Wat. Res. Vol. 33, No. 4, pp.

979±988, 1999
# 1999 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: S0043-1354(98)00294-2 0043-1354/99/$ - see front matter

REACTIVE DYES REMOVAL FROM WASTEWATERS BY

ADSORPTION ON EUCALYPTUS BARK: VARIABLES
THAT DEFINE THE PROCESS
L. C. MORAIS1, O. M. FREITAS1, E. P. GONC° ALVES2, L. T. VASCONCELOS1 and
C. G. GONZAÂLEZ BEC° A2*
1
Centro de Investigac° aÄo de Engenharia Aplicada (C.I.E.A.), Instituto Superior de Engenharia do Porto,
4200 Porto Codex, Portugal and 2Centro de QuõÂ mica da Universidade do Porto (CEQUP/ICETA),
Faculdade de Engenharia, 4099 Porto Codex, Portugal

(First received October 1997; accepted in revised form June 1998)

AbstractÐAn attempt to help solving the pollution problem caused by the presence of reactive dyes in
textile e‚uents, was undertaken. Owing to the fact that eucalyptus bark is a very abundant, inexpen-
sive, forest residue in the authors' country, Portugal, it was decided to experiment with it as a potential
adsorbent for a certain type of the supracited pollutants used in cellulose ®bers dyeing (Remazol BB).
In order to study the variables that de®ne the process, 24 and 25 factorial experimentations were carried
out, in two stages. In the ®rst one, the variables selected were: temperature, initial pH, sodium chloride
concentration and initial dye concentration/bark concentration ratio. As tests showed that the last par-
ameter did not describe the adsorption process, another stage followed, keeping the ®rst three variables
and substituting initial dye concentration and bark concentration separately for their ratio. It was veri-
®ed that all the variables studied had signi®cant in¯uence on the adsorption process at the level of 1%
probability. The in¯uence order was: initial dye concentration>bark concentration>initial pH>so-
dium chloride concentration>temperature. Parallel adsorption tests, under similar conditions, carried
out with a commercial activated carbon and with eucalyptus bark, showed for the last one an adsorp-
tion capacity about half of that of the ®rst one. It is considered that the obtained results are quite
encouraging, though there is still a long way to go before de®nitive conclusions may be withdrawn. #
1999 Elsevier Science Ltd. All rights reserved

Key wordsÐreactive dyes removal, eucalyptus bark adsorbent

INTRODUCTION cally, consequently having a more and more nega-

In the last twenty years, an increase of attention
has been directed towards the environment. It has
been increasingly noticeable the growing awareness
of the public in general and particularly of the
entrepreneurs, concerning the conditions that have
a negative impact over the environment, water qual-
ity being one of their main preoccupations (Leonas
and Leonas, 1994). This change is due to the de-
terioration, both qualitative and quantitative, of the
existing natural resources caused by the high
untreated discharges from industrial activities and
urban agglomerates.
Color removal from textile e‚uents has been the
target of great attention in the last few years, not
only because of its toxicity but mainly due to its
visibility (ICI Watercare, 1991).
Colored e‚uents have been produced, most cer-
tainly, since the dyeing technique was invented.
Through hundreds of years, obviously, the scale of
production and the nature of dyes changed drasti-
*Author to whom all correspondence should be addressed.
[E-mail: cbeca@fe.up.pt].
979
tive impact in nature (Waters, 1995).
Recent estimates indicate that, approximately,
12% of synthetic textile dyes used each year are lost
during manufacture and processing operations and
that 20% of these lost dyes enter the environment
through e‚uents that result from the treatment of
residual industrial waters (Weber and Stickney,
1993).
Reactive dyes are, commercially, a very import-
ant class of textile dyes, whose losses through pro-
cessing are particularly signi®cant and dicult to
treat. Thus, in the case of cellulose ®bers dyed with
these dyes, nearly 50% may be lost to the e‚uent
(Easton, 1995). These dyes, that didn't react with
the ®bre and are in an hydrolysed state (Cooper,
1993), encounter themselves in a di€erent and irre-
trievable form in the resulting e‚uent.
Both their low ®xation level and removal rate in
treatment stations make these dyes a unique and
separate class that must be treated as such. Indeed,
though the durability of these dyes, from the chemi-
cal and photolytic point of view, may have
increased, they became less treatable through the
biological aerobic procedures, remaining in the
980 L. C. Morais et al.
e‚uent even after the biological treatment given in
the waste waters treatment stations. One possible
treatment for the removal of soluble dyes by an
aerobic procedure may be biodegradation, but reac-
tive dyes constitute an exception (Waters, 1995).
Although some existing technologies may have a
certain eciency in the removal of reactive dyes,
their initial and operational costs are so great, that
they constitute an inhibition to dyeing and ®nishing
industries. On the other hand, low cost technologies
don't allow a wishful color removal or have certain
disadvantages, being, consequently, necessary their
integration in a more complex and complete treat-
ment plan, therefore more onerous, such as a com-
bination of biological, chemical and physical
procedures (Yeh et al., 1993). Hence, research has
been directed to other non-conventional materials
and procedures of color removal that will combine
e€ectiveness with cheapness (Laszlo, 1994).
Adsorption is one of the most e€ective physical
processes for the removal of color and treatment of
textile e‚uents. In the adsorption procedures (and
for the case of waters and waste waters) the conven-
tional material used is activated carbon (Yeh et al.,
1993) but, due to its high price it is not used on a
great scale.
Several researchers have been studying the use of
alternative materials that, although less ecient,
involve lower costs. Among the studied alternative
materials are agricultural, forest and animal by-pro-
ducts such as: peat (Coupal and Lalancette, 1976),
wheat bran, sawdust, tree barks (Michelsen et al.,
1992), chitin and others (Laszlo, 1994). However, as
these studies haven't yet, satisfactorily, solved the
problem in hands, it was decided to experiment
with eucalyptus bark as an adsorbent of reactive
dyes for cellulose ®bers (Remazol BB) due to the
fact that it is a very abundant and inexpensive ma-
terial in Portugal.
EXPERIMENTAL
Materials
Adsorbent. The eucalyptus bark used (from Eucalyptus
globulus) was collected in September 1995 from adult trees
in a region situated in the south-east of Oporto, Portugal.
Hydrolysed dye solution. Hydrolysed Remazol BB, a
vinyl sulfone type reactive dye (D±SO2 ±CH.CH2 where D
represents the cromophore group), was used in this study.
Methods
Preparation of the adsorbent. The bark was dried at
room temperature and delivered in strips with an electric
saw. Then it was ground in a cutting mill (Fritsch) whose
grinding chamber was ®tted with a sieve (1 mm holes).
The eucalyptus bark samples were not sieved or classi®ed
in several size fractions because, owing to the ®brous
nature of the material, sieving would be ine€ective, allow-
ing passage of long ®bers even through small holes. The
obtained bark was kept under conditioned air.
Preparation of the hydrolysed dye solution. Once the
reactive dyes, present in the e‚uents, are hydrolysed, it
was necessary to be aware of the industrial dyeing con-
ditions to predict the ®nal concentration of the hydrolysed
dye in the residual dyeing bath. To simulate industrial
processing the following operating conditions were con-
sidered:
. Temperature 508C, time 1 h.
. [NaCl]: 30, 45 and 60 g lÿ1; 30, 40 and 50 g lÿ1 (low,
medium and high levels for 1st and 2nd stages, respect-
ively).
. Amount of NaOH: 0.4 ml lÿ1 of a 10 M solution.
. Amount of Na2CO3: 1.6 g lÿ1.
. [dye]: 100, 350 and 500 mg lÿ1; 100, 300 and 500 mg lÿ1
(low, medium and high levels for 1st and 2nd stages, re-
spectively).
The appropriate amounts of dye, NaCl and Na2CO3
were weighed and transferred into a beaker and the appro-
priate volumes of NaOH solution and distilled water were
added to it. The solution was heated on a stirring plate at
508C, for 1 h. After cooling, pH of the solution, with a
value about 11, was adjusted adding 12 M HCl solution,
using a pH meter. The solution was transferred to a volu-
metric ¯ask and the volume was completed with distilled
water of the same pH.
Hydrolysed dye solutions, prepared under the selected
conditions, showed to be stable at 20 218C for long
periods of time, at least one month, if adequately stored.
They were kept in glass containers, wrapped with silver
sheets and maintained inside a wooden cupboard at
20218C.
Scanning curve. It was necessary to choose a sound and
expeditious method to determine the dye concentration in
solution before and after the contact with the adsorbent.
As the dye concentration may be controlled through color,
the chosen method was spectrophotometry in the visible
region.
The scanning curve, that is to say, the graph absorban-
ce = f(l) at pH 1.00, 2.00, 2.50, 4.00 and 5.00, allowed us
to determine the most suitable wavelength for absorbances
measurement. The selected wavelength was 609 nm.
Choice of the way to separate the bark from the solution.
The separation of the bark from the solution in which dye
concentration had to be determined, brought severe pro-
blems. As a matter of fact, both decantation and centrifu-
gation (33000 rpm) failed, owing to the presence of
extremely light (dust-like) particles. Ordinary ®ltering
media did not solve the problem either, because they
adsorbed the dye in a complex, aleatorious form. Fil-
tration through glass ®bre ®lters GF/A was successful as it
was observed that after having ®ltered 20 ml of suspen-
sion, dye adsorption on the ®lters did not occur anymore.
Procedure for the adsorption tests. In order to carry out
the adsorption tests it was necessary to know the equili-
brium time for the process. With this purpose several pre-
liminary kinetic tests were done. The adsorption tests were
performed as follows:
The appropriate amount of eucalyptus bark was
weighed and transferred into a beaker; 100 ml of hydro-
lysed dye solution with adequate characteristics were
added to it. The beakers were placed in a thermostatic
oven or bath, with orbital stirring (170 rpm) at the
selected temperature. After a period of time, close enough
to the equilibrium one, the beakers were withdrawn from
the oven or bath and the suspensions were ®ltered through
glass ®bre ®lters, GF/A. Then, the ®ltrates were analyzed
in a thermostatized room at 20218C by spectropho-
tometry in the visible region. Absorbances at 609 nm were
recorded. As pH could have in¯uence on absorbance it
was determined just before spectrophotometric analysis.
The adsorption capacity of the bark was determined
from the concentrations di€erence of the solution, at the
beginning and at equilibrium.
 Dye adsorption on eucalyptus bark
DEFINITION OF THE VARIABLES THAT DESCRIBE THE
ADSORPTION PROCESS
Factorial experimentation: choice of the variables
and of the domain to study
A thorough literature search concerning the util-
ization of eucalyptus bark for the treatment of
industrial e‚uents, containing hydrolysed reactive
dyes, showed no published papers, though there
were references to research work dealing with the
adsorption capacity of natural and chemically trea-
ted peat (Coupal and Lalancette, 1976; Smith and
Mark, 1978; D'AÂvila et al., 1992) and wood (Poots
et al., 1976).
As a consequence, it was necessary to carry out
tests to obtain information about the parameters
that could signi®cantly in¯uence the adsorption
process. The choice of the variables to study was
based on theoretical knowledge, concerning the sev-
eral factors that may in¯uence the adsorption exten-
sion and on information dealing with works
performed with similar materials (Vasconcelos,
1989).
Besides the theoretical aspects, there was a practi-
cal one that it was convenient to explore. As a mat-
ter of fact, if a certain goal must be reached when
treating an e‚uent, it may be interesting to have
several options concerning the conditions to use.
Though, in some cases, certain operating conditions
may have to be practically imposed (for instance, if
it is undesirable to render the process quite oner-
ous) it was advisable to know something about the
in¯uence that possible variations on the operating
conditions could have in the adsorption extension.
The ®rst thing to do in a factorial experimen-
tation, is to decide which variables must be tested;
then, the choice of the high and low levels for each
one must follow.
This study was developed in two stages.
1st stage
Choice of Variables
Adsorbent's particle size: though this may a€ect
the adsorption process, it was not considered.
Actually, it is not safe to refer particles ``size'', in
this case, owing to their shape (longitudinal ®bers).
Temperature: an adsorption decrease with a tem-
perature increase is to be expected (physical adsorp-
tion) unless there are extrinsic processes that may
be altered with thermal variations. For instance,
Youssef, 1993, who studied the adsorption of acid
dyes on cellulose and some of its derivatives, veri-
®ed that the adsorption capacity increases with an
increase in temperature. This variable's e€ect was
studied.
pH: Youssef, 1993, showed that pH in¯uences
adsorption, favoring it at low pH values. This e€ect
was explored.
981
[NaCl]: because the dyeing baths with reactive
dyes contain sodium chloride, its in¯uence was
taken into consideration.
[Dye]initial: though this variable in¯uences the
adsorption process, its e€ect was not explored by
itself, but through its relation to the adsorbent's
concentration. The reason for this is given next.
[Dye]initial/[bark] ratio: research work carried out
by the authors' group, dealing with heavy metal ions
removal through adsorption processes, showed that,
under certain circumstances, for di€erent initial
solute concentrations but for the same [solute]initial/
[adsorbent] ratio, X, identical adsorption capacities
occurred (unpublished data). It was decided to
explore the in¯uence of X.
Domain's Choice
The choice of high and low levels for each vari-
able naturally depends on the domain to be
explored. As this study aims at utilizing eucalyptus
bark as a cheaper alternative adsorbent relatively to
traditional ones, the process must avoid everything
that may render it more onerous.
Concerning the variable temperature, since the
mechanism of the adsorption process was unknown,
it was necessary to perform the study in a rather
wide temperature range. The chosen domain was
25±708C.
Consulted bibliography (Laszlo, 1994) mentioned
the fact that some reactive dyes behave like acid
dyes; since for these it was shown that when using
cellulosic materials as adsorbents, adsorption
increased with a pH decrease, it would look natural
to choose a domain in the acid zone. Nevertheless,
as we knew nothing about Remazol BB's behavior,
the ®rst study should explore a wide range. It was
selected the pH 4.00±10.00 range. However, this
had to be changed because leaching of colored com-
pounds occurred above pH 5.00 rendering the spec-
trophotometric control impracticable. For the low
and high levels of pH, values of 2.00 and 4.00, re-
spectively, were chosen.
Dealing with sodium chloride and dye concen-
trations, domains were selected so as to correspond
to what normally occurs in dyeing baths.
For sodium chloride concentrations, values equal
to 30 and 60 g lÿ1, as low and high levels, respect-
ively, were chosen. These values are not signi®cantly
a€ected as a consequence of HCl solution addition
for pH adjustment (the variation in NaCl concen-
tration was lower than 5%).
The selected initial dye concentrations were 500
and 100 ppm. Therefore, the high and low levels for
X = [dye]initial/[bark] were 100 and 5 mg of dye/g of
dry bark, respectively.
As from an industrial point of view, it is prefer-
able to consume amounts of bark, as small as poss-
ible, low concentrations of adsorbent were chosen,
namely, 5 and 20 g lÿ1.
982 L. C. Morais et al.

Table 1. Domain to explore: values of chosen variables for factorial experimentation (1st stage)

Parameter High level Medium level Low level

Temperature (8C) 70 47.5 25

[H+]initial (mol lÿ1) (pH) 10ÿ2 (2.00) 5.10  10ÿ3 (2.30) 10ÿ4 (4.00)
[NaCl] (g lÿ1) 60 45 30
[dye]initial]:[bark] (mg dye/g of dry bark) 500:5 350:7 100:20

The levels of each variable for the 1st stage are [Bark]: even lower values than the ones, pre-
condensed on Table 1. viously chosen, were used (2 and 8 g lÿ1).
2nd stage The levels of each variable for the 2nd stage are
condensed on Table 2.

Choice of Variables Planning of the experiments

The di€erence in the variables choice in this stage
relatively to the previous one, consisted solely on
the fact that dye and bark concentrations were stu-
died separately. This study was a consequence of
having realized during the tests, that the [dye]initial/
[bark] ratio didn't de®ne the adsorption process,
that is, for a certain value of this ratio, several
adsorption capacity values were found (Morais,
1997). Thus, in this stage, the selected variables
were: temperature, pHinitial, [NaCl], [dye]initial and
[bark].
Domain's Choice
Temperature: this choice was made taking in due
consideration the results obtained in stage one. The
®rst stage showed that at lower temperatures the
adsorption capacities were higher and this was why
it was decided to explore a still lower temperature
range (6±188C).
pHinitial: it was previously veri®ed that when
working in the pH 2.00±4.00 range, there was an
increase in the adsorption capacity with a pH
decrease. Therefore, wanting to know if this ten-
dency was kept, the values 1.00 and 2.50 for the
low and high pH levels, respectively, were chosen.
It seemed interesting that the selected domain
should include part of the one studied before,
because there was an indication that in the studied
domain the answer was not linear (unpublished
data).
[NaCl]: according to the latest tendency to
decrease this electrolyte's concentration in the dye-
ing baths as much as possible, the value 50 g lÿ1
was chosen as the highest level for this variable.
[Dye]initial: the same levels used in the ®rst stage
were selected.
The purpose of the tests that followed, was to
determine the in¯uence level of the selected factors,
as well as that of their interactions, in order to opti-
mize the conditions to obtain a higher adsorption
capacity of Remazol BB.
A series of experiments was planned, according
to a factorial plan, and the results were analyzed by
the study of the variance (Volk, 1969; Miller, 1986;
Barker, 1994). Doing so with the factorial planning
of the experiments, it was possible to investigate
simultaneously the e€ect of all the variables.
If we call n the number of variables to be tested,
in order to measure the e€ect of all the variables
combinations when each variable is tested at a high
and a low level, 2n experiments will be needed. The
planning also included experiments in central
points, where the variables levels were the mean of
the high and the low level, so as to be able to detect
any lack of linearity in the experimental answer
between the two extreme levels.
Planning was done for 2n experiments. For each
experimentation factor, there were a test and its
duplicate. The number of tests needed was 32 and
64 for the 1st and 2nd stages, respectively, taking
into consideration that all the parameters should be
studied (four in the 1st stage and ®ve in the 2nd) at
a high and a low level. The study of the medium
level implied four more tests for each stage.
RESULTS AND DISCUSSION
Preliminary kinetic tests
These tests aimed at determining the equilibrium
time, that is to say, the contact time between the
two phases necessary to equalize adsorption and
desorption rates.

Table 2. Domain to explore: values of chosen variables for factorial experimentation (2nd stage)

Parameter High level Medium level Low level

Temperature (8C) 18 12 6
[H+]initial (mol lÿ1) (pH) 10ÿ1 (1.00) 5.18  10ÿ3 (1.29) 3.16  10ÿ3 (2.50)
[NaCl] (g lÿ1) 50 40 30
[dye]initial (ppm) 500 300 100
[bark] (g lÿ1) 8 5 2
 Dye adsorption on eucalyptus bark 983

Fig. 1. Mean adsorption capacity of the bark as a function of the contact time between the two phases
for high and low levels of all the factors.

In order to determine the equilibrium time, sev-
eral tests with their duplicates were carried out,
conjugating the extreme conditions of the variables
with signi®cant in¯uence in the adsorption process.
Adsorption capacities of the bark after contact
times equal to 0.5, 1, 2, 3 and 6 h were determined.
The results of these preliminary kinetic tests are
shown in Fig. 1.
It can be seen that after 6 h, the system nearly
reached equilibrium. For high level tests a variation
in adsorption capacity of about 6%, between 3 and
6 h occurred. For low level tests the situation after
6 h was a bit further from equilibrium. Therefore it
was decided to explore a larger period of time. It
was concluded that between 12 h and 3 days there
was a variation in adsorption capacity of, approxi-
mately, 6%. After 3 days and until 5 days no vari-
ation occurred. Consequently it was considered that
12 h should be a suitable period of time for the
tests.
Statistical analysis of the results of the factorial ex-
perimentation
The statistical technique used was the analysis of
the variance.
1st stage
Factorial Experimentation
The results of the factorial experimentation are
summarized in Table 3 and in Figs 2 and 3, where
the values of the several parameters are shown,
both for the medium level of all the variables and
for those situations with high and low temperature
level, respectively. As it can be seen pH®nal values
are within the pH range for which calibration
curves for the dye solutions were performed. For
this pH range, pH showed no in¯uence on absor-
bance. In Table 3 and in Figs 2 and 3, the indices
of the experimentation factor correspond to the
level of each variable (0, 1 and 2 indicate the med-

Table 3. Adsorption capacity for medium level of the variables (1st stage)

[H+]initial [NaCl] [Dye]initial/[bark] Experimental Adsorption capacity [H+]®nal

Temperature (8C) (M) (g lÿ1) (ppm/(g lÿ1)) factor (mg dye/g dry bark) (M)

47.5 5.1  10ÿ3 45 350/7 Y0000 19.0 4.6  10ÿ4

984 L. C. Morais et al.

Fig. 2. Adsorption capacity values obtained from the factorial experimentation (high level of the tem-
perature; 1st stage).

ium, low and high levels, respectively), in this order:
temperature, pH, sodium chloride concentration
and [dye]initial/[bark] ratio.
Statistical Analysis
Table 4 shows the values of the interactions and
e€ects, coecients, variance of the results and stat-
istical F for the four parameters studied and their
possible interactions. The variables under study,
temperature (8C), pH, sodium chloride concen-
tration (g lÿ1) and [dye]initial/[bark] ratio (mg of dye/
g of dry bark) are symbolized by T, pH, C and X,
respectively.
Observing Table 4 and using Fisher's test, not
only e€ects and interactions without meaning can
be eliminated, but the ones that have more in¯uence
on the adsorption process can be also veri®ed.
Comparing the calculated F values with Fisher's F

Fig. 3. Adsorption capacity values obtained from the factorial experimentation (low level of the tem-
perature; 1st stage).
 Dye adsorption on eucalyptus bark 985

Table 4. Statistical analysis of the results of the factorial experimentation (1st stage)

Causes of the variation Interactions and e€ects Coecients Variance of the results Statistical F

Variables T ÿ1.36 ÿ0.04 0.06 253.11

pH 3.26 0.10 0.33 1458.15
C 2.74 2.74 0.23 1027.93
X ÿ1.07 ÿ0.03 0.04 156.39

1st order interaction T±pH 0.30 0.01 0.00 12.69

T±C ÿ0.40 0.01 0.00 22.20
T±X 1.38 0.04 0.06 260.77
pH±C ÿ0.54 ÿ0.02 0.01 39.68
pH±X ÿ1.30 ÿ0.04 0.05 230.61
C±X 1.89 0.06 0.01 491.29

2nd order interactions T±pH±C 0.08 0.00 0.00 0.83

T±pH±X 0.12 0.00 0.00 1.94
T±C±X 0.93 0.03 0.03 117.93
pH±C±X 1.85 0.06 0.11 471.06

3rd order interactions T±pH±C±X 0.81 0.02 0.02 90.68

Central point is 4.77. Variance of the experimental error is 0.00.

values [F0.01 (1,17) = 8.40 and F0.05 (1,17) = 4.45]
for probability levels of 1 and 5%, it can be veri®ed
that, in both cases, all the variables in¯uence the
adsorption process. That is to say, all the par-
ameters have signi®cant in¯uence on the adsorption
process; pH and sodium chloride concentration are
the most weighty.
Dealing with temperature±pH±sodium chloride
concentration and temperature±pH±[dye]initial/[bark]
ratio interactions it was concluded that they do not
condition the adsorption process.
When two levels are used, it is possible that the
relation between the expected answers and the cor-
responding individual values of each factor in the
mathematical model is linear. The central point of
the plan is useful to get an estimate of all curvature
e€ects that may have a signi®cant value.
This curvature e€ect is given by the di€erence
between the obtained answer and the predicted
value by the 1st order model in all the factors. If
we consider just one factor, x (as the responsible
by the eventual curvature) its principal e€ect
would be:
Ex ˆ …rx ÿ r0 † ‡ …r0 ÿ r1 † ˆ rx ÿ r1 ,
where r0 is the value of the answer corresponding
to the central point, r1 is the value of the answer,
when all the variables are at low level and rx is
the value of the answer when the variable x is at
high level and the others are at low level.
Curvature e€ect exists if (r1ÿr0) ÿ (r0ÿrx)$ 0.
The results shown in Table 5 allow us to con-
clude that all the variables are responsible by
curvature.
2nd stage
Factorial Experimentation
The results of the factorial experimentation are
summarized in Table 6 which shows the values of
the several parameters at medium level of all the
variables and Figs 4 and 5 correspond to situations
at high and low temperature levels. Once more it
can be seen that pH®nal values for initial pH 1.00
show very scarce variations and, as such, are within
the pH range for which calibration curves for the
dye solutions were performed.
In Table 6 and in Figs 4 and 5, the indices of the
experimentation factor correspond to the level of
each variable (0, 1 and 2 indicate the medium, low
and high levels, respectively) in this order: tempera-
ture, pH, sodium chloride concentration, initial dye
concentration and bark concentration.
Statistical Analysis
Table 7 shows the values of the variance of the
results and of statistical F for the ®ve parameters
studied and their possible interactions.
The variables under study, temperature (8C), pH,
sodium chloride concentration (g lÿ1), initial dye
concentration (ppm) and adsorbent concentration
(g lÿ1) are symbolized by T, pH, C, S and A, re-
spectively.
Table 5. Checking of the existence of curvature e€ect (1st stage)
Parameter (r1ÿr0) ÿ (r0ÿrx)
T 0.266 r0=0.378
pH 0.787
C 0.651 r1=0.412
X 0.352
986 L. C. Morais et al.

Table 6. Medium level of the variables (2nd stage)

Experimental Adsorption capacity

Temperature (8C) [H+]initial (M) [NaCl] (g lÿ1) [dye] (ppm) [bark] (g lÿ1) factor (mg dye/g bark)

12 5.2  10ÿ3 40 300 5 Y00000 34.1

Fig. 4. Adsorption capacity values obtained from the factorial experimentation (high level of the tem-
perature; 2nd stage).

Fig. 5. Adsorption capacity values obtained from the factorial experimentation (low level of the tem-
perature; 2nd stage).
 Dye adsorption on eucalyptus bark 987

Table 7. Statistical analysis of the results of the factorial experimentation (2nd stage)

Causes of the variation Interactions and e€ects Coecients Variance of the results Statistical F

Variables T 21.97 0.34 7.54 62.77

pH ÿ363.03 ÿ5.67 2059.23 17138.62
C 95.17 1.49 141.52 1177.85
S 1019.63 15.93 16244.46 135200.00
A ÿ510.97 ÿ7.98 4079.54 33953.27

1st order interactions T±pH ÿ3.23 ÿ0.05 0.16 1.36

T±C ÿ121.03 ÿ1.89 228.88 1904.92
T±S ÿ9.77 ÿ0.15 1.49 12.41
T±A 45.23 0.71 31.96 266.04
pH±C ÿ259.23 ÿ4.05 1050.00 8739.00
pH±S ÿ372.77 ÿ5.82 2171.21 18070.60
pH±A 177.03 2.77 489.68 4075.53
C±S 146.23 2.28 334.11 2780.76
C±A ÿ17.97 ÿ0.28 5.05 41.99
S±A ÿ149.23 ÿ2.33 347.96 2896.03

2nd order interactions T±pH±C ÿ91.83 ÿ1.43 131.76 1096.63

T±pH±S ÿ35.37 ÿ0.55 19.55 162.69
T±pH±A 53.63 0.84 44.94 374.03
T±C±S ÿ121.97 ÿ1.91 232.45 1934.63
T±C±A 119.43 1.87 222.87 1854.89
T±S±A 57.37 28.69 51.43 428.02
pH±C±S ÿ170.17 ÿ2.66 452.47 3765.80
pH±C±A 155.63 2.43 378.45 3149.76
pH±S±A 202.37 3.16 639.90 5325.78
C±S±A ÿ43.83 ÿ0.68 30.02 249.2

3rd order interactions T±pH±C±S ÿ114.77 ÿ1.79 205.81 1712.96

T±pH±S±A 66.17 1.03 68.41 569.39
T±pH±C±A 101.43 1.58 160.75 1337.90
T±C±S±A ÿ149.23 ÿ2.33 347.96 2896.03
pH±C±S±A 99.77 1.56 155.53 1294.47

4th order interactions T±pH±C±S±A 119.17 1.86 221.90 1846.82

Central point is 30.81. Variance of the experimental error is 0.12.

After calculation of the variance it was necessary,
once more, to use Fisher's test in order to eliminate
e€ects and interactions without meaning and verify
which ones in¯uence more the adsorption process.
Comparing the calculated F values with Fisher's
values (F0.01 (1,33) = 7.56 and F0.05 (1,33) = 4.17) it
was veri®ed that all the variables in¯uenced the
adsorption process at the level of 1% (Volk, 1969).
The most conditioning variable in the adsorption
process was dye concentration. Adsorbent concen-
tration, pH and sodium chloride concentration had
smaller in¯uence; temperature was the least in¯uen-
cing variable.
Relatively to 1st order interactions, pH±dye con-
centration is the most conditioning, while tempera-
ture±pH interaction has no in¯uence on the
adsorption process. All the other interactions have
in¯uence at the level of 1%.
Higher order interactions in¯uence the adsorption
process at the level of 1%.
In order to verify the existence of linearity, a
study similar to that one carried out for the 1st
stage, was done. The results shown in Table 8 indi-
cate that linearity doesn't exist and all the variables
are responsible by the curvature.
The con®rmation of nonlinearity between the two
levels studied (1.00 and 2.50 for the variable pH)
together with the results obtained in the 1st stage
that also predicted a not linear answer, when low
and high pH levels were, respectively, 2.00 and
4.00, suggests the existence of a maximum adsorp-
tion capacity in the pH domain, contained in the
two previously mentioned domains. Actually, there
is an adsorption capacity increase when pH
increases from 1.00 up to 2.50 (2nd stage) and
decreases from 4.00 to 2.00 (1st stage).
Table 8. Checking of the existence of curvature e€ect (2nd stage)
Parameter (r1ÿr0) ÿ (r0ÿrx)
T ÿ32.7 r0=34.1
pH ÿ28.3
C ÿ30.5 r1=17.2
S 4.9
A ÿ42.9
988 L. C. Morais et al.
Due to the practical character, that it is intended
to give to this work, it is important to remember
that extrapolation of the obtained results to real
situations of e‚uents treatment, should take into
consideration several other variables with possible
in¯uence on the adsorption extension.
CONCLUSIONS
The study of the variables that describe the
adsorption process of Remazol BB on eucalyptus
bark, allowed to conclude that the selected par-
ameters, solute (dye) concentration (S), adsorbent
concentration (A), pH, sodium chloride concen-
tration (C) and temperature (T), had signi®cant in-
¯uence on the process with decreasing importance,
as indicated: S>A>pH>C>T. As a rule, all the
interactions had signi®cant in¯uence on the process.
In what concerns the way in which the variables
in¯uenced the process, it is advisable to emphasize
the following:
. Initial dye concentration: obviously, this par-
ameter will a€ect the adsorption capacity. As its
in¯uence was rather pronounced, it suggests
adsorption curves of a favorable type.
. pH: there was an adsorption capacity increase
when pH decreases from 4.00 to 2.00 and
increases from 1.00 to 2.50. This indicates, that
there is an optimum pH value, for which a maxi-
mum adsorption capacity shall occur, as it was
already shown in work developed by this group,
as well as a possible interpretation, that will be,
opportunately, communicated.
. Sodium chloride concentration: generally, it was
observed that an increase in its concentration
causes an adsorption capacity increase for
pHr 2.00 and decrease for pH 1.00.
. Temperature: for temperatures higher than the
room temperature, there was an adsorption ca-
pacity decrease with a temperature increase. For
temperatures lower than room temperature the
opposite was veri®ed, except for high sodium
chloride concentration and pH 1.00. Therefore,
there is no reason to carry out the adsorption
process at temperatures higher than the ambient
one, what is bene®cial for energy costs.
The best value recorded for the adsorption ca-
pacity was about 90 mg of dye/g of dry bark what
is rather interesting. It was obtained at pH 2.50 for
a sodium chloride concentration of 50 g lÿ1, an in-
itial dye concentration equal to 500 ppm, a bark
concentration equal to 2 g lÿ1 and at 188C.
Parallel tests performed under the same con-
ditions with eucalyptus bark and with an activated
carbon (SUTCLIFFE GAC 10±30) gave for the ®rst
adsorbent a value that was about half that of the
second one. This fact allows us to conclude that
eucalyptus bark has relatively good potential to
work as an alternative adsorbent to activated car-
bon.
Therefore, there are quite promising perspectives
for its utilization on an industrial scale, keeping in
mind that it is a very abundant forest residue,
whose price may be considered negligible when
compared with that of activated carbon.
The obtained results stimulate the prosecution of
this research, because there are still several very im-
portant aspects to clarify.
REFERENCES
Barker T. B. (1994) Quality by Experimental Design, 2nd
edn. Marcel Dekker, New York.
Cooper P. (1993) Removing color from dyehouse waste
waters: a critical review of technology available. J. Soc.
Dyers Colorists 109, 97±101.
Coupal B. and Lalancette J. M. (1976) The treatment of
waste waters with peat moss. Wat. Res. 10, 1071±1076.
D'AÂvila J. S., Matos C. M. and Cavalcanti M. R. (1992)
Heavy metals removal from wastewater by using acti-
vated peat. Wat. Sci. Tech. 26, 2309±2312.
Easton J. R. (1995) The dye maker's view. In Colour in
Dyehouse E‚uent, ed. P. Cooper. pp. 9±21. Society of
Dyers and Colorists, Bradford.
ICI Watercare (1991) Color in textile e‚uent, environmen-
tal brief No. 1. In Introduction to the Environmental
Brief. ICI colours textile dyes technology group.
Laszlo J. A. (1994) Removing acid dyes from textile waste-
water using biomass for decolorization. Am. Dyest. Rep.
83(August), 17±21.
Leonas K. K. and Leonas M. L. (1994) Textileprocess
wastewater permits: an update and strategies. Am.
Dyest. Rep. 83(March), 26±34.
Michelsen D. L., Fulk L. L., Wooby R. M. and
Boardman G. D. (1992) Adsorptive and chemical pre-
treatment of reactive dyes discharges, ACS Symposium
Series 518, Emerging technologies in hazardous waste
management III, Washington, DC, pp. 119±136.
Miller R. E. (1986) Exp. Des. Chem. Eng. 23(June), 113±
117.
Morais L. C. (1997) Treatment of simulated textile e‚u-
ents by adsorption techniques. M.Sc. thesis, University
of Oporto, Portugal.
Poots V. J. P., MacKay G. and Healy J. J. (1976) The
removal of acid dyes from e‚uents using natural adsor-
bents-II wood. Wat. Res. 10, 1067±1070.
Smith E. F. and Mark H. B., Jr. (1978) Chemically modi-
®ed peat as an economical means of water treatment.
Toxicol. Environ. Chem. Rev. 2, 237±256.
Vasconcelos L. A. T. V. (1989) Utilization of pine bark
for heavy metals removal in wastewaters. Ph.D. thesis,
University of Oporto, Portugal.
Volk, W. (1969) Applied Statistics for Engineers, 2nd edn.
McGraw Hill Book Company, New York.
Waters, B. D. (1995) The regulator's view. In Color in
Dyehouse E‚uent, ed P. Cooper), pp. 22±30. Society of
Dyers and Colourists, Bradford.
Weber E. J. and Stickney V. C. (1993) Hydrolysis kinetics
of reactive blue 19-vinyl sulfone. Water Res. 27, 63±67.
Yeh R. Y. L., Liu R. L. H., Chiu H. M. and Hung Y. T.
(1993) Comparative study of adsorption capacity of var-
ious adsorbents for treating dye wastewaters. Int. J.
Environ. Stud. 44, 259±284.
Youssef B. M. (1993) Adsorption of acid dyes by cellulose
derivatives. Am. Dyest. Rep. 82(April), 30±33.