PREDICTING THE EFFECTIVENESS OF
CHEMICAL-PROTECTIVE CLOTHING:MODEL
AND TEST METHOD DEVELOPMENT
-------
PREDICTING THE EFFECTIVENESS OF CHEMICAL-PROTECTIVE CLOTHING
MODEL AND TEST METHOD DEVELOPMENT
by
Abhoyjit S. Shown
Elizabeth F. Philpot
Donald P. Segers
Gary D. Sides
Ralph B. Spafford
Southern Research Institute
Birmingham, Alabama 35255
Contract No. 68-03-3113
Project Officer
Michael D. Royer
Releases Control Branch
Hazardous Waste Engineering Research Laboratory
Edison, New Jersey 08837
This study was conducted
under subcontract with
JRB Associates
McLean, Virginia 22102
WATER ENGINEERING RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
DISCLAIMER
The information in this document has been funded wholly or in part hy
the United States Environmental Protection Agency under Contract
No. 6R-D3-3113 to JRB Associates/SAIC. It has been subject to the
Agency's peer and administrative review, and it has been approved for
publication as an EPA document. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
ii
-------
FOREWORD
The U.S. Environmental Protection Agency is charged by Congress with
protecting the Nation's land, air, and water systems. Under a mandate of
national environmental laws, the agency strives to formulate and imple-
ment actions leading to a compatible balance between human activities and
the ability of natural systems to support and nurture life. The Clean
Water Act, the Safe Drinking Water Act, and the Toxic Substances Control
Act are three of the major congressional laws that provide the framework
for restoring and maintaining the integrity of our Nation's water, for
preserving and enhancing the water we drink, and for protecting the
environment from toxic substances. These laws direct the EPA to perform
research to define our environmental problems, measure the impacts, and
search for solutions.
The Water Engineering Research Laboratory is that component of EPA's
Research and Development program concerned with preventing, treating, and
managing municipal and industrial wastewater discharges; establishing
practices to control and remove contaminants from drinking water and to
prevent its deterioration during storage and distribution; and assessing
the nature and controllability of releases of toxic substances to the air,
water, and land from manufacturing processes and subsequent product uses.
This publication is one of the products of that research and provides a
vital communication link between the researcher and the user community.
As part of the Premanufacture Notification (PMN) program, which is
mandated by the Toxic Substances Control Act (TSCA), EPA's Office of Toxic
Substances evaluates the potential hazards posed by the manufacture of new
chemicals. In support of the PMN program, the current work was undertaken
to develop improved methodologies for assessing the effectiveness of
chemical protective clothing for preventing harmful exposures to new chemi-
cal substances.
Francis T. Mayo, Hi rector
Water Engineering Research Laboratory
iii
-------
ABblRACT
A predictive model and test method were developed for determining the
chemical resistance of protective polymeric gloves exposed to liquid
organic chemicals (solvents). The prediction of permeation through protec-
tive gloves by solvents was emphasized.
Several theoretical models and test methods applicable to estimating
permeation-related properties were identified during a literature review
and were evaluated in light of permeation tests performed during this study.
The models and test methods chosen were based on theories of the solution
thermodynamics of polymer/solvent systems and the diffusion of solvents in
polymers (as opposed to being based on empirical approaches). These models
and test methods were further developed to estimate the solubility, S, and
the diffusion coefficient, D, for a solvent in a glove polymer. Riven S and
D, the permeation of a glove by a solvent can be predicted for various
exposure conditions using analytical or numerical solutions to Pick's laws.
The model developed for estimating solubility is based on Universal
Quasichemical Functional-group Activity Coefficient for Polymers (UNIFAP)
theory, which is an extension of the Universal Quasichemical Functional-
group Activity Coefficient (UNIFAC) method for predicting phase equilibria.
The model recommended for estimating diffusion coefficients versus concen-
tration is the Paul model, which is based on free-volume theory. The pre-
dictive test method developed is a liquid-immersion adsorption/desorption
method that provides estimates of S and n. The models and test method chosen
were incorporated into an algorithm for evaluating protective gloves recom-
mended (in Premanufacture Notification [PMNl suhmittals to EPA) for use with
new chemicals.
Finally, limited confirmation of the developed models and test method was
secured by comparing estimated values of S and n with reported experimental
data and by using the estimated values to predict instantaneous permeation
rates, breakthrough times, and steady-state permeation rates for comparison
with experimental permeation data.
This report was submitted in fulfillment of Contract No. fift-03-3113 by
Southern Research Institute under the sponsorship of the U.S. Environmental
Protection Agency. The study was conducted under subcontract with JRR
Associates, McLean, Virginia 22102. This report covers the period January
1985 to September 1985, and work was completed as of September
iv
-------
CONTENTS
Abstract . iv
Tables vi
Figures . . . . viii
Acknowledgments. . x
1. Introduction 1
2. Conclusions and Recommendations 4
3. Literature Review 7
Chemical-protective gloves 7
Chemical-resistance data 8
Permeation data for polymers 15
Diffusion theory 22
Predictive models 29
Predictive test methods ' 30
A. General Approach to the Development of Predictive Models
and Test Methods 36
5. Predictive Models for Solubility 38
Flory-Hugging theory 38
UN1FAP theory 47
6. Predictive Models for the Diffusion Coefficient 60
Vrentas-Duda model 60
Paul model 61
7. Predictive Test Methods for Solubility and
the Diffusion Coefficient 67
Selection of predictive test methods 67
Immersion absorption/desorption tests . . 68
Vapor absorption tests 77
Permeation test data 81
8. Predictive Algorithms 83
Specific algorithm requirements 83
Approach to the algorithm 83
Input to the algorithm-. 84
Calculations of fundamental parameters 86
Calculation of cumulative permeation or permeation rate ... 86
Evaluation of protection criteria 87
Output or evaluation report 90
Confirmation 90
References 97
Appendices
A. Summary of Test Methods for Evaluating Protective Materials. . . . 105
B. The Derivation of the Modified Paul Model and a Computer
Program 110
C. Numerical Methods for Solving Diffusion Problems . . 121
D. Summary of Liquid-Immersion Absorption Data and Permeation- -
Test Results 125
-------
TABLES
Page
Typical Formulations for Natural-Rubber, Neoprene-Rubber,
and Nitrile-Rubber Chemical-Protective Gloves
2 Typical Formulations for Poly(Vinyl Chloride) Chemical-Protective
Gloves 10
3 Types of Data Reported in Twenty-Five Journal Articles Describing
Glove-Permeation Experiments. ' 14
4 Chemicals for which Permeation Data for Natural-Rubber Gloves Exist
in the Scientific Literature 16
5 Solubilities of Selected Organic Liquids in Polymers 18
6 Diffusion Coefficients for Selected Organic Liquids in Natural
Rubber 21
7 Standard Test Methods for the Evaluation of Chemical-Protective
Materials 32
8 Values OF A, A', and A" for Selected Solvents and Natural Rubber. . 41
9 Solubilities Calculated for Various Solvents in Natural Rubber
(Based on Flory-Huggins Theory and Solubility Parameters) .... 43
10 Solvent Data Used in UNIFAP Calculations 51
11 Solubilities Calculated for Various Solvents in Natural Rubber
(Based on "Vapor/Liquid" UNIFAP Theory) 53
12 Solubilities Calculated for Various Solvents in Natural Rubber
(Based on "Liquid/Liquid" UNIFAP Theory) 57
13 Identification of Glove Materials Used in Liquid-Immersion
Absorption and Desorption Tests 70
14 Summary of the Types of Liquid-Immersion Tests Conducted 73
15 Summary of Average Liquid-Immersion Absorption Test Data 74
16 Average Solubilities and Diffusion Coefficients Calculated from
Liquid-Immersion Absorption and Desorption Test Data 75
17 Vapor Absorption Data Obtained for Nitrile Rubber and Acetone ... 80
-------
TABLES
Number Page
18 Summary of Average Permeation-Test Data .............. 82
19 Comparison of Solubilities Calculated Using the UNIFAP Model or
Obtained Experimentally with Manufacturers' Chemical-Resistance
Guidelines ............................ 89
20 Comparison of Measured .Breakthrough Times and Steady-State
Permeation Rates with Those Predicted from Immersion Test Data. . 96
21 Viscosity and Specific Volume of Benzene as a Function of
Temperature ...........................
22 Other Parameters Used in the Calculation of Diffusion Coefficients
for Benzene in Natural Rubber .................. 115
23 Viscosity and Specific Volume of n-Heptane as a Function of
Temperature ........................... 116
24 Other Parameters Used in the Calculation of Diffusion
Coefficients for _n -Heptane in Natural Rubber ........... 117
25 Permeation Rate Versus Time Calculated Using the Crank-Nicolson
Method .............................. 122
26 Summary of Liquid-Immersion Absorption Test Data .......... 126
27 Summary of Individual Permeation-Test Results ........... 131
Vll
-------
FIGURES
Number Page
1 Comparison of solubilities calculated using Flory-Huggins
theory, Cj, with experimental solubilities, Cfi 44
2 Comparison of solubilities calculated using Flory-Ruggins
theory, Cj, with experimental solubilities, Cg 45
3 Comparison of solubilities, calculated using Flory-Huggins
theory, C'j, with experimental solubilities, Cg 46
4 Comparison of solubilities calculated using vapor/liquid UNIFAP,
C , with experimental solutilibi.es, Cg 55
5 Diffusion coefficient of benzene in natural rubber as a function of
volume fraction 63
6 Diffusion coefficient of n-heptane in natural rubber as a function
of volume fraction 64
7 Diagram of vapor sorption apparatus ... ... 79
8 Comparison of predicted and experimental permeation-rate curves
for benzene through natural rubber 92
9 Comparison of predicted and experimental permeation-rate curves for
acetone through nitrile rubber 94
10 Comparison of predicted and experimental permeation-rate curves for
acetone through nitrile rubber 95
11 Absorption and desorption curves for cyclohexane in butyl rubber. . 132
12 Absorption and desorption curves for toluene in buytyl rubber . . . 133
13 Absorption and desorption curves for cyclohexane in natural
rubber 134
14 Absorption and desorption curves for toluene in natural rubber. . . 135
15 Absorption and desorption curves for cyclohexane in neoprene
rubber 136
(continued)
vin
-------
FIGURES (continued)
Number Page
16 Absorption and desorption curves for toluene in neoprene rubber. . 137
17 Absorption and desorption curves for acetone in nitrile rubber . . 138
18 Absorption and desorption curves for toluene in nitrile rubber . . 139
-------
ACKNOWLEDGMENTS
The authors are indebted to Mr. John W. Gibson for his helpful discussions
concerning diffusion theory and polymers. The authors also gratefully
acknowledge the experimental work and data reduction performed by
Mr. William J. Jessen and Mr. Richard F. Collison. Others contributing to the
project were Ms. B. Lynn Reaves, Mr. James P. English, Mr. M. Dean Howard,
Dr. Herbert C. Miller, and Dr. Richard L. Dunn.
The authors are also thankful for the support provided by the
Environmental Protection Agency under Contract 68-03-3113 through Subcontract
33-956-02 with JRB Associates and to Dr. Kin F. Wong of the Office of Toxic
Substances (U.S. Environmental Protection Agency, Washington, DC).
LABORATORY NOTEBOOKS
The work described in this report is documented in the following Institute
laboratory notebooks: D0107, D0220, D0265 and D0301.
-------
SECTION 1
INTRODUCTION
Section 5 of the Toxic Substances Control Act (Public Law 94-469) requires
prospective manufacturers of new chemicals to submit Premanufacture
Notifications (PMNs), which are reviewed by the U.S. Environmental Protection
Agency's Office of Toxic Substances (OTS). PMN submittals often propose
specific chemical-protective clothing to limit the dermal exposure of workers
to toxic chemicals. Because OTS has only 90 days to complete each PMN review
and because testing by the manufacturer must be kept to a minimum, the
development of reliable models for predicting the performance of protective
clothing is desirable. Thus, the purpose of Task I under the contract was to
develop predictive models applicable to the evaluation of the chemical
resistance of protective clothing exposed to liquid organic chemicals.
No matter how sophisticated the models developed under this contract or in
future efforts, it is likely that there may often be insufficient data
available to allow a given model to make predictions as accurate as those
requested in the statement of work for the contract (±50% for permeation rate
and ±20% for breakthrough time). Thus, predictive test methods are also needed
that will allow the estimation of the permeation of chemical-protective
clothing under expected exposure conditions. The purpose of Task II of the
contract was to develop such methods (either by recommending the use of
existing test methods or by developing new ones).
The completion of Tasks I and II, as defined above, would have been
impossible within the six-month period allowed for the performance of the
technical effort without guidelines to limit the scope of the effort. The
primary guidelines developed for the current work were:
This work emphasized the prediction of permeation through
polymeric barrier materials, especially protective gloves.
Predictive models or test methods that would have required
significantly more experimental data or effort to execute than
permeation tests were eliminated from further consideration.
The models (and test methods) developed were based as much as
possible on theory (as opposed to empirical approaches).
The technical effort emphasized the development of algorithms or
approaches rather than computer programs.
-------
Each of these guidelines is discussed briefly below.
Permeation through polymeric barrier materials. The dermal exposure of
workers who are wearing protective clothing to toxic organic liquids may occur
primarily by two mechanisms: permeation and penetration. Penetration is a
physical process that involves the macroscopic flow or transport of liquid
through openings in the protective material. These openings may be natural
(for example, the open structure in fabrics), or they may be caused by wear
(due to, for example, abrasion or punctures) or by chemical degradation.
Because penetration cannot be readily modeled and because ASTM methods exist or
are currently being developed to measure the resistance of protective materials
to penetration, the current work emphasized permeation, that is, the diffusion
of chemicals through materials.
Simple predictive models and test methods. If a predictive model requires
data that are not commonly available or easy to determine experimentally, then
the model would be of little practical value. In other words, if permeation
tests were simpler to complete (that is, less time-consuming, safer, less
expensive, and so forth) than experiments to generate the data needed to use a.
given model, then it is likely that most manufacturers would choose instead to
perform permeation tests. The same general statements may be made in the
comparison of predictive test methods with permeation tests.
Theoretical basis of models and test methods. Because the contract
required quantitative predictions (for example, breakthrough times and
permeation rates) rather than qualitative predictions (such as good, fair,
poor, and so forth), it was necessary to base the predictive models and test
methods developed as much as possible on theory, that is, on the development of
rigorous mathematical expressions. Given a mathematical expression for
permeation rate versus time, it is simple to calculate, for example, a
breakthrough time (given its definition) or the cumulative permeation at a
given time. With an empirical approach, however, the transformation from one
measure of protection (such as breakthrough time) to another (for example,
steady-state permeation rate) is usually not possible.
Emphasis on algorithms. Prior to the development of "user-friendly"
computer programs (which will be required for utilization of the predictive
models in the PMN review process), algorithms or approaches to predicting
permeation using models or test methods must be developed. After these
algorithms have been defined, the programming effort is relatively simple.
Thus, although some programming was completed, this work emphasized the
development of algorithms.
After the guidelines outlined above were developed, the work was completed
in essentially the same order presented in this report. First, a review of the
existing literature on predictive models and test methods, as well as diffusion
theory, was conducted. This review included collecting information on
chemical-protective-clothing formulations, studying chemical-resistance
literature published by manufacturers of protective clothing, evaluating
chemical-protective information published in scientific journals and" reports,
and obtaining solubility and diffusivity data for polymers used to formulate
-------
protective clothing. (As shown by the discussion beginning on page 23,
solubility and diffusivity data allow permeation to be predicted.)
Following the literature review, a general approach to the development of
predictive models and test methods was formulated. This approach incorporated
the guidelines presented above as well as considered information collected
during the literature review.
After the formulation of a general approach to the study, predictive
models for the determination of solubilities were developed. This work was
conducted concurrently with the selection and evaluation of predictive models
for determining diffusivities (that is, diffusion coefficients). Also, test
methods that resulted in solubility and diffusivity data were evaluated.
Once the work described above was completed, algorithms for the evaluation
of protective polymeric barriers (such as gloves) were outlined. These
algorithms comprise Che predictive model developed under the contract. At this
stage of the work, the distinction between predictive models and predictive
test methods became unimportant. That is, the predictive algorithms functioned
on the basis of the data available, whether from a model or from a test.
After the predictive model (a set of algorithms) was developed, it was
used to make a limited number of predictions of breakthrough times and
permeation-rate curves, and these predictions were compared to experimental
data and manufacturers' recommendations.
-------
SECTION 2
CONCLUSIONS AND RECOMMENDATIONS
Models that have been developed for Che prediction of the permeation of
organic compounds through protective-glove polymers have often been based on
empirical approaches with little emphasis on diffusion theory. The current
work has demonstrated that predictive models and predictive test methods that
yield diffusion coefficients and solubilities may be used to estimate
permeation data such as breakthrough times (for a given definition),
»teady-state permeation rates, and permeation-rate curves. These fundamental
parameters may be estimated using the theoretical models or the simple test
methods described in this report. Only relatively simple diffusion theory and
mathematical methods are required to calculate permeation data from diffusion
coefficients and solubilities. These permeation data may then be used to
estimate the protection afforded by polymeric gloves recommended in PMN
submittals.
This work has shown the potential of theoretical methods for estimating
solubilities and diffusion coefficients of organic compounds in polymers used
in the manufacture of chemical-protective gloves. Theoretical models for
predicting these fundamental parameters are especially useful for new chemicals
because the models require only Limited physicochemical data, such as density
and viscosity as a function of temperature. For example, a model based on
UNIFAP theory needs only the densities of the organic compound and the polymer
at the temperature of interest to estimate the solubility of the compound in
the polymer (given, of course, the structure of the polymer and of the organic
compound and other limited information).
Because even limited physicochemical data may not be available for the
organic compound and polymeric glove of interest, it may frequently be
necessary to conduct experiments to determine the resistance of the glove to
permeation. The traditional method of determining the chemical resistance of a
protective-glove material is to conduct a permeation test. However, such tests
generally require the initial purchase of expensive analytical instrumentation,
such as gas chromatographs, and they must be performed by well-trained
technical personnel. The current research effort has resulted in the
application of a simple liquid-immersion test to the determination of data that
can be used to estimate solubilities and diffusion coefficients for organic
compounds in protective-glove polymers. This test method requires only the
purchase of a simple and relatively inexpensive analytical balance. Also,
personnel with' limited technical backgrounds can easily perform immersion
tests.
The liquid-immersion test described in this report may be used to estimate
permeation-rate curves (and, thus, parameters such as breakthrough times) even
when the identity and composition of the polymeric glove material and the
identity of the organic liquid are unknown. This predictive test method would
prove especially useful if a manufacturer recommends a protective glove based
on a proprietary polymer formulation. It could also be used for evaluating
-------
protective gloves with complex formulations that could not be modeled using
theoretical methods or when the physicochemical data necessary to use
theoretical methods are unavailable.
The work described in this report was a feasibility study. Although
models and test methods for the prediction of permeation data are presented,
the brief confirmation studies described were conducted only for unsupported
glove formulations. There is a need to extend the research effort to a wider
range of protective-glove formulations and organic chemicals. For example,
neither the theoretical models nor the immersion test can currently generate
data sufficient to allow the quantitative prediction of permeation rates for
organic chemicals through PVC-based protective gloves.
Most of the limited confirmation work conducted under this contract was
performed "manually." That is, although computers were used to model polymer/
solvent systems, to perform curve fits of experimental data, and to predict
permeation-rate data, the computer programs written were not integrated into a
single software package and they were not made "user friendly." Thus, work
under the confirmation task was tedious and time-consuming. For this reason,
it was not possible to evaluate all of the experimental immersion and
permeation data obtained under this contract, and the Paul model was used to
predict diffusion coefficients for only two polymer/solvent systems.
The limited development of integrated computer software during the current
feasibility study should not be considered unusual, particularly because this
study emphasized the identification and preliminary investigation of predictive
models and test methods. Thus, much of the effort consisted of reviewing the
literature and completing mathematical derivations.
Any continued research effort based on the work described here, however,
should begin with the consolidation of the programs written into a single
software package. This package should also include the predictive algorithms
outlined in Section 8 of this report. Thus, future confirmation work should,
to a large extent, begin to resemble the preparation and evaluation of PMN-
submittal-review software.
The predictive algorithms described in Section 8 emphasize diffusion
theory; thus, time-dependent permeation (or immersion) data are necessary to
confirm the algorithms. Because of the general lack of such data in the
chemical-resistance literature, the initial confirmation work under a future
contract should emphasize the use of the experimental data presented in the
separate data volume accompanying this report. However, additional
time-dependent permeation and immersion test data should be generated for a
variety of polymeric glove materials and organic chemicals.
It may also be possible to use semiquantitative data bases such as those
represented by manufacturers' chemical-resistance brochures or permeation
databases to assist in the confirmation of predictive algorithms. However,
such semiquantitative data bases should not strongly influence the development
of predictive algorithms, because these data bases often contain information on
the chemical-resistance of protective gloves of unknown origin and parameters
such as breakthrough times may not be properly defined.
-------
Future efforts to continue the work described here should include the
identification and use of computerized data bases of physicochemical data that
are needed to make predictions of diffusion coefficients and solubilities
[using, for example, the Paul model or Universal Quasichemical Functional-group
Activity Coefficients for Polymers (UNIFAP) theory]. Also, work should include
addressing the limitations of the UNIFAP theory and the modified Paul model
described in this report. In addition, the consideration of other theoretical
models should be encouraged.
Perhaps the major recommendation to result from the current study is that
there should continue to be a strong emphasis on the development of predictive
algorithms that are based as much as possible on the rigorous interpretation of
diffusion theory. This approach will allow permeation-rate and cumulative-
permeation curves to be calculated as desired; other permeation-related
parameters (for example, breakthrough times) can be determined from these
curves. It should be noted that rigorous predictive algorithms can always be
modified to yield simple correlations; however, the extension of an algorithm
based on empirical correlations to the calculation of quantitative data is
often difficult if not impossible.
-------
SECTION 3
LITERATURE REVIEW
A literature review was conducted Co develop a broad-based understanding
of the current state of knowledge concerning chemical-protective clothing and
to collect permeation data that could be used to evaluate predictive models
and test methods. The review included collecting and evaluating information
about chemical-protective-glove formulations, chemical-resistance data for
gloves, and solubility and diffusivity data for polymers commonly used in
fabricating protective clothing. In addition, the fundamentals of diffusion
theory were reviewed, and information on predictive models and test methods was
collected and evaluated.
The information collected during the literature review was obtained from a
number of sources, including journal articles, manufacturers' literature,
Government reports, and discussions with experts on diffusion in polymers and
protective clothing.
The key information collected during the literature review was summarized
in the monthly and quarterly reports prepared under this contract, and most of
this information is repeated in this report. In addition, copies of about
70 articles of particular relevance were provided to EPA during the contract
period.
One of the early key findings of the literature survey was that almost all
of the published chemical-resistance data concerns polymeric gloves. Thus,
because of the need for experimental data with which to evaluate predictive
models and test methods, the emphasis throughout the remainder of the contract
was placed on gloves.
CHEMICAL-PROTECTIVE GLOVES
The development of predictive permeation models (Task I) and test methods
(Task II) depended on an understanding of chemical-protective gloves. Thus,
a survey was conducted to identify glove manufacturers, construction and
manufacturing processes used by them, and glove styles and formulations. The
information obtained in the survey was based on the experience of Institute
staff in the manufacture of protective gloves, on manufacturers' literature,
and on references such as Waterman (I) and Blackley (_2) .
The information obtained in the survey of protective-glove literature
showed that seven glove formulations account for almost all of the commercial
market. These are poly(vinyl chloride), natural rubber, neoprene rubber,
nitrile rubber, butyl rubber, poly(vinyl alcohol), and polyethylene. Six glove
styles are in general use; these are unsupported, coated interlock,.coated
jersey, coated flannel, inner flocked, and thin, unsupported disposable. Also,
the manufacturing processes and construction methods,used vary greatly
-------
depending on the glove style. There are approximately 35 manufacturers of
chemical-protective gloves in the United States.
The development of predictive models and, to some extent, the development
of test methods are made difficult by the large number of glove manufacturers
and the variety of construction and manufacturing processes used by them.
Also, polymeric gloves are not "pure" polymers, but they include other
components such as fillers, plasticizers, stabilizers, pigments, and
antioxidants (for example, see Tables 1 and 2). The identities and amounts of
these additives in a given formulation will often have a major impact on the
rate of permeation of chemicals through the glove. This is particularly true
for poly(vinyl chloride) gloves, which often contain a higher percentage of
additives, especially plasticizers, than the PVC resins used to form the
gloves. PVC gloves have the largest share of the market in chemical-protective
gloves.
Even gloves that are formulated for the same application, but that are
made by different manufacturers, often have very different chemical-protective
properties. For example, Williams (_3) reported a variation of a factor of
three in the breakthrough times obtained for the permeation of carbon
tetrachloride through PVC gloves of the same thickness and formulated for the
same purpose, but made by three different manufacturers.
After a review of glove styles, construction methods, manufacturing
processes, and formulations, it was concluded that unsupported natural-rubber
gloves would be the simplest for which to begin the development of predictive
permeation models and test methods primarily because of the "purity" of the
typical formulation for such gloves. Thus, much of the modeling effort during
the remainder of the contract emphasized natural rubber.
It should be noted that the validity of the PMN approval, whether based on
a predictive model or on test data, may be nullified in the event a
manufacturer elects to make a change in the formulation of a glove. This
change may be dictated by economic considerations or by the availability of
materials, and the change may affect the permeability of the glove with respect
to the chemical against which it is supposed to provide protection. For this
reason, once PMN approval is granted, it should be made specific for a given
glove formulation from a specific glove manufacturer (both recommended, of
course, by the prospective manufacturer of the new chemical).
CHEMICAL-RESISTANCE DATA
After the development of a predictive model (or predictive test method),
its applicability must be confirmed by comparing predictions such as
breakthrough times (for a given definition), steady-state permeation rates, and
perraeation-rate-versus-time data with experimental chemical-resistance data
of the same type. Thus, the existing chemical-resistance literature was
searched for such quantitative data.
-------
TABLE 1. TYPICAL FORMULATIONS FOR NATURAL-RUBBER, NEOPRENE-RUBBER,
AND MITRILE-RUBBER CHEMICAL-PROTECTIVE GLOVES3'b
Component
Latex
Plasticizer
Zinc oxide
Clay
Color pigments
Antioxidant
Sulfur
Potassium hydroxide
Butyl zioate
Other accelerators
Ammoniated casein
Thickener
Wetting agents
Bactericide
Natural
Uns
93.3
0
0.9
0
1.4
1.0
1.0
0.5
0
1.4
0.5
0
0
0.02
rubber
Sup
83.5
4.0
1.6
0-5
1.5
0.8
0.8
0.5
0
1.6
0.5
0.08
0.08
0.02
Neoprene
Uns
83.0
4.0
4.2
4.0
1.5
0.8
0.8
0.1
0.8
0
0.3
0.5
0
0
rubber
Sup
80.5
4.0
4.0
4.0
2.0
0.8
1.0
0.1
0.5
2.0
0.5
0.1
0.5
0
Nitrile
Uns
93. 8C
0
0
0
1.5
1.0
0.2
0
0.5
2.5
0.5
0.5
0
0
rubber
Sup
91. Oc
0
2.5
0
1.5
1.0
0
1.0
1.0
1.0
0.2
0.4
0.5
0
aThe numbers in the table are in units of weight percent. "Uns" means
unsupported; "Sup" means supported.
See, for example, Waterman (_U and Blackley (2).
cNitrile gloves are prepared from a copolymer of about 40% acrylonitrile
and about 60% butadiene.
-------
TABLE 2. TYPICAL FORMULATIONS FOR POLY(VINYL CHLORIDE)
CHEMICAL-PROTECTIVE GLOVES*
General Purpose
Component
High-raolecular-weighc PVC resins
Phchalate-type plascicizers
Filler
Pigment
Heat stabilizer
Weight %
42-52
42-53
0-5
0.2
0-2.6
Solvent Resistant
Component
High-molecular-weight PVC resins
Butyl benzylphthalate;
polyglycol benzoates
Polyester plasticizer
Filler
Pigment
Heat stabilizer
Weight %
42-52
23-39
15-26
0-5
0.2
1.0-2.5
Low Temperature
Component
High-molecular-weight PVC resins
Phthalate-type plasticizers
Adipate-type plascicizers
Filler
Pigment
Heac stabilizers
Weight I
43-46
35-46
7-14
0-2
0.2
1,2-2.3
aSee, for example, Waterman (_1_) and Blackley (2)
10
-------
Manufacturers' Brochures
Manufacturers of chemical-protective gloves generally provide chemical-
resistance tables in their marketing brochures. These tables usually include
no permeation data but only recommendations such as excellent, good, fair,
poor, and not recommended. Often, no criteria for these ratings are given, and
thus, they are of little value in making or confirming quantitative predic-
tions.
Two exceptions to the qualitative data generally presented by manufac-
turers of gloves are those provided by North Hand Protection (Siebe North,
Inc.) and by Edraont (Beeton Dickinson and Company). Both the North and Edmont
brochures include measured breakthrough times. However, most details concern-
ing the permeation-test methods used to generate the data are not included in
the North and Edmont literature. For example, the North brochure defines
breakthrough time as "the elapsed time between initial contact of the liquid
chemical with the outside surface of the glove and the time at which the chemi-
cal can be detected at the inside surface of the glove by means of the
analytical technique." However, neither North nor Edmont include analytical
detection limits for any of the chemicals in their literature.
North also gives the maximum permeation rates observed during tests of
unspecified duration. The Edmont literature categorizes the maximum permeation
rate during a six-hour test as none detected, excellent, very good, and so
forth. Each qualitative maximum permeation rate reported by Edmont is defined
in terms of an order-of-magnitude quantitative range; for example, a maximum
permeation rate categorized as "good" means that the maximum rate observed was
in the range of l.S to 15 mg/(m2>sec).
There is no way to relate the breakthrough times and maximum
permeation-rate data reported for North and Edmont glovea to fundamental
parameters such as solubilities and diffusion coefficients, which are needed to
predict quantitative permeation data under a variety of exposure conditions.
It should be noted also that both North and Edmont recommend strongly that
users conduct their own permeation (and degradation) tests to determine the
suitability of a given protective glove for a specific application.
Data similar to those provided by North and Edmont are also available from
the Pioneer Industrial Products Division of Brunswick Corporation. These data
(as well as those of North and Edmont) are included in the second edition of a
two-volume document entitled "Guidelines for the Selection of Chemical
Protective Clothing" (4). The Pioneer data given in that report include
percentage weight change and percentage volume change as a function of time for
polymeric glove samples immersed in organic liquids. The thickness of each
glove sample is also presented in the report. However, neither the initial
weights nor the initial volumes of the samples are given; thus, the
weight-change and volume-change data cannot be used to calculate solubilities
in units of concentration (for example, g/cm3).
11
-------
Government Reports
A number of reports published under Government contract were reviewed.
These included the two-volume document entitled "Guidelines for the Selection
of Chemical Protective Clothing" (5}, which was originally published by
Arthur D. Little, Inc., under the auspices of the EPA and the American
Conference of Government Industrial Hygienists. Although the original document
included no quantitative data on the permeation of polymeric materials, a
revised version that includes breakthrough times and steady-state
permeation-rate data collected from a number of sources was issued recently
(4_). Tables of immersion data (including percentage weight changes and
percentage volume changes as a function of time) and diffusion coefficients
have also been added to the document. Almost all of the diffusion coefficients
reported are, however, for chemical-protective materials for which the
manufacturers are unknown.
The document referenced above also includes an overview of considerations
involved in the selection of chemical-protective clothing, a review of simple
permeation theory, and a list of ASTM methods currently used to evaluate
protective clothing. In addition, a summary of manufacturers' recommendations
and 280 references are presented.
SRI International recently published a report entitled "Studies to Support
PMN Review: Effectiveness of Protective Gloves" (b). The report includes a
brief introduction to simple permeation theory. It also presents tables and
figures that attempt to show a correlation between separation of solvent* and
polymer in solubility space and permeability coefficient. Observed
correlations were good to poor. Thus, the predictive value of such
correlations is questionable. Also, little correlation between separation in
solubility space and diffusion coefficient was observed.
One of the primary points made in the SRI International report is that
little permeation data that can be used in evaluating models currently exist.
Also, much of the data that do exist is of questionable value. For example,
breakthrough times reported for various glove/solvent combinations are of
limited use in the calculation of diffusion coefficients, because these times
depend on the sensitivity of the analytical methods employed. Also, on the
basis of simple diffusion theory, the so-called lag times rather than
breakthrough times should be used to calculate diffusion coefficients; the SRI
International report stated that diffusion coefficients based on breakthrough
times may be in error by as much as a factor of ten.
The SRI International report also includes estimates of diffusion
coefficients for various glove/solvent combinations. It also contains
permeability coefficients for glove/solvent combinations that were based on
reported steady-state permeation rates. Each steady-state permeation rate was
divided by the vapor pressure of the solvent at saturation and multiplied by
*The word "solvent" is often used in this report as a substitute for the phrase
"liquid organic chemical."
12
-------
the thickness of the glove polymer to yield an estimate of the permeability
coefficient. The SRI International report also includes brief discussions of
solubility parameters and how to calculate them; in addition, it contains 55
references.
Another document reviewed under the current contract was a draft copy of a
monograph being prepared by NIOSH (T). This monograph, which summarizes a
computerized data base, includes breakthrough times and steady-state permeation
data collected from a variety of sources. Although this report will represent
an impressive collection of information when it is published, it does not
contain all the data necessary (for example, diffusion coefficients and
solubilities) to evaluate a model that must make quantitative predictions, such
as permeation rate versus time. As illustrated by the discussion below, the
primary reason such data (for example, analytical detection limits) are not
included in the monograph is that these data are generally not available.
Journal Articles
Journal articles were collected that describe experiments in which
protective gloves were exposed to various solvents and other chemicals.
Table 3 gives a summary of the types of data contained in these articles that
specifically address glove permeation. These articles and the data included in
them are believed to be representative of the information on the permeation of
protective gloves that has been reported in the scientific literature.
Unfortunately, much of the information reported in the articles referenced
in Table 3 demonstrates a lack of appreciation of permeation theory by the
authors. Also, extracting useful data from the literature is often made
difficult by the careless use of terminology; for example, the terms
"permeation rate" and "cumulative permeation" are frequently used
interchangeably even though the significance of the two types of data and the
theories used to interpret the data differ considerably. Thus, most of the
data reported in the literature are of limited value in the validation of
either predictive models or test methods.
The most significant data that can be generated to allow the validation of
models and simple Cest methods are the permeation rate* (J) or the cumulative
permeation (Q) as a function of time. Unfortunately, little of these data
exists for protective' gloves. Host of the data that exist are in the form of
breakthrough times, which are dependent on analytical sensitivity and which
provide insufficient information to characterize fully the permeation being
studied.
A significant amount of the permeation data and interpretation of these
data is based on misunderstandings of simple permeation theory. For example,
the so-called normalized breakthrough time (t-/£2 or t / £2) and the product Q*&
listed in Table 3 have no basis, even in simple theory. Yet the concept of
*The correct terminology is permeation flux; however, the use of "rate" in
place of "flux" is widespread.
13
-------
TABLE 3. TYPES OF DATA REPORTED IN TWENTY-FIVE JOURNAL ARTICLES
DESCRIBING GLOVE-PERMENATION EXPERIMENTS
Data reported References
Permeation rate (J) versus time 3, 8
Cumulative permeation (Q) versus time 9-20
Lag tine (CL) 8
Steady-state permeation rate (Ja) 3, 8, 11, 13, 19, 21, 22
Permeation rate (J) at selected times 23, 24
Cumulative permeation (Q) at selected 12, 20, 25, 26
times
Ratios of products of cumulative 27
permeation (Q) and polymer
thickness (i)
Weight or volume changes at a given 8-10, 20, 28
time
Breakthrough times (t-) based on 3, 8, 13, 22, 23
permeation rate
Breakthrough times (t ) based on 3, 8-12, 14, 19-21
cumulative permeation
Normalized breakthrough times (c-/42 8-11, 20
or
Diffusion coefficients (calculated 11, 14
incorrectly)
Other 13af 15b, 16C, 18C , 28d, 29e, 30f, 31*
*Time at which the permeation rate is equal to one-half the steady-state
rate.
Percent equilibrium.
Distribution of nicrosanines between solvent, glove, and receiving fluid.
Percent loss of solvent through glove; percent change in length; other
physical effects.
cDroplet test data (generated using a so-called splash test).
Diffusion coefficient and solubility (incorrectly labeled solubility
coefficient) versus vapor concentration.
^Physiological symptoms versus time in glove-permeation tests using human
subjects.
-------
normalized breakthrough time is widely used, and such data are frequently
reported.
On first examination, there appears to be a significant (though not exten-
sive) body of data in the open literature on the permeation of protective-glove
materials by chemicals. For example, Table 4 lists 19 references for studies
of the permeation of natural-rubber-based gloves by 67 compounds. An
examination of the articles referenced in the table, however, revealed that
very little of the data (which represent about 130 permeation experiments)
would be of significant value in evaluating quantitative predictive models or
test methods for natural-rubber gloves. For example, the permeation data for
only three chemicals (pentachlorophenol, dimethyl sulfoxide, and benzene) in
three references listed in the table meet the following simple criteria:
That a permeation-related property (for example, permeation
rate, cumulative permeation, weight change, or volume change) be
reported as a function of time.
That the glove be clearly identified (that is, the glove
manufacturer, type, catalog number, and thickness should be
reported).
Even more revealing is that of more than 2000 glove/chemical-permeation
experiments found in 23 references, only about 52 of these experiments meet the
two criteria listed above.
A more extensive survey of the type described above was beyond the scope of
this contract. The limited survey results presented here are only intended to
demonstrate the lack of data suitable for evaluating quantitative, predictive
models or test methods.
Not all glove-permeation data in the literature are worthless for
evaluating predictive models. The data that exist are useful for comparing
trends in a homologous series of compounds or for a single compound in a series
of glove materials. Also, some of the data are sufficient to allow the
evaluation of models and test methods that are intended to identify gross
glove/chemical incompatibilities.
PERMEATION DATA FOR POLYMERS
Because little time-dependent permeation-related data were found to exist
for polymeric glove samples, the literature was reviewed to obtain such data
for polymers commonly used in the manufacture of protective gloves.
Approximately 118 journal articles were reviewed; of these, only 17 contained
data of potential use in the present study (32_-4£). These data included
solubilities, diffusion coefficients, and permeabilities (the product of the
diffusion coefficient and the solubility). The most relevant of the solubility
and diffusion-coefficient data in the journal articles referenced (as well as
in a report by ADL (49)) are presented in Tables 5 and 6.
15
-------
TABLE 4. CHEMICALS FOR WHICH PERMEATION DATA
FOR NATURAL-RUBBER GLOVES EXIST
IN THE SCIENTIFIC LITERATURE
Compound
Acetone
Acrylonitrile
Allyl glycidyl eCher
Aniline
Aroclor
Benzene
Butyl acetate
n-Butanol
Carbon tetrachloride
jj-Chloroaniline
Chloroform
m-Cresol
Cyclohexane
1 ,2-Dibromo-3-chloropropane
1 ,2-Dibromoe thane
Dichlororae thane
Dimethyl formaraide
Dimethyl su If oxide
Dioxane
Epichlorohydrin
Ethanol
Ethylene dibromide
Ethylene dichloride
Ethylene glycol
Ethylene glycol dinitrate
Freon TF
ji-Hexane
Isoamyl acetate
Methanol
Methylene bis(2-chloroaniline)
Methylene chloride
4 ,4 '-Methylene dianiline
Methyl ethyl ketone
Methyl iodide
Nitric acid, inhibited red fuming
Nitroglycerin
N-Nitrosobutyl methylamine
N-Nitrosodibutylamine
N-Nitrosodiethylamine
_N-Nitrosodimethylamine
N-Nitrosodipropylaoine
N-Nitrosoethyl methylamine
jJ-Nitrosomethyl pentylamine
N-Nitrosopiperidine
(continued)
16
Reference
14, 15, 18, 22, 27, 28
25
28
15, 27
10
15, 20a, 22, 29, 30
22
15
15, 22, 29
14
15, 22
22
22
25
8
14, 18
15, 27
12a, 15, 22
15, 22
8
14, 15, 18, 22
25
22
22
31
22
15
22
15, 22
14
15, 22
14
15, 22
15
23
31
18
16, 17
16-18
16-18
16-18
18
16
16, 17
-------
TABLE 4 (continued)
Compound
Reference
]f-Nitropyrrolidine 16, 17
N-Nitrosodi-sec-butylamine 18
Pentachlorophenol lla
Pentane 22
Perchloroethylene 8, 15
Phenol 9, 15, 22
Phenyl glycidyl ether 28
2-Propanol 15
Pyridine 15, 22
1,1,2,2-Tetrachloroethane 15
Tetrachloroethane 22
Tetrahydrofuran 15, 22
Toluene 15, 22, 28
j>-Toluidine 14
Trichlorobenzene 10
1,1,1-Trichloroethane 22, 28
1,1,2-Trichloroethane 15
Trichloroethylene 8, 22
Trifluoroethanol 22
Unsymmetrical dimethylhydrazine 23
Water 28
Water, tritiated 14, 15, 18
Xylene 22
aOnly these reference/chemical combinations meet the
acceptance criteria listed on page 15.
17
-------
TABLE 5. SOLUBILITIES3 OF SELECTED ORGANIC
LIQUIDS IN POLYMERS
Polymer Organic liquid
Butyl rubber Benzene
1 , 1-Dimethylhydrazine
Epichlorohydrin
Ethyl enimine
2-Nitropropane
Trichloroethylene
Natural rubber Acetone
Benzene
Benzyl alcohol
n-Butanol
^-Butanol
Carbon tetrachloride
Cyclohexane
Cyclohexanone
Diethyl carbonate
Ethanol
Ethyl acetate
2-Ethyl-l-butanol
Ethylenimine
ji-Heptane
n-Hexane
Reference
49
49
49
49
49
49
48
49
48
48
48
48
48
48
48
48
48
48
49
48
48
Solubility,
cmVcm^
0.662
0.138
0.044
0.169
0.021
1.46
0.125
3.91
0.147
0.114
0.391
5.21
3.67
3.55
0.770
0.0050
0.755
0.319
0.169
2.33
2.03
Temp . ,
K
295
295
295
295
295
295
297
295
297
297
303
297
297
297
297
297
297
297
295
297
297
(continued)
18
-------
TABLE 5 (continued)
Polymer Organic liquid
Isopropanol
Met Hanoi
Methyl ethyl ketone
ji-Pentanol
£-Pentanol
jn-Propanol
ji-Propyl acetate
Tetrachloroethylene
Tetralin
Toluene
Trichloroethylene
Neoprene Benzene
rubber
Carbon tetrachloride
Diisobutylene
1,1 -Dime thylhydrazine
Epichlorohydrin
Ethyl acetate
Methanol
Reference
48
48
48
48
48
48
48
48
48
48
48
40
49
40
40
49
49
40
40
Solubility,
cmVczn3
0.0406
0.0020
0.642
0.142
0.479
0.110
1.47
4.29
4.53
4.13
5.13
2.98
3.75
4.12
3.91
3.44
3.69
3.41
0.25
0.47
0.57
0.664
0.455
1.20
1.16
1.27
0.04
0.26
Temp . ,
K
297
297
297
297
297
297
297
297
297
297
297
298
328
355
295
298
328
355
298
328
355
295
295
298
328
355
298
328
(continued)
19
-------
TABLE 5 (continued)
Polymer
Nicrile
rubber
Polyethylene
Poly(vinyl
alcohol)
Organic liquid
Methyl ethyl ketone
2-Nitropropane
Trichloroethylene
1 , 1 -Dime thy Ihydrazine
Epichlorohydrin
2-Nitropropane
Trichloroethylene
Benzene
Epichlorohydrin
iv-Heptane
n-Hexane
2-Nitropropane
Trichloroethylene
Benzene
Trichloroethylene
Reference
40
49
49
49
49
49
49
35.36
49
49
35,36
35,36
49
49
49
49
Solubility.
cm3/cm3
1.42
1.52
2.02
0.289
0.993
0.631
0.24L
0.717
1.46
0.0343
0.0350
0.0355
0.295
0.092
0.0542
0.0535
0.0524
0.0264
0.0276
0.0290
0.065
0.039
0.035
0.028
Temp.
K
298
328
355
295
295
295
295
295
295
298
303
308
295
295
298
303
308
298
303
308
295
295
295
295
aThe solubilities reported are in units of cm3 of organic liquid per
of unswollen polymer. The densities of the organic liquids, which can be
found in chemical reference-data handbooks, may be used to convert the
solubilities given to units of g/cm3.
20
-------
TABLE 6. DIFFUSION COEFFICIENTS FOR SELECTED ORGANIC
LIQUIDS IN NATURAL RUBBER3
103
Organic liquid
Benzene
Carbon tetrachloride
Cyclohexane
Cjrclohexanone
n-Keptane
n-Hexane
Isooctane
Methyl ethyl ketone
Methyl isobutyl ketone
Tetralin
Toluene
o-Xylene
x Concentration.
moles/cm3
43.1
54.0
35.9
26.0
15.9
15.0
11.8
6.91
12.3
32.4
40.2
36.7
LO6 x Diffusion
coefficient, cm2/sec
2.37
1.67
2.01
1.80
1.78
0.66
0.74
3.07
4.06
2.73
1.95
4.09
1.67
0.82
2.36
2.75
1.78
Reference0
45
45
47
45
46
45
47
45
45
45
47
45
45
45
45
47
45
data were obtained at 303 K.
The concentrations reported are in. units of moles of organic liquid per
cm3 of unswollen polymer. (These data represent solubility data in
addition to those given in Table 5). Each reported diffusion coefficient
was experimentally determined at the corresponding concentration.
cThe column labeled "D x 10**" in Table 1 of Reference 47 should be
labeled "Dm x 106".
21
-------
Of some interest also are solubility data reported by Curry and McKinley
(44) for acetone and benzene in natural rubber as a function of the partial
pressure of the vapor to which the polymer was exposed. This reference
includes expressions for the concentration dependence of the diffusion
coefficients of acetone and benzene in natural rubber. (The data are not
presented here because the primary interest under the current effort is the
exposure of polymers to organic liquids rather than to vapors.)
Another article of interest was authored by Fujita (50) . It included a
review of diffusion theory as well as data originally published by Hayes and
Park (51) that demonstrate the concentration dependence of the diffusion
coefficient for benzene in natural rubber. (These data are included in
Figure 5 in this report.)
DIFFUSION THEORY
During the literature survey, diffusion theory was reviewed to ensure that
the critique of literature data and that proposed models and test methods were
based on the correct interpretation of theory. Several well-recognized
references were consulted ( 52_-5b) A brief summary of simple diffusion theory
is presented below.
Fick'a Laws
In general the permeation of polymeric gloves by an organic liquid may be
described by Pick's laws of diffusion:
j = -D3C/3X (1)
3C/at = ,5(Dac/3x)/ax (2)
where j is the flux through a plane perpendicular to the x axis, D is the
diffusion coefficient, and C is the concentration of the solvent in the
polymer.
To solve Equations 1 or 2 for quantities such as permeation rate, J,
versus time or concentration versus time and position in the polymer, it is
necessary to specify initial conditions (for example, C(x,0)sO) and boundary
conditions (for example, C(0,t) - S and C(fc,t)=0). After the initial and
boundary conditions are specified (and, usually, after other simplifying
assumptions are made), the differential equations may be solved to yield the
desired analytical solutions. Alternatively, numerical methods may be used to
"solve" the equations.
Permeation
The simplest theory that describes the permeation of a polymeric glove by
a solvent requires that the following assumptions be made:
Diffusion obeys Pick's laws.
22
-------
Diffusion occurs in only one dimension.
The diffusion coefficient, D, is independent of the solvent
concentration in the polymer.
The initial concentration of the solvent throughout the polymer
sample is zero.
The solvent has a finite solubility, S, in the polymer, and
this solubility is attained instantly at the surface of the
polymer in contact with the solvent.
The concentration of the solvent on the challenge side of the
polymeric glove sample is much greater than that in the
receiving fluid (gas or liquid) on the opposite side of the
sample throughout the measurement.
The thickness of the polymer is constant; that is, negligible
swelling occurs as the solvent dissolves in the polymer.
The temperature of the test apparatus and the polymeric sample
remains constant throughout the experiment.
The solvent does not react with the polymer or alter its
physical properties.
Almost all of the interpretation of permeation data for protective gloves
reported in journal articles is based on these assumptions.
If the assumptions above are true (and often one or more are not), then
the time-dependent diffusion equation, Equation 2, may be solved to yield:
J =
-------
Q - (DS/i){t - *2/6D _ (2£2/1I2D) £ [(.tf* exp{-m2ir2Dt)/ *2]/,n2J (4)
m=l
If Q-versus-C or J-versus-C data were generated and reported routinely,
then the fundamental parameters D and S could be determined. And these data
could be used for the evaluation of predictive models and test methods.
Because Equations 1 through 4 look complicated and require the use a
computer program, albeit very simple, to determine D and S, workers who evaluate
protective gloves usually seek approximations that may be useful for comparing
permeation data. One approach to determining the parameters D and S without the
aid of a computer is to transform Equation 3, using a method suggested by
Holstein (.55) , to yield the so-called early-time approximation:
ln(J.tl'2) , ln[2S(D/ir)l/2] - fc2/4Dt (5)
If this equation is used, a plot of ln(J»t1/2) versus 1/t yields a straight
line with, a slope equal to -£2/4D and a y-intercept equal to ln[2S(D/ir) 1/2] .
Thus, in this manner, D and S can be determined graphically.
At long times, Equation 3 reduces to the so-called steady-state
approximation:
(6)
where J^ is the steady-state permeation rate. Thus, in tests where
steady-state permeation is achieved, researchers have only to calculate the
product DS (also known as the permeability) to compare the relative
effectiveness of different types0 of protective gloves.
Although the calculations are simple, the determination of steady-state
permeation rates is of limited value because no time-dependent information is
obtained. Also, because of the breakdown of the assumptions listed above,
steady-state permeation may never be achieved for a given glove/solvent
combination. In addition, knowing the permeation rate for a given glove polymer
may be of much more significance at early times, as the protection afforded by
the glove begins to fail, than at long times when the steady-state permeation
rate may be far above that which results in acceptable exposure levels.
Steady-state permeation rates (or frequently, a detector signal
proportional to J^) are often multiplied by glove thickness ( &) to yield a
result which is directly proportional to the product DS (see Equation 6). Thus,
a comparison of J^-i values for various gloves enables their ability to protect
against permeation by a solvent at steady state to be compared. Unfortunately,
even this simple normalization has been incorrectly applied. For example,
Sansone and Tewari (27) multiplied cumulative-permeation (Q) data rather than
steady-state permeation-rate (J^) data times glove thickness ( £) in an
ill-founded attempt to correct for the effect of sample thickness in their study
of gloves obtained from several manufacturers.
Another useful approximation may be obtained by reducing Equation 4 for
long times to:
24
-------
<}= (DS/4){t - fc2/6D} (7)
If cumulative-permeation data obtained at long times (more correctly, when
steady-state permeation is achieved) are extrapolated to zero cumulative
permeation, then Equation 7 reduces to:
0 = (DS/£){tL - *2/6D} (8)
where t^ is the so-called lag time. And,
t = *2/6D (9)
Thus, if the lag time is determined, then the diffusion coefficient D may be
calculated.
The misinterpretation of Equations 8 and 9 has resulted in two major
errors in the literature:
Confusion of the lag time with breakthrough time.
"Normalizing" breakthrough-time data by dividing by j2,
Researchers often use an experimental method and analytical technique to
determine the time at which the permeation rate (J) or the cumulative
permeation (Q) exceeds a given value, usually the analytical detection limit.
And, as stated above, they frequently and incorrectly equate the breakthrough
time with lag time.
Mathematically, the breakthrough t-imes generally used (t and t- below) are
defined by the equations:
Qb =
-------
also recommends reporting the steady-state permeation rate, which can be used to
estimate the product DS. In addition, it recommends reporting a graph of
cumulat ive-permeation-versus-time data, which, if reported correctly, could be
used to determine D and S if the assumptions listed on pages 22 and 23 remain
valid throughout a given experiment. Also, more sophisticated theories than
describe above could be used to analyze the data when, for example, D is
concentration dependent. Unfortunately, most testing laboratories using ASTM
Method F739-81 do not report either J-versus-t or Q-versus-t data.
Absorption and Desorption
Other data reported in the literature include the percentage weight gain
at a specific time (usually 24 hr) for a polymer sample immersed in an organic
Liquid. Simple diffusion theory may be used to show that the weight gain as
a function of time for a thin planar sample of a polymer immersed in a liquid
is given by the equation:
Mt/M- " X ~ <8/n2) <2tm-l)-2 exp(-D(2m+l) 2B2t/ £2)
m=0
where M£ is the net weight gained at time t and M^ is the net weight gained at
long times (at equilibrium). (S may be calculated from the value determined
for M^.) However, the weight gain at a single time does not yield enough
information to allow the calculation of D and S.
The weight gained per unit volume at 24 hr (or at another fixed time) by a
polymer sample immersed in an organic liquid is often considered to be the
solubility, S. -However, most researchers who study the absorption of organic
liquids by protective gloves fail to prove that equilibrium is actually
achieved. That is, they 'do not determine the weight gained as a function of
time to show that equilibrium has been achieved. Thus, the weight gain of the
polymer at 24 hr cannot be assumed to equal the value of S.
The weight gained by a given polymer sample when it is immersed in an
organic liquid may not reach an equilibrium, [n fact, weight loss caused by
the leaching of additives from the polymer is not unusual (8_,2B} . Thus, the
effective value of S when the polymer sample is first immerTed~in the liquid
may differ considerably from the effective value of S at long times.
After a polymer sample immersed in a solvent has reached equilibrium, it
may be removed from the solvent and blotted to remove excess solvent, and then
its weight may be monitored as a function of time. The desorption of the
solvent from the polymer sample as a function of time is also described by
Equation 12, except that MC is now defined as the cumulative weight loss at
time t. As for absorption, the values of D and Ma (proportional to S) may be
determined by fitting the Mg-versus-t data to Equation 12.
As in permeation studies, approximations are often useful in
absorption/desorption studies. One such approximation (52), derived from
26
-------
Equation 12, is Chat the time, tl/2> for which Mt/Ma> = 1/2 is approximately
given by
} (13)
with an error of about 0.001%. Thus,
D = 0.049 42/t (14)
and if the half-time of a sorption process is experimentally determined, the
diffusion coefficient (concentration independent) can be readily calculated.
Unfortunately, polymer swelling and concentration-dependent diffusion
coefficients are common for many polymer/ solvent systems. The application of
Equations 12 -and 14 to absorption/desorption data for nonideal systems yields
an apparent diffusion coefficient averaged over the concentration range
corresponding to the particular experiment. This apparent diffusion
coefficient is often a good approximation to the integral diffusion
coefficient, ^^nt, given by
Dint " (1/C0> /0° Mc (15)
where 0 to CQ is the concentration range existing in the polymer sheet during
the absorption/desorpcion experiment. If the apparent diffusion coefficient is
approximately the same as the integral diffusion coefficient, then it can be
used (along with a value for S) to predict permeation-rate data for exposure
conditions similar to those in the sorption experiment.
The common practice in presenting data from absorption or desorption
experiments is to plot the ratio M^/H^ against the quantity tl/2/£, where MC
is the amount of a given solvent absorbed in or desorbed from a given polymer
sample for a time t from the start of the experiment, M is the equilibrium
weight gain of the polymer in the immersion experiment (related to the measured
solubility of the liquid in the polymer), and i is the thickness of the
polymeric glove sample. The resulting curve is called the reduced absorption or
desorption curve. The initial portion of these reduced curves is normally
linear, that is, the amount absorbed or desorbed is directly proportional to the
square root of time. This is true regardless of the relationship between the
diffusion coefficient and concentration, assuming Fickian behavior (see the
paragraph below) .
As discussed by Fujita (50) and by Crank (^2) , an apparent diffusion
coefficient can be calculated from the initial slope of the reduced absorption
or desorption curve according to the equation:
D = (n/16)l2 (16)
where I is the slope of the initial (linear) portion of the reduced sorption
curve .
27
-------
It should be noted that experimental curves of absorption and desorption
data for a liquid/polymer system can often demonstrate the nature of the
diffusion of the organic chemical in the test polymer (_50,5_2) . If the reduced
absorption-versus-time curve is equivalent to the reduced desorption-versus-time
curve, the diffusion is Fickian and the diffusion coefficient is concentration
independent. If a simple hysteresis is observed, then the diffusion may still
by classified as Fickian, but the diffusion coefficient is concentration
dependent. If the absorption curve shows an inflection point and the absorption
and desorption curves intersect, then the diffusion is non-Fickian or anomalous.
Non-Fickian behavior has been attributed to time-dependent effects on diffusion.
For a given polymer at temperatures above its glass-transition temperature,
Fickian behavior is usually observed, while at temperatures below its
glass-transition temperature, non-Fickian behavior is generally observed.
Numerical Methods
The equations presented above are analytical solutions that may be
obtained if the diffusion processes being studied are ideal. Often one or more
of the assumptions made fail. For example, the diffusion coefficient is
often dependent on the concentration of the organic compound in the polymer.
Equation 2 must then be written as:
3C/3t 3(D(C)3C/3x)/ax (17)
Usually the functional dependence of D on C is not known. For such cases,
numerical methods nay be used to determine the empirical dependence of D on C
from, for example, penneation-rate-versus-time data.
Because even very "simple" deviations from the assumptions given
previously may result in differential equations that cannot be solved
analytically, it becomes necessary to rely on numerical methods. As more
sophisticated predictive models and test methods are developed, there will be
an increasing need to use such techniques.
Applications
As stated previously, the approach to the development of predictive models
and test methods in this work involved the development of rigorous mathematical
expressions (or the use of numerical methods). These expressions may then be
used to make whatever calculations are necessary to demonstrate that a given
glove provides the desired protection against permeation. For example,
Equation 10 may be used to solve for the breakthrough time corresponding to a
given cumulative amount of an organic compound in the receiving fluid in a
closed-loop-mode permeation test. Also, Equation 11 may be used to solve for
the breakthrough time corresponding to a given concentration of an organic
compound in the receiving fluid in the open-loop mode (if the Clow rate of the
receiving fluid is known). Equation 6 may be used to determine the
steady-state permeation rate.
Obviously, if such quantitative data can be determined, then qualitative
permeation ratings can be established based on ranges such as those used by
Edmont in the definition of their permeation ratings.
28
-------
PREDICTIVE MODELS
Very few attempts appear to have been made to predict quantitatively the
permeation of polymeric gloves by organic compounds. Most modeling efforts
have been qualitative and empirical in approach. And they usually involve only
the consideration of solubility and not diffusivity.
Attempts have been made by a number of researchers (JJ-10) to correlate
weight or volume changes in glove materials at a fixed time after the immersion
of test samples in a chemical with normalized breakthrough times. These
correlations have yielded poor results at best. The primary reasons for the
poor correlations observed are:
The. type of experimental data reported in the literature is
generally not sufficient to demonstrate correlations.
Normalized breakthrough times (t-/i2 or t / &2) are based on a
misinterpretation of permeation theory.
Glove materials may lose weight or shrink in volume in contrast
to expectations of weight gains or volume increases on immersion
in an organic liquid.
Equilibrium may not be reached at 24 hr or at any other fixed
time during the immersion test.
The permeation literature includes numerous attempts to correlate
permeation-related observations, such as percentage weight gain on immersion,
with solubility parameters. The solubility parameter for an organic compound
(or polymer) is defined as the square root of its cohesive energy density
(which is the energy of vaporization per cubic centimeter); extensive tables of
solubility parameters exist (for example, Reference 56). Solubility parameters
may be related to solubility by the Flory interaction parameter (see
Section 4). Two compounds (or a compound and a polymer) are thought to be
miscible if there is a close match in their respective solubility parameters.
Also, they are generally considered to be immiscible if their solubility
parameters differ greatly.
As an example, Haxo, Nelson, and Hiedema (57) presented plots of
maximum percentage weight gain on immersion of polymer samples in various
organic liquids as a function of solubility parameter. In their work, they
indeed found examples of organic liquids that resulted in large weight gains
for a given polymer sample and that had about the same solubility parameter as
the polymer. However, they also identified other compounds whose solubility
parameters approximately matched that of the polymer, but there were little or
no weight gains observed for polymer samples immersed in these organic liquids
(see Figure 4 in Reference 57).
As described previously, SRI International (j6) found that correlations
between the permeability coefficient and the separation in solubility space of
Che solvent and the polymer (effectively, the difference in solubility
29
-------
parameters) ranged from good Co poor. Even worse results were obtained in the
current work in attempts to correlate solubility and separation in solubility
space (see Section 5).
Very few researchers concerned with protective clothing have considered
the effect of diffusivity (that is, the diffusion coefficient) on permeation.
However, the diffusion coefficient may strongly affect observed permeation
rates (or parameters such as breakthrough times) because it appears not only as
a multiplier in, for example, Equation 3 but also in the exponential terms. In
addition, the diffusion coefficient is often strongly coupled to the
concentration of the organic compound in the polymer. That is, as the
concentration of the compound in the polymer increases (or decreases), the
diffusion coefficient may change significantly (44,50,51).
Correlations of the type referenced above may be of some value for
screening gross incompatibilities between gloves and organic liquids. However,
if a model is to be generally useful, then it must be capable of quantitative
predictions of data such as permeation rate as a function of time.
Several models were identified during the literature search for estimating
solubilities and diffusivities. Discussions of these models are presented in
Sections 5 and 6 of this report.
PREDICTIVE TEST METHODS
During the first three months of the contract referenced above, written
descriptions of test methods were collected from several sources, including the
American Society for Testing and Materials (ASTM), the International
Organization for Standardization (abbreviated ISO), the British Standards
Institute, and the US Army. The test methods and types of data reported in
journal articles and in reports by other contractors and Government agencies
(for example, Arthur D. Little, Inc.; MIOSH; and SRI International) were also
reviewed critically.
In general, the permeation-test methods in use are satisfactory. However,
they are not universal. That is, extensive analytical methods development is
often necessary when applying a given test method to the measurement of the
permeation of a specific permeant (solvent or other liquid chemical). Thus,
the adaptation of permeation test methods currently in use to evaluate
protective gloves is usually quite expensive, time-consuming, and complex.
For Che reasons given above, the efforts on Task II emphasized the
identification of potential test methods that would be more universal than
those currently in use and that would allow the prediction of permeation rate
versus time. It should be noted that there are currently no test methods that
attempt to use simple experimental procedures to predict permeation rate versus
time.
30
-------
Summary of Test Methods
The existing standard test methods for evaluating chemical protective
clothing were recently compiled in Arthur D. Little's "Guidelines for the
Selection of Chemical Protective Clothing" (4_^5) . This compilation was based
on an earlier list in the ADL report entitled "Development of Performance
Criteria for Protective Clothing Used Against Carcinogenic Liquids" (49). With
the exception of the stress-crazing test and the transparency test, which are
specific for the evaluation of rigid plastics, all of the test methods listed
and discussed in these reports are applicable to the evaluation of protective
gloves. (The fabric and textile tests included in ADL's compilation are
considered applicable to the evaluation of fabric-supported gloves.)
An updated list of standard test methods that includes those given in the
two ADL reports, proposed ASTM test methods, British and ISO test methods, and
the Army's standard test methods for evaluating the resistance of protective
clothing to-permeation by chemical-warfare agents is given in Table 7. All of
the test methods listed in Table 7 except the ISO and British standards were
reviewed. Copies of the ISO and British standards were ordered but were not
received during the contract period. Brief descriptions of all of the methods
reviewed are given in Appendix A.
The test methods listed in Table 7 can be divided into two broad
categories according to the glove properties that are being evaluated:
chemical resistance of the gloves and mechanical properties of the gloves. The
glove properties included in the chemical-resistance category are permeation
resistance, penetration resistance, degradation resistance, and swelling and
solubility. The glove properties included in the mechanical-properties
category are tear resistance and strength, cut resistance, puncture resistance,
abrasion resistance, flexibility, ozone resistance, and UV resistance. (Ozone
resistance and UV resistance are technically chemical-degradation tests, but
they are not grouped in the chemical-resistance category because the tests do
not measure degradation caused by the liquid chemicals that the gloves were
developed to resist.)
Evaluation of Test Methods
Although both categories of tests are necessary, the primary purpose of
chemical-protective gloves is to prevent the exposure of workers to hazardous
liquid chemicals. Thus, the chemical-resistance category was considered more
important to this work than the mechanical-properties category. Within the
chemical-resistance category, the importance of the tests (in descending order)
was considered to be as follows: permeation resistance, penetration resistance,
degradation resistance, and swelling and solubility. This ranking of the tests
is consistent with a recent survey of the members of ASTM Subcommittee F23.30 on
Chemical Resistance that rated the priority of subcommittee objectives and
projects. Listed in descending order, these were: permeation, penetration,
degradation, decontamination, standard chemicals, splash, and particles.
Of the existing or proposed standard methods for evaluating chemical-
protective gloves, the most definitive is the permeation-resistance test. The
permeation resistance of a glove dictates the ultimate choice of the type
31
-------
TABLE 7. STANDARD TEST METHODS FOR THE EVALUATION
OF CHEMICAL-PROTECTIVE MATERIALS
Glove property
Test method
Title
Permeation resistance
ASTM Method F739-81
Penetration resistance
Degradation resistance
ASTM Draft Test
Method F739-8X
(Revision of ASTM
F739-81)
British standard test
Method 33 4724:1971
Draft ISO Method 6529
(Identical to BS 4724)
ISO Method 6530
CRDC-SP-84010
ASTM Draft Test
Method F903
ASTM Draft Test
Method Fxxx
Swelling and solubility ASTM Method D471-79
(continued)
Resistance of protective
clothing materials to
permeation by hazardous
liquid chemicals
Resistance of protective
clothing materials to
permeation by liquids or
gases
Resistance of air-
impermeable clothing
materials to penetration
by harmful liquids
Protective clothing
resistance to penetration
by dangerous liquid
chemicals
Clothing for limited
protection against
dangerous liquid
chemicalsresistance to
penetrat ionmarking
Laboratory methods for
evaluating protective
clothing systems against
chemical agents
Resistance of protective
clothing materials to
penetration by liquids
Test method for evalu-
ating protective clothing
materials for resistance
to degradation by liquid
chemicals
Rubber propertyeffect
of liquids
32
-------
TABLE 7 (continued)
Glove property
Test method
Title
Tear resistance and
strength
Cut resistance
Puncture resistance
Abrasion resistance
Flexibility
Ozone resistance
Ozone resistance
UV resistance
ISO Method 2025
ASTM Method D751-79
ASTM Method D412-83
ASTM Dl682-64
(Fed. 191A-5102)
ASTM D2261-83
ASTM Draft Test
Method Fxxx
ASTM Draft Test
Method Fxxx
ASTM Method D4157-82
(Replaces ASTM
D1175-71)
ASTM Dl388-64
ASTM Method D3041-79
ASTM Method 01149-81
ASTM Method G26-83
(Combination of two
previous methodsG26
and G27)
Lined industrial rubber
boots with general
purpose oil resistance
Standard methods of
testing coated fabrics
Rubber properties in
tension
Breaking load and
elongation of textile
fabrics
Tearing strength of woven
fabrics by the tongue
(single rip) method
(constant-rate-of-
extension tensile testing
machine)
Resistance to cut
Resistance to puncture
Abrasion resistance of
textile fabrics
(oscillatory cylinder
method)
Stiffness of fabrics
Coated fabricsozone
cracking in a chamber
Rubber deterioration
surface ozone cracking in
a chamber (flat specimen)
Operating ligtxp-exposure
apparatus (xenon-arc
type) with and without
water for exposure of
nonraetallic materials
33
-------
of chemical-protective glove material to be used in a. given application.
Penetration resistance is a manufacturing quality-control problem that is
independent of the identity of the solvent or chemical in contact with the
gloves. Resistance to chemical degradation is a materials-compatibility
problem; that is, the selection of a glove material should be made only from a
list of materials that have already been screened for compatibility with the
chemical of interest.
All of the standard permeation tests are based on a partition-cell method
in which the glove material to be evaluated is mounted in a test cell so that
the cell is divided or partitioned into two chambers by the sample. One side of
the test sample is exposed to a Liquid-challenge chemical, and the amount of the
chemical that permeates through the sample into an appropriate collecting fluid
(either gas or liquid) is monitored as a function of time.
ASTM Method F739-81, which is typically cited as the standard permeation-
test method, is rather general in scope. The major emphasis of the method is
the designation of a specific test cell to use in conducting the tests. No
specific analytical methods are recommended in the ASTM method, because the
analytical method must be chosen specifically for each chemical or class of
chemicals to be tested. Typical analytical methods that have been used include
gas, liquid, and ion chromatography; UV and IR spectrophotometry; the use of
radioactive isotopes; and wet chemical methods.
The current draft ASTM Method F739-8X contains two major additions to
Method F739-81modification of the test method to allow gas or vapor
challenges of the teat material and provision for using alternative test cells
that have been found to be equivalent to the ASTM reference cell. (A standard
method for experimentally determining the equivalency of test cells is
currently under development by ASTM Committee F23.) Otherwise, the revised
method is identical to Method F739-81.
Unlike ASTM Method F739-81, Che methods given in CRDC-SP-84010 (58) are
very specific. The methods specify test-cell design, test procedures, and
analytical methods for permeation tests using chemical-warfare agents. Criteria
for interpreting the test results are also given; these criteria can be
specified because both the physicochemical and physiological properties of CW
agents are known and well defined. Thus, the methods given in CRDC-SP-84010 are
not generally applicable to the evaluation of chemical-protective gloves against
a wide variety of organic liquids.
The major disadvantages of the standard permeation tests are the
specificity of the analytical method for the challenge chemical and the
frequent requirement for relatively complex and expensive analytical
instrumentation. A general, relatively inexpensive test method or a
"universal" analytical instrument is desirable for the routine experimental
determination of the resistance of gloves to permeation.
In Section 6 of this report are proposed gravimetric absorptioo/desorption
procedures for determining the solubility, S, and the diffusion coefficient, D,
of a solvent in a glove sample. Data such as the steady-state permeation rate
and lag time (which is inversely proportional to D) as well as the permeation
-------
rate and cumulative permeation as a function of time can be predicted once D and
S are known. Although a method based on such absorption/desorption procedures
may be more time-consuming than a direct permeation test and may require
sophisticated mathematical computation, the method would be general, relatively
simple, and inexpensive in terms of the analytical instrumentation (the only
requirement being a sensitive balance or a calibrated quartz spring and
cathetometer) and technical training required.
35
-------
SECTION 4
GENERAL APPROACH TO THE DEVELOPMENT OF
PREDICTIVE MODELS AND TEST METHODS
As a result of Che literature review and the requirements of the PMN
review process, the decision was made to base the development of models and
test methods on the separate and independent prediction of the diffusion
coefficient, D, and the solubility, S. The coupling of D and S through the
concentration dependence of the diffusion coefficient would be considered as
well. Once D and S are determined, then predictions of quantities such as
breakthrough times, cumulative permeation, and maximum permeation rate can be
based on analytical or numerical solutions of Pick's laws of diffusion.
(Diffusion that does not appear to obey Pick's laws, that is, anomalous
diffusion, is also possible. Such anomalous or non-Fickian diffusion was not
specifically considered in this work.)
The prediction of D and S may be based on theory (for example, the use of
solution thermodynamics to determine S) or on simple test methods capable of
yielding D and S (for example, the use of immersion tests). The development of
predictive models in this work was based on the independent determination of D
and S. Two theoretical methods for the prediction of solubilities were
evaluated, and these methods are described in Section 5 of this report. Two
approaches for the prediction of diffusion coefficients based on free-volume
theory were identified and also evaluated; these approaches are outlined in
Section 6 of this report.
As with theoretical models, the primary emphasis in the selection of test
methods was in the independent (if necessary) determination of estimates of
solubilities and diffusion coefficients. In addition, the test methods
developed emphasized simplicity, low cost, and the desire to use relatively
untrained technical personnel to perform the tests. In conjunction with the
development of predictive test methods, there was a need to conduct some
permeation tests in order to have reliable time-dependent data to verify both
test methods and predictive models.
After the selection of specific approaches to the prediction of D and S,
these approaches were evaluated by comparing predicted D and S values to values
published in the chemical literature.
Once the methods for the prediction of D and S were selected and
evaluated, simple diffusion theory (such as that described in Section 3) was
used to predict permeation-rate-versus-time curves. Then parameters such as
breakthrough times and steady-state permeation rates were determined from these
curves. The results obtained were first compared to manufacturers'
chemical-resistance and degradation-rating tables. These predictions were then
compared to the available experimental data.
In this initial approach, simple diffusion theory was employed. In the
future, as examples of failures of the predictive ability of the models and
36
-------
test methods are identified, the diffusion theory (as well as that used to
predict D and S) can be increased in sophistication as necessary to reduce the
frequency of the failures. The availability of ptfysicochemical data for
organic liquids and polymers must be considered.
Throughout the development of the predictive model, the emphasis was on
the development of algorithms as opposed to "user-friendly" computer programs.
Some programming was necessary to ease, for example, the calculations required
to estimate diffusion coefficients and to make possible calculations that
required numerical methods.
37
-------
SECTION 5
PREDICTIVE MODELS FOR SOLUBILITY
This section of Che report describes two approaches evaluated for the
estimation of the solubility of an organic compound in a polymer. These were
Flory-Huggins theory and UNIFAP theory.
FLORY-HUGGINS'THEORY
Discussion of the Theory
The thermodynamics of solutions is difficult to quantify in many cases-
however, many theories have been developed which accurately predict the
behavior of particular systems. Solution equilibrium depends on the Gibbs
energy of mixing, which consists of an enthalpy term and an entropy term. The
regular-solution theory of Scatchard and Hildebrand (59) has been successfully
used to characterize various nonpolar solutions where the components are of
similar molecular size. This theory assumes that Che Gibbs free energy of
mixing depends only on the enthalpy of mixing, that is, Che entropy of mixing
is zero (assuming no volume change on mixing). In conjunction with the
development of this theory, Hildebrand introduced the concept and the use of
solubility parameters. In reference to polymer solutions, Flory (60) and
Muggins (6±-63) took an alternative approach and initially assumed that Che
enthalpy of mixing is zero for solutions of small molecules and long-chain
molecules; solutions of this type are termed athermal solutions.
The Flory-Huggins equation (derived using statistical mechanics) for an
athermal solution of a solvent and a polymer is
In &1 = In *1 +
-------
If a term to account for the enthalpy of mixing, deviations from complete
randomness of mixing, and other factors is added to Equation 18, the
Flory-Huggins equation may be written as follows:
In 8j a In (frj + (l-Vj/Vj)^ + x*22 <19>
where x is the Flory interaction parameter for the polymer/ solvent system. The
Flory interaction parameter can be related to Hildebrand's solubility
parameters by the following equation:
X (v^RT) (6^63)2 (20)
where R is the universal gas constant (cal/K-mole) , T is the absolute
temperature (K), 6, is the solubility parameter of the solvent [(cal/cm3) l/2] ,
and 5, is the solubility parameter of the polymer [(cal/cm3) i/2] .
Flory-Huggins theory will account for polymer swelling; however, Equation 18
was derived assuming no volume change on mixing, which may not hold for highly
swelling systems.
If we assume that (l-v./\J2) = I for high-mo lecular-weight polymers and
given that *j+*2 = *» Equation 19 becomes:
In a = In * + (1-) + ^1"*2 (21)
Huggins (63) has shown that if X is larger than a critical value given
by:
X,. - (1/2)[1 + Ov/v^)"2]* (22)
then the calculated curve for activity, a^ versus the mole fraction of
polymer, x., (or versus *, or C2' where C? e(?ual3 tne concentration of the
polymer in the solution in moles/cm^) exhibits a minimum and a maximum,
indicating a phase separation. Examination of Equation 22 shows that the
limiting value of Xc " 0.5 for high-molecular-weight polymers. Thus,
for polymer/solvent systems, a phase separation (that is, a finite solubility)
can be expected when X>0.5. (When X<0.5, the solvent and the
polymer will be miscible in all proportions if the theory is applicable. At
values of X=0.5, the behavior will be uncertain.)
For the case when a phase separation occurs, a^sl, and Equation 21 reduces
to:
0 « In $L * (1-^) + x^1'*^2 (23^
The concentration of the solvent, C., in moles/cm3 of unswollen polymer is
related to $1 by the following equation:
39
-------
(24)
Thus, if X is known, *. can be calculated using Equation 23 and an
iceracive computation on a computer. The concentration C , which can be
calculated using Equation 24, will equal the solubility of the solvent in the
polymer.
As discussed above, the Flory interaction parameter, Xi
nonpolar systems can be calculated using Hildebrand's solubility parameters.
A refinement of solubility theory involves the use of three-dimensional
solubility parameters where { is given by
62 6d2 + 6p2 + 6h2 (25)
where 5d is for dispersion forces, 5- is for polar effects, and 5. is for
hydrogen bonding. The use of three-dimensional solubility parameters in the
calculation of x should make Flory-Huggins theory more applicable for polar
systems.
The Flory interaction parameter depends on the difference in
Hildebrand's solubility parameters (Equation 20). However, there is some
disagreement over the calculation of the "difference" in solubility parameters
when three-dimensional solubility parameters are used. For example, Bomberger
and coworkers (.6) reported a so-called "separation in solubility space," A, for
various solvents in polymeric glove materials. They stated that the A values
which they reported were calculated by Henriksen (64) or by themselves using
the following equation:
(26)
where the p and £ superscripts refer respectively to the polymer and the
liquid solvent. Even though Equation 26 is often used (56) to define A, the
equation actually used by Henriksen is:
A' - [4( $-tf) 2 + ( fiP-S*) 2 + ( freft 2] 1/2 (27)
where the factor of 4 is added to provide a spherical interaction volume.
Haasen and Beer bower (65) have suggested another method for calculating
the "difference" in solubility parameters. They state that <5_ and jj, are not
really separable and, therefore, the following equation should be used:
A" = [(<§-$}>* + 0.250(TP-T*)2P'2 (28)
where T - U2 + «g) l/z.
Calculation of Solubilities
Table 8 contains A, A1, and A" values calculated using Equations 26, 27,
and 28, respectively, for a given set of solvents and natural rubber. These
values were calculated using three-dimensional solubility parameters reported
40
-------
TABLE 8. VALUES OF A, A1, AND A" FOR SELECTED
SOLVENTS AND NATURAL RUBBER3
Solvent
Methanol
Ethanol
Zsopropanol
n-Butanol
n-Pentanol
Benzyl alcohol
n-Propanol
Ace Cone
2-Ethyl-l-butanol
t-Pentanol
Diethyl carbonate
Methyl ethyl ketone
Ethyl acetate
n-Propyl acetate
n-Hexane
n-Heptane
Tetralin
Cyclohexane
Cyclohexanone
Toluene
Tetrachloroethylene
Carbon tetrachloride
Trichloroethylene
A, (cal/cn3)l/2
9.1
7.0
5.1
4.7
3.7
3.8
5.6
4.3
3.5
5.2
1.1
3.7
2.1
3.7
4.0
3.9
2.2
3.6
2.3
2.5
3.1
3.4
1.0
A1, (cal/cm3)l/2
9.5
7.3
5.6
5.2
4.3
3.8
6.0
5.0
4.2
6.6
1.9
4.3
3.1
5.2
5.0
4.7
2.4
3.9
2.4
2.5
3.1
3.4
1.1
A", (cal/cn3)l/2
4.7
3.6
2.8
2.6
2.1
1.9
3.0
1.9
2.1
3.1
0.91
1.4
1.3
2.3
2.5
2.4
1.1
1.9
0.35
1.2
0.31
1.7
0.38
aThe separation in solubility space A was calculated using Equation 26.
A' and A" were calculated using Equations 27 and 28, respectively.
41
-------
by Hansen and Beerbower or by Barton ^56). In most cases, the A and A' values
in Table 8 are the same or very close, but significant differences are still
observed for a couple of the solvents. Large variations are observed when the
A" values are compared to the A and A' values.
Equation 20 [written x = ( v./RT)A2] and the A, A1, and A" values
given in Table 8 were used to calculate x> X*> and X* values,
respectively, at 298 K to be used in Equation 23. This latter equation was then
solved for *i, *i, and $1 values from which Cj, Cj, and C\ values were
calculated using Equation 24. The prime superscript indicates that A' was used
in the calculations, and the double-prime superscript indicates that A" was
used in the calculations. The results of these calculations are given in
Table 9. This table also includes solubilities (specified by C ) directly
calculated from experimental volume-fraction data reported by Paul and
coworkers (48).
Comparison of the Cj, C\, and Cj values in Table 9 shows that C] and C\
are generally similar, while c'{ is considerably greater. Comparison of the
calculated values with the literature values can be^made by visual inspection
of Figures 1,2, and 3, which present C\, C\, and C\, respectively, versus
C . The solid line in each plot represents GI equal to C for all possible
vafues of C . The dashed lines define the range of possible calculated
values which rail within one order of magnitude greater than or less than the
experimental values. Figures 1,2, and 3 all have approximately the same
number of calculated values within the dashed lines. The Cj and C} values
are generally smaller than the corresponding(Cex values, while the Cj values
are generally larger. Overall, the GI and GI values are better than the GI
values for the less soluble solvents, but the C^ and Ci values become more
scattered for the more soluble solvents. The Cj values appear to be better
for the soluble solvents. The prediction of miscibility in all proportions or
"infinite solubility" occurs more often for the C± values, indicating that the
c" values could possibly be used to screen out highly soluble solvents when
evaluating natural rubber as a protective-glove material.
An example of the dependence of the estimated solubility on the technique
used to calculate the separation in solubility space is given by the values
listed in Table 9 for the solubility of £-pentanol in natural rubber:
0.0000223, 0.00000110, and 0.000929 moles/cm3. AS another example, the use of
Equations 26 and 27 resulted in relatively low solubilities (0.000699 and
0.000308 moles/cm3, respectively) for methyl ethyl ketone in natural rubber.
However, the use of Equation 28 resulted in the prediction that methyl ethyl
ketone and natural rubber should be miscible in all proportions.
For the entire data set in Table 9, Equation 26 resulted in
solubilities in error by factors from 1.5 to 280 and a linear-regression
correlation coefficient of 0.15 (when compared to the experimental solubilities
42
-------
TABLE 9 SOLUBILITIES CALCULATED FOR VARIOUS SOLVENTS IN
NATURAL RUBBER1 (BASED ON FLORY-HUGCINS THEORY
AND SOLUBILITY PARAMETERS)
Literature solubility Calculated solubility0'
Solvent
Mechanol
Ethanol
Isopropanot
n-Butanol
n-Pentanol
Benzyl alcohol
n-Propanol
Acetone
2-Ethyl-l-butanol
t-Pentanol
Diethyl carbonate
Methyl ethyl ketone
Ethyl acetate
n-Propyl acetate
n-Hexane
n-Heptane
Tetral in
Cyclohexane
Cyclohexanone
Toluene
Tetrachloroethylene
Carbon tetrachlor ide
Trichloroethylene
105 * C.xp
6.92
8.59
52.6
125
130
162
146
169
259
437
636'
713
766
1270
1540
1580
3330
3380
3410
3870
4240
5370
5690
10* x C|
3.21
5.37
18.7
14.3
34.5
35.0
9.33
64.2
28.1
2.23
«
69.9
1400
28.6
8.57
5.71
266
39.7
689
299
136
96.8
-
10* x CJ
1.88
3.21
9.21
6.99
13.7
15.0
5.25
26.7
8.68
0.110
858
30.8
151
1.76
1.15
1.00
157
25.9
629
286
127
88.9
~
10s x Ci
392
396
606
675
724
2630
639
483
92.9
~
--
363
152
135
~
1380
--
~~
'The concentrations (solubilities) reported are in unit* of
moles/cm1 of unswollen polymer at 298 K.
''These values are calculated from the experimental volume-fraction data
given in Reference 48
cThese values are calculated using Flory-Huggins theory. The term C, means
that Equation 26 was used to calculate the separation in solubility
space, C". corresponds to Equation 27, and C'j corresponds to Equation 28.
^The svmbol "" means that no solubility could be calculated, because
these systems are predicted to be oiscible in all proportions.
43
-------
10,000
n
u
1.000
100
10
i i 11111| i i i 11 mi i i i i ni i] I
I I I I I I III I I I I I (III
i i n
i i i 11 ml/1 i i i i mil i i i i lull i i i M ml
10
100 1.000 10.000
10** x Cexp, moles/cm3
Figure 1. Comparison of solubilities calculated using Flory-Huggins theory, Cj. with
experimental solubilities. CexD.
44
-------
10.000
1.000
u
x
U
K
in
o
100
10
0.1
ITMI i n M iin i i i 11 in i i i niu
10
100 1.000
x C. moles/cm3
10.000
6749-3
Figure 2. Comparison of solubilities calculated using Flory-Huggins theory, Cj. with
experimental solubilities, Cexo.
-------
11
10.000
1.000
m
u
N.
M
U
K
100
10
I I I I ITTTT
I 1 I I Mill/ L I I I I III! I I I I II 1(1 I 1 I I I Illl
10
100 1.000
x Cexp. moles/cm^
10.000
1141-4
Figure 3. Comparison of solubilities calculated using Flory-Huggins theory, c'j, with
experimental solubilities, Cexp.
46
-------
included in.Che cable). Equation 27 resulted in solubilities in error by
factors from 1.3 Co 4000 and'a correlation coefficienC of 0.23. Equation 28
resulted in solubilities in error by factors from 1.9 to 80 and a correlation
coefficient of 0.073.
Because of the poor correlation of Che calculated solubilities with che
experimental solubilities, Che development of a method based on Flory-Huggins
theory and solubility parameters for the estimation of solubilities was not
pursued further.
UNIFAP THEORY
Discussion of the Theory
The UNIFAP equations, like Che Flory-Huggins equation, model liquids as
being in a solid-like or quasicrystalline state. In other words, both
approaches are based on Lacti.ce theories. These theories propose chat
molecules of a liquid are fixed in an ordered arrangement and chat Che behavior
of the liquid depends on che molecules' interaccions with their neighbors. If
the molecules show no preference in che selection of their neighbors, Chen a
completely random mixture exists, that is, ideal conditions are present.
However, this rarely occurs, and Guggenheim proposed in his quasichemical
theory of liquid mixtures Co correct for Chis nonideality (66). He described
the behavior of nonrandom systems with equally sized molecules. Flory and
Huggins independently used lattice theory to predict Che behavior of mixtures
whose molecules varied in size but were chemically similar (£6_,J£). Their work
resulted in a relatively simple one-parameter equation. However, because of
Che strong dependence of the Flory inCeraction parameter on composition and
several assumptions made in the derivation of the Flory-Huggins equation, it
does not give a good quantitative description of solubility.
Wilson extended Flory and Huggins1 work by examining the interactions
between different molecules (68). His semiempirical approach is based on che
local composition concept, which assumes that a liquid is not homogeneous on a
molecular level. Therefore, Che energies of interactions are significant and
must be taken into account. Wilson accounted for molecular interactions by
including two adjustable parameters for each pair of molecules in his
equation.
Abrams and Frausnitz's universal quasichemical equation, UNIQUAC
(universal quasichemical activity coefficients), uses local area fractions
instead of local mole fractions as used in the Wilson equation (68). UNIQUAC
provides no major improvements over che Wilson equacion for prediccing the
behavior of completely miscible vapor/liquid systems, but it allows a
prediction of Che behavior of liquid/liquid systems, even liquid/liquid
systems with more than two components. As in the Wilson equation, UNIQUAC uses
only two adjustable parameters for each pair of molecules, and with several
specific assumptions, che UNIQUAC equation reduces to the Flory-Huggins
equacion or any of the well-known equations derived from Guggenheim's work.
-------
Fredenslund et al. (Q,TQ) combined UNIQUAC with the solution-of-groups
concept chat a physical property can be predicted by summing the effects of the
functional groups (that is, the structural units) comprising the compound. The
resulting model, UNIFAC (UNIQUAC Functional-group Activity Coefficients),
contains two adjustable parameters per pair of functional groups instead of
containing two adjustable parameters per pair of molecules as in UNIQUAC.
Compared to UNIQUAC, this significantly reduces the number of interaction
parameters required to apply the model, because the number of possible
funccional groups is much smaller than the number of existing compounds.
Furthermore, this allows a quantitative description of the behavior of systems
for which no experimental data are available but that contain functional groups
whose energies of interaction have been experimentally determined. However,
this extrapolation technique holds only if the behavior of any given group is
not affected by the presence of the other groups within the molecule. Often
this assumption is not true; thus, UNIFAC is an approximate method.
As in the Flory-Huggins equation, the UNIFAC equation used to calculate
the liquid-phase activity coefficient for a given component consists of an
entropy term or a combinatorial part, resulting from the differences in the
size and shape of the functional groups in the mixture, and an enthalpy term or
residual part, resulting from the interaction energies between the functional
groups.
The combinatorial part of the UNIFAC equation is calculated from the
functional-group parameters known as normalized van der Waals group volumes and
interaction surface areas. These parameters are determined independently from
pure component, atomic and molecular structure data (70).
The residual part of the UNIFAC equation depends on the "concentrations"
of the functional groups and the interactions between the groups. Therefore,
the solution-of-groups concept plays an important part in calculating this
term. In addition, the residual term resembles the Wilson equation written in
terms of area and segment fractions.
The group-interaction parameters, a , in the residual part of the model
characterize the energy of interactions between the functional groups n and m.
They have the units of Kelvin, and for interactions between a given pair of
functional groups, there exist two distinct parameters (that is, annj*amn ^'
These parameters are determined empirically from experimental vapor/liquid
equilibrium data. An extensive compilation of such data are available on
magnetic tape from the University of Dortmund (Dortmund, West Germany); the
Dortmund data base contains vapor/liquid equilibrium data for systems meeting
the following requirements (70): the pressure is less than IS atm, and the
components only consist of water or organic compounds with a normal boiling
point higher than 273 K. Fortunately, tables of group-interaction parameters
are available in the literature (71-73).
As more phase-equilibrium data for vapor/liquid systems become available,
it is possible to estimate previously missing interaction parameters and to
improve the estimation of the parameters that were based on very limited data
(71).
48
-------
The UNIFAC model has two significant advantages: It is very simple
because the interaction parameters are not very strong functions of temperature
and pressure within the range of applicability; it is also very flexible and,
thus, very easily applied to a large number of systems because UNIFAC
parameters are available for a large number of different functional groups.
Overall, UNIFAC has been proven to predict satisfactorily activity coeffi-
cients in a large number of systems. However, UNIFAC does have limitations
(74). It can only be applied when:
The pressure is no more than a few atmospheres.
All components are well below their critical points.
The temperature of interest is in the range of.300 to 425 K
(80 to 300 *F).
No noncondensables or electrolytes are contained in the system.
No immiscible liquids are contained in the system.
No polymers are contained in the system.
Only components that contain ten or less different UNIFAC
functional groups are present.
Oishi and Prausnitz (75) extended UNIFAC to the calculation of solvent
activities in polymer solutions, and they referred to the modified theory as
UNIFAP (UNIFAC for polymer systems). They began by working in terms of
activities rather than activity coefficients; they felt mole fractions were
"awkward units" of concentration for polymer solutions because of the much
larger molecular weight of the polymer versus the solvent. In addition, they
added another contribution term to the model to take into account the changes
in free volume caused by mixing the solvent and the polymer. Thus, the
activity of the solvent is determined by
In 3j In a^ + In a f + In ajfv (29)
where a.c is the combinatorial activity (an entropy term), a .r is the residual
_-«_^/ ^ t \ _J_ f V J ^L _ f __* - ^ .*-_ - ^
1
activity (an enthalpy tern), and a. is the free-volume activity.
The equations necessary to apply the UNIFAP model are tedious. An
excellent summary of the equations and an illustrative calculation are given in
Reference 75. A copy of the UNIFAP software package written by Oishi and
Prausnitz, which is based on the equations given in Reference 75, is available
from the Friends of Chemical Engineering (University of California, Berkeley,
CA).
In general, the UNIFAP (UNIFAC for polymer systems) model has been proven
to predict activities satisfactorily for polymer/solvent systems (75,76).
However, as in the case of UNIFAC, the independence of the interaction
parameters to changes in temperature and to the effects of other functional
groups contained in the system may not always be a valid assumption. Thus,
49
-------
accounting for such effects in the determination of the group-interaction
parameters would greatly improve the reliability of UNIFAP in predicting
solubility (76) .
Calculation of Solubilities
The UNIFAP software obtained from the Friends of Chemical Engineering
yields activities at specific volume fractions. The UNIFAP software was
modified slightly to determine activities for various solvent volume
fractions. The modified software allows the determination of the volume
fraction of the solvent that yields an activity of one (that is, a =1).
Therefore, the solubility of the solvent in a polymer can be determined.
Using the modified UNIFAP program,- solubilities, Cuni, for various
solvents in natural rubber at 298 K were calculated. In general, the
temperature of the system, the densities of the solvent and the polymer, and
the functional groups comprising the solvent and the polymer were the only
required inputs of this method. From these inputs, a solvent volume fraction
was generated at the saturation condition (that is, for a^l). Finally, this
volume fraction, «. , was converted to solubility, Cun£ , in units of moles/cm 3
of unswollen polymer by using the following equation:
where vi is the molar volume of the solvent.
The data used in the UNIFAP calculations of solubilities are listed in
Table 10. The UNIFAP results and experimental solubilities, C , reported by
Paul et al. (48) are given in Table 11. In addition, the calculated values and
the experimental values are compared graphically in Figure 4. Several types of
regressions were performed on the data, but a linear regression provided the
best fit with a correlation coefficient, r, equal to 0.92. The equation given
in Figure 4 indicates that solubilities calculated using the UNIFAP model are
generally less than experimental values by about a factor of two (except for
very low solubilities and solubilities that indicate miscibility in all
proportions). Future work using UNIFAP should address this systematic error.
As shown by Che results in Table 11, the equilibrium solubilities for some
polymer/solvent combinations (for example, acetone or ji-pentanol in natural
rubber) can be described accurately by the UNIFAP modeT. However, solubili-
ties for many of the systems given can only be determined within a factor of
five. This error probably results from inaccuracies in the group-interaction
parameters in the data base used in performing the calculations or the
breakdown of some of the assumptions on which UNIFAP is based.
50
-------
TABLE 10. SOLVENT DATA USED IN UNIFAP CALCULATIONS
Molecular
Solvent weight, g/mole Density,3 g/ctn
Acetone
Benzyl alcohol
£-Butanol
_t-Butanol
Carbon tetrachloride
Cyclohexane
Cyclohexanone
Diethyl carbonate
Ethanol
Ethyl acetate
2-Ethyl-l-butanol
n- Heptane
n-Hexane
Isopropanol
Methanol
Methyl ethyl ketone
j»-Pentanol
_t-Pentanol
jn-Propanol
_n-Propyl acetate
Tetrachloroethylene
58.08
108.1
74.12
74.12
153.8
84.16
98.14
118.3
46.06
88.10
102.2
100.2
86.17
60.09
32.04
72.10
88.15
88.15
60.09
102.1
165.9
0.788
1.05
0.810
0.786
1.59
0.778
0.948
0.976
0.816
0.902
0.833
0.634
0.660
0.785
0.792
0.805
0.815
0.808
0.805
0.836
1.62
(continued)
3 Functional groups'9
1 CH3, 1 CH3CO
5 ACH, 1 ACCH2, 1 OH
1 CH3, 1 CH2, 1 OH
3 CH3, 1 C, 1 OH
1 CCl^
u wEl M
5 CH2, 1 CH2CO
2 CH3, 1 CH20, 1 CH2COO
1 CH3, 1 CH2, 1 OH
1 CH3, 1 CH2, 1 CHjCOO
2*CH3, 3 CH2, 1 CH, 1 OH
2fu e ru
OJlj, 3 *'tl2
7 fu i, rv
L l-Mj, 1 l,H2
2 CH3, 1 CH, 1 OH
1 CH3OH
Ipu i pu i pu p/\
~nji » 2 ' u'ji'U
1 CH3, 4 CH2, 1 OH
3 CH3, 1 CH2, 1 C, 1 OH
1 CH3, 2 CH2, 1 OH
1 CH3, 2 CH2, 1 CH3COO
1 0»C, 4 Cl(OC)
51
-------
TABLE 10 (continued)
Solvent
Tetralin
Toluene
Trichloroethylene
Molecular
weight, g/mole
132.2
92.13
131.4
Density,8 g/cm3
0.970
0.866
1.46
Functional
4 CH2, 4
5 ACH, 1
1 CH-C, 3
groups
ACH, 2 AC
ACCH3
Cl(C-C)
aThese are the densities at 298 K.
The symbol AC designates an aromatic carbon.
52
-------
TABLE 11. SOLUBILITIES CALCULATED FOR VARIOUS SOLVENTS IN NATURAL
RUBBERa'b (BASED ON "VAPOR/LIQUID" UNIFAP THEORY)
Solvent
Methanol
Ethanol
Isopropanol
jn-Butanol
n-Pentanol
Benzyl alcohol
ti-Propanol
Acetone
2-Ethyl-l-butanol
t-Butanol
t-Pentanol
Diethyl carbonate
Methyl ethyl ketone
Ethyl acetate
n-Propyl acetate
n-Hexane
n- Heptane
Tetralin
105 x Cun£ moles/cm3
26.5
42.4
81.6
93.3
103
27.5
100
185
114
81.5
96.9
202
271
411
581
*
*
*
105 x C moles/cm3
cxp y
4.92
8.59
52.8
125
130
142
146
169
259
414
437
636
713
766
1270
1540
1580
3330
(continued)
53
-------
TABLE 11 (continued)
Solvent 10s x Cuni moles/cm3
Cyclohexane *
Cyclohexanone *
Toluene *
TetrachLoroethylene *
Carbon Cetrachloride *
Trichloroethylene *
10s x C___ moles/cn>3
CAt* t
3380
3410
3870
4240
5370
5690
aThe UNIFAP calculations of the solubilities at 298 K were performed using
group-interaction parameters estimated by fitting vapor/liquid phase
equilibrium data to the UNIFAC equations.
The symbol "*" indicates that an activity of one was achieved only for a
solvent volume fraction equal to one. Thus, the polymer and the solvent
are predicted to be miscible in all proportions.
54
-------
o
o
X
600
500
400
300
200
100
1 1
i i
r = 0.92
Cuni = 0.39 Cexp+ 21.72
s
-
r
i i
I I
200 400 600 800 1000 1200 1400
x Ce«. moles/cm^
5 74 9-1
Figure 4. Comparison of solubilities calculated using vapor/liquid UNIFAP, Cuni,
with experimental solubilities, Cexp.
55
-------
1C should be noted that Che error in Che UNIFAP resales given here is most
significant in comparison Co very small and very large experimental
solubilities. In fact, aC extremely large solubilities, an accivity of one
(which is indicacive of reaching equilibrium and, Chus, defines the solubility)
is achieved only for a solvenC volume fraccion equal Co one. Thus, at large
solvent concentrations, UNIFAP predicts a greater solubility than is actually
observed experimentally. This phenomenon is expected because Che UNIFAP
calculations do noC Cake into account the amount of crosslinking in a polymer,
which may limit the swelling of the polymer. Rather than being a limitation,
if the UNIFAP model yields an activity of one only at a solvent volume fraction
of one, Chen the solvent and the polymer may be miscible in all proportions;
that is, only one phase may exist. Therefore, Che polymer probably would not
be suitable as a protective barrier material for the solvent. (Nonetheless,
future modifications of UNIFAP theory should consider the degree of
crosslinking.)
As discussed above, the group-interaction parameters are assumed Co be
independent of CemperaCure and independent of Che other groups within the given
molecule. However, these group-interaction parameters may not be constant, and
determinations of the dependence of Chese parameters on composition, pressure,
and Cemperature need to be made (76). Furthermore, the data base currently
being used is a table of group-interaction parameters obtained by fitting
experimental vapor/liquid phase-equilibrium data to the UNIFAC equations.
Theoretically, the parameters and equations used for the prediction of a
vapor/liquid system's behavior at equilibrium also can be used for the
prediction of a liquid/liquid system's behavior at equilibrium (TT) And it
has been shown that group-interaction parameters determined from vapor/liquid
equilibrium data typically give a deviation of approximately 9 moleX (78) when
used to determine equilibrium in liquid/liquid systems. However, using a
UNIFAC interaction-parameter Cable obCained from liquid/liquid equilibrium data
rather than vapor/liquid data would be expected to yield better results.
Magnussen, Rasnussen, and Fredenslund (78) presented a data base of UNIFAC
group-interaction parameters for predicting liquid/liquid equilibrium behavior
in 1981. These parameters were estimated by fitting experimental liquid/liquid
equilibrium data measured between 283 and 313 K to the UNIFAC equations. A
data base for the UNIFAP software that included these group-interaction parame-
ters was set up. This data base was a combination of the previously listed
vapor/liquid interaction parameters and the "new" liquid/liquid interaction
parameters. This combination was necessary because only a limited set of
reliable experimental phase-equilibrium data exists for liquid/liquid systems.
If only this small liquid/liquid data base were used Co calculate group-
interaction parameters, then there would be an insufficient set of them. That
is, prediccions could be made for only a limited number of polymer/solvent
systems.
Solubilicies calculated for several polymer/solvent systems using this
combined data base and the experimental solubilities, C , reported by
Paul et al. (48) are compared in Table 12. Several solubility predictions for
individual systems were in closer agreement using liquid/liquid interaction
parameters than when using vapor/liquid interaction parameters. F«r example,
Paul experimentally determined a solubility equal to 125 x 10~"5 moles/cm3 for
56
-------
TABLE 12. SOLUBILITIES CALCULATED FOR VARIOUS SOLVENTS IN NATURAL
RUBBER*>b (BASED ON "LIQUID/LIQUID" UNIFAP THEORY)
Solvent
Methanol
Ethanol
Isopropanol
n-Butanol
n-Pencanol
Benzyl alcohol
n-Propanol
Acetone
2-Ethyl-l-butanol
t-Butanol
t-Pentanol
Diethyl carbonate
Methyl ethyl ketone
Ethyl acetate
n-Propyl acetate
n-Hexane
n-Heptane
Tetralin
10 5 x Cun£ moles/cm3
30.8
58.8
110
124
135
101
138
76.3
138
105
197
93.0
101
*
*
*
*
*
105 x C___ moles/cm3
exp ,
4.92
8.59
52.8
125
130
142
146
169
259
414
437
636
713
766
1270
1540
1580
3330
(continued)
57
-------
TABLE 12 (continued)
Solvent 10s x Cuni> moles/cm^
Cyclohexane *
Cyclohexanone *
Toluene *
Tetrachloroethylene 259
Carbon tetrachlaride 128
Trichloroethylene 310
10 5 x C___ moles/cm3
exp ,
3380
3410
3870
4240
5370
5690
aThe UNIFAP calculations of the solubilities at 298 K were performed using
group-interaction parameters estimated by fitting vapor/liquid and
liquid/liquid phase-equilibrium data to the UNIFAC equations.
bThe symbol "*" indicates that an activity of one was achieved only for a
solvent volume fraction equal to one. Thus, the polymer and the solvent
are predicted to be miscible in all proportions.
58
-------
n-butanol in natural rubber. Using UNIFAP, solubilities were found to be
724 x 10~5 moles/cm3 and 93.3 x 10~5 moles/cm3 using liquid/liquid and vapor/
liquid interaction parameters, respectively. However, using liquid/liquid
parameters to calculate solubilities, several solvents previously determined Co
be totally miscible with natural rubber (using vapor/liquid parameters) were
predicted to show phase separation (that is, an activity of one was obtained at
a solvent volume fraction leas than one). In addition, some of the predicted
solubilities were low by factors as large as 42 in comparison to experimental
solubilities. Also, several types of regressions were performed on the
comparison of the calculated values and the experimental values, but no satis-
factory correlation resulted. Therefore, it is suggested that the vapor/liquid
interaction-parameter data base instead of the liquid/liquid parameter data
base be used to calculate solubilities until more thermodynamically consistent
liquid/liquid equilibrium data are available.
It should be noted that the selection of functional groups comprising a
compound is sometimes somewhat arbitrary because the set of functional groups
for which interaction parameters are available is limited. Therefore, a "best
guess" of the functional groups used to construct the compound oust be made.
It may be possible to construct a given compound from two or more "best-guess"
sets of functional groups. And the two sets of functional groups may yield
very different predicted solubilities using UNIFAP. For example, using
liquid/liquid interaction parameters and the functional groups 2 CH3, 1 Ct^O,
and 1 CH2COO for diethy1 carbonate generated an activity of one only for a
solvent volume fraction equal to one; that is, no finite solubility could be
predicted. However, using liquid/liquid interaction parameters and the func-
tional groups 2 CH,, 1 CH2, 1 CH20, and 1 COO for diethyl carbonate generated a
UNIFAP solubility equal to 93.0 x 10~5 moles/cm3 of unswoilen polymer. To
eliminate such "arbitrary" choices of functional groups, the set of functional
groups for which interaction parameters are available needs to be expanded.
As previously discussed, the UNIFAP model generally provides a good
estimate of solubilities for many polymer/solvent systems. With improvements
in estimating the values of the group-interaction parameters and in
standardizing the process of selection (or definition) of the functional groups
comprising the compounds, the reliability of this method will probably improve
significantly.
The UNIFAP software also needs to be extended to predict the behavior of
ternary and other larger systems to account for the presence of plasticizers
and various other additives in glove formulations. (Because the UNIFAP model
works with only functional groups, the model can describe the behavior of
ternary or larger systems; however, the software to perform calculations for
such systems has not yet been developed.) Furthermore, the UNIFAP model needs
to take into account the possibility of the polymer being crosslinked.
Currently, the model is limited to only uncrosslinked polymers. (As stated
previously, this limitation accounts for the predictions of total miscibility
given in Table 11 for some of the solvents in crosslinked natural rubber even
though experimental solubilities less than "infinity" have been reported.)
59
-------
SECTION 6
PREDICTIVE MODELS FOR THE DIFFUSION COEFFICIENT
Predicting Che permeation race of a solvent through a polymeric membrane
involves solving Che governing mass transport equations. If the simple
diffusion theory presented in Section 3 is applicable, it is necessary only
to solve Pick's first and second laws with the appropriate boundary and initial
conditions. An important term appearing in Pick's laws is Che diffusion
coefficient. Numerous theories and correlations are reported in the literature
for predicting the diffusion coefficient, D, of various substances under
various conditions (7j). For solvent diffusion in polymers, however, no single
theory has yet been universally accepted. One of the more widely used
approaches to predicting diffusion coefficients is based on free-volume
theory.
Free-volume theory depends on a concept in which solvent diffusion is
related to "holes" or "free volume" which exists throughout the polymer bulk.
Free volume is defined as the volume in a polymer bulk not occupied by polymer
molecules themselves. Because of random molecular notion, individual
free-volume elements are constantly being collapsed and recreated, but the
total free volume does not change as a result of this motion. If it is assumed
that solvent diffusion occurs due to the random notion of solvent molecules
through the free volume, then relationships can be derived to predict the
diffusion coefficient. Also, because the solvent molecules will occupy the
free volume as they diffuse into the polymer, free-volume theories can
generally account for concentration-dependent diffusion coefficients. This
capability is especially important in polymer/solvent systems.
Two diffusion theories based on free-volume concepts are discussed below.
The two do not constitute the entire set of free-volume theories, but are meant
to be representative of them. The first of the two theories discussed was
developed by Vrentas and Duda (80-82) and is more complex than the second,
which was developed by Paul (£377 Because of the relative simplicity of the
Paul model, its application was emphasized in the current work, the goal of
which was to demonstrate the feasibility of the development of predictive
models.
VRENTAS-DUDA MODEL
The Vrentas-Duda model can predict diffusion coefficients over a majority
of the possible solvent volume-fraction range. Because the model uses
specific information about the polymer/solvent system at small solvent volume
fractions, it is especially accurate at very small solvent volume fractions
(* <0.1). At very large volume fractions {* >0.9), free-volume concepts are no
longer valid and, therefore, the Vrentas-Duda model is inaccurate. At
intermediate volume fractions, the model is able to predict diffusion
coefficients fairly accurately.
60
-------
A major drawback to the Vrentas-Duda model, however, is the large amount
of experimental data necessary to compute the diffusion coefficient, D. The
data needed include:
Ratio of the number of surface sites for the polymer segments Co
the number of surface sites for the solvent. (This can be
estimated from molecular dimensions.)
Energy interchange parameter. (This can be derived from
enthalpy of dilution data for the mixture.)
Entropy interchange parameter. (This can be derived from
reduced, residual chemical-potential data near zero polymer
segment fraction. The reduced, residual chemical-potential data
are derived from chemical-potential-versus-teraperature data.
Polymer segment fraction refers to a parameter defined using
specific volume and other characteristic data.)
Viscosity of the solvent as a function of temperature.
Viscosity of the polymer as a function of temperature and
molecular weight.
Specific volume of the solvent and the polymer at 0 K.
Polymer glass-transition temperature as a function of molecular
weight.
Diffusion coefficient of the solvent in the polymer near zero
solvent concentration.
Solvent and polymer molecular weights.
A very complex series of calculations is required to estimate D using the
Vrentas-Duda model. Because of the rather extensive set of required data, some
of which may often be difficult to obtain, and because of the complexity of the
calculations, the Vrentas-Duda model was considered less satisfactory than the
Paul model for initial attempts to predict D.
PAUL MODEL
The Paul model (83) for predicting diffusion coefficients is also based on
free-volume theory. One advantage of this model over the Vrentas-Duda
model, however, is the relatively small amount of data needed to calculate the
diffusion coefficient. These data include:
Viscosity of the solvent as a function of temperature.
61
-------
Density (or specific volume) of Che solvent as a function of
temperature.
Density (or specific volume) of the solvent and the polymer at
the temperature of interest and at absolute zero (0 K).
Critical volume of the solvent (see Appendix B).
Molecular weight of the solvent and the polymer.
Solvent chemical potential as a function of solvent volume
fraction. [These data can be derived from UNIFAP data (see
Appendix B)].
Given the data listed above and the following assumptions (which represent a
modification of Paul's model), diffusion coefficients can be calculated.
The average solvent molecular velocity is proportional to T1/2.
The free volume is randomly distributed among all units of mass
in the solution.
The only volume which is not freely distributed is the molecular
volume and the interstitial volume associated with a random
packing of the molecules. (The molecular volume and the
associated interstitial volume is approximated by the volume at
0 K.)
Polymer segments have a negligible chance of refilling a void
space compared to a solvent molecule. (This assumption makes
the Paul model less accurate for solvent volume fractions less
than 0.1).
The polymer self-diffusion coefficient is negligible.
The excess volume of mixing is zero at all concentrations.
The polymer is nonglassy and, thus, above its glass transition
temperature.
A detailed derivation of the modified Paul model is given in Appendix B.
A computer program written to perform the necessary calculations is also
included in this appendix.
The concentration-dependence of the diffusion coefficient for benzene
in natural rubber at 298 K, calculated using the modified Paul model-, is
illustrated in Figure 5. Similar data for n-heptane in natural rubber at 298 K
are shown in Figure 6. The input data used to perform these calculations are
given i Appendix B. (The UNIFAP calculations that were necessary used the
vapor/liquid data base.)
62
-------
0.75
0.65
0.55
J
0.45
O
x
in
S 0.35
U
ui
O
(J
S 0.25
C0
0.15
0.05
0.05
APPROXIMATE PREDICTIVE
RANGE: 0.1<4>i <0.9
PREDICTED USING
THE PAUL MODEL
s\ \ -
7 EXPERIMENTAL DATA
*S [HAYES AND PARK (15)1 \
\~\
0.00 0.20 0.40 0.60 0.80 1.00
BENZENE VOLUME FRACTION, <*>, 6749-16
Figure 5. Diffusion coefficient of benzene in natural rubber as a function of volume fraction.
63
-------
0.80
0.70
0.60
N
- 0.50
Q
x
1 0.40
u
o
u
2
S 0.30
C/5
0.20
0.10
0.00
0.00
APPROXIMATE PREDICTIVE
RANGE: 0.1 <<* <0.9
PREDICTED USING
THE PAUL MODEL
0.20 0.40 0.60
jvHEPTANE VOLUME FRACTION.
0.80 1.00
4749-1T
Figure 6. Diffusion coefficient of n_-heptane in natural rubber as a function of volume fraction
-------
Figure 5 also shows an experimental curve reported by Hayes and Park (51)
for benzene in natural rubber at 298 K. The maximum value of D reported is
close to the value predicted by the modified Paul model. However, the volume
fraction at which the maximum occurs is not in agreement; the theoretical curve
appears to be shifted to the left. No plausible explanation has yet been
developed for this disagreement.
D.R. Paul (45) reported experimental diffusion coefficients for benzene
and ji-heptane in natural rubber of 2.3? x 10"6 cm2/sec and 3.07 x 1C"6 cm2/secf
respectively, at 303 K; the diffusion coefficient for benzene was determined at
a volume fraction of 0.8, and the diffusion coefficient for n-heptane was
determined at a volume fraction of 0.7. The modified Paul model predicts that
the diffusion coefficients are 2.72 x 10~8 cm2/sec for benzene and
2.71 x 10~£ cm2/sec for n-heptane at volume fraction of 0.8 and 0.7, respec-
tively. Comparison of D?R. Paul's reported diffusion coefficients with those
calculated using the modified Paul model suggests that the model has only
limited applicability. However, much more extensive evaluations of the Paul
model must be made before discarding it in favor of more sophisticated models.
The predicted D-versus-*. curves in Figures 5 and 6 exhibit a maximum, a
phenomenon observed experimentally by others (5^,j[l ,]J4^8_5). The curves also
suggest that the solvent diffusion coefficient approaches zero as the solvent
volume fraction approaches one. Because free-volume theory requires a consi-
derable amount of polymer-polymer contact, this result is not surprising.
Below a minimum polymer volume fraction, the concept of free-volume within a
polymer bulk is no longer valid (Bl). For small polymer volume fractions,
theories for dilute and infinitely dilute polymer concentrations are valid.
Berry and Fox (86) report that the minimum polymer mass fraction, which may be
related to volume" fraction, for which free-volume calculations are acceptable
is given by:
w2 - 4/(l + 0.2 MW21'2) (31)
where HV. is the molecular weight of the polymer. For natural rubber, the
polymer molecular weight is approximately 68,100, assuming a degree of polymer-
ization of 1000. Thus, the minimum polymer mass fraction for which free-volume
calculations are valid for natural rubber is about 0.08.
On first examination, use of the modified Paul model appears to be a
limited approach to predicting diffusion coefficients. However, because only
benzene/natural-rubber predictions were checked with D-versus-* experimental
data, no firm conclusions can yet be drawn. Future work should include
extensive checking of the model against experimental data. Because of the
relatively small amount of experimental data reported in the literature, the
thorough confirmation of the model will be difficult without experimentally
'determining diffusion-coefficient data for a series of solvents in a variety of
polymers. An alternative is to use diffusion-coefficient data calculated with
the Paul model and solubility data to calculate permeation-rate-versus-time
65
-------
curves (see Section 7). By comparing these data to experimental permeation-
rate curves, the Paul model could be indirectly confirmed or refuted. Criteria
such as defined breakthrough times, lag times, and steady-state permeation
rates calculated from the predicted permeation-rate curves could also be
compared to experimental data to indirectly confirm the modified Paul model.
66
-------
SECTION 7
PREDICTIVE TEST METHODS FOR SOLUBILITY AND
THE DIFFUSION COEFFICIENT
Under Task II of the contract, two potential predictive test methods that
could be used to estimate the solubility and diffusion coefficient of an
organic liquid in a polymeric glove material were investigated. Permeation
tests for selected glove/solvent combinations were also conducted to generate
data that could be used to demonstrate the validity of the predictive methods
studied.
SELECTION OF PREDICTIVE TEST METHODS
As discussed in Section 3 of this report, the most accurate, precise, and
reliable method for "predicting" the permeation rate of an organic liquid
through a glove material as a function of time is a direct permeation test,
such as ASTM Method F739. However, the apparatus that is needed to perform
permeation tests may be costly and conceptually complex; thus, such tests must
usually be performed by well-trained chemical professionals. The major
components of a permeation-test system include a permeation test cell; a gas or
liquid stream or reservoir to collect the solvent after it permeates through
the test sample; a sophisticated, sensitive, and usually expensive analytical
instrument for quantifying the solvent, and possibly a temperature-control
system. For these reasons, it is desirable to develop alternative predictive
test methods that are simpler and less expensive than direct permeation tests
and that can be performed by personnel with less technical training than is
normally required for permeation testing.
The first step in the selection of alternative methods was a review of
the existing standard test methods for evaluating chemical-protective clothing.
The test methods reviewed were described in Section 3 and in Appendix A of this
report. As stated previously, these methods can be divided into two broad
categories: chemical-resistance tests and mechanical-properties tests. The
glove properties included in the chemical-resistance category are permeation
resistance, penetration resistance, degradation resistance, and swelling and
solubility. The glove properties included in the mechanical-properties
category were tear resistance and strength, cut resistance, puncture
resistance, abrasion resistance, flexibility, ozone resistance, and UV
resistance. However, no existing standard test method other than the direct
permeation test was identified that would yield solubilities and diffusion
coefficients of organic liquids in polymeric glove materials.
A review of experimental methods described in the scientific literature,
however, revealed that absorption and desorption methods have been developed
and used extensively for many years by academic and industrial researchers to
study the diffusion of permanent gases and organic vapors in polymer films
(see, for example, Reference 50). In addition, exact analytical solutions of
Fick's laws that describe absorption and desorption have been developed. On
67
-------
the basis of this experience, two test methods for evaluation and development
were selected; these were a saturated-vapor absorption/desorption method and a
liquid-immersion absorption/desorption method. Each of the methods is a
gravimetric procedure that was expected to yield both the solubility, S, and
the diffusion coefficient, D, of a solvent in a polymeric glove sample in a
single experiment. Although these methods may be more time-consuming than
direct permeation tests and may require relatively sophisticated mathematical
computation, they are experimentally quite simple. Thus, the analytical
instrumentation required is relatively inexpensive and unsophisticated, and
personnel with only a minimum of technical training can conduct the tests.
IMMERSION ABSORPTION/DESORPTION TESTS
Description of the Test
In the immersion test, the weight of an organic liquid absorbed by a
polymer sample totally immersed in the liquid was measured as a function of
time with an analytical balance. To begin the test, the dimensions of a
polymeric glove sample were measured, and the sample was weighed accurately and
immersed in the organic liquid. After a specified time, the sample was removed
from the liquid, the excess liquid was blotted from the exposed surfaces of the
glove sample, and the sample was weighed. The sample was then reimmersed in
the liquid and reweighed following the same procedure at specified intervals.
The weighing and reimmersion procedure was repeated until a constant weight was
attained. After equilibrium absorption or saturation was reached, the
cumulative desorption of the organic chemical from the glove material was
determined by monitoring the weight loss of the saturated sample as a function
of time with an analytical balance.
Determination of Solubilities and Diffusion Coefficients
The solubility, S, of the organic liquid in the polymeric glove material
was determined by dividing the equilibrium weight gain of the glove sample
during the immersion test by its unswollen volume. The diffusion coefficient,
D, was estimated from the time-dependent absorption and desorption data by an
approximate method described by Crank (52) and by numerically fitting
Equation 12 to the entire absorption or desorption curve. These two methods
are described briefly below. (A more extensive discussion of these
data-reduction techniques was presented in Section 2 of this report.)
Direct Curve Fit
The absorption or desorption of a substance into or from a planar sample
(if edge effects are negligible) can be described by the equation:
M../M = 1 - (8/ir2) 7 (2m+l)~2 exp(-D(2m+l)2n2t/£2) (12)
t
-------
(effectively, after equilibrium has been reached), D is the diffusion coeffi-
cient (cm2/sec), and I is the thickness of the sheet (cm). This equation is
applicable to the liquid immersion test if the diffusion is effectively one
dimensional (thin samples) and the diffusion coefficient is concentration
independent. Experimental time-dependent absorption or desorption data can be
fit, with the aid of a computer, to Equation 12 using nonlinear curve-fitting
techniques to yield the diffusion coefficient.
Initial Rates of Absorption and Desorption
As stated previously in Section 3, the common practice (50) in presenting
data from absorption or desorption experiments is to plot the ratio M^/M^
against the quantity tl/2/i, where MC is the cumulative amount of a given
solvent absorbed in or desorbed from a given polymer sample at time t from the
start of the absorption or desorption experiment. M^ is th'e equilibrium weight
gain of the polymer in the immersion experiment (related to the solubility of
the liquid in the polymer), and t is the thickness of the original, unswollen
polymer sample. The resulting curve is called the reduced absorption or
desorption curve.
As discussed by Fujita (.50) and by Crank (£2), an apparent diffusion
coefficient can be calculated from the initial slope of the reduced absorption
or desorption curve according to the equation:
D - (ir/16)l2 (16)
where D is the apparent diffusion coefficient and I is the slope of the initial
(linear) portion of the reduced sorption curve. In this report, D and I are
denoted D and Ia, respectively, for an absorption experiment and Dd and Id,
respectively, for a desorption experiment.
Test Samples
Liquid-immersion absorption and desorption tests were conducted using five
protective-glove materials and four solvents. The glove materials used were
butyl rubber, natural rubber, neoprene rubber, nitrile rubber, and poly(vinyl
chloride). All of the gloves were unsupported. The solvents used were
acetone, cyclohexane, isopropanol, and toluene. The natural-, neoprene-, and
nitrile-rubber gloves were manufactured by the Edmont Company and were obtained
from a local supplier. The PVC gloves were manufactured by the Pioneer
Company, and the butyl-rubber gloves were made by the Norton Company; both of
these types of gloves were obtained from the stockroom at the Institute. The
manufacturer, supplier, and style number of each glove used in the immersion
and desorption tests are given in Table 13. The solvents used in the
measurements were reagent-grade chemicals obtained from the Institute
stockroom. Two brands of solvents were used during the test program
Mallinckrodt (Paris, KY) and EH Science (Cherry Hill, NJ).
The glove materials used in the tests were considered representative of a
variety of commercially available, unsupported protective gloves in common use.
The chemicals used in the tests were selected because they are common,
solvents that encompass several chemical functional groups.
69
-------
TABLE 13. IDENTIFICATION OF GLOVE MATERIALS USED IN LIQUID-IMMERSION
ABSORPTION AND DESORPTION TESTS
Glove
Manufacturer
Nominal
Style No. thickness, mil
Source
Butyl rubber
Poly(vinyl
chloride)
Neoprene
rubber
Natural
rubber
Nitrile
rubber
Siebe Norton, Inc.
N. Charleston, SC
Pioneer Industrial
Products Co.
Willard, OH
Edraont
Coshocton, OH
Edraont
Coshocton, OH
Edraont
Coshocton, OH
B-224 25 SRI stockroom
V-5 Quixam 5 SRI stockroom
29-875 19 Southern Safety
Products, Inc.
Birmingham, AL
26-680 21 Southern Safety
Products, Inc.
Birmingham, AL
37-165 22 Southern Safety
Products, Inc.
Birmingham, AL
70
-------
Absorption tests were conducted with each glove material with each
solvent. However, because of the low solubility of some of the solvents'in
some of the gloves, the duration of some of the immersion tests, and the
apparent extraction of some component of the PVC gloves (as discussed in the
next section), desorption tests were conducted with only ten of the
glove/solvent combinations. All of the tests were performed as simply as
possible using common laboratory equipment. Also, all of the tests were
purposely conducted at ambient temperature and relative humidity in the
laboratory with no provision for temperature or humidity control to keep the
test requirements as simple as possible.
Each glove sample tested was a flat, circular sample punched from a
protective glove with an arch punch. The diameter of each sample was 1-3/8 in.
The nominal thickness of each glove material except PVC was approximately
20 mil (0.0508 cm). The nominal thickness of the PVC gloves was 5 mil
(0.0127 cm). The diameter of each glove sample was measured with a stainless
steel ruler calibrated in units of 0.01 in. (0.0254 cm). The thickness of each
glove sample was measured with a Starrett dial gauge that was capable of
measuring accurately a thickness of less than 0.2 mil (0.000508 cm). Five
thickness measurements were made uniformly over the surface of each glove
sample. Each reported sample thickness was the average of the five
measurements.
Test Procedure
In the absorption tests, each glove sample was weighed to the nearest
0.0001 g on a top-loading Sartorius analytical balance. The weighed glove
sample was then immersed in approximately 50 mL of a given solvent in a
wide-mouth, screw-cap jar. At timed intervals, the sample was removed from the
jar, quickly and lightly blotted between two sheets of ashless filter paper,
placed in a.tared, wide-mouth weighing bottle, and weighed on the analytical
balance. After some weighings, the diameter and thickness of the test sample
were also measured. The sample was then re immersed in the solvent in the jar.
The weighing and reimmersion at timed intervals was continued until a constant
weight was observed. For some tests, a constant weight was not attained, even
after three weeks.
The desorption tests were run with samples that had reached equilibrium
solubility (that is, attained a constant weight) in a given solvent. In the
desorption tests, a solvent-saturated sample was removed from the solvent,
quickly and lightly blotted between two sheets of ashless filter paper, and
mounted on a tared wire tripod (constructed from 18-gauge copper wire) on the
pan of the analytical balance. The tripod-mounted sample was left on the
balance pan for the remainder of the desorption test, and the weight of the
sample was monitored and recorded as a function of time at ambient temperature
and relative humidity. The glass sliding doors on the side and top of the
weighing chamber on the balance were left slightly open. A low airflow was
maintained through the chamber by means of an aspirator pump connected to the
opening in the door on the top of the weighing chamber.
71
-------
The raw data collected in Che tests were sample weights as a function of
absorption time or desorption time. The data.were analyzed as described
above.
Immersion Test Results
The immersion absorption and desorption tests that were conducted during
the study are listed in Table 14. Three main types of sorption behavior were
observed:
Each PVC sample lost weight from immersion in each solvent.
Such weight loss indicates the extraction of some component of
the PVC formulation (probably a plaaticizer) by the solvent.
The neoprene-, nitrile-, butyl-, and natural-rubber glove
samples each showed small weight gains over a long period
(several days) with two of the four solvents.
The neoprene-, nitrile-, butyl-, and natural-rubber glove
samples each showed large, rapid weight gains over a period of
hours with the other two of the four solvents.
Desorption tests were conducted with the eight glove/solvent com-
binations that exhibited large, rapid solvent uptake by the glove sample
during the immersion tests and with two glove/solvent combinations that
exhibited small, slow uptakenatural-rubber/acetone and neoprene-rubber/
acetone.
The data obtained in the liquid-immersion absorption tests are summarized
in Table 15. The table includes for each test the average initial dimensions
of each glove, the range of test conditions (temperature and relative
humidity), the average initial weight of each glove, and the average weight
gain at equilibrium saturation for each glove/solvent combination. A summary
of these data for each glove/solvent absorption test conducted are presented in
Table 26 in Appendix D.
The time-dependent absorption and desorption data for each glove/solvent
combination tested are given in the data tables included as a separate volume
with this report. Absorption and desorption curves for selected tests are
plotted as reduced sorption curves and given in Appendix D (see Figures 11
through 18).
Solubilities
According to the principles on which the immersion test is based, the
weight gain of the glove sample at long times (that is, at equilibrium)
should yield the solubility of the solvent in the polymer. Estimated
solubilities (in g/cm3 of unsvollen polymer) obtained in this work are included
in Table 16. A comparison of the average solubilities (in moles/cm3)
determined in this work for the four solvents used versus solubilities
calculated from the data of Paul et al. (48) is given below for natural
rubber:
72
-------
TABLE 14. SUMMARY OF THE TYPES OF LIQUID-IMMERSION TESTS CONDUCTED
Glove
Butyl rubber
Natural rubber
Neoprene rubber
Nitrile rubber
Poly(vinyl chloride)
Solvent
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyc lohexane
Isopropanol
Toluene
Acetone
Cyc lohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
Absorption
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Desorption3
X
X
X
X
X
X
X
X
X
X
aDesorption tests were conducted for solvent/glove combinations that
achieved equilibrium during absorption tests.
73
-------
TABLE 15. SUMMARY OF AVERAGE LIQUID-IMMERSION ABSORPTION TEST DATA
Glove
Butyl rubber
Natural rubber
Neoprene rubber
Nitrite rubber
Poly(v inyl
chloride)
Initial
Solvent diameter, cm.
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
laopropanol
Toluene
Acetone
Cyclohexane
laopropanol
Toluene
Acetone
Cyclohexane3
Isopropanol3
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
3.53
3.53
3.53
3.56
3.56
3.56
3.56
3.56
3.56
3.56
3.56
3.56
3.56
3.56
3.53
3.56
3.61
3.56
3.56
3.56
Init Lai
thickness, cm.
0.0597
0.0582
0.0592
0.0587
0.0632
0.0660
0.0655
0.0653
0.0478
0.0480
0.0483
0.0485
0.0541
0.0566
0.0554
0.0559
0.0160
0.0145
0.0218
0.0147
Init ial
weight, g
0.6604
0.6455
0.6555
0.6484
0.6323
0.6612
0.6486
0.6478
0.6367
0.6405
0.64S8
0.6418
0.5635
0.5981
0.5890
0.5932
0.1930
0.1718
0.2187
0.1791
Maximum Temperature
weight gain, g range, *F
0.0268
1.5491
0.0024
1.0264
0.0958
1.7633
0.0314
2.0761
0.1912
0.4176
0.0274
1 . 5090
0.9707
0.0548
0.1285
0.7904
-0.0529
-0.0510
-0.0635
-0.0273
70-79
67-78
70-79
67-78
69-79
69-78
69-79
68-78
70-79
69-78
69-79
68-78
69-78
70-79
70-79
69-78
70-72
70-72
70-72
70-72
R.H.
range, Z
63-86
55-75
63-86
55-74
51-88
51-80
51-86
51-80
52-87
52-73
52-86
52-73
53-74
51-86
51-86
53-72
51-80
51-80
51-80
51-80
"Some <»f the nitrile-rubber samples immersed in these solvents continued to gain weight even after 530 hr.
-------
TABLE 16. AVERAGE SOLUBILITIES AND DIFFUSION COEFFICIENTS CALCULATED FROM
LIQUID-IMMERSION ABSORPTION AND DESORPTION TEST DATA
Glove
Butyl rubber
Natural rubber
Neoprene rubber
Nitrile rubber
Solvent
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Isopropanol"
Toluene
Solubility,
g/cm3
0.0457
2.7261
0.0041
1.7691
0.1527
2.6877
0.0489
3.2067
0.4044
0.8781
0.0574
3.1388
1.8204
0.0982
0.2340
1.4377
Diffusion coef f icient ,a> >c cm2 /sec
10fi x Da
0.0056
-0.15
0.0071
0.38
0.40
0.25
0.042
0.53
0.25
0.037
0.0013
0.35
0.42
0.00025
0.00035
0.059
10b x Dd
_.
0.33
0.28
0.38
0.49
0.26
0.26
0.054
0.20
0.45
0.20
10b x Da
0.0050
0.18
0.00043
0.32
0.69
0.26
0.044
0.58
0.52
0.040
0.00014
0.41
0.44
0.00030
0.00071
0.098
10b x D^
_ _
0.28
0.21
0.45
0.48
0.29
0.31
0.043
0.21
0.48
0.15
"The term D is the diffusion coefficient estimated from the initial slope of the reduced absorption curve.
D, is the diffusion coefficient estimated from the initial slope of the reduced desorption curve. All data
points (including (0,0)1 for which Mt/M<>) <0.6 were used in these calculations; if no data points meeting this
criterion except (0,0) existed, then the calculation was based on (0,0) and the first data point above
bThe term Da is the diffusion coefficient calculated from a curve-fit of data obtained in an absorption test.
D! is the diffusion coefficient calculated from a curve-fit of data obtained in a desorption test.
a
cThe symbol " " means that the experiment indicated was not conducted.
''some of the nitr ile-rubber samples immersed in these solvents continued to gain weight even after 530 hr.
-------
Solvent Paul et al. Southern Research
Acetone 0.00169 0.00263
Cyclohexane 0.0338 0.0319
Isopropanol 0.000528 0.000814
Toluene 0.0387 0.0348
It may be seen that the solubilities determined in this work agree well with
Paul's for toluene and cyclohexane, for which high solubilities were obtained.
There is less agreement in the two data sets for acetone and isopropanol;
however, these solvents are not very soluble in natural rubber.
It should be emphasized that it is important to obtain time-dependent
weight-gain data in immersion tests that are conducted to determine the
solubilities of organic liquids in polymeric glove materials. Time-dependent
data are necessary to establish the attainment of equilibrium weight gain.
Although in many cases a true solubility may be obtained by simply immersing a
glove sample in an organic liquid for a fixed period of time (24 hr, for
example) and then weighing the sample, this single-data-point technique may
often give incorrect results if equilibrium weight gain has not been attained.
In the immersion tests conducted in this study, for example, the weights of
some of the nitrile samples immersed in cyclohexane and some of the nitrile
samples immersed in isopropanol were still increasing at the time the final
weights were measured (530 hr). Thus, an equilibrium weight gain was not
established, and a true solubility could not be determined.
The major problem in determining the solubility of an organic liquid in an
unsupported, polymeric glove material (and the major potential problem in the
entire immersion test method) is the possible extraction of components of tha»
glove formulation by the immersion solvent. The extraction of components can
often be deduced from the liquid-immersion absorption data; for example, the
PVC samples in this work that were immersed in solvents lost weight. However,
it is conceivable that simultaneous solvent absorption and component extraction
could occur during immersion and, thus, that the net weight gain of the glove
sample may not be entirely due to the absorption of solvent by the glove.
Hence, some caution must be used in interpreting immersion test results. (It
should be noted, however, that if the solvent extracts a component of the glove
formulation and a weight loss is observed, the glove will probably not provide
the required chemical resistance.)
Diffusion Coefficients
Solvent diffusion coefficients estimated from the immersion data by the
two methods described above [that is, from the initial slope of the reduced
sorption curve (Equation 16, Section 7) or by a fit of the reduced sorption
curve to Equation 12, Section 7] are listed in Table 16 for each glove/solvent
combination. There is little data in the literature that contains
experimental values for diffusion coefficients of organic liquids in polymeric
glove materials with which to compare the diffusion coefficients determined in
the immersion tests. For the two glove/solvent combinations for which we found
literature values of the diffusion coefficient (natural-rubber/cyclphexane and
natural-rubber/toluene), the diffusion coefficients estimated from the immer-
sion absorption and desorption test data are approximately an order of magni-
tude smaller than the values reported in the literature. However, the diffu-
76
-------
sion coefficients calculated for each glove/solvent combination with each
method of estimation are, in general, self-consistent and in close agreement
with apparent diffusion coefficients calculated from the direct permeation
tests discussed in this section.
It should be emphasized that the values of the diffusion coefficients
determined from the immersion absorption and desorption data are based on
approximations that contain several simplifying assumptions. In the curve-fit
method, in particular, a concentration-independent diffusion coefficient is
assumed. From the frequent lack of coincidence of the absorption and
desorption curves shown in the representative figures in Appendix D, however,
the diffusion coefficients of the solvents in most of the gloves tested appear
to be concentration dependent. Although more sophisticated treatments of the
absorption and desorption data are possible, it was decided to first explore
the potential of simple data-reduction methods that would yield parameters
useful for predictive purposes.
VAPOR ABSORPTION TESTS
Liquid-immersion absorption tests cannot be readily automated; thus,
they are labor intensive although they are very simple to conduct. In an
attempt to avoid this problem, a brief investigation of a vapor absorption/
desorption procedure, which could easily be automated, was conducted.
Theoretically, at atmospheric pressure, the process of solvent diffusion into a
polymer in contact with a liquid is the same as for a polymer in contact with a
saturated vapor. Therefore, the saturated-vapor test and the immersion test
should provide the same data; however, differences in test results may be
observed for some liquid/polymer systems. For example^ a liquid may leach
additives from a polymer to a greater extent than does" vapor.
Description of the Test
In the vapor absorption test described below, a thin polymeric glove
sample was suspended from a sensitive quartz spring, which was calibrated for
extension versus load. The sample and spring were enclosed in a chamber
containing an organic vapor maintained at the vapor pressure of the liquid at
"ambient" temperature. A cathetometer was used to observe the spring extension
as a function of time until a constant extension was observed. The data
obtained were then converted to weight gain versus time or, in other words,
cumulative absorption versus time. Time-dependent vapor absorption and
desorption data are treated and analyzed in the same manner as for the
immersion tests described above.
Test Apparatus and Procedure
The same glove materials and solvents that were used in the immersion
tests were scheduled for use in saturated-vapor absorption and desorption
tests. As discussed below, however, because of the long time required to
conduct the vapor absorption test, only a single absorption test was conducted
with a single glove/solvent combination (nitrile rubber/acetone) during the
test program. (No desorption tests were conducted.)
77
-------
The test apparatus used in the vapor absorption test is shown in Figure 7.
The apparatus consisted of a calibrated fused-quartz spring (Ruska Instrument
Corporation, Houston, TX) suspended from a hook on the inside upper end of a
custom-fabricated jacketed condenser (M.B. Watson Scientific Glassblowing,
Tuscaloosa, AL), a round-bottom flask that contained approximately 25 mL of
acetone, a water bath to prevent rapid changes in the temperature of the
solvent, and a cathetometer to monitor the extension or compression of the
quartz spring.
The quartz spring used in the vapor absorption test was precalibrated by
the manufacturer to give a spring extension of 1 mm for a 1 mg load. The maxi-
mum extension of the spring was 500 mm. Prior to the vapor absorption test,
the spring was calibrated over a range of 10 to 500 mm with a set of precision
milligram weights (Bitronics, Inc., Bethlehem, PA). The extension of the
spring was measured with a sensitive cathetometer graduated in divisions of
0.1 mm. The cathetometer was focused on a "crosshair" reference mark near the
end of the spring.
For the test conducted, a 1-cm by l-cm square sample was cut from a
nitrile-rubber glove, and a small hole was punched through the sample near the
center of one of the edges. The test sample was mounted on a hook at the
bottom of the quartz spring through the hole in the sample. The vertical
position of the reference mark on the quartz spring was measured through the
cathetometer. About 25 mL of acetone was poured into the 100-mL round bottom
flask, and it was connected to the bottom of the condenser. The flask was
immersed into the water bath as shown in Figure 7. The extension of the quartz
spring as the glove sample absorbed acetone vapor was then monitored with the
cathetometer as a function of time over a ten-day period.
Vapor Absorption Test Results
By the time the test method was selected and the required appparatus was
designed, fabricated, assembled, and calibrated, little time remained under
the contract for testing. As stated previously, because of the long duration
required (approximately two weeks) for a single vapor absorption test, only one
such test with a single glove/solvent combinationnitrile rubber with
acetonewas completed. The data obtained in this test are given in Table 17.
As the data in the table indicate, the measured absorption of acetone by
the nitrile-rubber sample fluctuated as a function of time. Consequently, no
useful time-dependent absorption data were obtained in the test.
We attribute the fluctuations obtained in the vapor absorption test
primarily to the lack of temperature control of the test apparatus. With the
addition to the apparatus of a constant-temperature bath to control the
temperature of the reservoir of liquid solvent and the circulation of the water
from the bath through the outer jacket of the condenser, reliable time-
dependent absorption data should be obtained.
The vapor sorption method is experimentally more complex and time-
consuming than the liquid-immersion absorption/desorption test method described
above, and the vapor sorption method is applicable only to volatile solvents.
78
-------
GLASS STOPPER WITH HOOK
CALIBRATED
QUARTZ
SPRING
ALIGNMENT
"CROSS HAIR-
ON QUARTZ
SPRING
JACKETED
REFLUX
CONDENSER
ROUND-
BOTTOM
FLASK
SOLVENT
GLOVE SAMPLE
(1 cm x 1 cm)
CATHETOMETER.
WATER BATH
6749-15
Figure 7. Diagram of vapor sorption apparatus (not drawn to scale).
-------
TABLE 17. VAPOR ABSORPTION DATA OBTAINED
FOR NITRILE RUBBER AND ACETONE8
Time, min Weight, mg Weight gain, mg
0
77
147
197
327
1,234
1,457
1,711
2,681
2,897
3,227
4,484
5,777
9,887
10,397
11,354
11,848
12,763
13,287
14,253
56.55
55.90
55.90
56.41
58.45
83.13
82.79
83.57
97.47
93.39
89.50
94.89
100.43
105.38
95.18
103.98
101.40
107.91
100.82
107.52
0
-0.65
-0.65
-0.14
1.90
26.58
26.24
27.02
40.92
36.84
32.95
38.34
43.88
48.83
38.63
47.43
44.85
51.36
44.27
50.97
aThe initial dimensions of the test sample were
1 cm x 1 cm x 0.0508 cm. The temperature was in
the range of 21 to 28 °C during the test.
80
-------
Thus, it does not appear to be a widely applicable alternative to permeation
testing.
PERMEATION TEST DATA
Direct permeation tests were conducted with each of the glove materials
(except PVC) and each of the organic solvents (except isopropanol) to generate
permeation-rate-versus-time data to compare with predictions made from the
results of the immersion and vapor sorption tests. Direct permeation tests
were conducted because sufficient quality time-dependent permeation data to
compare with the gravimetric methods proposed do not exist in the literature.
The direct permeation tests were conducted according to the standard ASTM
Method F739. The tests were conducted in a 1-in. "ASTM permeation test cell"
from Pesce Lab Sales (Kennett Square, PA). All tests were conducted at room
temperature using dry nitrogen gas as the collection fluid. Each permeation
test was conducted in an "open-loop" mode; that is, fresh nitrogen was
continually swept at a flow rate of 100 mL/min across the unchallenged surface
of the test sample. At timed intervals, an aliquot of the gas stream from the
test cell was sampled with a gas-tight syringe. The sampled portion of the gas
stream in the syringe was then injected directly into a Hewlett-Packard Model
5790 gas chromatograph for the detection and quantitation of the solvent in the
gas stream.
In each permeation test, the test cell was immersed in a water bath
containing tap water at room temperature. The temperature of the water bath
was not controlled, but its temperature remained constant at approximately
20 ± I "C during each test. The temperature of the nitrogen gas stream that
was swept through the collection side of the test cell was preequilibrated to
the temperature of the water bath by flowing it through a 50-ft coil of
l/4-in.-OD copper tubing immersed in the water bath.
Prior to each permeation test, the gas chromatograph was calibrated for
the organic challenge liquid to be used in the permeation test by injecting
dilute solutions of the organic liquid in a suitable solvent into the gas
chromatograph. A standard calibration curve that covered the entire working
range of the method was determined with five concentrations of standard solu-
tions.
The test data generated in each open-loop permeation test were instantane-
ous permeation rate as a function of time. Average breakthrough times and
steady-state permeation rates obtained in these direct permeation tests are
given in Table 18. This table also includes diffusion coefficients and
solubilities obtained from curve fits of the permeation-rate data to
Equation 3 (Section 3).
81
-------
TABLE 18. SUMMARY OF AVERAGE PERMEATION-TEST DATA
00
to
Glove
Butyl rubber
Natural rubber
Neoprene rubber
Nitrite rubber
Solvent
Acetone
Cyclohexane
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Toluene
Acetone
Cyclohexane
Toluene
Thickness ,
cm
0.0615
0.0600
0.0589
0.0616
0.0619
0.0617
0.0617
0 .0480
0.0460
0.0478
0.0589
0.0582
0.0574
Steady-state
Breakthrough permeation rate.
time,8 tnin pg/(cm2«min)
>1115
55
24
17
19
>150, <1400
10
22
170
12
10
>1350
52
<0.47
437
396
34
572
1
782
151
26
642
962
<0.12
210
Calculated
dif fusion
coefficient ,
cm2/sec, 106 x D
0.085
0.19
0.17
0.22
0.0012
0.58
0.15
0.014
0.26
0.39
0.066
Calculated
solubility .
g/cm3
7.6
2.9
0.33
3.2
2.6
1.5
1.1
7.0
2.1
3.0
4.3
aThe minimum permeation rates that could be detected were: 0.47 ug/(cm2>nun) for acetone,
0.11 ug/(cm2.min) for toluene, and 0.12 Ug/(cm2.min) for Cyclohexane.
bThe symbol "--" means that no diffusion coefficient or solubility could be calculated.
-------
SECTION 8
PREDICTIVE ALGORITHMS
The purpose of the contract was to develop models and test methods that
would allow the prediction of the protection from permeation afforded by
polymeric gloves in contact with liquid organic chemicals. The previous
sections of this report have included a summary of the results of a literature
survey of topics relevant to this purpose and descriptions of the theory and
test methods developed to make the required predictions.
In this section of the report, the key results presented previously are
used to construct or suggest algorithms or approaches to the evaluation of
polymeric gloves proposed as protective clothing for use with specific
chemicals.
SPECIFIC ALGORITHM REQUIREMENTS
Among the objectives of this work were that the model or predictive
algorithms developed must be relatively simple; must not require data other
than that provided in the PMN submittal, in handbooks or in data bases, or
obtained in simple, reproducible tests; and'tnust be applicable to a wide
variety of chemical substances and polymeric gloves. In addition, the
algorithms must be capable of predicting such quantities as maximum (or
steady-state) permeation rates; breakthrough times; 1-hr and 8-hr cumulative
exposures; and permeation rates versus time. It must also be able to place the
protective glove into one of a series of qualitative groupings such as the
following:
The glove will protect against dermal contact with the chemical
in question for a limited period (1 hr) or an extended period
(full work shift).
The glove will not protect against dermal contact even for a
limited period.
Experimental data are needed before a prediction can be made.
APPROACH TO THE ALGORITHM
As stated previously in this report, the approach used in the development
of the predictive ability outlined above was to base predictions as much as
possible on the theory of diffusion in polymers. Thus, this work emphasized
the development of models and simple test methods that allow the prediction or
determination of diffusion coefficients and solubilities. Given these
fundamental parameters, any of the required predictive calculations can be
completed using, for example, diffusion equations such as those presented in
Section 3.
83
-------
1C should be noted that the predictive algorithms given in this section of
the report are based only on simple diffusion theory, and this will limit their
usefulness. However, because the development of any predictive algorithm is a
stepwise process, it is logical to begin with the simplest approach. The
predictive*algorithms developed can easily be improved in sophistication under
subsequent research efforts conducted in conjunction with confirming the
applicability of the algorithms. That is, predictions should be made and
compared with available experimental data, and the results should be studied to
determine needed improvements to the model. The model (that is, the
algorithms) would then be modified, and new predictions would be made and
compared to the experimental data, and so forth.
Another point to be made is that the algorithms described here are not
"user friendly." That is, for example, the suggested input may not be in a
form that would be easily handled by someone processing the PUN submittal.
Also, the only "software" that currently exists (except for very simple
programs) is described in the various sections of this report. The development
of software that would be easy for an "untrained" person to use would not be
particularly difficult, but it would be time-consuming.
INPUT TO THE ALGORITHM
The input to the predictive algorithm includes data provided in the PHN
submittal as well as data that may be readily available from handbooks or
data bases. In addition, the parameters (and associated quantitative values)
used to evaluate the degree of protection must be specified. Examples of input
data required by the UNIFAP program (which is a part of the predictive
algorithm) to calculate solubilities are:
The temperature of interest.
The structure, molecular weight, and density of the liquid
organic chemical.
The structure, molecular weight, density, and degree of
polymerization of the polymeric glove material.
Other input data* may include those required to calculate diffusion
coefficients using the modified Paul model (which is also incorporated in the
predictive algorithm):
Viscosity of the solvent as a function of temperature.
Density (or specific volume) of the solvent as a function of
temperature.
*The list of input data given in Section 6 for the Paul model included other
parameters that may be determined within software and, thus, be transparent to
the user.
84
-------
Density (or specific volume) of the solvent and the polymer at
the temperature of interest.
The molecular weights of the solvent and the polymer.
Still additional input data may be those obtained in simple immersion tests
such as those described in Section 7:
Initial dimensions of the polymeric glove samples (thickness and
area) for each test.
Temperatures at which experiments were conducted.
Weight-change-versus-time data for each immersion test.
For the sake of completeness, other data may be input to the predictive
algorithm. These may include, for example, the manufacturer or vendor catalog
number and other miscellaneous information describing the polymeric glove.
Even with all of the input data listed above, it will not be possible for
the predictive algorithm to assess the protection afforded by a given polymeric
glove unless some quantitative criteria are specified. These criteria may
include one or more of the following:
The maximum permeation rate per unit area allowed within a
specified period (1 hr, 8 hr, and so forth).
The maximum cumulative exposure per unit area allowed within a
specified period (1 hr, 8 hr, and so forth).
The minimum breakthrough time allowed (based on a specified
analytical sensitivity to simulate open-lo'op or closed-loop
permeation tests of a known area of sample).
Maximum permeation-rate ranges to simulate manufacturers'
qualitative recommendations (for example, 0.15 to
1.5 mg/(m2«aec) to be equivalent to Edmont's permeation rating
of "excellent").
Maximum percentage weight gain within a fixed period (for
example, 2 hr) to simulate a degradation test.
The protection criteria specified by the user of the predictive algorithm
should be based, if possible, on well-established safety criteria, such as
those published by NIOSH, the American Conference of Government Industrial
Hygienists, or EPA.
85
-------
CALCULATIONS OF FUNDAMENTAL PARAMETERS
After data of the type listed above are input to the predictive algorithm,
it will calculate the fundamental parameters needed to make the desired
predictionsparameters such as solubility (using UNIFAP), solubility (using
the 24-hr immersion-test data), the solvent diffusion coefficient versus
solvent volume fraction (using the modified Paul model), or the apparent
solvent diffusion coefficient (using immersion-test data). The approach used
by the predictive algorithm will depend on the data input to it. For example,
if only immersion-test data are provided in the PMN submittal (and the glove
material is not identified), then the solubility and the diffusion coefficient
cannot be calculated using the UNIFAP or Paul models, respectively. Also, only
an apparent diffusion coefficient can be determined from the immersion-test
data using the technique recommended in this report; that is, concentration-
dependent diffusion cannot be modeled. It should be noted, however, that the
apparent diffusion coefficient calculated from immersion-test data for systems
exhibiting concentration-dependent diffusion coefficients may often be
satisfactory for predictive purposes.
Results of the calculation of diffusion coefficients and solubilities
using the predictive models and test methods developed in this work were
presented in previous tables (11 and 16) and figures (4, 5, and 6) in this
report. As stated above, the specific calculations performed by the predictive
algorithms will depend on the data input. If insufficient information is input
to perform the required calculations, then the predictive algorithm will inform
the PMN submittal reviewer that additional information, is required; the
algorithm could be designed to specify the type of missing data that must be
supplied.
CALCULATION OF CUMULATIVE PERMEATION OR PERMEATION RATE
After the algorithm has yielded the solvent solubility and its apparent
diffusion coefficient, these data may then be used to calculate the permeation
rate versus time (using, for example, Equation 3) and the cumulative
permeation versus time (using, for example, Equation 4). If there is
sufficient data for the predictive algorithm to yield the solvent diffusion
coefficient as a function of solvent volume fraction (as well as to yield the
solubility), then numerical methods similar to the example given in Appendix C
must be used to determine permeation rate and cumulative permeation as a
function of time from Fick's laws, given the initial and boundary conditions
(see Section 3). An example of the calculation of permeation rate versus time
using a numerical method as well as a computer program for the calculation is
also given in Appendix C. (This simple program will not handle concentration-
dependent diffusion coefficients.)
If all of the input data listed above were available, then the
calculation of J versus t and Q versus t could be performed using several
combinations of fundamental parameters and methods. The most likely
combinations are:
86
-------
UNIFAP solubility, Paul diffusion coefficients, numerical
analysis.
Immersion solubility, Paul diffusion coefficients, numerical
analysis.
UNIFAP solubility, immersion diffusion coefficient, analytical
solutions.
Immersion solubility, immersion diffusion coefficient,
analytical solutions.
It would be preferable to perform calculations for all of the combinations
listed above or for as many combinations as possible. Any estimate of the
protection afforded by a recommended polymeric glove could then be
"safe-sided." That is, a predicted failure to meet the specified protection
criteria for any combination of input data would result in the rejection of
the polymeric gloves recommended in the PMN submittal. It may obviously be
desirable to weight evaluations in favor of experimental data, such as
immersion-test data.
EVALUATION OF PROTECTION CRITERIA
After the calculation of permeation rate (J) and cumulative permeation (Q)
as a function of time, these data may be used to determine whether the
protection criteria input to the algorithm have been met (that is, whether
the polymeric glove will provide the desired protection). For example, if the
criteria specified that the permeation rate per unit area shall not exceed
1-mg/(m2
-------
TABLE 19. COMPARISON OF SOLUBILITIES CALCULATED USING THE UNIFAP
MODEL OR OBTAINED EXPERIMENTALLY WITH MANUFACTURERS'
CHEMICAL-RESISTANCE GUIDELINES3)b
Polymer
Natural
rubber
Butyl
rubber
Neoprene
rubber
Degradation
Solvent 10s x Cuni 10s x Cexp rating0
Mechanol
Ethanol
Isopropanol
n-Butanol
£-Pentanol
Benzyl alcohol
ri-Propanol
Acetone
2-Ethyl-l-butanol
£-Butanol
t^Pentanol
Diethyl carbonate
Methyl ethyl ketone
Ethyl acetate
n-Propyl acetate
n-Hexane
ii-Heptane
Tetralin
Cyclohexane
Cyelohexanone
Toluene
Tetrachloroethylene
Carbon tetrachloride
Trichloroethylene
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
26.5
42.4
81.6
93.3
103
27.5
100
IBS
114
81.5
96.9
202
271
411
581
*e
*
*
*
*
*
*
*
*
322
*
29.7
*
*
*
69.8
4.92
8.59
52.8 (81.4)
125
130
142
146
169 (263)
259
414
437
636
713
766
1270
1540
1580
3330
3380 (3190)
3410
3870 (3480)
4240
5370
5690
(78.7)
(3240)
(6.8)
(1920)
(696)
(1040)
(95.5)
(3410)
E
E
E
E
NA
NA
E
E
NA
NA
NA
NA
G
C
F
NR
NA
NA
NA
NA
NR
NA
NR
NA
NA
NA
NA
NA
G
NA
E
NR
Permeation
racing*1
E, NN
VG, NN
E, NN
G
NA
NA
VG
F, NN
NA
NA
NA
NA
P. NN
C, NN
F, NN
NA
NN
NA
NN
NA
NN
NN
NN
NN
NA
RR
NA
NA
F, NN
NN
E
NN
(continued)
88
-------
TABLE 19 (continued)
Polymer
Nitrile
rubber
Poly(vinyl
chloride)
Solvent
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
lo5,cunl .....^
(3130)
(117)
(389)
(1560)
,8 (..)»
* ( )
49.5 { >
* (-)
Degradation
? rating0
NR
NA
E
F
NR
NA
G
NR
Permeation
rating"
NN
RR
E, RR
F, NN
NN
NN
E
NN
aThe units of the calculated equilibrium solubilities, CUR^, and the
experimental solubilities, Cex_, are moles/cm^.
The values in parentheses under the column labeled "10s x Cex " were
determined during immersion tests conducted under the current effort; the
other data under this heading were calculated from data reported in
Reference 48.
cThese are Edmont degradation ratings; E means excellent; G means good; F
means fair; NR means not recommended; NA means not available.
For the Edmont ratings, E means excellent (permeation rate <0.15 mg/m2/sec);
VG means very good (permeation rate <1.5 mg/m2/aec); G means good
(permeation rate <15 mg/m2/sec); F means fair (permeation rate
<150 mg/m2/sec); P means poor (permeation race <1500 mg/m^/sec); ND means
none detected. The double letters refer to ratings used by ADL (4_); RR
(recommended) means that a large amount of test data indicates exTellent
chemical resistance; NN (not recommended) means that a large amount of test
data indicates poor chemical resistance; NA means rating not available (or
conflicting ratings are reported by ADL).
eThe symbol "*" in this column means that an activity of one was achieved
only for a solvent volume fraction equal to one. (That is, the solvent and
the polymer are predicted to be miscible in all proportions.)
The symbol "--" in this column means that no interaction parameters were
available for the polymer/solvent pair indicated.
UNIFAP solubilities for solvents in poly(vinyl chloride) were based on
the polymer structure alone; the presence of plasticizer was not
considered.
The symbol "" in this column means that weight loss was observed in the
immersion tests with poly(vinyl chloride). ~
89
-------
If only "steady-state" or 24-hr solubilities were available from immersion
tests, similar correlations between large solubilities and manufacturers'
recommendations would be expected. Experimental (or immersion-test
solubilities) are also shown in Table 19. Again, there is generally a good
correlation between manufacturers' recommendations and experimental
solubilities.
OUTPUT OR EVALUATION REPORT
The output of the predictive algorithm could include:
A reiteration of all of the data input to the algorithm.
A list of the methods used to calculate solubility and
diffusivity data.
A pass/fail report for all protection criteria.
Recommendations such as the suitability of the glove for a 1-hr
period or for extended periods.
Recommendations for the submittal of more data on which to base
an evaluation.
Other output could include calculated J-versus-t or Q-versus-t curves,
D-versus-C curves, and tables of such data. A regulatory style report could
also be issued by the computer.
CONFIRMATION
As stated above, the development of a predictive algorithm is an iterative
process. That is, the algorithm must be tested or confirmed at various states
in its development. An anticipated task to confirm the models and predictive
test methods developed under this contract was not completed due to the
unavailability of funds.
Some confirmation work has been described previously in this report. This
work included the comparison of calculated diffusion coefficients to values
published in the scientific literature, the comparison of solubilities
calculated using UNIFAP and from immersion-test data to solubilities previously
reported, and the comparison of calculated solubilities to manufacturers'
degradation and permeation ratings.
Presented below are results obtained in additional efforts to demonstrate
the feasibility of the suggested predictive models and test methods described
in this report. These efforts included:
The prediction of a permeation-rate-versus-time curve for
benzene in natural rubber using a solubility calculated with the
UNIFAP program, concentration-dependent diffusion-coefficient
90
-------
data calculated with the modified Paul model, and a numerical
analysis method.
The prediction of permeation-rate-versus-time curves for several
solvents in a series of glove materials using apparent diffusion
coefficients and solubilities obtained in liquid-immersion tests
and using Equation 3.
The use of these calculated permeation-rate-versus-time curves
to predict breakthrough times and steady-state permeation rates
for the glove/solvent combinations used in the liquid-immersion
tests and the comparison of these predictions with values
obtained in permeation tests.
The comparison of two predicted permeation-rate-versus-time
curves with data obtained in permeation experiments.
It should be noted that much more confirmation work than presented here
could be completed using the experimental data obtained under the current
contract. Again, the current effort was devoted primarily to the
identification of potential predictive models and test methods and the limited
demonstration of their feasibility. More extensive confirmation work and
additional refinements to the models and test methods proposed here should be
the subject of a future contract.
Prediction of a Permeation-Rate Curve Using Theoretical Models
The concentration dependence of the diffusion coefficient for benzene in
natural rubber calculated using Che modified Paul model was given in Figure 5.
The solubility of benzene in natural rubber (0.685 g/cm3 of swollen polymer)
was calculated using the UHIFAP software. These theoretical data were then
used to predict a benzene permeation-rate-versus-time curve for an unsupported,
0.046-cm-thick natural-rubber glove (see Figure 8). The calculations needed
were performed using a numerical analysis method similar to that described in
Appendix C; however, the method was modified to account for a concentration-
dependent diffusion coefficient.
Figure 8 includes a permeation-rate curve calculated from experimental
cumulative-permeation data reported by Weeks and McLeod (20) for benzene
through an unsupported natural-rubber glove of the same thickness as used in
the theoretical calculations. Although there is considerable scatter in the
experimental data, the relative agreement between the experimental and
predicted curves is obvious.
Prediction of Permeation-Rate Curves Using Test Methods
Table 16 includes apparent diffusion coefficients and solubilities
calculated from data generated in liquid-immersion tests using 16 glove/solvent
combinations. Average apparent diffusion coefficients defined by [(Dfl+Dd)/2]
and solubilities and Equation 3 were used to predict permeation-rate-versus-
91
-------
300
E
isi
200
100
oc
HI
&
PERMEATION DATA
(EXTRAPOLATED FROM
REFERENCE 20)
W""
PREDICTED USING
THEORETICAL MODEL
10
15
TIME, min
20
25
5749-11
Figure 8. Comparison of predicted and experimental permeation-rate curves for
benzene through natural rubber.
92
-------
time data for acetone in natural rubber and in nitrile rubber. (Permeation
tests were also conducted for these glove/solvent combinations.) The predicted
permeation-rate data for acetone through a 0.061-cm-thick nitrile-rubber-glove
sample are shown in Figure 9. Also shown in the same figure are permeation-
rate data obtained in an actual permeation experiment using the same glove/
solvent combination (see Test No. D031-45 in Table 27 in Appendix D). The
general agreement between the major features of the predicted curve and the
actual permeation-rate curve is obvious. The predicted permeation-rate curve
for acetone through a 0.061-cm-thick natural-rubber glove is compared to the
measured permeation-rate curve (see Test No. D0301-28 in Table 26) in
Figure 10.
Due to the lack of time, plots of all of the predicted permeation-rate
curves or the actual permeation data could not be prepared. However, the use
of the constants (D and S) in Table 16 and Equation 3 to calculate these curves
is relatively simple, and the permeation-rate data obtained in permeation
experiments are given in a separate data volume. Thus, more comparisons such
as those shown in Figures 9 and 10 could easily be made.
Permeation-rate data, predicted as just described, were examined to yield
breakthrough times (based on the permeation-rate sensitivities reported in
Table 18) and steady-state permeation rates. These predictions and the average
observed breakthrough times and steady-state permeation rates measured in
permeation experiments are given in Table 20.
93
-------
1200
1000
E
N* 800
a.
< 600
O
ff
400
200
I I
PERMEATION DATA
10
20
30
40
TIME, min
50
PREDICTED FROM
IMMERSION DATA
60
70
80
5749-19
Figure 9. Comparison of predicted and experimental permeation-rate curves for acetone through
nitrite rubber.
94
-------
PREDICTED FROM
IMMERSION DATA
PERMEATION DATA
Figure fO. Comprison of predicted and experimental permeation-rate curves for acetone
through natural rubber.
95
-------
TABLE 20. COMPARISON OF MEASURED BREAKTHROUGH TIMES
AND STEADY-STATE PERMEATION RATES WITH THOSE
PREDICTED FROM IMMERSION TEST DATA3
Glove
Natural
rubber
Neoprene
rubber
Nitrile
rubber
Solvent
Breakthrough time, min
Measured Predicted
Steady-state permeation
rate, ug/(cm2*min)
Measured
Predicted
Butyl
rubber
Acetone
Cyclohexane
Isopropanol
Toluene
>1113
55
24
MO*
6
>106
5
<0.47
437
~
396
0.26
670
0.030
59
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
Acetone
Cyclohexane
Isopropanol
Toluene
17
19
>150, <1400
10
22
>170
12
10
M350
52
7
5
130
4
6
28
10*
4
3
>106
>106
11
34
572
1
782
151
26
642
962
<0.12
210
57
910
1.9
1200
130
49
0.095
1100
880
0.025
0.088
200
*The symbol "" means that no permeation test was conducted with this
glove/solvent combination.
bThe predicted breakthrough times are based on the minimum permeation
rates that could be detected in permeation tests: 0.47 ug/(cm2«min) for
acetone, 0.12 yg/(cm2'oin) for Cyclohexane, and 0.11 ug/(cm2«min) for
toluene, and 0.31 ug/(cm2-min) for isopropanol.
96
-------
REFERENCES
1. Waterman, R.R. (Ed.). R.T. Vanderbilt News. 34(2); 1972.
2. Blackley, D.C. High Polymer Latexes, Their Science and technology, Vol.
2, Testing and Applications. Applied Science Publishers Ltd., London;
1966.
3. Williams, J.R. Permeation of Glove Materials by Physiologically Harmful
Chemicals. Am. Ind. Kyg. Assoc. J. 40(10): 877-882; 1979.
4. Schvope, A.D.; Costas, P.P.; Jackson, J.O.; Weitzman, D.J. Guidelines for
the Selection of Chemical Protective Clothing, 2nd ed. Volume I: Field
Guide. Vol. II: Technical and Reference Manual. Prepared by Arthur D.
Little, Inc., for the U.S. Environmental Protection Agency (under
subcontract), 1985 March.
5. Schvope, A.D.; Costas, P.P.; Jackson, J.O.; Weitzman, D.J. Guidelines for
the Selection of Chemical Protective Clothing. Volume I: Field Guide.
Vol. II: Technical and Reference Manual. Prepared by Arthur D. Little,
Inc., for the U.S. Environmental Protection Agency (under subcontract),
1983.
6. Bomberger, D.C.; Brauman, S.K.; Podoll, R.T. Studies to Support PMN
Review: Effectiveness of Protective Gloves; Final Report. Prepared by
SRI International under Contract 68-01-6016 with the US Environmental
Protection Agency, Washington, DC, 1984 September.
7. Roder, M.M. Chemical Protective Clothing Monograph. Prepared by the
National Institute for Occupational Safety and Health (NIOSH). Draft
copy; 1985 January 31.
8. Stampfer, J.F.; McLeod, M.J.; Betts, M.R.; Martinez, A.M.; Berardinelli,
S.P. Permeation of Eleven Protective Garment Materials by Four Organic
Solvents. Am. Ind. Hyg. Assoc. J. 45(9): 642-654; 1984.
9. Bennett, R.D.; Feigley, C.E.; Oswald, E.O.; Hill, R.H. The Permeation by
Liquid Coal of Gloves Used in Coal Liquefaction Pilot Plants. Am. Ind.
Hyg. Assoc. J. 44(6): 447-452; 1983.
10. Stampfer, J.F.; McLeod, M.J.; Betts, M.R.; Martinez, A.M.; Berardinelli,
S.P. Permeation of Polychlorinated Biphenyls and Solutions of These
Substances Through Selected Protective Clothing Materials. Am. Ind. Hyg.
Assoc. J. 45(9): 634-641; 1984.
11. Silkowski, J.B.; Horstman, S.W.; Morgan, M.S. Permeation Through Five
Commercially Available Glove Materials by Two Pentachlorophenfll
Formulations. Am. Ind. Hyg. Assoc. J. 45(8): 501-504; 1984.
97
-------
12. Schwope, A.D.; Randel, M.A.; Broome, H.G. Dimethyl Sulfoxide Permeation
Through Glove Materials. Am. Ind. Hyg. Assoc. J. 42(10): 722-725; 1981.
13. Williams, J.R. Evaluation of Intact Gloves and Boots for Chemical
Protection. Am. Ind. Hyg. As-soc. J. 42(6): 468-471; 1981.
14. Weeks, Jr., R.W.; Dean, B.J. Permeation of Methanolic Aromatic Amine
Solutions Through Commercially Available Glove Materials. Am. Ind. Hyg.
Assoc. J. 38(12): 721-725; 1977.
15. Sansone, E.B.; Tewari, Y.B. The Permeability of Laboratory Gloves to
Selected Solvents. Am. Ind. Uyg. Assoc. J. 39(2): 169-174; 1978.
16. Walker, E.A.; Castegnaro, M.; Garren, L.; PignaCelli, B. Limitations to
the Protective Effect of Rubber Gloves for Handling Nitrosamines. In:
Proceedings of a Conference on Environmental Aspects of N-Nitroso
Compounds; 1977 August 22-24; Durham, NH. pp. 535-543.
17. Cough, T.A.; Webb, K.S.; McPhail, M.F. Diffusion of Nitrosamines Through
Protective Gloves. In: Proceedings of a Conference on Environmental
Aspects of N-Nitroso Compounds; 1977 August 22-24; Durham, NH.
pp. 531-534.
18. Sansone, E.B.; Tewari, Y.B. The Permeability of Laboratory Gloves to
Selected Nitrosamines. In: Proceedings of a Conference on Environmental
Aspects of N-Nitroso Compounds; 1977 August 22-24; Durham, NH.
pp. 517-529.
19. Henry III, N.W.; Schiatter, C.N. The Development of a Standard Method for
Evaluating Chemical Protective Clothing to Permeation by Hazardous
Chemicals. Am. Ind. Hyg. Assoc. J. 42(3): 202-207; 1981.
20. Weeks, Jr., R.W.; McLeod, M.J. Permeation of Protective Garment Material
by Liquid Benzene and by Tritiated Water. Am. Ind. Hyg. Assoc. J.
43(3): 201-211; 1982.
21. Mikatavage, M.; Que Hee, S.S.; Ayer, H.E. Permeation of Chlorinated
Aromatic Compounds Through Viton and Nitrile Glove Materials. Am. Ind.
Hyg. Assoc. J. 45(9): 617-621; 1984.
22. Nelson, G.O.; Lum, B.Y.; Carlson, G.J.; Wong, C.M.; Johnson, J.S. Glove
Permeation by Organic Solvents. Am. Ind. Hyg. Assoc. J. 42(3): 217-225;
1981.
23. Abernathy, R.N.; Cohen, R.B.; Shirtz, J.J. Measurements of Hypergolic
Fuels and Oxidants1 Permeation Through Commercial Protective
MaterialsPart I: Inhibited Red Fuming Nitric Acid and Unsymmetrical
DimethyIhydrazine. Am. Ind. Hyg. Assoc. J. 44(7): 505-513; 1983.
24. Johnson, T.C.; Merciez, W.D. Permeation of Halogenated Solvents Through
Drybox Gloves. Report RFP-1608. Prepared by the Dow Chemical Company on
Contract AT(2901)-1106 for the Atomic Energy Commission; 1971 April 16.
98
-------
25. Sansone, E.B.; Tewari, Y.B. Penetration of Protective Clothing Materials
by 1,2-Dibromo-3-Chloropropane, Ethylene Dibromide, and Acrylonitrile.
Am. Ind. Hyg. Assoc. J. 39(11): 921-922; 1978.
26. Calingaert, G.; Shapiro, H. Permeability of Protective Glove Materials to
Tetraechyllead and Ethylene Bromide. Ind. and Eng. Chen. 40(2): 322-335;
1946.
27. Sansone, E.G.; Tewari, Y.B. Differences in the Extent of Solvent
Penetration Through Natural Rubber and Nitrile Gloves from Various
Manufacturers. Am. Ind. Hyg. Assoc. J. 41(7): 527-528; 1980.
28. McFee, D.R. How Well Do Gloves Protect Hands Against Solvents? A.S.S.E.
J. IX(5): 1-16; 1964.
29. Sansone, E.B.; Jonas, L.J. Resistance of Protective Clothing Materials to
Permeation by Solvent "Splash". Environ. Res. 26: 340-346; 1981.
30. Sansone, E.B.; Tewari, Y.B. The Permeability of Protective Clothing
Materials to Benzene Vapor. Am. Ind. Hyg. Assoc. J. 41(3): 170-174;
1980.
31. Hogstedt, C.; Stahl, R. Skin Absorption and Protective Gloves in Dynamite
Work. Am. Ind. Hyg. Assoc. J. 41(5): 367-372; 1980.
32. Williams, J.L.; Hopfenberg, H.B.; Stannett, V. Water Transport and
Clustering in Poly(Vinyl Chloride), Poly(Oxymethylene), and Other
Polymers. J. Macroool. Sci. B3(4): 711-725; 1969.
33. Sweeney, R.F.; Rose, A. Factors Determining Rates and Separation in
Barrier Membrane Permeation. I & EC Prod. R. & D. 4(4): 248-251; 1965.
34. Simril, V.L.; Hershberger, A. Permeability of Polymeric Films to Organic
Vapors. Mod. Plas.: 97-102, 150-158; 1950 June.
35. Aboul-Nasr, O.T.; Huang, R.Y.M. Diffusivity and Solubility of Organic
Vapors in Modified Polyethylene Films; II. Diffusivity Studies. J.
Appl. Polyra. Sci. 23: 1833-1849; 1979.
36. Aboul-Nasr, O.T.; Huang, R.Y.M. Diffusivity and Solubility of Organic
Vapors in Modified Polyethylene Films: III. Solubility Studies. J.
Appl. Polym. Sci. 23: 1851-1862; 1979.
37. Sefcik, M.D.; Schaefer, J.; May, F.L.; Raucher, D.; Dub, S.M. Diffusivity
of Gases .and Main-Chain Cooperative Motions in Plasticized Poly(Vinyl
Chloride). J. Polym. Sci. Polym. Phys. Ed. 21: 1041-1054; 1983.
38. Barrer, R.M.; Skirrow, G. Transport and Equilibrium Phenomena in
Gas-Elastomer Systems: I. Kinetic Phenomena. J. Polym. Sci. 3(4):
549-563; 1948.
99
-------
39. Podk&wka, J.; Puchalik, A. Comparative Evaluation of Diffusion
Coefficients for Gases and Vapors of Organic Substances Through
Polyethylene Membranes Determined by Adsorption and Desorption Upstream
Time Lag Method. J. App. Polym. Sci. 27: 1471-1478; 1982.
40. Mueller, W.J. Permeability of Rubber to Organic Liquids. Rubber Age. .
81: 982-987; 1957.
41. Kulkarni, S.S.; Stern, S.A. The Diffusion of C02, CH^, C2H , and C^ in
Polyethylene at Elevated Pressures. J. Polym. ScL. Polym. Phys. Ed. 21:
441-465; 1983.
42. Lebovits, A. Permeability of Polymers to Gases, Vapors, and Liquids.
Mod. Plas.: 139-150, 194-213: 1966 March.
43. Schott, H. Solubility Parameter, Specific Molar Cohesion, and the
Solubility of Ethylene Oxide in Polymers. Biomaterials. 3: 195-198;
1982.
44. Curry, J.E.; McKinley, M.D. Transient Permeation of Organic Vapors
Through Elastomeric Membranes. J. Polym. Sci. Polym. Phys. Ed. 11:
2209-2225; 1973.
45. Paul, D.R.; Ebra-Lima, O.M. Pressure-Induced Diffusion of Organic Liquids
Through Highly Swollen Polymer Membranes. J. Appl. Polym. Sci. 14:
2201-2224; 1970.
46. Paul, D.R.; Ebra-Lima, O.M. The Mechanism of Liquid Transport Through
Swollen Polymer Membranes. J. Appl. Polym. Sci. 15: 2199-2210; 1971.
47. Paul, D.R.; Paciotti, J.D.; Ebra-Lima, O.M. Hydraulic Permeation of
Liquids Through Swollen Polymeric Networks. II. Liquid Mixtures. J.
Appl. Polym. Sci. 19: 1837-1845; 1975.
48. Paul, D.R.; Garcin, M.; Gannon, W.E. Solute Diffusion Through Swollen
Polymer Membranes. J. Appl. Polym. Sci. 20: 609-625; 1976.
49. Coletta, G.C.; Schwope, A.D.; Arons, I.J.; King, J.W.; Sivak, A.
Development of Performance Criteria for Protective Clothing Used Against
Carcinogenic Liquids; Final Report. Prepared by Arthur D. Little, Inc.,
for the National Institute for Occupational Safety and Health under
Contract 210-76-0130, 1978 October.
50. Fujita, H. Diffusion. In: Encyclopedia of Polymer Science and
Technology, Vol. 5. John Wiley & Sons, Inc., New York; 1966. pp. 65-81.
51. Hayes, M.J.; Park, G.S. The Diffusion of Benzene in Rubber. Part 2~High
Concentrations of Benzene. Trans. Faraday Soc. 52: 949-955; 1956.
Hayes, M.J.; Park, G.S. The Diffusion of Benzene in Rubber. Part 1Low
Concentrations of Benzene. Trans. Faraday Soc. 51: 1134-1142; 1951.
100
-------
52. Crank, J. The Mathematics of Diffusion. Oxford University Press, New
York; 1956.
53. Jost, W. Diffusion in Solids, Liquids, and Gases. Academic Press, New
York; 1952.
54. Crank, J.; Park, G.S. (Ed.). Diffusion in Polymers. Academic Press,
London; 1968.
55. Holstein, T. Westinghouse Research Report 411-9-D.
56. Barton, A.P.M. CRC Handbook of Solubility Parameters and Other Cohesion
Parameters. CRC Press, Inc., Boca Raton, Florida; 1983.
57. Kaxo, H.E., Jr.; Nelson, N.A.; Miedema, J.A. Solubility Parameters for
Predicting Membrane-Waste Liquid Compatibility. In: Proceedings of the
Eleventh Annual Research Symposium on Land Disposal of Hazardous Waste;
1985 April 29-May 1; Cincinnati, OH.
58. Waters, M.J. Laboratory Methods for Evaluating Protective Clothing
Systems Against Chemical Agents. Report CRDC-SP-84010. Prepared by the
Chemical Research and Development Center, 1964 June.
59. Hildebrand, J.H. Solubility. XII. Regular Solutions. J. Am. Chera. Soc.
51: 66; 1929.
60. Flory, P.J. Thermodynamics of High Polymer Solutions. J. Chem. Fhys. 9:
660; 1941. Flory, P.J. Thermodynamics of High Polymer Solutions. J.
Chem. Phys. 10: 51; 1942.
61. Huggins, M.L. Solutions of Long Chain Compounds. J. Chem. Phys. 9: 440;
1941.
62. Huggins, M.L. Some Properties of Solutions of Long-Chain Compounds. J.
Phys. Chen. 46: 151; 1942
63. Huggins, M.L. Theory of Solutions of High Polymers. J. Am. Chem. Soc.
64: 1712; 1942.
64. Henriksen, H. Selection of Materials for Safety Gloves, Polymer Membranes
to Protect Against Contact with Epoxy Products. Directorate of Labor
Inspection, Copenhagen; 1982. (Translation from National Institute for
Occupational Safety and Health, Morgantown, WV).
65. Hansen, C.; Beerbower, A. Solubility Parameters. In: Kirk-Othmer:
Encyclopedia of Chemical Technology, Supplement Vol., 2nd ed. Wiley &
Sons, New York; 1971. pp. 889-910.
66. Prausnitz, J.M. Molecular Thermodynamics of Fluid Phase Equilibria.
Prentice-Hall, Inc., Engleuood Cliffs, New Jersey; 1969.
101
-------
67. Heil, J.F.; Prausnitz, J.M. Phase Equilibria in Polymer Solutions. AIChE
J. 12(4): 678-685; 1966.
68. Abrams, D.S.; Prausnitz, J.M. Statistical Thermodynamics of Liquid
Mixtures: A New Expression for the Excess Gibbs Energy of a Partly or
Completely Miscible Systems. AIChE J. 21(1): 116-128; 1975.
69. Fredenslund, A.; Jones, R.L.; Prausnitz, J.M. Group-Contribution Esti-
mation of Activity Coefficients in Nonideal Liquid Mixtures. AIChE J.
21(6): 1086-1099; 1975.
70. Fredenslund, A.; Gmehling, J.; Rasmussen, P. Vapor-Liquid Equilibria
Using UNIFAC: A Group-Contribution Method. Elsevier Scientific
Publishing Company, New York; 1977.
71. Skjold-Jorgensen, S.; Kolbe, B.; Gmehling, J.; Rasmussen, P. Vapor-Liquid
Equilibria by UNIFAC Group Contribution. Revision and Extension. Ind.
Eng. Chem. Process Des. Dev. 18(4): 714-722; 1979.
72. Gmehling, J.; Rasnussen, P.; Fredenslund, A. Vapor-Liquid Equilibria by
UNIFAC Group Contribution. Revision and Extension. 2. Ind. Eng. Chem.
Process Des. Dev. 21: 118-127; 1-982.
73. Macedo, E.A.; Weidlich, U.; Gmehling, J.; Rasmussen, P. Vapor-Liquid
Equilibria by UNIFAC Group Contribution. Revision and Extension. 3.
Ind. Eng. Chem Process Des. Dev. 22(4): 676-678; 1983.
74. Wilcox, R.F. Expanding the Boundaries of UNIFAC. Chemical Engineering,
October 17, 1983; 66-68.
75. Oishi, T.; Prausnitz, J.M. Estimation of Solvent Activities in Polymer
Solutions Using a Group-Contribution Method. Ind. Eng. Chero. Process Des.
Dev. 17: 333-339; 1978.
76. Tseng, H.S.; Lloyd, D.R. Influence of Polymer Molecular Weight on
Selected Thermodynamic Properties of Polymer/Solvent Systems and the
Application of the UNIFAC theory. Polymer. 25: 670-680; 1984.
77. Georgeton, G.K.; Sommerfeld, J.T. Estimating vapor-liquid equilibria with
UNIFAC. Chemical Engineering, October 17, 1983; 61-65.
78. Magnussen, T.; Rasmussen, P.; Fredenslund, A. UNIFAC Parameter Tables for
Prediction of Liquid-Liquid Equilibria. Ind. Eng. Chem. Process Des. Dev.
20: 331-339; 1981.
79. Reid, R.C.; Prausnitz, J.M.; Sherwood, T.K. The Properties of Gases and
Liquids, 3rd ed. McGraw-Hill, New York; 1977.
80. Vrentas, J.S.; Duda, J.L. Diffusion in Polymer-Solvent Systems. I.
Reexamination of the Free-Volume Theory. J. Poly. Sci. Polym. Phys. Ed.
15: 403-416; 1977.
102
-------
81. Vrentas, J.S.; Duda, J.L. Diffusion in Polymer-Solvent Systems. II. A
Predictive Theory for the Dependence of Diffusion Coefficients on
Temperature, Concentration and Molecular Weight. J. Polym. Sci. Polym.
Phys. Ed. 15: 417-439; 1977.
82. Vrentas, J.S.; Duda, J.L. Diffusion in Polymer-Solvent Systems. III.
Construction of Deborah Number Diagrams. J. Polym. Sci. Polym. Phys. Ed.
15: 441-453; 1977.
83. Paul, C.W. A Model for Predicting Solvent Self-Diffusion Coefficients in
Nonglassy Polymer/Solvent Solutions. J. Polym. Sci. Polym. Phys. Ed. 21:
425-439; 1983.
84. Chalykh, A.E.; Vasenin, R.M. Diffusion in Polymer-Solvent Systems.
Polym. Sci. USSR. 8: 1908-1916; 1966.
85. Rehage, G.; Ernst, 0.; Fuhrmann, J. Fickian and Non-Fickian Diffusion in
High Polymer Systems. Discuss. Faraday Soc. 49: 208-221; 1970.
86. Berry, G.C.; Fox, T.G. The Viscosity of Polymers and Their Concentrated
Solutions. Adv. Polym. Sci. 5: 261; 1968.
87. Loflin, T.; Mclaughlin, E. Diffusion in Binary Liquid Mixtures. J. Phys.
Chem. 73: 186-190; 1969.
88. Cohen, M.H.; Turnbull, D. Molecular Transport in Liquids and Glasses. J.
Chem. Phys. 31: 1164-1169; 1959.
89. Dullien, F.A.L. Predictive Equations for Self-Diffusion in Liquids; A
Different Approach. AIChE J. 18: 62-70; 1972.
90. Haward, R.N. Occupied Volume of Liquids and Polymers. Macromol. Sci.
Revs. Macromol. Chem. C4: 191-202; 1970.
91. Flory, P.J. Thermodynamics of Polymer Solutions. Discuss. Faraday Soc.
49: 7-29; 1970.
92. Likey, P.E.; Gambill, W.R. Physical and Chemical Data. In: Chemical
Engineers' Handbook. McGraw-Hill, New York; 1973. p. 3-212.
93. Lange's Handbook of Chemistry, Physical Properties. McGraw-Hill, New
York; 1979. pp. 10-127 and 10-128.
94. Rodriguez, F. Principles of Polymer Systems. McGraw-Hill, New York;
1982.
95. Smith, J.M.; Van Ness, H.C. Introduction to Chemical Engineering
Thermodynamics, 3rd ed. McGraw-Hill, New York; 1975. pp. 569-570.
96. Viscosity: Pure Liquids. In: International Critical Tables, Vol. VII.
McGraw-Hill, New York; 1930. p. 219.
103
-------
97. Selected Values of Physical and Thermodynaraic Properties of Hydrocarbons
and Related Compounds. American Petroleum Institute Project 44, Carnegie
Press, Pittsburgh, PA; 1953.
98. Gerald, C.F. Applied Numerical Analysis, 2nd ed. Addison-Wesley
Publishing Company, Reading, MA; 1980. pp. 390-432.
104
-------
APPENDIX A
SUMMARY OF TEST METHODS FOR EVALUATING PROTECTIVE MATERIALS
SUMMARY OF MECHANICAL-PROPERTIES TESTS
ASTM Method D751-79 -- Standard Methods of Testing
Coated Fabrics
The purpose of this method, first published in 1943, is to test coated
fabrics for a vide variety of mechanical properties. The method describes
standard procedures for determining length, width, thickness, and mass of
coated fabric samples, breaking strength, elongation, bursting strength, tear-
ing strength (by both a pendulum method and a tongue tear method), hydrostatic
resistance, adhesion of coating to fabric, tack-tear resistance, low-
temperature bend strength, low-temperature impact strength, and seam strength.
In terms of glove evaluation, Method D751-79 is directly applicable to the
evaluation of fabric-supported rubber gloves.
ASTM Method D412-83 Standard Test Methods for Rubber
Properties in Tension
This method, first published in 1935, is used to determine the tensile
properties of rubber at various temperatures. The method describes the
specifications of the testing machine (such as an Instron tester) and the teat
chamber, the preparation of the test specimens, and the procedures for
determining tensile strength, tensile stress, ultimate elongation, and tensile
set. The method is directly applicable to the evaluation of unsupported rubber
glove material or elastomeric sheets or films.
ASTM Method D16B2-64 -- Standard Test Methods for Breaking
Load and Elongation of Textile Fabrics
This test method, first issued in 1959, is used to determine the breaking
load and elongation of textile fabrics with a tensile test machine using the
grab, raveled-strip, and cut-strip methods. The grab test is a tension test in
which only a part of the width of the fabric specimen is gripped in the clamps
of a tensile testing machine. The raveled-strip test is a tension test in
which the full width of the specimen is gripped in the clamps and the specified
specimen width is secured by raveling away yarns. The cut-strip method is a
tension test in which the full width of the specimen is gripped in the clamps
and the specimen width is secured by cutting the fabric. This method is not
directly applicable to the evaluation of glove samples but may have some
utility in evaluating the fabric portion of fabric-supported gloves.
ASTM Method D2261-83 Standard Test Method for Tearing Strength
of Woven Fabrics by the Tongue (Single Rip) Method (Constant-Rate-
of-Extension Tensile Testing Machine)
Originally issued in 1964, this method describes procedures for the deter-
mination of the tearing strength of woven fabrics by the tongue (single rip)
105
-------
method using a recording constant-rate-of-extension (CRE) tensile testing
machine. In the method, tearing strength is defined as the force required to
continue or propagate a lengthwise tear started previously in the specimen.
The method may be applicable to the evaluation of the fabric portion of fabric-
supported gloves.
ASTM Draft Test Method Fxxx Test Method for Resistance to Cut
This draft ASTM test method describes procedures for determining the
resistance of a rectangular test specimen (either a single layer or a
composite material) to static cut by measuring the force required to cause a
sharp-edged blade to cut the surface of the test specimen. The test method
defines the test procedure, test apparatus, blade dimensions, the size and
conditioning of the test specimen, and the positioning of the test specimen on
the test apparatus.
ASTM Draft Test Method Fxxx A Test Method for Resistance to
Puncture
This draft ASTM test method describes procedures for determining the
resistance of a rectangular test specimen (either a single layer or a
composite material) to puncture by measuring the force required to cause a
pointed penetrometer to puncture the material specimen. The test method
defines the test procedure, the test apparatus, the dimensions of the
penetrometer, and the test specimen size, condition, and position in the test
apparatus.
ASTM Method D4157-82 Standard Test Method for Abrasion
Resistance of Textile Fabrics (Oscillatory Cylinder Method)
This method defines a standard procedure for measuring the abrasion resis-
tance of textile fabrics by subjecting the test specimen to unidirectional
rubbing action under known conditions of pressure, tension, and abrasive
action. The test is conducted in a special apparatus, described in the method,
that contains an oscillating cylinder section. The method may be useful in
evaluating the fabric portion of fabric-supported protective gloves.
ASTM Method D1388-64 Standard Test Methods for Stiffness of Fabrics
Originally issued in 1956, ASTM Method D1388-64 describes two test
methods for determining the stiffness of fabrics, particularly woven fabrics:
the cantilever test and the heart loop test. Both methods are based on the
bending of a fabric in one plane under the force of gravity. The method may be
useful in evaluating the fabric portion of fabric-supported protective gloves.
ASTM Method D3041-79 Standard Method for Testing Coated
Fabrics Ozone Cracking in a Chamber
This method defines a standard procedure for determining the resistance of
elastomer-coated fabrics to cracking when exposed to an atmosphere containing
ozone. Each test specimen is kept under a controlled surface strain, and the
ozone concentration in the test chamber is maintained at a fixed value. The
106
-------
method is directly applicable to the evaluation of fabric-supported rubber
gloves.
ASTM Method Dl149-81 Standard Method for Rubber Deterioration
Surface Ozone Cracking in a Chamber (Flat Specimen)
Originally issued in 1951, this method defines a standard procedure for
determining the resistance of vulcanized rubber to cracking when exposed to
an atmosphere containing ozone. Each rubber specimen is kept under a surface
tensile strain, and the ozone concentration in the test chamber is maintained
at a fixed value. The method is directly applicable to the evaluation of
unsupported rubber gloves.
ASTM Method G26-83 Standard Practice for Operating
light-Exposure Apparatus {Xenon-Arc Typ^e) With and
Without Mater for Exposure of Nonmetallic Materials
This method is a combination of two previous ASTM MethodsG26 and G27.
The method describes the basic principles and operating procedure for
exposing samples of nonmetallic materials to ultraviolet radiation with a
xenon-arc light source. The method is concerned only with the exposure method
and does not cover sample preparation, test conditions, or evaluation of
results. The method is applicable to the evaluation of both supported and
unsupported rubber gloves.
SUMMARY OF CHEMICAL-RESISTANCE TESTS
ASTM Method F739-81 Standard Test Method for Resistance
of Protective Clothing Materials to Permeation by Hazardous
Liquid Chemicals
This relatively recent method defines standard test procedures for deter-
mining the resistance of protective-clothing materials to permeation by
hazardous liquid chemicals in direct, continuous contact with the normal outer
surface of the material specimen. The permeation resistance of the test speci-
men is determined by measuring the breakthrough time of the challenge chemical
through the test sample and then monitoring the subsequent permeation rate of
the chemical through the sample.
The method specifies the use of a specially constructed glass
permeation-test cell. When mounted in the test cell, the material specimen
acts as a barrier separating the liquid challenge chemical from a collecting
medium. The collecting fluid, either a liquid or a gas, is sampled and
quantitatively analyzed for hazardous permeant as a function of time after
initial liquid contact. Both the initial breakthrough time and the permeation
rate of the hazardous chemical are determined from the time-dependent chemical
analysis of the collecting fluid by means of direct calculations or graphical
analysis.
107
-------
ASTM Draft.Method F739-8X (Revision 4) Standard Test Method
for Resistance of Protective Clothing Materials to Permeation
by Liquids or Gases
This draft test method is an extension of Method F739-81 that incorporates
a standard test procedure for determining the resistance of protective
clothing materials to permeation by potentially hazardous gaseous chemicals in
continuous contact with the normal outer surface of the material. Otherwise
the method is essentially identical to Method F739-81. An earlier revision of
the draft test method included as an appendix a statistical procedure for
determining the equivalency of permeation-test cells of different designs.
Under the current revision of test method F739-81, the equivalency appendix has
been deleted from the method and is slated to be developed into a separate test
method.
ASTM Draft Method F739-8X (Draft 3) Standard Test Method
for Resistance of Protective Clothing Materials to Permeation
by Liquid, Liquid Splashes, and Gases
This is the latest draft revision of Method F739-81. The method includes
a standard procedure for determining the resistance of protective-clothing
materials to permeation by liquid chemicals in intermittent contact with the
normal outer surface of the material. The remainder of the method is essenti-
ally identical to Draft Method F739-8X (Revision 4) discussed above.
Permeation Test Methods Contained in CRDC-SP-84010 Laboratory
Methods for Evaluating Protective Clothing Systems Against
Chemical Agents
CRDC-SP-84010 is a special publication (58) issued by the US Army Chemical
Research and Development Center. The publicaTIon specifies standard test
procedures for evaluating the resistance of protective clothing materials to
permeation by chemical-warfare agents. The methods specify test cells design,
test procedures, analytical methods, and criteria for interpreting the test
results.
ASTM Draft Method F903 (Revision 7) New Standard Test Method
for Resistance of Protective Clothing Materials to Penetration"
by Liquids
This proposed test method defines standard procedures for determining the
resistance of protective clothing materials to visible penetration by liquids
in direct, continuous contact with the normal outer surface of the material.
The method determines resistance to penetration only and not the resistance to
permeation or chemical degradation. The method involves mounting a sample of a
protective material in a specially designed test cell so that the sample
divides the cell into two chambers. The normal outer surface of the sample is
then exposed to a liquid chemical under pressure (2 psig), and the normal inner
surface of the sample is observed (through the transparent cover plate of the
test cell) for visible penetration of the liquid.
108
-------
ASTM Draft Method Fxxx Test Method for Evaluating Protective
Clothing Materials for Resistance to Degradation by Liquid
Chemicals
This proposed test method is a semiquantitative method for determining
the resistance of protective-clothing materials to degradation by liquid
chemicals in direct, continuous contact with the normal outer surface of the
material. The method consists of measuring the thickness, weight, and elonga-
tion of a material specimen, exposing separate identical specimens of the same
material to a liquid chemical, and measuring the thickness, weight, and elonga-
tion of the additional specimens to identify changes resulting from contact
with the liquid chemical.
ASTM Method D471-79 Standard Test Method for Rubber Property
Effect of Liquids
Originally published in 1937, this well-established method measures the
comparative ability of rubber and elastomeric materials to withstand the
effect of liquids. The method involves immersing a sample of a material in a
liquid chemical at a constant temperature and measuring selected physical
properties of the sample as a function of immersion time. Any deterioration of
the material sample is determined by noting the changes in physical properties
before and after immersion in the test liquid over various time intervals. The
physical properties that are measured during the test are weight, volume,
thickness, soluble extracted matter, tensile strength, elongation, hardness,
breaking strength, burst strength, tear strength, and adhesion (for coated
fabrics).
109
-------
APPENDIX B
THE DERIVATION OF THE MODIFIED PAUL MODEL AND A COMPUTER PROGRAM
DERIVATION OF THE MODIFIED PAUL MODEL
The Paul model for predicting the solvent self-diffusion coefficient is
based on free-volume theory. A self-diffusion coefficient may be defined as
the diffusion coetficient of a component within itself. No concentration
gradient exists. A mutual-diffusion coefficient may be defined as the diffu-
sion coefficient of a component within another, for example, a solvent within
a polymer. It can be shown that (87) :
D* [f(D1,x1,D2,x2)/RTl Ouj/aln XI)T(P (32)
where D is the mutual diffusion coefficient; D. and D. are the solvent and
polymer self-diffusion coefficients, respectively; x1 and x2 are solvent and
polymer mole fractions, respectively; R is the universal gas constant; T is the
absolute temperature; u, is the solvent chemical potential; and P is the pres-
sure. The quantity f(D1,xI,D2,x2) represents a function yet to be determined.
Vrentas and Duda (80-8£) proposed that because DZ is much smaller than DI
for most systems, D2 may be neglected in f(DL,x ^D-.Xj) over a large concentra-
tion interval from almost pure polymer to about 85* solvent in some systems.
In this range, they suggest the use of:
D » (x^/RTKai^/ain a^x p (33)
Paul's model, an extension of the Cohen- Turnbull model (88) , is based on the
following expression:
Dj = J DQ1 Tl/2 exp (-YV*/V£) (34)
where J is the jump-back factor described below, D. is a constant evaluated
from experimental data, and y is a numerical factor between 1/2 and 1, which
accounts for the fact that a given free-volume element is available to more
than one molecule. The term V* is the specific critical volume for diffusion,
which is the minimum free-volume element necessary for diffusion, and Vj is the
specific free volume of the mixture. The parameters J, DQ., y, v., and v^
must be evaluated.
The jump-back factor, J, represents the ratio of the probability in the
mixture that a solvent diffusive jump is successful to the probability in a
pure solvent that a diffusive jump is successful. A successful jump by a
solvent molecule is defined as one which is not immediately followed by a
110
-------
return jump to its original position. Analytically,
z
J -[(*+!)/«] [n/(n*D] P(s=n) (35)
where z is the number of sites available to a diffusing solvent molecule
(usually 4!>] l"»2z'1 <36>
Figures which graphically represent Equation 36 can be found in the litera-
ture (£3). It should be noted that J(z=10)/J(z=4) is less than 2 for ^jO.05.
This implies a weak dependence of J on z except at very low solvent concentra-
t ions .
The parameters D , v"f , and the product *p* may be evaluated using vis-
cosity and specific-volume data for the pure solvent. Only a limited amount
of binary data are also needed, but these may be approximated from equations
of state. Paul (83) used a correlation presented by Dullien (89):
H,D° - (RT/v.) (0.124 x 10-16 mol2/3) v|/3 (37)
ill * i
where n, is the solvent viscosity at temperature T, D^ is the self-diffusion
coefficient of pure solvent, v is the solvent molar volume at temperature T,
and v is the solvent critical molar volume. For pure solvent (*!), Equation
34 reduces to:
D0 Dfli Ti/2 exp {^/[Vj - v^O K)] } (38)
Note that
vf Vj - v^O K) (39)
has been used. Equation 39 implies that the free volume is identically zero
at 0 K (absolute zero). The value of v^O K) may be estimated by a. variety of
methods ^90). Substituting Equation 38 into Equation 37 and
taking the ratio of the resulting equation evaluated at T to the same
equation evaluated at Tref, some reference temperature, yields:
-------
cosity and specific-volume (or molar-volume) data are present in the literature
for most solvents. However, these data would have to be determined if they are
unavailable.
Knowing >v* we can evaluate D from Equations 37 and 38. The only
remaining parameter is v"f which may oe expressed as:
5f = v - v(0 K> (41)
where v is the specific volume for the mixture.
If v is not available in the literature, a rough estimate of this quantity
may be obtained using the Flory-Prigogne theory (91) of excess volume:
VE = v - (w.vO + w,vO)
11 2 2 (42)
where VE is the excess specific volume and *>l and w2 are the weight fractions
of the solvent and the polymer, respectively. The terms vj and v2 are the
specific volume of pure solvent and pure polymer, respectively.
Rough estimates are also possible by setting VE = 0. This may be done
because generally at room temperature, the maximum value of VE is less than 1%
of the actual volume whereas v^ is about 152. Increasing the temperature
increases VE but vf increases more rapidly (83) . For the calculations
presented in this report, VE was assumed to be zero.
One drawback to Che Paul model is Chat the value of D calculated is less
accurate when the solvent volume fraction, 4^, is less than 0.1. This is
because the choice of z affects J more strongly at small solvent volume frac-
tions than at large volume fractions. More significantly however, the polymer
segments may contribute substantially to the refilling of voids even if solvent
molecules are nearby; thus, an assumption made in Paul's derivation of
Equation 34 is invalidated (£3). Under such conditions, that is, for $^0.1,
the Vrencas-Duda model should be used.
After the calculation of D., the only remaining quantity to be evaluated
in order to obtain D is the derivative shown in Equation 33. This may be
calculated by noting that:
Xj)TjP RTCSln a^Bln x^p (43)
where a is the activity of the solvent in the mixture.
The expression on the right-hand side of Equation 43 may be evaluated from
a -versus-x. data generated by the UNIFAP program. This information can then
be used with D. and Equation 33 to calculate the mutual-diffusion coefficient,
D.
112
-------
COMPUTER PROGRAM FOR THE CALCULATION OF D
A computer program based on the Paul model was written using FORTRAN 77,
and a DEC 2040 computer was used to perform the series of calculations needed
to estimate the diffusion coefficient, D. A copy of the program is included
at the end of this section. Because time limitations, the software was not
made "user friendly"; however, it can be easily followed.
The first parameter to be calculated by the program is the quantity -yv*.
Viscosity-versus-temperature and specific-volume-versus-temperature data in
Equation 40 are used by a linear-regression subroutine. Next Equations 37 and
38 are used to calculate the constant D . The jump-back factor is calculated
by Equations 35 and 36. UNIFAP data are used to calculate the derivative in
Equation 32 by noting the relationship shown in Equation 43. Next, D is calcu-
lated using Equation 32. A library graphics package is used to plot D as a
function of solvent weight fraction. Tables 21 through 24 show the data used
in the calculations presented in Section 6 of this report.
113
-------
TABLE 21. VISCOSITY AND SPECIFIC VOLUME OF BENZENE
AS A FUNCTION OF TEMPERATURE
Temperature, K
284C
293
303
313
323
333
343
353
Viscosity,8 cp
0.75
0.65
0.57
0.50
0.66
0.39
0.35
0.31
Specific volume, cm3/g
1.1254
1.1376
1.1516
1.1660
1.1809
1.1963
1.2124
1.2291
aFrom Reference 92.
bFrom Reference 93.
cData used as the reference condition in Equation 40.
114
-------
TABLE 22. OTHER PARAMETERS USED IN THE CALCULATION OF DIFFUSION
COEFFICIENTS FOR BENZENE IN NATURAL RUBBER
Parameter
-0
vl
-0
V2
Vj <0 K)
5 2 (0 K)
Cl
Tinc
MW1
MM,
z
Value
1.1109 cn3/g
1.0753 co3/g
0.9115 cm3/g
-0.9 cm3/g
258.7 cm3/g
298 K
78.11
68.12a
6
Reference
93
94
90
90
95
aGiven as the molecular weight of a monomer unit
115
-------
TABLE.23. VISCOSITY AND SPECIFIC VOLUME OF n-HEPTANE
AS A FUNCTION OF TEMPERATURE
Temperature, K
289C
300
311
322
333
344
355
366
378
389
Viscosity,* cp
0.2271
0.2156
0.2045
0.1947
0.1857
0.1775
0.1700
0.1631
0.1567
0.1508
Specific volume, crnVg
1.4544
1 .4748
1.4960
1.5181
1.5412
1.5653
1.5907
1.6174
1.6456
1.6754
aFrom Reference 96.
bFrora Reference 97.
cData used as the reference condition in Equation 40.
116
-------
TABLE 24. OTHER PARAMETERS USED IN THE CALCULATION OF DIFFUSION
COEFFICIENTS FOR n-HEPTANE IN NATURAL RUBBER
Parameter
-0
vl
.0
V2
Vj (0 K)
V2 (0 K)
v
Tint
MW .
MW2
z
Value
1.4327 cn»3/g
1.0753 cm3/g
1.0941 cm3/g
sO.9 cm3/g
432.0 cm3/g
298 K
100.20
68.12a
6
Reference
93
94
90
90
94
aGiven as the molecular weight of a monomer unit.
117
-------
c
c
^"
c
C THIS PROGRAM MILL CALCULATE DIFFUSION COEFFICIENTS FOR VARIOUS
C SOLVENT/POLYMER COMBINATIONS AS A FUNCTION OF SOLVENT
C CONCENTRATION,
C
C
C PROGRAM WRITTEN BY ABHOYJIT BHOHN 8/85
C
C
C
C ALLOCATE VARIABLES
C
INTEGER OUT
REAL TENP(30).VISC(50).SVOLC50),X(50),Y(30),N.J,MW1,NW2,XP(50),
* YP(«0)
INTECEB Z
R s 8.11441
C
C
C . OPEN DATA FILES
C
OPEN (WHIT»1.FILE='DIFFDAT.DAT',T»PE«'OLD')
OPES (UNITsZ.FILEs'UNIFDAT.DAT'.TYPE-'OLD1 .DEVICCo'DSK')
C
C
C INPUT DATA
C
READ (1,*) N
DO 100 IBI.N
100 READ (I,*) TEKPCI),YI3Cm,SYOLU)
READ ti,«) vio,V20,viox,v30K,vic
READ (1.*) TREF,VREF,SREF,TXNT,MM1,MW2,Z
READ (1,*) Hia.HlE.Wll
C
c
C. -.-.CALCULATE CAMHA-VSTAR
C
00 200 I»1,N
X(I) 1/(SVOL(X)-V10K) - 1/CSREF-V10K]
200 Yd) LOG (VISCtD/VREF SVOLd)/SREF SORT(TREF/TEKP(X) ))
CALL LINREC (N,X. Y. SLOPE, YINT.CORR)
GAVS « SLOPE
C
C
C CALCULATE CONSTANT D01
C
00 100 I«1,N
X(I) SORT (TEMPtin EXP(-CAVS/(SVOL(I).V10K))
JOO Yd) « 0.124E-16 R TENP(I) / SVOL(I) / VISC(I) V1C«»C2./J.)
* 1.0E9 / HU1
CALL LINREC (h.X. Y. SLOPE. YXHT.CORR)
D01 SLOPE
C
C
C.. -START CONCENTRATION-DEPENDANT DIFFUSION COEFFICIENT CALCULATIONS
C
118
-------
WRITE (1,1)
1C H 0
00 400 H1BU1B,U1C,W1I
1C 1C + 1
c
c
C.......READ UN1FAP DATA AND CALCULATE DERIVATIVE OF A VS X
C
350 REAR (2,»> HEI1.ACT1
IF (NISTdOO.«£!!).NE.NINT(100.»W1)} GOTO 350
READ (2,*) VEI2.ACT2
XUEIl B Meil/MWl/(MEIl/MHl+(l.wEIl)/HW2)
XWEI2 B WEI2/MU1/(UEI2/HU1+(1»UEI2)/MU2)
DERV B (LOGCACT1)-LOG(ACT2)) / (LOGCXNEI1)-LOC(XWEI2))
VF HI (V10-V10K) * (l-Hi)*(V20-V20K)
VFRAi M*V10/(W1*V10*(1-W1)«V20)
C
C
C.....CALCULATE J
C
CALL FACT (Z.I1)
SUN 0.
DO 170 1=1,Z
CALL FACT (1,12)
CALL FACT (Z-I.I3)
P B FLOATCIl/I2/I31»vrPAl»»I»(l,-VFRAn»»(l-I3
370 SUN B SUN * I/(I»1.)»H
J (I*i,)/I*SUH
C
c
C.CALCULATE 0
C
Di
XI
D
XP(IC)
YP(IC)
J*D01*SORT(TINT)*EXP(.6AVS/VF)
tl.-Xl)*Dl«DERV
VFRAI
D
400 WRITE (3,2) Kl.VFRAI,D
C
C
C--FORMAT STATEMENTS
C
1 rOR»AT (IX.'WEIGHT FRACTION',5X,'VOLUHt FRACTION',5X.
* 'HUTUAL-DXFF COEFF (CN»*2/S)>)
2 FORMAT (1X,F10.4,10X,F10.4,15X.E14.7)
C
C
C-....CLOSE FILES, CALL PLOTTING ROUTINE, AND END PROGRAM
C
CLOSE CUNITBl)
CLOSE CUNITBP2)
CALL PLOTS (0,0,6)
CALL PLOT (0.0,0.0,-))
CALL SAML06 (XP.YP.IC,1,1,1,1,1,II)
CALL PLOT (12.0,0.0,-999)
END
C
C
C.....LINEAR REGRESSION SUBROUTINE
C
119
-------
SUBROUTINE LIMBEC (DN,OX,DT,DSLOPC,DTINT,DCORR>
C
C--ALLOCATE AND RESET VARIABLES
DIMENSION DX(SO).DY(SOJ
SUMX 0.
o.
0,
SUHX2 0.
5UHVJ 0.
C
C.....CALCULATE VARIOUS SUMS
C
00 100 iBl.ON
SUMX « SUMX + DX(I)
SUMY SUMY * DY(I)
SUMXY s SUNXY * DX(I)*OY(I}
SUHX2 5UM2 * t>XCI)»«2
100 5UMY2 SUHT2 4 OYtI)»«2
C
C
C"CALCULATE SLOPE.YINT,CORR
DSLOPE (SUNXY«SUMX«5U)ir/DN)/CSUMX2-SUMX**2/ONl
DYINT c
DCORR OSLOPE'SDEVX/SDEVY
END
C
C--FACTORIAL SUBROUTINE
C
SUBROUTINE FACT tJi,J2)
J2=l
DO 100 J«2,J1
100 J
END
120
-------
APPENDIX C
NUMERICAL METHODS FOR SOLVING DIFFUSION PROBLEMS
Permeation-rate-versus-time data may be calculated using Che fundamental
parameters D and S and analytical solutions of Pick's first and second laws for
given initial and boundary conditions (for example, see Equation 3 in this
report). However, frequently the initial and boundary conditions are complex
and phenomena such as concentration-dependent diffusion must be considered.
For these reasons, an exact analytical solution to a given diffusion problem is
often not possible. Thus, during the current effort, numerical methods for
solving such problems were explored briefly.
Presented below is a trivial example in which the Crank-Nicolson implicit
finite differences method (98) was used to determine the permeation rate
versus time for a given D (which was held constant) and a given C , the
concentration of solvent on one side of a polymeric membrane. The
concentration of solvent in the receiving fluid on the other side of the
membrane was assumed to be zero. The initial concentration of solvent
throughout the membrane was assumed to be zero, although the computer program
listed will accept any valid concentration profile.
The parameters needed for the calculation of permeation rate versus time
are:
n, which is the number of nodes at which the concentration will be
calculated within the membrane.
Cfi, which is the solvent concentration on one side of the membrane,
usually taken as the solubility of the solvent in the polymer.
i, which is the thickness of the membrane.
At, which is the time increment.
The maximum time to which calculations should be carried out.
Table 25 shows results from calculations done by two computer systems, an
Osborne I microcomputer using BASIC and a DEC 2040 using FORTRAN. A comparison
of permeation rates versus time calculated numerically with those calculated
using the exact analytical solution is also given in the table. The results
obtained with the two computers differ slightly for two reasons. First, the
DEC 2040 is based on a higher-bit microprocessor than the Osborne I; this
enables the DEC 2040 to carry operations to higher significant figures.
Second, the method of actually solving the equations set up by the
Crank-Nicolson scheme was different in the two programs. (The program listed
after the table was written for the DEC 2040.)
121
-------
TABLE 25. PERMEATION RATE VERSUS TIME CALCULATED
USING THE CRANK-NICOLSON METHOD*>b>c
^^
Time,
sec
1800
2025
2250
2475
2700
2925
3150
3375
3600
3825
4050
5275
4500
Calculated permeation race
Osborne I
0.06868
0.1084
0.1555
0.2081
0.2643
0.3227
0.3818
0 .4408
0.3988
0.5554
0.6100
0.6625
0.7127
DEC 2040
0.06779
0.1070
0.1539
0.2065
0.2628
0.3214
0.3807
0.4399
0.4981
0.5548
0.6095
0.6621
0.7123
, ng/(cm2.sec)
Analytical
0.06713
0.1067
0.1536
0.2061
0.2623
0.3207
0.3799
0.4389
0.4970
0.5536
0.6083
0.6608
0.7110
Relative difference, %
Osborne I
2.3
1.6
1.2
1.0
0.8
0.6
0.5
0.4
0.4
0.3
0.3
0.3
0.2
DEC 2040
1.0
0.3
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
aThe solubility used was 0.01078 g/cm3, the diffusion coefficient was
6.59 x 10~10 on2/sec, and the thickness of the polymer sample, £, was
0.00472 cm. The initial condition was C(x,0) =0; the boundary conditions
were C(0,t) - 0.01078 g/cm* and CU.t) - 0.
bFor the Osborne I calculation, the number of x intervals was 40 and At was
6.25 sec. The software was written in BASIC. The execution time was
about 5 min.
cFor the DEC 2040 calculation, the number of x intervals was 40 and At was
225 sec. The software was written in FORTRAN. The execution time was
1.96 sec.
"^Thirty terms of the infinite series analytical solution were used in this
calculation.
122
-------
c
c
C THIS PROGRAM WILL CALCUATE CONCENTRATION PROflLES WITHIN A MEMBRANE
C EXPOSED TO CONSTANT BOUNDRY CONDITIONS USING FINITE DIFFERENCES.
C XT WILL ALSO CALCULATE THE FLUX AT X«L,
C
C
PROGRAM WRITTEN BY ABHOYJIT BHOWN 8/85
ALLOCATE VARIABLES
DIMENSION UOLD( 100), UNC«( 100) ,W(100),C( 1001, BBC 100), D( 100)
REAL LBC.L
C
C
C...READ PARAMETERS
C
OPEN (UNITsl.FILFB'PARAN.DAT'.TYPEa'OLD1)
N « 40
LBC * 0.
RBC a 1.0
THICK 1.
L 0.00472
CS 0.01078
C
C
C SET BOUNDARY CONDITIONS
C
UOLDClJaLBC
UOLD(N]«RBC
UNEU(1)BLBC
UNEii(N)BRBC
C
C
C SCT INITIAL CONDITIONS
C
WRITE O.D.O.
WRITE (3,»)
WRITE (3,2).1,UOLD(1)*CS
DO 100 H2.N-1
UOLDCI)«0.
100 WRITE (J,2),I,UOLD(I)*C5
WRITE (3.2),N,UOLO(N)»CS
WRITE (3,»)
WRITE C3,»)
WRITE O.»)
C
C
C --INITIALIZE ITERATION
C
DXFF 6.59E-10
TNAX 4500,*DIFF/L*«2
TINC 225,*DlFF/L»*a
DELX THICK/(N-1)
-------
R TIMC/DELX**2
A R/2
C R/2
C
C
C-START TIKE
C
DO 700 TsTIHIT.-THAX.TIMC
C
C .. -INITIALIZE THOMAS ALGORITHM
C
DO 200 I&2.X-1
200 D(I)
D(2)
W(21
C(2)
88C2)
(R/2*uaLD(I-l)-(R-l)"UOLDrl)+R/2*UOLDCX+l))
D(2)-A»UOLOC1)
D(N-l)-C*UOLDCN)
1/B
W(2)*0(2)
W(2)*C
C
C..........START THOMAS ALGORITHM
C
00 300 1=3,N-l
«a) i/(B-A«BB(x-m
BB(1)
300 G(X) i
UNEW(M-l) Gttl-1)
DO 400 X«N-2,2,-l
400 UNEW(X) a G(X)-BB(X)*UNEW(I+1)
C
C
C...... PRINT VALUES
C
WRITE (3.1),T*L**2/DXFF
WRITE (3.*)
DO 600 IBI.N
WRITE (3,3),I,I»L,UKEW(I)*CS
600 UOLD(X) UKEN(I)
FLUX UULO(2)*OIFF»CS/L/OLLX»1.E9
WRITE (3,«)
WRITE (3.4) FLUX
WRITE (3,*)
WRITE (3.*)
WPXTE (3,*)
700 CONTINUE
C
C
CFORMAT STATMENTS
C
1 FORMAT (IX,'TIME '.F10.0)
2 FORMAT (IX.'NODE '.I3.SX,'FEXLD «',F10.6)
1 FORMAT (IX,'NODE *'.I3,SX,'X =',F10.6,3X,'FEILD .F10.6)
4 FORMAT (U.'FLUX «',F10.6)
C
C
CCLOSE FILES AND END PROGRAM
C
CLOSE (UHXTBl)
END «*
-------
APPENDIX D
SUMMARY OF LIQUID-IMMERSION ABSORPTION
DATA AND PERMEATION-TEST RESULTS
125
-------
TABLE 26. SUMMARY OF LIQUID-IMMERSION ABSORPTION TEST DATA
Dimeter, in
Clove Solvent Teat Number Initial
Butyl Acetone D0220-27-3
rubber 00220-45-3
00220-46-1
00265-93-1
00265-93-3
00265-93-5
Cyclohexane D0220-27-I
00220-45-1
00220-46-1
00265-36
00265-54
00265-74
laopropanol 00220-27-4
00220-45-4
D0220-46-4
00265-93-2
00265-93-4
0026 5-93-6
Toluene 00220-27-2
00220-45-2
00220-46-2
00265-38
00265-56
00265-76
.37
.38
.39
.40
.40
.40
.37
.40
.36
.41
.40
.39
.39
.38
.39
.40
.40
.40
.39
.40
.38
.41
.40
.40
Final
1.41
1.42
1.42
1.42
1.42
1.42
2.24
2.24
2.26
2.18
2.19
2.21
1.41
1.41
1.39
1.40
1.40
1.40
1.98
1.95
1.99
1.92
1.96
1.96
Thickneaa, oil
Initial
25.62
22.5
21.24
21.78
25.6
24.5
24.3
20.4
24.52
23.36
22.94
22.12
24.8
21.0
24.1
26.1
22.4
21.54
22.1
23.1
21.0
22.18
25.48
24.80
Final
25.98
22.9
21.4
22.06
26.08
24.86
36.8
29.54
36.0
35.64
35.12
34.54
25.0
20.7
23.96
26.11
22.36
21.62
29.8
30.2
27.2
29.64
35.32
35.30
Temp.
range,
F
70-76
70-76
70-76
74-79
74-79
74-79
73-76
70-75
71-76
67-68
74-77
76-78
70-76
70-76
70-76
74-79
74-79
74-79
73-76
70-75
71-76
67-71
74-76
76-78
R.H.
range,
X
65-71
63-71
63-65
66-86
66-86
66-86
66-71
63-71
66-71
55-60
62-72
63-75
65-71
63-71
63-65
66-86
66-86
66-86
66-71
63-71
66-71
55-61
62-72
67-74
Init lal
weight,
8
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.7155
.6343
.5971
.6100
.7132
.6921
.6743
.5740
6925
.6622
.6461
.6237
.7012
.5919
6767
.7315
6252
.6064
.6168
6507
.5914
6246
.7068
6999
Max .
weight
gain. Solubility,
8 8/l
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
1
1
.0281
.0257
.0232
.0255
0295
.0287
.6162
.4282
.6738
.5368
.5185
.5209
.0025
0019
0020
.0022
.0025
.0032.
.9771
0132
.9373
9677
3330
.1090
0.0454
0.0466
0.0439
0.0464
0.0457
0.0464
2.7533
2.7753
2.8676
2.5711
2.6240
2.7650
0.0040
0.0037
0 0033
0.0033
0 0044
0.0059
.7780
.7731
.8210
.7051
.7646
.7727
(continued)
-------
TABLE 26 (continued)
Diameter, in
Clove Solvent Test Number Initial
Natural Acetone D0220-II6-1
rubber D0220-130-3
D0220-13I-3
00265-92- 1
D0265-92-3
D0265-92-5
D0265-102
00265-104
D0265-I1I
D0265-120
D0265-122
Cyclohexane D0220-I16-I
D0220-I30-I
D0220-I3I-I
00265-50
00265-66
00265-86
laopropanol 00220-11 6-4
002 20- 130-4
00220-1)1-4
00265-92-2
00265-92-4
00265-92-6
Toluene 00220-116-2
D0220-I30-2
00220-131-2
00265-48
00265-70
00265-88
.40
.39
.40
.40
.40
.40
.40
.40
.40
.40
.40
.41
.40
.40
.40
.40
.39
.40
.40
.40
.40
.40
.40
.40
.40
.40
.40
.40
.40
Final
1.45
1.4)
1.46
1.46
1.47
1.46
1.45
1.45
1.45
1.46
2.23
2.18
2.24
2.21
2.17
2.17
1.43
1.43
1.42
1.42
1.41
1.40
2.28
2.31
2.32
_.
2.34
2.27
Thickneai, mil
Initial
25.14
23.96
23.82
24.9
23.82
26.3
25.40
24.28
25.48
24.66
25.78
25.86
25.46
25.9
25.88
26.78
26.14
26.08
25.86
27.74
24.5
26.3
24.1
25.30
25.34
25.7
26.74
24.38
26.72
Final
26.0
25.06
24.84
25.9
24.80
27.44
..
25.60
26.48
25.52
26.84
38.7
38.20
38.4
40.12
40.9
40.46
26.4
26.70
27.82
25.04
26.6
24.5
38.6
40.74
39.6
.-
40.5
43.52
Temp.
range ,
F
69-73
70-71
70-73
74-79
74-79
74-79
14-76
76
7S-76
78-79
76-77
69-73
70-71
70-73
69-70
73-76
77-78
69-73
70-71
70-73
74-79
74-79
74-79
69-73
70-71
70-73
68-70
71-76
74-78
R.H.
range ,
I
52-80
51-74
51-76
66-86
66-86
66-86
80-88
78-82
73-76
73-74
68-69
52-80
51-74
52-76
66-69
59-61
63-71
52-80
51-74
51-76
66-86
66-86
66-86
52-80
51-74
52-76
67-71
61-75
63-72
Init lal
weight ,
R
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.6466
.6162
.6068
.6332
.6065
.6649
.6531
.6177
.6404
.6161
.6544
.6655
.6528
.6699
.6514
.6674
.6604
.6536
.6502
.6816
6267
.6652
.6141
.6121
.6512
.6567
6537
.6207
6724
Max
weight
Rim.
R
0
0
0
0
0
0
0
0
0
0
0
0
0
0
(1
0
0
2
2
2
2
2
2
.0919
0831'
.0832
.1012
.0955
1099
.0979
0931
.0932
.1037
.1009
8424
.6602
.7526
7640
.8177
.7429
.0)09
0306
.0)19
0)07
.0)4)
0)02
0086
0591
.0732
0637
.1494
.1026
Solubility.
g/cml
0.1449
0.1395
0.1385
0.1611
0.1589
0.1656
0.1528
0.1520
0.1450
0.1667
0.1551
2.7844
2.5850
2.6825
2.7020
2.6907
2 6811
0.0470
0.0465
0.0489
0 0497
o osi;
0.0497
.1472
2212
1979
.0594
.4949
.1194
(continued)
-------
TABLE 76 (continued)
Dimeter, in
Clove Solvent Test Number Initial Final
Neoprene Acetone D0220-71-3
rubber 00220-93-1
D0220-94-3
00265-94-1
00265-94-3
00265-94-5
00265-106
00265-112
00265-118
00265-124
Cyclohexane 00220-73-1
D0220-93-I
00220-96-1
00265-46
D026S-62
00265-82
Iiopropanol 00220-71-4
D0220-93-4
00220-94-4
00265-94-2
00265-94-4
00265-94-6
Toluene 00220- 7 1-2
00220-93-2
00220-94-2
00265-44
00265-64
00265-84
.39
.39
.38
.40
.40
.40
.40
.40
.40
.40
.40
.39
.39
40
.40
.40
.39
.39
.40
.40
.40
.40
.40
.39
.39
.40
.40
.40
.52
.54
.SO
.55
.55
.57
.54
.58
.55
.55
.76
.78
.78
.80
.73
.77
.44
.42
.43
.42
.41
.43
.28
.29
.33
.26
.26
Thickneaa
Initial
18
18
IB
IB
19
19
19
19
18
19
18
18
19
19
19
19
IB
18
19
19
19
IB
19
19
19
19
18
19
.58
.32
.40
.59
.12
.4
.24
.06
.52
.08
.54
.58
.16
.02
.04
.28
.74
.70
.24
.8
.5
.2
.2
.04
.68
.26
.54
.16
, nil
Final
20.12
20.46
20.42
20.74
21.08
21 .36
21 .06
21.64
20.80
20.98
23.06
23.00
24.46
24.04
23.92
24.70
19.1
19.16
19.54
20.12
20.06
19.2
29.1
29.40
30.68
28.92
30.20
Temp.
range.
F
70-76
71-73
70-73
74-79
74-79
74-79
74-75
76
78-79
76-77
71-76
69-73
70-73
69-74
73-76
73-78
70-76
69-73
70-73
74-79
74-79
74-79
71-76
69-73
70-73
68-70
71-75
73-78
R.H.
range.
X
52-73
52-73
52-72
66-86
66-86
66-86
80-87
71-74
73-74
68-69
52-73
53-73
52-72
66-71
62-67
59-63
52-73
52-73
S2-72
66-86
66-86
66-86
52-73
52-73
52-72
65-71
62-73
62-64
Inn ill
weight .
R
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
6224
6206
6195
6310
.6385
6477
6165
6995
6156
6360
.6219
6351
.6463
.6447
.6661
.6491
.6303
.6333
.6512
.6689
.6554
.6358
6490
6441
.6666
.6365
.6141
.6403
Max
weight
gain,
g
0 1662
0 1915
0.1595
0 1976
0 2026
0 2054
0 2013
0 1908
0 1958
0 1993
0.4025
0 4006
0 4167
0 4220
0.4117
0 4119
0.0279
0.0224
0.0232
0.031)
0.0287
0.0110
.5244
.5030
.6134
.4840
.4280
501)
Solubt 1 ity,
g/ca.1
0.3597
0.4204
0.3537
0.4214
0 4200
0 4197
0.4189
0.1968
0.4191
0 4140
0 .8606
0.8670
0.8746
0 8795
0 8988
0 8880
0.0599
0.0482
0.0478
0.0627
0.0581
0.0675
3.1473
3.1745
3 2968
3.0544
3 0533
3.1062
(continued)
-------
TABLE 26 (continued)
Clove
Htcrilc
rubber
Dimeter, in
Solvent Teat Humbei Initial Final
Acetone 00 2 20- SO- J
D0220-5I-4
D0220-52-3
D0265-42
D0265-S8
00265-78
Cycloheiiane 00220-50-1
00220-S1-2
D0220-52-1
D026S-91-I
D0265-91-J
D026 5-91-5
laopropanol 00220-50-4
D0220-SI-I
00220-52-4
D0265-9I-2
00265-91 -4
D026S-9I-6
Toluene DO 2 20- 50- 2
00220-Sl-J
D0220-52-2
D0265-40
D0265-6I
00265-80
.40
.39
.39
.40
.40
.40
.40
.40
.37
.40
.40
.40
.39
.39
.37
.40
.40
.40
.40
.38
.39
.40
.40
.40
.99
.93
.94
.84
.92
.89
.45
.43
.43
.44
.46
.43
.10
.11
.SO
.51
.50
.SI
.88
.85
.89
.81
.86
.84
Thtekneat
Initial
20.7
19.5
21.9
20.40
21.92
23.20
21.4
22.9
22.0
2). 4
21.7
22.3
21.2
23.0
19.4
22.7
21.5
23. S
20. S
20.66
21.7
20.70
24.72
23.54
, .11
Final
29.0
26.3
28. 5
28.06
31 .02
32.02
22.1
23.6
23.1
24.68
22.98
23.0
23.34
2S.72
20.9
2J. 18
23.64
26.14
27.48
28.1
29.7
27.82
32.78
32.04
Temp.
range .
F
70-76
71-7S
70-7)
69-71
74-78
77-78
70-76
70-75
70-75
74-79
74-79
74-79
70-76
70-76
70-75
74-79
74-79
74-79
70-76
71-75
70-76
69 -M
74-77
74-78
R.H.
range,
S
51-72
63-66
63-66
64-70
63-72
66-74
53-72
51-71
51-71
66-86
66-86
66-86
53-72
65-71
SI-M
66-86
66-86
66-86
53-72
63-66
63-66
64-70
62-72
59-64
Initial
Height.
g
0.557S
0.5382
0.5970
0.5318
0.5517
0 6048
0 5713
0.6116
0.5947
0 6305
0 5826
0 5979
0.5768
0.6243
0.5082
0 6153
o 5717
a f>373
0.5545
0 5663
0.5914
0.5617
0.6575
0.6277
M«x .
weight
gain,
8
0 9497
0.9451
1.0851
0.8854
0.9438
1 0154
0.0629
0 0556
0.0404
0.0544
0.0690
0.0463
0.1149
0.1482
0.0971
0.1375
0 1178
0.1557
0.7491
0.7607
0.7966
0.7218
0.8727
O.S4I6
Solubility,
g/cnl
.8187
.9490
.9925
.7205
.7068
7350
0.1165
0.0962
0.0760
0 0922
0.1260
0.0823
0.2180
0.2591
0.2072
0.2401
0 2172
0.2626
.4486
.5022
.4763
.3823
.3995
.4173
(continued)
-------
TABLE 26 (continued)
Dimeter, in Thickneia.
Clove
PolyCvinyl
chloride)
Solvent
Acetone
Cyclohexane
laopropanol
Toluene
Teat Number Initial Final Initial
00220- HO- 3
D0220-1S1-3
DO 220- 150-1
D0220-D1-1
00220-150-4
D0220-15I-4
D0220-150-2
D0220-15I-2
.41
.42
.39
.41
.39
.40
.40
.40
.24
.29
.23
.29
.26
.29
.24
.31
.70
.88
.14
.22
.42
.80
.20
.38
nil
Final
5
4
6
6
II
5
5
4
.24
.6
.04
.2
.80
.68
.18
.14
Temp.
range ,
F
70-71
69-72
70-71
70-72
70-71
70-72
70-71
70-72
R.H.
range ,
I
51-80
52-71
SI-BO
52-71
51-80
52-71
51-80
52-71
Initial
weight .
ft
0 2047
0.1814
0.1573
0.1864
0.2596
0.1778
0.1923
0 1660
Max
weight
gain. Solubility,
K g/cin'
-0.0511 --'
-0.0528
-0 0492
-0 0528
-0 0747
-0.0460
-0.0310
-0.0236
The symbol "--" in the column libeled "Solubility. g/cm3"
aolubility could be calculated.
eani that a weight loss waa observed, thui. no
-------
TABLE 27. SUMMARY OF INDIVIDUAL PERMEATION-TEST RESULTS
Clove
Butyl rubber
Natural rubber
Neoprene rubber
Nitrile rubber
Solvent
Acetone
Cyclohexane
Toluene
Acetone
Cyclohexane
Toluene
Isopropanol
Acetone
Cyclohexane
Toluene
Acetone
Cyctohexane
Toluene
Teat No.
D0301-40
D0301-53
D0301-56
D0301-59
00301-87
DO 30 1-84
D0301-24
00301-27
D0301-28
00301-73
D030I-74
D030I-76
DO 301 -66
D0301-70
D030I-71
D0301-95
D0301-34
D030I-38
D0301-39
D0301-62
D0301-8S
00301-88
D030I-93
D030I-80
00301-44
00301-45
D0301-48
D030I-91
D0301-7B
-t =i
Thickness,
cm
0.0615
0.0607
0.0597
0.0597
0.0625
0.0589
0.0610
0.0630
0.0607
0.0602
0.0617
0.0638
0.0635
0.0617
0.0599
0.0617
0.0688
0.0480
0.0472
0.048
0.0493
0.0485
0.0483
0.0478
0.0572
0.0584
0.0612
0.0607
0.0574
Breakthrough
time,8 min
>1I15
50
50
57
63
24
18
16
18
18
21
IB
10
8
11
>150
20
24
22
60
>60
>I60
135
12
10
8
12
>I350
52
Steady-state
permeation rate.
ug/(cm?*inin)
<0.47
460
565
403
311
396
30
34
37
572
528
476
775
760
812
I
171
149
132
55
16
7
61
642
984
933
969
<0. 12
210
Calculated diffusion
coefficient. cra'/sec
1.
0.
0.
0.
1.
1.
1.
1.
1.
1.
2.
7.
5.
4.
1.
1.
1.
1.
3.
3.
3.
1.
2.
3.
4.
3.
6.
2
93
72
60
9
5
8
7
3
8
5
1
9
3
2
6
3
4
4
3
4
7
6
8
2
7
6
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
10~7
ID"7
lo-'
IO-?
10"
10-'
io- 7
io-7
io- 7
io-7
io-7
lot7
JO"7
io-7
io-9
io-7
IO'7
io-7
JO-"
ID"9
io-'
IO-»
io-7
io-7
io-7
io-7
10"B
Calculated
soliibil ity,
R/cn.3
5
6
7
10
2
0
0
0
2
4
2
1
1
2
2
I
1
0
0
7
17
3
2
2
2
3
-
4
.4
.7
.9
.9
.46
.28
.26
.9
.6
.4
.2
.4
.0
.6
.0
.2
.98
.78
.6
.1
'
.A
.7
.4
-
.3
"The minimum permeation rates that could be detected were: 0.47 ug/(cmz*min) for acetone, 0.12 ug/(cm2-min)
for Cyclohexane, and O.I I iig/(cm2-min) for toluene.
bThp symbol "" means that the calculation could not be performed.
-------
1 2
<*>
s>
iiiiiiiiiiiiiii ii r
ABSORPTION
00
I I I I I I I I L
( 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 3O 32 34 36 38
TIME'-0. mmv'
Figure 11. Absorption and desorntion curves lor cycloltexfine in butyl rubber
-------
1 2
u>
LO
1 I I I I I I I I | I I I I I I I I
o.o
I O
I I I I I I I I I I I I I I I I I I
4 6 8 10 12 14
16 18 20 22
TIME1'1. mm''2
24 26 28 3O 32 34 36 38
Figure 12. Absorption and desorption curves for toluene in butyl rubber
-------
1 2
1 O
08
06
O4
02
OO
1 | I I I I I \ \ \ I I I I I I I I
I I I I I I I I I I I I I I I I I L
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 3O 32 34 36 38
TIME1/', mm*
Figure 13. Absorption and drsorption curves for cyclohexane in natural rubber
-------
1 2
iiiiiiiii i i i i i i i i r
0 Z 4 6
8 10 12 14 16 18 2O 22 24 26 28 3O 32 34 36 38
TIME*7'. min'/' BTIS-II
Figure 14. Absorption and dcsorption curves tor toluene in natural rubber
-------
1 2
1 0
08
06
04
O2
T\ii\iiiiiii\ i i i r
/
s
ABSORPTION
OESORPTION
I I I I I 1 I I \ I I I I I L
00
I 0 1
2 3 4 5 6 7
8 9 10
TIME1"*. minv'
11 12 13 14 15 16
17 18
6719-7
Figure 15. Absorption anil desorption curves for cydohexane in neoprene rubber
-------
iiiiiiiiiiiiiiiir
ABSORPTION
v
DE3CRF7SON
00
J I \ I I I I I I I I I I I I » I
2 ? 4 5 67 8 9 10 11 12 13 14 IS
TIME1* mm1/4
16 17 18
6749-6
Figure 16. Absorption and desorption curves for toluene in neoprene rubber.
-------
1 2
LO
OO
1 O
1 I I I I I I I I I I I I I I I I T
ABSORPTION
/
OO
1_ I I I I I I I I I I I I I I
8 10 12 14
16 18 20
TIMEy*.
22 24 26 28 3O 32 34
36 38
6749-t
Figure 17. Absorption and desorption curves for acetone in nitrite rubber
-------
12
LO
1 O
08
06
04
02
OO
I
"1 I I I I I I I I I I I I I 1 I I T
ABSORPTION
DESORPTION
1 1 L I I I I I I I I I I
O 2 4 6 8 10 12 14
16 18 20 22
TIME*, minI/J
24 26 28 3O 32 34 36
38
-10
Figure 18.' Absorption and desorplion curves lor toluene in nilnle rubber
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing}
1 REPORT NO.
EPA/600/2-86/055
2.
PB86-209087/AS
3. RECIPIENT'S ACCESSION-NO.
Effectiveness of Chemical-
Protective Clothing: Model and Test Method
Development
5. REPORT DATE
September 30, 1986
6. PERFORMING ORGANIZATION CODE
, _it S. Bhown, Elizabeth F. Philpot,
Donald P. Segers, Gary D. Sides, Ralph B. Spafford
8. PERFORMING ORGANIZATION REPORT NO.
SoRI-EAS-85-835
5749-F
10. PROGRAM ELEMENT NO.
9 PI
: AND ADDRESS
ERFORMING ORGANIZATION NAME AND 4
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35255-5305
11. CONTRACT/GRANT NO.
68-03-3113
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Water Engineering Research Laboratory
26 W. St. Clair Street
Cincinnati, OH 45268
13. TYPE OF REPORT AND PERIOD COVERED
Project Report (1/85- 9/85)
14. SPONSORING AGENCY CODE
IS. SUPPLEMENTARY NOTES
A predictive model and test method were developed for determining the chemical
resistance of protective polymeric gloves exposed to liquid organic chemicals." The
prediction of permeation through protective gloves by solvents was based on theories
of the solution thermodynamics of polymer/sol vent systems and the diffusion of solvents
in polymers. These models and test methods were further developed to estimate the
solubility, S, and the diffusion coefficient, D for a solvent in a glove polymer.
Given S and D, the permeation of a glove by a solvent can be predicted for various
exposure conditions using analytical or numerical solutions to Pick's Laws.
The model developed for estimating solubility is based on group-contribution
methods for predicting phase equilibria, while that for estimating diffusion coeffi-
cients versus concentration is based on free-volume theory. The predictive test
method developed is a liquid-immersion/desorption method that provides estimates of S
and D.
Limited confirmation of the developed models and test method was secured by com-
paring estimated values of S and D with reported experimental data and by using the
estimated values to predict instantaneous permeation rates, breakthrough times, and
steady-state permeation rates for comparison with experimental permeation data.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSATi Field/Croup
a DISTRIBUTION STATEMENT
19 SECURITY CLASS (ThisReport/
21 NO. OF PAGES
20. SECURITY CLASS (Thupage)
22. PRICE
EPA Form 2220-1 (9-73)
------- |