Professional Documents
Culture Documents
J. RAMSBOTHAM
Koninklijke/Shell Laboratorium (Shell Research B.V.), Amsterdam (The Netherlands)
Contents
1. Introduction
During the last ten years, considerable advances have been made in the
development of techniques which can assist the solvent formulator in his
work. Many of these have been separately described in the literature.
However, in the majority of cases, the solvent formulator has an interest in a
combination of these techniques rather than in any single one. It is therefore
important to him to have a series of simple, accurate guidelines which allow
him to arrive, without unnecessary trial and error, at a cost-optimised solvent
blend within the constraints he may care to apply. The object of this review
is to present such a series of techniques which we have successfully used for
the past few years for technical service and trouble-shooting work.
Technical criteria for designing a solvent blend for a paint system may
be summarised as follows:
Candidate solvents when combined must have a volatility suitable for
the applicational technique.
114
They must form a true solvent for the polymer or binder system.
Solutions often have to conform to flashpoint regulations and attention
must be paid to other solution properties such as viscosity.
Solvent must evaporate from the paint fihn in such a way as to mini-
mise flow defects such as orange peeling and not cause the resin to
precipitate out of solution.
Unnecessarily high levels of retained solvent in the final film must be
avoided.
The techniques necessary for following the above scheme through are
now discussed.
From the earliest times, formulators have tried to quantify the ‘power’
or ability of their solvents to dissolve things. Many of these traditional
methods were developed for the hydrocarbon extracts used with natural
resins, and factors such as the kauri-butanol number and aniline point are
still widely quoted [l] . However, certain coatings such as shellac, and later
nitrocellulose, required the use of additional oxygenated or chemical
solvents whose power far exceeded that of the hydrocarbons. Other criteria
of measuring solvent power came into vogue [l] and a vast amount of
literature relating to nitrocellulose in particular is found. However, during
the past 20 years, coatings formulators have achieved access to such a wide
range of synthetic resins and polymers that a general method for describing
the power of various solvents towards them is absolutely vital. Such a general
method is provided by solubility parameters.
made much easier. The work of Henry [4] will hopefully help towards this
end. Perhaps the best known parameter systems are those due to Hansen
[ 5 - 71 and Nelson and his co-workers [S, 91.
Using a concept of ‘homomorphs’, Hansen postulated that the cohesive
energy density of a liquid, C - a measure of the total intermolecular forces
- could be related to dispersion, Cd, polar, C, and hydrogen bonding, C.&,
components by :
c = Cd + c, + Ch
The Hildebrand solubility parameter 6 is related to the cohesive energy
density, C = 62, and this then gives rise to the expression:
Hansen derived values for the parameter components and a selection are
summarised for more common solvents in Table 1.
By contrast, the parameter system developed by Nelson and his co-
workers is a much more empirical type of development than that of Hansen.
The Hildebrand solubility parameter is retained together with a fractional
polarity term’proposed by Gardon [lo] and a hydrogen bonding index [ 81.
Values;for these parameters are summarised in Table 2.
P= 5 v,P,
x=1
TABLE 1
Hansen’s solubility parameters - solvents
TABLE 1 (Continued)
TABLE 2
Solubility parameters - Nelson, Hemwall and Edwards
Individual hydrocarbons
n-Pentane 7.02 0 0
2-Methylbutane 6.75 0 0
n-Hexane 7.27 0 0
Methylpentanes 7.0 - 7.1 0 0
Dimethylbutanes 6.7 - 6.9 0 0
n -Heptane 7.5 0 0
Methylhexanes 7.25 0 0
3 -Ethylpentane 7.37 0 0
Trimethylbutane 6.95 0 0
n-Octane 7.54 0 0
Methylheptanes 7.35 0 0
Ethylhexanes 7.5 0 0
Dimethylhexanes 7.1 - 7.4 0 0
Ethyl methylpentanes 7.3 0 0
Nonane 7.64 0 0
Decane 7.74 0 0
Undecane 7.81 0 0
Dodecane 7.92 0 0
Cyclopentane 8.1 0 0
Ethyl cyclopentane 7.94 0 0
Methyl cyclopentane 7.85 0 0
Trimethyl cyclopentanes 7.3 - 7.55 0 0
Cyclohexane 8.2 0 0
Ethyl cyclohexane 7.96 0 0
(Continued overleaf)
118
TABLE 2 (Continued)
Ketones
Acetone 10.0 0.695 12.5
Methyl ethyl ketone 9.3 0.510 10.5
Methyl isobutyl ketone 8.4 0.315 10.5
Methyl isoamyl ketone 8.3 0.255 10.9
Cyclohexanone 9.9 0.380 13.7
Ethyl amyl ketone 8.2 0.223 8.5
Pent-o-xone 8.5 0.190 12.5
Diisobutyl ketone 7.8 0.123 9.8
Diacetone alcohol 9.2 0.312 0
Isophorone 9.1 0.190 14.9
Mesityl oxide 9.0 0.332 12.0
Esters
Ethyl acetate (85 - 88%) 9.6 0.182 4.9
Ethyl acetate (99%) 9.1 0.167 8.4
Isopropyl acetate 8.6 0.100 8.5
nPropy1 acetate 8.75 0.129 8.5
set-Butylacetate 8.2 0.082 8.3
Isobutyl acetate 8.3 0.097 8.7
nButy1 acetate (90 - 92%) 9.0 0.118 5.7
(Continued on facing page)
119
TABLE 2 (Continued)
Alcohols
Methyl alcohol 14.5 0.388 -19.8
Isopropyl alcohol 11.5 0.178 -16.7
n-Propyl alcohol 11.9 0.152 -16.5
secButy1 alcohol 10.8 0.123 -17.5
Isobutyl alcohol 10.7 0.111 -17.9
nButy1 alcohol 11.4 0.096 -18.0
Methyl isobutyl carbinol 10.0 0.066 -18.7
n-Amy1 alcohol 10.9 0.074 -18.2
Cyclohexanol 11.4 0.075 -16.5
2-Ethylhexanol 9.5 0.042 -18.7
Miscellaneous
Dichloromethane 9.7 0.12 1.5
Chloroform 9.3 0.017 1.5
1,2-Dichloroethane 9.8 0.043 1.5
1,2,2,2-Tetrachloroethane 10.4 0.092 1.5
2-Nitropropane 9.9 0.72 4.0
Dimethylformamide 12.1 0.772 18.9
Chlorobenzene 9.50 0.058 2.7
A third solvent can then be added to this ratio and a similar procedure
repeated between point ‘X’ and the point characterising the new solvent.
(In this case solvent fractional polarities would need to be calculated at each
stage.) This approach gives something of a ‘feel’ for which solvents could be
combined to reproduce the solubility parameters of a given existing blend.
120
t9
+5
XILENE (0001,
+3
-3
-6
-9
-12
-15
-18
7 a 9 IO II 12
nlLoE9RlNo SOLUBILITY WRIMETER
RAO
.
-12 - NON-SOLVENTS
NON-SOLVENTS
Fig. 3. Solubiiity map for Epikote 1004 (solids > co. 30% wt.).
Fig. 4. Solubility map for Epikote 1001 (solids > ca. 30% wt.).
125
+6 -
$3 -
SOLVENTS
O-
-3 -
-6 - -6
-9 - -9
-15
Fig. 5. Solubility map for Epikote 1007 (solids > ca. 30% wt.).
Fig. 6. Solubility map for Epikote 1009 (solids > cu. 30% wt.).
+4
-4
-0
NON-SOLVENT
-12
APPRBIMI\TE END-POINTS
OF DIL”TK)N RATIO DATA
-16
-20
I I
---- _---_
s 9 10 II 12
HILDEBRAND SOLUBLIN PARAMETER
Fig. 8. Solubility map for Varcum 6820 phenolic resin (Reichhold) (solids > co. 10% wt.).
In practical terms, this means that when using any solubility map it is
useful to know the minimum solids contents for which it applies to avoid
formulating partial solvents. In the case of mixed binder systems, formula-
tions are conducted in the ‘area of overlap’ of the solubility maps, and the
solids content is taken, for absolute safety, as that of the lowest resin con-
centration.
B-0 - CH2
R .O.CH,.CH,.OH +
k dH,
‘0’
127
He suggests that this could account for the fact that these solvents are
miscible with water on the one hand (6 = 23.5) and aliphatic hydrocarbons
on the other (6 = cu. 7.5). All solubility parameter systems content them-
selves with one set of values for the solubility parameters, however. In the
majority of cases these serve very well with organic solvent based coatings
with the exception of situations where an Oxitol is blended with a paraffin
hydrocarbon, i.e. an extremely non-polar, non-hydrogen-bonding ‘weak’
solvent. Nunn [ 171 has also examined some anomalous predictions obtained
using solubility parameters. In summary to this section it must be stressed
that poor predictions using solubility parameters/solubility maps are rather
an exception than a rule!
TABLE 4
Relative evaporation rates of solvents and effective viscosities (n-butyl acetate = 1)
Effective ASTM
viscosity D.3539.76
(cP,20 “C)
Alcohols
Methyl alcohol _ 1.9
Denaturated ethanol 1.05 1.54
Isopropanol (IPA) 1.10 1.5
Normal propanol (NPA) 1.40 0.94
Secondary butanol (SBA) 1.40 0.89
Isobutanol (IBA) 1.80 0.64
Normal butanol (NBA) 1.60 0.47
Methyl isobutyl carbinol (MIBC) 1.8 0.28
Diisobutyl carbinol (DIBC) - 0.036
Hexylene glycol (HG) 0.007
Esters
Methyl acetate 0.42 5.1
Ethyl acetate 0.45 4.2
Isopropyl acetate 0.52 3.5
Normal propyl acetate 0.59 2.14
Secondary butyl acetate 0.65 1.75
Isobutyl acetate (98%) 0.68 1.45
Normal butyl acetate 0.68 1.0
Methyl amyl acetate 0.87 0.47
Amy1 acetate (95% mixed isomers) 0.83 0.38
Ethyl lactate - 0.2
Ketones
Acetone (DMK) 0.31 5.6
Methyl ethyl ketone (MEK) 0.41 3.7
Methyl isobutyl ketone (MIBK) 0.55 1.6
Cyclohexanone 2.0 0.32
Ethyl amyl ketone (EAK) 0.8 0.28
Pent-o-xone (ME-6K) 1.2 0.27
Methyl cyclohexanone - 0.2
TABLE 4 (Continued)
Effective ASTM
viscosity D.3539.76
(cP,20 “C)
Miscellaneous
Diethyl ether - 11.0
Isopropyl ether (IPE) - 6.81
Methylone dichloride - 6.35
Tetrahydrofuran - 4.85
Trichloroethane - 4.7
Trichloroethylene - 3.22
1,4-Dioxan - 1.65
Perchlorethylene - 1.29
Dichloropropane - 1.06
2-Nitropropane - 1.04
Water - 0.38
Hydrochrbons(‘)
Hexane 7.2
SBP 62162 7.0
SBP 2* 5.7
SBP l* 5.6
Cyclohexane 4.5
SBP 80/110 4.27
Benzene 4.12
SBP 5* 3.92
Heptaue 3.62
SBP 4* 3.51
SBP 3* 2.7
Toluene 2.0
SBP 100/140 1.9
Rubber solvent* 1.12
Xylene 0.76
SBP 140/165 0.59
SBP 6* 0.41
SBP ll* 0.29
Shellsol A 0.2
White spirit (LAWS) 0.19
Shellsol E 0.14
Shellsol TD 0.14
White spirit (115 OF flash) 0.12
Shellsol T 0.1
Shellsol AB 0.06
Distillate* 0.05
Solvent 300* 0.029
Shellsol N 0.02
Shellsal RA 0.017
TABLE 4 (Continued)
Effective ASTM
viscosity D.3539.76
(CP, 20 “C)
Shellsol K 0.013
Shellsol R 0.013
Solvent 350* 0.011
Shell Ink Solvent 2325 0.007
Shell Ink Solvent 2426 0.003
Shell Ink Solvent 2729 <0.003
Shell Ink Solvent 2831 < 0.003
(l)Many of these are ‘mixed solvents’ and relative evaporation rates quoted should be
considered only as a guide to solvent behaviour rather than absolute values.
*Grades available only in U.K.
Actual viscosities may be used as effective viscosities.
Walsham and Edwards [22] have shown that this type of approach
predicts the evaporation characteristics of blends of pure solvents very well.
Moreover, Sletmore [23] pointed out that since the chemical functional
groups of solvents played a far more significant role in determining the
value of the activity coefficient than the hydrocarbon part of the molecule,
solvents could be classed into three major groups, viz. hydrocarbons, ketones
and esters (carbonyl group), and alcohols and glycol ethers (hydroxyl
group).
Figures 9 - 11 illustrate a representative series of activity coeffi-
cients [24] .
Y ESTERS/KETONES
50
25 25
0 0
0 IO 20 30 40 50 60 70 80 60 100 0 10 2u 30 40 50 6a 70 60 90 100
C HYDROCARBONS (% VOLUME 1 C ESTERS/KETONES (% VOLUME)
0
0 IO 20 30 40 50 60 70 60 90 100
C ALCOHOLS/GLICOL ETHERS (%“OLUMEI
Fig. 12. Surface tension driven cellular convection (Bhard cell formation).
within the films occur. Very often this so-called cellular convection phe-
nomenon or BCnard cell formation can be minimised by attention to solvent
design. Moreover, these cellular convection currents can also cause pigment
separation. Cellular convection has been reviewed by Hansen and Pierce
[ 27, 281, while Kresse [ 291 and Haselmeyer [ 301 have given practical
treatments of pigment sedimentation phenomena.
dI’
at2 -- dT Iv.13
( )
where a is the temperature gradient across the film, t is the film thickness,
dI’/dT is the change in surface tension as a function of temperature, 17is the
kinematic viscosity, and D is the density.
In practical terms this implies that one must:
reduce a, i.e. avoid a blend composed entirely of highly volatile
solvents,
not attempt to apply excessively thick films,
minimise (dI’/d!I’), and
134
When designing solvent blends, the above set of criteria can be helped
along by ensuring that involatile, true solvents are present in sufficient
quantity to give an adequate degree of flow-out at the end of the wet stage
of film formation.
r2P, -W/s
where r is the radius of the pigment particle, D, and D, are the densities of
pigment and solution and q the viscosity. Pigment sedimentation phenomena
will tend to occur with:
particles of widely differing size (check dispersion or flocculation),
pigments’of differing densities, and
too low a viscosity medium (see Section 4.2).
through the bulk film. The pioneer work in this field was published by
Hansen [35] who applied Fick’s laws of diffusion to the process. Much work
has subsequently been done on the topic, the bulk of it using thermoplastic
rather than thermosetting systems [ 201. However, our experience has been
that conclusions for thermoplastic systems seem to hold quite well for
ambient temperature cured thermosetting formulations. Several general
conclusions which have an obvious practical importance can be made:
The mechanism of solvent release in the dry stage is unrelated to the
mechanism of release in the wet stage, i.e., solvents which are very volatile
do not necessarily have high diffusion coefficients.
Solvent retention levels increase as a function of the square of the film
thickness and are concentration-related [ 361.
Inclusion of plasticiser reduces the level of retained solvent.
From the point of view of the solvent formulator, the first of the
preceding points is probably of most interest. While the actual composition
of the solvent blend changes, sometimes radically, in the wet stage of film
formation, it is primarily the shape and size of the solvent molecules that
determines how they will be released during the dry stage. (There are odd
exceptions to this rule where there is an intensely strong affinity between
polymer and solvent, e.g. glycol ethers/nitrocelluloses and benzene/colo-
phonium!) Using this line of reasoning and reviewing all published data
available led Newman and Nunn [ 201 to draw up the list adapted in Table 5.
Although solvents will clearly be retained at differing levels in different
polymers, the degree to which they are retained relative to each other does
not change significantly from polymer to polymer. Table 5 is therefore an
extremely handy guide for selecting solvents for a reformulation where
solvent retention has proved to be a problem. The council of perfection is
to determine the diffusion coefficients of solvents in polymers, but data
currently available appear to be unreliable [ 371.
4. Solution properties
4.1 Flashpoin ts
It is unfortunate that there is no prediction technique available for
flashpoints which is so reliable that actual measurements can be largely
dispensed with. Most published techniques, moreover, deal with the solvent
136
TABLE 5
An approximate order of increasing retention for individual solvents
Methanol 1.9 40
Acetone 5.6 73
Methyl Oxitol 0.53 79
Methyl ethyl ketone (MEK) 3.7 90
Ethyl acetate 4.02 97
Ethyl Oxitol 0.38 97
n-Heptane 3.62 146
Butyl Oxitol 0.08 130
nButy1 acetate 1.0 132
Benzene 4.12 88
Methyl OxitaI acetate 0.32 117
increasing Ethyl Oxitol acetate 0.20 135
retention Dioxan 1.65 85
Toluene 2.0 106
2-Nitropropane 1.04 90
Methyl isobutyl ketone (MIBK) 1.6 124
Isobutyl acetate 1.45 133
2,4-Dimethyl pentane 5.6 148
Cyclohexane 4.5 108
Diacetone alcohol 0.15 123
Pent-o-xone 0.27. 143
Methyl cyclohexane 3.5 126
Cyclohexanone 0.32 103
v Methyl cyclohexanone 0.2 122
F=-_
Flu1 +~,vz+ ..I
71 72
where F is the blend flashpoint, F, that of pure component x, u, its volume
fraction and -yx the activity coefficient calculated from Figs. 9 - 11.
Reference [39] contains reference to other work on this topic.
The above expression predicts that the flashpoints of non-ideal com-
binations such as hydrocarbon/alcohol mixtures can fall below that of either
component. This same effect can be observed especially with low concentra-
137
TABLE 6
Flashpoints of epoxy resin solutions in ethyl Oxitol. Flashpoints measured in ethyl Oxitol
(ethyl glycol ether) using an automatic Pensky-Martens Closed Cup apparatus by ASTM
D93
Pure Oxitol 0 41
Epikote 1001 10 44
Epikote 1001 30 31
Epikote 1009 10 44
Epikote 1009 30 30
-20
0 20 40 w W 100
9. VOLUME CHLWNATED
SOLVENT IN KETONE
TABLE 7
Influence of non-flammable solvents on flashpoints
10
15 (“-“,+
l,l,l-trichloroethane 5 32
10 31
15 30
20 29
show. Similar types of results have been observed with systems containing
water (Table 7). It is tempting to hypothesise that at low concentrations of
water or chlorinated hydrocarbon, the structure of the flammable com-
ponent(s) is broken and the resulting higher vapour pressure leads to a lower
flashpoint. However, it is clear that at higher concentrations combustion is
simply not supported. Without doubt, this aspect of solvent system design
would benefit considerably from a methodical series of studies.
4.2 Viscosity
There have been at least 80 relevant publications on predicting
viscosities since 1965 and clearly these cannot all be dealt with in this short
review. The prediction of the actual viscosity of a polymer solution can be
carried out using the concept of intrinsic viscosity; most recently Bagda has
shown how this could be related to solubility parameters [41]. However, in
many formulations the question is not so much what the actual viscosity will
be but rather, given a choice of a number of solvents, which will give the
lowest viscosity and how will they affect the viscosity relative to each other.
Rocklin and Edwards have recently published details of a technique which
meets this need quite admirably [ 421. These workers use a modified version
of the Kendall equation to predict viscosities:
where 7~is the blend viscosity, u, the volume fraction of component x and
n, is the so-called ‘effective viscosity’ of component x. This effective
viscosity applies mostly to highly structured solvents such as the alcohols
139
and glycol ethers; other classes of solvent have effective viscosities which are
virtually the same as their ‘normal’ viscosities. The effective viscosity is the
viscosity a solvent would have if most of its (hydrogen-bonded) structure
was broken, as is the case in most solvent blends where the concentration of
-OH containing solvent is < cc. 30%. Values published by Rocklin and
Edwards are given in Table 4.
The philosophy behind using these figures is that if two solvent blends
are both powerful solvents for a polymer or resin, then the blend having the
lower viscosity will give the lower solution viscosities at a given concentra-
tion. In other words, if in doubt, the preceding equation can be used to
predict solvent viscosities. More commonly, it is adequate to simply choose
solvents in a blend having the lowest effective viscosities where this is an
important formulation parameter.
5. Concluding remarks
Many of the preceding techniques involve linear relationships or factors
which can be linearised, and there is no doubt that much work can be saved
by setting up computer programmes to carry out linear optimisation work.
However, this is not an absolute must and, as has been indicated, blends can
be built up starting with a basic pair of solvents and adding to them.
Invariably, formulators use commercial hydrocarbon solvents which are
often mixtures of compounds since they are distillation fractions. Typical
properties of the hydrocarbon solvents referred to in this review are given in
the Appendix.
In producing a review such as this, certain areas of solvent technology
which really would repay further study become clear. Flammability related
legislation involves flashpoints, and a series of practical guidelines which are
more reliable than those presented here would be’a boon to the paint
formulator .
Finally, I would like to thank all my associates in the Shell laboratories
at Amsterdam for their help in making this publication possible.
Appendix
Typical properties of selected hydrocarbon solvents
(Continued)
References