Determination of the Hydrophile-lipophile Character of Surface
Active Agents and Oils by a Water Titration
H. L. GREENWALD, G. L. BROWN, and M. N. FINEMAN
Rohm & Haas Co., 5 0 0 0 Richmond S t , Philadelphia, Pa.
The titration by water of a dioxane-benzene solution
of an oil or a surface active agent to a cloud end point
can be used to determine a value related to the hydro-
phile-lipophile halance character of the oil or surface
active agent. Details of such a titration are given and
the method is compared w-ith other rating procedures.
E MULSIONS formed with an oil and water usually require
the presence of an emulsifier, if a useful stability is to be
attained. Singly or in combination, hundreds of surface active
agents are available which may potentially aid emulsion forma-
tion and stability, Although this variety contributes flexibility,
certain principles of selection are needed, if the task of choosing
an emulsifier is to be rendered manageable.
In so far as emulsifiers are concerned, theoretical considera-
tions indicate two types of desiderata. First, there should be
a proper balance, in the emulsifier, betneen attraction for the
water phase and attraction for the oil phase. That this balance
alone is not sufficient is indicated by considering the possibility
of using alcohols as emulsifiers. A simple alcohol of proper
balance (such as methanol or ethanol) could be found, but it
would exhibit too little interfacial activity to be an effective
emulsifier; it would not concentrate a t the interface to a great
enough degree. Secondly, one end of the emulsifier molecule
should be strongly attracted by the first phase or excluded by the
second and the other end of the molecule should be excluded
by the first phase or attracted by the second. This ought to
give rise to a high interfacial excess and very low interfacial
tension, Low interfacial tension is expected to improve stability
with respect to both coalescence and creaming ( 7 ) . Improve-
ment in creaming is anticipated because, for a given energy input
in forming the emulsion, the lower the surface tension the smaller
the emulsion particle size.
Of the two desiderata, only the interfacial tension has been
capable of quantitative description. The other characteristic
has defied such clear specification, although it has long been
recognized that the balance between the polar and nonpolar
portions of a surface active agent molecule is an important factor
in the effectiveness of an emulsifier. Clayton (3) lists patents
dating back to 1933 embodying this concept. The same concept
is implied by some of the theories of emulsification and emul-
sion stability ( 4 4 ) . Work by Schulman and Leja (12) on the
use of solid particles in the stabilization of emulsions can also
be interpreted in these terms. Winsor ( I d ) , in a series of papers,
discusses the hydrophilic and lipophilic solvent affinities of
“amphiphilic” species and the effects of these properties on
hydrotropy, solubilization, and related emulsification processes.
Winsor (15’) points out the importance of the balance of these
properties in spontaneous emulsification and in the establish-
ment of the continuous phase in an emulsion.
A systematic ranking of emulsifiers and of oils with respect to
the hydrophile-lipophile balance (HLB) was undertaken by
Griffin (8). By comparing the type of emulsion (oil-water or
water-oil) and the stability of emulsion formed by emulsification
of a series of oils in the presence of the surface active agent, the
latter was ranked. The scale chosen ranged from 1 to 40 with
hydrophobic materials having a low number and hydrophilic a
high number. Approximately 75 emulsions were made for the
1693
determination of each HLB number. Inspection of the tablr
of HLB numbers indicates qualitative agreement, with expec-
tations based on the chemical structure of a member in a given
series of compounds. Thus increasing the ethylene oxide con-
tent of a polyoxyethylene ester increases the HLB number,
whereas increasing the size of the hydrocarbon group decreases it.
The desirability of some measure of hydrophile-lipophile bal-
ance is then well established. Enough is known of Griffin’s
method to indicate that it is a time-consuming procedure. Not
enough is known of this procedure to perform independent de-
terminations of the hydrophile-lipophile balance of emulsifiers
using the same scale as his, although in a recent paper Griffin
(9) gives an equation for calculating HLB values for some
types of nonionics.
The use of solubility properties offers a clue as t o the hydro-
phile-lipophile balance of an emulsifier or oil. Solvents for use
in the paint and varnish industry are ranked by tests which in-
dicate that the solvent power of oils increases with increasing
polarity and polarizability. The petroleum thinner index of
1953 (IO) gives values of aniline point, kauri-butanol value,
and composition expressed as percentage of paraffins, naphthenes,
and aromatics for 400 commercial petroleum thinners. Scheflan
and Jacobs (11) describe these tests in some detail and give tables
of kauri-butanol values for pure hydrocarbons. I n the kauri-
butanol test, 20 grams of a solution of gum kauri in butanol is
titrated n i t h the solvent to be tested. The end point in this
titration is a cloud such that 10-point Century type is illegible
when viewed through a flask. One of the various standards used
in this test is toluene = 105 cc. On this scale mineral spirits =
35 and solvent naphtha = 85. The aniline and mixed aniline
points are the minimum critical solution temperatures of equal
volumes of aniline and solvent and of two volumes of aniline
mixed with one volume each of n-heptane and solvent, respec-
tively (1).
I n working with nonionic emulsifiers prepared by coupling a
given alkyl phenol to a polyoxyethylene chain, it is found that
longer hydrophilic chains are required for optimum emulsification
in passing through the series: mineral oil, xylene, toluene, ben-
zene. Thus an examination of the series of oils and of their
preferred emulsifiers shows that the more polar the oil, the more
water-soluble the preferred emulsifier (2, 8). Solubilities of
these oils in water are of a magnitude which does not allow easy
observation, hoxever, and a magnified scale is desirable.
This report presents a study undertaken to design a solubility
scale capable of ranking emulsifiers and oils in a manner com-
mensurate with the matching that results from emulsification
experiments. The ultimate purpose is to predict emulsion per-
formance by an independent method,
EXPERIiMENTAL
As the hydrophile-lipophile balance of an oil or emulsifier
appeared to be related to the solubility characteristics, an index
of these was sought. A titration might be used if a suitable
solution of the oil or emulsifier could be titrated with a liquid
of limited solubility, the end point being the appearance of a
second phase.
Selection of Solvent System. I n searching for a suitable solvent
the requirements considered were:
1694 ANALYTICAL CHEMISTRY

1. Compatibility with a wide range of emulsifiers and, if pos- One-gram samples are accurately neighed into 125-mI. Erlen-
sible, oils. meyer flasks, and 30 ml. of solvent is added by pipet to dissolve
2. Ability to cover this range of materials by titration with the sample. Distilled water is then titrated into this clear solu-
one precipitating agent (preferably water). tion from a buret until the first persistent turbidity.
3. Distinctness of the end point in terms of the spread be-
tween compounds. Using this procedure, 11 solvents were tried on various oils and
4. Absence of complicating factors such as a gelation and emulsifiers (Table I). The uncertainties indicated in the table,
fading of the end point. and throughout the rest of this report, are mean deviations, usu-
In the preliminary work, thr follon ing procedure was arbitrarily ally obtained from but two determinations. For the octyl
adopted as standard. phenyl polyoxyethylene series the symbol OPE, is used, the sub-
script referring to the number of moles of ethylene oxide appended
to each alkyl phenyl radical. None of the solvents filled the re-
quirements very well.
Because none of these compounds proved to be satisfactory :i+
solvents, mised solvent systems 1%-eret8ried. .is diosane \vas t,oo
hydrophilic and phenyl Cellosolve too hydrophobic whrn used
alone, mixtures of t,hese were investigated. The results of this
investigation, given in Table 11, indicated that the 50: 50 solu-
tion gave too short a ppread between menhers of the octyl
phenyl pol>-oxyethylene series. A better range was ohta,inetl
by using 25y0 phenyl Cellosolve in diosane. Runs with t \ V ( J
lots of dioxane showed that this solution gave results in sltt,i+
factory reproducibility in terms of the spwing between commer-
cial members of the OPE, series. .It thip stage both c~ommercinl
dioxane and the 25y0 Cellosolve solution n ~ w used in othpr
investigations.
.kiter the 25y0 phenyl Cellosolve had been used for a while, it
was found that the reproducibilitj- aiid the distinctness of the
end point n-ere poor, particularly a-hen ionic Pmulsifiers were
tested. The data obtained n-ere often unsatiductory, and the
operator \vas uncertain of the results and forind the titrat,ions
difficult and time-consuming. In view of this, the toluene- :tncl
benzene-containing samples of diosane were tried, as indicated
in Table 11. The 3% benzene solution gave some of tlie sanie
trouble and the 5% benzene solution had too short a range br-
tween numbers. Optimum behavior as ohtained with a 4%
WEIGHT PERCENT E T H Y L E N E OXIDE
benzene solution, so this was accepted as the standard.
Figure 1. Octll phenll polyoxyethylene series A question might be raised as to the significance of a wn-atcr

Table I. Vater Titration Values in Candidate Solvents

(MI.)
Sol\ ent Xylene Atreol 9 OPE6 OPE-.s OPEs.; OPEI?.~
Dimethyl forinaniide 1 4 2 f 1 1 0 7 I 1 13 7 1 0 . 3 > 200 .... ...
terl-Butyl alcohol 40 9 + 0 1 4 2 & 0 0 1 3 1 . 8 1 0 . 3 >zoo . . . ..
Butyl Cellosolve 2 6 9 I O 1 4 O & O 1 >zoo 1200
Phenyl c a r h ~ t o l 6 3 I 0 3 7.3 I O 5 7.6 r 0 0 7 6'I'O 9.1.g1
n-Hexyl carbitol 1 o o i o 3 11 8 f 0.4
Pure dioxanea 1 3 2 f O l 1 4 1 0 49:9,* 0 2 > 200 . .
Corn. dioxane 1 2 5 f 0 1 1 2 I O 4 3 . 2 I O 2 > 200 . .
Phenyl Cellosol\ e ..... 2 0
Methyl hexyl ketone ..... 0 95
OPE8 ..... 9.9 ( r e r y viscoiis)
ti-Hexyl Cellosolre ..... 4.33 ...
a ,Purified by method of Fieser (6). The final distillation, ox-er sodiunr, was through a rolumn with a &foot packed
section. Distillate was collected in a 0.5' C . range.

__ ~~ -

Table 11. Water Titration Values in llixect Solvents

(AI1 )
Soli ent SJstem None Xylene Atreol 9 OPE6 OPE- 6 OPE9 7 OPE12.3 Blend .la Blend B n

, .. 20 1 1.4 40.2 40.6 50.3 55.0 45.3 i0.3 13.6 i 0 . 2

030 zeo5 19.0 1.0 f 0 . 2 38 8 44.8 48.7 52 3 .... ....
19.5 Z t 0.1 1 0 6 i O 5.2 I O 16.4 + 0.1 1 7 . 4 2~ 0 17.5 i 3 18.4 i0 .... ....
(lot 2 ) .... .... 8.2 I O 8.5 i0 8.81 0 . 1 8.9 i0
10Tc benzene-90TC dioxane 9 08 .... .... .... .... .... .... ....
5 7 c henzene-957* dioxane
(lot 3) 17.85 I 0 . 0 8 .... .... .... .... .... .... 14 35 i 0 . 0 6 12 5 * 0.3
4% benzene-96% dioxane
(lot 3) 21.7 I O . 1 9.1 I O . 2 1 . 5 1 0 . 1 18.4 + 0 . 1 21.4 +0 2 3 . 1 I O . 3 23 0 = 0.2 15 30 + 0.03 19.90 + 0
3L7, benzene-9iY0 dioxane
(lot 3) 2 9 . 0 ze 0 . 2 .... .... 16.09 + 0 . 0 1 12.3 I 0 . 3
2 5% benzene-97 5 % di-
oxane 35.0 .... .... . . .
Blend of nonionic and a.nionic einulsifiers
V O L U M E 28, N O . 11, N O V E M B E R 1 9 5 6 1695

Table 111. Change in Water Titration \-dues with Table Y. Effect of Temperature
Dioxane Sample Variation
(Solvent. 4% benzene in dioxane)
Solvent OPE5 None Xylene
Solii t e 30" C . , 311. 200 c., 411. 411. c
Dioxane
Lot 1 43.' i o 2 12 5 1 = 0.07 OPE5 18.77 i 0 . 0 2 18.0 f0.3 0 08
41.56 +
0 Oli ..... OPEs.;
Atreol 9
2 3 . 1 =z 0 . 2
1 . 8 7 i 0 02
22.1 1 0 . 2
1.36 0.04
0 10
0 0.5
Lot 2 41 26 1 0 02 .. 12 ~ j $ ' i ' 0 , 0 2
Lot 3 41 8 .... Xylene 9 92 =t 0 . 0 4 9 . 0 9 1 0.0Ti 0.08
Lot 4 titi = 0 00 , . . ..., E.C.a 7.60 i 0.01 7 . 0 8 i. 0 . 0 8 0 05
Lot 19 45 0 ... . .... None 22.6 i0 . 3 20.2 i 0 2 0.24
4 7 brnzene -9U% dioxane
Lot 3 18 40 zt 0 00 2 1 . 7 IO . l a 9 35 =t 0.08 a Emulsifiable concentrate containing 13 poiinds of l'osapiaene I J ? ~ gallon
.,.. 22 1 i 0 , o e . of kerosine.
Lot 4 .... 2 1 . 6 4 + o 04a 9 5i'1'0.05
.... 21.79 +O.OSa 9 . 4 1 I0 . 0 2
Composition of the c n . 47; benzene solution was adjusted t o give t h e
desired water number \\-lien first made. Second line for lot 3 was a deter- Table VI. Effect of Speed of Titration
mination 5 days after the first: for lot 4, 2 inontlis after the first.
(Solvent. 4 7 benzene in dioxane)
\Vat er XII iiiher
-. __
~

Solnte I) i>d,'min. 1 inl. Inin.

OPE%: 22 81 32 89
Table I \ . End Point Experiments in Determinations of Sorbitan nionolariratP i 41 7 .st?
Water Tihration kalue .loci, anionic-60cc nonionic, 12.84 12 so
(Sol\ ent 2jC: p11rn)l Cellosol\ e in dioxane)
"r (7 E irst Cloud, Dense Cloud,
Anionic Yonionic 311. 311. Table YII. Water Sunibers of Emulsifiers
12.5 87 5 9 i l 42.7 i0 2 311.)
25.0 75 0 7 1 10.9 1, 8 1 0 1
37. ,5 62 5 7 . 8 1-0.8 13 9 1 0 . 3 Water Titration Water S o .
50.0 50 0 7.9 10 7 11 4 i 0 . 7 Dioxane (4%, Benzene,
02,s 37 s fl.40 1 0 01 11 5 i 0 2 Compound as Solvent) 96% Dioxane)
75 0 2.5 0 8.2 =0 R 10 2 o i 0
OPE0 11.6 i 0 . 1
OPE1 1 8 1 1 0 1 116.7
3.8 1 i0.1
OPE8 22.5 f0 1
OPE5 432io.' 18.4 + 0.1
OPE7.s >zoo 21.4 f 0
OPEs.r 2'3.0 i 0 . 3
0PEn.a 23.6 f 0 . 2
(OPEia-CHz-). >loo 24.3 f 0.1
n-CiaHsrCsHeEa 13.1 k 0.1
number value greater than that of the solvent system alone Branched CuHwCsHa-E5 15.0 f 0 . 1
tert-CiwiHa;-rr NEsa 8 5 i 0 1 9 . 4 i0 . 2
(column 2 in Table 11). This observation can be expected \Then- tert-Cis-zrI-Iar-ra NEiaa 11.3 10 12.0 10
tert-Cis-xHa;-rs NEzsa 12 6 0.2 13.7 f 0 3 +
ever the emulsifier is more hydrophilic than the solvent system n-Hexadecyl + n-octadec,yl alcohol
Polyethylene glycol (mol. wt. 300)(Ea.&)
8.4
22.8 10.0
0.2 +
used. Polyethylene glycol (mol. wt. 400) (Es.7) 2 3 . 1 i 0.3
Because commercial dioxane contains impurities in fairly high Polyethylene glycol (mol. wt. 600)(Eii.:) 22 5 1 0 . 0
concentration, it is important to consider the effect of batch to a Ethylene oxide adduct of a priniar)- amine.
batch variation in dioxane on water number. That this is found,
and indeed accounts for a variation from 41.3 to 46.7 for the
I n t e r number of OPES in dioxane as solvent, is seen in Table 111.
-1s indicated in Table 111, after the first several solvent lots,
hatches of 4% benzene in dioxane were adjusted in benzene con-
tent to give water numbers, run on the solvent system alone, in
the range 21.6 to 21.8. Data given 011 lot 4 dioxane indicate that
the effect of aging a given sample of dioxane for 2 months is not
great.
Relat.ed to the problem of selecting the solvent is the problem
of obtaining a suitable end point. It. was previously noted that
the desired end point was a pronounced turbidity, not the first
appearance of a cloud. The difference between these points is
illustrated by the data obtained on a series of blends with 25%
phenyl Cellosolve in dioxane as the solvent (Table IT). I t is
obvious that the dense cloud point is more reproducible and is
the one which indicates the hydrophile-lipophile balance of the
emulsifier blend, -4t the dense cloud point there is a sharp change
from emulsion droplets just large enough to be visible to droplets
too small to be sren individually. I t is this change in particle
size that seems to be responsible for the marked change in appear-
ance. The nature of the change is unknown, but in appearance
it is somewhat reminiscent of emulsion inversion.
In order to assess the possibilit'y of doing the determinations
in the open laboratory, the effect of temperature on water number
was determined. Table V gives the water numbers of five repre-
sentative solutes a t 20" and 30" C. As the change is roughly
0.08 ml. per degree, it would seem that * l o C. thermostating
would be adequate but that it is not satisfactory to work a t ordi-
narj- room temperature (22" to 30" C.). No explanation for the
rather large temperature coefficient of the water number of the
WATER NUMBER
wlvent system alonca is offered. Figure 2. Polyhydric alcohol esters
1696 ANALYTICAL CHEMISTRY

The possible effect of speed of titration was investigated by specific chemical factors in addition to hydrophile-lipophile
taking duplicate samples and titrating one a t the rate of 6 ml. per balance operate in both cases. I n any event, it is clear that
minute and the other a t 1 ml. per minute. Results of this experi- whichever solvent is used the ordering of the members in each
ment (Table VI) indicate no significant effect. family is correct.
Water Numbers of Surface Active Agents and Oils. The water Figure 2 is a plot of HLB number (8) us. water number for 18
numbers of several emulsifiers were determined as listed in Table surface active agents in two families of polyhydric alcohol esters:
VII. It is presumed that the low water numbers obtained for the (1) ethylene oxide adducts and (2) those without ethylene oxide.
polyethylene glycols as compared with OPE12.a are due to solu- A fairly good straight-line relationship between the two types of
bilization of the benzene by the latter. determinations is obtained for each family. The possibility
The data obtained for the octyl phenyl polyoxyethylene series exists that the HLB number for one of these families was deter-
(OPE,) are plotted in Figure 1, along with the titration values mined by the emulsion type formed in the given oils and for the
determined using dioxane as the solvent. It is seen (Table VII) other family by emulsion stabilities. Since the HLB numbers are
that the effect of the presence of benzene in the dioxane is to based on performance in emulsions, the data in Figure 2 suggest
decrease all of the titration values and to enable the method to that the water number will give a t least a fair indication of the
be employed over a wider range of ethylene oxide content. performance of a surface active agent as an emulsifier.
Comparison of the data obtained in the two solvents for OPE, Results of determinations of n-ater number on a group of oils
and the amine E16 indicates that the choice of solvent system mill are reported in Table VIII. The table of water numbers of the
have some effect on the relative results for different families of oils can be interpreted simply as indicating that increasing aro-
compounds. This might be true in emulsification as well, as matic character increases the vater number. Correlation be-
tween the aniline or mixed aniline point and water number is
good, except in the case of OPE1. Thus it appears that water
number parallels aniline points only when the aliphatic-aromatic
balance in hydrocarbons is considered.
If surface active agents differing in hydrophile-lipophile bL1-
Table VIII. Water Numbers of Oils ance range from the ones used in these tests are of interest, it is
(MI.) probable that a suitable spread between numbers of a series can
hlixed be obtained by altering the benzene content of the solvent. Thus
Aniline0 Anilinea for determinations on OPE, compounds with z above 10 it might
Water S o . Pt., C. Pt.. C.
be desirable to spread the upper end of the scale by going down
Bayol N-150 (Esso) 1 . 0 1 10 . 0 1 Ca. 108 .. to 3 or 3.5% benzene.
Dormex (Cities Service) 1 . 3 10 . 1 103.7 ..
s/v Prorex M (S,ocony Vacuum) 1.43 1 0.06 95.5 .. Application to Emulsifier Selection. The more polar an oil, the
Atreol 9 (Atlantic) 1 . 4 9 1 0 06 101.7 ..
Insectasol (deodorized kerosine) 5.78 1 0.09 74.8 more hydrophilic is the best emulsifier for that oil emulsified in
Nonyl (tripropylene) benzene 6.50 + 0.01 ...
Velsicol AR-55 (Velsicol Corp.) 7.35 + 0.05 3i:5 water. As a test of this principle and as an illustration of the
Sovacide 544-C (Socony Vacuum) 7 . 5 7 10 . 0 3 26.8 utility of the water number procedure, the following experiment
H A N 132 (Esso) 7.8 10.1 25.1
Xylene 9 . 5 1 1 0.02 10.8 was performed.
OPE1 1 3 . 8 =e 0 25.2
5 Aniline and mixed aniline points run b y H. F. Majka. hlixed aniline A number of oil solutions of agricultural toxicants were listeil
points r u n on samples whose aniline point was too low t o be determined. in order of increasing water number (Table X).
Water numbers of blends of water-soluble emulsifier C and
water-insoluble emulsifier D were determined (Table IX).
The oil solutions were emulsified in four different waters rang-
Table IX. Water Numbers of Blends ing in hardness from 1000 to 0 p.p.m. using the C-D emulsifier
% 70 blends. A water-oil ratio of 85: 15 to 96:4 by volume was used
Emulsifier Emulsifier Vater and the emulsifier content was from 3 to 670 of the weight of the
D C NO.
oil. These ratios were determined by the field use requirements.
100 0 1 3 . 2 zt 0 . 2
80 20 1 4 . 0 10 . 0 Emulsion stability was determined by observing the rates of
70 , 30 1 4 . 7 10 . 8 creaming, clearing, and free oil formation following the procedure
60 40 1 5 . 0 10 . 1 of Brown and Rilev (8). From these data the best emulsifier for
50 50 1 6 . 1 10 . 2
40 60 17.0 k 0.0 each oil-water system’was judged. These best emulsifiers are
0 100 23.0 1 0.0 listed in Table X.

Table X. Performance of Blends of Emulsifier C and Emulsifier D

% of Emulsifier C in Blend Giving
Toxicant-Oil Solution Best Performance
Water 1000 p.p.m. S a v y hard Stoneaille Distilled
Lb. No. hard water water (soft) water water
4
6
Toxaphene-kerosine
Toxaphene-kerosine
6.3 *
1
7 1 1 0 1
<O
<O <O
<O
..
<0
<O
6 Chloi dan-kerosine 7 2 1 0 1 <O .. 0 <O
a Toxaphene-kerosine 7 6 1 0 2 <O .. 0 10
25% DDT-aromatic naphtha 77zk01 0=10 .. 10 <0
8 Chlordan-kerosine 8 2 1 0 2 <O .. 40 10
25% DDT-aromatic naphtha 8 2 1 0 1 10 10 .. 0
4 2,4-D butyl ester-aromatic naphtha 5 1 + 0 1 40 10 .. 10=20
4 2,4-D isopropyl ester-aromatic nap1itha 9 3*0 1 30 20 .. 20
4 Toxaphene 5 3zko 1 30 20 , . 20
2 DDT-xylene
25% DDT-xylene
1 . 2 r-BHC
9.4 * 0.2
9.6 i0.0
>40
40
20=40
.. 30
20
30
2 DDT-xylene
0 . 8 r-BHC
1 . 3 3 D D T-xylene
9.9 * 0.2 50 10==20=30 .. 40
1 r-BHC-x lene 1 0 . 0 10 . 1 >60 30 40
2 2,4-D T z F ester 10.2 1 0 1 40 20=46=50 .. 30=40
2 2,4,5-T T H F ester-xylene
4 and > indicate t h a t in judging creaming d a t a i t appeared t h a t less, or more, respectively, Emulsifier C
would have improved t h e performance where further testing in the desired direction was not done.
V O L U M E 2 8 , NO. 11, N O V E M B E R 1 9 5 6
Inspection of Table X indicates that the water number of the
oil solution correlates well, although not perfectly, with the rrater-
soluble component content of the "best" emulsifier blends. The
water numbers of the toxicant-oil solutions are in the expected
order in general-a saturated chlorinated hydrocarbon dissolved
in kerosine is the lowest and the xylene solution of an ester the
highest. Water numbers of the emulsifier blends given in Table
I S are also consistent with expectations, 100% water-insoluble
D being the lowest and 100% n-ater-soluble C the highest.
Performance data in the 1000-p.p.m. water when compared
with the data in the softer waters show that increasing the water
hardness moves the scale for the best emulsifier towards higher
water numbers. This can be explained as due to the salting-out
effect which makes an emulsifier less hydrophilic in the presence
of hard water than soft water.
The determination of water number offers the emulsion formu-
lator a means of ordering oils and emulsifiers and thus of estimat-
ing how a new oil or new emulsifier fits into his past experience.
ACKNOWLEDGMENT
The authors are grateful to J. IC Beemer, P. J. Kline, and John
Refractive Index of Liquids at Elevated Temperatures
JAMES L. LAUER and RICHARD W. KING
Sun oil Co. Research Laboratory, Norwood, Pa.
The determination of refractive index at elevated tem-
peratures using commercially available instruments has
been hampered by the lack of reliable calibration stand-
ards. Two instruments have been constructed for the
absolute determination of refractive index at elevated
temperatures. One instrument is useful for deter-
minations in the temperature range 25" to 55' C., and
the other from 60' to 110' C. Data obtained using these
instruments are presented for three hydrocarbons pre-
pared by API 44 at the Carnegie Institute of Technology
to serve as calibration standards and six hydrocarbons
on loan from API 42 at the Perms)-lvania State Univer-
sity. Air refractive indices of these compounds at 80"
and 100" C. have been determined for six wave lengths
of the visible spectrum. The accuracy with which the
Eyliman function, C = (n2- l)/d(n +
0.4), represents
the relationship between refractive index and density
has also been investigated for two hydrocarbon liquids
from room temperature to 100" C.
T HE refractive index of liquids is a physical property so
easily determined with accuracy that it has become a
standard for their characterization. It would appear logical,
therefore, to apply it t o -ivaxes and similar materials a t tempera-
tures above their melting points. However, accurate refrac-
tive indices are diacult to obtain a t temperatures much above
ambient because of the difficult,y of maintaining the sample and
part of the measuring instrument a t a uniform temperature.
Every commercial refractometer exhibits its own characteristic
deviations when used a t elevated temperatures. Reliable cali-
bration standards are necessary, therefore, before commercially
available instruments can be used with confidence.
Hollow-prism spectrometers have been reported as best
suited for the absolute determination of the refractive index of
liquids (9, IO). Accordingly, two such instruments were con-
1697
Huey for having done much of the experimental work reported
herein.
LITERATURE CITED
Am. Soc. Testing Materials, ASTM Tentative Method 1012-
49T.
Brown, G. L., Riley, G. C., A p . Chemicals 10, No. 8 , 34 (1955).
Clayton, W., "Theory of Emulsions and Their Technical
Treatment," 4th ed., Blakiston, Philadelphia, 1943.
Ibid., Chap. VI.
Cockbain, E. G., RZcRoberts, T. S., J . Colloid S c i . 8 , 440 (1943).
Fieser, L. F., "Experiments in Organic Chemistry," D. C.
Heath, Boston, 1941.
Greenwald, H. L., J . SOC.Cosmetic Chemists 6 , 164 (1955).
Griffin, W. C., Ibid., 1, 311 (1949).
I b i d . , 5 , 249 (1954).
Katl. Paint, Varnish and Lacquer Assoc., Sci. Sect. Circ. 761
(July 1953).
Scheflan, L., Jacobs, 11. B., "Handbook of Solvents," Van
Kostrand, New York, 1953.
Schulman, J. H., Leja, J., T r a n s . F a r a d a y SOC.50, 598 (1954).
Winsor, P. A, Ibid., 44, 376 (1948).
I b i d . , 46, 762 (1950), and preceding papers.
for review January 13, 1955. Accepted August 2, 1956.
RECEIVED
structed. One was designed for use in the temperature range
25" to 55" C. and the other from 60" to 110" C. Both utilized
a Gaertner spectrometer but were fitted with measuring prisms
and constant-temperature jackets of different design. Thus t h e
entire temperature range from 25' to 110' C. could be covered
conveniently and accurately. With this equipment refractive
indices were measured at elevated temperatures for three com-
pounds of high purity supplied by API Project 44 a t the Carnegie
Institute of Technology and intended to serve as calibration
standards. Incidental to supplying data on these standards for
laboratory refractometers, refractive indices were measured
a t several different temperatures and wave lengths for a number
of hydrocarbons which had been prepared by API Project 42
a t the Pennsylvania State University (8). The accuracy with
which the Eykman function ( I , d ) , C = (n2- l ) / d (n 0.4) +
expresses the relationship between the density and refractive
index of organic liquids was also investigated for t v o hydrocar-
bons.
APPARATUS AND PROCEDURE
All spectrometric methods suitable for precise refractometry
involve the determination of the angular deviation with respect
t o some fixed direction of a collimated beam of light which has
been refracted by a system of plane surfaces, usually two in
number, One of the most convenient and often used methods of
measurement is that of minimum deviation. Its application to
the precise refractometry of liquids has been thoroughly de-
scribed (IO). In brief, the method depends upon the fact that
the angles of incidence and emergence are equal when the orien-
tation of the prism containing the liquid sample is such that the
deviation of the refracted ray is a minimum. The refractive
index is calculated from the angle of minimum deviation and the
prism angle.
Because the former angle is determined by moving the tele-
scope arm of the spectrometer with respect t o the collimator,
the method becomes inapplicable a t temperatures above 50' C.,