Chemical Engineering and Processing 43 (2004) 803810

Kinetics of hexamine crystallization in industrial scale

A. Alamdari , F. Tabkhi
Department of Chemical Engineering, School of Engineering, Shiraz University, 71345 Shiraz, Iran
Received 9 December 2002; received in revised form 13 January 2003; accepted 18 April 2003

Abstract
The values of kinetic parameters for hexamine crystallization from aqueous solution, necessary for control of industrial crystallizers were
calculated in this study. Hexamine is an intermediate chemical mainly used in pharmaceutical and resin industries. In Meissner process, the
major industrial method of hexamine production, formaldehyde and ammonia in gaseous states are introduced into the reaction vessel, and
crystalline hexamine is continuously produced. The reaction between formaldehyde and ammonia is very fast and thus the whole process is
controlled by the crystallization stage. However, the kinetic information reported in the literature for industrial crystallization of hexamine is
very limited. Mechanistic modeling of hexamine crystallization was performed using mass and population balance equations. These equations
were solved using the backward difference method. Kinetic parameters in the growth and nucleation rate equations were determined by fitting
the mechanistic model to the industrial data obtained in the present study from an industrial MSMPR crystallizer. To determine the population
densities and supersaturations, both the solid and liquor phases were sampled during the transient step of particle size evolution. To optimize
the parameters, the differences between the model-predicted data and the plant data were minimized using the nonlinear unconstrained method
in an optimization subroutine. The growth and nucleation rate equations for hexamine crystallization in industrial scale were calculated as
G = 0.7S1.47 and B = 9138MT0.74 S0.551 , respectively. The growth and nucleation rate equations obtained in this study can be used to simulate
the crystallization stage of hexamine plants.
2003 Elsevier B.V. All rights reserved.
Keywords: Hexamine; Crystallization; Kinetics

1. Introduction

producing soluble hexamine followed by hexamine crystallization is presented by the overall equation of

Hexamine (CH2 )6 N4 , or hexamethylenetetramine is used

in pharmaceutical industries as a primary feed material, and
in chemical industries as an intermediate material. It is well
soluble in water and, like some other tertiary amines, has an
inverse solubility at low temperatures [1].
The most common production method of hexamine in industrial scale is the reaction of formaldehyde with ammonia
or ammonium salts [2]. This reaction in aqueous solution
at room temperature is very fast and produces considerable
amount of heat and water. The temperature at which the
reaction can take place with a high efficiency has been reported as between 0 and 90 C [1,3]. Side reactions due to
temperature rise may be prevented by cooling. The reaction

6CH2 O(g) + 4NH3(g) C6 H12 N4(s) + 6H2 O(l) + q

Corresponding author. Tel.: +98-911-718-4316;

fax: +98-711-628-7294.
E-mail address: alamdari44@yahoo.com (A. Alamdari).

0255-2701/$ see front matter 2003 Elsevier B.V. All rights reserved.
doi:10.1016/S0255-2701(03)00085-0

(1)

where q is 744.8 kJ mol1 of hexamine [3].

An optimum method to produce hexamine from formaldehyde and gaseous ammonia is the Meissner process in which
the reaction and crystallization stages take place simultaneously to produce crystalline hexamine. Due to a high
dissolution rate of ammonia and formaldehyde in the aqueous solution and a high reaction rate of hexamine production, the crystallization stage controls the Meissner process
[3].
The overall growth rates of spheres of compressed polycrystalline hexamine with diameters of 10 and 16 mm
in aqueous solutions were studied by Bomio et al. at a
temperature of 30 C and in a supersaturation range of
0.00060.002 kg kg1
solution [4]. They observed a linear dependency of growth on supersaturation for a relative solution/sphere velocity under 84 mm s1 and a slightly higher

804

A. Alamdari, F. Tabkhi / Chemical Engineering and Processing 43 (2004) 803810

order dependency as the velocity exceeded 84 mm s1 .

Therefore, they concluded that at lower velocities the diffusion is controlling and at higher velocities both the diffusion
and surface reaction are controlling the growth process.
For different solution/sphere velocities, they calculated the
growth rate coefficient as in the range 0.00250.0493 kg
m2 s1 and the growth order in the range 0.951.16. They
also calculated the activation energy for hexamine growth
as 50.2 kJ mol1 for the temperature range 24.535 C.
The growth kinetics of hexamine crystallizing from pure
ethanol at 25 C were studied by Bourne and Davey [5]. They
used a fluidized bed, a rotating disk, and a single crystal system and found an independent growth rate of crystal size.
They reported that the growth rates attainable from ethanolic solutions are much higher than those from the aqueous
solutions at the same absolute supersaturations. They also
noted that measurements made at 25 C on the two different
experimental systems of their study suggested no influence
of hydrodynamics on the growth rates. However, secondary
nucleation which substantially affects the CSD of product
strongly depends on hydrodynamics and hence on the scale
of equipments.
In another study Bourne et al. [6] using a rotating disc
system correlated the surface integration kinetic data of hexamine crystallizing from aqueous solutions at 30 C in the
absence of diffusional effects as:
MA = 0.383(wi ws )

1.41

(2)

where MA is the mass transfer rate of hexamine (kg m2

s1 ), wi is the interfacial solution concentration, and ws
is the saturation concentration (kghexamine kg1
solution ). They
also calculated the activation energies for the growth of
hexamine as in the range 28.960.7 kJ mol1 depending on
the rotational rate of discs. They reported that the values of
activation energy were influenced by mass transfer.
The growth and nucleation rates of hexamine crystallizing from methanol and from an aqueous solution containing
76% 2-propanol in a bench scale MSMPR crystallizer at
steady state were studied by Myerson et al. [7]. Rohani and
Bourne used experimental data of hexamine crystallizing
from alcoholic solutions in the literature to develop a general model for a seeded cooling batch crystallizer based on
population and mass balances [8].
However, the values of kinetic parameters used in the rate
equations of growth and nucleation for hexamine crystallizing from water in industrial scale, have not been reported.
These parameters are necessary for simulation of crystallization process and control of crystal size distribution (CSD)
of product.

Fig. 1. Flow diagram of the industrial process of hexamine production.

Table 1
The composition of the stream containing formaldehyde
Component
Mass fraction

N2
0.3985

H2 O
0.2816

CH2 O
0.2589

CH3 OH
0.021

CO2
0.0249

2. Industrial crystallizer of hexamine

A stream containing formaldehyde, water vapor, and inert
gases with a flow rate of 0.344 kg s1 is injected to a 15.5
m3 crystallizer at 100120 C (Fig. 1). The composition of
the stream is reported in Table 1. In another stream ammonia
gas with a purity of 99.8 mass percent in stoichiometric
proportion to formaldehyde, enters the crystallizer where
these reactants dissolve in aqueous solution and react with
each other to produce hexamine. The operating conditions
are listed in Table 2.
A double propeller mixer suspends the crystals in the solution. The high flow rate of gases entering at mid-height
of the body of the reaction vessel also enhances mixing.
Therefore, the vessel behaves as an MSMPR crystallizer. A
side stream continuously leaves the crystallizer and enters
an evaporator under a pressure lower than that of the crystallizer. The stream, after being concentrated sufficiently in
the evaporator, returns to the crystallizer to complete the cycle. This concentration is necessary to vaporize the water
produced by the reaction, and to create supersaturation as a
driving force for nucleation and growth of hexamine crystals
in the solution. The product stream from the crystallizer is
sent to a centrifuge filter by a positive displacement pump.
The filtrate, containing very fine particles passed through
the meshes of the centrifuge basket, is sent to a dissolution

Table 2
The operating conditions of the hexamine plant
Crystallizer pressure (bar)

Evaporator pressure (bar)

Crystallizer temperature ( C)

Residence time (h)

0.25

0.59

6070

12.88

A. Alamdari, F. Tabkhi / Chemical Engineering and Processing 43 (2004) 803810

tank, in which fines are dissolved by dilution and heating.

These particles are expected to be produced through the
mechanism of contact nucleation. The clear solution from
the dissolution tank is sent to a head tank, and thereafter is
returned to the crystallizer. The dissolution of fines prevents
their return to the crystallizer and maintains the MSMPR
conditions. The solution level in the crystallizer is controlled
by a simple hydrodynamic controller installed between the
head tank and the crystallizer. However, the variation of the
suspension volume is generally within the range of 5.5%.

Dynamic information necessary for calculating the kinetic

parameters of crystallization, was obtained during the course
of start up of the industrial hexamine crystallizer. As the
reaction progressed and evaporation continued, spontaneous
nucleation created new crystals which they began to grow.
Thus the size distribution of crystals was evolving before
the steady state conditions were reached.
Consecutive sampling from the suspension was carried out
and crystals were separated from the liquor in each sample
using a Buchner funnel under vacuum conditions. Crystals
were well washed by hexamine saturated methanol and were
dried in a laboratory oven. The CSD was measured using
sieve analysis and the hexamine concentration in the filtrate
from each sample was determined by gravimetric method,
after the complete water evaporation from the sample. Since
there was only one solute component in the solution, the
gravimetric method seemed to be accurate enough to measure the hexamine concentrations. This method has been reported as an accurate method for measuring concentrations
of KCl in controlled cooling crystallization operations [9].
To obtain an insight into the morphology of hexamine
crystals, scanning electron microscopy (SEM) photomicrographs were prepared using a Cambridge 360 electronic
microscope.

4. Mathematical modeling
The form of the growth rate equation was assumed as:
G = KG S g

(3)

where KG is the growth rate coefficient and S is the supersaturation. S was calculated as the difference of the liquor concentration and the saturation concentration. The secondary
nucleation rate equation was assumed as:
B = KN MTi S j

tallizer with constant volume of solution in the absence of

crystal agglomeration and breakage is [10]:
n
n
n
+G
+Q
= 0
t
L
V

(4)

where KN is the nucleation rate coefficient and MT is the

suspension density of crystals in the solution. Assuming
a size independent growth rate of crystals, the transient
population balance equation in a continuous MSMPR crys-

(5)

where Q is the volumetric flow rate of suspension leaving the crystallizer as product and V is the volume of the
suspension in the crystallizer. Q and V remained almost
constant with time at the conditions of the industrial crystallizer under study. An initial condition for the population
balance equation is:
n(0, L) = f(L)

3. Industrial data

805

(6)

where f(L) is the size distribution of crystals initially created in the solution by spontaneous nucleation due to a
high supersaturation. f(L) was measured by sieve analysis
of crystals in a sample taken from the crystallizer contents
at time zero, after spontaneous nucleation had occurred.
During the transient period of start up, and thereafter at
steady state period, secondary nucleation creates nuclei in
the solution. The nuclei density represented by the following
equation is another boundary condition of the population
balance equation:
n(t, 0) = n = B /G

(7)

The variations in the solution temperature were recorded

over time during the course of start up. The range of variations was from 60 to 67 C. The temperature values necessary for modeling were calculated by linear interpolation
between the recorded values.
The mass balance equation for hexamine component was
written as:



d
w + MT
w + MT
V
= m
r +m
hQ
1/s + MT /c
dt 1/s + MT /c
(8)
where m
r is the rate of hexamine production by reaction
and m
h is the mass flow rate of hexamine returned to the
crystallizer from the head tank. The last term on the right
hand side of Eq. (8) is the mass flow rate of hexamine in
both solid and liquid phases taken out from the crystallizer.
c and s are the crystal and the mother liquor densities,
respectively, and w is the mass fraction of hexamine in the
clear solution. The initial condition of Eq. (8) is the mass
fraction of hexamine in the initial solution at time zero, w0 .
The mass flow rate of hexamine produced by reaction was
calculated from the mass flow rate of formaldehyde entering the crystallizer. The hexamine mass flow rate entering
the crystallizer from the head tank was calculated by balancing the hexamine mass around the centrifuge, the fines
dissolution tank, and the head tank.
The simultaneous solution of the mass and population balances predicts the CSD of product at different times. Kinetic
parameters of the crystallization process were determined
by fitting the model generated set of CSD to the measured

806

A. Alamdari, F. Tabkhi / Chemical Engineering and Processing 43 (2004) 803810

set of CSD. The objective function, defined as the following

equation, calculates the difference between these two sets of
distributions:
fobj =

11
5 


(P exp (tk , Lj ) P mod (tk , Lj ))2

(9)

k =1j =1

where P is the mass of particles smaller than size L in unit

mass of solid free solution and was calculated as:
 L
c kv L3 ndL
(10)
P = 1018
0

The conversion factor of 1018 has been introduced to adjust the units of crystal size in the equation. Optimum kinetic parameters were obtained by minimizing the objective
function.

5. Solution of model
Initially, an arbitrary value for each kinetic parameter was
assumed. The solution of model equations yielded the population densities of crystals at time increments from the beginning. Using the model predicted population densities of
crystals at different times, the values of objective function
corresponding to assumed parameters were calculated. The
values of parameters for least value of objective function
were determined using unconstrained nonlinear minimization.
Since the model equations were nonlinear, they were
solved numerically. The backward finite difference method
was used to prevent divergence in calculations. The programs were written using the package of matlab and the
parameter optimization was carried out using fminsearch
routine in this package.

6. Results and discussion

The optimized values for kinetic parameters of hexamine crystallization under the industrial conditions specified
are given in Table 3. It is pertinent to note that the hexamine crystals suspended in the solution were assumed not to
stick together considerably and thus, the agglomeration was
considered negligible. This assumption was checked by visualization of crystals under the microscope. No significant
agglomeration was observed, but crystals appeared mostly
individual and unattached. This observation also showed no
significant crystal breakage. Fig. 2 shows representative phoTable 3
The optimized values of the kinetic parameters of hexamine crystallization
KG

KN

0.7

1.47

9138

0.74

0.551

tomicrographs of hexamine crystals sampled from the industrial crystallizer under this study.
The initial CSD measured experimentally was fitted with
a smooth exponential function, f(L), in order to minimize
errors due to fluctuations in the input data to the model:
f(L) = 3.25 105 (1 e0.022L )

(11)

The size evolution in the industrial crystallizer, which was

followed by measuring CSD at five different times during
the transient period of operation, is shown in Fig. 3. The figure also shows the set of CSD predicted by the model using
the optimized values of the kinetic parameters. As the crystallization process continues, after about 20 h, the size distribution of particles in the crystallizer is expected to approach
the steady CSD suggested by the model shown in Fig. 4.
Hexamine concentrations in the solution predicted by the
model, and those measured experimentally in the industrial
unit are shown in Fig. 5. Both the model predicted and the
plant data showed a peak in concentrations of hexamine in
the solution during the transient period. The reaction of ammonia with formaldehyde and the accompanied evaporation
taking place in the solution both continuously increase the
concentration of hexamine. On the other hand, mass deposition of hexamine on the surface area of suspended crystals
diminishes the concentration. At the beginning of the process, there was not enough surface area available to consume
hexamine generated in the solution; therefore, the concentration increased greatly. This led to a higher supersaturation, which accordingly increased the growth and nucleation
rates. Higher rates of these mechanisms resulted in more
crystal surface area available for mass deposition of hexamine (Figs. 6 and 7). Therefore, as enough surface area
was generated, the mass deposition prevailed the concentration increase by reaction and evaporation, and the concentration started to decrease. As the steady state condition was
reached, an equilibrium state was established between the
concentration rise of hexamine by reaction and evaporation
and the concentration fall by mass deposition.
Comparison of the model predicted data with the plant
data in both Figs. 5 and 7 shows the consistency of the results and the accuracy of the calculated values of the kinetic
parameters. It is worth noting that the industrial data of hexamine concentrations in the clear solution and suspension
densities of crystals shown in these figures were not directly
used in the definition of the objective function for minimization in Eq. (9). Therefore, a good agreement of the model
predicted data with the plant data presents the validity of the
mechanistic model.
Since the temperature in the present industrial operation
was almost constant around 65 C and the hexamine solubility in water at the conditions of the operation is approximately independent of temperature, a peak in supersaturation values similar to that of concentration values with time
was expected. The calculated values of supersaturation are
shown in Fig. 8. The same behavior was expected for the
growth rate; however, calculations showed that the nucle-

A. Alamdari, F. Tabkhi / Chemical Engineering and Processing 43 (2004) 803810

Fig. 2. Representative photomicrographs of hexamine crystals sampled from the industrial crystallizer.

807

808

A. Alamdari, F. Tabkhi / Chemical Engineering and Processing 43 (2004) 803810

Fig. 3. Cumulative size distribution at different times during the evolution of particle size, comparison between the model predictions and the plant data.

Fig. 4. Cumulative size distribution predicted by the model using the optimized parameters.

Fig. 5. Hexamine concentrations in the solution with time, comparison between the model predictions and the plant data.

A. Alamdari, F. Tabkhi / Chemical Engineering and Processing 43 (2004) 803810

809

Fig. 9. Variations of growth rate during the course of plant operation,

predicted by model using optimized parameters.
Fig. 6. Predictions of surface area available in the solution.

Fig. 7. Suspension densities in the solution with time, comparison between

the model predicted data and the plant data.

Fig. 10. Variations of nucleation rate with time, predicted by model using
optimized parameters.

ation rate behaved differently (Figs. 9 and 10). Due to increasing supersaturation at the first few hours of operation,
the nucleation rate was expected to increase and approached
its maximum value. Thereafter, no considerable decline in
nucleation rate was calculated. This was due to large mass
of crystals created after about 10 h. The generation of secondary nuclei due to mass of existing crystals, compensated
the reduction of nucleation caused by lower values of supersaturation as the process approached towards the steady
state conditions.

7. Conclusions
The process of hexamine crystallization in an industrial
MSMPR crystallizer was modeled using the population and
mass balance equations. Minimizing the differences between
the model-predicted data and the plant data, the kinetic parameters of growth and nucleation rates were calculated.
These parameters may be used to optimize the quality and
the quantity of hexamine product crystallized from aqueous
solutions in industrial plants.

Appendix A. Nomenclature
AT
B
CSD
f(L)
fobj
Fig. 8. Variations of supersaturation with time, calculated by the model
using the optimized parameters.

g, i, j
G

total surface area of crystals available

in solution (m2 kg1
solution )
nucleation rate (# (kgsolution s)1 )
crystal size distribution
the population density of initial
crystals (# (m kgsolution )1 )
objective function of optimization
2
((kg kg1
solution ) )
orders in kinetic equations (dimensionless)
growth rate of crystals (m s1 )

810

KG
KN
kv
L
MSMPR
MA
MT
m
h
m
r
n
n
P

q
Q
S
SEM
t
V
w
wi

w0
ws

A. Alamdari, F. Tabkhi / Chemical Engineering and Processing 43 (2004) 803810

growth rate coefficient (m s1 kg1.47

hexamine
kg1.47
solution )
nucleation rate coefficient (# s1 kg0.74
solid
0.291
kg0.551
hexamine kgsolution )
volumetric shape factor of crystals (#1 )
crystal size (m)
mixed suspension mixed product
removal
mass deposition rate in surface
integration (kghexamine m2 s1 )
suspension density of crystals in solution
(kgsolid kg1
clearsolution )
mass inflow rate of hexamine from head
tank to the crystallizer (kg s1 )
production rate of hexamine by
reaction (kg s1 )
population density of crystals
(# (m kgsolution )1 )
population density of nuclei
(# (m kgsolution )1 )
mass of crystals smaller than size L
in unit mass of solid free solution
(kg kg1
solution )
heat generation in the reaction of
hexamine production (kJ mol1
hexamine )
volume outflow rate of suspension from
the crystallizer (m3 s1 )
supersaturation (kghexamine kg1
solution )
scanning electron microscopy
time (s)
volume of crystallizer and evaporator (m3 )
hexamine concentration in solution
(kghexamine kg1
solution )
interfacial hexamine concentration between
the solid surface and the solution (kghexamine
kg1
solution )
initial concentration of hexamine in
solution (kghexamine kg1
solution )
the saturation concentration of hexamine
(kghexamine kg1
solution )

Greek letters
c
s

crystal density (kg m3 )

solution density (kg m3 )

Subscripts
g
j
k
l

gas
jth sieve used in size analysis
kth set of data points
liquid

Superscripts
exp
mod

experiment
model

References
[1] E.M. Smolin, L. Rapoport, The Chemistry of Heterocyclic Compounds, S-Triazines and Derivatives, Interscience Publication, New
York, 1959.
[2] J. Kirk, D.F. Othmer, Encyclopedia of Chemical Technology, third
ed, vol. 2, John Wiley & Sons, New York, 1978, p. 511.
[3] F. Meissner, E. Schwiedessen, D.F. Othmer, Continuous production
of hexamethylenetetramine, Ind. Eng. Chem. 46 (1954) 724.
[4] P. Bomio, J.R. Bourne, R.J. Davey, The growth and dissolution of
hexamethylene tetramine in aqueous solution, J. Cryst. Growth 30
(1975) 7785.
[5] J.R. Bourne, R.J. Davey, The growth of hexamethylene tetramine
crystals from ethanolic solutions, J. Cryst. Growth 34 (1976) 230
238.
[6] J.R. Bourne, R.J. Davey, H. Gros, K. Hungerbhler, The rotating
disk configuration in the measurement of crystal growth kinetics
from solution, J. Cryst. Growth 34 (1976) 221229.
[7] A.S. Myerson, S.E. Decker, F. Weiping, Solvent selection and batch
crystallization, Ind. Eng. Chem. Proc. Des. Dev. 25 (1986) 925929.
[8] S. Rohani, J.R. Bourne, A simplified approach to the operation of
batch crystallizer, Can. J. Chem. Eng. 68 (1990) 799806.
[9] H.A. Mohameed, B. Abu-Jdayil, M. Al Khateeb, Effect of cooling
rate on unseeded batch crystallization of KCl, Chem. Eng. Process.
41 (2002) 297.
[10] A.D. Randolph, M.A. Larson, Theory of Particulate Processes, Academic Press, New York, 1988, p. 59.