See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/244063494

Optimization of the process variables for the synthesis of cardanol-based

nivolac phenolic resin using response surface methodology

Article in European Polymer Journal · August 2007

DOI: 10.1016/j.eurpolymj.2007.05.033

CITATIONS READS

39 183

4 authors:

Ranjana Yadav Archana Ashika Devi

Forest Research Institute Dehradun Fiji National University
43 PUBLICATIONS 284 CITATIONS 4 PUBLICATIONS 203 CITATIONS

SEE PROFILE SEE PROFILE

Garima Tripathi Dr. Deepak Srivastava

Indian Institute of Technology Kanpur Harcourt Butler Technological Institute
28 PUBLICATIONS 1,024 CITATIONS 98 PUBLICATIONS 1,539 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Ranjana Yadav on 16 June 2022.

The user has requested enhancement of the downloaded file.

 EUROPEAN
POLYMER
European Polymer Journal 43 (2007) 3531–3537
JOURNAL
www.elsevier.com/locate/europolj

Optimization of the process variables for the synthesis

of cardanol-based novolac-type phenolic resin
using response surface methodology
Ranjana Yadav, Archana Devi, Garima Tripathi, Deepak Srivastava *

Department of Plastic Technology, H.B. Technological Institute, Kanpur 208 002, Uttar Pradesh, India

Received 12 January 2007; received in revised form 22 May 2007; accepted 29 May 2007
Available online 2 June 2007

Abstract

Models capable of predicting the maximum extent of conversion (p) of cardanol-based novolac-type phenolic resin,
have been developed using response surface methodology to determine the optimum reaction conditions. Three-dimen-
sional response surface and their contour plot were drawn. The maximum extent of conversion (98.93%) was predicted
when the cardanol was condensed with formaldehyde (molar ratio 1:0.652) at 119.84 °C for a time period of 3 h with
the catalyst (e.g., citric acid) concentration of 1.988% of total volume of cardanol and formaldehyde. The pH of the reac-
tion mixture was maintained at 3.0. These predicted values for optimum process conditions were in good agreement with
experimental data.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Cardanol-based novolac-type phenolic resin; Response surface methodology; Surface and contour plot; Central composite
rotatable design

1. Introduction found its niche in terms of an appropriate industrial

Cashew Nut Shell Liquid (CNSL), an agricul-
tural byproduct abundantly available in the coun-
try, is one of the few major and economic sources
of naturally occurring phenols and can be regarded
as a versatile and valuable raw material for polymer
production. Cardanol, a natural alkyl phenol from
CNSL and a potential natural resource material
for biomonomers, cannot even today be said to have
*
Corresponding author.
E-mail address: deepak_sri92@rediffmail.com (D. Srivastava).
application. By far the greatest amount of work on
polymeric materials derived from CNSL or carda-
nol has been with their use in the manufacture or
modification of phenolic resins [1–5]. The phenolic
nature of the material makes it possible to react
under a variety of conditions to form acid-catalyzed
novolac resin. Several acid catalysts having two car-
boxylic groups, separated by a chain of more than
five carbon atoms, have been extensively studied
by several researchers in the past [6–9]. Oxalic acid
has also been used as an acid catalyst [7] to produce
novolac-type phenolic resin with a pH range of
0.5–1.5. Acid-catalyzed formylation reaction of

0014-3057/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.eurpolymj.2007.05.033
3532 R. Yadav et al. / European Polymer Journal 43 (2007) 3531–3537

cardanol and formaldehyde is very rarely studied, Table 1

particularly tricarboxylic acid catalyzed reaction [8]. Physical characteristics of cardanol
Response surface methodology (RSM) is reported Property Calculated values Literature
to be an effective tool for optimizing a process, as values
Distilled From open
highlighted by various workers [10–13]. Some good cardanol market
examples of appropriate applications of this tech- Viscosity at 30 °C (cp) 41.5 40.2 38.2
nique in dairy products (cheese) [10,11], polyethylene Specific gravity (g/cc) 0.900 0.854 0.873
blends [12], grafting of methylmethacrylate (MMA) Iodine value (wijs) 288.6 280.4 279.8
Moisture content at 3.2 2.8 2.6
on nylon-6 fibers [13], etc., are the optimization of 100 °C (%)
process variables.
It is evident from the literature that no work has
been reported so far for the optimization of process
variables of the condensation reaction of cardanol
CITRIC ACID
and formaldehyde. The present investigation was,
therefore, undertaken to optimize the process vari-
ables, viz., mole ratio of cardanol-to-formaldehyde,
CARDANOL FORMALDEHYDE
catalyst concentration, reaction temperature, reac-
tion time, and pH of the reaction mixture, for the
production of cardanol-based novolac-type pheno-
REACTION MIXTURE
lic resin using RSM by adopting a five-level, five-
factor central composite rotatable design (CCRD).
Second-order model was used to generate three-
dimensional response surfaces for the extent of con-
version of novolac resin. VISCOUS MASS
(NOVOLAC RESIN)
2. Materials and methods

2.1. Materials COOLED AND PROCURED

Cardanol was obtained from M/s Satya Cashew
Pvt. Ltd., Chennai, India. Formaldehyde (37% solu-
tion) from M/s Qualikem Industries, New Delhi was
used for formylation, Oxalic, succinic and citric acid
from M/s S.D. Fine chemicals, Mumbai, was used
as catalysts. Methanol (BDH) was used to dissolve
the catalyst.
2.2. Analysis of cardanol
Cardanol, obtained from the distillation of com-
mercial CNSL under reduced pressure (1 mmHg)
and collected at 206–208 °C, was subjected to exten-
sive analysis, viz., for iodine value, viscosity, specific
gravity, etc., as per the procedure mentioned in IS
standard 840–1964 (refer Table 1).
2.3. Synthesis of cardanol-based novolac-type
phenolic resin
Cardanol-based novolac-type phenolic resin was
prepared according to the flow diagram shown in
Fig. 1 by a method similar to that adopted by Knop
FOR FURTHER ANALYSIS
Fig. 1. Flow diagram for the preparation of cardanol-based
novolac resin.
and Schieb [14] for phenol based novolac resin. The
levels of molar ratios of cardanol and formalde-
hyde, catalyst concentration, reaction temperature,
reaction time and pH of reaction mixture were var-
ied according to the experimental design. These val-
ues were suggested by previous studies carried out in
our laboratory. Free-formaldehyde content and
acid value of the reaction mixture were checked
after every 45 min to see the completion of the
methylolation reaction [8]. The reaction product
was cooled and dried under the vacuum at 60 °C
overnight before purification by column chromato-
graphy. A resin solution prepared with n-hexane,
charged to the silica gel column chromatographic
purification, was adopted mainly to remove unre-
acted components, impurities, etc., from the methyl-
olated cardanol. Purification was effected using the
elutent mixture of ethyl acetate–benzene (60:40).
The purified resin was analyzed by infra-red (IR)
 R. Yadav et al. / European Polymer Journal 43 (2007) 3531–3537 3533

and gel permeation chromatographic (GPC) Table 3

analysis. Central composite design arrangement and response
Experiment Variable levels Response
number
X1 X2 X3 X4 X5
2.4. Experimental design
1 1 1 1 1 1 0.73
2 2 0 0 0 0 0.41
The study is based on the hypothesis that the 3 0 2 0 0 0 0.62
extent of conversion is functionally related to pro- 4 0 0 0 0 2 0.43
cess variables, and attempts to fit a multiple regres- 5 0 0 2 0 0 0.62
sion equation describing the response, i.e., p. Table 6 1 1 1 1 1 0.52
2 lists variables in the descending order of assumed 7 1 1 1 1 1 0.61
8 1 1 1 1 1 0.60
importance as process variables. Care was taken to 9 1 1 1 1 1 0.78
ensure independent operation of the variables, since 10 1 1 1 1 1 0.77
an error in the choice of variables could lead to an 11 0 2 0 0 0 0.65
indeterminate solution. 12 0 0 0 2 0 0.51
The design dependent up on the symmetrical 13 1 1 1 1 1 0.66
14 0 0 2 0 0 0.70
selection of variation increments about the central 15 1 1 1 1 1 0.70
composition. These levels of variation were chosen 16 1 1 1 1 1 0.82
to be within the reasonable range, since interpreta- 17 2 0 0 0 0 0.79
tion of the result was validously within the experi- 18 1 1 1 1 1 0.58
mental limits. The levels selected were also based 19 1 1 1 1 1 0.41
20 0 0 0 2 0 0.51
on the conclusion of previous studies. The incre- 21 1 1 1 1 1 0.60
ments of variation for each variable spaced around 22 1 1 1 1 1 0.61
the center point along with the equation relating the 23 1 1 1 1 1 0.73
actual and coded ratios are presented in Table 2. By 24 1 1 1 1 1 0.64
substituting these coded for solution, process vari- 25 0 0 0 0 2 0.43
26 1 1 1 1 1 0.72
ables were coded for solution of the multiple regres- 27 0 0 0 0 0 0.49
sion equation. A central composite rotatable design 28 0 0 0 0 0 0.47
(CCRD) (Table 3) was adopted. In this design, 29 0 0 0 0 0 0.51
experiments were randomized in order to minimize 30 0 0 0 0 0 0.49
the effects of unexplained validity in the observed 31 0 0 0 0 0 0.50
32 0 0 0 0 0 0.52
response due to extraneous factors. The function
was assumed to be approximated by a second-
degree polynomial equation (Eq. (1)). were the linear, quadratic and cross-product regres-
X
5 X
5 X
5 sion terms, respectively.
Y k ¼ bk 0 þ bk i X i þ bkii X 2i þ bkij X i X j
i¼1 i¼1 i6¼j¼1

ð1Þ 2.5. Analysis of data

where bk0 is the value of fitted response at the center Multiple regression analysis was conducted for
point of design, i.e., point (0 0 0), and bki ; bkii , and bkij fitting the model represented by the equation to

Table 2
Variables and their levels for central composite design
Independent variables Symbols Levels
Coded Actual 2 1 0 +1 +2
a
Mole ratio X1 x1 0.4 0.5 0.6 0.7 0.8
b
Catalyst concentration (%) X2 x2 0.5 1.0 1.5 2.0 2.5
c
Reaction temperature (°C) X3 x3 90 100 110 120 130
d
Reaction time (s) X4 x4 2700 5400 8100 10,800 13,500
e
pH X5 x5 1.0 2.0 3.0 4.0 5.0
X1 = (x1  0.6)/(0.1); X2 = (x2  1.5)/(0.5); X3 = (x3  110)/(10); X4 = (x4  8100)/(2700); X5 = (x5  3.0)/(1.0).
3534 R. Yadav et al. / European Polymer Journal 43 (2007) 3531–3537

the experimental data. Maximization or minimiza- Table 4

tion of the polynomial thus fitted was performed Estimated coefficients of fitted quadratic equation for response
based on t-statistics
by numerical technique, using the mathematical
optimizer procedure of Quattro Pro12 of Word Per- Coefficients Estimated coefficients
fect Office 12 (M/s Corel Corporation, USA) that bk 0 0.48
deals with constraints. The mapping of the fitted bk 1 0.120833
bk 2 0.0575
response was achieved using STATGRAPHICS bk 3 0.08083
Centurion XV version 15.1.02 (M/s StatPoint Inc., bk4 0.06417
USA). The response surfaces and contour plot for bk11 0.07688
these models were plotted as a function of two vari- bk22 0.094375
ables, while keeping other variables at the optimum bk33 0.106875
bk44 0.031875
level. bk45 0.05375
R2 0.935
3. Results and discussion p < 0.05 (2.262); df = 9.

3.1. Diagnostic checking of the fitted model

bars with lengths proportional to the absolute val-
Regression analyses for different models indi- ues of the estimated effects divided by their standard
cated that the fitted quadratic models accounted values. The chart includes a vertical line at the crit-
for more than 93.7% of the variations in the exper- ical t-value for a 95% confidence level. A bar cross-
imental data, which were found to be highly signif- ing this vertical line corresponds to a factor or
icant. Multiple regression equations were generated combination of factors that have a significant effect
relating extent of conversion to coded levels of the in the response.
variables. Model was developed as follows: The regression coefficients are shown in Table 4,
as well as the correlation coefficient obtained for the
p ¼ 0:48 þ 0:120833X 1 þ 0:0575X 2 model. The correlation coefficient for extent of con-
version, p (R2 = 0.937) is quite satisfactory for
þ 0:08083X 3 þ 0:0641X 4 þ 0:076875X 21
response surfaces. The mole ratio of cardanol and
þ 0:094375X 22 þ 0:106875X 23 formaldehyde, catalyst concentration, reaction tem-
þ 0:031875X 24 þ 0:05375X 4 X 5 perature and reaction time have positive linear and
quadratic effect on the extent of conversion. The
ðdf ¼ 9; R2 ¼ 0:937Þ ð2Þ
interaction effect of reaction time and pH is highly
All main effects, linear and quadratic, and interac- significant on p at 95% level.
tions were calculated for each model. The estimated
effects were used to plot a standardized Pareto
3.2. Analysis of a variance
Chart for the model (Fig. 2); the chart consists of
When a model has been selected, an analysis of a
X1
variance is calculated to assess how well the model
+ represents the data. An analysis of a variance for
X3X3
-
X2X2 the response is presented in Table 5. To evaluate
X1X1 the goodness of the model, a F-value test was con-
X3 ducted. The F-value for extent of conversion was
X4
X2
X4X5 Table 5
X4X4 Analysis of variance for the proposed model
X2X3
Response Source of df Sum of Mean F-value
X1X2
variation squares square
0 2 4 6 8 10 Extent of Regression 9 0.164 0.018 12.36*
Standardized effect conversion, p Residual 21 0.162 0.008
Total 30 0.327
Fig. 2. Standard Pareto chart for the estimated effects of extent
*
of conversion, p. p < 0.05 (3.02) for df = 9.
 R. Yadav et al. / European Polymer Journal 43 (2007) 3531–3537 3535

12.36. On this basis, it can be concluded that the

selected model adequately represents the data for
extent of conversion. From analysis of residuals, it
is possible to conclude that they were randomly dis-
tributed around zero, and there was no evidence of
outliers.

3.3. Conditions for optimum response

Model was useful in indicating the direction in

which to change variables in order to maximize
extent of conversion. The optimum conditions to
yield maximum extent of conversion are presented
in Table 6. The optimum value of p was found to Fig. 3. Surface and contour plot between mole ratio and catalyst
be 0.9893 which was higher than the highest value concentration.
amongst the calculated values based on the experi-
mental design. The response surfaces in Figs. 3–12
is based on the aforesaid model for p (Eq. (2)) with
three variables kept constant at the optimum level
and varying the remaining two within the experi-
mental range. The surface plot along with the
contour plot of extent of conversion (p) of a carda-
nol-based novolac resin as a function of a mole ratio
of cardanol-to-formaldehyde and catalyst concen-
tration has been shown in Fig. 3. It is clear from
the figure that the increase in molar ratio or catalyst
concentration beyond the optimum value of extent
of conversion, increased the value of p. Further, at
fixed level of a mole ratio, the change of p showed Fig. 4. Surface and contour plot between mole ratio and reaction
a parabolic pattern with catalyst concentration temperature.
and vice-versa. Similar effect was observed with
mole ratio and reaction temperature (Fig. 4). The
increase in p was found to be very sharp with change
of reaction time and mole ratio (Fig. 5). The pH of
reaction medium changed extent of conversion line-
arly with a mole ratio (Fig. 6). Also, at fixed level of
a mole ratio, the effect of pH on p was found to be
uniformly increasing. Fig. 7 showed the surface and
contour plots of p as a function of catalyst concen-
tration and reaction temperature whereas Figs. 8
and 9, demonstrates the effect of catalyst concentra-

Table 6
Optimum conditions for maximum extent of conversion, p
Process variables Coded values Uncoded values
Mole ratio 0.521 0.6521
Catalyst concentration (%) 0.976 1.988 Fig. 5. Surface and contour plot between mole ratio and reaction
Reaction temperature (°C) 0.984 119.84 time.
Reaction time (s) 1.000 10,800
pH 0.000 3.000
tion with reaction time or pH, respectively, on the
Maximum value of extent of conversion, p = 0.9893
extent of conversion, p. The increase of reaction
3536 R. Yadav et al. / European Polymer Journal 43 (2007) 3531–3537
Fig. 6. Surface and contour plot between mole ratio and pH.
Fig. 7. Surface and contour plot between catalyst concentration
and reaction temperature.
Fig. 8. Surface and contour plot between catalyst concentration
and reaction time.
Fig. 9. Surface and contour plot between catalyst concentration
and pH.
Fig. 10. Surface and contour plot between mole ratio and
reaction time.
temperature and catalyst concentration decreased
the value of p up to an optimum value and then
increased.
The surface plot of p as a function of reaction
temperature and reaction time is shown in Fig. 10.
The figure clearly evidenced that the reaction tem-
perature and reaction time affected p in a similar
fashion as that affected by mole ratio and reaction
time (Fig. 5). Change of reaction temperature with
pH (Fig. 11) demonstrated that the extent of conver-
sion increased progressively with reaction tempera-
tures at higher pH while at lower temperatures
change due to pH did not produce any significant
effect. The value of p linearly decreased with pH
and increased with reaction time (Fig. 12).
 R. Yadav et al. / European Polymer Journal 43 (2007) 3531–3537
Fig. 11. Surface and contour plot between reaction temperature
and pH.
Fig. 12. Surface and contour plot between reaction time and pH.
4. Conclusion
It may concluded that using RSM, with a mini-
mum number of experiments, can effectively opti-
mize the condensation reaction of cardanol and
formaldehyde to produce novolac-type phenolic
resin. Computerized computations and model build-
ing will go a long way to unrevealing the complexity
of the process with different sets of conditions. The
View publication stats
3537
maximum extent of conversion (98.93%) was pre-
dicted when the cardanol was condensed with form-
aldehyde (molar ratio 1:0.652) at 119.84 °C for a
time period of 3 h with the catalyst (e.g., citric acid)
concentration of 1.988% of total volume of cardanol
and formaldehyde. The pH of the reaction mixture
was maintained at 3.0. These predicted values for
optimum process conditions were in good agree-
ment with experimental data. The condensation
reaction of cardanol and formaldehyde with other
catalyst systems such as succinic and oxalic acidc
under the obtained optimum conditions to ensure
good reproducibility between different experiments.
Acknowledgements
This work was sponsored by Council of Science
and Technology, Uttar Pradesh, Lucknow. Also,
we thankfully acknowledge the contributions ren-
dered by M/s Satya Cashsew Pvt. Ltd., Chennai
for providing cardanol as gift samples.
References
[1] Menon ARR, Pillai CKS, Sudha JD, Mathew AG. J Sci Ind
Res 1985;44:324.
[2] George J, Mitsutashi M, Yusyase O, Shimizi K. Adv Mater
2001;13(10):715.
[3] Prabhakaran K, Narayan A, Pvithram C. J Eur Ceram Soc
2001;21(6):2873.
[4] Niimuru N, Miyakoshi T. Int J Polym Anal Char
2003;8(1):47.
[5] Huong NLe, Nieu NH, Tan TTM, Griesser UJ. Angew
Makromol Chem 1996;243:77.
[6] Lee YK, Teak DJK, Hwang S, Refarlovich M, Sokolar AA.
J Appl Polym Sci 2003;89:2589.
[7] Vlosova EG, Medvedeva LN, Vlasova NV. Lakokrasochnye
Materiatly Ikh Primenenie 1990;5:858.
[8] Devi A, Srivastava D. J Appl Polym Sci 2006;102:2730.
[9] Urbanski J, Czerwinkski W, Ianika K, Majewsta F, Zowall
H. In: Hand book of analysis of synthetic polymer and
plastic. Chickester: Ellis Horword Ltd. Pub.; 1977. p. 179.
[10] Pintado ME, Malcata FX. Food Res Int 2000;33:821.
[11] Thakur S, Saxena DC. Lebensm Wiss Technol 2000;33:354.
[12] Srivastava D, Garg R, Saxena DC, Mathur GN. Macromol
Mater Eng 2000;283:81.
[13] Srivastava D. J Polym Eng 2002;22:457.
[14] Knop A, Pilato L. Chemistry and application of phenolic
resins. Berlin: Springer; 1979.