DESIGN OF A WATER-DISPOSABLE
research grants
UI-00651
EC-00033
Clemson
University
-------
This report has been reviewed by the U.S. Environmental
Protection Agency and approved for publication. Approval
does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Pro-
tection Agency, nor does mention of commercial products
constitute endorsement or recommendation for use by the
U.S. Government.
-------
DESIGN OF A WATER-DISPOSABLE GLASS PACKAGING CONTAINER
Part I: Protective Oxide Coatings for Glasses by Chemical Vapor Deposition
Part II: Mechanical Properties of Water-Soluble Sodium Silicate Glasses
Part III: The Rate of Dissolution of Sodium Silicate Glasses in Aqueous Solutions
These interim reports (SW-llrg) on work performed under
Research Grant No. UI-00651 (Parts I and II) and under Grant No. EC-00033 (Part III)
to the Clemson University
were prepared by SAMUEL F. HULBERT, C. CLIFFORD FAIN, and MICHAEL J. EITEL
and have been reproduced as received from the grantee.
U.S. ENVIRONMENTAL PROTECTION AGENCY
1971
-------
An environmental protection publication in the solid waste
(
management series (SW-llrg).
For sale by the Superintendent of Documents, U.S. Government Printing Office
Washington, D.C. 20402 - Price $1.75
Stock Number 5502-0046
-------
FOREWORD
PACKAGING WASTES constitute a significant and growing fraction of
the total solid waste load, and a major component of this total
is the glass container.
In a study done for this agency, the Midwest Research Insti-
tute found that shipments of glass containers have grown from 20.2
billion units in 1958 to 29.4 billion units in 1966.] They pro-
jected a consumption rate by 1970 of 34.9 billion units and by
1976 of 45.7 billion units. During the period from 1966 to 1976
they expect the shipments of glass beverage containers to more
than double, as nonreturnable containers account for a larger and
larger share of the market.
We applaud the nationwide program by which glass container
manufacturers are supporting recycling activities. Clearly,
though, other avenues must be explored, including the concept of
reducing solid wastes at the source—reducing the quantities of
waste that will ultimately require storage, collection, transport,
processing, and disposal.
Under the authority of the Solid Waste Disposal Act of 1965,
amended by the Resource Recovery Act of 1970, research has been
undertaken for the study and solution of problems caused by
packaging wastes. The research work discussed in this report
illustrates the innovative and imaginative approach being brought
to bear on the problem. The development of a water-soluble glass
container, presented in this three-part work, has high potential
for promoting source reduction, and thereby improving the quality
of our environment.
Darnay, A., and W. E. Franklin. The role of packaging in solid
waste management, 1966--1976. Public Health Service Publication
No. 1855. Washington, U.S. Government Printing Office, 1969. 205 p.
m
-------
PREFACE
THIS REPORT was prepared by the Division of Interdisciplinary
Studies, Clemson University, Clemson, South Carolina, under
U.S. Public Health Service Research Grants UI-00651 and
EC-00033.
The work reported in Part I was performed in fulfillment
of the thesis requirements for the degree of Master of Science
in Ceramic Engineering by Martin M. Cooper under the direction
of Dr. C. Clifford Fain. The work reported in Part II was
performed in fulfillment of the requirements for the degree of
Master of Science in Materials Engineering by Charles Wesley
Jennings under the direction of Dr. Samuel F. Hulbert. That
reported in Part III was performed in fulfillment of the re-
quirements for the degree of Master of Science in Ceramic
Engineering by David Taylor Ballenger, also under the direction
of Dr. Hulbert.
IV
-------
PARTI
PROTECTIVE OXIDE COATINGS FOR GLASSES BY CHEMICAL VAPOR DEPOSITION
-------
ABSTRACT
A method of coating water soluble silicate glasses with oxide films by
chemical vapor deposition was studied. The work represents the initial phase of a
•
three year investigation to determine a feasible method of processing glass con-
tainers after use in order to relieve a source of solid waste currently reaching
gigantic proportions nationally.
A detailed description of the system was provided to include coating equip-
ment design features, input material requirements, and techniques of coating.
Particular attention was given to key process control parameters such as temperature
control, evaporation rates, and vapor transport procedures.
Coatings of silica, titania, and other oxides were readily obtained by
pyrolytic decomposition of organic ester vapors. The reaction kinetics of selected
materials were analyzed in terms of the process conditions used to derive them.
For example, the effects of deposition temperature, carrier gas flow rate, reactant
vapor concentration in the carrier gas stream, and substrate orientation were corre-
lated with deposition rates and coating efficiencies.
High quality coatings of both silica and titania were obtained on soluble
glass substrates by exercising proper judgment in the choice of process conditions
to be used. Oftentimes the best coatings were obtained at the expense of longer
•
coating times and lower efficiencies. The films produced, however, were found to
be continuous, highly resistant to various forms of corroding media, and adherent
to the surface of the substrate.
-------
The titania films were most readily applied as uniform thin films at a growth
rate of approximately 6000 A per hour between temperatures of 500 F - 800 F.
It was determined that the film structures were readily controlled by proper atten-
\
tion to process parameters. The silica films were generally applied under more
restricted conditions, typically between 1000° F - 1050° F at about 4000 A per
hour.
Both types of oxide coatings were found to produce undesirably high
residual stress levels, primarily because of a mismatch of thermal expansion
coefficients between the coating and the substrate. In addition, many coatings
exhibited microstructural variability primarily caused by uncontrollable coating
conditions such as poor control on vaporization and gas delivery processes,
impurities, substrate surface variations, localized temperature fluctuations, and
others. The variable coating microstructures in conjunction with high residual
stress levels are believed to promote a generally higher defect level such as
microcracks and pinholes, thus contributing to a premature breakdown of the
continuous barrier under water attack. The fact that tentative investigations
revealed a wide scatter in the times required for water to penetrate and dissolve
a coated specimen tends to amplify this observation.
Vi i
-------
PARTI
TABLE OF CONTENTS
Chapter Page
I. INTRODUCTION 1
II. REVIEW OF LITERATURE 3
A. Solid Waste Control 3
B. Soluble Silicate Glasses 4
C. Coating Considerations 6
III. OBJECTIVES OF RESEARCH 14
IV. EXPERIMENTAL PROCEDURE 15
A. Coating Apparatus Design
B. Other Design Considerations 25
C, Input Material Considerations 26
D. Deposition Techniques 31
E. Analytical Procedures 34
1. Microscopy 34
2. Phase Analysis 34
3. Coating Thickness and Uniformity 35
4. Coating Efficiency 33
5. Coating Stresses • • 40
V. RESULTS AND DISCUSSION OF RESULTS 42
A. Effect of Process Variables Upon Reaction Kinetics «~
1. Substrate Temperature 43
2. Coating Time at Different Temperatures 43
• * •
vm
-------
Chapter Page
3. Carrier Gas Flow Rate and Substrate Orientation. . . 46
4. Reactant Gas Concentration . . . , 46
B. Coating Stresses 48
C. Coating Uniformity 53
D. Substrate Considerations 54
E. Characterization Of SiO2 Coatings 58
F. Characterization Of TIO2 Coatings 62
G. Other Coatings 72
H. Initial Corrosion Studies 76
VI. CONCLUSIONS 78
VII. RECOMMENDATIONS FOR FUTURE WORK 80
LITERATURE CITED 84
APPENDICES
APPENDIX I. DETERMINATION OF PLATING VAPOR
CONCENTRATION IN THE CARRIER GAS STREAM. .
APPENDIX II. DETERMINATION OF COATING EFFICIENCY. ... 90
APPENDIX III. DETERMINATION OF RESIDUAL DEPOSIT STRESS . 92
IX
-------
PARTI
LIST OF TABLES
Table Page
I. Chemical Properties of Plating Compounds 30
It. Optical Properties of Coatings and Substrates 39
III. Mechanical Properties of Coatings and Substrates 41
IV. X-Ray Diffraction Data for Coatings Obtained 63
V. Chemical Durability of Coatings Obtained 66
-------
PARTI
LIST OF FIGURES
Figure Page
1. Schematic of the chemical vapor deposition apparatus 16
2. Photograph of the chemical vapor deposition apparatus 17
3. Schematic of the temperature control circuit 20
4. Plot of evaporation rates of tetraisopropyl titanate under
various conditions 23
5. Photograph of the dual furnace assembly 27
6. Schematic of the two substrate positions used 33
7. Analysis of film thickness and uniformity shown by interference
colors 36
8. Plot of deposition rates at different temperatures 44
9. Plot of deposition rates at two different temperatures for varying
coating time periods 45
10. Plot of deposition rates at different substrate orientations and
gas flow rates 47
11. Plot of deposition rates at different reactant vapor concentrations
at an overall constant flow rate 49
12. Photograph of residual stresses in a SiO2 deposit 50
13. Photomicrographs of residual stresses in a Ti03 deposit 52
14. Photograph of a representative deposition pattern for SiO2 deposits . 55
15. Photomicrographs of atmospheric attack upon a soluble silicate
surface at three time periods 57
16. Plot of deposition rates at different temperatures for SiOo deposits . 60
17. Photomicrograph of surface topography of a typical S«O2 deposit . . 61
Xi
-------
Figure Page
18. Photomicrograph of Cristobalite crystals 64
19. Photomicrograph of columnar growth habit of SiO2 Deposits .... 65
20. Plot of TiO2 structure at different temperatures and reactant 69
vapor mass flow rates
21. Photomicrographs of typical anatase microstructures 71
22. Photomicrographs of secondary growth features in TiO2 deposits . . 73
23. Photomicrograph of surface topography of amorphous TiO2
deposit 74
24. Photomicrograph of a TiO2~SiO2"TiO2 duplex coating 75
-------
CHAPTER I
INTRODUCTION
The great natural resources of our nation have been ingeneously utilized by
a resourceful people to create a vast industrial capability which today provides an
expanding population with the highest standard of living on earth. This epic growth
has been accompanied, however, by a constantly expanding quantity of material
wastes emanating from our nation's communities. By virtue of the current magni-
tude of the problem we are now urgently compelled to find new methods to process
or neutralize these wastes rather than simply digest or absorb them as in the past.
Our people can no longer afford to ignore the effects of belching factory smoke,
gaseous emissions from vehicles, foul liquid contaminants dumped into our streams,
and solid wastes piled up at refuse dumps or discarded along our roadways, for to
do so will subject much of our citizenry to potentially severe health hazards as well
as contaminate the beauty of our national landscape.
Recent reports (1) indicate that the nation has already taken conclusive
steps to control gaseous and liquid wastes. In contrast, however, it is pointed out
that we are just beginning to focus our attention on the problems related to solid
wastes. The Federal Government is now providing funds and leadership for research
and planning programs to curb the undesirable effects of inadequate solid waste
control. Essential to this effort is a method of processing incombustible, unreactive
glass containers after use. The potential impact of such a development upon solid
waste control may be readily visualized by considering that approximately 135 glass
-------
bottles and jars are produced each year for every person in the United States. This
amounts to about 27 billion glass containers per year (2).
This paper represents the initial segment of a projected three year study to
formulate a method of reducing the source of glass container waste by utilizing a
water soluble glass encapsulated in a thin, impervious coating which resists corro-
sion by the environments normally encountered by the glass packaging industry.
After the consumer empties the container, the continuous physical barrier may be
broken and the water soluble glass dissolved. It is anticipated that the decompo-
sition could be made to occur gradually over several months' duration, as at a
municipal refuse dump, or relatively rapidly, perhaps within several minutes, by
adjustment of the chemical and physical parameters of the base glass and its over-
lay coating.
This paper describes the use of chemical vapor deposition processes for
*
applying chemically inert metallic oxide coatings to soluble glass substrates. A
detailed analysis of the method is presented and its potential feasibility for vitreous
container waste control is discussed.
-------
CHAPTER II
REVIEW OF LITERATURE
A. Solid Waste Control
The problem of solid waste control in our country is rapidly approaching
critical proportions. This statement surprises many people, for generally speaking,
the public is not aware that there is a problem. A vivid illustration of public apathy
can be seen in the attitude toward the throw-away glass bottle which relieves the
producer, distributor, and consumer of handling costs and inconvenience, but cre-
ates more burdens for refuse collectors. Unfortunately, most people are not even
aware of the additional problems that have been created by this convenience item.
The refuse collector is painfully aware of his increased burdens, however. The use
of non-returnable glass bottles is increasing (3), and no immediate solutions are in
sight, although some states have begun to introduce legislation (4) in an effort to
curb their production.
It is estimated (5) that glass comprises five to ten percent by weight of all
refuse ending up on municipal dumps. Of course, scrap glass represents a major
solid waste problem because, unlike much refuse, it is non-degradable. In addi-
tion to being aesthetically objectionable, discarded glass containers may harbor
rodents and catch other vermin. They may catch and hold water in which mosqui-
toes may breed and subsequently transmit malaria, yellow fever, dengue, mosquito-
borne encephalitis, and filariasis (6)-
-------
At present, non-combustible glass containers can best be disposed of by
sanitary landfill techniques. Obviously, this does not represent the complete
answer, for convenient sites are not always available.
In a major effort to generate fresh approaches directed toward solution of
the problem, the Federal Government passed the Solid Waste Disposal Act, which
became law on October 20, 1965 (1). The Congress assigned responsibility for
administering the Act to the Secretary of Health, Education and Welfare and to
the Secretary of the Interior. It is interesting to note that similar legislation to
combat water and air pollution has been in effect for twenty, and twelve years,
respectively (1).
This work represents one facet of the total program which has been imple-
mented in selected universities and industrial research facilities throughout the
country.
B. Soluble Silicate Glasses
Glass containers are carefully formulated today to yield high chemical dura-
bility. The resistance which glass offers to various corrosive agents is a property of
great practical significance, according to Morey (7). This eminent glass authority
further points out that the chemical durability of glass is oftentimes the chief reason
for its preference over competing materials. As an example, he cites the use of
glass containers which are used in enormous numbers for the distribution of commod-
ities ranging from milk to medicine to acids. In this respect, he concludes that the
superiority of glass leaves it without a competitor.
-------
Of course, the hypothesis of this paper requires that a vitreous container
must be soluble in order to be readily eliminated after use. Thus, a water soluble
glass, with satisfactory optical, mechanical and chemical properties for container
use is required. Fortunately, a vast market exists today which makes use of water
soluble glasses. Vail (8) discusses the diverse uses for these alkaline silicate
materials, which may serve such purposes as adhesives, cleaners, cements, defloc-
culants, and protective coatings.
Vail (9) points out that the largest quantity of soluble glass produced for the
soluble silicate industry is the eutectic composition between sodium disilicate and
quartz, which is readily dissolved by water in less than one hour. Potassium sili-
cate glasses are also known to be readily soluble.
The properties of the alkali silicate glasses have been investigated, but
generally the reasons were for scientific curiosity rather than to achieve any prac-
tical objective. Their properties tend to change continuously as a function of com-
position rather than show the abrupt type of discontinuity characteristic of hetero-
geneous crystalline systems. The solution rate depends primarily upon the ratio of
alkali to silica in the glass (9).
The alkali silicate glass system would offer some inherent advantages if
adopted for glass container production. As reviewed by Vail (8), sodium silicate
solutions are now added to polluted waters, as a flocculant in the settling and fil-
tering steps of water purification. At the same time, these solutions tend to form
a thin film on the inside of water pipes which would protect them against water
-------
corrosion. In addition, it is noteworthy to add that no adverse effects from potable
waters which have been sodium silicate treated have been detected.
The production of alkali silicate glasses is simple and cheap. Most of the
material made in this country is easily produced by firing mixtures of sand and
sodium carbonate in open-hearth type furnaces of regenerative or recuperative
design with gas or oil. The clear glass reaction product is then drawn directly into
water to produce silicate solutions (9). The objective of this study is to use a
similar procedure with the exception that the vitreous silicate will be made into a
container and used prior to putting it into solution.
C. Coatmg Considerations
The choice of a coating mechanism for this work entailed the critical eval-
uation of several methods. These methods can best be categorized in terms of the
size of the particles which are being used to produce the coating structure (10).
They are (a) atomic species; (b) particulate material; and, (c) sheet material. On
this basis, the following categories provide a reasonable framework for discussing
coating methods,
1. Atomic species processes
a. Vacuum evaporation
b. Cathode sputtering
c. Electrode position
d. Chemical vapor deposition
2. Particulate material processes
a. Flame and plasma spraying
b. Enamelling
c. Painting
d. Trowelling
-------
3. Sheet material processes
a. Laminating
b. Cladding
Chemical vapor deposition was used as the coating mechanism in this study
because it is believed to be the best method for producing thin, impervious oxide
coatings of desired quality. Lesser considerations governed ttie choice to some
3xtent, such as potential adaptability to other materials systems, construction,
•naintenance, and production costs, and scale-up considerations.
Chemical vapor deposition may be defined (11) as a method of plating in
/vhich deposits are produced by heterogeneous gas-solid chemical reactions at the
surface of a heated substrate. A volatile compound of the element or substance to
DC deposited is vaporized and the vapor thermally decomposed, or reacted with
Dther gases or vapors at the substrate to yield non-volatile reaction products which
deposit on the substrate, as a coating.
Blocher (12) refers to chemical vapor deposition, vacuum evaporation,
cathode sputtering, and electrodeposition as "molecular forming" processes, that
is, they tend to build up coatings or deposits atom-by-atom upon a substrate. As
a result, these processes have in common the potential of yielding dense deposits
of controlled thickness and orientation. Such properties are essential for a sound
protective coating.
Each of the molecular forming processes has its advantages and limitations.
However, when these are analyzed, chemical vapor deposition comes out quite well
by Blocher's comparisons. To illustrate, chemical vapor deposition is capable of
yielding the greatest variety of products at rates of deposition equivalent to or
-------
exceeding those of the other techniques. It has the greatest throwing power, that
is, the ability to deposit uniformly on relatively complex shapes. Blocher regards
the major drawback of chemical vapor deposition as the fact that the substrate
must usually be heated to relatively high temperatures, which can lead to residual
stresses in the deposit and subsequent mechanical failure provided the differential
thermal expansion coefficients between the substrate and the coating are sufficiently
high. The other techniques have the advantage of maintaining the substrate at
ambient temperatures.
Accountius (13) regards chemical vapor deposition as still largely a labo-
ratory operation, except in a relatively few instances. He states that the deposi-
tion of each material is unique, and in nearly every case, further study of the ther-
modynamics, kinetics, heat transfer, chemistry, and solid state phenomena are
required to elevate chemical vapor deposition from the status of an art to that of a
science.
In spite of its many unexplored facets at present, the process has evolved
quite rapidly from a technological standpoint in the last fifteen years. One may
gain an accurate idea of its recent growth by comparing the 1955 state-of-the-art
survey entitled Vapor Plating (14) with its 1966 revision, Vapor Deposition (]5\
both prepared under the auspices of Battelle Memorial Institute, an organization
which has spearheaded the bulk of fundamental and applied research that has gone
into the field.
The use of chemical vapor deposition is already well entrenched in protec-
tive coatings technology. It is interesting to note, in fact, that the first significant
8
-------
historical application of the technique grew out of a need for a protective coating
material in the incandescent lamp industry in the last two decades of the nine-
teenth century (11). Sawyer and Man (16) deposited pyrolytic carbon, and Ayls-
worth (17) and deLodyguine (18) deposited metals onto carbon filaments. Unfor-
tunately, this particular application was never very satisfactory because of the
brittle carbon core, and interest in chemical vapor deposition as a coating tool
soon subsided for several years.
Since 1935, however, several interesting applications of chemical vapor
deposition have been devised to protectively coat metals, alloys, and non-metallic
materials against erosion and abrasion, general corrosion, and high temperature
oxidation. These have been reviewed by Krier (19). Countless materials are cur-
rently protected by siliconizing, chromizing, nitriding, and other like processes
which make up a vast commercial endeavor. There are also a large number of
research and development programs active today. Emanating from these programs
are such developments as new silicide and aluminide coatings for the aerospace
industry, pyrolytic carbon coatings for rocket nozzles and re-entry vehicle surfaces,
and others. These coatings are usually chemically vapor deposited by pack cemen-
tation or fluidized bed methods, two sub-processes of chemical vapor deposition
which hold high promise for future applications.
Another use of chemical vapor deposition has been recently discovered in
the atomic energy field. Small spherical fuel particles such as uranium dioxide are
coated with oxides or carbides by fluidized bed techniques to produce a dense,
impervious barrier to fission products or other reaction products (20). The fission
-------
product retention of these coatings has been found to be significantly low. Char-
1 ^o
acteristically only a few parts per million (ppm) of Xe Ov5 fission gas generated in
neutron activation is released on post irradiation heating. As a primary result of
this caliber of performance particulate fissionable material encapsulated in a vapor
deposited coating is believed to have the potential for solving many problems in
reactor technology.
Curiously, as mentioned previously, the molecular forming processes such
as chemical vapor deposition have the ability to control the structure of a coating.
For example, a coating may be made to exist in either a crystalline or amorphous
form, or a combination of both, depending upon the proper selection of deposition
parameters. Walton (21) explained the phenomena by pointing out that no orien-
tation is produced when a single bond (i.e., an atom joined to a cluster by a single
bond) is stable. Single bonds rather than multiple bonds occur provided the sub-
strate temperature is low enough or the reactant gas incidence rate high enough.
If, on the other hand, the supersaturation is low enough for multiple bonds to be
stable, but not single bonds, an oriented deposit can result. Interpreted another
way, if one is obtaining crystalline films on a given substrate, the process may be
adjusted to yield amorphous or glassy films by increasing the deposition rate and/or
decreasing the substrate temperature.
Several investigators have discussed whether amorphous films should be
termed glasses. Secrist and Mackenzie (22) have proposed that there is no dis-
tinction, based upon their work. They compared the critical structure indicators
10
-------
of amorphous silica prepared by conventionally cooling from a melt and from var-
ious deposition techniques and found no significant changes.
This exciting discovery means that glassy materials may be prepared by such
processes as sputtering, vacuum evaporation, and chemical vapor deposition as
well as by the classical method of cooling from a melt.
The concept has radically altered ideas on what substances are capable of
forming glasses. Many oxides and non-oxides are capable of being made into
glasses by the molecular forming techniques although most of them cannot be
obtained as glasses in bulk form by cooling from a melt. Sinclair (23) states that
essentially any material which can be deposited by the molecular forming processes
can be made into amorphous or glassy films.
An amorphous film has inherent advantages for a coating application such
as the one proposed in this paper. Sinclair (23) points out, for example, that
glass films have no grain boundaries and are isotropic. For this reason, they are
currently .used extensively in the field of thin-film microelectronics as diffusion
barriers to keep water and ionic contaminants from junctions.
On the other hand, Sinclair (23) points out that the properties of glass films
are greatly effected by methods of preparation, techniques, and conditions. He
illustrates the effect of impurities by stating that a monolayer of water absorbed
onto a substrate surface, if incorporated into a 1000 Angstrom (A) silica film,
would represent 0.5 mole percent of the composition. Additional water soluble
surface contaminants such as sodium ions would have the dual effect of being
impurity sources in their own right and also causes of increased water absorption
11
-------
on the surface. He concludes the the predeposition cleaning procedure is very
important.
Holland (24) generally takes a dim view of the feasibility of protecting
glass surfaces by vapor depositing oxide films onto the substrate. Citing results
from sputtering and vacuum evaporation techniques, he states that it is difficult to
produce films completely free from pinholes. He also observes that only a limited
range of film materials have suitable optical properties to allow their use.
Overall, the writer believes that selected oxide coatings, chemically inert
to water and other corroding media, offer a splendid source of potential for
research in the field of coating soluble glasses by chemical vapor deposition. The
literature contains several references to work which has been recently accomplished
in this area on relatively durable compositions. For example, thin titania films are
now commonly applied to glass bottles after annealing primarily to impart a more
effective scratch-resistant finish (25). The method involves the use of a volatile
organic titanate which is easily pyrolyzed to effect the reaction. The resulting
film is extremely thin and transparent and is said to impart a substantial degree of
additional durability. The method is also applicable to the oxides of aluminum and
Tirconium (26).
Badger (27) has discovered that silica deposited on glass surfaces by treat-
ment with ethyl orthosilicate or other alkoxysilanes, at a temperature between the
melting point and the softening point of the glass, increases the chemical durability
of the glass. The same system has been devised for aluminum oxide, deposited from
-------
aluminum isopropoxide, and is said to work for any oxide which forms constituents
of conventionally prepared durable glasses.
Mixed oxides may be deposited by using single oxide deposition processes
and mixing the metal compounds in the plating atmosphere (28). Such procedures
are readily adaptable to producing borosilicate glass coatings produced by the
mixed vapors from ethyl orthosilicate and triethyl borate.
Several oxides have been coated sequentially onto a substrate to form
composite layers (28). Such materials as silica, alumina, titania, and zirconia
have been mentioned. This latter method would possibly offer advantages in mini-
mizing the effects of pinholes and other growth defects.
13
-------
CHAPTER ill
OBJECTIVES OF RESEARCH
The objective of this research was to define the capabilities of chemical
vapor deposition as a tool for coating soluble glass substrates with metallic oxide
films. The work represents a segment of a projected three year study which will
deal with methods of processing glass packaging containers after use. In order to
accomplish the objective of this investigation, the research was divided into three
phases, as discussed below.
The first phase involved the design and construction of a mechanism cap-
able of depositing metallic oxide films from the vapor phase onto soluble glass
substrates. The proper selection of equipment components and materials necessarily
involved an extensive research effort and numerous calculations to assure that satis-
factory coatings could be attained over a wide range of controlled conditions.
The second phase of the research effort was directed toward characterizing
selected coatings in terms of various process parameters, as well as accumulating
basic data designed to reveal the overall feasibility of the system's capability of
performing its ultimate task by considering various pertinent properties of the com-
posite produced.
The third phase of this work involved a critical assessment of the system
investigated. Its basic limitations were discussed, and suggestions were given for
possible improvements to be incorporated into the current facility.
14
-------
CHAPTER IV
EXPERIMENTAL PROCEDURE
A. Coating Apparatus Design
The vapor deposition apparatus designed and constructed for this study is
shown schematically in Figure 1 and photographically in Figure 2. The choice of
components was carefully screened to provide a system capable of depositing a
wide spectrum of materials over a range of controlled conditions. At the same
time, it was deemed advisable to maintain a simplified and straightforward design
concept which can be ultimately tailored to scaled up production operations with
minimum difficulty.
A mullite-lined tube furnace served as the reaction chamber for this inves-
tigation. The combustion tube, purchased from McDanel Refractory Porcelain
Company, measured thirty-six inches in length. Its cross sectional dimensions were
2 1/4 inches inside diameter with a 1/8 inch wall thickness. The ends of the tube
were readily closed off with rubber-gasketed aluminum doors supplied by the same
company. O-ring compression seals were mounted on the doors to provide 1/4 inch
gas tight lead-throughs for the entrance and exit of gases. Thermocouples and
other control components were also introduced into the interior of the reaction
chamber via the same type lead-throughs. The thermocouple wires extending
through the double bore porcelain insulator tubing were sealed with epoxy cement
to prevent gas leakage. The sealed door arrangement provided a gas tight system
capable of being dynamically monitored or controlled during a coating run.
15
-------
PRESSURE GAUGE
FLOWMETER
ORGANIC LIQUID
COMPRESSED GAS
FURNACE
COMBUSTION TUBE
SUBSTRATE
EXHAUST-
•SCRUBBER
THERMOCOUPLE TUBE
AND POTENTIOMETER
•CONTROLLER
Figure 1. Schematic of the chemical vapor deposition apparatus
-------
Figure 2. Photograph of the chemical vapor deposition apparatus
-------
A review of other designs (11) suggested that the cross-sectional dimen-
sions of the chamber should be as small as practically possible for improved control
effects. This consideration is thermodynamically related to gas mixing uniformity
and heat transfer effects within the tube (29). It was recognized at the outset of
this investigation that two primary control parameters would be temperature uni-
formity of the substrate on all external faces and a uniform supply of the reactant
gas and carrier gas mixture to the substrate. A compact reaction chamber with a
small internal volume enhances the control of such variables.
The tube was fired with a 230 volt, 3200 watt Lindberg "Hevi-Duty" labo-
ratory combustion furnace. The furnace measured 25 inches along the axial length
of the tube. The unit incorporated eight 1/2 inch by 16 inch series connected
silicon carbide heating elements which were arranged transverse to the axis of the
tube, four above and four below, on four inch horizontal and vertical centers, to
form a twelve inch firing zone.
Excellent temperature control of the unit was readily achieved through the
use of a Barber Coleman Model 477 Capacitrol electronic indicating controller to
drive a Model 621A silicon controlled rectifier for the proportional control of load
power. The solid state circuitry of the Model 477 makes it possible to eliminate
temperature overshoot and undershoot by properly adjusting the proportional control
action and reset features of the instrument.
The control thermocouple was composed of platinum and platinum alloyed
,>
with 13 percent rhodium wires. The junction was maintained in contact with the
external face of the combustion tube at the center of the firing zone. The final
18
-------
location of the controlling sensing element was determined after many trial and
error experimental runs which consisted of monitoring the response and recovery
characteristics of the above control system to arbitrary changes in temperature and
gas flow with the thermocouple junction located inside the firing zone of the com-
bustion tube as well as against the outside face.
For the purpose of the above trials, a monitoring thermocouple was con-
nected to a potentiometer with its junction positioned inside the firing zone of the
combustion tube. This thermocouple was also subsequently used during actual
coating runs to record the substrate temperature.
In general, it was found that the control characteristics for an externally
heated system under stable conditions were excellent for both thermocouple loca-
tions. However, process lag and cycling were less for the external thermocouple
location when an arbitrary process change was introduced. Mullite heat reflectors
were placed inside the tube at the ends of the firing zone to help balance the tem-
perature profile, and with this arrangement, it was possible to maintain plus or
minus 2 degrees Fahrenheit (° F) at a given spot and plus or minus 15° F at different
locations within the firing chamber under stabilized conditions. A schematic
representation of the control circuit is shown in Figure 3.
This system as described will operate successfully to 2700 F. However,
longer life of the heating elements and thermocouple wires is expected if the tem-
perature does not exceed 2500° F. The unit is readily responsive to input power
changes. Heating and cooling rates above 350° F per hour are considered dele-
terious to extended component life.
19
-------
ro
o
MODEL
477P
INDICATING
CONTROLLER
THERMOCOUPLE
LOAD
MODEL
62IA
POWER
CONTROLLER
Figure 3. Schematic of the temperature control circuit
-------
Control of the plating vapor concentration in the carrier gas stream over a
range of flow rates is important in chemical vapor deposition processes, for such
variables partially govern deposit thickness, uniformity, orientation, and purity.
For the purposes of this investigation, cylinders of prepurified grade compressed
gases were purchased from Matheson Company to serve as the transport vehicle.
These gases were pressure regulated on Matheson equipment in two stages to yield
a resultant gas force which could be controlled in integral units of water pressure
up to 35 inches. This controlled pressure was fed into the system through a flow-
meter calibrated to read between 0-10 cubic feet of air per hour (cfh) in increments
of 0.25 cfh up to 2 cfh and in increments of 1.0 cfh at higher flow rates. A cali-
bration chart supplied by the vendor allowed different gases to be used with the
same flowmeter by applying correction factors to the settings used. The resultant
system yielded a highly reproducible, control-sensitive gas supply to the vaporizer.
The liquid plating agents used in this study were vaporized into the carrier
gas from a simple 125 milliliter (ml) Pyrex flask. The flask was sealed at the top
with a rubber stopper containing two sections of 1/4 inch Pyrex tubing, one to allow
for the entrance of the carrier gas and the other for the exit of the resultant gas
mixture. A constant temperature water bath was utilized to elevate the vapor
pressure of the liquid for those runs that required a higher concentration of plating
vapor in the gas stream mixture entering the reaction chamber. Higher temperatures
(between 200°F - 500° F) were achieved by using an electrical blanket around the
flask, connected through an autotransformer for temperature regulation. The liquid
21
-------
temperature Inside the vaporizer was read with a mercury thermometer, vertically
suspended by the rubber stopper at the top of the flask.
The carrier gas was admitted to the vaporizer by bubbling through the
liquid, or alternately, by simply passing it over the liquid surface. The type pro-
cedure used was found to influence the evaporation rate significantly. In the first
case, the carrier gas becomes saturated with plating vapor, whereas in the second
case saturation does not necessarily occur. Thus, the bubble-type vaporizer was
found to give a constant concentration of plating vapor in the carrier gas over a
range of flow rates whereas the non-bubbling type was largely uninfluenced by
flow rate except at higher flows and at temperatures near the boiling point of the
liquid. Figure 4 depicts the relationship for tetraisopropyl titanate, one of the
plating compounds used.
Both methods generally worked satisfactorily for the range of conditions used
in this study. However, high vaporization temperatures and high flow rates leading
to excessive plating vapor saturation in the gas stream oftentimes resulted in spray
formation or misting conditions and subsequent coating defects on the substrate with
either method. This situation was partially remedied by wrapping the downstream
surfaces of the vaporizer and gas lines with electrical heating tape and packing the
interior of the flask with loosely fitting binder free glass cloth to lessen the possi-
bility of the vapor condensing and subsequently reevaporating on downstream sur-
faces. Such measures however, were only marginally successful and as a result it
was concluded that more sophisticated vaporizer design concepts would be required
22
-------
10.0
5.0
X
UJ
1 1.0
5
V.
-------
under such conditions. Li (30) and Powell (11) have reviewed some advanced
design concepts used in this field.
The vaporizer equipment described thus allowed a given plating compound
to be vaporized at a given rate and mixed under controlled conditions with a
selected carrier gas. The ability to regulate the concentration of the plating
vapor in the resultant gas stream is highly important, as previously mentioned.
Appendix 1 contains a sample calculation illustrating the method used to derive
plating vapor concentration in the gas stream under a given set of conditions.
After mixing, the effluent gas stream exited the vaporizer by forced con-
vection and passed into a 1/4 inch stainless steel tube leading into the interior of
the reaction chamber via an O-ring compression seal mounted on the entrance door.
The stainless steel tube was water cooled along its length by a copper coil. The
entrance and exit water leads of the coil were also extended through O-ring seals
and connected to a nearby water source and drain system. The purpose of the water
cooling was to protect the various vapor species from premature decomposition,
primarily at higher temperatures and lower flow rates. In many cases, water cooling
was not required, and a 1/4 inch Pyrex glass inlet tube was substituted under those
conditions to promote cleanliness and lessen the possibility of side reactions. The
length of the inlet tube was sufficiently long to allow its terminal end inside the
reaction chamber to be pre-adjusted laterally with respect to the substrate position
in the firing zone. This arrangement allowed a range of distances between the gas
outlet and the substrate to be selected for orientational studies.
24
-------
After reaction at the substrate, the spent gases were exited through the
rear end of the sealed combustion tube, bubbled through 1/2 inch of water level
in a 1000 ml Pyrex flask and exhausted through a fume hood collection system
erected over the deposition apparatus.
B. Other Design Considerations
As an adjunct to the method of vaporization described above, a method of
vaporizing solid materials or liquids with very low vapor pressures was derived.
This procedure involved the use of higher temperatures (500° F - 1800° F) supplied
by a 120 volt, 1650 watt nichrome wire-wound combustion tube furnace located
upstream from the deposition furnace previously described and supplied with a
common mullite tube 57 inches in length extending through both furnaces. The
smaller furnace was independently controlled with a 120 volt General Radio auto-
transformer supplying power to the system as dictated by a Barber Coleman Model
472 Capacitrol indicating time proportioning controller. A chromel-alumel thermo-
couple vertically located in contact with the external surface of the combustion
tube at the center of the 12 inch firing zone served as the sensing element.
For this system, evaporation was carried out inside a 1 inch by 1 inch
mullite crucible which contained the substance to be evaporated. The crucible was
placed into the center of the firing zone, and a carrier gas passed over its surface.
The actual vaporization temperature was read with a chromel-alumel thermocouple
located 1/2 inch from the evaporating surface.
Intimate mixing of the gaseous species was difficult to attain, and the
method was generally unsatisfactory. However, the system has merit for high
25
-------
temperatures and should qualify as a satisfactory method if procedures to better
direct the gas flow for improved mixing effects are installed. The external arrange-
ment of the complete system is shown in Figure 5.
Efforts to meter liquids dropwise into the gas tight chamber for evaporation
proved fruitless. A titrimeter charged with the plating substance was installed
above the evaporation furnace and the liquid head produced drops of liquid which
entered the chamber by gravity, being collected in a crucible under the influence
of controlled heat as described in the preceding paragraphs. The rate of metering
did not lend itself to close control by this method since the head pressure of the
titrimeter constantly changed. No coatings of value were obtained in this way.
C . Input Material Considerations
The compressed gases used in this study were prepared by Matheson Com-
pany. Nitrogen was the primary gas employed, with argon and helium used to a
lesser extent. As in all chemical vapor deposition processes, the relative purity
level of the carrier gases largely influences coating characterizations and reaction
kinetics. The presence of water vapor in these gases was considered especially
deleterious because of the pronounced tendency of tetraisopropyl titanate to hydro-
lyze. As can be seen by a comparison of the molecular weights, 284.2 for tetra-
isopropyl titanate and 18.0 for water, only a small amount of water is required to
reduce a relatively large amount of the plating compound. All compressed gases
used were prepurified grade or higher with a purity level of greater than 99.99
percent.
26
-------
I
Figure 5. Photograph of dual furnace assembly
-------
The plating compounds used in this study were selected for their ability to
yield controlled coatings of desired composition through the medium of chemical
vapor deposition. The physical mechanism consists of causing the vaporized plating
compound to decompose or pyrolyze at the heated substrate, leaving the nonvolatile
components of the compound behind as a chemically bonded coating. Obviously,
a coating compound must possess certain characteristics to enable it to be processed
in this way. Ideally, it should be a liquid at room temperature with a sufficiently
high vapor pressure to obviate the necessity for elevated vaporizer temperatures.
Its purity content must be high to minimize the possibility of undesirable contami-
nants collecting on the substrate and also to prevent the possibility of side reactions
neutralizing the plating vapor. Lastly, the source material must be chemically
stable even though it is volatile, that is, it must not prematurely decompose, and
it should contain only one nonvolatile component. Two organometallic compounds
which readily fit these criteria were selected for this investigation. They are
tetraisopropyl titanate, Ti(C3H7O)4, known as TPT, and ethyl orthosilicate,
Si(OC2H5)4, known as ETOS. In addition, several runs were made using aluminum
isopropoxide, Al (C3H7O)3, and ethyltriethoxysilane, (C2H5)Si(OC2H5)3. These
materials are all relatively stable liquids with boiling points ranging from 322° F
to 478° F. All of these materials yield oxide coatings with the associated metallic
cation. Their proposed reaction mechanisms are listed below, in the form of
chemical equations.
Ti(C3H70)4 —» Ti02 + 2C3H6 + 2C3H7OH
Si(OC2H5)4 ,> SiO2 + hydrocarbon gases
(C2H5)Si(OC2H5)3 » SiQ2 + hydrocarbon gases
2AI(C3H70)3 — * AI203 + 3C3H6 + 3C3H7OH
28
-------
The purity levels of the plating compounds used were the highest obtain-
able for stock items from the major chemical supply houses. Generally, this is
specified to be greater than 90.0 percent pure. The writer believes that future
work in this field should incorporate even higher purity levels if possible to further
preclude the possibility of side reactions and impurity deposits. The chemical
complexity of these materials requires that they be stored under cool, dry conditions,
No evidence of deterioration with time was noticed. Table I may be consulted for
chemical information pertaining to the plating compounds.
Glass microscope slides were employed in this study to serve as substrates
for all process evaluations. The 1 inch by 1 3/4 inch microslides were composed of
Pyrex glass manufactured by W.H. Curtin and Company. These slides were care-
fully individually washed with warm, soapy water, rinsed, and dried prior to use.
The soluble glass compositions were produced as 1 1/4 inch by 1/4 inch
discs. Raw materials consisted of commercial grade sodium carbonate and silicon
dioxide weighed and mixed in the proper proportions to yield the desired oxide ratio
after melting. The range of compositions studied varied between 43-57 weight
percent SiC^. Powders were intimately mixed with a V-blender mixer, poured into
a mullite crucible and fired to 2500° F. The resultant fused mixture was removed
from the furnace after 30 minutes and with the aid of hand tongs was poured into a
steel die and pressed into the desired shape. The pressed pellet was rapidly removed
from the die and transported to an annealing furnace at 800 F. After stabilization
at that temperature for thirty minutes the furnace was deenergized and slowly cooled
29
-------
TABLE I
CHEMICAL PROPERTIES OF PLATING COMPOUNDS
co
o
MATERIAL (REFERENCE)
FORMULA
MOLECULAR
WEIGHT
SPECIFIC MELTING
GRAVITY POINT, °F
BOILING POINT, UF
AT A VAPOR PRESSURE OF
60 mm. 400 mm. 760 mm.
Terra isopropyl titanate (25) TUCqHjO)^.*
Ethyl orthosilicate (37)
Ethyltriethoxysilane (37)
Aluminum isopropoxide (38)
Si(OC2H5)4**
(C2H5)Si(OC2H5)3*
A,(C2H70)3*
284.2
208.3
192.3
204.3
0.955 68
0.935 -121
0.889
1.058 223
300 410 450
333
188 282 322
478
'Supplied by K & K Laboratories, Inc.
^Supplied by Fisher Scientific Company
-------
The resultant stress-free product was subsequently removed from the furnace and
ground and lapped on conventional metallographic processing equipment using 80,
240, 400, and 600 grit silicon carbide particles. Mechanical polishing was
accomplished with 1 micron alumina by Linde. The finished discs were visually
inspected at 80x under vertical illumination. Samples exhibiting scratches, pits,
or other defects were repolished. The finished product was then stored in a desic-
cator prior to use.
D. Deposition Techniques
The reaction chamber was charged with glass substrates at room temperature
to prevent stress cracking the samples. Immediately after sealing the system, a 1.0
cfh flow of nitrogen sweeping gas was initiated to purge the system. After a five
minute purge, the furnace power was turned on, When the substrate temperature
reached the specified value as read on the monitoring thermocouple located 1/2 inch
away from its surface, the compressed gas flow was adjusted to a specified level and
rerouted through the plating compound vaporizer, thus introducing a reactant gas
mixture into the deposition zone. The gas mixture was allowed to flow through the
furnace for a specified length of time determined by the thickness of the coating
required. The furnace was then turned off and purging was resumed until room tem-
perature was reached and the sample removed.
,>
Two methods of holding the substrates in the deposition chamber were devised
Those substrate?, subjected to the upper temperature limits for glass were simply laid
on one face onto a 2 inch by 6 inch sheet of corrugated alumina. The alumina
holder tended to protect the substrate from a tendency to slump at the higher
31
-------
temperatures. The major drawback to this method was that it required an additiona
coating cycle for the other side of the substrate.
Lower temperature coating operations were alternately carried out by sus-
pending the substrate vertically with the front face perpendicular to the gas stream
The substrate was held by the hooked ends of a short length of wire which was spiral
wrapped in the middle portion around the outside of the monitoring thermocouple
insulator tubing for anchoring purposes. In this manner, minimum surface contact
with the substrate was maintained. This mode of orientation allowed some coating
to be deposited on the back side of the sample at higher gas flow rates. In addi-
tion, Hi is method of suspension allowed the substrate to be rotated during the coatinc
operation, thereby improving the overall coating uniformity to a high degree. Rota-
tion of the sample during a run was accomplished by exerting a torque on the moni-
toring thermocouple insulator tubing extending through the exit end of the reaction
chamber. The insulator tubing was linked to the heat reflector and substrate, and
twisting it through a given angle transformed the same rotation to these component
inside the chamber. The arrangement of the two substrate positions is shown sche-
matically in Figure 6.
It was found that several successive runs tended to produce such contami-
nation products as non-adherent white powder particles on the interior surfaces of
the chamber. These were analyzed and found to be either TiO2 or SiC^, depending
upon the plating material used. Black soot (presumably carbon) and oily spots
(partially polymerized silane from ethyltriethoxysilane) were also observed. These
substances were easily removed with a warm detergent solution, with the exception
32
-------
•ALUMINUM DOOR
MULLITE COMBUSTION TUBE
TO POTENTIOMETER
TO EXHAUST
0-RING COMPRESSION SEAL
GAS INLET TUBE
HEAT REFLECTOR
THERMOCOUPLE TIP
SUBSTRATE
CO
CO
•
B
ft
\
\
\
\
\
1
s
s
s
\
s
>
1 /^
' S/SSSSSS/SS/S S/SSSSS/S S S S S S / / / / / SAT S S
/
. ./.
^jj^2^SEB^n_
4
' / / SS / S ///////S////// //////// SIS /// //s
1
&\j\j*j i r\j"i i i_
s sssssssss
•
///////// V
B
yy xy y yy v^y yyv y y<
0
^
fe
3
s
s
s
J
1
In
%
CM
ALUMINA SUBSTRATE HOLDER
Figure 6. Schematic of the substrate positions used. The top view shows the substrate in a vertical position.
The bottom view shows the substrate in a horizontal position.
-------
of the carbon, which was removed only under high temperature oxidizing condi-
tions. Silica glass was also found inside the tube after a series of runs. It was
removed periodically by scrubbing with a weak solution of hydrofluoric acid.
These contaminants did not noticeably alter any results, although their effect on
coating defects was not definitely established. Jordan (31) reported no deleterious
effect from silica contamination using a similar process for coating semiconductors,
where purity level is critical.
E. Analytical Procedures
1. Microscopy
The coatings obtained were viewed under a vertically illuminated Leitz
Panphot me tall ©graphic microscope, and photographic work was accomplished on
this equipment with the aid of a Nikon photomicrographic attachment coupled to
a 35 millimeter (mm) Leica camera. The photomicrographs yielded information pri-
marily related to the defect structure and orientation of the deposits.
2. Phase Analysis
Chemical phase analyses of the TiO2 coatings were determined with the aid
of X-Ray diffractometry and spectroscopy, using Norelco instrumentation. The
structure of the deposit was determined by observing the presence or absence of
diffraction lines for a given material, using copper K-alpha radiation, 30 kilovolts
and 10 mil I lamps. TiO2 deposits which failed to diffract X-Rays were declared to
be amorphous; those deposits which exhibited characteristic diffraction peaks were
crystalline.
34
-------
Crystalline SiO2 was determined by the same method outlined previously.
The silica glass was detected by X-Ray diffraction and infra-red spectroscopy. The
broad, diffuse pattern revealed by the diffractometer was not considered to be very
definitive, so the spectroscopy technique was primarily relied upon. A Perkin-
Elmer 621 Grating Infrared Spectrophotometer was utilized for the production of
these data. It was necessary to use a substrate which was transparent to infrared
energy, so glass was replaced by a 10 mil wafer of semiconductor grade silicon,
which served as the base for the deposited coating.
3. Coating Thickness and Uniformity
The TiOo coatings provided an additional dimension to the analytical tech-
niques used because they were found to produce optical interference colors when
deposited on glass. The presence of these colors is related to the differential refrac
tive index between the substrate and the coating, as well as the coating thickness.
An excellent review of the subject is provided by A. Vasicek (32).
Generally, the color fringes may be interpreted in the same way one reads
a topographic map, except in this case the differential thickness between bands of
the same color equals one-half of the wavelength of light transmitted through the
film producing that color. Figure 7 illustrates the concept. The target effect in
photograph (A) resulted from the plating vapor contacting the substrate in a concen-
trated stream rather than diffusing out to produce a more uniform color, as seen in
(B). Both of these samples contained roughly the same amount of coating by weight.
However, the coating on the left varied progressively in thickness across its surface,
35
-------
to
a
(A) (B) (C) (D)
Figure 7. Analysis of TiOo film thickness and uniformity shown by interference colors. All films were derived by bubbling
nitrogen gas through TPT at 212° F. Exhibit (A): Carrier gas flow 4 cfh; deposition temperature 575° F; substrate
orientation 4 inches; time 15 minutes. Exhibit (B): Carrier gas flow 10 cfh; deposition temperature 765° F; sub-
trate orientation 10 inches; time 30 minutes. Exhibit (C): Carrier gas flow 10 cfh; deposition temperature 765° F;
substrate orientation 10 inches; time 5 minutes. Exhibit (D): blank sample.
-------
whereas (B) was coated relatively evenly, resulting in an average thickness of
approximately 3000 A. Photograph (C) depicts an extremely thin, uniform coating,
approximately 500 A in thickness, which may be contrasted with the uncoated
sample (D). Whereas the thin films (less than 1000 A) were highly reflective, as
a silvered mirror, if crystalline, or merely tinted if amorphous, the thick films (over
50,000 A) were opaque and reflected little light.
Thus, the TiO2 films were easily interpreted visually for uniformity and
thickness. Such was not the case for SiC>2 films which yielded no colors because
of a close similarity of refractive indices between the coating and the substrate.
Alternately, film thickness was measured by weighing the specimen before
and after coating on a Mettler grammatic balance. The weight gain represented
the amount of coating deposited on the substrate, accurate to 0.0001 grams. The
average coating thickness was computed from the following formula:
w
t =
\d) (A) (10-8)
where
t = coating thickness in Angstrom units
w = weight gain of substrate in grams
d = density of coating in grams per cubic centimeter
A = area of coating surface in square centimeters
Use of the foregoing relationship yielded an average thickness for all exposed
surfaces coming in contact with the plating vapor. Thus, to be used with validity,
steps were taken to insure that all surfaces of the substrate received equal coating
37
-------
treatment by rotating the sample and reversing its orientation as required, or alter-
nately to mask out all surfaces except the one being coated. This last procedure
was accomplished on microslides by pressing two faces together as in a stacking
sequence, coating the composite, and separating the two microslides after coating
to reveal one uncoated surface for each slide, as well as one coated surface. Of
course, the weighing procedure is unable to distinguish between a diffusion coating
and an overlay coating. For this reason, its use is restricted to lower temperature
coating operations (below 1500 F).
Table II lists the refractive indices and densities of all coatings and sub-
strates used in this investigation.
4. Coating Efficiency
Chemical vapor deposition processes inherently produce low deposition
efficiencies which are consequently a major cost control factor, especially since the
plating substances involved are oftentimes very expensive (13). Therefore, the
efficiencies yielded by this process were carefully studied. The coating efficiency
for a given run was derived by comparing the weight gain of the substrate during a
coating run to the weight loss of the coating material in the substance being vapor-
ized and expressing the result as a percentage. The amount of plating vapor in the
substance being vaporized is derived from a balanced chemical equation of the
reaction. Gravimetric measurements were obtained on the Mettler balance des-
cribed previously. The method of computing coating efficiency is illustrated in
Appendix II.
38
-------
TABLE II
OPTqCAL PROPERTIES OF COATINGS AND SUBSTRATES
INDEX OF
REFRACTION
DENSITY,
g/cm3
I . Coatings
A. Si O2 glass
B.
C.
anatase
amorphous
II. Substrates
A. Pyrex Microslides
B. !.ONa2O-1.3SiO2
C. 1.3Na2O-l.OSiO2
1.43(22)
2.54(39)
1.46 (7)
1.51 (7)
1.52(7)
2.20(22)
3.84(39)
2.23(7)
2.42 (7)
2.43(7)
39
-------
5. Coating Stresses
As mentioned earlier, the fact that vapor deposits are applied at elevated
temperatures usually leads to residual stresses, primarily caused by thermal expan-
sion coefficient differences between the deposit and the substrate. Powell (33) has
pointed out that stress analysis in chemical vapor deposition systems is complex
because thermal expansion stresses tend to mask out smaller stresses which may arise
r
from other sources. In addition, elevated coating temperatures sometimes allow
/
stress relief to take place by diffusional and other relaxational processes. The use
of glassy materials introduces an additional dimension of complexity into any stress
analysis calculation, because their mechanical properties generally present anomalies
which are well known (7). The strength properties of glass are largely controlled by
surface conditions.
Table III contains the mechanical property constants of the various materials
used in this study. They may be used in conjunction with the calculation of residual
stresses between a coating and substrate. Appendix III contains an explanation of
the procedure used to derive the residual stress.
40
-------
TARE III
MECHANICAL PROPERTIES Of COATINGS AND SUBSTRATES
MODULUS Of
ELASTICITY, E,
psi
COEFFICIENT OF
THERMAL EXPAN
SION,*, /°F
TENSILE
STRENGTH,
psi
I. Coatings
A.
Glass
B. TiOo Amorphous
C. TiO Anatase
D. TiO2 Rutile
II. Substrates
A. Pyrex Microslides
B. 1.ONa2O-1.3SiO2
C. 1.3Na2O-l.OSiO2
10 x 10* (40) 0.55 x 10~6 (40) 7250 (40)
41 x 1^(41) 4.f x 10"6 (41) 7500 (41)
f x 106(7)
10 x 10^(7)
10 x 106(7)
3.2 x 10"°(7) 8900(7)
14.0 x 10"6(7)
14.0 x 10"6(7)
41
U.S EPA Headquarters Library
1200 Pennsylvaniavenue
ton, »X
-------
CHAPTER V
RESULTS AND DISCUSSION OF RESULTS
A. Effect of Process Variables Upon Reaction Kinetics
The coating process variables associated with a given vapor deposition
system are interrelated in a complex manner, and are uniquely dependent upon the
geometry of the unit, as well as upon the chemical characteristics of the coating,
materials and the substrate used. It is therefore important to understand that
although the general relationships among the process variables of a given system
j
are valid for all systems, the translation of absolute data from one system to another
is not.
i
I
For this work, a series of TiO2 coating runs were made over a range of con-
trolled conditions in order to study the effects of important variables upon deposi-
tion rates and coating efficiencies. These variables and their limits of investiga-
tion were as follows:
VARIABLE LIMITS
1. Substrate temperature, ° F 450-1085° F
2. Coating time, minutes 5-75 minutes
3. Carrier gas flow rate/ cfh 2-10 cfh
4. Substrate orientation, inches* 2-12 inches
5. Reactant gas concentration in
carrier gas (ppm) at a given flow
rate 19-2300 ppm
*Substrate orientation may be defined as the lateral distance in inches between the
effluent end of the input gas tube and the front face of the substrate.
42
-------
coatings from tetraisopropyl titanate were selected for this study
because of their ease of application to glass substrates over a wide range of con-
ditions. By contrast, SiO2 coatings from ethyl orthosilicate are energetically
feasible only within a narrow range of temperatures immediately below the softening
point of the substrate.
1. Substrate Temperature
Figure 8 illustrates the strong influence that the temperature of the substrate
exerts over the deposition rate and coating efficiency. A linear relationship was
found to exist for the range of conditions investigated. It is especially interesting
to observe that for the temperature range shown the efficiency can be increased by
a factor of four merely by operating at the upper temperature extreme rather than
the lower. It is anticipated that the curve shown would begin to deviate from
linearity at higher temperatures with an accompanying decrease in plating efficiency
because of premature decomposition of the vapor species inside the reaction chamber.
2. Coating Time at Different Temperatures
Figure 9 depicts the interesting effect exhibited by coating time upon depo-
sition rates at different temperatures. The coating grown at the lower temperature
was characterized by an initial period of slow growth, known as an induction period,
and a subsequently increased rate of steady-state growth thereafter. This dual growth
rate effect was absent in the higher temperature coating which grew at a faster rate,
as seen by the slopes of the two curves. The overall coating efficiency for the low
temperature coating was severely damaged by the presence of the induction period.
43
-------
12
X
LU
I-
LU
-I
O
5
UJ
h-
<
or
o
Q.
LU
O
CM
O
7-
6-
> 5
o
2
UJ
O
iZ 4.
u_
LU
2 3-
o
o
2-
I-
400
600 800
DEPOSITION TEMPERATURE (°F)
1000
1200
Figure 8. Plot of TiC^ deposition rates and coating efficiencies at different deposition temperatures. The data
was derived by bubbling nitrogen gas through TPT at 10 cfh; vaporization temperature 70° F; substrate
orientation 4 inches; time 60 minutes.
-------
tn
30
COATING
45
(IVItNUTES)
60
75
Figur© 9.
Plot of TiO2 deposition rates at two different deposition fr©mperatur©s for varying coating time p©nods.
The data was derived by bubbling nitrogen gas through TPT at 10 cfn; vaporization temperature 70° F;
substrate orientation 4 inches.
-------
3. Carrier Gas Flow Rate and Substrate Orientation
The carrier gas flow rate plays an important role in coating reaction kinetics,
as shown by Figure 10. The relationship is especially important when considered as
a function of the substrate orientation. The plot shown illustrates the effect of two
different flow rates upon the coating efficiency at different substrate orientations.
Coating reactions at a given gas flow rate were generally enhanced by decreasing
the mean average time of residence of the reactant vapors in the deposition zone
prior to contacting the substrate. This of course, was readily accomplished by
decreasing the substrate orientation. Curiously, however, this relation did not
hold at extremely short distances, presumably because the increased cooling effect
of the impinging gases on the substrate was sufficient to retard the reaction.
A change in gas flow promoted marked changes in the reaction rate—sub-
strate orientation profile, as shown. Reaction rates under low gas flow conditions
were highly sensitive to the substrate orientation. To illustrate, for the data shown,
the coating deposition rate and efficiency at two cfh varied from a maximum at two
inches to no reaction at all at six inches, whereas these parameters did not change
appreciably in this range at the higher flow rate, although the maximum attainable
reaction rate at the higher flow was less than one-half that which could be developed
by the lower gas flow condition.
4. Reactant Gas Concentration
TiC>2 coatings were readily obtained over a range of reactant gas concen-
trations in the carrier gas stream at total flow rates ranging from two to ten cfh
46
-------
SUBSTRATE ORIENTATION (INCHES)
Figure 10. Plot of TiO2 coating efficiencies at different substrate orientation and carrier
data was derived by bubbling nitrogen gas through TPT; vaporization
temperature 620° F; time 5 minutes.
—•« flow rates
212° F;
-------
The deposition rate was found to vary linearly with concentration at a given total
fVow rate, as shown by Figure 11. The coating efficiency for these runs was gen-
ertilly the same. The quality ef the coatings produced at the different concentra-
tion $ was significantly affected, however. At high concentrations, the adherence
was i x>or and the coatings examined shewed split and cracked areas. These samples
were t requently covered with a loosely -adherent brownish colored powder which
wai" fou nd to be a mixture of TIO2 and carbon. Many ©f the coatings produced
exhibitec ^ a dirty cast, indicative of carbon inclusions in the deposit. None of
these advt *rse conditions were found in etepesits grown under similar conditions at
lower reaci ;°nt 9Qs concentrations.
B. Coating . Stresses
The cc Bating stresses developed by SiC>2 coatings applied to either Pyrex
microslides or & oluble glass substrates were compressive in nature, and very pro-
nounced, as wou 'd generally be expected from an inspection of the physical con-
stants involved in the stress calculation as well as the high deposition temperatures
employed. An exeu "nple of a highly stressed SiO2 coating is shown in Figure 12.
In this case, the coating was applied t© a Pyrex microslide at 1220° F. During the
initial stages of the cooling cycle sufficient differential stress developed to warp
the four corners of t he substrate while #ie glass was plastic enough to flow. Further
cooling with the gla: ss in a warped, rigid condiHon subjected the coating to a com-
bined compressive shisar stress which caused it to craze.
Depending u pon conditions, several i>iO'2 coated soluble glass substrates
cracked in tension u nder the influence of the residual stress developed during the
48
-------
10
350
1100
1655"
Figure 11
REACTANT GAS CONCENTRATION (PPM)
Plot ef TiC>2 deposition rotes at different reactant vapor concentrations. The data was derived by
mixing the carrier gas with TPT under both bubbling and non-bubbling conditions and the vaporization
temperatures regulated between 70° F - 212° F; carrier gps flow rate 8 cfh; deposition temperature
780° F; substrate orientation 4 inches; time 15 minutes.
-------
rn
o
Figure 12. Photograph of high residual stress in SiO2 deposit. The coating was derived from ethyl orthosilicate
vaporized into nitrogen gas at 1 cfh; vaporization temperature 70° F; deposition temperature 1220° F;
substrate orientation 3 inches; time 14 hours.
-------
cooling cycle, and others were sufficiently weakened to allow manual breakage by
application of a torque to the stressed sample.
Despite these obvious drawbacks, satisfactory coatings were developed by
resorting to low-temperature slow growth techniques. In this case, temperatures
o
did not exceed 1050 F, and the resultant product demonstrated adequate mechan-
ical properties.
TiO2 coatings were found to develop smaller compressive stresses on soluble
glass substrates, because the temperature of deposition was generally much lower
and the thermal expansion coefficients were more closely matched. Coatings
applied to borosilicate microslides actually developed a small tensile stress. Yet
even under these much improved conditions, TiC>2 coatings could be made to pro-
duce a definite crack pattern, provided the coating conditions were not properly
controlled. Figure 13 illustrates two TiO2 coatings produced at 780° F and a
reactant gas concentration of 2150 ppm in the gas stream flowing at a rate of eight
cfh. In this case, the concentration of the reactant vapor distributed within the
carrier gas was very high and non-uniform. In fact, small droplets of tetraisopropyl
titanate actually contacted the substrate prior to decomposing, and it is believed
that this factor was primarily the reason for the failure of the coating. Figure 13(B)
shows the mode of failure to be a tensile stress in the coating.
In general, edges along the substrate surfaces developed stress concentra-
tions, and coating could be made to flake off along these points under relatively
low abrasive shear stresses. This was especially noticeable for the thicker films of
SiO2 (over 75,000 A) and did not appear prevalent on the thinner coated specimens,
51
-------
'
' I
(A) (B)
Figure 13. Photomicrographs of residual stresses in a TiC^ deposit. Exhibit (A): thin coating, Exhibit (B): thick
coating. These coatings were derived by bubbling nitrogen gas through TPT at 8 cfh; vaporization
temperature 212° F; deposition temperature 780° F; substrate orientation 4 inches; time 15 minutes.
One inch on photomicrographs equals 88.9 microns.
-------
C. Coating Uniformity
Coating uniformity was found to depend chiefly upon the ability of the
system to deliver a constant supply of plating vapor to all surfaces of the substrate
under uniform temperature conditions. Generally, the substrate temperature was
well balanced for all conditions investigated, with the design previously described.
A uniform dispersal of the plating vapor to all surfaces of the substrate was
more difficult to attain, especially for the TiO2 coatings. The deposit tended to
build up primarily on the front face of the vertically-positioned substrate with a
characteristic pattern, most notably the concentric circle effect referred to earlier,
which was a direct consequence of the non-turbulent flow characteristics of the
system. Geometric limitations precluded Hie possibility of practically achieving
turbulent flow within the tube. As measured by the Reynolds number, NRC/ the
criteria for turbulent gas flow within the system would have required a cost prohi-
bitive carrier gas flow rate of almost 200 cfh. In addition, the cooling effect from
this amount of gas flow would probably have rendered temperature control inside the
tube more difficult to attain.
For the above reasons, the control of coating uniformity was achieved in
other ways. Periodic rotation of the substrate within the coating chamber during a
run was found to even out the deposition pattern of TiO2 coatings considerably. In
addition, the act of increasing the substrate orientational distance relative to the
effluent end of the input gas tube while simultaneously increasing the carrier gas
flow rate improved the TiOj coating uniformity to a high degree by allowing the
gases to disperse and more intimately mix prior to contacting the substrate. Although
53
-------
this procedure had a detrimental effect on coating efficiency (see Figure 10) the
uniformity for thin coatings was greatly enhanced (see Figure 7).
The influence of generally higher temperatures for the SiC>2 coatings
tended to more effectively disperse the plating vapor prior to contacting the sub-
strate, through the medium of thermal energy. As a consequence, relatively uni-
form SiO2 overlay coatings were developed without resorting to the physical and
mechanical dispersion methods described for TiC^- Figure 14 illustrates the depo-
sition pattern for a thick film of SiO2 deposited onto a thin wafer of semiconductor
grade silicon. The coating is approximately 40,000 A in thickness and is interpreted
through the medium of interference colors to be very uniform in the central portions
of the wafer with a gradual tapering off in thickness around the edges. Thinner
coatings would probably be even more uniform.
D. Substrate Considerations
The important role played by the substrate in vapor deposition operations
has been well documented in the literature. In addition to its already discussed
influence on residual stresses, the physical and chemical properties of a given sub-
strate are capable of influencing the nucleation kinetics of a given system (34),
the structure and properties of the deposit formed (35 and 36) and the bonding
strength of the deposit (11).
These important effects are believed to be operative in the system discussed
in this paper. Furthermore, the use of soluble silicate substrates for vapor deposited
coatings represented the addition of another major variable to be considered. These
54
-------
en
en
• .
(A) ^ ^ (B)
Figure 14. Photograph of a representative deposition pattern for SiC>2 deposits. Exhibit (A): coated silicon wafer
Exhibit (B): blank sample. The deposit was derived by bubbling nitrogen through ethyl orthosilicate;
carrier gas flow rate 2 cfh; vaporization temperature 70° F; deposition temperature 1350° F; substrate
orientation 3 inches; time 12 hours.
-------
materials, which have a profound affinity for moisture, must be handled and pro~
cessed under special conditions if controlled coatings are to be obtained on their
surfaces. No reports were found of previous vapor deposition work dealing with
this class of materials.
One may readily gain an insight into the dynamic solution kinetics of a
typical composition used in this study by referring to the sequential photomicro-
graphs of Figure 15, which illustrate three stages of corrosion of a sodium silicate
composition containing 43 percent SiC>2 by weight which was attacked by atmos-
pheric moisture under controlled room conditions of 70 percent relative humidity
and 70° F. The samples shown from left to right were photographed (a) immediately
after removal from the desiccator, (b) 30 minutes later, and (c) 1 1/2 hours later.
The identification of the mechanism of corrosion, and the correlation of
corrosion rates with induced physical and chemical changes are currently being
studied and will be reported in a separate paper. For the purposes of this work,
the accumulated residue shown on the exposed glass surface is believed to be a
precipitated alkaline compound which formed on the surface as a non-adherent solid
after reaction with either the water or carbon dioxide contained in the atmosphere.
The presence of this residue during a coating run generally resulted in
poorly bonded deposits which cracked or spoiled during the cooling cycle, pre-
sumably because the crystalline contaminant effectively blocked the ability of the
plating vapor to form a continuous, chemically bonded film along the glass surface.
Furthermore, the residue usually promoted coating structures which micro-
scopically appeared to exhibit more variability or defect areas than those produced
56
-------
m
(A) (B) (C)
Figure 15. Dissolution kinetics of sodium silicate glass by atmospheric attack with time. Exhibit (A): fresh surface.
Exhibit (B): 30 minutes later. Exhibit (C): 90 minutes later. One inch on photographs equals 107
microns.
-------
on clean surfaces. It is uncertain whether these effects are related to heterogeneous
nucleation sites or other effects.
It was determined that improved coatings were attained on ground and
polished substrates. Those substrates which were not ground and polished generally
exhibited some warpage and the surfaces were microscopically undulating. These
conditions generally produced non-continuous, variable coatings. Removal of all
microscopic contours from the glass surface by grinding to produce a generally flat
surface was found to be more important than polishing the surface to a high reflec-
tance level, in terms of tfie coating quality achieved. These effects may be
explained by considering that the residual shear stress components of the film and
the glass were decreased by a flat surface. Furthermore, it may be assumed that
all mechanically worked surfaces, regardless of the degree of polish, contain some
roughness; slightly microscopically roughened surfaces would probably promote
coating adhesion.
E. Characterization of SiC>2 Coatings
The highest quality SiC>2 glass films were obtained at deposition temperatures
between 1000° F and 1050° F, yielding an average coating growth rate of approxi-
mately 4000 A per hour. Higher deposition temperatures created higher residual
stresses in the deposits, and lower temperatures were found to result in excessively
slow reaction rates and films which were more clouded in appearance. The films
were grown by bubbling nitrogen gas at two cfh through ethyl orthosilicate con-
tained in a flask at room temperature. The substrate was positioned upon one face
in the center of the deposition zone approximately three inches from the effluent
58
-------
end of the input gas tube. The concentration of the reactant gases in the vapor
mixture approximated 800 ppm under these conditions, and the efficiency averaged
2 percent.
Figure 16 shows the relationship between deposition temperature and the
rate of reaction for SiO2 coatings. It is noted that the reaction rate reached a
maximum at 1150 F and decreased at higher temperatures because of premature
decomposition of the plating vapor prior to contacting the substrate.
Various analytical schemes were employed to classify the films obtained.
Visual inspection revealed the appearance of a coating which was clear and glassy
in nature. This observation was reinforced by microscopic examination, which also
brought out the topographical details as seen in Figure 17.
The composition of the film was checked by infrared absorption. The spectra
showed a strong absorption band at 9.2 microns and lesser bands at 12.5 and 21.9
microns, thus corresponding with the known bands for silicon dioxide glass (42).
No evidence of non-stoichiometric silicon oxide could be found.
X-Ray diffractometer scans showed a broad, diffuse band at 4.3 A, thus
indicating the presence of SiO2 glass (43). Some of the higher temperature coatings
were found to have a slightly crystalline nature. This was generally more prevalent
on the bottom face of the substrate where slower deposition rates, higher temperatures,
and impurity nucleating sites probably occurred at the interface between the substrate
and the substrate holder. Under these conditions, the presence of cristobalite was
confirmed by X-Ray analysis. The crystalline material was deposited with a high
degree of preferred orientation, as observed by the fact that diffraction occurred for
59
-------
o
X
UJ
I-
z
2
en
o
o
Q.
LJ
O
o1
900 1000 iioo saoo 1300
DEPOSITION TEMPERATURE (*F)
Figure 16. Plot of SiC>2 deposition rates at different deposition temperatures. The data was derived by bubbling
nitrogen gas through ethyl orthosiiicate at 1 cfh; vaporization temperature 70° F; substrate orientation
3 inches; time 14 hours.
-------
Figure 17. Photomicrograph of surface topography of a typical SiO2 deposit.
The coating was derived by bubbling nitrogen gas through ethyl
orthosilicate at 2 cfh; vaporization temperature 70° F; deposition
temperature 1135° F; substrate orientation 3 inches; time 17 hours.
One inch on photomicrograph equals 53.3 microns.
61
-------
only one of the major intensity peaks for that material—at an interplanar spacing
of 4.15 A. Table IV lists the important diffraction peaks and their intensities for
this substance. Optical examination showed the crystals to be large in size,
approximately 100 microns in diameter. Figure 18 depicts several cristobalite
crystals which have grown and coalesced together. No crystals of comparable size
were found on the top face of the substrate -
Figure 19 reveals the curious appearance of a columnar growth habit which
was found to typify fractured samples, as well as the presence of an interlocking
(mechanical) bond at the interface on roughened substrate surfaces. Based on the
excellent adherence of SiO2 coatings on smooth as well as rough surfaces, it is
believed that the basic source of bond strength is derived from a chemical bond.
The SiO2 as shown by Table V, exhibited a high degree of chemical inert-
ness. In order to measure this property, films were microscopically observed before
and after 30 minute soaks in 50 percent concentrated sulfuric, nitric, hydrochloric,
and acetic acids, as well as in sodium hydroxide. No chemical deterioration was
observed in these media. Only hydrofluoric acid was found to degrade the film,
and it could be removed very rapidly by exposure to this material.
F. Characterization of TiO2 Coatings
Whereas optimized S?O2 coatings were obtained under a somewhat limited
range of conditions, such was not the case for TiO2 coatings, as previously discussed.
Furthermore, it was determined that TiO2 could generally be deposited in either
crystalline or amorphous form at faster rates and with lower residual stresses developed
at the substrate interface than SiO2- Consequently, it was determined at the outset
62
-------
TABLE IV
X-RAY DIFFRACTION DATA FOR COATINGS OBTAINED (44)
CO
CRYSTALLINE
MATERIAL
MINERALOGICAL
NAME
INTERPLANAR SPACING AND RELATIVE
INTENSITIES OF THREE STRONGEST LINES
SPACING, (A) RELATIVE INTENSITY
H ANNA WALT
CARD FILE
INDEX NUMBER
SiO2
TiO2
Ti203
Cristobal ite
Anatase
Titanium
Sesquioxide
4.15
2.53
1.64
3.51
1.89
2.38
1.68
0.90
1.85
100
80
60
100
33
22
100
100
70
4-0359
4-0477
2-1359
-------
Figure 18. Photomicrograph of cristobalite crystals. The coating was derived
by bubbling nitrogen gas through ethyl orthosilicate at 2 cfh;
vaporization temperature 70° F; deposition temperature 1220° F;
substrate orientation 3 inches; time 11 hours. One inch on
photomicrograph equals 53.3 microns.
64
-------
\
Figure 19. Photomicrograph of SiC>2 coating—substrate interface showing
columnar growth habit of SiC^- The substrate is oriented below
the coating. The coating was derived by bubbling nitrogen gas
through ethyl orthosilicate at 1 cfh; vaporization temperature
70 F; deposition temperature 1050 F; substrate orientation 3
inches; time 48 hours. One inch on photomicrograph equals
53.3 microns.
65
-------
TABLE V
CHEMICAL DURABILITY OF COATINGS OBTAINED
COATING TESTED CORRODING AGENT
SiO2 glass H2SO4
HNO3
HC1
HC2H3O2
HF
NaOH
TiO2 amorphous H2SO4
HNO3
HC1
HC2H3O2
HF
NaOH
TiO2 anatase H2SO4
HNO3
HC1
HC2H302
HF
NaOH
EFFECT
Insoluble
Insoluble
Insoluble
Insoluble
Soluble
Insoluble
Insoluble
Insoluble
Insoluble
Insoluble
Insoluble
Insoluble
Insoluble
Insoluble
Insoluble
Insoluble
Insoluble
Insoluble
66
-------
of this work that TiO2 should make a good candidate for coating soluble glass,
if not alone then certainly as a prime coat for a composite coating composed of,
for example, TiC>2 and SiO2. In the latter case, it is believed that the TiO2
would buffer the stress distribution over two interfaces rather than one, and would
also probably enhance the overall protective characteristics of the overlay barrier.
Corollary studies in these areas are currently in progress and will be discussed in
a subsequent paper.
For applications as either single or composite coatings, the use of extremely
thin, uniform, amorphous TiO2 films would be desired because they would be least
objectionable from an optical standpoint. Because of the wide divergence in
refractive indices between substrate and coating, interference colors become
prominent as the thickness increases over 1000 A. An amorphous, uniform film
in this thickness range would reduce the overall light transmission through the
piece, which would appear tinted to the eye. Crystalline films, however, would
tend to reflect the light by scattering, thus decreasing the light transmission to an
even greater extent. Unfortunately, no method was devised to retain the amorphous
structure of TiO2 prime coats during the subsequent SiO2 deposition at higher
temperatures, and the amorphous phase was ruled out except for external coating
applications.
The structure of the deposits obtained over a range of conditions was studied
by X-Ray diffraction. The structure of the TiO2 deposits was found to depend pri-
marily upon deposition temperature and to a lesser extent upon the mass flow rate of
67
-------
the reactant vapor. Figure 20 illustrates the relationship between these two
structure determining variables as derived from a systematic investigation of
approximately one hundred samples. The boundary line, however, is shown
hatched because its accuracy is believed to be limited to plus or minus 25° F.
The transition between amorphous and crystalline deposits was found to occur at
620° F plus or minus 25° F for the range of tetraisopropyl titanate mass flow rates
investigated. Coatings formed below the transition line were amorphous and those
above were crystalline. Curiously, the transition temperature was found to decrease
at extremely low mass flow rates, and several samples analyzed in that range were
found to contain various oxygen deficient modifications of TiO2/ principally
* Crystalline deposits derived from higher mass flow rates were all stoichio-
metric. Table IV contains the essential X-Ray diffraction data for the crystalline
forms of titanium oxide found in this study.
The stoichiometric crystalline deposits consisted of preferentially oriented
anatase. This fact was established by examining the location and intensities of
the three major diffraction peaks for that material and comparing the data to that
produced during an actual X-Ray scan. Two of the three lines, at interplanar
spacings of 3.51 A and 1.89 A could always be identified; the line at 2.38 A was
always characteristically absent, thus indicating preferred orientation of the
deposit.
High quality, uniform thin films of TiCX> were readily produced by bubbling
nitrogen gas at ten cfh through tetraisopropyl titanate maintained in a temperature-
controlled flask at 212 F. The substrate was vertically positioned inside the center
68
-------
lOOOr
800-
u_
o
uj600
cr
or
UJ
a.
5
UJ
400
200
CRYSTALLINE
AMORPHOUS
I 2 3
TPT FLOW RATE (MOLES / MINUTE X IO"4)
Figure 20. Plot of TJO2 structures at different temperature and reactant vapor mass flow rates
-------
of the deposition zone with its front face perpendicular to the impinging gas stream
issuing from the end of the input tube ten inches away. Although the coating
efficiencies were low (less than 3 percent), the uniformity of the coatings was
excellent and the associated film properties such as adherence and chemical
inertness were quite good between 500° F - 800° F. The conditions described allowed
amorphous coatings to be formed at the lower temperatures at growth rates of approxi-
mately 4000 A per hour, and crystalline films were produced at the higher temperatures
at growth rates of approximately 6000 A per hour.
The overall microstructural uniformity of the crystalline deposits was very
sensitive to the coating conditions used. Many of the crystalline samples produced
were found to contain obvious variations when viewed under the microscope.
Several of the most extreme conditions have previously been mentioned in con-
junction with coatings that cracked or spoiled when subjected to a concentrated,
poorly mixed vapor. More typically, the variations observed were much more
subtle, as shown in Figure 21, which depicts two regions of the same sample.
Exhibit (A) was taken from an area which received the direct impact of a poorly
diffused, reactant vapor-rich gas stream resulting in many small grains. Exhibit
(B) was located on the periphery of this impinging stream and consequently received
less cooling effect as well as a more dilute dosage of reactant vapor. It is charac-
terized by a lesser number of larger grains. The effects to be expected from a
consideration of the principles of nucleation and grain growth are exactly as
represented by the two microstructures shown.
70
-------
cz
o
m
••••/ , • -:•'.'•' . SiSjC?^^ i 1
•'--. ;: /. 9 : c^O^M-!^S
•;•.• •• ',.;„',; r • : - '. '• , •?'•
•' v. * i •. i' • K#< •• :-, . " •.'•-'*.'> '}•', {. -. •
'*^- ^. ', vl- '- *'"'" W* '^ i\ * -'''•"•' V -'- ' ••'>%' "'V*'. •'•-''' ••"'"'
< O ™
0^09^
13 ro O- CX
V^tS ^« * s A^a^rS *—*•' * * v - ** ^- 4
^^;V\''*^.l3^^ S ^ .>.'.. "^-Yw
**?* - * »'''.' ft•J*W*l'a?%4": ',i,.,'3.':-i«-'%.ivJi I ' • • " '~f..."'_.
.,.
'- -
»«^i:-lwi*l', i^ef j
•*,*< p -.
i^'^l'jTT'*/1.**-,*1.S->V i-> • *( *.•
*^iiS£vw 2*K ^ *£*&£%
•*4&/at «fc ^%" v / ' • ^i'1^"
o?
cr
-
(A)
Figure 21 . typical anatase microstructures taken from different areas of same sample. Exhibit (A) received a more
concentrated dosage of reactant vapor than Exhibit (B), but was also cooled more by the impinging gas
stream. The photomicrographs were derived by bubbling nitrogen gas through TPT at 10 cfh; vaporization
temperature 70° F; deposition temperature 690° F; substrate orientation 4 inches; time 60 minutes. One
inch on photomicrograph equals 88.9 microns.
-------
Crystalline coatings which were slowly grown were sometimes found to
contain one and two dimensional growth features incorporated into the deposit.
Figure 22 shows two examples of dendritic growth.
A general characteristic of the TiC>2 coatings was their smooth, continuous
appearance. This was especially true for the amorphous films, as shown by Figure
23. This photomicrograph shows a cracked film with some portions stripped away
to reveal the substrate. It was selected to portray the effects denoted above, which
are readily brought out because of the visual contrast.
Figure 24 shows an interesting view of a duplex coating. In this instance,
two 15,000 A layers of TiO2 were deposited around a 45,000 A layer of SiC>2.
The photomicrograph indicates that both coatings were applied evenly and
continuously.
The chemical durability of the TiC>2 films was observed and generally found
to be excellent. The same reagents employed for measuring the chemical inertness
of the SiC>2 coatings were used for both amorphous and crystalline TiO2« The results
are shown in Table V.
G. Other Coatings
Several runs were made with ethyltriethoxysilane as the source material
for the SiC>2 coatings. The coatings were applied in a similar manner to the SiC>2
films derived from ethyl orthosilicate, except that the optimum temperature of
deposition was slightly higher, around 1200° F - 1250° F. Oily traces of partially
72
-------
• •*, f
'"•':;V t' #'''* "^
"
.
(A) ^ (B)
Figure 22. Photomicrographs of secondary growth features in TiC>2 deposits. Exhibit (A): dendrites. Exhibit (B): crosses
These coatings were derived by bubbling nitrogen gas through TPT at 10 cfh; vaporization temperature 70° F;
deposition temperature 976° F; substrate orientation 4 inches; time 60 minutes. One inch on photomicrograph
equals 88.9 microns.
-------
Figure 23. Photomicrograph of surface topography of amorphous TiC>2 film.
Coating was derived from nitrogen gas bubbled through TPT at
10 cfh; vaporization temperature 212° F; deposition temperature
620° F; substrate orientation 4 inches; time 5 minutes. One inch
on photomicrograph equals 53.3 microns.
74
-------
I
en
TiOo —
SiO2 —
TiO2 —
Substrate
Figure 24. Photomicrograph of a duplex coating. Structure is composed of two 15,000 A layers of TiOo with 45,000 A
layer of SiO2 sandwiched between. The TiC>2 coatings were derived by bubbling helium gas through TPT at
9 cfh; vaporization temperature 70° F; deposition temperature 700° F; substrate orientation 4 inches; Hme 3
hours. The SiO2 coating was derived by bubbling helium gas through ethyl orthosilicate at 4 cfh; vapori-
zation temperature 70° F; deposition temperature 1100° F; substrate orientation 3 inches; time 12 hours. One
inch on photomicrograph equals 53.3 microns.
-------
decomposed si lane could be found inside the combustion tube for lower deposition
temperatures. No differences were observed in the SiC>2 films produced from the
two different materials. Because of this factor and the significantly higher unit
cost of the silane, it was not explored further.
Aluminum oxide was deposited from aluminum isopropoxide vapors at depo-
sition temperatures between 550° F - 700° F. The deposits obtained were appar-
ently amorphous A1203. The isopropoxide was not easily vaporized in the flask
that was used for the other plating agents, primarily because of the fact that the
material required heating to 223° F in order to melt it. The coating process used
resembled the technique employed for coating substrates with TiC^. The deposi-
tion rates were slower, however. A^Og low deposition temperature coatings
offer some potential, and should be investigated further.
H. Initial Corrosion Studies
Since the various films produced on Pyrex micros! ides were found to be
highly resistant to boiling water as well as other corrosive media, the major
remaining question concerning the utility of the films revolves around their ability
to form a continuous, impenetrable barrier between the soluble glass and the
corrodent. A thorough analysis of this important subject is currently in progress
and will be reported on at a later date. Nevertheless, several tentative obser-
vations may be made at this point, based on experimental runs made during this study.
Uncoated soluble glass discs were readily dissolved in tap water between
room temperature and the boiling point of water. The rate of dissolution was found
76
-------
to be highly sensitive to the amount of alkali in the glass, the temperature of the
water, and the amount of agitation. Typically, 57 percent sodium oxide and
43 percent silicon dioxide compositions were completely dissolved in less than 15
minutes by agitated cold tap water, whereas the most resistant composition con-
taining 43 percent Na2O typically required over 90 minutes to accomplish the
same task. However, water at 140 F was found to completely dissolve the two
compositions in less than five minutes, and 30 minutes, respectively.
The application of both SiO2 and TiC>2 coatings to these materials was
found to increase the durability substantially. Generally, however, there appeared
to be a very wide variation in the protective quality of both types of films. Some,
in fact deteriorated almost as rapidly as if they were uncoated while others demon-
strated an ability to withstand the ravages of 140° F water for several hours without
deteriorating.
It is thought that any break or discontinuity in the film structure allows
moisture to attack the interfacial bond between the substrate and the coating at a
localized point. Once initiated, the process spreads rapidly across the entire
surface, destroying the protective film.
77
-------
CHAPTER VI
CONCLUSIONS
1. The coating apparatus designed and constructed for this investigation is a
sound laboratory tool, capable of producing many types of chemically depos-
ited coatings over a range of controlled conditions. Its utility is not restricted
to this particular system; rather, it has a broad capability for research work in
many areas of this technology, such as diffusion masking for semiconductors,
and others.
2. Soluble silicate glass compositions are readily coated by chemical vapor depo-
sition processes. The surface condition of the glass must be closely controlled
in order for satisfactory coatings to result.
3. Metallic oxides derived from selected organic esters provide excellent coating
materials for soluble glass substrates. The high vapor pressure and chemical
stability of the mother liquid readily allows coatings to be produced by pyrol-
ysis reactions at the heated substrate surface.
4. Titanium oxide coatings are readily produced over a wide range of controlled
conditions such as deposition temperature, time, carrier gas flow rate,
reactant vapor concentration, substrate orientation, and others. The coatings
produced can be amorphous or crystalline, depending upon conditions. These
coatings are chemically stable, and form c chemical bond with the glass
surface.
78
-------
5. Silicon oxide glass coatings are produced under somewhat more restricted
conditions; however, the coatings formed likewise possess excellent properties.
The problem of residual stress in the coated sample is a greater hazard with
SiC>2 coatings because of generally higher deposition temperatures and a
greater mismatch of thermal expansion coefficients.
6. The high residual stress level contributes to microscopic defects in the finished
coatings. Generally uncontrollable coating conditions also promote defects.
These are typified by variable vaporization rates, non-uniform dispersion of
the vapor species, uneven substrate temperatures, variable substrate surfaces,
and impurities.
U 8 IPA Headquarters Library
Mai! code 3404T
79 -J200 Pennsylvania Avenue NW
Washington, DC 20460
202-566-0556
-------
CHAPTER VII
RECOMMENDATIONS FOR FUTURE WORK
The feasibility of ultimately applying the technology derived from this
investigation to the solution of glass container waste control problems is bright;
however, many basic problems have been found which must be solved in the labo-
ratory prior to significant scaled-up operations becoming a reality.
The significance of this work was that it demonstrated that oxide coatings
may be readily applied to soluble silicate glass substrates by chemical vapor depo-
sition processes with some flexibility in the selection of process parameters used.
This represents in itself an important early finding, fundamental to much of the
work which will follow. Corollary studies are presently being pursued in the areas
of glass corrosion kinetics, mechanical properties, and other types of coating media
As rapidly as possible, new information must be disseminated to all interested par-
ties, to assure that the central objective of the overall project remains in focus
and that wasted motion is minimized.
The major product deficiencies denoted in this work were coated specimens
characterized by high residual stresses and variable microstructures. The author
believes that these problems must be resolved prior to further significant advances
in this area. The solutions may involve new materials systems, equipment modifi-
cations, new procedures of coating, or a combination of all. In all likelihood,
each of the corollary study areas will be correspondingly affected by any changes
that take place.
80
-------
In terms of the specific problems just recounted, a high residual stress level
obviously damages the mechanical and chemical durability properties of the com-
posite structures. A need exists to more closely match the thermal expansion
coefficients of the materials used in order to decrease these effects. As a general
class of materials, the oxides do not match up satisfactorily with the glass compo-
sitions used; in fact, the residual stresses for several coatings produced in this study
were calculated and found to exceed the theoretical mechanical stress limits of the
substances involved. The mechanism of stress relief which allowed the coating and
substrate to remain intact is unknown at the present time, but the example clearly
illustrates the point.
Alternately, methods may be devised to coat substrates at much lower tem-
peratures. Klerer (45) has reported that amorphous SiO2 films may be readily
formed from the vapor phase at deposition temperatures as low as 150 F - 250° F by
going through a radical-coupling step. This technique would represent an inter-
esting solution to the problem and should be investigated.
It is surmised that the variability that currently exists in the coating micro-
structures is at least partially responsible for the erratic tentative results which
have been obtained from an early study of the corrosion rates of the coated samples.
Defects have been microscopically observed in many coatings at relatively low
magnification; this suggests that the defects which are likely present on a truly
microscopic scale as well are producing pinholes and other faults which are the
early failure points during a corrosion evaluation.
81
-------
The question remaining to be answered is how to effectively minimize or
even erase these variations. The author believes that the problem of structural
defects can be largely defeated by devising an improved vaporizer and vapor
delivery system. A basic requirement in this process is that the substrate must
receive a constant, well-mixed and diffused vapor supply to all surfaces. It is
well known that local concentrations of reactant vapor in the total vapor stream
under the influence of elevated temperatures will upset the thermodynamic balance
of the system and possibly cause growth defects to occur during the growth process.
Another area which should be upgraded is that of input materials. Higher
purity coating reagents than the ones employed in this work should be obtained and
used for future studies if possible. Special precautions should be installed to mini-
mize contamination at all levels and from all potential sources. The soluble sili-
cate substrates must be prepared with particular emphasis placed on mixing uni-
formity, annealing, and surface condition. It is believed that each of these items
plays some part in determining the relative number of defects in a coating.
Solution of these problem areas would provide a giant step in the right
direction. Still remaining, however, is the matter of basic cost. The process is
obviously inherently expensive, and the investigator is limited in the possible
significant approaches that might defray costs. During the course of this study,
few conditions were explored that yielded a coating efficiency above 10 percent.
The best coatings that were produced were less than 5 percent, in fact. Reclqim
systems would elevate this figure somewhat, but the basic cost factors such as
expensive plating agents, sophisticated control equipment, highly trained
82
-------
operators, and the like probably point to a basic problem which can only be
improved upon by diligent effort—not solved completely. To illustrate this last
point, compressed dry air should suffice as a substitute for the carrier gases
employed, thus resulting in some savings.
The author believes that a logical long range approach to the program at
this stage of its development would be to design a glass composition which might
be dissolved by the elements, as at a refuse dump, over a period of several months'
duration. This composition would likely possess a lower expansion coefficient
than the ones used in this work; consequently, it is likely that a stress-compatible
coating system could be worked out easier for this material, and the resultant
coated product would thereby possess improved mechanical properties. Since the
durability of the glassy material would also likely be upgraded to a significant
degree, it is possible that the coating microstructural sensitivity factor would be
correspondingly reduced, growth defects might be less critical, and thinner
(cheaper) coatings would thereby suffice. This approach seems to make more prac-
tical sense at the present time than any other.
83
-------
LITERATURE CITED
1. Breidenbach, A.W., "Research Activities Of The Solid Wastes Program Of
The Public Health Service—A Status Report," U.S. Department of Health,
Education and Welfare, Washington, D.C., 1-2, (1968).
2. President's Science Advisory Committee's Report Of The Environmental Pollu-
tion Panel, "Restoring The Quality Of Our Environment/1 Washington,
D.C., HI, (November, 1965).
3. Glass Industry, 92 (1): 40, (1968).
4. Alexander, T., "Where Will We Put All That Garbage?" Fortune Mag.
126(5): 151, (1967).
5. Anonymous, "Municipal Refuse Disposal," American Publ ic Works Association,
47-52, (1966).
6. Anonymous, "Proceedings Of The National Conference On Solid Waste Research,"
American Public Works Association, (1963).
7. Morey, G.W., The Properties Of Glass, Reinhold Publishing Corp., New York,
(1954).
8. Vail, J.G., Soluble Silicates: Their Properties And Uses, Volume II, Rein-
hold Publishing Corp., New York, (1952).
9. Vail, J.G., Soluble Silicates: Their Properties And Uses, Volume I, Reinhold
Publishing Corp., New York, (1952).
10. Plunkett, J.D., "NASA Contribution to the Technology of Inorganic Coatings,"
NASA SP-5014 Technology Survey, Washington, D.C., (1964).
11. Powell, C.F., "Chemical Vapor Deposition," Vapor Deposition, Edited by
C.F. Powell, J.H. Oxley, and J.M. Blocher, Jr., John Wiley and Sons,
Inc., New York, (1966).
12. Blocher, J.M., Jr., "The Promise And Problems Of Chemical Vapor Deposi-
tion," Chemical Vapor Deposition, Defense Metals Information Center
Report 170, Battelle Memorial Institute, Columbus, Ohio, (1962).
13. Accountius, O.E., "Vapor Deposition," Critical Evaluation Of Ceramic
Processing At Subconventional Temperatures, Edited by S.D. Brown, Air
Force Materials Laboratory Report TR-67-194, Wright-Patterson Air Force
Base, Ohio, (1967).
84
-------
14. Powell, C.F., Campbell, I.E., and B.W. Gonser, Vapor Plating, John
Wiley And Sons, Inc., New York, (1955).
15. Powell, C.F., Oxley, J.H., and J.M. Blocher, Jr., Vapor Deposition,
John Wiley And Sons, Inc., New York, (1966). "
16. Sawyer, W.E., and A. Man, U.S. Patent 229, 335 (June 29, 1880).
17. Aylsworth, J.W., U.S. Patent 553,296 (January 21, 1896).
18. deLodyguine, A., U.S. Patent 575,002 (January 12, 1897).
19. Krier, C.A., "Protective Coatings," Vapor Deposition, Edited by C.F.
Powell, J.H. Oxley and J.M. Blocher, Jr., John Wiley And Sons, Inc.,
New York, (1966).
20. Blocher, J.M., and J.H. Oxley, "Chemical Vapor Deposition Opens New
Horizons In Ceramic Technology," Bull. Am. Ceram. Soc., 41 (2): 81-84,
(1962).
21. Walton, D., "The Orientation Of Vapor Deposits," Phil. Mag., 82 (7):
1671-1679, (1962).
22. Secrist, D.R., and J.D. Mackenzie, "Identification Of Uncommon Non-
crystalline Solids As Glasses," J. Am. Ceram. Soc., 48 (9): 487-91,
(1965).
23. Sinclair, R.W., "Inorganic Glass Films—Their Preparation, Properties, And
Uses," The Glass Industry, 92 (1): 22-28, (1968).
24. Holland, L., The Properties Of Glass Surfaces, John Wiley And Sons, Inc.,
New York, (1964).
25. Anonymous, "Tyzor Organic Titanates," E.I. Dupont deNemours And Co.,
Wilmington, Del.
26. Deyrup, A.J., U.S. Patent 2,881,780 (April 22, 1958).
27. Badger, A.E., U.S. Patent 2,881,566 (April 14, 1959).
28. Powell, C0F., "Chemically Deposited Nonmetals," Vapor Deposition,
Edited by C.F. Powell, J.H0 Oxley, and J.M. Blocher, Jr., John Wiley
and Sons, Inc., New York, (1966).
85
-------
29. Oxley, J.H., "Transport Processes," Vapor Deposition, Edited by C.F.
Powell, J.H. Oxley, and J.M. Blocher, Jr., John Wiley And Sons, Inc.,
New York, (1966).
30. Li, C.H., "Epitaxial Growth Of Silicon," J. Electrochem. Soc., 109(10):
952-57, (1962).
31. Jordan, E.L., "A Diffusion Mask For Germanium," J. Eiectrochem. Soc.,
108(5): 478-81, (1961).
32. Vasicek, A., Optics Of Thin Films, Interscience Publishers, Inc., New York,
(1960).
33. Powell, C.F., "Stresses In Deposits," Vapor Deposition, Edited by C.F.
Powell, J.H. Oxley, and J.M. Blocher, Jr., John Wiley and Sons, Inc.,
New York, (1966).
34. Hirth, J.P., "Condensation Processes," Vapor Deposition, Edited by C.F.
Powell, J.H. Oxley, and J.M. Blocher, Jr., John Wiley and Sons, Inc.,
New York, (1966).
35. Coffin, L.F., "Structure-Property Relations For Pyrolytic Graphite," J. Am.
Ceram. Soc., 47 (10): 473-78, (1964).
36. Gretz, R.D., "Structure Of Deposits," Vapor Deposition, Edited by C.F.
Powell, J.H. Oxley, and J.M. Blocher, Jr., John Wiley and Sons, Inc.,
New York, (1966).
37. Handbook Of Organometallic Compounds, Edited by H.C. Kaufmann, D.
Van Nostrand Company, Inc., New York, (1961).
38. Handbook Of Chemistry and Physics, Edited by R.C. Weast, The Chemical
Rubber Company, Cleveland, O., (1967).
39. Turnbull, R.C., and W.G. Lawrence, "The Role Of Titania In Silica Glasses,"
J. Am. Ceram. Soc., 35 (2): 48-53, (1952).
40. Sosman, R.B., The Properties Of Silica, Reinhold Publishing Corp., New
York, (1927).
41. Engineering Properties Of Selected Ceramic Materials, Edited by J.F. Lynch,
C.G. Ruderer, and W.H. Duckworth, The American Ceramic Society,
Columbus, O., (1966).
42. Pliskin, WeA., "The Evaluation Of Thin Film Insulators," Thin Solid Films
2 (1): 1-26, (1968). " '
86
-------
43. Law, H.B0, "The Formation Of Insulating Layers By The Thermal Decompo-
sition of Ethyl Silicate," The Review Of Scientific Instruments, 20 (12):
958, (1949). ~~~
44. Index To The X-Ray Powder Data File, Edited by Joseph V. Smith, ASTM
Special Technical Publication 48-H, American Society For Testing
Materials, Philadelphia, Pa., (1959).
45. Klerer, J., "A Method For The Decomposition Of SiO2 At Low Temperatures,"
J. Electrochem. Soc., 108(11): 1070-71, (1961).
87
-------
APPENDIX I
DETERMINATION OF PLATING VAPOR CONCENTRATION
IN THE CARRIER GAS STREAM
For the purposes of this illustration, assume that it is desired to obtain the concen-
tration of tetraisopropyl titanate (TPT) in nitrogen gas flowing through the system
at a rate of two cubic feet per hour.
1. Calculate the number of moles of carrier gas utilized per unit
of time, using the following relationship from the ideal gas law:
_
ng'RT
where n = the number of moles of carrier gas
Pa = pressure in atmospheres
V = volume in liters per unit of time
R = universal gas constant
T = temperature in degrees Kelvin
It may be assumed that the carrier gas is exhausted to the atmosphere
for purposes of calculation. The universal gas constant expressed in
the choice of units used is 0.082 I iter-atmospheres . The volume is
mole -degree K.
given in cubic feet per hour and must be converted to liters per hour
by a conversion factor of 28 liters per cubic feet. Substituting in the
equation yields
n - (DP* 28) - , , , t -,
n - M Q32\ (300) — ~ moles ot nitrogen gas per hour
2. The next step is to determine the amount of TPT used during that time
interval. This is determined by weighing the flask before and after
the specified time interval. Assuming that 3.8206 grams of TPT were
-------
used, one may divide this number by the molecular weight of TPT
to determine the number of moles of TPT used.
- we'9n* l°ss 'n grams _ 3.8206
nTPT " molecular weight ~ 284.2 " °'0134
3. The resultant concentration is expressed as the ration of the number
of moles of plating compound to the number of moles of carrier gas.
^ nTPT 0.0134 A „,,
C= = 0-55 =0.0061
ng 2.2
The resultant gas stream is composed of 6.1 parts of TPT to 1000 parts
of carrier gas or 6100 parts per million (ppm).
89
-------
APPENDIX II
DETERMINATION OF COATING EFFICIENCY
For the purposes of this illustration, assume that the coating efficiency of TPT is
required during a run.
1. Determine the loss in weight of the vaporizer containing the liquid
plating substance. Assume that this loss amounts to 4.7394 grams of
TPT over a given period of time.
2. Determine the gain in weight of the substrate during the same period
of time. Assume that this gain is 0.0731 grams of TiO2.
3. Write a balanced chemical equation for the reaction taking place to
compute the yield of the reaction, that is, the number of moles of
TiO2 formed by the pyrolysis of 1 mole of TPT.
Ti(C3H7O)4 > TiO2 + 2C3H6 + 2C3H7OH
In this case, one mole of TiO2 is formed from one mole of TPT, or
expressing it in terms of molecular weights, 1 mole of TiO« (79.9
grams) is created every time 1 mole of TPT (284.2 grams) is decom-
posed. Expressed as a fraction, the TiO2 yield from the reaction is
simply 1^ =0.281
4. Multiply the TiO2 yield fraction by the number of grams of TPT
vaporized to obtain the number of grams of TiO2 contained in the
TPT that vaporized. For this example,
(4.7394) (0.281) = 1.332
90
-------
5. The coating efficiency is expressed as a fraction of the TiOo con-
tained in the vaporizing TPT divided into the TiO2 deposited on the
substrate. For this run,
0.0731
Coating Efficiency = , 330 = 0.548 = 5.48%
91
-------
APPENDIX ill
DETERMINATION OF RESIDUAL COATING STRESS
If a substrate and coating are under isothermal conditions during applica-
tion, and the thermal expansion coefficients of both vary in approximately the
same manner, the coating stress Sj, induced by thermal variations in a coating
applied to a flat plate, assuming no plastic flow occurs, can be estimated (33):
=
ST -
EB+2Ec(tc/tB)
where
Sy - coating stress in pounds per square inch
En - elastic moduli of base in pounds per square inch
E£ - elastic moduli of coating in pounds per square inch
AT = temperature change from deposition temperature, in degrees
Fahrenheit
°*B' °^C = l'near thermal expansion coefficients of the base and coating
respectively, in reciprocal degrees Fahrenheit.
*C' *B = thickness of the coating and base respectively, in inches
If AT («*B - «<£) is positive, a tensile stress will be induced in the coating
and a compressive stress in the substrate. If the term is negative, tfie reverse will
occur.
92
-------
PART 11
MECHANICAL PROPERTIES OF WATER-SOLUBLE SODIUM SILICATE GLASSES
-------
ABSTRACT
A method for mechanically testing sodium silicate glass by the
bending test was presented. Five glass compositions between 1.ONa^O*
l.OSiO and 1.0-Na^O'1.6$iCL are investigated. The strength was ex-
amined as a function of four variables: composition, re-anneal time,
atmospheric exposure, and coatings.
This work represents the second segment of a two year investiga-
tion to determine a feasible method of processing glass containers
after use.
The description of fabricating and testing rectangular glass bars
3/8" x 3/8" x V was described, with the author's interpretation of
the results.
Sodium silicate glasses between the composition range l.ONa^O-
1.3Si02 and 1 .ONa20-1.65iO-2 possess adequate strength to be utilized
as a container material. However, only after the completion of con-
current studies on dissolution kinetics and coatings can a just ap-
praisal of the feasibility of this project be made.
-------
PART II
TABLE OF CONTENTS
Chapter Page
I. INTRODUCTION ]
II. REVIEW OF LITERATURE 2
Solid Waste Disposal 2
Soluble Silicate Glasses ^
Physical Properties 5
Strength of Glass 12
Measurement of Strength 13
III. OBJECTIVE OF RESEARCH 16
IV. EXPERIMENTAL PROCEDURES , 17
Melt Preparation 17
Specimen Fabrication 17
V. PRESENTATION AND DISCUSSION OF RESULTS 27
Composition Analysis 27
Effect of Re-annealing 31
Atmospheric Exposure 31
Coatings > 36
VI. CONCLUSIONS 40
VII. CONTINUANCE OF RESEARCH 41
LITERATURE CITED 43
APPENDIX ....'.'.., 45
-------
PART 11
LIST OF TABLES
Table Page
I. Composition by Weight of Glasses Investigated 18
II. Annealing Temperature for Glass Compositions
under Investigation 26
iv
-------
PART II
LIST OF FIGURES
Figure Page
1. Refractive Index Versus Composition 8
2. Density Versus Composition 7
3. Linear Thermal Expansion Below the Transformation
Temperature q
k. Linear Thermal Expansion Above the Transformation
Temperature
5. Transformation Temperature ................ 11
6. Photograph of Cylindrical Disk Mold ........... 20
7. Photograph of Disk Specimens being Prepared , ...... 21
8. Photograph of Rectangular Specimens being
Prepared ....................... 22
9. Photograph of Bubble Defects in Rectangular
Specimens ....................... 23
10. Photograph Showing the Effect of the Addition of
As?0_, as a Fining Agent, on the Incidence of
Bubbles in the Test Specimens ............. 25
11. Plot of Bending Strength Versus Composition ....... 28
12. Photograph of the Fracture Patterns for the
Specimens from the Two Types of Molds ......... 30
13. Plot of Percent Change in Strength Versus Re-Anneal
Time for a 1 .ONa20 ' 1 . 3Si02 Composition ........ 32
14. Plot of Percent Change in Strength Versus the Time
of Atmospheric Exposure ................
15. Illustration of How Surface Flaws are Eliminated
after Exposure to the Atmosphere ........... 35
16. Series of Photographs Showing the Absorption of
Moisture by the Surface of a 1 .ONa^O- 1 .2S50
Composition ..................... 37
-------
CHAPTER 1
INTRODUCTION
The use of glass has so saturated the market of commodities distri-
bution, that glass has become one of the largest and most troublesome
constituents of solid waste disposal. The high chemical durability that
glass possesses makes it indispensable in the containment of corrosive
systems, but renders it a formidable opponent on the trash heap. The
formulation and design of a glass container whose durability and sus-
ceptability to environments can be controlled would greatly assist in
the degradation of such containers.
The solubility of sodium silicate glasses has been observed as a
natural oddity in the field of glass technology. Morey (l) points out
that special attention should be called to the Na^O-ZSiO^ quartz eutec-
tic. There is a remarkable melting point lowering of about 1000°F after
the addition of 25% Na«0. The low temperature of this eutectic is of
vital importance to glass technology. More importantly, Morey suggests
that glass of this composition would be excellent "were it not easily
attacked by water."
If a suitable coating could be developed that provided an impene-
trable barrier for water vapor, certain sodium silicate glass could be
utilized as container material.
This paper is the second in a series of papers that investigates
the feasibility of using sodium silicate glass as a method of improving
the degradement of solid waste by rendering glass easily disposable.
-------
CHAPTER 11
REVIEW OF LITERATURE
Glass making is older than recorded history. No one can say when
or where or who first discovered how to make glass. Perhaps the dis-
covery resulted from an experiment, but more likely the discovery was
accidental. At some time in history the conditions existed where some-
one heated sand and soda over a hot fire and upon cooling discovered
this vitreous substance - glass.
Sol id Waste Disposal
The United States is facing a staggering problem in trying to dis-
pose of the immense amounts of solid wastes that are the unwanted resi-
due of our nation's achievements in production and consumption. Our
cities and their surrounding urbanized areas are already bearing the
brunt of our explosive growth and increased industrial activity. This
country has to deal with 3-5 billion tons of solid waste each year.
This consists of 48 billion cans, 26 billion bottles, more than 30
million tons of paper, 4 million tons of plastics, and 100 million
rubber tires weighing a million tons (2, 3). With this soaring problem
in solid waste disposal becoming more critical, the general public is
not aware of the problem. A vivid illustration of public apathy can
be seen in the attitude toward the non-returnable glass bottles which
relieves the producer, distributor, and consumer of handling cost and
inconvenience, but creates more burden for refuse collectors. It is
unfortunate that most people are not aware of the additional problems
that have been created by this convenience item. The refuse collector
2
-------
is aware of his increased burden, however, and the use of non-returnable
bottles is increasing at an alarming rate (^).
Initial action was instigated by passage of the Solid Waste Dis-
posal Act of 1965, Title 11 of Public Law 89^272, which President John-
son signed on October 20 of that year. This new legislation directs
the Secretary of Health, Education, and Welfare to initiate, encourage,
and support a national program aimed at discovering and evaluating
better methods of coping with the solid waste problem (5, 6).
Containers, with respect to solid waste disposal procedures, can
be divided into two classifications: combustible and non-combustible.
Containers made from paper, wood, or plastic are the major portion of
the organic compounds of refuse. Combustible containers have heat value
and when dry will burn freely in incinerators without forced draft and
without fuel.
A non-combustible container is any container that is unburnable at
ordinary incinerator temperatures (1300° to 2000°F). Although metal
containers undergo slow disintegration by oxidation, this process is
far too slow to be employed in connection with incinerator procedures.
Carelessly stored, used containers are aesthetically objectionable and
may harbor rodents and other vermin. Open cans and bottles catch and
hold water in which mosquitoes can breed, so that many individual
•
citizens unknowingly but actively encourage the proliferation of these
disease carrying pests (7).
Funds and leadership for research and planning programs to curb the
undesirable effects of inadequate solid waste control are starting to
be provided (8).* Essential to this effort is a method of processing
-------
incombustible, unreactive glass containers after use. The potential
impact to such a development upon solid waste control may be readily
visualized by considering that approximately one hundred thirty-five
glass bottles and jars alone are produced each year for every person
in the United States. This amounts to twenty-six billion glass con-
tainers each year (3). The control of the disposal of glass would be
a tremendous step toward the control of all solid waste.
Soluble Silicate Glasses
With the development of glass blowing techniques, the formulation
of glass containers was directed to yield high chemical durability.
The resistance which glass offers to various corrosive agents is of
great practical significance. Morey (1) points out that the cKemical
durability of glass is often times the chief reason for its preference
over competing materials.
There are six important types of glass used today: soda-lime,
lead alkali, borosi1icates, aluminosi1icates, 36% silica, and fused
silica. The most common kind of glass is soda-1ime-si1ica glass com-
posed of silica sand, soda, ash, and limestone. The addition of in-
creasing amounts of soda to silica while decreasing other constituents
significantly lowers the melting and softening temperatures. However,
the chemical resistance is also impaired, and if enough soda is used
the glass is readily attacked by many environments (1).
In the past the properties of alkali silicate glasses, like sodium
silicate glass, have been investigated more from scientific curiosity
than from practical scientific investigation. Soluble silicates have
been desired more for other industrial uses than for glasses. They
-------
have been used for such applications as adhesives, cleansers, cements,
deflocculants, and protective coatings (9).
The alkali silicate glass system would offer some inherent advan-
tages if adopted for glass container production. Vail (9) points out
that sodium silicate gels are now added to polluted water as a
flocculant in the settling and filtration steps of water purification.
At the same time these gels form solutions that tend to form a thin
film on the inside of water pipes which protect them against water
corrosion. These are important by-products of soluble silicates.
Morey (1) has the most complete record of the investigation of
sodium silicate glasses. However, the range of his investigation is
limited between 100% Si02 - 0% Na20 and 70% Si02 - 30% Na20, and this
did not extend into the range of compositions under investigation.
Morey points out that with the high SiO^ composition, a sodium disili-
cate (Na^O'ZSiO^) >s formed that is easily made into a glass, while as
the soda content increases, a second compound is formed: sodium meta-
silicate (Na~0*SiO?) . It has been shown by a number of workers (10, 11,
12) that the range of glass formation in the Na20 - Si02 system is con-
tinuous from SiO? to a limiting composition close to that of the meta-
silicate Na70-Si09. The limiting composition depends on the experimen-
tal conditions. As is to be expected, the smaller the melt and the
greater the rate of cooling, the higher the Na20 content at which the
material can still be made as glass free from any crystalline material.
The work of Moore and Carey (11), working with 6 gram melts, established
the limiting composition as kl mole percent Na20. Imaoba and Yamazaki
(12), in a general survey of glass forming regions in simple silicate
-------
systems, found it possible to make glasses containing up to 58.8 weight
percent Na?0 (57.8 mole percent) on a scale of 1-2 grams. The different
results for the Na?0-SiO? system clearly show the effect of the scale
of melting, and illustrate the importance of specifying as clearly as
possible the melting and cooling conditions when reporting the region
glass formation in a system. It may be significant that the limit in
the Na-0-SiCL system coincides almost exactly with the composition at
which Na?0 becomes the primary crystalline phase, that is, the first
phase to crystallize from the melt. At this composition a change will
occur in the nature of the crystallization process which will be re-
flected in the kinetics of the process. Scholes (13) substantiates
the fact that sodium metasilicate can be obtained as a glass, but only
in small quantities. However, to meet the specifications necessary for
use in the container industry, it is only necessary for these glasses
to meet the qualifications of a water soluble container and not those
that make it a glass.
Physical Properties
The knowledge of the physical properties of sodium silicate glass
is somewhat limited. There have only been a few researchers that have
done extensive work with these glasses because its solubility has
greatly hindered any thorough investigation. Not only have the inves-
tigations been limited, but they have produced significant discrepancies
in these researchers' results. Some properties that are easier to ob-
tain than others have been sufficiently substantiated. These include
changes in density, refractive index, and linear thermal expansion
»
below and above the transformation temperature. These properties are
-------
2.6 O -
2.20
100
0
90
10
80
20
70
30
60
40
50
50
40
60
% Si
% N
COMPOSITION
FIGURE I.
DENSITY VS COMPOSITION
-------
1.52 -
00
X
UJ
o
O
<
-------
X
H
yj
26
24
22
CO
z
CL
X
UJ
-I
2
ce
• - I
20
18
16
14
12
1O
1
100
0
I
90
10
s
so
20
I
70
30
i
60
40
1
50
50
I
4O
6O
% Si
FIGURE 3.
COMPOSITION
LINEAR THERMAL EXPANSION BELOW THE
TRANSFORMATION TEMPERATlfRE.
-------
o
X
o
CO
CL
X
UJ
IT
UJ
80
70
60
50
40
30
20
1
I
1
I
o BUREAU OF STANDARDS
SAMSOEN
1
1
100
0
FIGURE 4.
90 80 70 60 50
10 20 30 40 50
COMPOSITION
LINEAR THERMAL EXPANSION ABOVE THE
TRANSFORMATION TEMPERATURE
40
60
SI0
-------
o
e
111
OC
CC
UJ
a.
450
Ul
400
O
350
o
b.
CO
300
a BUREAU OF STANDARDS
O SAMSOENS
100
0
90
10
80
20
70
30
60
40
50
50
40
60
%C i
O 1
FIGURE 5.
COMPOSITION
TRANSFORMATION TEMPERATURE
VERSUS COMPOSITION
-------
shown in Figures 1, 2, 3, and 4 respectively and are taken from
Morey (1). An example of the discrepancies in results is seen in
Figure 5 which shows the change in the transformation temperature with
changes in composition. The results from Samsoens (14) and the Bureau
of Standards (1) are used.
Strength of Glass
"Even when conducted with utmost care, studies of the mechanical
properties of glass are plagued by what seems to be an inherent and un-
avoidable scatter, a condition stemming from the susceptibility of the
glass surface to flawing by either mechanical or chemical action (15)."
it is practically impossible to predict the strength of a particu-
lar specimen of glass, except within very broad limits. Only the
average obtained from a large number of strength measurements has real
significance. An average strength figure alone does not characterize
the property measured since it is also necessary to know the extent to
which individual measurements are likely to depart from the mean. This
fact makes it necessary to approach the measurement of the strength of
glass with a statistical procedure.
It is widely accepted that glass fails only under tensile stres-
ses. The strength of glass is conditioned by the existence of flaws or
weak points in the material. Several observations regarding the
strength of glass are summarized below.
(1) The strength of a massive glass specimen measured by
bending or tension experiments is much less than pre-
dicted by any theory which assumes that strong inter-
molecular forces co-operate in a regular way.
12
-------
(2) The strength of similar pieces of glass broken in
tension may vary widely and only average figures
hold any significance.
(3) The strength of glass is easily affected by the
treatment of the material, and in particular the
factors affecting the conditions of the surface,
such as environments, and annealing times and
temperatures.
The Griffith Flaw Theory (2k) provides an explanation for the difference
between the theoretical strength and the experimentally observed results
from the presence of small cracks or flaws around which a strong stress
concentration arises when the solid is stressed. The first and third
statements above can be used to substantiate the Griffith Flaw Theory.
The strength of massive glass specimens may also vary as a function of
the geometrical shape and size. Holland and Turner (16) have shown
that for specimens tested in bending, the flexural strength decreases
as the thickness is increased.
In summary, any attempt to find the strength of glass as a func-
tion of any other parameter is a very difficult task. Some (17) feel
that there is no correlation among glass parameters, such as strength
versus composition, while there are others (1, 19, 20, 21, 23) that
disagree.
Measurement of Strength
Two methods can be used to measure strength of brittle material.
The most direct way to measure strength is to take a rod or fiber with
enlarged ends and load it with a gradually increasing weight until the
13
-------
specimen breaks. If the cross sectional area of the specimen is A, and
the load is W, then the nominal breaking stress is W/A. The value of
W/A obtained by this method of testing will usually be much greater
for fine fibers than for large rods due to the greater probability of
encountering a flaw in a larger cross sectional area than a smaller
one. The direct measurement of tensile strength by rod or fiber
method is not so convenient as the indirect method using a bending
test. For the direct test it is necessary to work with specimens with
enlarged ends while with the bending test it is not.
A rod can be easily tested by bending it under a load and simply
measuring the load required to break the specimen. If the load W
pounds is applied at the center of a span L inches in length, the
maximum bending moment developed is WI_A inch-pounds. This maximum
bending moment occurs in the cross section of the beam directly under
the load, that is, at the center of the span. If the section modulus
of the cross section is A, then the maximum tension developed Is WL/4A.
This parameter is called the "modulus of rupture." Assuming the beam
to be symmetrical above and below the central axis, the value of A wi 11
be 21/d, where 1 is the flexural moment of Inertia, and d is the depth
of the beam. For different geometrical configuration the value of A
will change. In the rectangular specimen used, the value of A is
2
bd /6 where b and d are the specimen dimensions respectively, and the
r\
value of the modulus of rupture for rectangular beams is 3WL/2bd .
As Scholes (13) emphasized in his work on the strength of glass,
the beam may not break directly under the load, where the stress is
greatest, but at some other point along the span. This happens because
14
-------
in glass the weakness of its surface is being tested, and not the
strength of the hulk samples. This fact holds more importance when a
>
glass, like the sodium silicate glass, is attacked by the constituents
of the atmosphere.
Rudnick, Hunter, and Holden (21) used a third method for deter-
mining the strength of brittle materials, that of the diametral-compres-
sion test. However, this is limited to a specific fracture pattern.
15
-------
CHAPTER 111
OBJECTIVES OF RESEARCH
The basic objective of this research was to expand the knowledge
of the mechanical properties of sodium silicate glass. This work
represents the second segment of a projected two year study which will
deal with the feasibility of, and methods for, the degradement of glass
packaging containers. The research was composed of three phases.
The initial phase concerned a method of fabricating suitable test
specimens that when tested would yield representative mechanical data.
This included the selection of methods for testing.
The second phase concerned physical testing of specimens and the
interpretation of results. This was achieved by relating strength as
a function of four variables: composition, anneal time, exposure time,
and coatings.
The final phase involved a critical assessment of the present
procedure and recommendations for certain changes in both the method
and procedure. The basic limitations of present methods are discussed
with ideas for future research presented.
16
-------
CHAPTER IV
EXPERIMENTAL PROCEDURES
Melt Preparation
The preparation of melts consisted of mixing the desired quantity
of Baker Analyzed Reagent Grade Na^CO- and Pennsylvania Pulvarizing Com-
pany's highest quality silica. The ratio necessary to yield the desired
composition was calculated from the basic equation:
+ b Si02 + c C0£ + + d Na20- e Si02
where a, b, c, d, and e are the molar ratio determined by the desired
composition. The percent by weight for the five compositions investi-
gated are given in Table 1.
The preparation was accomplished by weighing out the desired quan-
tities of material and thoroughly mixing them in a mortar and pestle.
The mixture was initially fired in common fireclay crucibles. However,
erosion of the crucible at the glass-air interface necessitated the use
of another crucible material. The selection was a high quality alumina
crucible. The desired composition and the alumina crucibles were heated
slowly in a silicon carbide resistance furnace to reduce the problem of
thermal fracture to the crucible. After the melting temperature of
1250°C was reached, this temperature was maintained for a minimum of
sixty minutes to allow for adequate soaking.
Specimen Fabrication
The fabrication of adequate test specimens proved to be the most
formidable obstacle encountered. Initially, cylindrical^ shaped disks
one and one-half inches in diameter and approximately three-eighths of
17
-------
TABLE 1
COMPOSITION BY WEIGHT
OF GLASSES INVESTIGATED
GLASS COMPOSITION
1 .0 Na20 •
1 .0 Na20 •
1 .0 Na20 •
1.0 Na 0 •
1 .0 Na 0 '
1.0 Si02
1.1 Si02
1.2 Si02
1.3 Si02
1.6 Si02
WT % Si02
49.2
51.6
53.7
55.6
60.7
WT % Na20
50.8
48.4
46.3
44.4
39.3
18
-------
an inch thick were cast in the steel mold shown in Figure 6. These
disks were made to fulfill a three-fold purpose. It was hoped that
these disks could be used as a suitable sample to determine strength
through diametral testing, solution rates, through rate of dissolution,
and to, determine the degree of workability of the glass. These disks
proved unsatisfactory for the diametral testing technique because they
broke with the pattern that was not typical for tensile failure. The
diametral compression test is limited to only tensile failures so
another testing method had to be utilized. Other test methods were
investigated for strength determination even though the disks continued
to be used for solubility studies. ASTM,industry, and other researchers
(17, 18, 19, 20) provided several testing methods for the modulus of
rupture of glass but these tests were for sheets of glass and were not
applicable. The lack of correlation between the strength of bulk sam-
ples and the strength of fibers tested in tension prevented the use of
tensilely testing fibers.
The,bending test was then chosen to yield strength in the form of
modulus of ruptures (MOR). This was accomplished by molding rectangu-
larly shaped bars and breaking these bars in three point loading. The
initial material used for the mold was graphite because it was readily
available and glass does not wet it. Another mold of mild steel was
later made. Several molds were made in an attempt to find a practical
size. It was decided that bars three-eighths inch by three-eighths
inch by four inches could both be adequately fabricated and tested.
Figures 7 and 8 show specimens being molded in both types of molds.
Initially the bars were of poor quality as seen in Figure 9. How-
19
-------
Figure 6. Photograph of initial disk mold
20
-------
Figure 7- Photograph of a disk specimen being poured.
-------
ro
PO
Figure 8. Photograph of bar specimen being poured
-------
i O
Figure 9. Photograph of the bubble defects encountered in initial test specimens
-------
ever, as the proper technique was developed the quality improved. No
fining agent was needed in the earlier melts because the disk specimens
were pressed with sufficient force to squeeze the bubbles from the
glass. No prevision was made to press the rectangular specimens so 2%
by weight of As?0~ was added to help sweep the bubbles from the melt.
Figure 10 shows the effect of various amounts of As^O, on the incidence
of bubbles.
During the initial attempts to mold the disk specimens, the mold
was heated by a gas or electric heater to raise the temperature of the
mold close to that of the annealing temperature. However, this had no
effect on the quality of the test specimens so the heater was removed.
After the glass specimens were cast, they were immediately placed in an
annealing furnace and held at the annealing temperature suggested by
Morey and listed in Table 11. The annealing temperature was maintained
for thirty minutes and then the furnace was shut down and allowed to
cool with the specimen inside. Upon reaching atmospheric temperature,
the furnace was opened and the samples were removed and placed in a
desiccator for storage and transportation to the testing area.
The three point bending method was used for the bending test.
This was performed on a Baldwin-Tate-Emery testing machine. A span of
two inches, and a loading rate of .05 inches/minute was used. The data
was collected on the basis of readings to the nearest pound.
The same procedure was used for the preparation, fabrica'tion, an-
nealing, and storage of all specimens tested.
24
-------
l\3
c n
Figure 10. Photograph showing the effect of the addition of As9CL, as a fining agent, on the incidence
£- j
of bubbles in the test specimens ( From top to bottom respectively, 0%, i%, 1%, 2% As?0 ).
-------
TABLE 11
COMPOS
1 .0 Na?0 •
1,0
1.0
1.0
1 .0
noN
1.0 Si02
1.1
1,2
1.3
1.6
ANNEAL ING 'TEMPERATURE °C
415
440
450
450
420
26
-------
CHAPTER V
PRESENTATION AND DISCUSSION OF RESULTS
Since the strength of glass is not understood in the same detail
as the strength of other brittle materials, it is very difficult to
explain exactly what happens under every condition. The strength of
glass is primarily controlled by flaws on the surface and secondarily
those in the interior. It is this author's intent to first present
his results and then give the explanation he feels the results show.
Composition Analysis
The first systematic investigation of the effect of composition on
the strength of glass was that of Gehlhoff and Thomas (19) in 1926.
Some (17) felt that it was almost impossible to draw convincing conclu-
sions from their results, where others (1, 19, 20, 21, 23) have con-
sidered the effect of composition on strength.
The results of the present work on the effect of composition on
the strength of glass is shown in Figure 11. It should be noted, that
the results from Morey (l), and Gehlhoff and Thomas (23) are also
depicted and the results from Morey's work with low Na20 content is
used. The points represented by the triangle are the averages, with the
range given as a bar above and below the average. The results for the
strength of the specimens from both the graphite mold and the steel
mold are shown.
The rectangular specimens made in the steel mold had a smoother
surface than those produced in the graphite mold. Thus, it seemed
logical that those specimens would be stronger. However, upon testing,
27
-------
ro
oo
CM
CO
00
O
O
O
O
LJ
a:
CD
Z
O
Z
111
ffl
32
28
24
20
(6
12
8
4
• GEHLHOFF AND
THOMAS
Q MOREY
^ GRAPHITE MOLD
O STEEL MOLD
I RANGE OF POINTS
100
0
1
90
10
1
80
20
t
1
70
30
«s\ tji e*t\& i-
I
60
40
n f\ k.i
I
50
50
1
40 (
60
% Si02
% Na20
FIGURE If. BENDING STRENGTH (MOR) VERSUS COMPOSITION
-------
the specimen from the steel mold revealed much lower strengths than
those from the graphite mold. This difference is also seen in Figure
11. Also, there was a distinct difference in the fracture patterns
between the two types of specimens. Figure 12 shows the different
fracture patterns. The difference in thermal conductivity between the
steel mold and the graphite mold may have produced the differences in
the strengths by altering the cooling rates for the specimens. This
change in the cooling rate also led to the difference in fracture
patterns. The leaching of iron or other substances from the steel mold
may be another factor that influenced the surface of the specimens and
reduced their strengths. These impurities would form complex glasses
on the surface that could lead to a heterogenity between the surface
and interior glass.
The rapid decline of the strength of the specimen from the graphite
mold with an additional increase in the Na«0 content past bk% is due to
the saturation of the SiCL structure's interstitices by sodium ions.
Coupled with this fact, the Na20 phase is becoming the primary crystal-
line phase, that is, the first phase to crystallize from the melt.
These processes would produce large clusters of Na ions or crystalline
Na20 that would disrupt the Si02 structure and reduce the strength. The
compositions of higher silica contents fit the projected curve of the
work by Morey.
The change in color with the change in composition was a noted
visable change. The specimens of higher silica content, 1.0 Na20.1.4 Si02
and above, possessed a clear and very transparent color. However, as
the N»?0 content was increased, the presence of an aqua-bluish tint
29
-------
0
:•
Figure 12. Photograph of the fracture patterns for the specimens from the two types of molds. The pattern
representative of the steel mold is on the left, and the pattern representative of the graphite
mold is on the right.
-------
appeared and increased until the 1.0Na20- 1.0Si02 specimen became very
blue-green. The change in the color could be from the presence of the
sodium ion, other ions leached from the crucible by the higher Na+
concentration, or a change in the valence state of the iron impurities.
Effect of Re-annealing
Since the specimens from the steel mold displayed significantly
lower strength than specimens from the graphite mold, they were re-
annealed to determine what effect this would have on their strengths.
Figure 13 shows the results from reannealing over a 2k hour period.
The composition used for this experiment was 1 .ONa-0- 1 ^SiO- because
concurrent studies indicate that this composition exhibits the dissolu-
tion kinetics required for adequate hydrolyzing. The relief of resi-
dual strains left from the first annealing period resulted with an in-
crease in strength. However, after a period of nucleation, there is
sufficient grain growth of Na^O crystals to reduce the strength by
disrupting the SiO« structure. Eventual complete devitrification
would follow. This crystal growth would then decrease the strength as
is shown after a re-annealing period of seven hours. This does not indi-
cate that these specimens would require that length of time for anneal-
ing but indicates that more research should be done on mold materials
and annealing schedules.
Prolonged Atmospheric Exposure
The 1 .ONa-0-1 .3SiCL composition was also used in this experiment.
The specimens were periodically removed from a desiccator and exposed
to the atmospheric conditions dictated by a dry bulb temperature of 77°F
31
-------
CO
ro
h-
O
z
LU
IT
CO
O
LU
CD
LU
0
Z
O
111
O
flC
LU
Q.
-25 -
O AVERAGE
I RANGE OF POINTS
8
10 12
TIME
14 16
HOURS
18 20 22 24
FIGURE 13. EFFECT OF RE-ANNEALING A I.ONa20-l.3Si02 COMPOSITION
AT 45O°C ON ITS BENDING STRENGTH.
-------
and a wet bulb temperature of 71°F (relative humidity of 6]%). Figure
]k shows the percent change in strength for the times indicated. Ini-
tial examination of this figure indicates a peculiar situation, a
decrease and then an increase in strength.
These changes in strength can be explained through the examination
of Figure 15, which shows the cross section of a specimen at four
important time periods. The strength of glass is primarily surface
dependent so the introduction of surface flaws decreases the strength
by increasing the probability of propagating a crack through the sample,
Figure ISA shows the unflawed surface of a molded sample. Figure 15B
shows the flawed surface caused by abrasion from atmospheric dust and
water attack. These additional flaws disrupt the surface and cause a
decrease in the strength of the sample. As the silicate glass surface
continues to absorb moisture and other atmospheric gases, it erodes
these areas of higher energy, or flaws with small radii of curvature.
This erosion, as seen in Figure 15C, forms a layer across the surface
of the sample. The removal of these highly stressed flaws results in
an increase in strength. Applying the Griffith Flaw Theory to the con-
ditions present in Figures 15B and 15C, the actual stress concentration
at a crack tip can be expressed by:
am =
P
and
am = actual stress for failure
a = theoretical stress for failure
33
-------
O
Z
UJ
a:
CO
CO
UJ
CD
O
100
50
-50
-100
0
1
1
1
2
1
3
1
4
I
5
i
6
1
7
FIGURE 14.
TIME , HOURS
PERCENT CHANGE IN STRENGTH
OF ATMOSPHERIC EXPOSURE
VS THE TIME
-------
ATMOSPHERE
GLASS
~
A. SURFACE AS FORMED
GLASS
B. SURFACE AFTER BEING FLAWED
SILICA-CARBONATE LAYER
C. SURFACE AFTER SHORT EXPOSURE
SILICA-CARBONATE LAYER
D. SURFACE AFTER PROLONGED EXPOSURE
FIGURE 15. ILLUSTRATION OF HOW SURFACE FLAWS
ARE ELIMINATED AFTER PROLONGED
EXPOSURE TO THE ATMOSPHERE
35
-------
c = one half the axis length of the crack
p = radius of curvature of the crack tip
it can be shown that as the radius of curvature of the crack tip in-
creases the strength increases by decreasing the probability of propo-
gating the crack through the specimen. When there is sufficient erosion
of the glass to produce a reduction in the cross sectional area as il-
lustrated in Figure 15D, there is a reduction in the overall strength.
The surfaces shown in Figure 16 support the author's theory on the
change in strength. As seen in this series of pictures, the layer
grows across the surface until it completely covers the surface. This
completion occurs between four to five hours and coincides with the
leveling off and decline of the curve shown in Figure 14.
The author is not in a position to say of what this surface layer
is composed. However, Phillips (21) says that sodium silicate is
leached from the glass and undergoes hydrolysis so that the layer con-
tains sodium hydroxide and colloidal silicic acid. In the presence of
carbon dioxide a further reaction occurs between this gas and the al-
kali, resulting in the formation of sodium carbonate and a surface
layer of finely divided silica.
Coating Evaluation
There are three types of coatings under investigation for use as
a moisture barrier. Organic epoxies, ion exchanged surfaces, and a
chemical deposited surface are under simultaneous study. It is impos-
sible to conclusively evaluate their effect on strength at this time.
The only coating available for testing was the organic epoxy. This
was applied to a heated specimen (200°C) by passing the specimen-through
36
-------
CO
....... ••
:•,.
'
A. 30 Minutes B. 60 Minutes C. 90 Minutes
Figure 16. This series of photographs shows the absorption of moisture by the surface of a 1.ONa20•1.2Si
glass and the growth of the resulting layer across the surface (2000X).
-------
CO
oo
0. 120 Minutes
E. 180 Minutes
F. 2MD Minutes
Figure 16. Con't
-------
a fluidized bed of the powdered resin. Upon cooling, the specimens
were placed in a desiccator and stored until testing. As in the other
tests, the 1.ONa20*1-3Si02 composition was used. The results of
testing yielded a decrease in strength by approximately ]2%. This is
a very incomplete experiment at this time; however, continuation of
present work should yield more pertinent information concerning coatings
39
-------
CHAPTER VI
CONCLUSIONS
1. The glass compositions from the 1 .ONa20-1.3Si02 to the l.
1.6SiO range seem to possess sufficient strength and ease of
workability to be used as a container material. However, no
decision for the selection of a definite composition can be made
until all system parameters are evaluated.
2. The technique of molding glass test specimens is not the optimum
procedure for producing samples because it is too time consuming.
Another method, like pulling glass rods, should be developed to
mass produce specimens.
3- The formation of the silica-carbonate layer on the surface of
the glass after prolonged exposure indicates that some type of
atmospheric control will be necessary during the production
process.
k. The change in strength between the specimens from the two types
of mold indicates that more research into mold materials is
needed.
5. The change in strength of the specimen formed in the steel mold,
after a period of rean.neal i nq, indicates that an optimum period
of annealing exists'and continued research in this area is needed.
6. The. addition of 2% by weight of As^O- is needed as a bubble re-
lease agent to sweep the melts of bubbles prior to molding.
40
-------
CHAPTER VII
CONTINUATION OF RESEARCH
At present there are corollary studies being pursued in the areas
of glass dissolution kinetics, organic coatings, ion exchanged sur-
faces, and a continuation of the work on chemical vapor deposition
(25). All these areas will influence the composition selection and
determine the feasibility of this project.
This work has answered but a few of the questions concerning the
feasibility of using a sodium silicate glass as a water soluble con-
tainer. However, it has been a guide in determining what paths will
be followed in future research. The molding of glass specimens is
not the optimum method for producing test samples. This author sug-
gests that the mass production of uniform samples by a pulling tech-
nique, similar to the one used by Watanbe and associates (26), be
given serious consideration as a faster and more efficient method.
The work on dissolution kinetics is vitally important in select-
ing the desired composition. This is very time consuming work that
encounters many obstacles. But, after the results are analyzed, that
work will combine with this work on mechanical strength to yield the
selection of a composition that possesses sufficient mechanical
strength and adequate solubility. Also important is the coating that
acts as a vapor barrier to protect the glass surface. This coating
must resist chipping or scratching from handling, and yet be feasibly
applied and reasonable in cost. These components must all function
ss an integral system before success is achieved.
41
-------
This work has justified the continuation of research on this
project, while indicating that additional work is needed on mechanical
properties and new research should be investigated on viscosity and
other properties that will be of vital importance to the glass manu-
facturing industry.
42
-------
1. Morey, G. W. , The Properties of Glass. Reinhold Publishing
New York, (195*0.
2. Anonymous, "Why the U.S. Is In Danger of Being Engulfed by its
Trash," U.S. News and World Report, Washington D C
(September 8, 1969).' ''
3. President's Science Advisory Committee's Report of the Environ-
mental Pollution Panel, "Restoring the Quality of Our En-
vironment ," Washington, D. C., 141, (November, 1965).
4. Glass Industry, 92(1):40, (1968).
5. Anonymous, "Solid Waste Handling in Metropolitan Areas," U. S.
Department of Health, Education, and Welfare, Washington,
D. C., (February, 1964).
6. Anonymous, "Proceedings--The Surgeon General's Conference on
Solid Waste Management," U.S. Department of Health, Educa-
tion, and Welfare, Washington, D. C., (July, 196?).
7. Anonymous, "Proceedings of the National Conference on Solid Waste
Research," American Public Works Association, (1963).
8. "Research and Training Grants," U.S. Department of Health, Educa-
tion, and Welfare, Washington, D. C., (1968).
9- Vail, J. G., Soluble Silicates: Their Properties and Uses, Vol.
_H_, Reinhold Publishing Corp., New York, (1952).
10. Rawson, H., Inorganic Glass Forming Systems, Academic Press, New
York, (T9&7T
11. Moore, H., and Carey, M. , "Limiting Composition of Binary Glasses
of the Type xR20'SiO? in Relation to Glass Structure,"
Journal of the Society of Glass Technology, 35:43-57, (1950-
12. Imaoka, M., and Yamazaki , T., "Survey of Glass-Forming Regions in
Simple Silicates," Journal of the Ceramic Association of
Japan, 61:13-14, (1963).
13. Scholes, S. R., Modern Glass Practices, Industrial Publications,
Inc., Chi cago, (1951).
14. Samseons, M. 0., Journal of the Society of Glass Technology, 182,
517 (1926).
15. Pranatis, A. L. , "Coaxing Effect Furing the Dynamic Fatigue of
Glass," Journal of the American Ceramic Society, Vol. 52 (6),
340-341 (1968).
16. Holland, A. J. and Turner, W. E. S., Journal of the Society of
Glass Technology, 20, 72 (1936).
U 8 EPA Headquarters Library
43 ' Mail code 3404T
1200 Pennsylvania AvenijeNW
Washington. DC 204oi>
202-566-0556
-------
17. 1969 Book of ASTM Standards, Part 13, Refractories, Glass and^
Other Ceramic Materials, Published by the American Ceramic
Society for Testing and Materials, Philadelphia, Pa., (1969)
18. Anonymous, "Glass Strength Test," Glass Industry, 45(2), 80-81,
(1964).
19. Guyer,
Gl
(1930).
E. M., "Mechanical Properties of Some Rolled and Polished
ass," Journal of the American Ceramic Society, J13:624,
20. Williams, A. E., "The Mechanical Strength of Glazing Glass,"
Journal of the American Ceramic Society, J6:980, (1923).
21. Phillips, C. J., Glass: The Miracle Maker, Pitman Publishing
Corp., New York, (1941),
22. Rudnick, A., Hinter, A. R., and Holden, F. C., "An Analysis of
the Diametral-Compression Test," Materials Research and
Standards, Vol. 3:4, 283-289, (April , 1963) .
23. Gehlhoff, G., and Thomas, M., "An Investigation of the Effect of
Composition on the Strength of Sodium Silicate Glass,"
Zeitschrift Fver. Physik., 7:105, (1926).
2k. Kingery, W. D., introduction to Ceramics, Wiley Publishing Co.,
New York, (I960).~
25. Cooper, M. M., "Protective Oxide Coatings for Glasses by Chemical
Vapor Deposition," Progress Report No. 1 submitted to United
States Public Health Service under Research Grant No.
UI00651, (April, 1961).
26. .Watanabe, M., Caporali, R. V., and Mould, R. E., "The Effect of
Chemical Composition on the Strength and Static Fatigue of
Soda-lirne Glass," Physics and Chemistry of Glasses, Vol. 2,
Mo. 1, (February, 1961).
44
-------
APPENDIX
TABULAR RESULTS OF VARIOUS TESTS
Although the graphical results are displayed during the text of
the thesis, this section was added to clarify any questions concerning
the number of specimens tested yielding usable data, numerical averages,
and standard deviations for each experiment.
NO. SPECIMENS NUMERICAL STANDARD
TEST
Bending Strength
Versus Composi-
tion using
graphi te mold.
Bending Strength
Versus Composi-
tion using steel
mold .
Effect of Re-
anneal ing on
Bending
Strength .
COMPOSITION
1 .0 Na20 • 1 .6 Si02
l.P Na20 • 1.3 Si02
1 .0 Na20 • 1 .2 SiO
1.0 Na20 ' 1.1 Si02
1 .0 NaJD • 1 .0 Si02
1.0 Na20 • 1.6 Si02
1.0 Na20 • 1.3 Si02
1.0 Na20 • 1.2 Si02
1.0 Na20 "1.1 Si02
1 .0 Na20 ' 1 .0 Si02
1.0 Na20 • 1.3 Si02
TESTED
11
18
9
10
7
12
23
19
11
8
23 at 0 hr
6 at 2 hr
5 at 4 hr
6 at 7 hr
4 at 12 hr
5 at 18 hr
4 at 23 hr
AVERAGE
29,320
30,850
22,010
17,870
13,990
6,880
8,750
10,960
12,030
12,790
13,980
11,930
13,2/0
1 13,980
13,650
11,800
10,910
DEVIATION
2,390
1,810
2,520
2,670
1,900
1,470
970
1,620
1,270
1,080
760
1,860
2,010
870
1,170
1,170
1,840
45
-------
NO. SPECIMENS NUMERICAL STANDARD
TEST COMPOSITION TESTED AVERAGE DEVIATION
o- 1 .0 Na00 ' 1.3 Si00 23 at
2 2
ure ^ at
chere 6 at 1
0 hr
1 hr
3/4 hr
8,750
5,510
6,810
970
690
560
longed Exposure
ic the Atmosphere
5 at 2 1/2 hr 9,180
5 at 3 1/2 hr 11,370 860
k at k 1/2 hr 12,680 1,370
3 at 5 1/2 hr 10,500 2,3^0
5 at 7 hr 12,650 1,750
46
-------
PART III
THE RATE OF DISSOLUTION OF SODIUM SILICATE GLASSES IN AQUEOUS SOLUTION
-------
ABSTRACT
The rate of reaction of soluble silicate glasses with aqueous
solutions was studied. The work reported is part of an investigation
into the feasibility of processing glass containers after use so that
one large source of solid waste can be relieved and possibly elimin-
ated.
The apparati used in studying the rate of reaction of the glasses
with water is described in detail. Two basic methods of rate deter-
mination are discussed. These two methods are measurement of the
sodium ion concentration in solution, and continuous measurement of
the total bulk weight loss. These two parameters are plotted versus
time and related to the rate of reaction of the glasses with the water
The glasses studied were simple sodium silicate glasses in the
range 1.0 Na20. 1.0 Si02 to 1.0 Na20 . 21. SiO-. The glass batches
were melted at 2050°F and annealed at 800°F. Surface characteristics
of samples were difficult to control, but were generally uniform.
These samples were dissolved, and the rates of reactions with water
under various experimental conditions were plotted. Particular
emphasis was placed on the variables affecting the rates of reaction.
The variables observed were the effects of time, temperature, volume
of solvent, and turbulence.
The rates of dissolution were found to increase with temperature
and turbulence. The effect of the volume of solvent to which the
glass is exposed was less than expected.
-------
It was found that glasses of a rapid dissolving sodium silicate
composition dissolved as a simple solution process in the same man-
ner as a sugar crystal. In more durable glasses, there is the forma-
tion of a boundary layer through which sodium ions and water mole-
cules must diffuse for solution to occur.
ill
-------
PART III
LIST OF FIGURES
Figure Page
1. Photograph of the mold used to form the glass disks 18
2. Photograph of the soluble glass before and after
KOLDMOUNT has been applied 20
3. Photograph of the experimental apparatus used to
measure water attack under static conditions 22
k. Photograph of the Beckman Sodium Ion Analyzer 23
5. Photograph of the experimental apparatus used to
measure water attack under turbulent conditions 26
6. Plot of sodium ion concentration versus time at 72 C
under static conditions 30
7. Photograph of the boundary layer forming on the
* surface of more durable glasses 32
8. Photographs of the breakdown of the boundary layer
of more durable glasses 33'
9. Plot of sodium ion concentration versus time at 90 C
under static conditions 35
10. Plot of sodium ion concentration versus time at 65 C
under turbulent conditions
37
11. Plot of sodium ion concentration versus time at 70 C
under turbulent conditions ................ 38
12. Plot of sodium ion concentration versus time at 75 C
under turbulent conditions ................ 39
13. Plot of bulk weight loss versus time at 2^°C in 300
mill iliters of water ................... 42
14. Plot of bulk weight loss versus time at 72°C in 300
mi 11 iliters of water
15. Plot of bulk weight loss versus time at 24°C in ]k liters
of water
16. Plot of bulk weight loss versus time at 72°C in 300
mi 11 iliters of water for a more durable 68% silica
glasses
vi
-------
PART III
LIST OF TABLES
Table Page
I. Analysis of the Silica Raw Material Used 1&
II. Analysis of the Sodium Carbonate Raw Material
Used 17
-------
PART III
TABLE OF CONTENTS
Chapter Page
I. INTRODUCTION ........................ 1
II. LITERATURE SURVEY
III. EXPERIMENTAL PROCEDURES .................. 15
IV. PRESENTATION AND DISCUSSION OF RESULTS ........... 28
V. CONCLUSIONS ........................ 51
VI. RECOMMENDATIONS FOR FUTURE WORK .............. 53
LITERATURE CITED ......................... 55
APPENDICES ............................ 57
APPENDIX I. Solutions ................... 57
APPENDIX II. The Solution Process ............. 55
IV
-------
FI9ure Page
17. Photograph showing the similarity of concentration
gradients flowing from the surface of sugar and
rapidly soluble glass
vn
-------
CHAPTER I
INTRODUCTION
In a society which has its share of problems such as air and
water pollution, conservation of natural resources, and deterioration
of the cities, there now is another problem which must be met and
solved if Americans are to go on enjoying their present standard of
living. The problem is called "solid waste management."
Solid waste management is not a new problem, but it is odd that
many people do not realize that it even is a problem today. If a
person happens to live near a city dump, he may be quite aware of the
problem, for it strikes home on two sides. He finds that his property
value is not what it might otherwise be and also the environment may
t
be poor due to air pollution or the unsightly dump.
However, most people go about their daily lives never giving
thought to what the "garbage man" does with their solid waste. All
they ask is that it be carted off regularly. In general, the public
has been apathetic until their trash cans stay full for an extra day.
The amount of solid waste produced in a country is closely cor-
related to the standard of living of the people. All prognosticators
are pointing to a boom in the 1970's for the United States economy.
As bad as our solid waste problems are today, everv indication is that
they are going to become much greater, unless some positive steps are
taken immediately. Government has already recognized the problems and
started to move to aid in solving them in numerous ways. It still
-------
remains for the public and industry to become as concerned as they
should be about the magnitude of the problem. Municipalities and
industry need a public mandate to get together and try to alleviate
the problem.
The importance of convenience to the consumer has now become
firmly established as a guiding principle in the design of a packag-
ing material. With this in mind, beverage container manufacturers
are going more and more to the disposable bottle. The Midwest Research
Institute has projected that between 1966 and 1976 the number of
beverage containers in use in the United States will double to an es-
timated 62.9 billion units (1). In this period the use of cans and non-
returnable bottles will also double. The problem is: what do we do
with these wastes?!
It should be realized that it is going to be necessary to deal
with this problem on a long term basis. At the local level, sanitary
landfills are the primary method of disposal. However, this is only
a temporary means of disposal which holds a remarkable similarity to
the housewife sweeping the dust under the carpet. It may be put out
of sight temporarily, but it is still there. This all points to one
of the basic problems concerning sanitary landfills. This problem is
the settling rate which occurs when raw refuse is used. "Compaction
of the various packaging and other materials generally is not complete,
creating undesirable voids which trap liquids and gases. Along with
decaying organic matter, this commonly results in noticeable settling
of the landfill for perhaps 15 to 25 years even in a properly operated
landfill" (2).
-------
Perhaps the answer to the problem lies in taking a different ap-
proach. Instead of trying to devise cheap methods of disposal as is
the general practice today, it may prove beneficial to work on means
of attaining long-range results even if they prove more costly than
the short-range operations.
It is this approach that is being taken at Clemson University.
Work is being done to perfect a one-way container, made of water-
soluble glass, which when discarded can be broken and dissolved.
While in use the container must be covered by a protective coating
both inside and out to prevent any degradation. Later this coating
can be broken and the bottle can be dissolved.
This research, supported by a Solid Wastes Program grant, could
aid in the waste disposal problems resulting from the increasing use
of nonreturnable glass containers. It is estimated that 26 billion
of these containers are discarded each year in the United States (2).
The objective of this research was to give insight into the factors
affecting the dissolution of sodium silicate glasses. The variables
which were studied were the effects of composition, temperature, ex-
posure time to solvent, turbulent conditions in the system, and
sodium concentration in the solution. The data collected was used to
determine the mechanisms which control the solution process.
These results give criteria as to what type of glass is more
suitable for implementation on a practical basis in the glass in-
dustry as a water-soluble glass container. It also serves as a guide
for the avenues of approach which would prove most beneficial to
future researchers in this area.
-------
CHAPTER I I
LITERATURE SURVEY
There has been a substantial amount of work done in the area of
the reaction of glass with aqueous solutions, or the chemical dura-
bility of glasses.
Silvermann (3) lists some general factors to consider in the chem-
ical durability of any glass. These factors are:
(1) State of division: The amount of surface exposed by a
body determines how fast it reacts.
(2) Condition of the glass surface: If the glass surface is
continuous, the effect will be different than if the surface has
been chipped, scratched, or abraded.
(3) Time of exposure to the solvent: Exposure time may deter-
mine the extent of external or internal influence on stability.
(4) Temperature: Conditions during firing and after fabrica-
tion affect durability.
(5) Pressure: External pressure during treatment or use and
internal pressure of contents or strain may affect stability.
(6) Concentration of extraneous reactants
(7) Radiation
(8) Mold and Bacteria
(9) Composition
F. R. v Bichowsky (k) has presented six mechanisms or steps which
take place theoretically at some time during the "weathering" of
-------
glass in an aqueous environment. These mechanisms are:
(1) Adsorption of water on the surface
(2) Diffusion of adsorbed water molecules into the body of the
glass
(3) Reaction of the dissolved and adsorbed water with the glass,
forming a surface film of a character depending on the glass composi-
tion
(4) Soaking up of water by this film thereby getting more
reaction
(5) Extraction of soluble salts from this surface film
(6) Solution of the silica skeleton film in the strongly alka-
line solution formed.
It has been shown that the extent to which any glass may be
attacked or partially dissolved by water is greatly dependent upon
its composition. Sodium disilicate has long been melted for the manu-
facture of water glass. This composition is used because it is water
soluble. Sodium metasilicate glasses dissolve even more rapidly.
Whenever metallic oxides are introduced, insoluble silicates are formed
which decrease the rate of solution. This rate is also affected by
other factors such as water impurities, striations in the glass in
which the ratio of alkali to silica varies from the main composition
of the glass, and strain which is set up by cooling through an un-
controlled temperature cycle. Of course such physical things as
amount of exposed surface area and the degree of circulation of the
liquid will affect the rate of solution (5).
-------
Morey (6) and Scholes (7) are of the opinoin that ft fs Incorrect
to speak of the solubility of glass as one would discuss the solubility
of salt. The action according to them is one of diffusion and disin-
tegration and not one of true solution. All that can be measured and
compared is the rate of attack of the solution on the glass, not the
solubility of the glass in that medium. The problem arises because
the glass changes its internal composition. "The mechanism of the
attack of water upon an ordinary glass cannot be accurately set forth
in the present state of the knowledge. But it is highly selective" (7).
When silicate glass is exposed to water, there is the transport
of water constituents into the glass. The reaction which follows is
dependent upon the composition of the glass and the capacity of the
water to act as a sink for metal hydroxides (8). There are two types
of general attack to be considered: attack on non-bridging oxygen
atoms by reagents with an electron deficit and attack on the network
silicon atoms by reagents with electron excess. Acids and neutral
solutions are called electrophi1ic reagents. They attack positions
with excess electrons. This type of attack is typified by a form
of ion exchange between the glass and attacking solution (9). When in
contact with a liquid aqueous phases, glasses which contain alkali
cations will undergo an ion exchange mechanism in which the liquid
water acts as a sink for alkali hydroxide by dissolving them (8).
Alkalines are nucleophilic reagents. They attack positions with
a deficit of electrons. Alkaline attack is usually typified by de-
gradation of the glass network.
-------
Water is a very good solvent of glass because it can attack the
glass both electrophi1ically and nucleophi1ically. Water ionizes to
form H ions which are electrophi lie and OH~ ions which are nucleo-
phi lie. Even a neutral water molecule can be weakly nucleophi He (9).
In an ordinary commercial glass, Scholes (6) says that sodium sili
cate is leached from the glass and undergoes hydrolysis. The solution
which is formed contains sodium hydroxide and colloidal silicic acid.
The sodium hydroxide which is produced on the surface reacts with car-
bon dioxide in the air while the silica separates out. The surface
film then becomes a deposit of crystals of sodium carbonate together
with finely divided silica. If the attack has not proceeded too far,
the film can be washed off bv a dilute acid or even by water, but if
attack is severe, the surface is permanently ruined. The relative
insolubility of silica causes it to exert a protective action which
causes the attack to proceed more slowly as time goes on.
Most literature concerns itself with solving the problem of
making a glass which is less susceptible to attack by aqueous solu-
tions. Therefore, it is necessary to use the information obtainable
in reverse to get a soluble glass. The literature is also of some
assistance in specimen preparation in the laboratory.
It has been shown that the range of glass formation in the
Na 0-SiO system is continuous from Si02 to a limiting composition
close to that of the 'metasiIIcate Na20-Sl02. The limiting composition
depends on the experimental conditions. The Na20 content of the glass
can be increased as the melt becomes smaller and the cooling rate be-
-------
comes greater. Imaoka and Yamazaki found It possible to make glasses
containing up to 57.8 mole per cent Na«0 on a scale of 1-2 grams. It
is essential that in all work the melting and cooling conditions be
reported when talking about the region of glass formation in a system.
The limit to glass forming compositions in the Na20-Si02 system
coincides almost exactly with the composition at which Na20 becomes
the primary crystalline phase or the first phase to crystallize from
the melt. At this composition, sodium begins to crystallize before the
glass crystallization can be stabilized. This change will be reflected
in the kinetics of the process (10).
The compos it ion wi11 have a pronounced effect on the rate of
devitrification. Dietzel and Wickert (11) studied this effect by
making measurements on twenty compositions in the Na-O-SIO^ system
containing 0-42 mole per cent Na20. "The stability decreases markedly
in the range 0 to 0.5 mole per cent Na20. Then in the range above
10 mole per cent Na20, it increases to a maximum at 23 mole per cent
Na20, a composition close to the eutectic between silica and sodium
disilicate, Na^SiJDJ'Oo). The explanation for this type of curve
is that the stability will decrease as more Na-0 is introduced because
the extra non-bridging oxygens will break down the continuity of the
silica network. Low stability at those compositions corresponding to
the compounds Na^iO,. and Na2$iO. occurs because there is a close
similarity between the structure of the melt and that of the correspond-
ing crystal. This condition enables crvstal1ization to occur readily
(10).
8
-------
an-
Bowen (12) acknowledges the problems associated with obtaining
the proper annealing temperature for a glass. It is necessary to
neal the glass to free it from excessive strain. This strain can have
a marked effect on the internal properties of the glass and its chemi-
cal susceptibility to hygroscopic attack. The choice of temperature
is quite important. To be able to choose the proper annealing temper-
ature, it is necessary to know the crystallizing power for various
temperatures of the glass to be annealed, in order to choose a temper-
ature that will not induce devitrification. Both annealing and
devitrification will take place faster at higher temperatures. In the
same manner, lowering the temperature may not help because there is
no definite lower limit of devitrification temperatures and the rate
of annealing may be too slow if the temperature is too low. Composi-
tional changes, of course, can be made which will affect the anneal-
ing rate or decrease the rate of devitrification. Cox (13) points
out that those glasses higher in soda, low in barium, and high in
silica, tended most readily to devitrification. Little devitrification
was noticed in glass high in potash content, if arsenic was not present.
There are many literature sources which relate methods of in-
creasing durabi1ity of a glass. It is known that solubility and the
tendency to weather are increased directly as the lime to soda ratio
in a glass (14). Baillie (15) and Vail (5) both point out that a
glass becomes less durable as the simple silicate content is in-
creased. It has been shown that there is better chemical durability
for soda glasses than for potash glasses made up to the same molecular
-------
formula, However, a glass composed of both alkalies is more durable
than one containing either NaJ) or ICO alone (6). Sen and Tooley (16)
state that the maximum durability of a glass composed of simple 1^0
and Na20 is at a K20/Na20 ratio of 2.6/KO by weight.
It is widely known that the presence of alumina in the glass will
retard the weathering of the glass. The addition of 1.5 to 2.5% alum-
ina to the glass composition will improve the chemical durability
significantly. The greatest benefit will occur when the alumina is
one-eighth the soda content (17).
The reaction product of the proper quantities of sodium carbonate
and silica will be a clear glass of two types. One type is neutral
glass, Na20 to SiO- ratio of 1/3.3. It is pale bluish or greenish in
color. The other type is alkaline glass, ratio of 1/2.1, and its
color is usually yellowish (5).
Niggli (18) studied the reaction of sodium carbonate and silica
at 898°-956 Centigrade and one atmosphere carbon dioxide. All CO-
was expelled and equilibrium could be approached from both directions.
It was also found that the amount of C02 displaced per mole of silica
was greater the smaller the amount of silica in the melt.
Howarth, ej^ aj_ (19), studied the reaction from equimolar to a
composition of 4 silica:! soda. The equimolar mixture took 40 hours
for complete expulsion of the C02 at 700 C. At temperatures above
900°C, the speed of reaction in all the mixtures increased rapidly.
Commercial soluble silicate glasses are for reasons of economy made
at temperatures in the range of 1200 to 1^00 C. Preliminary heating
10
-------
at about 700 C allows most of the carbon dioxide to escape. Niggli
found that it takes a ratio more siliceous than 1 alkaline oxide to
2 Si02 to expel all of the C02 from glasses at 900 to 1000°C (5).
There have been many attempts made at measuring the durability
or weathering or rate of attack of a glass by its environment. Morey
is quite skeptical of any such results which may be obtained by the
numerous methods now is use and probably justifiably so. The re-
quired precise control of experimental conditions is one reason for
such skepticism. The tests used in the past range from visual clas-
sification of "not affected" and "very affected" to some very precise
and elaborate titration measurements recommended by the Committee on
the Chemical Durability of Glass of the American Ceramic Society.
All have some common points of agreement and some differences. All
recognize the importance of controlling surface area; of attaining
homogeneity in the glass; and of controlling external factors such
as temperature, exposure time, and concentration of solution. The
basic differences lie in the method of determining how much deteriora-
tion has taken place. One school of thought maintains that the best
method is the direct method of weighing both before and after exposure,
The other way is to measure the amount of attack by measuring the
quantity of alkali,.yielded to the solvent under definite conditions.
This has been done by the use of flame photometry, by measuring
electrical conductivities, and by using elaborate titration techniques
Morey is quick to bring attention to the fact that any comparison of
the results of one observer with those of another Is rarely possible,
11
-------
and in any case the validity of any such comparison requires proof.
Morey dealt with systems of wide ranges of temperatures and
pressures and decided that true phase equilibria do not exist in
systems involving water and complex silicate minerals. When per-
forming an experiment on the reaction of a glass with aqueous solutions,
several mechanisms may be involved in any experiment. The rate-con-
trolling processes may vary during the experiment as well as with time
and temperature, with the ratio of glass surface area to the volume
of solution, and particularly to the way the pH of the solution is
controlled. The basic objective of an experiment in this area is
to control the time and temperature of the experiment adequately
and exercise control over the compositional change of the attacking
medium (20).
Douglas and Isard (21) suggest that the process of dissolving
a soda-1ime-si1ica glass is diffusion controlled. A model has been
developed in which the rate controlling process is the diffusion of
Na through the glass to the surface in order to react. This work
found in general a linear relation of the quantity of alkali extracted
from the glass to the square root of time in the early stages of the
reaction. In later stages it was found that the quantity of alkali
was linearly dependent on time.
In later work, Rana and Douglas (22) found also that the rate
of reaction varies to a close approximation / t in the early stages,
but later the rate of reaction becomes constant. It was also found
that in binary silicate glasses the ratio of silica to alkali removed
-------
during the reaction is much less than the ratio of the constituents
in the glass. It was shown that the change in dependence of the rate
of reaction from /T~ to linearity takes place in a short time inter-
val. This "time of changeover" varies from glass to glass and from
temperature to temperature. The amount of alkali removed is approxi-
mately constant up to the changeover.
Haller (8) proposes that "water uptake follows initially a para-
bolic law and is accompanied by a volume expansion. The rate of up-
take is affected by the previous hydration history of the samples,
suggesting that an irreversible change of the glass structure is
associated with the diffusion of the water into the glass." The
model proposed by Douglas and El Shamy (20) suggested that a leached
layer is formed on the surface of the glass. Silica is then dis-
solved at the leached layer-water interface while sodium ions dif-
fuse across the leached layer from the glass into the water. The
change in rate from root time dependence to linear reflects that as
time passes diffusion will slow as the leached layer becomes thicker.
The rate of extraction of alkali from the glass is independent
of pH+ when it lies between 1 and 9. If the pH Is greater than 9,
the rate of extraction decreases with increasing pH . The effect on
silica extraction is opposite. The rate of extraction increases with
increasing pH+. It was found that pure water below 100°C does little
compared to alkaline solutions. There is a strong influence of al-
kali level on the rate of silica attack. This work served to con-
firm Morey's deduction that the action of water on glass Is a com-
13
-------
plicated decomposition reaction and not a normal solution process.
This is shown by the occurrence in solution of species chemically
different from the original glass which impl ies some rearrangement
of chemical bond occurs. It was also shown that aklali ions are,
always, extracted in excess of the silica. This leaves an alkali-
deficient leached layer which is not appreciably different in volume
from the original glass. At the boundary between the leached layer
and the solution, there are sites available to cations. It was
surmised that the diffusion process must be a flow of hydrogen ions
from the solution and an equivalent flow of sodium ions from the
glass into the solution. It is believed that a glass of low chemical
durability would have a large leached layer-solution zone and a
smaller leached layer-glass zone (20).
14
-------
CHAPTER III
EXPERIMENTAL PROCEDURE
The glasses studied were simple sodium silicate glasses with
Na20/Si02 ratios in the range 1.0/1.0 to 1.0/2.1. These glasses were
chosen because of their susceptibility to water attack and the ease
with which they could be processed.
The raw materials used were of reagent grade. Their analyses are
shown in Tables 1 and 2. These materials were hand mixed in a porce-
lain mortar pestle. The batches were fired in both alumina and high
temperature porcelain crucibles. The firing temperature was 2050°F
with a soak time of approximately thirty minutes. All firing was
done in a small globar heated furnace.
After soaking for 30 minutes, the glass was formed into small
disks. The mold used to form the disks was made of steel (See
Figure 1). The glass was poured, pressed, allowed to cool slightly
in the mold,, and annealed at 800°F for thirty minutes. The furnace
power was turned off, and the furnace was allowed to cool at approxi-
mately 158°F per hour. After cooling, the glass disks were removed
and placed in a desiccator until such time as the tests were made.
The surface area of the sample exposed to the water in experiments
had to be constant, or some geometric time factor had to be considered
to account for changes in the surface area. It was decided to try to
isolate one surface of a piece of glass and work with that surface
only. In this way several things would be accomplished: the surface
15
-------
TABLE I
SILICA
Silica
Chloride
Sulfate
Calcium
Heavy metals (as lead)
I ron
Magnesium
Potassium
0.0005%
0.0030%
0.0100%
0.0010%
0.0050%
0.0005%
0.0050%
16
-------
TABLE 11
SODIUM CARBONATE
Sodium Carbonate
Insoluble matter
Lost on heating at 285°C
Chlorine
Nitrogen compounds
Phosphate
Sulphate
Ammonium hydroxide precipitate
Arsenic precipitate
Calcium and magnesium precipitate
Heavy metals (as lead)
I ron
Potassium
Silica
99.5000%
0.0100%
0.5000%
0.0010%
0.0010%
0.0005%
0.0020%
0.0100%
0.0001%
0.0100%
0.0005%
0.0005%
0.0050%
0.0050%
17
-------
Figure 1. Photograph of the mold used to form the glass disks
18
-------
area exposed to the water at any time would be known, and the labora-
tory fabrication problem of producing samples with identical total
volume (or total surface area) each time could be circumvented.
A disk shape was selected because it had the advantage of being
a regular shape which was easily fabricated. All that was needed
was a cylindrical glass mold in which a sample of any desired thick-
ness could be formed. Having chosen to work with the disk shape, it
soon became apparent that, if the sides and bottom of the disk were
covered, it would leave only one surface of known area exposed to the
aqueous solution. It, therefore, became standard procedure to cover
the side and bottom of the disk with an insoluble epoxy resin (trade-
name KOLDMOUNT) before each test in order to isolate one surface
area (See Figure 2).
There were three experimental methods used to measure the rate
of reaction of the glass with water. The first two involved the
measurement of the sodium ion concentration in the aqueous solution.
The third method employed the measurement of the total bulk weight
lost by the sample during exposure to water.
To analyze the rate of reaction of the sodium silicate glass with
water, the sodium ion concentration of the water both before and after
exposure to the glass surface was measured. This sodium concentration
was plotted versus time to give some idea of the rate of reaction of
the glass when exposed to water.
In these experiments, the water to which the glasses were ex-
posed contained only about 18 parts per billion sodium ion concentra-
19
-------
Figure 2. Photograph of the soluble glass before and after KOLDMOUNT
has been applled.
-------
tion. Therefore, eighteen ppb was used as the zero point or starting
point for sodium concentration in all readings taken.
A coated glass disk was placed in a flask to which 200 mi 111-
liters of distilled water had been added. This flask was placed in
a controlled temperature bath and left for a certain time interval
(See Figure 3). When the appropriate time interval had elasped, the
sample solution was extracted in approximately the same manner each
time. The solution was analyzed for the sodium ion content present
by the Beckman Sodium Son Analyzer. In this manner it was possible
to study the rate of dissolution at different temperatures and at
different times.
The Beckman Sodium Son Analyzer has the capability to measure
from 1 to 10,000 parts per billion sodium ion concentration. This
analyzer is a process instrument that can make a continuous measure-
ment of the concentration of sodium ions in a solution by determining
the difference in potential developed between two electrodes, a silver/
silver chloride electrode and a sodium ion electrode, immersed in
the solution being analyzed (See Figure 4).
The operating principle of the analyzer is that the sodium ion
measurement is made with a glass electrode-reference electrode pair.
The glass electrode is of a special glass that is very responsive to
changes In sodium ion concentration. This response comes in the form
of a potential which is proportional to the log of the sodium ion
concentration present In the sample. A reference electrode is also
used to complete the circuit and provide a constant reference potential
21
-------
Figure 3-
Photograph of the experimental apparatus used to measure
water attack under static conditions.
22
-------
i
Figure 4. Photograph of the Beckman Sodium Ion Analyzer
23
-------
ihe sodium Ion electrode is also sensitive to hydrogen ions
(pH ). In order to permit effective sodium ion measurement regardless
of the pH of the sample, it is necessary to control the pH of the
sample. In the Sodium Ion Analyzer, a simple system provides for ad-
justment of pH by the addition of anhydrous ammonia. The electrode
is not sensitive to the ammonia ion or other cations, with the excep-
tion of silver, which was not present in the sample.
The system operates in the following manner. The sample passes
through the flow chamber which contains the sodium ion electrode and
the reference electrode. Sample flow is controlled and the pH of the
system is raised by bleeding ammonia into the stream through a capil-
lary tube. As the sample passes through the chamber, the sodium
electrode senses changes in sodium ion concentration. The difference
between the fixed potential of the reference electrode and the vary-
ing potential of the sodium electrode results in a signal which is
sent to a high impedance amplifier. The amplifier gives a direct
meter scale readout. It is a system similar to many standard pH
systems currently in use (23).
The concentration of sodium in solution after a given time was
thus determined for a reasonably static situation. It then was neces-
sary to determine what effect turbulent conditions would have on the
rate of attack of the glass by water.
In this part of the study it was desirable to control the tur-
bulent conditions as much as possible. A system was developed which
could pump the water at a constant rate into the system at a given
24
-------
point. The system in this instance was a water vat of approximately
10 gallons volume. It was found that an aquarium pump could be set
up to work very adequately. A tube from the pump was attached so that
water flowed into the system 1/8 inch above the glass sample, which
had been coated as explained before and mounted on a stainless steel
pedestal. Samples of the solution were withdrawn from the same
position in the system each time. This withdrawal was from a point
1/4 inch above the sample. The water vat was heated by a resistance
element. In addition, the temperature was controlled to ±1°C.
By using this apparatus (See Figure 5), the rate of water attack
upon the soluble glass at given times and temperatures could be ob-
served under turbulent conditions. The sample was exposed to the
water at a constant temperature, under constant turbulent conditions,
for varying times. Sample solutions were extracted and then analyzed
for sodium ion concentration by the Sodium Ion Analyzer, and plots
were made of sodium ion concentration versus time.
The rate of water attack upon the bulk glass composition was
tested. This procedure entailed suspending a glass sample in water
under varying conditions. While being exoosed to water attack, the
weight of the sample suspended in the solution could be monitored
and the overall weight loss of the bulk composition could be deter-
mined at variable times. There was data collected under the follow-
ing conditions: (1) the sample was placed in a small amount of water
(300 ml) at 24°C; (2) the sample was placed in a large amount of water
liters) at 24°C; (3) the sample was placed in a small amount of
25
-------
':«»
Figure 5-
Photograph of the experimental apparatus used to measure
water attack under turbulent conditions.
26
-------
water (300 ml) at a high temperature (72°C); (k) the sample was placed
in turbulent water (14 liters) at 2^°C; and (5) a different glass com-
position was placed in 300 ml of water at 24°C and compared with the
more soluble composition.
The samples were in each case weighed continuously by an Ainsworth
analytical balance that was connected to a strip chart recorder which
plotted the weight loss versus time. This work was done so that the
results of the decomposition of the total bulk could be compared to
the loss of sodium ions from the sample.
Spectrographic analyses were run on solutions which had been ex-
posed to different glass compositions. The glass samples were ex-
posed to water for various times, and sample solutions were withdrawn.
A drop of this solution was then analyzed by use of a Bausch and Lomb
dual grating spectrograph. The resulting photographic plates were
analyzed for the relative intensities and were qualitatively related
to the amount of the element present in the solution. These compari-
sons coupled with the other data enabled an understanding of how the
attack of water upon soluble glass proceeds.
27
-------
CHAPTER IV
PRESENTATION AND DISCUSSION OF RESULTS
The samples used in this experimental research were sodium sili-
cate glasses of varying composition. In the discussion text these
samples will be referred to as being of a rapid dissolving, a moderate-
ly soluble, or a more durable composition. The rapid dissolving
glasses are those having a Na20/Si02 molar ratio of between 1.0/1.0
and 1.0/1.3. The moderately soluble glasses are those which have a
ratio of 1.0/1.3 to 1.0/1.6, and the more durable compositions have
molar ratios greater than 1.0/1.6. Any of these glass compositions
will become more durable, less soluble, if there are impurities present
in the glass.
The first experimental measurements were made on a moderately
soluble glass disk exposed to distilled water at a temperature of
72 C. The only current flow or turbulence in the system was that set
up by a thermal gradient in the system. This gradient was minimized
by the immersion of the flask in the water bath to give more isotherm-
al conditions.
In these measurements, the disks were oriented horizontally on
the bottom of the flask. Although strictly speaking it cannot be said
that the surface remained exactly constant, the change in surface
area was so slight as to not warrant consideration. This statement
can be made because it was observed that the glass dissolved rather
evenly across the exposed face. This characteristic helps to minimize
28
-------
any additional concentration caused by an increase in surface area
due to a curved surface being exposed rather than a planar surface.
It was noted that after a certain period of exposure to the water a
surface film or coating was being deposited on the glass disk. This
surface deposit remained for the entire duration of the experiment.
If the glass was allowed to completely dissolve, the film would re-
main in the bottom of the epoxy shell. After dehydration, the resi-
due became a brittle, white, flaky substance. Other workers have
indicated in the literature that the substance is probably a complex
sodium carbonate-fine grain silica residue.
Samples of the solution in the flask were withdrawn at various
times and analyzed for sodium ion concentration. These concentrations
were then plotted versus time to yield the curve seen in Figure 6.
The ordinate in this figure is sodium ion concentration in parts per
million as measured by the Beckman Sodium Ion Analyzer. The abscissa
is time of exposure to water in hours. The time at which the sample
is placed into the water in the flask is considered the zero point.
Initial sodium ion concentration is taken as 18 ppb, the initial con-
centration of sodium in the distilled water. If the concentration
of sodium is considered C, and the time is considered T, then the slope
of the curve dC/dT will yield the rate at which sodium ions are coming
into solution from the glass. This may be the rate of solubility of
the glass, or it may be the rate of water attack on the glass surface.
Its exact correlation with the solubility of the glass is not known.
All that can be said is that the rate observed is the rate at which
29
-------
soor
400 -
3OO -
CO
o
z
UJ
o
z
o
o
o
z
200 -
100 -
34
TIME, hours
FIGURE 6. PLOT OF SODIUM ION CONCENTRATION VERSUS TIME AT 72°C UNDER
STATIC CONDITIONS.
-------
sodium ions are going into solution under these specific conditions.
It is quite possible that slight changes in the experimental condi-
tions will yield changes in this rate.
In Figure 6, it can be seen that from the time the sample was
placed into the water until after about 2 hours of exposure, the rate
of increase of the sodium ion concentration was constant. The initial
increase was caused by the removal of sodium ions from the glass by
the aqueous solution. The removal of these sodium ions from the sur-
face causes the glass to develop a glass-solution interface which is
depleted of alkali ions and is rich in silica. The macroscopic ef-
fect of this leaching is the formation of a boundary layer through
which sodium ions and water molecules must diffuse for the dissolu-
tion to proceed (See Figure 7). As the boundary layer is formed,
the rate slows to an almost negligible increase in sodium ion con-
centration because diffusion is occurring.
After three hours exposure time, the rate begins to increase
again. This rapid increase in rate can be attributed to two factors.
One factor is that as more sodium enters the solution the more basic
the solution becomes. This basic solution is much more corrosive to
the actual silica framework than water alone. As the framework is
corroded at a faster rate, more sodium is released into solution at
a faster rate.
The increase in rate at this point is also caused by the break-
down of the boundary layer (See Figure 8). After the boundary layer
has been exposed to the water for a time, it begins to deteriorate
31
-------
(A)
(B)
(C)
(D)
Figure 7•
Photographs of the boundary layer forming on the surface
of a more durable glass (A) originally, (B) after 1 hour,
(C) after 2 hours, (D) after k hours.
32
-------
(E)
(F)
(G)
(H)
Figure 8.
Photographs showing breakdown of the boundary layer on a
more durable glass (E) after 5 hours, (F) after 6 hours,
(G) after 7 hours, (H) after 8 hours.
33
-------
and allow water to diffuse through it at a faster rate. This deterio-
ration then allows the glass matrix to be attacked at a faster rate
and yields the higher sodium concentration in solution.
This theory can be supported by spectrographic analysis for the
silica content of a solution. For times coinciding with the initial
leaching of sodium from the glass and for times coinciding with the
formation of a boundary layer, it was observed that there is very
little increase in the amount of silica in solution. However, for
times coinciding with the breakup of the boundary layer and alkaline
attack of the silica framework, there is a very substantial increase
in the amount of silica in solution. This indicates that there is an
extended period of exposure during which very little silica goes into
solution followed by a period during which silica is rapidly dissolved.
After about 5 hours, the rate gradually slows as there is degrada-
tion of the surface of the sample. At this time the sample has been
corroded to the extent that it no longer presents a continuous surface
to the solution. Water is able to go completely through the sample
and attack the sample on more surfaces. Also the sample has been
almost completely eroded away. Therefore, at this point the test
ce" as to be valid.
Figure 9 shows the effect obtained by increasing the temperature.
The immediately noticeable effect is that the reaction takes place
much faster at higher temperatures. For example, after one hour of
exposure at 72 C there was approximately 50 ppm sodium concentration
in solution. After one hour at 90 C there was an average 115 Ppm
34
-------
600 r
GO
in
O
O
500 -
400 -
300 -
200 -
100 -
O
I 23456
TIME, minutes
FIGURE 9. PLOT OF SODIUM ION CONCENTRATION VERSUS TIME AT
9O°C UNDER STATIC CONDITIONS.
-------
sodium concentration In solution. This represents a very marked ef-
fect by temperature increase of just 18°C. This increase in rate
is reflected in the greater slope (dC/dT) of the curve. There is no
change in the rate of sodium uptake over the duration of the experi-
ment. The sodium concentration increases at a linear rate. This in-
dicates that at higher temperatures there is not an induction period
involved during which diffusion through a boundary layer slows the
reaction. This effect could be due to several reasons. It could be
that the boundary layer in fact never forms at this temperature. The
author considers this very unlikely from the standpoint of the sur-
faces which have been studied and photographed (See Figure 7). It is
possible that at this temperature a boundary layer does form, but it
is broken down by water turbulence so that it has no effect on the
reaction rate. Finally, at this temperature viscosity of the water
may have been reduced sufficiently for it to diffuse through any
boundary layer that is present at such a fast rate that the boundary
layer diffusion is no longer a rate controlling step. The pronounced
effect of temperature is demonstrated by the results shown in Figures
6 and 9- With the exception of temperature, these two experiments
were performed under the same static conditions.
The effect of turbulence on the rate of reaction of the glass
with water was observed next (See Figures 10, 11, and 12). The most
evident result of the turbulence is a dramatic increase in the rate of
sodium uptake. From Figure 6, it can be seen that it takes one hour
under static conditions for 50 ppm sodium ion concentration to show
36
-------
CO
--J
Hi
O
200 -
160 -
120 -
8
12 16
TIME, minutes
20
24
28
FIGURE IO. PLOT OF SODIUM ION CONCENTRATION VERSUS TIME AT 65°C UNDER
TURBULENT CONDITIONS.
-------
200 r
CO
CD
or
z
UJ
o
o
o
160 -
120 -
1 1
3
1 ,
4
1 ,
8
1
12
1 i
16
1
20
1
24
i 1
28
TIME, minutes
FIGURE II. PLOT OF SODIUM ION CONCENTRATION VERSUS TIME AT 70°C
UNDER TURBULENT CONDITIONS.
-------
2OOr
00
cc
I-
z
UJ
O
z
o
o
I60h
I20h
8 12
TIME, minutes
16
20
24
FIGURE 12. PLOT OF SODIUM ION CONCENTRATION VERSUS TIME AT 75°C
UNDER TURBULENT CONDITIONS.
-------
up in solution. In Figure 10, it is found that under turbulent
conditions there is 50 ppm sodium ion concentration in solution after
only eight minutes. This is an Increase of 7.5 times as fast under
turbulent conditions. At this point, once again the buildup of a
boundary layer is sufficient to have a leveling effect on the curve.
The rate of increase of sodium ion concentration becomes quite small
until twenty minutes of exposure time has passed. At this time, there
is once again a very rapid increase in the slope of the curve. This
increase is probably due to the breakdown of the barrier layer. This
deterioration occurs rapidly because of the turbulent conditions around
the surface. The constant flow of water against the surface causes
enough abrasion to deteriorate the boundary layer and expose the
glass to attack as exhibited by the very steep slope after 20 minutes
exposure in Figure 10.
Figures 10 and 11 show again in a clear manner the influence on
the attack process of even very minor temperature changes. If the
point at which the slope begins to increase again after the induction
period is used as a reference, it can be seen that an increase in
temperature serves to shift the reaction to a shorter time along the
abscissa. For example, in Figure 10 (run at 65 C) that change in slope
occurs at 20 minutes exposure time. In Figure 11 (run at 70 C) that
slope change occurs'after only 16 minutes exposure time. In Figure 12,
the time has been reduced to only 14 minutes of exposure to water at-
tack. So it can be seen that the effect of turbulence and temperature
in the system is very pronounced. These results indicate that tur-
40
-------
bulent conditions promote the dissolution of the glass. All of the
results discussed thus far were obtained by analyzing the solution in
contact with moderately soluble glass after certain exposure times.
This analysis was made with the Beckman Sodium Ion Analyzer.
The remainder of the results were concerned with changes in the
total bulk composition of the glass. In particular, the weight loss
of the glass was monitored by an Ainsworth analytical balance which
weighed the sample continuously. This weight loss was plotted versus
time of exposure to water. Typical results of this analytical tech-
nique can be seen in Figure 13.
In Figure 13, a sample of rapidly dissolving glass was placed in
300 ml of distilled water at 2k C. The sample was weighed continuous-
ly. The resulting curve is linear until 120 minutes of exposure has
occurred. At this time, the sample had decomposed to the extent
that there was no longer a continuous surface presented to the cor-
roding medium. By this it is meant that the surface crumbled and
depleted to the extent that the test could not be considered valid
for longer times.
The effect of turbulence was attempted to be shown, but the ex-
perimental apparatus could not be operated continuously due to the
swaying effect caused by the water flow. Two weight measurements
were made after 20 and 50 minutes of exposure under turbulent condi-
tions. These measurements are superimposed on Figure 13- There is
not enough data to make any conclusions except that, as in the pre-
vious experiments, turbulence seems to have increased the rate of
41
-------
ro
to
QC
O
CO
CO
O
-J
O
UJ
8.0 -
0°
O
O w
4.0
8.0
120 160
TIME, minutes
2OO
240
FIGURE 13. PLOT OF BULK WEIGHT LOSS VERSUS TIME AT 24°C IN 300
MILLILITERS OF WATER.
-------
attack on the glass.
Next, the effect of an increase in temperature was observed
using the weight loss measurement method. The water was preheated
to 72°C and then brought into contact with the glass sample. The
effect was as expected. The glass lost weight at a much faster rate
at 72 C than at 2^°C (See Figure H). An example of this is that at
2^°C the glass sample had lost 1.7 grams after 50 minutes of exposure.
However, if exposed to 72°C water for only 5 minutes, the same glass
lost 1.8 grams. The rate of bulk weight loss at this temperature
once again displays linear characteristics.
The effect of the volume of solvent (or corroding medium) to
which the sample is exposed was also investigated. It was of interest
to observe if the amount of water to which the glass was exposed had
a great effect on the rate of attack. Of particular interest was
whether or not the process was self-limiting in small volumes of
water. If so, it would indicate that the water was becoming saturated
with the solute and causing the reaction to change, probably to slow
or stop completely.
It was found that, if the sample were placed in \k liters of
water instead of in 300. ml of water, there was very little change in
the rate of weight loss observed. If the curves in Figures 15 and 13
are compared, it is found that they could almost be superimposed over
each other without any change. The only effect that was observed was
that with the larger volume of water (14 liters) the loss took place
at only a minutely faster rate; the increase was to be expected. The
43
-------
cr
o
CO
(0
o
o
UJ
5t
12 16
TIME, minutes
24
28
FIGURE 14. PLOT OF BULK WEIGHT LOSS VERSUS TIME AT 72«C IN 300 MILLI LITERS
OF WATER.
-------
10.0
en
CO
<0
CO
o
g
UJ
O
O
O
O
r* O
r • '
> 40
. i i
80
i I i
120
1
160
i I
200
i 1
240
TIME, minutes
FIGURE 15. PLOT OF BULK WEIGHT LOSS VERSUS TIME AT 24«C IN
14 LITERS OF WATER.
-------
slightness of this increase is surprising. The lack of effect of
solute volume is shown in the following example: After 40 minutes,
1.7 grams weight loss is observed in 300 ml of 24°C water. If the
sample is exposed to 14 liters of 2k C water, the weight loss in-
creases only 0.2 gram to 1.9 grams weight loss. After 140 minutes
in 300 ml of water, the weight loss is 6.4 grams; in 14 liters, it
is 6.5 grams.
This very small difference in weight loss between the two
volumes leads to the conclusion that the amount of solvent present
has very little effect on the rate of attack on the glass sample.
At least this appears to be true for the amounts of water used in
these experiments. It is possible that at even smaller volumes
of water there is a self-limiting effect. However, these experiments
indicate that, as long as there is enough solvent present to dissolve
the glass, it does not matter how much excess solvent is present in
the system.
The weight loss measurement was also plotted versus time for a
less soluble glass. This glass was 68 per cent silica and 32 per
cent soda. The results of this measurement showed an initial loss,
followed by an extended period of very little or no weight loss and
a final rapid weight loss (See Figure 16). The various sections of
the curve can be explained by applying the boundary layer diffusion
theory outlined previously. In fact, this curve is remarkably similar
to those obtained by sodium ion analysis. It appears to follow the
same sequence as the rate of sodium ion increase shown in Figure 6.
46
-------
I.O
X
<
K
o
o
tal
O.8
-CX
C i
* IO
1
20
I
30
1
4O
i
SO
1
6O
i
7O
1
SO
TIME, hours
FIGURE 16. PLOT OF BULK WEIGHT LOSS VERSUS TIME AT 72»C IN 3OO MILLILITERS
OF WATER FOR A MORE DURABLE 66H SILICA OLASS.
-------
There is an initial weight loss, a leveling-off effect (a boundary
layer induction period), and a final sharp increase in the rate of
weight loss.
If the results of the weight loss measurements of rapidly dis-
solving glass (Figure 1^) and more durable glass (Figure 16) are com-
pared, it is observed that there is a difference in the way that the
two types of glass dissolve. The rapid dissolving glasses lose weight
at a constant linear rate. More durable glasses go through a change
in surface characteristics which causes it to dissolve in steps.
The constant linear rate of weight loss in rapidly dissolving
glasses is indicative of a true solution process (See Appendix I).
By simple solution it is being implied that there is no breakdown
of the glass structure into its individual components as far as the
solvent is concerned. Both soda and silica complexes are being at-
tacked as one single phase much as a single sugar crystal is dis-
solved in water (See Figure 17 and Appendix II). This theory is
verified by spectrographic analysis which shows that in rapidly dis-
solving glasses the silica concentration in solution increases at a
relatively constant rate. In more durable glasses the silica con-
centration remains low until such time as the silica framework is
attacked by solution. This attack is exhibited by a dramatic increase
in silica concentration in solution.
The conclusion reached from this comparison is that sodium
silicate glasses which dissolve rapidly without the formation of a
boundary layer dissolve by a mechanism quite similar to true solution.
48
-------
Figure 17•
soluble glass (B) .
49
-------
In more durable glasses there is the formation of a boundary layer
which causes the dissolution of the glass to be more complex. This
process can be described as the preferential leaching of sodium ions,
followed by the formation of a boundary layer, and ultimately al-
kaline attack on the remaining silica network.
50
-------
CHAPTER V
CONCLUSIONS
1. The attack of water on a rapidly soluble silicate glass released
sodium ions into the solution at a linear rate. Weight loss also
followed a linear relationship versus time. This linearity re-
presents a simple solution process and not a complex decomposi-
tion process as observed by experimenters working with more com-
plex glasses.
2. In more durable but yet soluble glasses, there is the formation
of a boundary layer after a certain time of exposure to water.
It then becomes necessary for the water molecules to diffuse
through the boundary layer before the solution process can con-
tinue.
3. The effect of temperature is that the rate of reaction of the
glass with the water is greatly increased as the temperature
is increased. This result is to be expected if the dissolution
of the glass is an endothermic reaction. It was also found that
temperature can affect the boundary layer and water viscosity to
the extent that water can diffuse through to the glass surface
at a fast enough rate to maintain a constant rate of reaction with
the glass.
*». The effect of turbulence was demonstrated to be a substantial
increase in the rate of attack of the water on the glass surface
under turbulent conditions. The explanation of this is that
turbulence produces more abrasion on the surface. Subsequently,
-------
the water molecules are able to cluster around the sodium sili-
cate molecules more easily and carry them into the solution.
5. Volume dependency was shown to not be as great as was originally
thought. The solution process did not seem to be self-limiting
in small volumes of water. In fact, the amount of solvent pre-
sent appears to have very little effect at all on the rate of
attack on the glass. Larger volumes of water did show a slight
increase in the amount of glass dissolved, but the increase is
not very significant.
52
-------
CHAPTER VI
RECOMMENDATIONS FOR FUTURE WORK
The principles observed in this investigation should ultimately
aid in the utilization of soluble sodium silicate glass in the glass
industry as a partial answer to the solid waste dilemma.
The significance of this work is that it demonstrated the feasi-
bility of producing a rapidly dissolving glass which could be easily
decomposed without undue pollution of water resources.
Having established this fundamental ability to produce such a
glass, correlated study can be accomplished to develop more suitable
glasses and study the solution characteristics of those glasses.
Corresponding studies are being conducted in the area of coating
materials and procedures for various soluble glass compositions.
The major deficiencies of the rapid dissolving glass appeared
in two areas, fabrication and the ability to be coated. The solu-
bility of the glass is such that it will easily meet any requirements
as to speed of dissolution. In fabrication, however, the glass
proved to be quite unstable requiring close attention to prevent
devitrification. The forming of nucleation sites and subsequent
devitrification was almost immediately evident if fabrication proce-
dures were not closely controlled. Problems in the area of-coatings
developed because of the extreme hygroscopic!ty of the glass surface.
The solution of these problems will likely involve the develop-
ment of more overall suitable glass compositions, equipment modifica-
tions, better coating procedures and materials, and probably a com-
53
-------
bination of all of these.
Better quality control over the firing conditions and forming
conditions of the glass is a definite requirement at this stage. A
furnace is required which will not allow contamination of the batch
and has better control over the heating and soaking periods. With
more sophisticated forming equipment, sample quality control could
be upgraded to eliminate sample variations in uniformity, annealing
stresses, and surface condition. The elimination of these problems
would greatly enhance the potential for success of the project.
The author is of the opinion that, for the optimum development
of the project, new glass compositions which would dissolve more
slowly should be investigated. A small pilot plant setup would be
extremely helpful. This would be the next logical step if progress
is to take place in the fabrication of a soluble glass container.
54
-------
LITERATURE CITED
1. Abrahams, J. H, "Packaging Industry Looks at Waste Utilization."
Glass Container Manufacturers Institute, Inc., Washington,
D. C.
2. Vaughan, R. D. "Packaging and Solid Waste Management." U. S.
Department of Health, Education, and Welfare, Consumer
Protection and Environmental Health Service.
3. Silvermann, A. "Chemical Durability of Glass: An Outline."
Bui 1. Am. Ceram. Soc. 30:399. 1951.
4. Bichowsky, F. R. v. "Note on the Mechanics of the 'Weathering1
of Glass." J^. Am. Ceram. Soc. 3:309- 1920.
5. Vai1, J. G. Soluble Si 1icates: Thei r Properties and Uses,
Volume I. Reinhold Publishing Corp., New York. 1952.
6. Morey, G. W. The Properties of Glass. Reinhold Publishing
Corp., New York. 1954-
7. Scholes, S. R. Modern Glass Practice. Industrial Publications,
Chicago. 1952.
8. Haller, W. "Kinetics of the Transport of Water Through Sili-
cate Glasses at Ambient Temperatures." Physics and
Chemistry of Glasses. 1(2):46-51. I960.
9. Budd, S. M. "The Mechanisms of Chemical Reaction Between Silica
Glass and Attacking Agents." Physics and Chemistry of
Glasses. 2(4):111-114. 1961.
10. Rawson, H. Inorganic Glass-Forming Systems. Academic Press,
New York. 196?.
11. Dietzel, A. and H. Wickert. Glastech. Ber. 29:1-4. 1956.
12. Bowen, N. L. "Devitrification of Glass." J_. Amer. Ceram. Soc.
2:262. 1919-
13. Cox, S. F. "Devitrification of Glass." J_. Amer. Ceram. Soc.
2:576. 1919-
14. Ford, K. L. "The Weathering of Glass Containers." £. Amer.
Ceram. Soc. 5:837. 1922.
15. Baillie, W. L. "The Effect of Additives on Chemical Durability
of Glasses." J. Soc. Glass Tech. 6:68-101. 1922.
55
-------
16. Sen, S. and F. V. Tooley. "Effect of Na-O/K-0 Ratio on Chemical
Durability of Alkal i-Lime-SM ica Glasses." £. Amer. Ceram.
Soc. 38:175. 1955.
17. Lyle, A. K. , W. Horak, and D. E. Sharp. "The Effect of Alumina
Upon the Chemical Durability of Sand-Soda-Lime Glasses."
J_. Amer. Ceram. Soc. 19: 1^*2. 1936.
18. Niggli, P. "The Reaction of Sodium Carbonate and Silica."
J_. Amer. Ceram. Soc. 35:1693. 1913.
19- Howarth, J. T. , W. Maskill, and W. E. S. Turner. "The Rate of
Reaction Between Silica and Sodium Carbonate at Different
Temperatures and the Process of Vitrification." J_. Soc.
Glass Tech. 17:25. 1933.
20. Douglas, R. W. and T. M. M. El-Shamy. "Reactions of Glasses
with Aqueous Solutions." J_. Amer. Ceram. Soc. 50:1-7.
1967.
21. Douglas, R. W. and J. 0. Isard. "Action of Water and of Sulfur
Dioxide on Glass Surfaces." J. Soc. Glass Tech. 33(15*0:
289-335.
22. Rana, M. A. and R. W. Douglas. "The Reaction Between Glass
and Water." Physics and Chemistry of Glasses, 2(6):
179-205. 1961.
23. Beckman Instruction Manual: Trace Sodium Ion Analyzer. Beck-
man Instruments, Inc. Fullerton, California.
2k. Maron, S. H. and C. F. Prutton. Principles of Physical Chemistry
The MacMillan Company. New York. 1958.
25. Sienko, M. J. and R. A. Plane. Chemistry. McGraw-Hill Book
Co., Inc. New York. 1957.
56
-------
APPENDIX I
SOLUTIONS
When several nonreacting substances are mixed, three possible
types of mixtures may be obtained: (a) a coarse mixture, (b) a col-
loidal dispersion, or (c) true solution. In a coarse mixture, the
individual particles are readily discernible and may be separated
from each other by mechanical means. In a colloidal dispersion, the
particles are much finer and the heterogeneity is not as apparent.
It is, however, not a homogeneous dispersion. The true solution
possesses the characteristic that the constituents cannot be
separated by mechanical means. A true solution is a physically
homogeneous mixture of two or more substances.
The extent to which a substance will dissolve in another varies
greatly with different substances and depends on the nature of the
solute and solvent, the temperature, and the pressure. The effect
of pressure is usually small unless gases are being used. However,
temperature usually has a marked effect. The direction in which the
solubility of a substance in a solvent changes with temperature
depends on the heat of solution. If the substance dissolves exo-
thermal ly, the solubility at higher temperatures will decrease.
If, however, the dissolution is an endothermic reaction, the solu-
bility will increase as the temperature is raised.
A general rule is that likes dissolve likes. Which means that
compounds similar in chemical character are more readily soluble in
U S BPA Headquarters Library
57 Mai! code 3404T
1200 Pennsylvania Avenue NW
Washington, DC 20460
202-566-0556
-------
each other than those of entirely different chemical character.
The extent to which solids dissolve in liquids varies greatly
with the nature of the solid and liquid, the temperature, and to
some extent the pressure on the system.
The influence of temperature on the solubility of a solute
in some solvent is usually quite pronounced. Because most solutes
tend to absorb heat on solution, they are more soluble at higher
temperatures. The opposite is true of exothermic reactions, but
all behavior is not regular in this respect.
Whenever the solubility of any substance is plotted versus
temperature, the curve is continuous until there is a change in the
nature of the solid phase. As soon as the solid changes, a break
will appear in the solubility curve and a new solubility curve will
originate from this point. These breaks are characteristic of the
substance involved.
The particular change in the nature of the solid phase involved
could be a number of things. It could be a transformation from one
crystalline form to another; it could be a change from a hydrate
to an anhydrous salt; or it could be a change from one hydrate to
another (24).
58
-------
APPENDIX I I
THE SOLUTION PROCESS
When sugar is placed in a beaker of water, the sugar dissolves.
In this process of solution, the sugar lattice is ripped apart, and
the individual sugar molecules are distributed throughout the body
of the solution. The entire process can be theorized to occur in
the following way: First, a sugar molecule must be pulled from the
surrounding molecules. Since there is molecular attraction between
molecules in the crystal lattice, work must be done to overcome this
intermolecular attraction (solute-solute attraction). Second, water
molecules have to be displaced to accommodate the sugar molecule.
This displacement also requires the expenditure of energy, since
there is molecular attraction between water molecules (solvent-
solvent attraction). If both of these processes require energy, they
cannot be the driving force for the solution to occur. There has to
be a third step which provides the energy that is needed. This step
is called solvation. Solvation arises from the attraction between
the sugar molecules and the water molecules. This sugar-water
attraction provides almost enough energy to overcome the sugar-sugar
attraction and the water-water attraction. The last bit of energy
necessary is supplied by a reduction in the average kinetic energy
of all the molecules in the system.
The actual state of the particles once they get into the solu-
tion is one of the really challenging problems facing chemists
59
-------
today. Progress is very slow. Present theory assumes that solution
occurs by some such mechanism as the following: In their random
thermal motion, solvent molecules continually bounce against the
surface of the solute. Solution occurs when solvent molecules carry
off solute molecules by clustering about them. A sugar molecule
with its associated cluster of water molecules is called a solvated
molecule and is believed to move off into solution as a unit. Sol-
vent molecules are continually exchanging places with other solvent
molecules in the surrounding area of solvent even though a strong
attraction for the solute particle is present. The character of
the solvated particles is constantly changing and this tends to
make any study of the system difficult.
* U. S. GOVERNMENT PRINTING OFFICE : 1971 O - 446-907
60
------- |