Optimization of adiponitrile hydrolysis in subcritical water using an orthogonal array design

Article abstract of DOI:10.1007/s10953-008-9362-3

A study of the hydrolysis of adiponitrile (ADN) was performed in subcritical water to research the dependence on experimental conditions. An L₂₅(5⁶) orthogonal array design (OAD) with six factors at five levels using statistical analysis was employed to optimize the experimental conditions for each product in which the interactions between the variables were temporarily neglected. The six factors were adiponitrile concentration (ADN c, wt%), temperature (T), time (t h), percentage of additives (reactant/additive, wt/wt%), additives (A), and pressure (p, MPa). The effects of these parameters were investigated using the analysis of variance (ANOVA) to determine the relationship between experimental conditions and yield levels of different products. The results showed that (ADN c) and T had a significant influence on the yields of adipamide, adipamic acid, and adipic acid at p<0.05. Time was the statistically significant factor for the yield of 5-cyanovalermic acid at p<0.05 and (ADN c) was the significant factor for the yield of 5-cyanovaleramide at p<0.1. Finally, five supplementary experiments were conducted under optimized conditions predicted by the Taguchi method; the results showed that the yield obtained of each product was no lower than that of the highest in the 25 experiments. Carbon balance was calculated to demonstrate the validity of the experimental technique and the reliability of the results. Based on the experimental results, a possible reaction mechanism was proposed.

Full text of DOI:10.1007/s10953-008-9362-3

J Solution Chem (2009) 38: 241–258
DOI 10.1007/s10953-008-9362-3
Optimization of Adiponitrile Hydrolysis in Subcritical
Water Using an Orthogonal Array Design
Peigao Duan · Yuanyuan Wang · Yan Yang · Liyi Dai
Received: 20 August 2008 / Accepted: 7 October 2008 / Published online: 2 December 2008
© Springer Science+Business Media, LLC 2008
Abstract A study of the hydrolysis of adiponitrile (ADN) was performed in subcritical
water to research the dependence on experimental conditions. An L25(56) orthogonal array
design (OAD) with six factors at five levels using statistical analysis was employed to op-
timize the experimental conditions for each product in which the interactions between the
variables were temporarily neglected. The six factors were adiponitrile concentration (ADN
c, wt%), temperature (T ), time (t h), percentage of additives (reactant/additive, wt/wt%),
additives (A), and pressure (p, MPa). The effects of these parameters were investigated us-
ing the analysis of variance (ANOVA) to determine the relationship between experimental
conditions and yield levels of different products. The results showed that (ADN c) and T
had a significant influence on the yields of adipamide, adipamic acid, and adipic acid at
p < 0.05. Time was the statistically significant factor for the yield of 5-cyanovalermic acid
at p < 0.05 and (ADN c) was the significant factor for the yield of 5-cyanovaleramide at
p < 0.1. Finally, five supplementary experiments were conducted under optimized condi-
tions predicted by the Taguchi method; the results showed that the yield obtained of each
product was no lower than that of the highest in the 25 experiments. Carbon balance was
calculated to demonstrate the validity of the experimental technique and the reliability of
the results. Based on the experimental results, a possible reaction mechanism was proposed.
Keywords Adiponitrile hydrolysis · Orthogonal array design · Subcritical water · Carbon
balance · Mechanism
1 Introduction
Hydrolysis is a chemical reaction or process in which a chemical compound is broken
down by reaction with water. Nitriles command special attention because of their appear-
ance in industrial waste streams and in the product spectra from the reaction of aliphatic-
P. Duan · Y. Wang · Y. Yang · L. Dai (
)
ꢀ
Shanghai Key Laboratory of Green Chemistry and Green Process, Department of Chemistry, East China
Normal University, Shanghai 200062, People’s Republic of China
e-mail: lydai@chem.ecnu.edu.cn

242
J Solution Chem (2009) 38: 241–258
NO2-containing compounds in high-temperature water [1]. Most of the nitriles are indus-
trially produced as intermediates and building blocks in organic synthesis and as organic
solvents [2]. Therefore, it is very important to know how fast they hydrolyze because, in
this way, waste streams can be cleaned. The hydrolysis of nitriles is also an area of great
synthetic significance for the preparation of amides (e.g., acrylamide or nicotineamide) and
carboxylic acids (e.g., R-(−)-mandelic acid and S-(+)-ibuprofen) in view of industrial ap-
plications and pharmacological interest in both classes of these compounds. At present,
there is no completely satisfactory method for the conversion of nitriles to amides or acids.
In practice, the reaction between nitriles and water would be so slow as to be completely
negligible. Nitriles can be hydrolyzed under either acidic or alkaline conditions to make
carboxylic acids or their salts [3, 4]. The final result is similar in both cases, but the exact
nature of the final product varies depending on the conditions used for the reaction. Acid-
catalyzed hydrolysis is used commercially for making amides from nitriles, and by further
heating with water the amides can be hydrolyzed to acids with the production of ammo-
nium salts [5]. Both the production of amides and acids, catalyzed by acid, can result in the
formation of ammonium salts as by-products, which present a difficult handing and waste
disposal problem. Increasing concern about water pollution can only mean an increasing and
costly by-product disposal problem for the acid-catalyzed hydrolysis process. Hydrolysis of
nitriles using alkali metal hydroxides was also reported in the literature [6, 7]. In this case,
instead of getting an ammonium salt if the reaction only involved water, a sodium salt would
be produced, which results in a sodium ion disposal problem and thus does not avoid the
ecological problem [8]. Ammonia gas is given off as well. For example, sodium hydroxide
can be used for the hydrolysis reaction, but again the sodium ion disposal problem exists. It
is also known that ammonia at high temperature and for extended times will hydrolyze ni-
trile but this process also has problems. In the first place pressure equipment is required and
in any long reaction time, yields suffer because by-products are frequently colored bodies
which degrade product quality.
Water is an important material in the chemistry. It is generally agreed that water is an
environmentally benign solvent, and its potential as a solvent depends on its unique system
of hydrogen bonding. As water is heated, its normally strong hydrogen bonds are weakened,
allowing dissociation that forms acidic hydronium ions and basic hydroxide ions [9, 10]. At
the subcritical stage, the ion product, Kw, is up to three orders of magnitude larger than what
it is at room temperatures and pressures. Subcritical water has properties similar to those of
polar organic solvents such as ethyl alcohol or acetone. Its dielectric constant drops from
80 to 20, and its density drops from 1 to 0.7 g·cm−3. Simply from the dissociation of water
into acidic and basic ions, much larger amounts of acid and base are produced in subcritical
water. We can use these acidic and basic ions to run acid-catalyzed and base-catalyzed re-
actions without the addition of mineral acids. Subcritical water also offers advantages in the
most costly part of most chemical processes: separation of the chemical products [11, 12]. In
subcritical processes, reaction products can be removed by cooling the water and reducing
the pressure, dropping them out of the solution. Water-soluble catalysts can also be reused if
they remain in the water. Use of subcritical water can also eliminate production of salts that
result from the neutralization of acids used in conventional separations. Disposal of these
salts can be a significant environmental issue.
Nitriles can hydrolyze in supercritical or subcritical water without the addition of any cat-
alysts [13–15]. Thus far, many investigations have concentrated on the hydrolysis reactivity
and kinetics of aliphatic and aromatic mononitriles, but few studies have been undertaken on
dinitriles. We have successfully realized the hydrolysis of iminodiacetonitrile in subcritical
water without the addition of any catalysts and the preparation of iminodiacetic acid salts as-
sisted with additives [16, 17]. Adiponitrile (ADN) is the one of the important alkyl dinitriles,

J Solution Chem (2009) 38: 241–258
243
which can hydrolyze partially or completely to produce valuable intermediates. Tradition-
ally, the hydrolysis of ADN needs chemical catalysts which promote the hydrolysis reaction.
Representative classes of catalysts primarily include nitrile hydratase, phosphoric acid or its
salts, aliphatic or aromatic acid, etc. [18, 19]. The strong self-dissociation ability of subcrit-
ical water can offer rich sources of both hydronium cations and hydroxide anions, making
it a potentially safe and environmentally compatible solvent, catalyst or reactant, in which
hydrolysis reactions of ADN can be conducted. A statistical method was used to investigate
the influence of the experimental parameters on the yield of different products.
Orthogonal array design (OAD), also known as the Taguchi method, is believed to in-
corporate the advantages of the simplex method and factorial design [20, 21]. It arranges
different factors for effective optimization of the experimental conditions. The results of the
OAD experiment can be statistically treated with two ways: analysis of variance (ANOVA)
and direct observation analysis [22, 23]. In the ANOVA, the effects of different factors on
response functions can be evaluated by both the significance (F ratio) and PC% (percentage
contribution) values. In other words, the importance of a factor or the interaction among
different factors can be estimated from the significance and PC% values. Direct observa-
tion analysis is also called range analysis, the fluctuation range and tendency of response
functions versus the levels of different factors can be directly observed from a broken linear
plot. Using ANOVA and direct observation analysis of the experimental results, factors that
significantly affect the output responses can be found, and optimal parameters for an analyt-
ical procedure can be obtained. Use of OAD can also simplify the experimental procedures
without affecting the quality of the results. Known as a chemometric method, OAD has
been widely applied for the optimization of analytical procedures in recent years [23–25].
Here for the first time, we use OAD for the optimization of conditions for adiponitrile hy-
drolysis in the subcritical water system. In the present study, a L25(56) OAD procedure was
performed with six factors (adiponitrile concentration, temperature, time, percentage of ad-
ditives, additives, and pressure) at five levels since it was suitable for the conditions being
investigated, and each experiment was repeated three times under the same conditions.
2 Experimental Sections
2.1 Materials
5-Cyanovaleramide (5-CVAM), 5-cyanovaleric acid (5-CVAC) and adipamic acid (ADAA)
were prepared and purified according to methods reported in the literatures [26, 27]. Adi-
pamide (ADAM) and adipic acid (ADA) were purchased from Aldrich. Adiponitrile was
obtained from Alfa Aesar. Ammonia solution (25 wt-%), terephthalic acid and sodium sul-
fate were commercially available from China National Medicines Corporation Ltd. All the
chemicals were used as received. Freshly double-distilled water was used throughout the
experiments.
2.2 Procedures
Experiments were conducted in a 16.5 cm3 316 stainless steel batch reactor consisting of
a body and fastening cap. A common inlet and outlet was provided for charging and evac-
uating the reactor. The maximum possible working conditions of this reactor were 500 ◦C
and 50 MPa. The temperature of the solution in the reactor was indirectly measured by a
thermocouple inserting into a dip nozzle which was installed at about the middle of the

244
J Solution Chem (2009) 38: 241–258
Fig. 1 Schematic representation
of batch reactor for subcritical
water generation
reactor (Fig. 1). The reactor was first loaded with reactant and double-distilled water and
different amounts of additives before being sealed. The water loadings were selected such
that 95% of the reactor volume would be occupied with liquid water at the reaction temper-
ature. Water densities were taken from the steam tables [28]. The mass of reactant charged
into the reactor was calculated based on wreactant/wwater × 100% at a given reactant concen-
tration. The concentrations of reactant and additives in the solution were determined based
on the solubility at room temperature. Prior to each experiment, the reactor was purged with
high purity nitrogen for about 30 min to eliminate any air. The method of pressure control of
the reaction system involved changing the amount of high purity nitrogen in the top of the
reactor. Pressures given in the Table 2 are the total pressures which include the partial pres-
sure of nitrogen and the water saturation vapor pressure at the desired temperature, ensuring
that nearly a liquid phase was present throughout the reaction. It is preferable to insure, by
adapting the pressure condition, that a liquid phase is present throughout the reaction. Under
the experimental conditions, roughly no lower than 95% of the reactor volume was occupied
by the liquid phase. In the case of the addition of aqueous ammonia, there was no change in
the concentration of ammonia in the reaction solution after it was purged with high purity
nitrogen as established by the titration with standard hydrochloric acid.
The loaded reactor was submerged in a pre-heated molten salt (components of sodium
nitrate, potassium nitrate, sodium nitrite) tank preheated at 100 ◦C above the desired tem-
perature. If the preheated temperature was set at the desired value, longer time would be
required to reach the desired temperature. For example, at a given temperature of 250 ◦C,
the salt bath was set at 350 ◦C and when the bath reached 350 ◦C, the loaded reactor was
submerged directly into it. After about one minute, the temperature in the reactor reached
100 ◦C, then it increased rapidly for four minutes; the last minute was the transition to steady
state. Five minutes later, the temperature reached 250 ◦C as described in Fig. 2. The tem-
perature range of the process was determined by the hydrolysis degree of the ADN within
the investigated range of time. At temperatures below 250 ◦C, the reaction rate was gener-
ally too low to be economically useful. At temperatures above 330 ◦C, significant amounts
of by-products were formed. The reaction was carried out at a residence time of from two

J Solution Chem (2009) 38: 241–258
245
Fig. 2 Temperature profile in the
reactor and salt bath tank as a
function of time (set temperature:
◦
250 C)
Fig. 3 Temperature profile in the
reactor at quenching stage as a
function of time (reaction
◦
temperature 250 C)
to six hours. A residence of below two hours generally gave too low a conversion, longer
residence time also resulted in the formation of by-products. About five minutes later, the
temperature in the reactor reached the desired temperature corresponding to zero reaction
time. The temperature was controlled to maintain isothermal conditions within 1◦C by an
Omega temperature controller (precision of 1◦C) assisting with a thermocouple inserting
a dip nozzle into the reactor. At the end of each reaction, the reactor was removed from the
molten salts tank and rapidly quenched in an ice bath water to stop the reaction. The tem-
perature profile in the reactor at the quenching time is described in Fig. 3. One minute later,
the temperature in the reactor reached the room temperature. The cooled system was first
depressurized and the final reaction mixture was poured out of the reactor. Reactors were
tested for leaks by weighing before and after each reaction. Hydrolysis products were recov-
ered by rinsing the reactor with pure water three times and collected as the final resulting
mixture.
The contribution of the preheating stage where the hydrolysis of ADN may proceed to
some extent was subtracted by conducting blank experiments. The results of these blank

246
J Solution Chem (2009) 38: 241–258
experiments revealed that lower-level conversion only occurred in the case of temperatures
higher than 330 ◦C. While estimating the kinetic parameters, these conversions were duly
accounted for. These experimental results were favorable for obtaining reaction kinetics
information. The uncertainty in these estimated product yields reported herein was about
95% confidence interval for all of the products based on three replicate experiments.
The liquid samples obtained were quantitatively analyzed by a HPLC (Water 1525HPLC)
equipped with an ODS C18 column (4.6 mm i.d., 250 mm, particle size 5 μm). The con-
ditions used for the analysis were as follows: mobile phase of pH = 2.5, 25 mmol·dm−3
H3PO4 containing 1.1%v/v methanol in water; flow rate of 1.0 mL·min−1; and a detection
wavelength of 200 nm. Product identification was achieved by comparing the results with
those for a standard solution of pure compounds and inspecting the high-performance liquid
chromatography-mass spectra.
2.3 Orthogonal Array Design
To seek optimal conditions for the yield of each product, the effects of some parameters on
the hydrolysis of ADN were investigated. A L25(56) OAD was used in the present hydroly-
sis procedure to optimize the experimental conditions for obtaining a high yield of different
products. It was noteworthy that appropriate selection of the factors of OAD could simplify
the experimental procedures without affecting the quality of results. The design involved six
factors at five levels as shown in Table 1. The order of experiment was obtained by inserting
parameters into columns of an orthogonal array and the experimental order was made ran-
dom and is presented in Table 2. The interactive effect of the parameters was not taken into
consideration because some preliminary tests showed that they could be neglected. Each of
the 25 experiments was performed in triplicate, corresponding to a total of 75 tests. For the
additives factor, we chose different kinds of additives (including no additive) to investigate
their effects on the hydrolysis rate of ADN and the distribution of the reaction products. We
observed the pH effect with the addition of ADA, terephthalic acid (PTA), NH3·H2O and
the neutral salt of Na2SO4. The analysis of data when there is no interaction includes (1) de-
termining the optimal conditions, (2) identifying the individual influence of each factor, and
(3) estimating the performance of the hydrolysis of ADN at the optimal conditions.
6
Table 1 Experimental parameters and their levels in L (5 ) OAD
25
Parameters
Levels
1
2
3
4
5
ADN concentration (wt-%)
1
2
4
6
8
◦
Temperature ( C)
250
2
270
3
290
4
310
330
6
Time (h)
5
a
Percentage of additives
2
4
6
8
10
b
Additives
No
ADA
20
PTA
27
NH ·H O
Na SO
2 4
3
2
Pressure (MPa)
14
35
43
a
Ratio of additive/ADN (wt/wt)
b
Without additives

J Solution Chem (2009) 38: 241–258
6
247
Table 2
L (5 ) OAD matrix randomize experimental plan table
25
a
p
No.
ADN con.
(wt-%)
T
t
A/R
Additive
◦
( C)
(h)
(wt/wt)%
(MPa)
1
2
1
1
1
1
1
2
2
2
2
2
4
4
4
4
4
6
6
6
6
6
8
8
8
8
8
250
270
290
310
330
250
270
290
310
330
250
270
290
310
330
250
270
290
310
330
250
270
290
310
330
2
3
4
5
6
3
4
5
6
2
4
5
6
2
3
5
6
2
3
4
6
2
3
4
5
2
None
14
20
27
35
43
43
14
20
27
35
35
43
14
20
27
27
35
43
14
20
20
27
35
43
14
4
ADA
3
6
PTA
4
8
NH ·H O
3
2
5
10
6
Na SO
2
4
6
NH ·H O
3
2
7
8
Na SO
2
4
8
10
2
None
ADA
PTA
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
4
10
2
ADA
PTA
4
NH ·H O
3
2
6
Na SO
2
4
8
None
4
Na SO
2
4
6
None
ADA
PTA
8
10
2
NH ·H O
3
2
8
PTA
10
2
NH ·H O
3
2
Na SO
2
4
4
None
ADA
6
a
Pressure was the total pressure of the partial pressure of nitrogen and the water saturation vapor pressure at
desired temperature
3 Results and Discussions
3.1 Identification of Intermediate Compounds and Final Products
There were five compounds resulting from the hydrolysis of ADN, including 5-cyanovaler-
amide, adipamide, 5-cyanovaleric acid, adipamic acid, and adipic acid detected by HPLC
(Fig. 4). Table 3 shows the analysis of hydrolysis products at different experimental condi-
tions. Under the conditions of higher temperatures and pressures, 5-cyanovaleramide result-
ing from the first step of ADN hydrolysis would preferably be converted to adipamide than
to 5-cyanovaleric acid, so some 5-cyanovaleric acid remained in the products along with the
proceeding of the reaction.

248
J Solution Chem (2009) 38: 241–258
Fig. 4 Chromatograph of the different hydrolysis products of adiponitrile in subcritical water assisted with
◦
6% NH ·H O at a temperature of 250 C, time of 3 h, pressure of 30 MPa. A adipic acid, B adipamide,
3
2
C 5-cyanovaleric acid, D 5-cyanovaleramide, E adipamic acid
Table 3 Hydrolysis product analysis
No.
ADAM
ADAA
5-CVAM
5-CVAC
ADA
b
c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
1.20
1.67
11.67
16.45
16.30
2.61
14.99
15.84
1.21
13.73
0.20
0.09
2.17
0.75
1.09
0.59
13.02
1.67
0.11
4.33
0.30
5.29
0.15
3.36
1.16
3.47
7.18
7.29
0.24
0.27
0.28
ND
ND
ND
1.63
1.17
2.94
0.10
0.01
2.33
2.44
2.91
0.99
1.75
5.01
3.31
3.62
1.14
6.30
7.82
9.60
7.18
1.50
7.20
12.13
13.00
6.36
0.76
0.30
4.92
1.30
ND
0.06
0.11
0.54
ND
ND
ND
23.43
20.99
25.12
8.93
11.20
25.76
17.96
20.64
7.35
32.77
35.82
42.48
28.08
4.69
24.81
28.18
31.92
13.56
1.86
ND
1.40
4.30
2.80
22.07
17.74
17.20
2.07
3.13
15.63
1.80
4.45
2.31
2.86
13.89
2.58
2.09
1.47
7.90
6.78
0.30
0.72
ND
0.17
0.40
0.32
0.01
0.50
0.02
0.58
ND
0.02
0.40
0.22
0.08
ND
0.30
0.80
a
The uncertainty in these estimated products yield reported herein is about 95% confidence interval for all of
the products
b
Percentage of weight (wt-%)
c
“ND” signifies that the material was not detected

J Solution Chem (2009) 38: 241–258
249
Fig. 5 Carbon balance for the
ADN hydrolysis in
high-temperature water
Table 4 Carbon balance for the hydrolysis of ADN at high temperature water
t (h)
ADN
ADAM
ADAA
5-CVAM
5-CVAC
ADA
−2
−2
−2
−2
−2
−2
0
2
3
4
5
6
9.30 × 10
7.21 × 10
6.03 × 10
4.05 × 10
3.03 × 10
2.73 × 10
0
0
0
0
0
0
−3
−3
−3
−3
−3
−2
−2
−2
−2
−2
−2
−2
−2
−2
−2
1.10 × 10
1.49 × 10
3.92 × 10
2.69 × 10
1.93 × 10
0.15 × 10
1.07 × 10
1.51 × 10
1.49 × 10
1.24 × 10
1.37 × 10
1.45 × 10
1.72 × 10
1.26 × 10
0.71 × 10
0
−3
0
0.28 × 10
4.48 × 10
1.19 × 10
0.81 × 10
−3
−3
−3
−3
−3
−3
0.06 × 10
0.10 × 10
0.50 × 10
3.2 Carbon Balance
The carbon balance of the hydrolysis process was calculated to demonstrate the effective-
ness of the experimental technique and the reliability of the results. The total carbon charged
into the reactor was the content of carbon in adiponitrile. The total carbon in the hydroly-
sis reaction mixture is the sum of the masses of carbon in adiponitrile, 5-cyanovaleramide,
5-cyanovaleric acid, adipamide, adipamic acid and adipic acid. The carbon mass was cal-
culated based on the content of these compounds in the reaction mixture detected by high
performance liquid chromatography. Changes of the overall carbon weight in the reaction
mixture between particular stages can be observed in Fig. 5. The content of carbon in each
product is also listed in Table 4. The specification of the process enables the process to be
monitored and provides an easy comparison of the participation of the main substances in
the whole hydrolysis scheme. The same type of carbon balance was calculated for all exper-
iments. If the amounts of carbon in the reactant and the reaction mixture were computed,
the ratio of carbon charged into the reactor to the carbon in the final reaction mixture can
be determined as C% mass balance = (Output C ÷ Input C) × 100%. Due to possible side
reactions, especially at high temperatures and longer reaction times, the average deviation
of the carbon balance was 46% as presented in Fig. 5.
3.3 ANOVA analysis of the OAD
The purpose of the OAD was to select the optimal experimental parameters for obtain-
ing a maximum yield of each product that would form adipamide rather than convert to
5-cyanovaleric acid, resulting in little of the latter remaining in the products. Statistical
analysis of variance (ANOVA) was performed to see whether the process parameters were
statistically significant. The significance (F ratio) analysis is shown in Table 5 where the

250
J Solution Chem (2009) 38: 241–258
6
Table 5 ANOVA analysis and percent contribution for the yield of each products in the L (5 ) OAD
25
matrix
a
b
ꢀc
d
Source
of variation
SS
Degree of
freedom
Mean square
F
SS
PC%
5-CVAM
**
ADN c (wt-%)
194.9
181.8
49.4
50.4
69.3
95.3
24.3
665.3
4
4
4
4
4
4
4
28
48.7
45.5
12.4
12.6
17.3
23.8
12.4
–
4.40
4.09
1.00
1.29
1.65
1.98
–
170.6
157.5
25.1
27.0
25.3
3.8
◦
T ( C)
t (h)
A/R (wt/wt)%
A
26.1
3.8
45.0
6.8
p (MPa)
Error
Total
70.9
150.6
665.3
10.7
22.6
100.0
–
ADAM
*
ADN c (wt-%)
170.7
86.2
24.0
16.2
8.4
14.4
8.4
328.3
4
4
4
4
4
4
4
28
42.7
21.6
6.0
4.0
2.1
3.6
2.1
–
20.24
10.22
2.85
1.92
1.00
1.71
–
162.3
77.8
15.6
7.8
49.4
23.7
4.8
◦
*
T ( C)
t (h)
A/R (wt/wt)%
A
2.4
0.0
–
p (MPa)
Error
Total
6.0
1.8
50.4
328.3
12.9
100.0
–
ADAA
*
ADN c (wt-%)
983.4
1508.7
202.5
170.5
249.5
117.9
117.9
3232.5
4
4
4
4
4
4
4
28
245.9
377.2
50.6
42.6
62.4
29.5
29.5
–
8.34
865.5
1390.8
84.6
26.8
46.7
2.6
◦
*
T ( C)
12.80
1.72
1.45
2.12
1.00
–
t (h)
A/R (wt/wt)%
A
50.0
1.5
131.6
0.0
4.2
–
p (MPa)
Error
Total
589.5
3232.5
18.2
100.0
–
5-CVAC
ADN c (wt-%)
T ( C)
t (h)
A/R (wt/wt)%
A
0.07
0.39
0.66
0.12
0.23
0.08
0.03
1.59
4
4
4
4
4
4
4
28
0.02
0.10
0.17
0.03
0.06
0.02
0.02
–
1.00
5.67
9.58
1.70
3.36
1.19
–
0.04
0.36
0.63
0.09
0.20
0.05
0.18
1.59
2.6
◦
22.7
40.9
5.8
13.8
3.5
*
p (MPa)
Error
Total
11.0
100.0
–
ADA
ADN c (wt-%)
T ( C)
t (h)
A/R (wt/wt)%
A
*
207.6
254.8
127.1
27.2
154.8
167.6
15.2
4
4
4
4
4
4
4
51.9
63.7
31.8
6.8
38.7
41.9
3.8
7.64
9.37
192.4
239.6
111.9
12.0
21.2
26.6
11.7
1.3
◦
*
4.68
1.00
5.69
6.16
–
139.6
152.4
91.2
14.6
16.0
9.6
p (MPa)
Error
Total
954.4
28
–
–
954.4
100.0
a
Sum of square
b
*
**
F
= 6.39 ( p < 0.05), 4.11 ( p < 0.1)
critical
c
Sum of square-sum of square of error
d
ꢀ
Percentage contribution = (SS /SS
) × 100
total
**
*
,
Significant factor

J Solution Chem (2009) 38: 241–258
251
critical F value for each process parameter indicates which parameter has the most signif-
icant influence on the yield of different hydrolysis products. For example, if one of the F
values is larger than those of any others, it has the greatest effect on the yield of the given
product due to the change of the process parameters. It can be seen that (ADN c) and T have
significant influences on the yields of adipamide, adipamic acid and adipic acid at p < 0.05,
where each p value represents twice the probability that the true value has any value with
a sign opposite to that of the observed value; t was the statistically significant factor on the
yield of 5-cyanovalermic acid at p < 0.05;(ADN c) was the significant factor on the yield
of 5-cyanovaleramide at p < 0.1. The F value for all factors is bigger than the extracted
F value from the table. This means that the variance of all factors is significant compared
with the variance of error, which indicates the variance of error has no meaningful effect
on the response. These analysis of variance results are further confirmed by the analysis of
percentage (P C%) contribution contained in Table 5. The P C% is obtained by the equation
as follows:
P C% = SSꢀ/SStotal × 100%;
Mean square = SS/degree of freedom
where SSꢀ signifies the sum of squares and is calculated by (SS − SSerror), SS is the indi-
vidual sum of squares, SSerror is the sum of squares of error, and SStotal is the total sum of
squares of all factors. Obviously, the value of SStꢀotal is equal to SStotal in statistics, whereas
SSeꢀrror is obtained from the difference of SStotal and the sum of SSꢀ of the considered fac-
tors. The results imply that (ADN c) is the most important factor contributing to the yield
of 5-cyanovaleramide (27.0%) and adipamic acid (49.4%); T is the most significant factor
contributing to the yield of adipamide (46.7%) and adipic acid (26.6%); the t percentage
contribution is the biggest to the yield of 5-cyanovaleric acid (40.9%).
Apparently the data analysis of P C% is in good agreement with the conclusion obtained
from the significance analysis discussed above. Three factors, including (ADN c), T and t
had remarkable influences on the hydrolysis reaction products distribution. Control of the
hydrolysis product distribution can be realized based on the significance of each factor on
different product yields.
3.4 Direct Observation Analysis of the OAD
The impact of parameters on the yield of each product can also be statistically treated with
direct observation analysis to evaluate the effect or the importance of a given factor. In the
direct observation analysis of the OAD, the average yield (k) of each product at each level
of a factor is obtained from the data listed in Table 3, and the results are listed in Table 6.
The mean value of each product yield for the corresponding factors and at each level was
calculated according to the assignment of the experiment. For example, the performance
statistics of the first value of k1 (9.19) for the 5-CVAM is thus the average of those obtained
from the experiment with experiment numbers of 1, 2, 3, 4 and 5. Experimental conditions
for the first value of k2 (0.94) for 5-CVAM are the conditions of the experiment with numbers
of 6, 7, 8, 9 and 10, and so on. The mean values of the five levels of each factor (e.g.,
temperature) reveal how the product yields will change when changing the levels of that
factor. In all cases, it should be noted that each reported product yield was the average
of five measurements, for each of which the parameter of interest was keep constant, and
the other parameters were changed (Table 2). It is impossible that the effects of the other
variables, which are not constant in the data used to calculate an average k, are likely to
convolute the interpretation of the effect of the examined variables. The optimal conditions

252
J Solution Chem (2009) 38: 241–258
Table 6 The average products yield (k) at each level of a factor and ranges
ADN c
T
t
A/R
Additive
p
◦
(wt/wt)%
( C)
(h)
(wt/wt)%
(MPa)
5-CVAM
k
k
k
k
k
9.19
0.94
3.89
2.69
3.05
8.26
8.11
5.24
1.13
4.11
0.97
7.15
6.11
3.53
4.03
4.34
1.75
4.37
4.29
4.42
1.21
5.35
4.48
4.14
3.29
6.72
2.36
4.94
2.45
4.36
3.74
7.03
3.11
5.55
1.12
5.91
1
2
3
4
5
Range
ADAM
k
k
k
k
k
1.42
1.74
2.97
6.48
7.89
6.48
3.29
5.78
5.90
4.60
0.90
5.06
5.51
4.59
2.62
3.49
4.27
2.89
4.27
3.72
2.68
4.64
4.72
2.05
3.79
3.33
4.31
3.98
5.07
1.74
2.96
3.73
4.19
5.30
4.22
2.34
1
2
3
4
5
Range
ADAA
k
k
k
k
k
9.65
14.09
24.97
25.83
20.74
5.00
20.38
15.68
13.87
19.54
22.87
7.31
17.71
16.89
14.95
22.87
18.12
7.92
15.88
18.47
20.81
13.43
22.06
8.63
14.60
19.53
9.09
1
2
3
4
5
15.69
16.58
28.77
19.94
19.12
20.71
16.79
6.11
Range
20.84
5-CVAC
k
k
k
k
k
0.06
0.02
1.28
0.14
1.00
1.26
0.14
0.65
0.39
0.04
1.28
1.24
0.02
0.24
0.22
1.49
0.53
1.47
0.77
0.41
0.84
0.34
0.14
0.70
0.30
0.88
0.73
0.53
0.06
0.82
1.21
0.15
0.33
0.04
0.73
1.17
1
2
3
4
5
Range
ADA
k
k
k
k
k
1.30
6.34
10.25
5.06
4.16
8.85
4.42
4.09
3.01
3.60
11.88
8.87
6.15
4.05
9.17
5.28
2.38
6.79
6.07
7.05
3.86
4.64
5.40
3.20
6.46
5.99
9.15
3.87
1.56
7.59
2.62
2.58
6.86
9.63
4.48
7.06
1
2
3
4
5
Range
for the yield of each product are presented in Table 7. The determination of the optimal
conditions for each product is based on the maximum k value of that particular parameter
given in Table 6. According to these findings and the significant effects by the OAD method,
results for the combination of levels could be predicted, and the predications are confirmed

J Solution Chem (2009) 38: 241–258
253
Table 7 The optimal conditions for the yields of each product
◦
Optimal parameters
ADN c wt-%
T ( C)
t (h)
A/R (wt/wt)
Additive
ADA
p (MPa)
5-CVAM
ADAM
ADAA
5-CVAC
ADA
1%
8%
6%
4%
4%
250
290
290
330
250
2
2
6
5
4
8%
8%
8%
6%
4%
25
25
25
10
25
Na SO
2
4
4
Na SO
2
ADA
PTA
Table 8 Experimental and optimized yield of each product
Yield wt-%
ADAM
ADAA
5-CVAM
5-CVAA
ADA
a
Experimental
13.00
13.56
42.48
43.05
15.84
15.90
0.80
0.80
22.07
22.49
b
Optimized
a
The highest yield in the experiment
b
Optimized conditions by the Taguchi method
by some experiments. The results from the confirmatory experiments showed that under
the optimal conditions, obtained from the ANOVA analysis, the yield of each product is
no lower than that of the maximum point in each graph, which are presented in Table 8.
Optimal combination of the process parameters can be predicted using ANOVA analysis
and the performance characteristics.
From the direct observation analysis of the results in Table 6, some data are scattered
or deviate unexpectedly, but it allowed us to identify conditions that optimize each of the
products. This work and approach would be useful to other high-temperature water systems
for optimizing product yields. The significance of each parameter on the hydrolysis products
is primarily based on the ANOVA, direct observation analysis is just an additional way to
validate the ANOVA results.
3.5 Effect of ADN Concentration on Product Yield
The effect of reactant concentration on the yield of each product was investigated with vary-
ing amounts of ADN in water as shown in Table 6. As can be seen, when the concentration of
ADN increases, the yield of 5-CVAM decreases, and the yield of ADAM, ADAA and ADA
increase up to a maximum followed by a decrease. There was a strong effect of the reactant
concentration on the yields of 5-CVAM, ADAM and ADAA. At lower reactant concentra-
tions, the hydrolysis product was primarily 5-CVAM, but at higher reactant concentrations,
ADAM, ADAA and ADA were the major products. There is no obvious effect of the change
of the reactant concentration on the yield of 5-CVAC. A possible explanation for this result
is that at lower reactant concentration, the formation rate of the 5-CVAM is faster than its
hydrolysis rate, which means that hydrolysis of 5-CVAM is the rate-determining step. With
increasing reactant content in the water, the accumulation of products resulting from the first
hydrolysis step also increases, and the hydrolysis rate of 5-CVAM increases, resulting in a
higher yield of the secondary products, suggesting autocatalytic kinetics for this reaction
system.

254
J Solution Chem (2009) 38: 241–258
3.6 Effect of Temperature on Product Yield
Variation of temperature affects the properties of the water and the interaction among addi-
tive, reactant and water molecules as well as the distribution of the products, accordingly.
Lower temperature is advantageous for the formation of 5-CVAM. At a temperature of about
275 ◦C, high yields of ADAM and ADAA were obtained. If the temperature increased grad-
ually, ADAM and ADAA would convert to ADA. At higher temperature, ADA was the main
product and also the final reaction product, consistent with the results of the ANOVA analy-
sis. Generally, higher temperature favors the yield of adipic acid. Increasing the temperature
leads to contrary effects for the production of adipic acid. The ion product decreases with in-
creasing temperature at ranges from 250 to 330 ◦C, resulting in a lower concentration of H+,
but higher pH is disadvantageous for the yield of ADA as discussed in the section “Effect
of the Kind of Additives on Product Yield”. On the other hand, increasing temperature can
accelerate the reaction rate, since the hydrolysis reaction is an endothermal reaction and the
equilibrium will shift in a positive direction following Le Chatelier‘s Principle, resulting in
the increase of ADA yield. The fact that the yield shows a somewhat decreasing trend to
290 ◦C under these experimental conditions is a negative effect because of the decrease of
the ion product with increasing temperature exceeds the enhancement effect. The combined
result is a decrease in the yield of ADA, because of increasing temperature in the range
250–290 ◦C. At higher temperatures ranging from 290 to 330 ◦C, the negative effect on the
ion product is less than the enhancement effect of increasing the temperature; the combined
result is an increase of ADA yield.
3.7 Effect of Time on Product Yield
The effect of time on the yield of different products was realized by conducting the hydrol-
ysis of ADN at different reaction time intervals. The variation range of k of each product
with the change of time is listed in Table 6. There was no remarkable influence of time on
the yield of 5-CVAM, ADAM, ADAA and ADA except for 5-CVAC. More specifically, the
yield of 5-CVAM on the whole was decreased, suggesting that 5-CVAM did not exist for
long in subcritical water. There was a similar variation trend of the yield of ADAM and
ADAA, both in lower yields at the same reaction time of four hours. Although the yield
of 5-CVAC mainly depended on reaction time, the maximum value was only 1.49 wt-%,
so the preparation of 5-CVAC was not accomplished with the adjustment of time. Longer
reaction time was unfavorable for the accumulation of ADA possibly because of its thermal
decomposition.
3.8 Effect of A/R Ratio on Product Yield
As can be seen in Table 6, a change of the additive content in the reaction system had no
obvious influence on the yield of each product, that is to say whether the additives existed or
not, hydrolysis of ADN could take place freely. Experiments on the hydrolysis of ADN in
subcritical water without the addition of any additives in entries 1, 8, 15, 17 and 24 proved
this result.
3.9 Effect of the Kind of Additives on Product Yield
Introduction of the additives into the reaction system could change the pH or the ionic
strength. In the presence of acid or base, the pH in the reaction solution would be changed.

J Solution Chem (2009) 38: 241–258
255
Increase of the concentration of H+ or OH− in the reaction system could accelerate the
hydrolysis of ADN. The investigated results showed that the existence of acidic or basic ad-
ditives could only promote the hydrolysis rate of ADN, resulting in a shorter reaction time
from the beginning of the reaction to the end. Additives have no significant influence on
the yields of all products just as described in Table 6. There is a little increase of the yield
of ADAM with the addition of Na2SO4, possibly because water molecules solvating the
ionic species of the dissolved salt and the local density of the surrounding water is expected
to increase. Increase of the water density around the salt species results in an increase of
the dielectric constant of this region, which causes an increase of the hydrolysis reaction
rate of ADN. PTA was the most effective additive for the production of ADA, whereas ba-
sic or neutral salts were the least effective additive. ADA was the optimal additive for the
preparation of 5-CVAC, suggesting that low pH was suitable for the presence of 5-CVAC in
high-temperature conditions. There is no trend to follow for effects of the additives on the
yield of 5-CVAM.
3.10 Effect of Pressure on Product Yield
Generally, the variation of pressure on the reaction system leads to an increase of the wa-
ter density, directly resulting in an increase of the H+ concentration; on the other hand,
increasing the water density can promote the dissociation of PTA, also increasing the H+
concentration. Pressure plays an important role in the production of ADA, but not as well
as for the other products presented in Table 6. An increase in pressure causes an increase
in the fluid density and thus it has a double effect: an increase of the concentration of H+
results in an increase in the yield of ADA; higher pressure is unfavorable for the hydrolysis
reaction according to Le Chatelier’s Principle. If the enhancement effect of increasing the
H+ concentration due to increasing pressure is less than the negative effect of reacting in
such a way as to produce fewer molecules due to the increase of pressure, the position of
equilibrium will move in the direction so that the pressure is reduced again. The combined
result is the decrease of ADA yield.
3.11 By-products in the Reaction at above 330 ◦C
Formation of by-products was suggested under the conditions of longer reaction time and
higher temperature as shown in the carbon balance analysis and the discussion section of
effect of temperature on product yield. Under the conditions of longer reaction time and
higher temperature, the final hydrolysis reaction product was primarily ADA which is a di-
carboxylic acid. It can be produced by the consecutive oxidation to lower molecular weight
dicarboxylic acids [29]. The oxidation of dicarboxylic acids was also accompanied by ox-
idative decarboxylation, leading to the formation of monocarboxylic acids. Total yields of
five products reached 11% as shown in Table 3 for run No. 25. The low yield value suggested
the presence of other by-products, and the carbon balance analysis at different temperatures
also gave a proof of this result.
4 Analysis of the Reaction Mechanism
In the first hydrolysis step of ADN, there is no competitive reaction pathway, and it only
involves the transformation of the cyano group to the amide group. If the two cyano groups
in adiponitrile were attacked simultaneously, no 5-cyanovaleramide would be formed. Actu-
ally, the higher content of 5-cyanovaleramide was detected at lower temperatures and shorter

256
J Solution Chem (2009) 38: 241–258
Fig. 6 Global reaction scheme for the hydrolysis of ADN in NCW
reaction times, suggesting that the two cyano groups transform to amides in sequence. There
could be two possible reaction pathways for the hydrolysis of 5-cyanovaleramide: first if
the cyano group or the amide group was attacked initially and second if the cyano group
was attacked first, adipamide was primarily formed and detected with higher content in the
products; on the contrary, if the amide group was firstly attacked, 5-cyanovaleric acid was
primarily formed and detected with higher content in the products. The experimental re-
sult was that the content of adipamide was much higher than that of 5-cyanovaleric acid in
the products, suggesting hydrolysis of 5-cyanovaleramide proceeded basically by the trans-
formation of the cyano group to the amide group. Hydrolysis of adipamide is similar to
that of adiponitrile in that only one of the amides was attacked to form the adipamic acid
which could further be hydrolyzed to form adipic acid. Based on the hydrolysis steps men-
tioned above, a hydrolysis mechanism of adiponitrile can be proposed as described in Fig. 6.
The concentration of ADN is the significant factor on the yield of 5-cyanovalericamide, the
lower content of ADN in the solution, and the higher yield of 5-CVAM. There are two steps
to follow for 5-cyanovalericamide, step2 and step3. The concentration of ADN and temper-
ature are the significant factors on the yield of adipamide. Higher concentrations of ADN
and moderate temperature are the preferable conditions for a higher yield of adipamide.
Reaction time is a considerable factor for the formation of 5-cyanovaleric acid. Under the
conditions of moderate concentration and temperature, 5-cyanovaleric acid and adipamide
can convert to adipamic acid, which is more stable than 5-cyanovaleric acid in the investi-
gated conditions. The formation rate of 5-cyanovaleric acid is slower than of its hydrolysis
rate, resulting in lower concentrations of 5-cyanovaleric acid in the final products. As sug-
gested by the results in Table 6, higher temperature is advantageous for the reaction step6,
indicating the relative stability of adipic acid at higher temperature.
5 Conclusions
In this study, an orthogonal array design (OAD) was employed to optimize the experimental
conditions of adiponitrile hydrolysis in subcritical water. Influences of six factors on the
yields of different products were considered. Results of the OAD experiments were treated
statistically by ANOVA and direct observation analysis. The results showed that (ADN c)
and T had significant influence on the yields of adipamide, adipamic acid and adipic acid

J Solution Chem (2009) 38: 241–258
257
at p < 0.05; T was the statistically significant factor for the yield of 5-cyanovalermic acid
at p < 0.05, and (ADN c) was the significant factor for the yield of 5-cyanovaleramide at
p < 0.1. The carbon balance was calculated. Based on the results for each product detected
and its yield under different conditions, a possible reaction mechanism was proposed.
Acknowledgements The authors gratefully acknowledge the financial support the Ministry of Science
and Technology of China (2007AA05Z101). Sincere thanks to Professor Phillip E. Savage (University of
Michigan) who gave us valuable suggestions on the paper writing.
References
1. Izzo, B., Harrell, C.L., Klein, M.T.: Nitrile reaction in high-temperature water: Kinetics and mechanism.
AIChE J. 43, 2048–2058 (1997)
2. Sorokin, D.Y., Pelt, S., Tourova, T.P., Takaichi, S., Muyzer, G.: Acetonitrile degradation under haloalka-
line conditions by Natronocella acetinitrilica gen, nov., sp. nov. Microbiol. 153, 1157–1164 (2007)
3. Krieble, V.K., Noll, C.I.: The hydrolysis of nitriles with acids. J. Am. Chem. Soc. 61, 560–563 (1939)
4. Hall, J.H., Gisler, M.: A simple method for converting nitriles to amides. Hydrolysis with potassium
hydroxide in tert-butyl alcohol. J. Org. Chem. 41, 3769–3770 (1976)
5. Neugebauer, W., Schmidt, L.: Process for hydrolysis of nitriles with recovery of nitrogen values derived
from the nitrile group. US Patent 492,345 (1970)
6. Peterson, R.E., Cannan, T.M., Lisa, R.E.: Method of producing carboxylic acid salts. US Patent 410,783
(2002)
7. Prihod’ko, R., Sychev, M., Kolomitsyn, I., Stobbelaar, P.J., Hensen, E.J.M., van Santen, R.A.: Layered
double hydroxides as catalysts for aromatic nitrile hydrolysis. Microporous Mesoporous Mater. 56, 241–
255 (2002)
8. Norton, R.V.: Process for aromatic carboxylic acids. US Patent 3,873,610 (1975)
9. Akiya, N., Savage, P.E.: The roles of water for chemical reactions in high-temperature water. Chem. Rev.
102, 2725–2750 (2002)
10. Kruse, A., Dinjus, E.: Hot compressed water as reaction medium and reactant: Properties and synthesis
reactions. J. Supercrit. Fluids 39, 362–380 (2007)
11. Patrick, H.R., Glaser, R., Griffith, K.: Near-critical water: A benign medium for catalytic reactions. Ind.
Eng. Chem. Res. 40, 6063–6067 (2001)
12. Chang, Y.L., Liu, B., Shen, B.: Orthogonal array design for the optimization of supercritical fluid extrac-
tion of baicalin from roots of Scutellaria baicalensis Georgi. J. Sep. Sci. 30, 1568–1574 (2007)
13. Harrell, C.L., Moscariello, J.S., Klein, M.T.: The absence of wall effects during benzonitrile hydrolysis.
J. Supercrit. Fluids 14, 219–224 (1999)
14. Krämer, A., Mittelstädt, S., Vogel, H.: Hydrolysis of nitriles in supercritical water. Chem. Eng. Technol.
22, 494–500 (1999)
15. Tewari, Y.B., Goldberg, R.N.: Thermodynamics of the hydrolysis reactions of nitriles. J. Chem. Ther-
modyn. 37, 720–728 (2005)
16. Duan, P.G., Wang, X., Dai, L.Y.: Noncatalytic hydrolysis of iminodiacetonitrile in near-critical water—
a green process for the manufacture of iminodiacetic acid. Chem. Eng. Technol. 30, 265–269 (2007)
17. Duan, P.G., Yang, Y., Dai, L.Y.: Preparation of iminodiacetic acid salts from iminodiacetonitrile in sub-
critical water. J. Chem. Chem. Eng. 1, 21–26 (2007)
18. Moreau, J.L., Azza, S., Bigey, F.: Application of high-performance liquid chromatography to the study
of the biological transformation of adiponitrile. Analyst 116, 1381–1383 (1994)
19. Hayes, R.A., Marks, D.N.: Process for the hydrolysis of adiponitrile and the production of nylon. US
Patent 6,075,117 (2000)
20. Tauguchi, G.: System of Experimental Design, vols. 1–2. Krus International, New York (1990)
21. Zhu, G.H., Ju, H.X.: Determination of naproxen with solid substrate room temperature phosphorimetry
based on an orthogonal array design. Anal. Chim. Acta 506, 177–181 (2004)
22. Roy, R.K.: A Primer on Taguchi Method. Van Nostrand Reinhold, New York (1990)
23. Safarzadeh, M.S., Moradkhani, D., Ilkhchi, M.O.: Determination of the optimum conditions for the
leaching of Cd–Ni residues from electrolytic zinc plant using statistical design of experiments. Sep.
Purif. Technol. 58, 367–376 (2008)
24. Díez, C., Barrado, E., Marinero, P.: Orthogonal array optimization of a multiresidue method for cereal
herbicides in soils. J. Chromatogr. A 1180, 10–23 (2008)

258
J Solution Chem (2009) 38: 241–258
25. Chamsaz, M., Safavi, A., Fadaee, J.: Simultaneous kinetic-spectrophotometric determination of car-
bidopa, levodopa and methyldopa in the presence of citrate with the aid of multivariate calibration and
artificial neural networks. J. Anal. Chim. Acta 603, 140–146 (2007)
26. Wiley, R.H., Morgan, H.S.: Partial hydrolysis of the adiponitrile and sebacontriule by hydrogen peroxide.
J. Org. Chem. 15, 800–801 (1950)
27. Zil’berman, E.N.: Products of reaction of adipic acid with ammonia. Zh. Obshchei Khim. 25, 2127–2132
(1955)
28. Cengal, Y.A., Boles, M.A.: Thermodynamics. McGraw-Hill, New York, pp. 905
29. Jin, F.M., Cao, J.X., Kishita, A., Enomoto, H., Moriya, T.: Oxidation reaction of high molecular weight
dicarboxylic acids in sub- and supercritical water. J. Supercrit. Fluids 44, 331–340 (2008)