with the lower viscosities at higher tempera-

tures tends to increase cavitation. Ultimately, however, an

increase in the vapor pressure of the liquid with increasing

temperature will result in the formation of larger bubbles

which can no longer remain within the resonating liquid.

Thus the gas will be forced out of the liquid, and cavita-

tion will cease. Degassing the solvent also has been found

to decrease the extent of cavitation such that vacuums must

be applied to the liquid in order to resume cavitation. At

any particular ultrasonic frequency, a threshold of acoustic

power is necessary to induce cavitation, the absolute value

of which varies greatly with the history of the cavitation

medium (66,67,69), and, in general, for a given frequency

higher acoustical powers result in greater cavitation inten-

sities (75). The increase in ultrasonic frequencies ultim-

ately halts cavitation because the rarefaction period becomes

too short to allow the formation of voids (66,67). Lastly,

cavitation gradually decreases with length of sonication

time due to degassing of the solvent. In order to maintain

a high extent of cavitation within an ultrasonic bath,

either the water must be aerated, or fresh aerated water

must be added (66).



Reversed Phase Stationary Phases
from Ultrasonic Cavitation


Bondjonk and Han (60) showed that silanes could be

coupled over lithium wire in the presence of ultrasonic

waves as shown below.









2R SiC1 ~L ) R3Si-SiR
3 Li 3 3


It is reasonable to assume that RP stationary phase bonding

could also be achieved using ultrasonic cavitation, and

optimism regarding the potential for the production of

analytically superior stationary phases is not unfounded

considering the vigorous nature of cavitation process. The

pressure shocks attained with cavitation could serve to dis-

tribute the alkyl chains more evenly over the surface of the

core material and could also help to drive the endcapping

reagents deep into the pores of the silica surface. Both

occurrences would result in a more uniform stationary phase

material with higher chromatographic efficiencies. It was

the objective of the following experiments to test the

feasibility of producing stationary phases with ultrasound

and to compare the ultrasonic products with traditional

refluxed phases. An investigation of the effects of reac-

tion times, acoustical energy, and reaction temperatures was

performed as was a study of reproducibility of the ultra-

sonic reaction process.














CHAPTER III

ULTRASOUND REACTIONS WITH SILICA



General Experimental Conditions


Reagents

All solvents were HPLC grade from Fisher Scientific

(Fair Lawn, NJ) with the exception of MeCl2 and diethylether

which were reagent grade. The MeCl2 was distilled from P205

immediately prior to use.

Chromatographic test solutes were dissolved in HPLC

grade MeOH and were used without purification as purchased

from the following:


Phenol Mallinckrodt (St Louis, MO)

Acetophenone Fisher

Nitrobenzene Eastman (Rochester, NY)

Benzene Mallinckrodt (Paris, KY)

Toluene Fisher

Ethylbenzene Fisher

n-Propylbenzene Alfa (Danvers, MA)

n-Butylbenzene Eastman

Naphthalene Eastman

2-Aminonaphthalene Aldrich (Milwaukee, WI)


The silanes and lutidine were purchased as follows and

were used without further purification:






33

Dimethyloctadecylchlorosilane Petrarch (Bristol, PA)

Trimethylchlorosilane, 98% Fluka (Hauppauge, NY)

2,6-Lutidine, 98% Fluka


All chromatographic silica was from a single lot of

Davisil [W. R. Grace, (Baltimore, MD)] synthetic amorphous

silica, grade 641Lcox182, with an absolute surface area of

300 m2/g, a mean particle diameter of 20-30 Um,an average

pore diameter of 150 A, and a mean pore volume of 1.10 mL/g.

The silica was leached in 0.1 M HNO3 for 24 h at reflux tem-

perature, washed with copious volumes of water, and dried at

2500C under vacuum for 24 h prior to use.


Instrumentation

All HPLC analyses were performed on a Spectra-Physics

(San Jose, CA) 8700 or 8700XR system fitted with a Beckman

(San Ramon, CA) model 153 fixed wavelength (254 nm) UV detec-

tor and a Houston Instruments (Austin, TX) model D500

recorder. The methylene selectivity and k' values for the

seven columns, which are presented on Page 95 were collected

using a Kratos (Ramsey, NJ) Spectroflow 757 variable wave-

length absorbance detector operated at 254 nm. All samples

were injected via an Altex (Beckman) injector fitted with a

5 uL loop.

All CHN analyses were performed by the departmental

microanalysis service and were performed in triplicate to

ensure accuracy of the results unless otherwise stated. The

calculations of bonding densities were performed using Equa-

tion 1 with 347.1 g for the molecular weight of silane and








300 m'/g as the surface area. The scanning electron micro-

graphs of the reaction products were generously provided by

Dr. John Novak of Alcoa. Magnification is unknown.

The ultrasonic baths used were Bransonic (Shelton, CT)

model B32 cleaners operating at a frequency of 55 kHz and

were used without any modification. Reactions were performed

by suspending flasks so the level of the reactants was below

the level of water contained within the bath.

Columns were packed with a Shandon (Keystone Scientific,

State College, PA) HPLC slurry packer, in a downward direc-

tion at 6000 psi unless otherwise stated. Column blanks of

4.6 mm internal diameter (id) stainless steel were fitted

with column ends and 2 pm frits from SSI (State Coollege, PA).

All columns were approximately 15 cm long.


Reaction Conditions

A system was designed which allowed fresh aerated tap

water to be continuously pumped into the ultrasonic bath.

This was necessary in order to ensure the integrity of the

cavitation as degassing occurred with the passage of time,

and to ensure that the temperature within the bath was con-

stant and well below reflux temperatures. Sensors continu-

ously monitored and controlled the water level within the

bath by turning on and off a pump as necessary. The sche-

matic for the water level monitoring device may be found in

the Appendix. To allow stirring of the reaction within the

bath, a rod was attached to a motor. A magnet was installed

onto the end of the rod and was submerged in the bath. As








the motor turned, the mounted magnet rotated a stirring bar

within the reaction flask.

The reagents for the reactions were mixed in an Aldrich

(Milwaukee, WI) glove bag which had been well purged with N2,

and a dry N2 atmosphere was maintained at all times within

the reaction flasks. All reaction glassware was silanized

using HF and trimethylchlorosilane to ensure that the C18

silanes would bond to the silica rather than to the glass.

To ensure that all reactions proceeded to completion, an

excess of octadecyldimethylchlorosilane to silica based on

5 moles OH-/m' silica was added unless otherwise stated. A

twofold stoichiometric excess of 2,6-lutidine relative to

moles of silane was added as a scrubber for the HC1, and

approximately 50 to 100 mL of solvent was added depending

upon the quantities of silica used. All reactions were per-

formed in dual arm flasks so that CHN samples could be

removed periodically from the mixture without disturbing the

N2 flow into the flask. Refluxed reactions were implemented

in silicone oil baths, were stirred using a magnetic stirrer,

and were run simultaneously with the ultrasound where appro-

priate.

The reaction products were washed with a series of sol-

vents as discussed in Chapter I. The phase material was

then dried at room temperature under vacuum for 24 h and end-

capped with trimethylchlorosilane using the same procedure

as the C18 bonding, and washed and dried.








Precision of Elemental Analysis


Because the CHN results were so vital to the study, the

precision of the departmental analysis service was monitored

by submitting samples from a single lot of stationary phase

material over a period of some months. The values reported

are presented in Table 1. For the 48 samples analyzed, the

values for %C ranged from 5.64 to 6.38, and the mean, median,

standard deviation, and relative standard deviation were

6.01 %C, 5.99 %C, 0.18 %C, and 2.95%, respectively. The 95%

confidence interval for the sample was calculated to be

6.010.051 using Equation 13 (76) where,for populations of

more than 30 samples, the 95% confidence interval is given by


1.96 os
i c -(13)
/ N

where os = standard deviation
X = mean
N = number of samples


The values ranged from 5.93 to 6.22 %C for samples analyzed

on a single day, and the mean and standard deviation for the

measurements were 5.95 and 0.12 %C, respectively. These

studies indicate that there is an absolute error in the CHN

measurement of 0.37 %C, and thus, for heavily loaded sam-

ples, the relative error is low and the process suitable.

For lightly loaded samples, however,the relative error is

substantial, and the values reported may not be accurate.

For this reason, all values of %C were reported uncorrected

for C content in the starting material, i.e., the bare silica.







Table 1

CHN Precision



Date of Date of
Number Analysis %C Number Analysis %C

1 1/16/85 5.82 25 1/21/86 6.12
2 1/24/85 6.06 26 1/21/86 6.13
3 1/30/85 5.83 27 1/21/86 6.16
4 2/11/85 5.91 28 2/26/86 6.22
5 2/20/85 5.91 29 2/26/86 6.31
6 3/26/85 6.30 30 5/2/86 5.75
7 4/1/85 6.14 31 5/2/86 5.75
8 5/8/85 5.64 32 5/6/86 6.01
9 7/1/85 6.24 33 5/6/86 5.98
10 7/3/85 6.31 34 5/6/86 6.00
11 8/19/85 5.76 35 6/2/86 6.04
12 9/11/84 5.73 36 6/2/86 5.85
13 11/19/85 6.38 37 6/2/86 5.94
14 11/26/85 6.04 38 6/4/86 5.94
15 11/26/85 5.99 39 6/4/86 5.86
16 11/26/85 5.96 40 6/4/86 6.08
17 12/17/85 6.21 41 6/17/86 5.98
18 12/17/85 5.93 42 6/17/86 6.08
19 12/17/85 6.08 43 6/17/86 5.76
20 12/17/85 6.22 44 6/17/86 5.92
21 12/17/85 6.15 45 6/17/86 5.99
22 12/18/85 6.20 46 6/26/86 5.84
23 12/18/85 5.86 47 6/26/86 5.89
24 12/18/85 6.15 48 6/26/86 5.86

Mean = 6.01 Standard Deviation = 0.18
Median = 5.99 Relative Standard Deviation 2.95%





38

Ultrasound vs. Refluxed C18 Silica



Experimental

For this experiment tandem reactions were run to produce

C18 silica to compare the efficiency of ultrasound stationary

phases with traditional refluxed phases. The material was

reacted by mixing the following:


Reagents Refluxed Ultrasound

g silica 13.37 16.38
g silane 12.61 12.39
g lutidine 7.64 7.54
mL MeC12 100 100


Samples were taken for CHN at 2, 15, 20, 24 and 30 h and

washed and dried in the same manner as the bulk product. A

sample of starting material, bare leached silica, was also

submitted for analysis.

Half of each batch of the C18 material was retained for

further experimentation, and the rest was endcapped via the

respective methods by mixing the following:


Reagents Refluxed Ultrasound

g C18 silica 7.99 8.24

g silane 3.67 5.20
g lutidine 7.60 5.60
mL NeC12 50 50


The reactions were allowed to proceed for 24 h, after which

time the material was treated as before and submitted for

elemental analysis. Samples of virgin silica and both





39

endcapped materials were submitted for scanning electron

microscopy to verify that the particles were not being

crushed or altered via the rigorous ultrasonic process.

The pre-endcapped and endcapped materials were slurried

in CHC13 and packed into columns with 50/50 MeOH/CHCl3, MeOH,

and 50/50 MeOH/H20 as packing, conditioning and equilibration

solvents, respectively. The columns were equilibrated with

50/50 MeOH/H20 until the retention times for the toluene test

solute were constant. An h vs. flow rate curve was run on

the endcapped columns to test the efficiencies of the two

columns, and a test mixture of phenol, acetophenone, nitro-

benzene, benzene, and toluene was separated on all four of

the columns.


Results

The CHN results from the C18 bonding reactions are shown

in Figure 4. The material plotted at time = 0 is the leached

silica. The results at this point indicate that the refluxed

reaction was slightly more efficient at 17.8 %C than the

ultrasound with 17.4 %C. Unfortunately,only single samples

of the materials were analyzed for CHN, so the statistical

differences between the final %C of the materials cannot be

discussed. After endcapping, the carbon content of both

materials decreased, the refluxed material containing

16.2 %C, and the ultrasound 15.6 %C. This decrease is most

probably due to a loss of polymeric material that was physi-

sorbed but not actually bound to the surface during the end-

capping process. The surface coverages calculated from












































0
401








, r03
0

4C H
cCC
C









C) 0






-CI




CO

HL,














O

x <

Li
0 0
Xj
li- n


Kv


I^
como\
NN--" l


I


f-



I-
Ld


z
0
0


VL
o_
nu
0-









0


I









Equation 1 show that the two materials were nearly equivalent.

The value for the refluxed C18 silica was 2.84 pmoles/m' and

2.72 Lmoles/m' for the ultrasound material.

Chromatographically, however, the endcapped materials

were vastly different as is evident from the h vs. flow rate

curves plotted in Figures 5 and 6. Values for the reduced

plate height were calculated from Equations 4, 5 and 9. The

refluxed column attained a minimum h value of 3.4 at

0.3 mL/min which is quite reasonable by today's criteria that

a column with a reduced plate height of 2 to 6 is considered

a good column (41,45), and a column with reduced plate

heights of less than two is exceptional.

The endcapped ultrasound column was superior to the

refluxed column by a factor of two with an h value of 1.5 at

0.3 mL/min. A maximum value of 2.3 was attained at

3.0 mL/min (the pump flow limit for this column) compared

with 6.3 at 2.0 mL/min for the refluxed. This is an impor-

tant aspect to the commercial LC market because of the two

columns, the ultrasound would have significantly reduced

analysis time with superior efficiency. The test mixture

was separated in the order phenol, acetophenone, nitroben-

zene, benzene, toluene as shown in Figures 7-10. The reduced

plate heights on the columns for the toluene peaks were as

follows:


where EC = endcapped,

































0
O



0 C) U~
U X
0 II


C C



0 o



U .0

> 0E
C *


c.0 c
C.0 C
nC 0

oa
O L)
,-J (








0 a




Ut




Co

tCi

























U











LL.
Ld
0













0

LJ
-Ir



w
L_


I I I I I
Uw LO 11 N


C
E


-N
I-


I,


































> 0
> -C fl

Z; CD C

u 0 ) I



; C
o3 -c

-C


. D

) U -U








4O
CCC)
C *

rO EU





CCC


C) C-H C
3 *0








CO -




Ml

*H



























Figure 7. Test mixture on the non-endcapped
refluxed C18 silica column.
Mobile phase = 50/50 MeOH/H20.





REFLUXED C18 SILICA


0 4 8 12 16 20 24 28
TIME (min)


AUFS


J



























Figure 8. Test mixture on the endcapped
refluxed C18 silica column.
Mobile phase = 50/50 MeH/H.
Mobile phase = 50/50 MeOH/H20.








ENDCAPPED REFLUXED
CIg SILICA


0 4 8 12 16 20 24


AUFS


TIME(min)





























Figure 9. Test mixture on the non-endcapped
ultrasound C18 silica column.
Mobile phase = 50/50 MeOH/H20.





52

LOW POWER ULTRASOUND
C18 SILICA




























-[ A


I I I
O 4 8 12 16 20
TIME (min)


I4
24 28


AUFS































Figure 10. Test mixture on endcapped
ultrasound C18 silica. Mobile
phase = 50/50 MeOH/H20.









Column h

Refluxed C18 2.4
EC Refluxed C18 2.4
Ultrasound C18 6.8
EC Ultrasound C18 2.5


The value for the nonendcapped ultrasound column is quite

high, However, this column was used extensively prior to this

test. Scanning electron micrographs of the materials are

presented in Figures 11-13. The photographs clearly demon-

strate that the irregular silica was undamaged in the ultra-

sound process. In fact,a comparison of Figures 12 and 13

indicates that more fines were present in the endcapped

refluxed sample. The possibility exists that the vigorous

ultrasonic process actually dislodged fines which had adhered

to larger silica particles so that the fines could be removed

during the washing process.



Reproducibility of Column Packing


In order to ensure that the differences in the two

columns in this study were actually due to differences in the

phase materials and not to variations in the packing proce-

dure, four columns were packed with the nonendcapped ultra-

sound material from the above experiment. The packing pro-

cedure was standardized as follows:


1) Slurry 2.0 g phase material in 30 mL CHC13 and
sonicate to disperse.

2) Pour the slurry into the reservoir and fill to
volume with CHC13.





































Figure 11. Leached silica. Magnification
unknown.



































Figure 12. Endcapped refluxed C18 silica.
Magnification unknown.


































Figure 13. Endcapped ultrasound C18 silica.
Magnification unknown.









61















r








r



















~la~:







j
I
c
rjr


c


















1'






62

3) Pack at 6000 psi with 150 mL CHC13 in the upward
direction.

4) Condition the column using 150 mL 50/50 MeOH/CHC13
in the downward direction.

5) Equilibrate the column with approximately 150 mL
MeOH, use more MeOH if a CHC13 odor is still appar-
ent.

6) Remove any excess packing material, finish the
column head by filling any divots in the column
bed and leveling the bed with the top rim of the
stainless steel tubing, and install the end fit-
tings.

7) Equilibrate the column on the HPLC system until
retention times and baselines are consistent.
Generally, this required 200 mL of solvent.


The Shandon column packer malfunctioned during the pack

ing of the second column. However, this column was finished

and analyzed with the others. The columns were equilibrated

with 40/60 ACN/H20 and evaluated using results obtained from

repetitive injections of a toluene test probe at 0.5 mL/min.

The results for the four columns are presented in Table 2

and in Figure 14. The error bars indicate 1 standard devia-

tion. The mean reduced plate height, standard deviation, and

relative standard deviation for the individual columns are

presented below the data for each column. These values

indicate the precision of repeated injections on a single

column. The large deviation in values for column 2 are, no

doubt, due to the packer malfunction. In fact, the precision

of the last two columns, which were packed after the check

valves and high pressure seals were replaced, far surpass

even the first column, indicating that the last two columns

were more efficiently packed. That column 1 was more poorly








Table 2

Column Packing Statistics
for Pre-endcapped Ultrasound C18 Silica



COLUMN 1

B/A 1.06 1.08 1.30 1.32 1.31
N 1910 2010 2220 2180 2300
H (cm) 0.0079 0.0074 0.0068 0.0069 0.0065
h 3.2 3.0 2.7 2.8 2.6

h = 2.80.27
Std dev = 0.22
RSD = 7.9%


COLUMN 2

B/A 1.16 1.19 1.24 1.34
N 3390 3420 2720 2509
H (cm) 0.0044 0.0044 0.0055 0.0060
h 1.8 1.8 2.2 2.4

S= 2.00.48
Std dev = 0.30
RSD = 15%


COLUMN 3

B/A 1.05 1.03 1.03 1.05 1.08
N 2300 2500 2460 2370 2410
H (cm) 0.0065 0.0060 0.0061 0.0063 0.0062
h 2.6 2.4 2.4 2.5 2.5

h = 2.50.10
Std dev = 0.084
RSD = 3.4%


COLUMN 4

B/A 1.11 1.05 1.09 1.07 1.05
N 2590 2690 2640 2530 2600
H (cm) 0.0058 0.0056 0.0057 0.0059 0.0058
h 2.3 2.2 2.3 2.4 2.3

h = 2.30.088
Std dev = 0.071
RSD = 3.1%











































>1 *H
-3 0






*H C

*H
-C
*H $-i
U (0i






3 -o
-U C
Om

di
+ip






o-,

CO

u4

c-H

C- -i
3U









0 *
CO
E -a






4-4

CO

CCi-0
-4


30
U] ^







30


fcr












WwJ


L-)



o <
GZ0



O4


I I I I I I I
rO c'






66

packed may explain the increase in tailing experienced midway

through the run. Packing additional columns from this sta-

tionary phase material would have proven useful.. However, the

supply was exhausted with the fourth column. Repacking col-

umns with used material would not have been valid, because

microbial growth could have occurred during use of the phase,

or crushing of the material may have occurred during the

initial packing process. In either case, the material would

no longer have had the same characteristics as the material

in the original four columns.

A comparison of the mean reduced plate heights of 2.8,

2.0, 2.5, and 2.3 for the four columns, respectively, indi-

cates the reproducibility of packing multiple columns from

the same material. The Student's T (77,78) calculation

which indicates that the population mean will fall within

the 95% confidence level limits as given by


U= A (14)
ZN

where X = mean
t = Student's T value
N = number of measurements
s = standard deviation


was used to generate the ranges following the mean reduced

plate heights in Table 2. Inspection of these ranges shows

that all of the columns are within the ranges of the other

columns with the exception of Columns 1 (lower limit 2.5)

and 4 (upper limit 2.4). These values may not be statistic-

ally different considering the increase in asymmetry (B/A

value) of the peaks obtained for Column 1 as the experiment






67

progressed. If Column 2 is neglected, the range of h values

given by the 95% confidence interval for all three of the

remaining columns collectively does not approach the range

of h values obtained between the endcapped ultrasound and

endcapped refluxed columns. This indicates that the differ-

ences in the two columns were due to differences in the pack-

ing materials rather than variations in the packing process.



Effect of Acoustic Power on C18 Reactions



Experimental

The relative acoustic powers of two Bransonic 55 kHz

cleaners were measured by submerging a piezoelectric trans-

ducer in the water of an operating bath. The resultant volt-

age was measured with an oscilloscope. Assuming that the

frequencies of the baths were equivalent, then the impedance

of the transducer would be constant for the measurement, and

the relative acoustic power of the baths could be related by



Z -- and Z = (15)
1 2


and


Z1 = Z2 (16)


then


E 2 (17)
1 2






68

where P = acoustical power
E = measured voltage
Z = impedance


Absolute values could not be obtained for either bath. How-

ever, one bath was found to be three times more powerful than

the bath which had been used in the original C18 bonding

reactions discussed in the previous sections. Tandem reac-

tions were run in the two baths by weighing and mixing the

following:


Reagents Low High

g silica 10.43 10.02
g silane 10.55 11.30
g lutidine 7.93 6.73
g MeC12 117.84 116.09


Samples were removed for CHN analysis at 6, 9, 22, and

24 h, and a sample of leached silica was submitted for

comparison. The material was treated as previously described

and then endcapped by reacting the following:


Reagents Low High

g C18 silica 7.97 11.24
g silane 6.30 6.33
g lutidine 8.91 8.24
g MeC12 134.55 134.55


Samples were removed for CHN at 3, 6, 11, 22, and 24 h, and

additional samples of the C18 material produced in the bond-

ing reaction were resubmitted for comparison with the endcap-

ping products.






69

A column was packed with the endcapped product from the

high power bath using the same series of solvents and process

as in the packing reproducibility study. The column was

equilibrated with 50/50 MeOH/H20, an h vs. flow rate curve

was obtained with toluene as the test probe, and a test mix-

ture sample was also separated. The column was then equilib-

rated with 40/60 ACN/H20 and the h vs. flow rate curve was

rerun. A column was also packed with the non-endcapped high

power material, equilibrated with 50/50 MeOH/H20, and was

used to separate the test mixture.


Results

The CHN results in Figure 15 indicate that for all sam-

ples the higher acoustical power produced greater carbon

loading. One interesting point to observe is that the 24 h

product of the bonding reaction, which was identical to the

sample listed at time = 0 for the endcapping reaction, has

quite different %C values. Six samples were removed from

each batch of 24 h C18 bonding material, half of which were

submitted for CHN analysis with the bonding samples and half

with the endcapping samples. The %C values reported were as

follows:


High Power Ultrasound Reaction

Finished C18 bonding product Initial endcapping material

27.5 25.1
27.0 25.1
25.1 24.2



































0)
co
C C

--
C 0
3 ri
o Co
Su


0
C. U


o
c-t



o c




CO
m 3














E CO
Co












CO
Co
44 C






72

The value of 24.2 was discarded with the Q test with 90% con-

fidence.


Low Power Ultrasound Reaction

Finished C18 bonding product Initial endcapping material

13.6 12.9
17.9 12.6
19.8 12.6


For this material all of the values were retained with the

Q test. Resubmission of the samples was not performed

because after the CHN analysis the integrity of the sample

could no longer be verified, and the results reported would

not be reliable.

A significant improvement in extent of bonding was

noted with higher acoustic power. For the endcapped high

power sample with 16.3 %C, the surface coverage is

2.87 umoles/m'. This compares to 13.07 %C or 2.19 umoles/m'

for the endcapped low power sample. These results indicate

that further studies of the effect of acoustic power are war-

ranted, but, unfortunately, this was beyond the scope of

these experiments.

The h vs. flow rate curves for the endcapped high power

product are presented in Figures 16 and 17. An acceptable

minimum h value of 2.3 was obtained at 0.3 mL/min with

50/50 MeOH/H20. In order to improve this value, a less

viscous 40/60 ACN/H20 mobile phase was substituted for the

MeOH/H20. The viscosity of the mobile phase is important in

efficiency considerations, because reduction of the solvent





























O
N

-V '~-
C 0
3- O
0001


COC



U v
a)

> 4-1 Lfl
- -




30
w a) II





SUS
>C, CO

.CO
o ro
- 'H
aa

T -H





00E
c-s:
4 CO






ODC
*Hc
Crl
0 -C 0

1C
300



C0 C CO)
-cCO
CO C -


& 0
0a) C-
00 4-1 (9)



'C


a)
3-
CO
*r-l






On
C:,






































or
C1-
-C)

I


L-

UD



Q-


E
E
w



I-
u
-J


I I I I
WD W) -






























0


CD
2
>0 0
04 CO <




:io



U 3
O 0 II
coo

0 0
(n






0 0 E0

0-I





c 3





















ai
4i-

CC
4:4m




























LU





0-
I

0



0z
LO
nob
< <



LiD


I I I I I
LO ;1 --1 (


E
E


I-


LI-






77

viscosity tends to increase the diffusion coefficient of a

solute. This will serve to decrease the resistance to mass

transfer term, which is C of the van Deemter equation pre-

sented in Equation 6 of Chapter 1, by decreasing the time

required for the solute to penetrate through the mobile

phase to the interfacial boundary between the mobile and

stationary phases. This penetration is essential before the

solute can interact with the stationary phase. The viscosity

and diffusion coefficient of a solute for a 50/50 MeOH/H20

mobile phase are 1.4 centipoise and 0.38 x 10-5 cm2/s,

respectively, and for 40/60 ACN/H20 the values are 0.98 cen-

tipoise and 0.54 x 10-5 cm'/s (5).

The improvement with the ACN/H20 mobile phase is appar-

ent in Figure 17. The minimum h of 2.2 was attained at

0.6 mL/min, and the slope of the curve was decreased, thereby

permitting the use of the column with the ACN/H20 mobile

phase at higher flow rates with little loss of efficiency.

As seen in Figures 18 and 19, the test mixture was well

separated with a mean h value for toluene of 4.4 and 5.4 for

the pre-endcapped and endcapped columns, respectively. The

components eluted in the order presented above, i.e., phenol,

acetophenone, nitrobenzene, benzene,and toluene. Again,

these columns would be acceptable for commercial use.



Reproducibility of Ultrasonic Reactions


Because the coverages were not identical for the ultra-

sonic acoustic power study, an experiment to test the
































Figure 18. Test mixture on the non-endcapped
high power ultrasound C18 silica
column. Mobile phase = 50/50
MeOH/H 20.







HIGH POWER
ULTRASOUND
C18 SILICA


TIME (min)


AUFS


0 4 8 12 16 20 24 28 32 36
































Figure 19. Test mixture on the endcapped high
power ultrasound C18 silica column.
Mobile phase = 50/50 MeOH/H20.






ENDCAPPED
HIGH POWER
ULTRASOUND
C18 SILICA















_J _. !Li




. . \ -i


0 4 8 12 16 20


TIME (min)


AUFS


24 28








reproducibility of ultrasonic reactions was designed. For

this test reagent quantities, with the exception of MeCl2,

were matched to the values used for the high power reaction

in the above study, and three separate reactions were run.

The MeCl2 was varied by a factor of 0.5 to see the effect of

dilution of the reagents. The higher power bath was chosen

because increased C loadings of the phases would reduce the

relative error in the CHN analysis. The reaction between

the following proceeded for 22 h:


Reagents Reaction 1 Reaction 2 Reaction 3

g silica 5.01 1.84 1.78
g silane 5.65 2.08 2.07
g lutidine 3.37 1.27 1.24
g MeC12 116.09 43.60 27.55


Eight samples of the final product were submitted for CHN

from each of the three reactions, and the mean values are

presented in Figure 20 with the results from the same bath

used in the power study included for comparison. The mean

values and 95% confidence limits of 13.20.3 %C,

14.20.7 %C, and 13.80.5 %C compare very well considering

the error in CHN analysis and the inherent error in position-

ing the flask over the transducers. The concentration of

MeC12 seems to have little effect on C loading; the 13.2

and 14.2 values were obtained using high volumes of MeCl2,

and 13.8 and 16.4 (from the acoustical power experiment)

were obtained using low volumes.

































U >1
-Hl O

0 4Or l
CCC
C *




z 0

rc U


E H



CO


u 0 au

- E- C
S-1 u4 -i


C4 U CU

m *0 O

'H 44C 0



CQ Uo





C 0 e C

S(UC C C

ciuc



C3






o0
ar o-1 r
Ec























z
LLO
O
0 -


mo
QLZ

O
D 0
0 <

Qr r)


/////i N


S// / / / / / /


K / / / / / / / ^


Y / / / / i / / //






85

Ultrasound with Heat


It was beyond the scope of this research to alter the

ultrasonic frequency or the acoustic power of the individual

baths, both of which would have proved an interesting conclu-

sion to this set of experiments, instead, the effect of tem-

perature was investigated. Two reactions were run in the

low power bath for this study, one at 230C, performed during

the Power Study set of experiments, and one at 280C. The

five degree difference was chosen because the lower power

bath had begun heating during the long reaction periods, most

probably due to excessive wear. For this experiment a very

slow flow of incoming tap water was maintained into the low

power bath so that the heating occurred. The reagents were

as follows:


Reagents Ambient Heated

g silica 10.43 10.50
g silane 10.55 10.54
g lutidine 7.93 7.98
g MeC12 117.84 117.92


Samples of the final product were submitted for CHN, and

the results are presented in Figure 21. The difference in %C

is quite substantial, 12.7% for the low temperature and 16.4%

for the high. These results indicate that further studies of

the effect of temperature on ultrasound are justified because,

as presented in Chapter II, increased heat will increase

ultrasonic cavitation consequently producing a more energetic

system. It appears from this experiment that increased

































C





C










OM
u
.l)
0
4n











1-H
oj










0 )
*-C
3-3
InC










cr



COl















1-1



CU












w
0
LLJ
T- I-

i M
z <
V O Z S

0 a F1








cavitation does indeed give increased loading. An interest-

ing experiment would be to heat the higher powered ultrasonic

system, possibly with a closed circulation system. Fresh

water should still be pumped into the system to ensure that

cavitation is not diminished as degassing occurs within the

bath.



Commercial Refluxed C18 Column



In order to compare the efficiency of the ultrasonic

phases prepared in the experiments discussed above, a virgin

Brownlee (from Spectra-Physics) C18 10 cm cartridge column

(number A2348-030 17834) was equilibrated with 50/50 MeOH/H20

and tested for efficiency by running the h vs. flow rate

curve which is presented in Figure 22. This column was quite

efficient with a minimum h of 2.2 at 0.3 mL/min and a value

of 3.3 obtained at a flow of 2.5 mL/min. Both of the end-

capped ultrasonic phases, the low and high powered, run under

the same conditions compare quite favorably with the

commercial column. The efficiency of the low powered column

exceeded that of the commercial column (minimum h = 1.5) and

the efficiency of the high powered column matched that of

the commercial (minimum h = 2.2). A comparison of the steep-

ness of the slopes of the curves, important for most analyt-

ical applications where efficient separations must be attained

in a minimum of time, showed that the columns compared quite

well with the low powered column again superior (maximum

h = 2.3 at 3.0 mL/min) and the high powered column slightly






































ci

0 l




u
U

.0
YE



o O-











O
> o









-0 ( 1
U a
0) S



ccO
4-- 4 ,-

COC-


-O 4i II

0 00

SH 'cO
-C -J a-













N
N




4-r
-H




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs