Manufacturing-Processing

THE T ALOFLOC DECOLORIZATION PROCESS

M. C. Bennett, F. J.
Gardiner, J. C. Abram, and J. T. Rundell
Tate & Lyle Ltd
Group Research & Development
Keston, Kent, England

ABSTRACT

Cationicsurfactants can be used to precipitate colorant and other anionic

impurity from washed raw melter liquors. This provides the basis for a new
process of simultaneous defecation and decolorization. To be effective, the
surfactant must have a strong cationic centre, for example, the quaternary
ammonium group, and a n-alkyl hydrocarbon chain of not less than 16 C atoms.
The 'electrical nature of the interaction has been studied by particle electro-
phoresis and by examination of the net anionic charge on colorant fractions.
These techniques have shown that most effective precipitation takes place up
to the point of zero zeta potential, and that the cationic surfactant tends to
precipitate those fractions of color which carry the greatest anionic charge.
The activity of cationic surfactants has been studied as a function of molecular
structure. Results for 6 series of cationics are given and these indicate that
dioctadecyl dimethylammonium chloride (TALOFLOC) is the most active sur-
factant. The color-TALOFLOC precipitate is difficult to separate by simple filtra-
tion, but is readily separated by a phosphatation flotation process. The TALO-
FLOC cationic surfactant provides a 2nd advantage in that air bubbles are
attached strongly to the floc. Use of a few ppm of a special anionic polymer
as a flotation aid causes a remarkable flotation effect, 50% clarification occur-
ring in less than a minute.

INTRODUCTION

Much of the high molecular weight impurity in raw sugar liquor is

anionic in character. For example, most color shows some acidic function.,
carboxylic acid gums are well-known and the amylopectin fraction of cane starch
contains some ester phosphate groups. The impurities are strongly hydrophilic
and most of them are highly water-soluble. An important feature of the defeca-
tion and decolorization processes in sugar refining is the ability to remove high
molecular weight anionic impurities.' These impurities are removed by specific
attachment to the defecation or decolorization agent.
The carbonatation process,for example, tends to remove any impurity
which can form a sparingly soluble calcium salt. The ability of carbonatation
to remove ash is well known: inorganic sulphate and phosphate are incorporated
within the calcium carbonate precipitate particles. Anionic colorant is also
removed in this way and is subsequently found distributed throughout the body
of each precipitated calcite particle. Bone char has a specific affinity for anionic
1569
1570 MANUFACTURING-PROCESSING

color because it is composed largely of hydroxyapatite which acquires a positive

charge by adsorption of calcium ions. Consequently, the bone char has a ten-
dency to adsorb negatively charged anionic color to preserve electrical neutrality.
With phosphatation, too, there is evidence to suggest that the flocculation effi-
ciency of calcium phosphate precipitated in situ is related' to the anionic charac-
ter of the colloidal impurities affected. In these examples, it is suggested that
specific attachment of the anionic impurity to the defecation or decolorization
agent takes place through a calcium salt linkage. .'
To extend this principle of impurity removal, it: would be necessary to
employ a cation which interacts more strongly than calcium with the organic
impurity anion. If, at the same time, the attachment of such a cation conferred
water insolubility upon theanionic impurity, then the impurity would be pre-
cipitated: defecation and decolorization would he brought about simultaneously
in a single operation. .
As a result of work ca~ried out in these laboratories over the past few
years, it is now known that many cations are capable of effecting such aprecipi-
tation. They are all members of the family described generally at cationic sur-
Iactants and are characterized by a strongly basic centre, for example, the quater-
nary ammonium group, coupled with a long chain hydrocarbon residue. The
cationic centre serves only as a means of attaching the hydrocarbon to the impur-
ity or colorant anion. It is the hydrocarbon which renders the colorant repellant
to water and which causes the color to be precipitated from solution.
The use of cationic surfactants as color precipitants in sugar refining (2)
has been the subject of a major development programme during the last 2 years.
Of the scores ofcationics which have now been tested, by far the most effective
from the point of view of price and performance isdioctadecyl dimethylammon-
ium chloride, which is manufactured for use in the sugar industry to a Tate &
Lyle specification and is available under the trade mark "TALOFLOC." Some
preliminary results from full scale refinery operations using TALOFLOC have
been reported earlier this year (3).
It is the purpose of this paper to present some of the physico-chemical
data upon which the use of cationics is based and to show the evidence upon
which TAL()FLOC was chosen as the most effective color precipitant.

EXPERIMENTAL PROCEDURE

The liquors and sugars used in this study have been taken from the
Thames Refinery of Tate & Lyle Ltd. In some cases, lie luors were taken directly
from the melter; in other cases, washed raw sugar was collected from the affina-
tion machines and dried at room temperature for storage. Experimental details
for each of the techniques used are presented more appropriately in each of the
sub-sections in Results and Discussion. AI\ color measurements were made at
420 nm after adjustment of the undiluted, sample to pH 7.5. Colors are reported
as the attenuation index a *420 in milliabsorbancy units (mau). Cationic surfac-
tant concentrations were determined by partition titration with the anionic
surfactant, sodium dioctylsulphosuccinate, using the competitive dye, bromo-
cresol blue, after the method of Barr, Oliver and Stubbings (1).
M. C. BENNETT, F. J. GARDINER, J. C. ABRAM, J. T. RUNDELL 1571

RESULTS AND DISCUSSION

The Precipitation of Color

One of the most common members of the family of cationic surfactants is
cetyltrimethylammonium bromide (CTAB). The color precipitation effect was
discovered using this material. A sample of 65% solids Natal liquor was treated
with CTAB solution to give a concentration of around 1 /Leq/g solids (364 ppm
on solids). The sample was divided into a number of aliquots, which were
heated at various temperatures up to 100 C and for various times. No visible
change occurred and the samples were, therefore, put on one side, together with
blanks not containing CTAB. After 4 days, floes had appeared in all the CTAB-
treated samples and after 14 days the flocculated material had settled out to give
a brilliant pale-colored; supernatant. The experiment was repeated, but this
time 2% kieselguhr (Dicalite Superaid) on solids was added to liquor which had
been heated for 30 min in the presence of CTAB. The sample was then filtered
and the color of the filtrate compared with that of a blank, filtered using the
same quantity of kieselguhr but not containing CTAB. The value for a * 420 in
the blank was 960, while that of the CTAB-treated sample was 266.
A series of experiments was then carried out in which it was found that
a) there was no difference between the effect at 10 min and that at 30 hr after
CTAB addition, and b) the sequence of kieselguhr and CTAB addition did not
affect the result.
The optimum CTAB concentration was located in the following manner.
Aliquots of 65% Natal liquor were treated with CTAB at room temperature,
and then heated at ca 75 C in a water bath. 2% kieselguhr was then added and
each filtered through 2Whatman 42 papers carrying a smear of kieselguhr.
In Fig. 1, the attenuation index a * 420 and the concentration of CTAB

'1200

0
COLOR
1·2

800
~
~

.s,'"
", ..
0

0'8
J.!!
s'"0 , ~
!J
400
.'/.
I
9
\
\
I
I
.
,S
III
,...
u
P \

p
/ • \
\ CTAB precipitated
\
0'4

,,
\
\
\
,If \
\
~
~
0. ......
-2 -1 0 2
10910 added C'TAB concentration (~eq/9)

Fig. 1. Precipitation of colorant from.65%Natal liquor with GrAB. Left-hand scale: filtrate
color. Right-hand scale: CTAB concentration in filtrate.
1572 s. MANUFACTURIN(;-,-PROCESSINC

removed from each sample are shown .asr:a function of the original CTAB con-
centration. It is seen that the filtrate color reaches a minimum at 1.2 p,eqjg (440
ppm), where the quantity of CTAB precipitated and removed by filtration is
at a maximum. Up to this point, all added CTAB is precipitated, the filtrate
content being effectively zero. For CTAB additions above 1.2 p,eqjg, CTAB
begins to appear in the filtrate and the color returns to its former value. The
result shown in Fig. I can be readily explained as follows. The precipitation of
colorant anions by CTAB is due to the attachment of the hydrophobic alkyl
(hexadecyl) chains as shown in Fig. 2, sketch (b). The colorant polyanion-

~$
e ---e
impurity polyanion
polyanlon/CTAB
complex
soluble polyanion/CTAB
complex
insoluble
soluble

Fig. 2. Diagrams representing the interaction between anionic' colorant and a cationic mono-
alkyl surfactant. Maximum precipitation is observed at stage (b), redissolution at stage (c).

CTAB complex is rendered insoluble and so separates from solution On filtra-

tion, both colorant and CTAB are removed in the filter cake. At higher addi-
tions of CTAB, a 2nd adsorbed layer of CTAB may form, so that any precipitate
formed will redissolve in much the same way as oil droplets can be dispersed in
water. This is shown in Fig. 2, sketch (c). When this system is filtered, the solu-
ble polyanion-CTAB complex passes through the filter medium and both com-
ponents are detected in the filtrate. It is clear from Fig. 2 that the polyanion can
be titrated with the CTAB and that precipitation will pass through a maximum;
The redissolution of precipitate in excess CTAB would impose a most
severe constraint on any application to commercial practice. The extent to which
redissolution will occur with any given cationic surfactant is, of course, a function
of the water solubility of the cationic itself and this is determined by the balance
between the hydrophilic quaternary centre and the hydrophobic hydrocarbon
residue. By increasing the hydrocarbon content of the cationic surfactant, it
should be possible to avoid redissolution of the precipitated colorant.
Fig. 3 shows the precipitation effect of the dialkyl quaternary ''TALO-
FLOC," which carries 2 octadecyl hydrocarbon chains on each quaternary
ammonium centre. Data are shown for 6 types of washed raw liquor and no redis-
solution is seen over the range up to 2,000 ppm (approx. 4 p,eqjg). Since greatest
interest attaches to the precipitation effect over the range 100-500 ppm T ALO-
M. -c. BENNETT, F. J. GARDINER, J. C. ABRAM, J. T. RUNDELL 1573

1400
- - • Australian

- - - DBWII

1200 --OBWIII

- - - - - t> Mixed cane/beet

- - ~ Mauritius
1000- i
- - eBWI m

800-

.'"
:;;:
oS
a:
9
8 600-

400

200

0'05 0'01 0'15

"!oTALOFLOC on solids

Fig. 3. Precipitation of colorant from 6 washed raw liquors with T ALOFLOC.

FLOC, where the decolorization lies in the range 40-70%, it is clear that with
this cationic surfactant there can be no risk of adding too much.

Electrophoresis of the Precipitated Colorant-Surfactant Complex

Previously, it was shown that the quantity of colorant precipitated from a
washed raw liquor depends upon the quantity of added cationic surfactant. With
the monoalkyl quarternary CTAB, the precipitation passes through a maximum
and further additions of CTAB cause the precipitate to redissolve. Fig. 2 Illus-
trates the well-known mechanism of this effect, which can be accounted for by
changes in the magnitude and sign of the electrical charge on the precipitated
particle. These changes can be followed using the technique of particle electro-
phoresis.
Most of the impurity in meltel' liquor is of sub-microscopic size, but some
of the larger particles are visible at magnifications of about X 400. These par-
ticles can be induced to move under a potential gradient at a rate which depends
on the magnitude of the potential gradient, the v~scosity and dielectric constant
1574 MANUFACTURING-PROCESSING

of the medium, and the electrokinetic or ~ (zeta) potential of the particle. The
electrophoretic mobility (ft) is given by the following equation:
.KD~
ft =-.--
7r7]

where K = a constant which is determined by the shape and size of the particles,
and the thickness of the electrical double layer; D _= dielectric constant of the
medium; 7] = viscosity of the medium; and ~. = zeta or electrokinetic potential,
The relationship between particle charge and zeta potential is very com-
plex, especially in this system. As the main interest in this work lies in the obser-
vation of change in zeta potential with addition of surfactant and, in particular,
in the concentration of surfactant necessary to produce zero zeta potential, no
attempt has been made to equate zeta potential with particle charge. In fact, as,
K, D, and 7] are effectively constant, all the results have been expressed in terms
of the electrophoretic mobility; ft.
In such a complex bio-colloidal system as this, the observed electrophoretic
mobility is determined by the nature of the particle surface rather than by the
composition of the particle itself. Since the suspension of particles observed in
the cell appears to be reasonably homogeneous, that is, most particles move with
the same mobility, it is reasonable to suppose that all particles present are cov-
ered by the same blanket of impurities adsorbed from solution in the liquor. Any
change in the electrical nature and solubility of anionic impurity, therefore,
affects all particles to a similar extent. .
Most of the work has been undertaken at room temperature, using 20%
solids melter liquor filtered through a BS 150 mesh screen. Electrophoresis mea-
surements were made using a Mattson cell (5) . The conclusions have been con-
firmed at higher sucrose concentrations in a thermostated capillary cell (Rank
Bros., Bottesham, Cambridge, England) .
Experiments were carried out in the following way: 65% solids washed
raw liquor was heated for 10 min at 75 C with the appropriate volume of sur-'
factant solution or the required concentration to give a final solution of 20%
solids concentration; after cooling to room temperature, this mixture was sep-
arated into 2 fractions; the Ist was filtered through kieselguhr and the percentage
color removed measured at pH 7.5 and 420 nm; the 2nd fraction was agitated in
a food blender for 30 sec to break up the large> floes and then placed in the
electrophoresis cell; measurements were made by observing a minimum of 5
particles moving through a distance of 40 microns under a maximum potential
of 5 volts/em; after each pass the current was reversed and the particle turned
in the reverse direction, following standard practice. Typical results are shown
in Fig. 4, 5 and 6.
Fig. 4 shows the effect of CTAB concentration on the mobility of the par-
ticles in a liquor of Jamaican origin. Because of the dilution, the CTAB concen-
tration is expressed in mM. Initially the impurity has a negative charge (an-
ionic), but as the concentration of CTAB added is increased, the magnitude of
the negative potential decreases to zero and then becomes increasingly positive.
This pattern has been observed subsequently for all cationic surfactants on all
cane melter liquors tested.
The relationship between color removal, CTAB concentration and electro-
M. C. BENNETT, F. ]. GARDINER,]. C. ABRAM, ]. T. RUNDELL 1575

+0·2

E 0
~

~~
-0,2
1e
u
'E
~ -0'4
:D
0
~

-2 -1 o
I0910CTAB conc~mM)

Fig. 4. The effect of added CTAB on electrophoretic mobility in Jamaican liquor diluted to
20% solids.

phoretic mobility for the system CTAB-Jamaican liquor is shown in Fig. 5. As

the concentration of CTAB added to the washed raw liquor increases, the amount
of insoluble colorant-surfactant complex removed increases to a maximum. This
maximum precipitation occurs at a CTAB concentration where the electrical
potential on the particles is effectively zero. A high potential, positive or nega-
rive, on the impurity is one of the factors leading to a high degree of stability in
colloidal systems. At low potentials the tendency for particles to flocculate
increases and the impurity-surfactant complex is more readily removed.
In the case of CTAB, the optimum precipitation of colorant lies very close
to the point of zero zeta potential. This is a function of the nature of CTAB.
and different cationics show different relationships.
It has been shown previously that T ALOFLOC, the dioctadecyl quater-
nary, does not show a maximum precipitation concentration. The relation
between its effect as a color precipitant and its effect on zeta potential is shown
in Fig. 6. Also shown in Fig. 6 is the effect of a cationic surfactant with a struc-
ture intermediate between that of CTAB (a single C 1 6 chain) and T ALOFLOC
(2 C 18 chains)'. This concerns didodecyldimethyl ammonium chloride (which
carries 2 C 1 2 chains) . It is seen that all 3 surfactants show a similar decolorization-
electrophoretic mobility relationship up to the point of zero potential. On the
positive side of the zero potential, the di-C 12 surfactant shows a behaviour inter-
mediate between CTAB and TALOFLOC. It causes redissolution of precipitated
colorant, but at higher concentrations than for CTAB. It is an important find-
ing that all 3 surfactants show the same color precipitation behaviour up to the
point of zero potential. As indicated in Fig. 2" this concerns the interaction
 I,"

1576 MANUFACTURING-PROCESSING

20

"C
QI
>.
o
E
~ 40
...
.2
o
o
#

60

80
I
I
I
I
Negative I Positive
I

1 2 3
CTAB cone. (mM)
Fig. 5. Color precipitation by addition of CTAB to ]amaicanliquor, indicating the point of
zero electrophoretic mobiltiy.

between anionic centre on the impurity and cantionic centre on the surfactant. All
3 surfactants referred to in Fig. 6 have the same quaternary nitrogen as the
cationic centre.
M. C. BENNETI, F. J. GARDINER, J. C. ABRAM, J. T. RUNDELL 1577

o
. I
9
I
\\ I
20 \\ I
\\ I
~ diC 12/
..... \
I
-0
Q)
..... \ I
>
0 40 .•.. \
E /
e ".\ I
:;
'0 .....\ /
u
'.. \
,$2 /
0
\\ /
60 /
I
I
,/
/
/
/
80 /

'"
\, _ ~)f~·" ..• ~~~1(l.~.~~~.?~~?~_.
"

Negative Positive

0,6 004 0·2 0 0·2 0-4 0'6 0'8

Electrophoretic Mobility(wsec./volt/cm.)

Fig. 6. The relation between color precipitation and electrophoretic mobility in Jamaican liquor
for 3 cationic surfactants, hexadecyl, didodecyl and dioctadecyl quaternaries.

The Relationship Between Anionic Charge on Colorant Molecules

and Precipitation by TALOFLOC
A wide variety of model colorants are available in these laboratories. Wil-
liams has used some of these to examine the action of cationic surfactants, and
has shown that the ease of colorant precipitation increases with the net charge
on the colorant anion. The model colorants used here were produced by degrad-
ing glucose in the presence of glycine (4). The product has been shown to cover
the range of properties identified in the anionic colorant types isolated from
washed raw liquors.
Ten ml of the colorant solution' (OD 420 nm = ca 20) was treated with
20 ml of 0.75% aqueous suspension of TALOFLOC. The color used here would
be equivalent to 200 ml of a washed raw liquor of color 1000 mau. At 65%
sp~ids, the liquor would contain about 170 g of sucrose, and the T ALOFLOC
 ~

1578 MANUFACTURING-PROCESSING

addition was, therefore, equivalent to 900 ppm on solids. This addition of

TALOFLOC was sufficient to precipitate 45<;'0 of the total colorant. The precipi-
tated colorant was recovered by dissolving the precipitate in ethanol and re-
precipitating with potassium bromide. Both the precipitated colorant and the
material not precipitated by surfactant were fractionated 'on a column of diethyl-
aminoethyl cellulose in the presence oJ 8 M urea. Elution with a gradient of
sodium chloride gave the profiles shown in Fig. 7 and 8. The peaks have been
numbered with the probable net anionic charge on the colorant molecules by
analogy with the work of Tomlinson and Tener (6) on oligonucleotides.

en
c:
o

~
.!:
c
.9
~c
Q)
u
e
o
u
c
'"(;
"8
Q)
>
:;:;
'"
Qj
0::

o Concentration ot Sodium Chor lde in Eluate(M) 0·45

(increasing net anionic charge of eluted colorants c-c-s-)
Fig. 7. Ion exchange cellulose fractionation of colorant precipitated by TALOFLOC.

5'Yo of colorant uncharged

til
c:
.Q
tJ
£ 4
.s
c
0

~c
Q)
u
c
0
u
c:
III
(;
"0
o
Q)

~
'"
Qj
0::

l
0 0-45
Concentiation of Sodium Chloride in Eluate(M)
(increasing net anionic charge of eluted cclorants c-s-)
Fig. 8. Ion exchange cellulose fractionation of colorant remaining after TALOFLOG treatment.
M. C. BENNETT, F. J .. GARDINER, J. C. ABRAM, J. T. RUNDELL 1579

Colorants with more highly charged molecules are seen to be concen-

trated in the surfactant precipitated material. Fractionation of natural sugar
liquor colorants on this ion exchange cellulose supports these results, for liquors
which are difficult to decolorize with TALOFLOC are found to contain relatively
low proportions of the more readily removed highly charged colorants.

The Relation Between Chemical Structure of the Cationic

and Activity as a Color Precipitant.
In the results presented previously, the magnitude of the decolorization
effect has been determined by simple filtration of 65% solids liquor, with and
without the added surfactant. This filtration requires the finest grade of kiesel-
guhr, so the filtration rate obtained is very slow. This will be discussed in greater
detail in the next section, where it will be shown that phosphatation is an excel-
lent scavenger system for the colorant-surfactant precipitate. The phosphatation
only requires the addition of phosphoric acid and lime to the surfactant-treated
liquor and provides a simple, quick and reliable method for separating the color-
ant precipitate. The following method has, therefore, been used to screen a
large number of different surfactants. The washed raw liquor was prepared at
65% solids-and treated with the required amount of each surfactant. Phosphoric
acid was then added to give a concentration 6f'OJJ2%P 20 5 on-solids; and the
liquor was limed to pHS.O. The liquor was then heated to SO C and, after a few
minutes, centrifuged at 1,000g for 15 min. Finally, the supernatant liquor was
polish filtered through 2 Whatman 42 papers precoatedwithDicalite Special
Speedflow kieselguhr. The filtrate color and surfactant concentration were
determined in the usual way.
Series 1. n-alkyl trimethylammonium bromides
Me R = C 1 2H 25
I C14H 29
R - N+ - Me Br- C 1 6H 3 3
I Ci sH 3 7
Me C 2 oH41
Results for this series of single straight chain quaternaries are shown in Fig. 9. It
is seen that activity as a color precipitant increases with increasing n-alkyl chain
length, but that there is little difference between surfactants containing 16 and
more alkyl C atoms. Minimum color values 'are given at various concentrations
of added surfactant, and these tend to decrease with increase in chain length from
C 1 2 to C 1 6 , The series of leakage curves shows clearly the improvement in sur-
factant performance with increase in n-alkyl chain length.
. Series 2. Hexadecyl cationics. In this series, the n-alkyl chain is kept
constant at C1 6 ' while other substituents are introduced. The trimethylammonium
derivative C1 6 - +NMe 3 is compared with the pyridinium derivative

. C" ~ 'NO
and the benzyl dimethylammoninm derivative C"N<Me, . CH'O. Results
are shown in Fig, 10, where it is seen that, although the 3 surfactants are similar
1580 MANUFACTURING-PROCESSING

500

C16

C12
~o

*.s~
l>:
o
o-'
u

4
en
.......
s
3.3- .
100
2 .S
~
U
t:
o
U

2 3 4
Added. surfactant~eq/g)

Fig. 9. Precipitation of colorant with n-monoalkyl trimethylammonium bromides, showing the

surfactant leakage concentration.

in performance, the activity does improve with increase in hydrophobic charac-

ter.
Series 3. Ethoxylated octadecyl quaternaries. f_,;

(Eto) n

I
e t8 - N+-Et Et.S0 4-
I .
(Eto) n

This is a range of relatively inexpensive quaternaries which are available as the

ethosulphate salts. It covers a wide range of. hydrophilic ethylene oxide chain
lengths (Eto) n- The series n = 2, 5, 9 and 15 has been used here to examine
the effect of increasing hydrophiliccharacter' in an otherwise unchanged cationic
surfactant.
The results given in 'Fig. 11 show that, as with the other: series, activity as
M. C. BENNETT, F. J. GARDINER, J. C. ABRAM, J. T. RUNDELL 1581

500

Trimethyl

400

Benzyldimethyl
0
300

II
'0
N
*.,.
I1J

0::
0
-'
0
u
200

3 ~
0-

'"
:1.
100
2
1i
~
;;:
S
U
c
0
u

2 3 4
Added surfactant (~eq/g)

Fig. 10. Precipitation of colorant with 3 hexaclecyl cationics, showing the surfactant leakage
concentration.

a color precipitant increases with hydrophobic character, but that greaterconcen"

trations of surfactant are required to achieve the lower filtrate colors. As before,
surfactant appears in the filtrate when the original addition exceeds that for
minimum color. I .

The results shown by Series 1, 2 and 3 all point to the requirement for
the molecule to he as hydrophobic as possible; this can be achieved by increasing
the number of alkyl chains in the molecule. Results for 2 dialkyl quaternaries
have been presented in earlier sections. Series 4 comprises 4 members of the
dialkyl series, while Series 5 comprises a number of tri- and tetra-alkyl derivatives.
In these experiments, the color-surfactant precipitate was separated by filtration
on a Millipore membrane (pore diameter 0.22 [.tm) , in absence of any phosphate
scavenger system.
Series 4. Dialkyldimethyl quaternaries. The surfactants were all of the
general formula
1582 MANUFACTURING-PROCESSING

400

n=5

300

'0
"..,.
C'l

'"0::
a
a-'
u

200 I
~~ () ()
()
'4

3~
100
I tr
O>
:>.
Q;
n;
2 §
.s
U
c
0
u

2
Added sUrfactant(~eq/g)
3
J
4

Fig. II. Precipitation of colorant with ethoxylated octadecyl quaternaries. The number of
ethoxy groups in each chain is indicated by n, and in the series shown n = 2, 5, 9 and 15.

R
I
N+-Me
R CI
I
Me
and in this series, R= C 1 0 ' C12> C 1 G and C 1 8 (TALOFLOq. Results are given
for 3 concentrations of each surfactant in Table 1. The results for both the di
C1 0 and the di C 1 2 derivatives show less decolorization at 0.1 % addition than at
0.05%. This indicates redissolution of the precipitate in higher concentrations
of the surfactant, as found for CTAB and all mono-alkyl derivatives. The di CHI
and di C 1 8 derivatives do not show any redissolution and the performance of the
di C 1 8 is superior to that of the di C w
Series 5. Trialkyl and tetra-alkyl quaternaries. In this series, each sur-
factant was used at a concentration of 0.05% on solids. The results are shown in
Table 2. In runs 1-3, the activity of the mono- di- and, tri-octadecyl derivatives
M. C. BENNETT, F. J. GARDINER, J. C. ABRAM, J. T. RUNDELL ]583
Table 1. The decolorization effect of dialkyldimethylammo-
nium chlorides. Untreated filtrate liquor colour : 1098 mau.

Surfactant

Concen- Filtrate Decolori-

tration color, a * 420 zation
Run T~pe (% on solids) (mau) (%)

1 0.02 759 31
2 R=Ci0 0.05 481 56
3 0.10 548 50

4 0.02 756 31
5 R=C12 0.05 473 57
6 0.10 558 49

7 0.02 719 35
8 R=C16 0.05 469 57
9 0.10 ,415 62

10 0.02 694 37
11 R=C18 0.05 463 58
12 0.10 354 68

Table 2. The decolorization effect of trialkylmethyl and tetra-

alkylammonium halides.

Run Surfactant Decolorization (%)

1 C18 N (Me) 3 Br 56
2 (C18) 2 N (Me) 2 Cl 62
3 (C18) 3 N Me Cl 55

4 (C8) 2 N (Me) 2 Br 37
5 (C8) 3 N Me Cl 40

6 (C 2) 4 N I 7
7 (C 4) 4 N Br 11
8 (4)4 N Br 27
9 (C 6) 4 N Br 34

are compared. It is seen 'that the dialkyl quaternary (TALOFLOC) shows the
best effect. The activity of the di- and tri-octylammonium surfactants, shown in
runs 4 and 5, is seen to be markedly inferior to that of the longer alkyl chain
derivatives. In runs 6-9, 4 tetra-alkylammonium compounds are compared. The
activity as a color precipitant increases strikingly with alkyl chain length, but it is
noted in Table 2 that 4 C 6 alkyl chains are considerably less effective than 1 C 1 8
chain.
Series 6. Other cationic centres. The color precipitation effect can be
demonstrated with many other types of cationic surfactants, but the really effec-
tive agents are all characterized byhaving 1 or more alkyl chains containing 16
or more carbon atoms. Some octadecyl primary and secondary amines have been
1584 MANUFACTURING-PROCESSING

found to be active, particularly when used as the acetate salts. However, the
activity is less than that of the corresponding quaternary dialkyl derivative.
(Eto) x H
A range of ethoxylated tertiary amines, of the typ~ R - N (
(Eta) y H
has been found effective when R = C I 8 and x + y = 2 or 5. In this series, the
ethoxy groups confer adequate water solubility on the surfactant molecule and,
in this respect, fulfill one of the functions of the strongly hydrophilic quaternary
ammonium centre.
Some experiments have indicated that when a small quantity of 1 of these
compounds is used, together with the normal amount of a quaternary, the result-
ing decolorization is somewhat greater than expected. This synergistic effect is
well-known for certain pairs of surfactants used in mineral are flotation processes.
The effect is shown by the results in Table 3. Some cyclic tertiary amines have

Table 3. Synergistic effect of ethoxylated tertiary amines on

TALOFLOC at 0.25% concentration, in absence of phosphata-
tion,

Tertiary surfactant

No. of ethoxy grps. Concentration Decolorization

Run (x + y) (%) (%)

I - - 77
2 2 0.02 84
3 2 0.05 83
4 2 0.07 83
5 2 0.10. 86
6 5 0.05 82

also been found to be quite active as color precipitants. In particular, glyoxali-

dene derivatives, some of which also contain several primary and secondary amine
groups, have been found to be as active as the corresponding long chain mono-
alkyl quaternary compound.

Separation of the Color-Surfactant Precipitate

It has already been mentioned that fine grades of kieselguhr filter aid are
required to retain the color-surfactant precipitate. Using coarse filter aids, the
precipitate particles pass through the filter medium and no color removal can
be achieved. In fact, in the early experiments carried out in these laboratories
the effect was missed for this reason.
Some indication of the maximum pore leakage diameter which can be
tolerated in the 'filter medium can be gained from successive filtration through
the range of millipore filters. This in no way represents a particle size distribu-
tion analysis of the precipitate because of partial closure of membrane pores dur-
ing filtration, but the technique does allow .some conclusions to be drawn con-
cerning alternative filter m e d i a . '
A 60% solids washed Mauritius raw liquor was subjected to successive
M. C. BENNETT, F. J. GARDINER, J. C. ABRAM, J. T. RUNDELL 1585

millipore filtration, the filtrate from one filtration being passed through the next
smaller pore diameter membrane. Another sample of the same original liquor
was treated with 360 ppm on .solids of TALOFLOC and the successive filtration
repeated. Results are shown in Fig. 12, where all filtrate colors are plotted as a

1500

1000

1··:···1
~ \···1; :,".

ti:""
rt:
.... -- ....

';;""-
/
....
9 / --- -r ~-,-
0
u ~~~..- ~-.. liquor + O' 036% Talofloc
500 1---1--. ",
___ ,....."..., +13ppm Floccuiant
liquor ...r'.\.. ,....Ct·····
+ 0·036% Talofloc
Special
.Speedflow .
Superai1 ~~ rpeedex.

-2 ,-1 0
I0910(pore diameter, urn)

Fig. 12. Successive millipore filtration of 60% washed raw Mauritius liquor. Filtrate colors are
plotted as a function of lOg10 pore diameter (,urn) for (a) untreated liquor, (b) liquor treated
with 360 ppm TALO'FLOC and liquor (b) further treated with 13 ppm of a polyacrylamide floc-
culant.

function of log 10 pore diameter. It is seen that in the T ALOFLOC treated

liquor, filtrates through coarse media can show an increase in color, this being
due largely to the turbidity caused by color-surfactant precipitate particles. Effec-
tive retention of the precipitate is assured with pore diameter below about 0.5
p'm, and this is why only the finest grades of kieselguhr, for example Dicalite
Superaid and possibly Special Speedflow, are satisfactory.
These findings have been borne out completely in a 2 tonjhr pilot plant
study carried out at the Montreal Refinery of Canada &: Dominion Sugar Co.,
under the supervision of S. Stachenko. Those grades of filter aid which allow
satisfactory rates of liquor throughput do not retain all the precipitate, so that
the observed decolorization is unsatisfactorily low.
To make use of the color precipitation reaction in refinery practice, it was
clearly necessary to devise some means for increasing the size of the precipitate
particles. Two such means have been examined, tl:l~",J~t involving the use of
a suitable high molecular weight anionic polymer, the·'Zna involving the use of
conventional defecation reactions, in particular phosphatation and carbonatation.
The beneficial effect of a polymer flocculant has been demonstrated by the
1586 MANUFAGTURING-PROCESSING

technique of successive millipore membrane filtration, and a typical result is

shown in fig. 12. The flocculant used was a partially hydrolysed polyacrylamide
at 13 ppm on solids. The increased flocculation was readily seen with the naked
eye, and effective separation of flocculated material could be achieved with very
coarse filter media (pore leakage diameter approximately 10· {-tm). Unfortu-
nately, the flocculation achieved by this means is not complete and some small
leakage of color-surfactant precipitate occurs. This can be seen in Fig. 12, where
lower colors are achieved in absence of the flocculant below a pore diameter of
1 {-tm. This effect has also been confirmed on the Montreal pilot plant where, at
T ALOfLOC additions up to 750 ppm, surfactant concentrations of up to 50
ppm appeared in the filtrate. With perfect filtration the leakage should be under
5 ppm. -

T ALOFLO C-Phosphatation Separation

When TALOFLOC-treated liquor, containing the whole of the color-
surfactant precipitate, is subjected to conventional phosphatation, complete floc-
culation is obtained. The amount of phosphoric acid required depends to a cer-
tain extent on the amount of TALOFLOC added, but generally lies in the range
0.02-0.05% P Z 0 5 on solids. As in conventional practice, after. addition of the
phosphoric acid, the liquor pH is adjusted to the required value (generally
around pH 7.5-8.0) by milk-of-lime, The essential requirement is that both
reagents must be extremely well mixed into the liquor stream. This is accom-
plished either by dosing the reagents separately into 2 in-line flow mixers in
series, or by dosing both reagents simultaneously into a well-stirred reaction
vessel. Both methods of reagent addition have been studied in the laboratory,
on the pilot scale, and on the full, refinery plant scale.
The flocculation produced in such a process is complete and, if the floc is
separated by flotation or by centrifuging, brilliant liquors are obtained. It is
clear that the in situ formation of calcium phosphate produces an excellent scaven-
ger for the color-surfactant precipitate. It is in this connection that the presence
of a cationic surfactant in the floc provides a 2nd and quite separate advantage.
The hydrocarbon chains of the surfactant allow the floc to' aerate merely by stir-
ring air bubbles into the system. This is the principle made use of in mineral
ore separation processes, where the special surfactant collector, adsorbed specif-
ically on only 1 type of ore particle, causes air bubbles to attach to those particles
only, thereby allowing a separation by flotation. In the sugar liquor system, the
surfactant is distributed uniformly through the calcium phosphate floc, hence,
air bubbles can be dispersed uniformly through the flocculated material.
It was expected that high flotation rates would be observed in such a sys-
tem, but this was not found to be the case. In fact, flotation" in the presence of
T ALOFLOC was observed to be somewhat inferior to that in absence of the
surfactant, at the same phosphate concentration. This effect was identified as
being due to the smaller size of the floes; although flocculation was complete, the
floes did not appear to grow to the size of those in conventional phosphatation
clarification systems .. As pointed out in the Introduction, it is believed that effec-
tive flocculation in phosphatation depends on the establishment of calcium salt
linkage with anionic impurity. It might be expected, therefore, that in the pres-
M. C.BENNE'IT, F. J. GARDINER, J. C. ABRAM, J. T. RUNDELL 1587

ence of a strongly cationic surfactant, the formation of those calcium linkages

would be affected.
The final stage in the development of a satisfactory separation method was
reached when it was shown that an addition of a few ppm of an anionic polymer
produced a spectacular growth in floc size and allowed visible flotation to occur
within a few minutes. In some systems examined, flotation to give 50% clear,
decolorized subnatant liquor occurs in less than 1 min. It is believed that the
extraordinary effect produced in this way is due to the strong interaction between
anionic centres on the polymer and cationic surfactant present in the floc.
This has established the basis for a sequence of operations, which is
described conveniently as the "TALOFLOC Phosphatation Process." The
sequence is as follows: 1) precipitation of color with cationic surfactant; 2) pri-
mary flocculation with (':alcium phosphate scavenger; 3) aeration of primary floc;
4) secondary flocculation with flotation aid.
An example of the process in operation at a number of levels of added
TALOFLOC surfactant is given inTable 4. This process has been under devel-

Table 4. Decolorization and flotation results for Mauritius liquor, colour 1162.
Phosphatation at 0.05% P205, flotation aid at 5 ppm, temperature 75 C.

Clarified liquor
TALOFLOC Flotation rate color Decolorization
(ppm) (% clear in 3 min) (a *420) (%)
200 67 505 57
400 70 429 63
600 74 374 68
800 80 338 71

opment during the last 2 years and an increasing amount of data from full-scale
refinery operations is becoming available. Jacobs clarifiers have been operated at
17 min retention time, while a new pilot clarifier designed specially for these
high flotation rates has been operated at Tate & Lyle's Thames Refinery at less
than 10 min retention time. At two other refineries, polish filtration of clarified
liquor was discontinued when it was shown that the liquor from the clarifier was
already brilliant.
In many cases, TALOFLOC is used jn conjunction with conventional
decolorization agents, such as bone char, granular carbon and powdered vegeta-
ble carbon. The savings achieved in these processes are roughly in proportion to
the decolorization provided by T ALOFLOC.

T ALOFLOC·Carbonatation Separation
Most cationic surfactants have a profound effect on the carbonatation reac-
tion and lead to extremely low filterabilities in the carbonatated liquor. With an
increasing hydrocarbon content and increasing hydrophobic character in the sur-
factant, the effect becomes less pronounced. It has now been shown that T ALO-
FLOC may be used in conjunction with carbonatation, provided it is dosed
directly into the 1st saturator rather than into the melter liquor.
Laboratory experiments have shown signi~cant gains in filterability, but
 1588 MANUFACTURING-PROCESSING

this effect has not been shown at any of the 5 carbonatation refineries where the
process has been under development.
The decolorization effect is approximately the same as that found in phos-
phatation systems, but the gain in bone char or carbon decolorization perfor-
mance is rather less. In carbonatation refineries operating a preliming stage,
there is virtually no saving in bone char and, therefore, no point in using T ALO-
FLOC. The reasons for this anomalous behaviour are not known but undoubt-
edly concern the chemical effects of high alkalinity and high temperature on
colorant and color precursors.

REFERENCES

1. Barr, T., J. Oliver, andW.···:V. Stubbings. 1948. The determination of surface active agents
in solution. J. Soc; Chern. Ind., 67:45. . .
2. Bennett, M. C. 1967; Improvements in and relating to sugar refining. British Patent No.
1224990.
3. Bennett, M. C. 1971. Preliminary results on the use of T ALOFLOC in sugar refining, Sugar
y Azucar 66 (10) :90-94.
4. Cookson, D., K. J. Parker, and J. C. WiHiams. 1970. Properties of sugar colorants removable
by ion-exchange resins. Proc, Tech. Sess. on Cane Sugar Refining Research (In Press) .
5. Mattson, S. 1928. Cataphoresis and the electrical neutralization of colloidal material. J.
Phys. Chem., 32:1532. .
6. Tomlinson, R. V., and G. M. Tener. 1963. The effect of urea, formamide and glycols on
secondary binding forces in the ion exchange chromatography of polyneucleotides on
DEAE cellulose. Biochemistry, 2:697.