Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

PTEC-12516; No of Pages 17

Powder Technology xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Powder Technology

journal homepage: www.elsevier.com/locate/powtec

Caking of crystals: Characterization, mechanisms and prevention


Mingyang Chen a, Songgu Wu a, Shijie Xu a, Bo Yu a, Mohannad Shilbayeh d, Ya Liu a, Xiaowen Zhu e,
Jingkang Wang a,b,c, Junbo Gong a,b,c,⁎
a
School of Chemical Engineering and Technology, Tianjin University, Tianjin, China
b
Key Laboratory Modern Drug Delivery and High Efficiency in Tianjin University, Tianjin, China
c
The Co-Innovation Center of Chemistry and Chemical Engineering of Tianjin, Tianjin, China
d
Faculty of Engineering, Western University, London, ON N6G 1G8, Canada
e
Tianjin University, Tianjin, China

a r t i c l e i n f o a b s t r a c t

Article history: Caking of crystals leads to a great deal of negative impact on the quality of products leading to extra cost to restore
Received 21 December 2016 the products to its original state before caking. It is therefore an undesired phenomenon that should be investi-
Received in revised form 15 April 2017 gated and inhibited. This review considers particle issues (particle science and technology) in a crystallization
Accepted 21 April 2017
view (crystallization science and engineering). 1. Characterization. Caking indices and the relevant characteriza-
Available online xxxx
tion methods are classified. Their advantages and disadvantages are summarized. Particularly, potential novel
Keywords:
characterization methods, e.g. image analysis technique and its application in analysis of caking ratio, are
Caking discussed. 2. Mechanisms. Caking process is organized into three stages: moisture sorption, then liquid bridge,
Crystal and finally crystal bridge. The mechanisms are reconsidered with recent studies (e.g. mass transfer, adhesion ef-
Particle fect of liquid bridge, kinetic of crystal caking) and crystallization science (e.g. nucleation and growth theory, poly-
Crystallization morphism, and phase transition). 3. Prevention. Based on particle and crystal engineering, prevention of crystals
Characterization caking is discussed through controlling ambient conditions (relative humidity, temperature, liquids screening,
Mechanisms and pressure) and improving anticaking property of crystals (particles size, particles shape, impurity, and anti-
Prevention
caking agent). A potential method of simulating crystal caking behavior based on discrete element method
(DEM) and caking mechanisms is presented. With the aid of simulation, the enhanced design of particle and crys-
tal engineering would be available, which could be a key to control these factors systematically.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction and elasticity [4–9], which is significantly different in comparison with


the caking mechanisms of crystals [1,5,10–13]. The relevant mecha-
In the crystallization field, most crystalline products are preferred to nisms of amorphous powders therefore cannot be adapted to crystals.
be produced as single particles with good flowability and stability. These In addition, polymorphic transitions, phase transitions, chemical reac-
crystals are easy to process prior to commercialization. Caking of crys- tions, and the debate of nucleation and growth of crystallization compli-
tals often occurs after a distinct drying step, e.g. transport, storage and cate the study of crystals caking. Hence, there remains a strong need to
some drying processes. It can be very problematic due to unusable or consider particle science and technology with crystallization science
additional equipment being required for de-caking [1]. Quite often 2– and engineering to overcome these issues (Fig. 1). As a foundation, par-
10% of the product is lost during this process [2,3]. To make matters ticle science and technology provides basic means of characterization,
even worse, the quality of products after the process of de-caking some- particle-particle interactions and particle system engineering. Further-
times differs significantly from the original ones, which may cause a se- more, the crystallization science and engineering contains nucleation
ries of problems in the following processes and commercialization. and growth theory, polymorphism, phase transition, crystal engineer-
Hence, in most cases, caking of crystals is undesired and should be ing, etc., which may give crystals caking a profound view on mecha-
prevented. However, most studies focus on the amorphous powders nisms and the prevention methods.
in the food engineering field instead of crystals. Caking of amorphous This review focuses on the caking of crystals, especially during trans-
powders is generally caused by glass transition, a variation of viscosity port and storage. In the following sections, characterization, mecha-
nisms, and prevention of crystals caking are summarized and
⁎ Corresponding author at: School of Chemical Engineering and Technology, Tianjin
discussed with the aid of recent studies. The motivation of novel caking
University, Tianjin, China. indices and devices in particle technology are discussed in the section of
E-mail address: Junbo_gong@tju.edu.cn (J. Gong). characterization. The mechanisms of crystals caking are reviewed from

http://dx.doi.org/10.1016/j.powtec.2017.04.052
0032-5910/© 2017 Elsevier B.V. All rights reserved.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
2 M. Chen et al. / Powder Technology xxx (2017) xxx–xxx

Fig. 1. Solution of crystal caking combined particle science and technology with crystallization science and engineering.

the viewpoint of crystallization science. The polymorphism, phase tran- techniques in recent years. The novel techniques for these indices are
sitions, chemistry reactions, nucleation and growth theory are consid- discussed in Section 2.2. Most of the statistical indices are not common
ered to discuss the possible research points. Based on particle and yet but can be valuable. The potential techniques of these indices and
crystal engineering, and the idea of “quality by design”, the factors for some potential ideas to obtain more precise and repeatable data are
caking prevention are summarized systematically. discussed for further applications in Section 2.3.

2. Characterization 2.1. Mechanical indices

Several indices (Fig. 2) are proposed in characterizing crystal caking. Many researchers focused on the mechanical indices in the past
Caking strength and compressibility are classified as mechanical indices. years due to good development of testers. These testers are easy to es-
Caking strength is used to characterize the force breaking the cake [13, tablish. Some are even commercial, e.g. shear testers, which motivate
14]. Compressibility shows the shrinkage of the specimen during cak- the investigation about mechanical indices, especially caking strength.
ing, where the pressure is related to the extent of powder caking [15]. Some groups have also developed devices that control humidity, tem-
Aspect ratio, circularity, convexity, sphericity and bluntness are mor- perature, pressure, and location of particles to fit more specific situa-
phological indices. They characterize the shape and structure of a parti- tions. These testers are summarized in Table 1.
cle. Aspect ratio (AR) is the ratio between the major axis and minor axis
of the ellipse encircling the object [16]. Circularity is the ratio of the 2.2. Morphological indices
object's projected area to the square of the perimeter of the object
[17]. Convexity is the perimeter of the convex hull of the object divided Particle morphology is one of the most important aspects in particle
by its perimeter [16,18]. It presents how spiky a particle is. Sphericity is science and technology. Caked particles have more complex morpho-
the ratio of the surface area of a sphere having the same volume as the logical characteristics, which is valuable to investigate. So far, utilizing
particle to the actual surface area of the particle [18,19]. Bluntness de- atomic force microscopy (AFM) technology has become the tendency
scribes how sharp or round the fine convex structures on the particle to investigate particle surface details in nanoscale. With the aid of
surface are [20]. In general, agglomerates present irregular shapes. AFM, Bode et al. [43,44] observed step pinning of NaCl crystal caused
These morphological indices of agglomerates differ a lot from the pri- by the anticaking agents, i.e. ferrocyanide and iron (III) meso-tartaric
mary individual particles, so sorting and filtering on these indices will acid. The AFM showed that the step heights are only one atomic layer
quickly identify the level of caking. (2.8 Å). Then, a blocking mechanism for crystal growth of sodium
Statistical indices include caking ratio, mechanical indices distribu- chloride crystals by ferrocyanide was suggested. Stellacci et al. [45,46]
tions, and morphological indices distributions. Caking ratio is to calcu- developed the high-resolution amplitude-modulation atomic force mi-
late the percentage of caked particles, which characterizes the degree croscopy (AM-AFM) to improve the precision of AFM significantly. The
of caking [21–23]. Distributions of mechanical and morphological indi- images could even show a single molecule or atom. In their subsequent
ces are statistical data from a mass of individual particles, which show work, this technology demonstrated that the stearic acid molecules
comprehensive views on the whole sample. tend to act as “pinning points” on the calcite's surface and slow down
Caking test methods have been developed based on these caking in- the crystal's restructuring kinetics [47]. These experiments utilizing
dices. Most available approaches tend to inform the operator whether a AFM have offered some possible ideas for anticaking agent. Besides,
product has caked or not. A highly desirable feature of a caking test is AFM also provides both morphological microstructure and interactions.
the ability to diagnose whether caking will occur in the future. The com- This advantage could provide insights into caking mechanisms. Howev-
mon and potential caking test methods (Fig. 2) for these caking indices er, using AFM is costly, time consuming [48], and limited to the study of
are introduced below. Mechanical indices are common in literature. The massive particles system. For this problem, optical and scanning elec-
conventional methods of these indices are summarized in Section 2.1. tron microscopes could be adopted. Previously, image analysis based
Morphological indices are less common than mechanical ones but on optical and scanning electron microscopes primarily depends on
could be utilized directly for caking due to the improvement of manual measurement of numerous photographs [49,50]. However,

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
M. Chen et al. / Powder Technology xxx (2017) xxx–xxx 3

Fig. 2. Caking indices and the relevant test methods.

now, after technical improvement of the optical and scanning electron individual data into a statistical one, it is time-consuming and always
microscopes and the image analysis method, more morphological indi- impossible to measure caking strength of each crystal bridge for a
ces can be obtained. This technology provides number-based statistics. mass of particles that have agglomerated into lumps. In fact, the
It means that the data are generated on individual particles as well as main reason that mechanical indices distributions are not common
on the sample as a whole (Fig. 3). This may motivate further research yet is due to these limitations. Although, Wahl et al. [3] presented a
on surfaces and crystal bridge of caked particles. Furthermore, the sta- simulation method to predict the crystal bridge strength distribu-
tistical indices may be analyzed through properly screening morpholog- tion, it still needs a lot of work to completely develop it. In regard
ical indices. In the following section, the statistical index, i.e. caking to the morphological indices distributions, thanks to the devices
ratio, shows this screening operation as an example. and software, e.g. Morphologi G3 and Particlematric (Phenom,
Netherlands), this kind of statistical indices could be estimated auto-
2.3. Statistical indices matically in minutes. Since it is easy to understand, instead of only
presenting morphological indices distributions by these techniques,
A mass of particles at any given time have to be characterized in in- an indirect test method for evaluating another important statistical
dustrial production, which means the tests for a single particle or parti- index, caking ratio, based on the analysis of these distributions is
cles system (e.g. crystal bridge bonding between particles) are not mainly introduced in the following.
effective enough. In this situation, it is of great importance to obtain Caking ratio is the proportion of caked particles (mass or number) to
some statistical data for understanding the products comprehensively. the whole collection of particles [21–23], which is very necessary to
Theoretically, the distributions of mechanical and morphological indices fully understand the extent of caking comprehensively. Thus, it is of im-
could also be regarded as statistical ones as well. It should be noted that portance for both scientific research and industrial manufacture. Two
there still remains a need to develop some effective measurement or examples are presented below to show the estimate of caking ratio
analysis methods for these statistical distributions. e.g. to convert based on morphological indices distributions.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
4 M. Chen et al. / Powder Technology xxx (2017) xxx–xxx

Table 1
Summary of common mechanical indices testers.

Caking indices Testers Summary

Caking strength Shear testers Carson et al. [24] made a general classification of shear testing techniques. Direct shear testers include translational and rotational, while indirect
ones include biaxial and triaxial. Because of easy operation and low investment, the translational shear testers, e.g. shear cell, have become an
industrial standard and been proven to be the most multifunctional shear tester for applications. The shear cell was originally developed by Jenike
[25]. It consists of two concentric rings [25,26], and is usually placed in an environmental chamber to control temperature and relative humidity
[27]. Now it is commonly used to rank granular materials according to their flowability [28].Other shear testers were reported in literatures
[6,24,28,29]. These testers offer unlimited strain, steady-state flow conditions, etc., which are capable of being automated and utilized to study the
shear flow of cohesive, caked particles.
Uniaxial tester A typical uniaxial tester provides compressive stress that is maintained for a specified time. The main advantage is that it provides a
fundamental measure of caking strength at a defined consolidation load [30,31]. The cell walls are smooth enough to reduce the friction
with samples for precision. However, such a device isolates the sample from ambient humidity. Consequently, making the humidity
difficult to be controlled. To overcome the problem of humidity, consumer oriented solids transport technology (COSTT) uniaxial tester
was developed [3]. It differs from typical uniaxial tester in two aspects: due to porous and sintered material, the tester allows fast diffusion
of moisture through the cell walls towards the sample, which indicates that the walls are fairly rough. In addition, the tester has been
equipped with a raised bottom to compensate for the Janssen effect [3]. After that, Calvert et al. [32] developed a uniaxial caking test
method that is capable of simultaneously conditioning and consolidating particle samples. However, uniaxial testers are still not able to
confirm whether the sample has reached environmental equilibrium. Recently, an extended uniaxial tester, the edinburgh powder tester
(EPT) was developed. The main differences between the EPT and previous uniaxial testers lies in the attention to mechanical details, the
level of care directed to the consolidation as well as failure load application, and the strategic intent. In the EPT, boundary effect is
minimized by allowing the powder sample to compress from both the top and bottom [33].
Micromanipula- In 2012, Haider et al. [34] developed MPT to study the influence of environmental conditions on the caking mechanism. It is designed for
tion measuring the compression forces between particles larger than 500 μm in diameter in contact under load and the adhesion forces upon
particle their separation. This device makes it possible to measure the effect of process conditions in pressure agglomeration on particle
tester (MPT) interactions and resulting adhesion: external pressure, compression velocity and physical state of the material. Although the sample used
is amorphous, it could be adopted for the investigation of crystals caking. However, it should be noted that the MPT cannot supply
information about forces in the piconewton range [35].
Device to apply The advantage of DAPP is the ability of applying precise pressure on only two particles. The movable pin of DAPP is connected to a piezo
pressure on electric load cell. The piezo cell and the flexible pin are moved backwards, thus pulling the particles apart by using a stepper motor. The
particles (DAPP) maximum force breaking the bridge between these particles is measured [3]. It therefore is able to investigate the interaction.
Force Force displacement tester measures the peak force required in breaking caked powder. It was applied to investigate the effect of
displacement temperature, relative humidity, and exposure time on caking strength [36,1,37].
tester
Blow tester The blow tester, in its multiple test form, can measure the forces holding the particles together while temperature and relative humidity
are held constant. It has some advantages, e.g. multiple trials can be performed on one sample, unlike methods such as friability or angle of
repose, where the sample is poured out of the container or onto a surface. It also enables the progression of the formation of liquid bridges
to be followed with time [38,39].
Compressibility Penetration test The penetration test method considers the unconfined yield strength of particles [40]. The punch was removed from the sample, and the die with
sample compact inside was placed in the Instron machine to carry out the penetration force measurement. Hassanpour [15] developed a ball
indentation method for evaluating the hardness of a powder. A sample is consolidated into a cylindrical die which is made of low friction materials
in order to reduce the effects of wall friction. The consolidated bed is then indented using a spherical indenter. An advantage of this approach is its
sensitivity and ability to measure the onset of caking. In recent years ball indentation has received further attention to improve our understanding
of this method [41].
ICI caking tester The ICI caking tester is a shallow and sealed mould. Caking is induced using pneumatic pressure. A displacement transducer has also been
used with this tester to evaluate the shrinkage of the specimen during caking [42]. After compression, the top cover of the tester is
removed and a load is applied to the top of the sample until a plug of material is displaced, the pressure at which this occurs can help
determine the extent of powder caking.

Example A: screen by circularity and convexity to determine the degree of caking. Using a sieve aperture that is larger
The circularity and convexity of caked particles differ a lot from un- than the constituent particles, will allow the free flowing material to
caked ones. They may be the key indices to separate the sample. First of pass it. A certain percentage of the material will not pass if the material
all, the circularity distribution (Figs. 4 and 5(a) and (c)) show that all has caked. The caking ratio here is given as the mass fraction of material
particles in part a have caked while the ones in part c didn't. that is retained on the sieve. However, there may be a disadvantage.
Fig. 5(b) shows that the index circularity could not separate the parti- Sieving causes collision among particles and friction between particles
cles well in part b. and sieve apertures. The collision and friction may break the particles
Secondly, the particles in part b of Fig. 4 were analyzed with convex- that caked slightly. As a result, the measured caking ratio may be
ity. Figs. 6 and 7 show that these particles were separated successfully. below the real one. To prevent this error, this sieving method should
Since all these data are number-based, the caking ratio could be estimat- only be utilized to measure the particles that caked badly while main-
ed precisely with number of particles. taining slight movement. Thus, it is preferable to use a sieve shaker ma-
Example B: screen by primary and secondary particles chine where the intensity can be controlled and maintained to allow for
Caked particle is secondary particle. Hence, the caked particles could repeated trials. As for the manual sieving, it is necessary to guarantee
be identified by analyzing primary and secondary particles. The key the same operator and procedure of movement. e.g. slope the sieves
technique is shown in Fig. 8. The spherical particles are successfully an- with given angle of inclination and given times to provide a constant
alyzed as secondary particles consisting of primary particles. Thus, ac- force because of gravity. Besides, the intensity of sieving could be con-
cording to the image analysis of number of primary and secondary trolled by the angle of inclination. No matter which sieving method is
particles before and after caking, the caking ratio could be evaluated. utilized, the two piles of separated particles should be checked via mi-
The two examples above show some potential ideas for evaluating croscope to make sure that the caked particles have been totally sieved.
caking ratio. These ideas should be properly modified to fit some specif- Sieving methods are low-cost and high-loading, which is preferred
ic conditions of samples, devices, and software. by industrial manufacturers. However, the unexpected breakage of
It should be noted that the common method of measuring caking caked particles during friction and collision leads to the errors on the
ratio is still sieving in recent literature [21–23]. Sieves have been used low side. On the other hand, although image analysis is high-cost and

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
M. Chen et al. / Powder Technology xxx (2017) xxx–xxx 5

Fig. 3. Optical microscope images of caked sodium phosphate and image analysis data of one of these crystals. The images and data were provided by a morphology analyzer (Morphologi
G3, Malvern Instruments Ltd., United Kingdom).

low-loading, it provides visual data processing technology, which helps situations, e.g. drying process with mother liquor or rinsing liquid on
obtain precise and repeatable results. These advanced technologies crystals surfaces [21]. In this process (Fig. 9), caking of crystals has
would contribute a lot in the further research. three typical stages: 1). moisture sorption, 2). liquid bridges effects,
3). crystal bridges formation [10]. The three stages are reviewed below.

3. Mechanisms 3.1. Moisture sorption

Generally, caking of crystals occurs during the absorption and de- Caking starts from the moisture sorption, which leads to the interac-
sorption of moisture content, e.g. storage and transport under high hu- tion of water with solids [52]. The crystals have the deliquescence point
midity, drying process with alcohol or other rinsing liquids on the RH0 and RHcc that is the critical RH at which capillary condensation re-
particles surfaces, etc. The liquid content plays an important role in sults in liquid bridge formation (Fig. 9, moisture sorption). The illustra-
the caking process [51]. Most crystals (carbohydrates, organic and tion (Fig. 10) shows that the moisture is absorbed slowly under RH0
inorganic acids and their salts, and various mineral salts) are polar sub- while it is much faster when exceeding RH0. Due to the difference, one
stances that dissolve in aqueous solutions. Conversely, apolar molecules of two main processes may be facilitated: capillary condensation or del-
interact primarily with apolar liquids such as chloroform, triacetin, car- iquescence. The two processes are reviewed below.
bon acids, carbon acid esters, phthalic acid and toluene [5]. In most stud- Capillary condensation can be described as the direct condensation
ies of crystals caking, water is widely adopted due to the typical caking of moisture at contact points of particles. In capillaries formed by the
process caused by moisture sorption and desorption under various con- contact points, the pressure is lower than surrounding atmosphere. If
ditions of humidity and temperature. Theoretically, this process is able the pressure is lowered enough to cause the vapor pressure to be great-
to show the integrated mechanisms of caking by controlling relevant er than the saturated partial pressure above the capillary surface, then
conditions, which therefore can be referred by other liquids or caking condensation will occur [39].
The Kelvin equation describes the occurrence of capillary bridges de-
pending on the relative humidity RH and the curvature of the meniscus
[11,39,52].
     
P RH γ∙V m 1 1 2∙γ∙ cosθ∙V m
ln ¼ ln ¼ ln ðaW Þ ¼ − ∙ − ¼−
p0 100 R∙T s l r k ∙R∙T
ð1Þ

Where P is external water vapor pressure, p0 is vapor pressure of


pure water, aW is water activity, γ is surface or interfacial tension, Vm
is molar volume of the liquid, R is gas constant, T is temperature, s is ra-
dius of meniscus, l is azimuthal radius of a meniscus, θ is contact angle of
liquid to solid surface, rk is Kelvin radius. Symbols are shown in Fig. 11.
The Kelvin equation can be used to predict the Kelvin radius where
Fig. 4. Circularity distribution of sodium phosphate particles by Morphologi G3. capillary condensation will occur for a range of water activities aW.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
6 M. Chen et al. / Powder Technology xxx (2017) xxx–xxx

Fig. 5. Optical microscope images of sodium phosphate by Morphologi G3. a) particles in part a of Fig. 4. b) particles in part b of Fig. 4. c) particles in part c of Fig. 4.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
M. Chen et al. / Powder Technology xxx (2017) xxx–xxx 7

Fig. 5 (continued).

Capillaries of smaller radius than the Kelvin radius will be full of water/ the chemical potential μ of pure liquid water in equilibrium with its
material solution [39]. vapor:
Deliquescence is a first-order phase transition process whereby a
substance dissolves in atmospheric moisture when the relative hu- μ ¼ μ 0 þ RT ln p0 ð2Þ
midity (RH) reaches or exceeds a certain critical value RH0 [53–57].
Allan et al. [58] summarized the RH0 measurement methods. Where μ0 is the standard chemical potential.
The aqueous solution is thermodynamically favored over the The difference in chemical potential creates a driving force for water
crystalline phase, which is known to occur sharply with increasing to collect on the surface of the crystal as long as the solute is present
relative humidity on reaching free energy equality between the ini- which can be expressed as:
tial (salt particle plus vapor) and final (fully dissolved solution  
drop plus vapor) states [59–61], resulting in spontaneous moisture ps
μ s −μ ¼ RT  ln ¼ RT  ln ðaW Þ ð3Þ
sorption and formation of a condensate film on the surface of the ma- p0
terial. The thermodynamics of deliquescence can be considered by Where μs is chemical potentials of the solution on the crystal sur-
faces, ps is vapor pressure of the solution.
The kinetics of deliquescence increase as the RH increases above RH0
depending on the ratio RH/RH0. The kinetic equation can be simplified as
follows:
 
RH
W ¼ ðC þ F Þ ln ð4Þ
RHo
Where W is the sorption rate, C and F represent the conductive and
radiative terms, respectively. C and F are found to contribute equally
to heat transport.
More details about the thermodynamics and kinetics of deliques-
cence are discussed in references [62,63].
The summary above shows that both capillary condensation and
deliquescence are facilitated depending on the RH. Hence, it is impor-
tant to note that special situations may occur due to the properties of
Fig. 6. Convexity contribution of sodium phosphate particles in part b of Fig. 4. the crystals with water (or other liquids in some conditions), e.g.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
8 M. Chen et al. / Powder Technology xxx (2017) xxx–xxx

Fig. 7. Optical microscope images of sodium phosphate by Morphologi G3. a) particles in part a of Fig. 6. b) particles in part b of Fig. 6.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
M. Chen et al. / Powder Technology xxx (2017) xxx–xxx 9

3.2. Liquid bridge

Moisture sorption leads to liquid bridges generating between parti-


cles. In a capillary condensation process, the water volume of a steady
liquid bridge could be estimated with Eq. (1). As for a deliquescence
process, the water volume increases continuously and finally results in
a total dissolution of the sample [62,63]. The liquid bridge, or liquid
film to some extent (Fig. 9, liquid film under deliquescence), therefore
keeps increasing with time. Komunjer et al. [70] described the kinetics
of absorption of water vapor and formation of solution with the Knud-
sen formula:

dn KP 0
¼ K ðP−pn Þ ¼ ðRH−RH0 Þ ð6Þ
dt 100

Where n is number of molecules or moles of moisture content enter-


ing the condensed phase, t is time, dndt
is therefore the rate of adsorption
of water onto the crystal surfaces. K is Knudsen coefficient, pn is non-
negligible pressure of condensed phase.
This theoretical approach showed a good agreement with the exper-
imental data. It was concluded that dn dt
remained constant until the crys-
tals completely dissolved. After that, the rate dropped slowly and finally
the equilibrium was reached when pn became equal to P.
Liquid bridges also present forces affecting contact between particles
[71], which may impact crystal bridge formation. The generated forces
are capillary forces first studied by Fisher [72] and Haines [73] in the
mid 1920s. In 2009, Butt reviewed the capillary forces between two par-
ticles or planes [74]. The capillary force f for sphere-sphere contact ge-
ometry (Fig. 11) can be calculated with Eq. (7):
  
1 1
f ¼ π ∙ γ ∙ R1 ∙ sinβ1 ∙ 2 sinðθ1 þ β1 Þ þ R1 ∙ sinβ1 ∙ − ð7Þ
s l

Where R1 is radius of a sphere, β1 and θ1 are filling angle and contact


angle of the liquid to the sphere with radius R1, respectively.
Various contact geometries, e.g. sphere-plane, core-core, core-plane,
plane-plane were also described in detail [74].
In addition, mathematical models of more complex liquid bridges
and alternative methods considering energy of particles system were
also studied. The liquid bridge does not only lead to the formation of a
liquid dimers as mentioned above, but also to more complex liquid
phase morphologies, e.g. liquid trimer, pentamer and tetrahedron.
Fig. 8. a) Scanning electron microscope (Phenom ProX, Phenom, Netherlands) of LiCO3 These liquid morphologies have been observed via the X-ray tomogra-
particles. b) Recognition of primary and secondary particles (Particlematric, Phenom,
phy technique [75–77]. Depending on the water content, these liquid
Netherlands).
morphologies lead to more complex states of cohesion: pendular (low
water content, forms in dimer, Fig. 12a), funicular (increasing the
water content, forms in trimer, pentamer and tetrahedron, Fig. 12b),
polymorphic transitions [13,64–66], phase transitions [67–69], chemi-
and capillary (nearly saturated, forms in trimer, pentamer and tetrahe-
cal reaction [59], etc. These phenomena change moisture sorption, as a
dron, Fig. 12c) [76,78]. So far, only the pendular state has been investi-
result, even further complicating the entire caking process. Listiohadi
gated systematically. In the 1990s, researchers started to study the
et al. [13] investigated the difference of moisture sorption behavior
funicular state with the liquid trimer, i.e. the fundamental structure of
among lactose polymorphs. They concluded that the polymorphic and
this state [78]. The investigation has been developed from two dimen-
isomeric properties of lactose play a significant role in its hygroscopicity.
sional to three dimensional funicular water bridge between three
The interconversion between the lactose polymorphs changes the α/β
spheres so far [79–83]. However, these studies are extremely limited
and anhydrous/monohydrate ratio, which has a great impact on mois-
due to the high complexity of the geometry. For the crystals, one
ture sorption and crystal bridge formation. Kwok et al. [59] investigated
method of determining the geometry is to create a particle geometry
the impact of sucrose hydrolysis on moisture sorption. One of the reac-
distribution, e.g. morphological indices distribution in Section 2.3.
tions is presented by the following chemical equation:
Nonetheless, the analytical solutions based on Young-Laplace law are
complicated. The alternative method is obtaining the index that pre-
sucroseðaqÞ þ H2 OðlÞ → glucoseðaqÞ þ fructoseðaqÞ ΔH h ð5Þ sents the equilibrium/average states of the samples. This concept is
¼ −14:93 KJ=mol often related to the study of energy. Based on the “Surface Evolver” by
Brakke [84], a software using calculus of variations to minimize the
inter-facial energy iteratively for the solutions of the Young-Laplace
Thermodynamically, the water vapor condenses onto the sucrose equation, Cater [85] and Wang [86] concerned the calculation of forces
particles surfaces and leads to the reaction. The moisture sorption be- for three spheres. The effects of inter-particle distance, water volume
havior of a mixture therefore differs from that of a pure substance. and contact angles were investigated qualitatively. For a same fixed

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
10 M. Chen et al. / Powder Technology xxx (2017) xxx–xxx

Fig. 9. Caking process under different conditions. RH is the environmental relative humidity value, RH0 is the deliquescence point of crystals, and RHcc is the critical RH at which capillary
condensation results in liquid bridge formation, which over time may solidify by mass transfer, without decreasing the RH.

total water volume, the capillary force is weakened by water bridge co- inefficient facing more complicated particle systems. For this situation,
alescence to a funicular bridge when the spheres are packed together, studies with the energy methods provide comprehensive and qualita-
but the situation may vary for different contact angles and inter- tive results, which may accelerate the development of the former. E.g.
particle distances. For the constant mean curvature condition, water in the near future, the basic conclusions about geometry of liquid trimer
bridge coalescence does not alter capillary force significantly when par- from the energy method would support further studies of the mathe-
ticles are packed closely, but the discrepancy is larger by increasing the matical models. It should be noted, however, that several problems of
gap [86]. Chen et al. [21,23] investigated the adhesion effects of liquid energy methods need to be solved. E.g. the Lifshitz-van der Waals
bridges thermodynamically utilizing the adhesion free energy that acid-base theory [87] used in adhesion free energy calculation is still
also considered the geometry. The capillary rise method used in the ex- controversial [88–92]. The basic parameters of liquids keep being re-
periments measured the average contact angles of the samples with dif- vised [87]. Besides, more basic parameters of common liquids need to
ferent geometry. Hence, the adhesion free energy calculated with the be measured.
contact angles is a statistical and thermodynamic result. It therefore
presents the average state of samples with geometry in consideration. 3.3. Crystal bridges
Adhesion free energy of liquid bridge between the particles denotes
the effects of repulsion or attraction. Results showed that for the one- The liquid bridge can be supersaturated due to vaporization of water
layer samples under static conditions, the critical value that avoids par- or variation of temperature, leading to recrystallization and then crystal
ticles contact is about 60 mJ/m2. bridge formation between crystals (Fig. 9, recrystallization in crystal
According to the summary, the effects of a liquid bridge depends on bridge) [1]. Interestingly, Kirsch et al. [93] observed that a time-
a vary of factors, e.g. liquid content, number of particles, location of par- dependent mass transfer into the liquid bridge took place, transforming
ticles, geometry of particles, particle shape, particle size, properties of it into a crystal bridge, which cannot be considered as the typical recrys-
particle surfaces, etc. Based on the Young-Laplace equation, the study tallization mechanism. This phenomenon is showed in Fig. 9 (mass
on two-particle and three-particle systems is able to establish the math-
ematical models and analyze each factor quantitatively. It is of great ne-
cessity for studying mechanisms. However, this method seems to be

Fig. 11. Sphere with radius R1 interacting with sphere with radius R2. Two spheres are
Fig. 10. Moisture sorption behavior of crystalline particles. Formation of sloping is caused bonded by a liquid bridge. s is radius of meniscus, l is azimuthal radius of a meniscus, θ
by RH, moisture sorption rate and time. is contact angle of liquid to solid surface, rk is Kelvin radius, β is filling angle.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
M. Chen et al. / Powder Technology xxx (2017) xxx–xxx 11

Fig. 12. Illustrations of three states of liquid presence in particles. Particles are simplified by spheres. a) Pendular state. Isolated liquid bridges are formed between particles. b) Funicular
state. Liquid bridges coalesce into liquid clusters which may connect more than two particles. c) Capillary state. Particles system is nearly saturated with only entrapped air bubbles.

transfer). It occurs during a long time period under strict instant condi- crystallization heat should be evaluated for the recrystallization process,
tions, especially relative humidity and temperature. Scholl et al. [10] etc. Particularly, due to the recrystallization of the liquid bridges pro-
agrees with the explanation of mass transfer, but no relevant compre- ceeding by growth starting from the constituent crystal particles and/
hensive evidence was proposed. In 2015, Vallatos et al. [94] assumed or by the formation of 3D nuclei within the liquid bridges, the evapora-
that the mass transfer takes place through diffusional transport within tion crystallization process could have great impact on the shape, size
a thin layer of solution on the particles surfaces. They applied nuclear and composition of crystal bridges. These differences from crystal
magnetic resonance (NMR) providing the first quantitative measure- bridges may cause a significant effect to the caking strength and caking
ments of diffusion in thin spherical films. Results showed that the mo- ratio. Common shape of crystal bridges is made from a number of
lecular diffusion coefficients in this film are the same as those found in microcrystallites, see Fig. 13(a). While in some special evaporation crys-
bulk solution. The thickness of the liquid layer composed of saturated tallization, needle-like aggregates generated in a droplet [99]. It means
urea (crystal sample, same with Kirsch [93]) solution covering the sur- that the crystal bridge can also be needle-like, see Fig. 13(b). Crystals
face of prills were estimated to be on the order of 10−5 m. As a result with phase transformation properties could also generate a crystal
of the strict conditions, more similar experiments with different bridge in different phases [67–69,103], see Fig. 13(c). Hence, it can be
substances should be carried out to ensure the repeatability and univer- concluded that crystal bridges can be quite complicated and should be
sality. On the assumption of reasonable mass transfer, this mechanism reconsidered with crystallization science instead of simplified models,
may be utilized under both constant and varying ambient conditions. e.g. cylinder-shape model.
In this case, the interaction between mass transfer and dissolution- As mentioned above, crystallization science is of importance in the
recrystallization should be considered and more studies in this field understanding of the formation of liquid bridges and crystal bridges,
are required. see Fig. 14. Supersaturation facilitates the crystal growth around the
Due to the diverse explanations of the crystal bridge and the contro- crystal surfaces, while nucleation occurs further away from these sur-
versial mechanisms of crystal nucleation and growth, so far the kinetics faces. The two-dimensional nucleation growth [104–107] and spiral
of crystal bridge formation studies focus on the simple model based on crystal growth (Burton, Cabrera and Frank theory, BCF) [108] then
the typical mechanism of dissolution-recrystallization. An experimental could be adopted for the growth. Recently, the controversy between
method on sorption-evaporation kinetics was developed by Komunjer the classical nucleation theory (CNT) [109,110] and the two-step nucle-
et al. [70] A theoretical approach describes the process of absorption ation mechanism [111,112] couldn't be ignored. The focus is liquid-
of water vapor and formation of solution-evaporation of water- phase precursors. CNT states that the density fluctuations are concomi-
nucleation and growth of the recrystallized solid. The foundation of tant with fluctuations in the order parameter, so that molecular packing
the kinetic model is based on the Knudsen formula. Eq. (6) shows that within clusters reflects the possible polymorphs of the solute [113]. The
the process is governed by the imposed driving force for evaporation two-step nucleation mechanism shows that the fluctuations in density
i.e., RH0-RH. The kinetic model was established by connecting the stages are disconnected from those in order, so that the initial clusters that
(absorption of water vapor and formation of solution-evaporation of form are liquidlike, and crystalline order appears later on [114–116].
water-nucleation and growth of the recrystallized solid) to the driving This controversy is motivating furhter research on crystal nucleation.
force with the relevant parameters of pressure and concentration [70]. Recently, Bode et al. [117] explained the caking of sodium chloride
The contrast between the theoretical and experimental curves provides dihydrate with crystallization science. They determined the morpholo-
useful insight into the kinetics. In addition, the stages affected by anti- gy of the dihydrate crystals with crystallographic structure. Both classi-
caking agent can be clearly inferred, which is a valuable method for cal and two-step nucleation theories were utilized for the nucleation
studying anticaking. The precision of this experimental method was rate. The growth rate was measured by taking micrographs. The results
improved by Langlet et al. [95] utilizing environmental scanning elec- showed that the two-step nucleation theory seems to better fit the nu-
tronic microscopy (ESEM). Coupled with dynamic vapour sorption cleation rate of sodium chloride dihydrate. The formation of crystal
(DVS), measurements were proved to be appropriate means of some bridge was inhibited by the effective anti-caking agent, Fe(III)-LTA,
quantitative and accurate determinations, e.g. RH0 of crystals and criti- due to the limitation of crystal growth rather than nucleation. Study of
cal supersaturation for nucleation. crystallization always depends on methods of fourier transform infrared
These studies are significant to give a view from the moisture sorp- spectroscopy (FTIR), raman spectroscopy, nuclear magnetic resonance
tion to crystal bridge formation. However, the explanations of crystalli- (NMR), molecular simulation, etc. In the near future, based on these the-
zation from liquid bridge to solid bridge in these studies are still not ories and methods of crystallization, further investigation of liquid and
enough. Though researchers are trying to establish the kinetics of nucle- crystal bridges would provide more novel views in caking mechanisms.
ation and growth, there are still more theories of crystallization science
that can be applied to crystal caking. E.g. the difference between sorp- 4. Prevention
tion and desorption curves is similar with the metastable zone in crys-
tallization [96]; the crystal bridge caused by evaporation may be The unique properties of crystals, e.g. critical value of humidity, spe-
inferred from the kinetics of evaporation crystallization [97–102]; the cific particle geometry, etc., should be considered for the prevention of

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
12 M. Chen et al. / Powder Technology xxx (2017) xxx–xxx

Fig. 13. Illustrations of crystal bridges generated under different conditions. a.) Crystal bridge made of microcrystallites. b). Crystal bridges made of needle-like crystals. c). Crystal bridge
under condition of phase transformation.

crystal caking. On the one hand, theoretically, controlling the moisture method is presented, and the factors that haven't been counted in yet
content is the direct thought of preventing caking. The moisture content are discussed for further study of DEM.
of the product is usually the most important factor in promoting caking.
For most caking mechanisms, the presence of moisture, in the form of
solution phases, is essential. Generally, the higher the degree of drying 4.1. Control ambient conditions
at the time of manufacture is, the less active the caking mechanisms
will be [118]. On the other hand, however, the ideal conditions of low 4.1.1. Humidity and temperature
moisture content cannot be always guaranteed in actual industrial pro- The co-effect of temperature and RH on the product when placed in
duction. Therefore, improving the anticaking property of crystals against storage influences caking tendency. Fig. 15 summarizes the water-solid
the unstable ambient conditions is another important consideration. interactions of crystals under different RH and temperature. Crystals
It must be noted that the experimental research for controlling these present critical values of RH0 and RHcc along with the relevant bound-
factors systematically is complicated. Instead, utilizing the simulation aries. It demonstrates that caking caused by deliquescence (RH0) is
technique would be effective. It is a necessary way to achieve the con- the most problematic, while the one caused by condensation capillary
version from scientific research to production and the idea of “quality (RHcc) is the easiest to be remedied. Caking can be significantly
by design”. To do this, the combination of particle and crystal engineer- inhibited when the co-effect of temperature and RH is under the RHcc
ing would be the key. According to the computational technique, e.g. boundary [2].
discrete element method (DEM) [119], the behavior of a crystal system As mentioned above, the RHcc and RH0 boundaries play key roles in
(particles packing, caking mechanisms, also crystallization) would be controlling ambient conditions in industrial production. They could be
simulated. Caking indices then could be predicted. With the systemati- measured by DVS. Depending on the substances, the RHcc value may
cally designed ambient conditions and crystal parameters, production not be obvious, then the RH0 would be the approximate crucial value.
would be properly prepared according to crystal engineering. So far, More precise measurement of the critical value of facilitating caking
there still remains a strong need to develop simulation methods that was presented by Billings et al. [39]. Based on Eq. (1), they developed
consider all the main factors. Hence, in this section, the research of a measurement of critical water activity (RH/100) through a diagram
each main factor is summarized, a simulation of caking based on DEM of relation between Kelvin radius and caking strength. The relevant

Fig. 14. Crystal nucleation and growth in a liquid bridge.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
M. Chen et al. / Powder Technology xxx (2017) xxx–xxx 13

Fig. 15. Caking behavior of crystals with RH changing under 25 °C.

critical values of RH and temperature for initiation of caking therefore 4.2.1. Particle size and shape
can be estimated. Most crystals are larger than 100μm. The effects of particles size and
shape should be both considered. Past studies have shown that large
particles with granulate or sphere-like shape present better properties
4.1.2. Liquids screening in inhibition of moisture sorption [23] and particles contact. These prop-
In addition to transport and storage, caking also occurs in drying pro- erties could reduce the caking ratio and caking strength.
cesses. In a drying process, the crystals contain a mass of the mother liq- For the simulation of caking behavior, utilizing computational tech-
uid or rinsing liquid which constitute the liquid bridges. The effect of nique, e.g. DEM, to simulate the particles packing is the first step. Hence,
liquid bridges is different based on the liquid used. Chen et al. [21,23] in- the particles size and shape are the most important factors that should
vestigated the effects of liquid bridge made from different liquids on the be primarily considered. The following shows a possible technique rou-
particles contact behavior. They concluded that some liquids present re- tine to predict caking ratio and strength with DEM. To focus on the rou-
pulsion, contrary to the water presenting attraction. The repulsion re- tine, the simulation was simplified by considering particles size only.
duces the chance of particles contact and then inhibits the crystal PFC3D software that is based on DEM was utilized to simulate the par-
bridges formation. Hence, the solid-liquid interaction should be consid- ticles packing with different particles size. Crystalline maltitol was
ered for the liquids screening. adopted as the test sample and the moisture-sorption property was in-
vestigated by DVS [70,95]. The results showed that the capillary conden-
sation point was 85% and deliquescence point was 92.5%. The humidity
4.1.3. Pressure recycle therefore was set to 90%–45%/day. Calculation of moist sorption
The pressure exerted at the bottom of a stack or pile of crystals has was based on the Kelvin equation. It was assumed that the valid diam-
an effect on caking. As with storage time, excessive and prolonged pres- eter of a crystal bridge is equal to half of diameter of the smaller crystal.
sure will result in an increase in caking. Under pressure, particle defor- Particles number was set to 100. Particles sizes were set to 225, 375,
mation may occur, resulting in increased contact area and surface 525, 675μm and 25 particles for each size.
tension [118,120]. In addition, the increased bulk density may lead to Fig. 16 shows the simulation result of caking behavior from 1 day to
a more serious effect of capillary condensation. 40 days. The color zones represent the caked particles, where a dark
Though no systematic and effective solutions have been proposed to color indicates larger caking strength. It shows that the crystals with
prevent the negative impacts of pressure on crystalline products, there smaller particles size caked first and formed strong crystal bridges rath-
have been a few studies focused on finding possible solutions. Pressure er quickly. On the other hand, Fig. 16(e) and (f) shows that the shapes of
should also be quantitatively considered to help solve problems such as caked particles from the simulation result were similar with the real
how to prevent crystals caking under static high pressure during storage ones.
or dynamic pressure during transportation. It is of great necessity to Based on Figs. 16 and 17 show the caking ratio of these crystals with
adopt the novel caking indices and simulation. time quantitatively. For the whole sample, caking ratio increased signif-
icantly in the first 10 days and then slowed down in next 30 days. The
reason is that the smaller particles contributed over 35% to the caking
4.2. Improve anticaking property of crystals ratio at the beginning and in the meantime lead to about 20% larger par-
ticles caking in the first 10 days. While in the next 30 days, most of the
As an effective method, anticaking agent changes production pro- smaller particles had finished caking and larger particles still caked
cess little. It therefore seems to be the preferred method to improve an- slowly, which slowed down the caking ratio. Hence, compared with
ticaking property. However, anticaking agent is limited by the high cost the larger ones, smaller particles show much more positive caking ten-
and difficulty of development and the consequent environmental prob- dency. To prevent crystals caking, it is therefore of great importance to
lems. To address these issues, the enhancement of crystal engineering improve particles size and narrow the particle size distribution.
should be considered first based on “quality by design”. With the specif- The simulation above shows a possible way to predict the caking in-
ic design of crystal size, shape and impurity, the anticaking property dices with DEM. It has been pointed out that particle shape is widely
would be truly improved. As a result, the usage of anticaking agent recognized to have a strong influence on the granular systems. To im-
would decrease or even be avoided. Hence, improving anticaking prop- prove the precision, particle shape therefore should be considered
erty should be considered comprehensively at the start of the design [121–124]. Lu et al. [119] reviewed the DEM for non-spherical particle
production process to achieve the requirement of high quality, low systems. Four methods have been developed to represent the non-
cost, and being environment friendly. spherical particle shapes in DEM: ellipses/ellipsoids, super-quadrics,

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
14 M. Chen et al. / Powder Technology xxx (2017) xxx–xxx

Fig. 16. Simulation of caking behavior by PFC3D and real caked particles image. a) 1 day, caking ratio = 0%; b) 2 days, caking ratio = 8%; c) 7 days, caking ratio = 48%; d) 19 days, caking
ratio = 69%; e) 40 days, caking ratio = 81%; f) real caked particles images under same simulation conditions (×40). The caking ratio was calculated with number of particles.

polygons/polyhedrons, composite spheres, and composite elements. problem to prevent nanoparticles caking (or aggregation to some ex-
The advantage and disadvantage of flexibility, numeral stability, and nu- tent). Most studies have focused on the deagglomeration process in-
meral efficiency for each method have been summarized for modellers stead of prevention.
to choose suitable modeling approach for their particular problem.
As for the very fine particles, especially nanoparticles, the effect of 4.2.2. Impurity
van de Waal's force makes the nanoparticles tend to form very strong Undesired impurity, e.g. small amount of different substances in
agglomerates and lose their high-surface area due to grain growth or crystalline or amorphous forms, also polymorphic transitions and
unavailability of the high surface area where it matters [125]. Studies phase transitions mentioned in Section 3.1, should be avoided. The im-
have shown that most nanoparticle agglomerates are rather large, ex- purity may worsen the anticaking property of crystals by deliquescence
tremely porous, and have a hierarchical fractal structure consisting of lowing. The presence of impurity may lower the RH0 and cause more
subagglomerates or aggregates at a number of different length scales, water to condense and initiate dissolution [61]. The effect of impurity
down to individual nano-size particles. As a result, it is still a serious can be investigated through the moisture sorption curve by DVS.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
M. Chen et al. / Powder Technology xxx (2017) xxx–xxx 15

models. According to the crystallization science, the mechanism of crys-


tal bridge formation will be further studied.
The prevention of crystals caking includes two aspects. On the one
hand, the ambient conditions should be controlled. Keep RH and tem-
perature lower than RHcc boundary to prevent moisture sorption. Re-
duce pressure to prevent crystals from deformation. Select a proper
rinsing liquid for the drying process that will reduce the adhesion effect
caused by liquid bridge. On the other hand, anticaking property of crys-
tals should be improved. Generally, large crystals with granular/spheri-
cal shape and high purity are hard to cake. However, the caking of nano-
crystals is still a serious problem. Anticaking agent is an effective mean
against caking under the condition that they are nontoxic, easy to han-
dle and only required in small amounts.
The simulation and prediction of crystal caking behavior would be
Fig. 17. Caking ratio (calculated with number of particles) vs. time. Small crystals include an important direction. It is able to estimate the extent of caking
225 μm and 375 μm maltitol. Large crystals include 525 μm and 675 μm maltitol. Caking under given conditions of temperature, humidity, particles size, parti-
ratio of whole crystals is the sum of the two.
cles shape, the amount of anticaking agent, etc. Thus, it may guide the
enhanced design of crystal engineering. The focus would be a combina-
4.2.3. Anticaking agent tion between computational techniques and caking experiments.
Anticaking agents are substances added into products to prevent
caking [126]. For crystals (e.g. ascorbic acid, sodium ascorbate, citric
acid, glucose, sucrose, and sodium chloride), common anticaking agents Acknowledgements
include calcium silicate, calcium stearate, silicon dioxide, tricalcium
phosphate, and corn starch [126]. Anticaking agents of crystals may The authors are grateful to the financial support of National Natural
function by one of several mechanisms: 1). Prevent moisture sorption Science Foundation of China (81361140344, 21676179 and 21376164),
and formation of liquid bridges. Some anticaking agents compete with National 863 Program (2015AA021002), major project of Tianjin
the host powder for moisture and/or create moisture-protective bar- (15JCZDJC33200), Major National Scientific Instrument Development
riers on the surface of hygroscopic particles or physical barriers between Project 21527812, and Major Science and Technology Program for Wa-
particles. 2). Smooth surfaces to eliminate inter-particle friction. 3). In- ter Pollution Control and Treatment (2015ZX07202-13).
hibit formation of crystal bridges as a crystal growth inhibitor [43,
127]. Anticaking agents are commonly used in small amounts, nontoxic References
and easy to handle. e.g. the current anticaking agent used for NaCl is fer-
[1] J.J. Fitzpatrick, N. Descamps, K. O'Meara, C. Jones, D. Walsh, M. Spitere, Comparing
rocyanide. A small amount can effectively block further crystal growth the caking behaviours of skim milk powder, amorphous maltodextrin and crystal-
and prevent particles from caking [44]. However, there are drawbacks. line common salt, Powder Technol. 204 (2010) 131–137.
It may decrease the efficiency in the production of chlorine gas with [2] U. Bröckel, M. Wahl, R. Kirsch, H.J. Feise, Formation and growth of crystal bridges in
bulk solids, Chem. Eng. Technol. 29 (6) (2006) 691–695.
NaCl by electrolysis. Therefore, a new anticaking agent was introduced: [3] M. Wahl, U. Bröckel, L. Brendel, H.J. Feise, B. Weigl, M. Röck, Understanding powder
Fe-mTA. It is almost as effective as ferrocyanide and eliminates the prob- caking: predicting caking strength from individual particle contacts, Powder
lem [43,44,117,128]. Technol. 188 (2) (2008) 147–152.
[4] E.K. Silva, S.V. Borges, J.M.G.D. Costa, F. Queiroz, Thermodynamic properties, kinet-
Screening anticaking agent is still based on the trial and error meth- ics and adsorption mechanisms of swiss cheese bioaroma powder, Powder
od. To decrease the difficulty and high cost of development, predicting Technol. 272 (2015) 181–188.
the effect of anticaking would become primary focus in further research. [5] S. Palzer, Agglomeration of pharmaceutical, detergent, chemical and food
powders-similarities and differences of materials and processes, Powder Technol.
The simulation mentioned in Section 4.2.1 may be an effective solution. 206 (1–2) (2011) 2–17.
It would be a key point to combine the simulation technology of parti- [6] M. Hartmann, S. Palzer, Caking of amorphous powders — material aspects, model-
cles packing with the properties of particles that are easy to measure ling and applications, Powder Technol. 206 (2011) 112–121.
[7] S. Palzer, The relation between material properties and supra-molecular structure
to predict the complex caking behavior.
of water-soluble food solids, Trends Food Sci. Technol. 21 (2010) 12–25.
[8] S. Palzer, Influence of material properties on the agglomeration of water-soluble
amorphous particles, Powder Technol. 189 (2) (2009) 318–326.
5. Conclusions [9] S. Palzer, The effect of glass transition on the desired and undesired agglomeration
of amorphous food powders, Chem. Eng. Sci. 60 (14) (2005) 3959–3968.
[10] S.K. Scholl, S.J. Schmidt, Determining the physical stability and water-solid interac-
Caking indices are classified as mechanical (caking strength and tions responsible for caking during storage of glucose monohydrate, J. Food Meas.
compressibility), morphological (aspect ratio, circularity, convexity, Charact. 8 (4) (2014) 316–325.
[11] D. Dopfer, S. Palzer, S. Heinrich, L. Fries, S. Antonyuk, C. Haider, Adhesion mecha-
sphericity, and bluntness), and statistical (mechanical indices distribu- nisms between water soluble particles, Powder Technol. 238 (4) (2013) 35–49.
tions, morphological indices distributions, and caking ratio) ones. The [12] M. Carpin, H. Bertelsen, J.K. Bech, R. Jeantet, J. Risbo, P. Schuck, Caking of lactose: A
relevant characterization methods of these indices are summarized. It critical review, Trends Food Sci. Technol. 53 (2016) 1–12.
[13] Y. Listiohadi, J.A. Hourigan, R.W. Sleigh, R.J. Steele, Moisture sorption, compressibil-
has been pointed out that image analysis should be utilized to obtain ity and caking of lactose polymorphs, Int. J. Pharm. 359 (359) (2008) 123-34.
both morphological and statistical indices in the further research. [14] P. Pierrat, H.S. Caram, Tensile strength of wet granular materials, Powder Technol.
Crystals caking caused by moisture sorption and desorption is 91 (2) (1997) 83-93(11).
[15] A. Hassanpour, M. Ghadiri, Characterisation of flowability of loosely compacted co-
discussed. The process is divided into three steps: moisture sorption, hesive powders by indentation, Part. Part. Syst. Charact. 24 (2) (2007) 117–123.
liquid bridge, and crystal bridge. Depending on the RHcc and RH0, capil- [16] E.M. Ålander, M.S. Uusi-Penttilä, Åke C. Rasmuson, Characterization of paracetamol
lary condensation (RHcc b RH b RH0) or deliquescence (RH N RH0) agglomerates by image analysis and strength measurement, Powder Technol. 130
(3) (2003) 298–306.
causes moisture sorption and leads to liquid bridge formation. The ef-
[17] P.G.C. Petean, M.L. Aguiar, Determining the adhesion force between particles and
fects of liquid bridge formation include dissolution of crystals and capil- rough surfaces, Powder Technol. 274 (2015) 67–76.
lary forces. Finally, recrystallization or mass transfer promotes crystal [18] C.F. Mora, A.K.H. Kwan, Sphericity, shape factor, and convexity measurement of
bridges. For further analysis of the crystal caking process, the model of coarse aggregate for concrete using digital image processing, Cem. Concr. Res. 30
(3) (2000) 351–358.
capillary forces should consider the actual geometry of particles. The [19] R.D. Barksdale, M.A. Kemp, W.J. Sheffield, J.L. Hubbard, Measurement of aggregate
study of mass transfer may provide more quantitative results and shape, surface area, and roughness, Transp. Res. Rec. 1301 (1991) 107–116.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
16 M. Chen et al. / Powder Technology xxx (2017) xxx–xxx

[20] E. Pirard, Shape processing and analysis using the calypter, J. Microsc. 175 (3) [55] G. Zografi, B. Hancock, Water-solid interactions in pharmaceutical systems, Pro-
(1994) 214–221. ceedings of International Congress on Pharmaceutical Sciences F.I.P. Stuttgart,
[21] M. Chen, S. Wu, W. Tang, J. Gong, Caking and adhesion free energy of maltitol: study- Medpharm Scientific, Germany 1994, pp. 405–419.
ing of mechanism in adhesion process, Powder Technol. 272 (2015) 235–240. [56] A.D. Mcnaught, A. Wilkinson, IUPAC Compendium of Chemical Terminology, Black-
[22] J. Cleaver, G. Karatzas, S. Louis, I. Hayati, Moisture-induced caking of boric acid well Science, Cambridge, 1997.
powder, Powder Technol. 146 (2004) 93–101. [57] A.M. Stoklosa, R.A. Lipasek, L.S. Taylor, L.J. Mauer, Effects of storage conditions, for-
[23] M. Chen, L. Lin, Y. Zhang, S. Wu, E. Liu, K. Wang, J. Wang, J. Gong, Mechanism and mulation, and particle size on moisture sorption and flowability of powders: a
inhibition of trisodium phosphate particles caking: study on particles shape and study of deliquescent ingredient blends, Food Res. Int. 49 (2) (2012) 78–791.
solubility, Particuology 27 (2016) 115–121. [58] M. Allan, L.J. Mauer, Comparison of methods for determining the deliques-
[24] J.W. Carson, H. Wilms, Development of an international standard for shear testing, cence points of single crystalline ingredients and blends, Food Chem. 195
Powder Technol. 167 (1) (2006) 1–9. (2015) 29–38.
[25] A.W. Jenike, Gravity flow of bulk solids, Bulletin No. 108, Utah State University, [59] K. Kwok, L.J. Mauer, L.S. Taylor, Kinetics of moisture-induced hydrolysis in powder
1961. blends stored at and below the deliquescence relative humidity: investigation of
[26] M. Röck, J. Schwedes, Investigations on the caking behaviour of bulk solids- sucrose-citric acid mixtures, J. Agric. Food Chem. 58 (22) (2010) 11716–11724.
macroscale experiments, Powder Technol. 157 (2005) 121–127. [60] M. Dupas-Langlet, M. Benali, I. Pezron, K. Saleh, L. Metlas-Komunjer, The impact of
[27] J.J. Fitzpatrick, K. Barry, P.S.M. Cerqueira, T. Iqbal, J. O'Neill, Y.H. Roos, Effect of com- deliquescence lowering on the caking of powder mixtures, Powder Technol. 270
position and storage conditions on the flowability of dairy powders, Int. Dairy J. 17 (2015) 502–509.
(4) (2007) 383–392. [61] J.A. Baird, R. Olayo-Valles, C. Rinaldi, L.S. Taylor, Effect of molecular weight, temper-
[28] D. Schulze, Powders and Bulk Solids: Behaviour, Characterization, Storage and ature, and additives on the moisture sorption properties of polyethylene glycol, J.
Flow, Springer, Berlin, 2008. Pharm. Sci. 99 (1) (2010) 154–168.
[29] A. Weuster, S. Strege, L. Brendel, H. Zetzener, D.E. Wolf, A. Kwade, Shear flow of co- [62] L.J. Mauer, L.S. Taylor, Water-solids interactions: deliquescence, Rev. Food Sci.
hesive powders with contact crystallization: experiment, model and calibration, Technol. 1 (2010) 41–63.
Granul. Matter 17 (2) (2015) 271–286. [63] L.J. Mauer, L.S. Taylor, Deliquescence of pharmaceutical systems, Pharm. Dev.
[30] M.E. Chávez, N.A. Santamaría, J.C. Gumy, P. Marchal, Moisture-induced caking of Technol. 15 (6) (2010) 582-94.
beverage powders, J. Sci. Food Agric. 91 (14) (2011) 2582-6. [64] P.P. Bag, C.M. Reddy, Screening and selective preparation of polymorphs by fast
[31] J. Schwedes, Review on testers for measuring flow properties of bulk solids, Granul. evaporation method: a case study of aspirin, anthranilic acid, and niflumic acid,
Matter 5 (1) (2003) 1–43. Cryst. Growth Des. 12 (6) (2012) 2740–2743.
[32] G. Calvert, N. Curcic, C. Redhead, H. Ahmadian, C. Owen, D. Beckett, A new environ- [65] W. Du, Q. Yin, J. Gong, Y. Bao, X. Zhang, X. Sun, Effects of solvent on polymorph for-
mental bulk powder caking tester, Powder Technol. 249 (11) (2013) 323–329. mation and nucleation of prasugrel hydrochloride, Cryst. Growth Des. 14 (9)
[33] S.C. Thakur, H. Ahmadian, S. Jin, Y.O. Jin, An experimental and numerical study of (2014) 4519–4525.
packing, compression, and caking behaviour of detergent powders, Particuology [66] Z. Liu, Q. Yin, H. Zhang, J. Gong, Investigation of the crystallization of disodium 5′-
12 (3) (2014) 2–12. inosinate in a water + ethanol system: solubility, nucleation mechanism, and crys-
[34] C.I. Haider, T. Althaus, G. Niederreiter, M.J. Hounslow, S. Palzer, A.D. Salman, A tal morphology, Ind. Eng. Chem. Res. 53 (21) (2014).
micromanipulation particle tester for agglomeration contact mechanism studies in [67] X. Zhang, C. Xie, Y. Huang, B. Hou, Y. Bao, J. Gong, Formation and transformation
a controlled environment, Meas. Sci. Technol. 23 (10) (2012) 105904–105915 (12). behavior of sodium dehydroacetate hydrates, Molecules 21 (4) (2016).
[35] C.I. Haider, M.J. Hounslow, A.D. Salman, T.O. Althaus, G. Niederreiter, S. Palzer, In- [68] S. Wu, H. Shen, K. Li, B. Yu, S. Xu, M. Chen, J. Gong, B. Hou, Agglomeration mecha-
fluence of environmental conditions on caking mechanisms in individual amor- nism of azithromycin dihydrate in acetone-water mixtures and optimization of the
phous food particle contacts, AICHE J. 60 (8) (2014) 2774–2787. powder properties, Ind. Eng. Chem. Res. 55 (2016) 4905–4910.
[36] J.J. Fitzpatrick, E. O'Callaghan, J. O'Flynn, Application of a novel cake strength tester [69] X. Zhang, Q. Yin, W. Du, J. Gong, Y. Bao, M. Zhang, Phase transformation between
for investigating caking of skim milk powder, Food Bioprod. Process. 86 (3) (2008) anhydrate and monohydrate of sodium dehydroacetate, Ind. Eng. Chem. Res. 54
198–203. (13) (2015) 3438–3444.
[37] J.J. Fitzpatrick, N. Descamps, Comparison of a uniaxial force displacement tester [70] L. Komunjer, C. Affolter, Absorption-evaporation kinetics of water vapour on highly
with a ring shear tester for assessing powder cake strength, Part. Sci. Technol. hygroscopic powder: case of ammonium nitrate, Powder Technol. 157 (1) (2005)
Int. J. 31 (2) (2013) 199–203. 67–71.
[38] A.H.J. Paterson, G.F. Brooks, J.E. Bronlund, K.D. Foster, Development of stickiness in [71] D. Hornbaker, R. Albert, I. Albert, A.L. Barabasi, P. Schiffer, What keeps sandcastles
amorphous lactose at constant T-T(g) levels, Int. Dairy J. 15 (5) (2005) 513–519. standing? Nature 387 (1997) 765.
[39] S.W. Billings, J.E. Bronlund, A.H.J. Paterson, Effects of capillary condensation on the [72] R.A. Fisher, On the capillary forces in an ideal soil; correction of formulae given by
caking of bulk sucrose, J. Food Eng. 77 (4) (2006) 887–895. W. B. Haines, J. Agric. Sci. 16 (1926) 492–505.
[40] P.C. Knight, S.H. Johnson, Measurement of powder cohesive strength with a pene- [73] W.B. Haines, Studies in the physical properties of soils: II. A note on the cohesion
tration test, Powder Technol. 54 (4) (1988) 279–283. developed by capillary forces in an ideal soil, J. Agric. Sci. 15 (1925) 529–535.
[41] M. Pasha, C. Hare, A. Hassanpour, M. Ghadiri, Analysis of ball indentation on cohe- [74] H.J. Butt, M. Kappl, Normal capillary forces, Adv. Colloid Interf. Sci. 146 (2009) 48–60.
sive powder beds using distinct element modeling, Powder Technol. 233 (2) [75] M. Scheel, R. Seemann, M. Brinkmann, M. Di Michiel, A. Sheppard, B. Breidenbach,
(2013) 80–90. S. Herminghaus, Morphological clues to wet granular pile stability, Nat. Mater. 7
[42] G.M. Walker, C.R. Holland, M.N. Ahmad, J.N. Fox, A.G. Kells, Granular fertilizer ag- (2008) 189–193.
glomeration in accelerated caking tests, Ind. Eng. Chem. Res. 38 (10) (1999) [76] K.N. Manahiloh, B. Muhunthan, Characterizing liquid phase fabric of unsaturated
4100–4103. specimens from X-ray computed tomography images, Unsaturated Soils: Research
[43] A.A.C. Bode, S. Jiang, J.A.M. Meijer, W.J.P.V. Enckevort, E. Vlieg, Growth inhibition of and Applications, Springer, Berlin, 2012.
sodium chloride crystals by anticaking agents: in situ observation of step pinning, [77] K.N. Manahiloh, B. Muhunthan, W.J. Likos, Microstructure-based effective stress for-
Cryst. Growth Des. 12 (12) (2012) 5889–5896. mulation for unsaturated granular soils, Int. J. Geomech. 16 (6) (2016), D4016006.
[44] A.A.C. Bode, V. Vonk, F.J.V.D. Bruele, D.J. Kok, A.M. Kerkenaar, M.F. Mantilla, Anti- [78] M. Urso, C. Lawrence, M. Adams, Pendular, funicular, and capillary bridges: results
caking activity of ferrocyanide on sodium chloride explained by charge mismatch, for two dimensions, J. Colloid Interface Sci. 220 (1999) 42–56.
Cryst. Growth Des. 12 (4) (2012) 1919–1924. [79] M.E.D. Urso, C.J. Lawrence, M.J. Adams, A two-dimensional study of the rupture of
[45] K. Voïtchovsky, J.J. Kuna, S.A. Contera, E. Tosatti, F. Stellacci, Direct mapping of the funicular liquid bridges, Chem. Eng. Sci. 57 (2002) 677–692.
solid-liquid adhesion energy with subnanometre resolution, Nat. Nanotechnol. 5 [80] K. Murase, T. Mochida, H. Sugama, Experimental and numerical studies on liquid
(6) (2010) 401-5. bridge formed among three spheres, Granul. Matter 6 (2004) 111–119.
[46] M. Ricci, P. Spijker, F. Stellacci, J.F. Molinari, K. Voïtchovsky, Direct visualization of [81] K. Murase, T. Mochida, Y. Sagawa, H. Sugama, Estimation on the strength of a liquid
single ions in the stern layer of calcite, Langmuir 29 (7) (2013) 2207–2216. bridge adhered to three spheres, Adv. Powder Technol. 19 (2008) 349–367.
[47] M. Ricci, J.J. Segura, B.W. Erickson, G. Fantner, F. Stellacci, K. Voïtchovsky, Growth [82] G. Gagneux, O. Millet, An analytical framework for evaluating the cohesion effects
and dissolution of calcite in the presence of adsorbed stearic acid, Langmuir 31 of coalescence between capillary bridges, Granul. Matter 18 (2016) 16.
(27) (2015) 7563-71. [83] C. Semprebon, M. Scheel, S. Herminghaus, R. Seemann, M. Brinkmann, Liquid mor-
[48] U. Zafar, C. Hare, A. Hassanpour, M. Ghadiri, Drop test: a new method to measure phologies and capillary forces between three spherical beads, Phys. Rev. E 94 (1–1)
the particle adhesion force, Powder Technol. 264 (3) (2014) 236–241. (2016), 012907.
[49] M. Maghsoodi, On the mechanism of agglomeration in suspension, Adv. Pharm. [84] K. Brakke, The surface evolver and the stability of liquid surfaces, Philos. Trans. R.
Bull. 2 (1) (2012) 25–30. Soc. Lond. A Math. Phys. Eng. Sci. 354 (1996) 2143–2157.
[50] J. Feeney, J.J. Fitzpatrick, Visualization of the caking behavior between two powder [85] L.Q. Chen, Mathematics of Microstructure Evolution, Minerals, Metals & Materials
particles, Part. Sci. Technol. 29 (5) (2011) 397–406. Society, Cleveland, 1995.
[51] H.J. Cheng, S.S. Hsiau, The study of granular agglomeration mechanism, Powder [86] J.P. Wang, E. Gallo, B. François, F. Gabrieli, P. Lambert, Capillary force and rupture of
Technol. 199 (3) (2010) 272–283. funicular liquid bridges between three spherical bodies, Powder Technol. 305
[52] A.W. Adamson, Physical Chemistry of Surfaces, John Wiley and Sons, New York, (2017) 89–98.
1963. [87] C.J. Van Oss, Interfacial Forces in Aqueous Media, second ed. Marcel Dekker, New
[53] M.K. Ghorab, K. Marrs, L.S. Taylor, L.J. Mauer, Water-solid interactions between York, 2006.
amorphous maltodextrins and crystalline sodium chloride, Food Chem. 144 (2) [88] J.H. Clint, Adhesion and components of solid surface energies, Curr. Opin. Colloid
(2014) 26–35. Interface Sci. 6 (1) (2001) 28–33.
[54] M. Kaiho, M. Chikazaw, T. Kanazawa, Relationship between adsorption of water- [89] W. Wu, G.H. Nancollas, Determination of interfacial tension from crystallization
vapor on ground sodium-chloride and caking of salt, Nippon Kagaku Kaishi 2 and dissolution data: a comparison with other methods, Adv. Colloid Interf. Sci.
(1974) 233–238. 79 (2–3) (1999) 229–279.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052
M. Chen et al. / Powder Technology xxx (2017) xxx–xxx 17

[90] C.D. Volpe, D. Maniglio, M. Brugnara, S. Siboni, M. Morra, The solid surface free en- [108] W.K. Burton, N. Cabrera, F.C. Frank, The growth of crystals and the equilibrium
ergy calculation: I. In defense of the multicomponent approach, J. Colloid Interface structure of their surfaces, Philos. Trans. R. Soc. Lond. A 243 (243) (1951) 299–358.
Sci. 271 (2) (2004) 434–453. [109] J.W. Mullin, Crystallization, Butterworth-Heinemann, Oxford, 2001.
[91] E.M. Ålander, Å.C. Rasmuson, Agglomeration and adhesion free energy of paracet- [110] D. Toroz, I. Rosbottom, T. Turner, D. Corzo, R. Hammond, X. Lai, K. Roberts, Towards
amol crystals in organic solvents, AICHE J. 53 (10) (2007) 2590–2605. an understanding of the nucleation of alpha-para amino benzoic acid from
[92] G. Hwang, J. Yang, C.H. Lee, I.S. Ahn, B.J. Mhin, New selection criterion for a base ethanolic solutions: a multi-scale approach, Faraday Discuss. 179 (2015) 79–114.
polar liquid in the Lifshitz-van der waals/lewis acid–-base approach, J. Phys. [111] A.S. Myerson, B.L. Trout, Nucleation from solution, Science 341 (2013) 855–856.
Chem. C 115 (25) (2011) 12458–12463. [112] P.G. Vekilov, Nucleation, Cryst. Growth Des. 10 (2010) 5007–5019.
[93] R.M. Kirsch, R.A. Williams, U. Bröckel, R.B. Hammond, X. Jia, Direct observation of [113] R.J. Davey, S.L. Schroeder, J.H. Ter Horst, Nucleation of organic crystals a molecular
the dynamics of bridge formation between urea prills, Ind. Eng. Chem. Res. 50 perspective, Angew. Chem. 52 (8) (2013) 2166–2179.
(20) (2011) 11728–11733. [114] A.F. Wallace, L.O. Hedges, A. Fernandez-Martinez, P. Raiteri, J.D. Gale, G.A. Waychunas,
[94] A. Vallatos, R.M. Kirsch, R.A. Williams, R.B. Hammond, X. Jia, U. Bröckel, M.M. S. Whitelam, J.F. Banfield, J.J.D. Yoreo, Microscopic evidence for liquid-liquid separa-
Britton, Probing composition and molecular mobility in thin spherical films using tion in supersaturated CaCO3 solutions, Science 341 (2013) 885–889.
nuclear magnetic resonance measurements of diffusion, Ind. Eng. Chem. Res. 54 [115] D. Gebauer, M. Kellermeier, J.D. Gale, L. Bergstrom, H. Colfen, Pre-nucleation clus-
(26) (2015) 6825–6830. ters as solute precursors in crystallization, Chem. Soc. Rev. 43 (7) (2014)
[95] M. Langlet, M. Benali, I. Pezron, K. Saleh, P. Guigon, L. Metlas-Komunjer, Caking of 2348–2371.
sodium chloride: role of ambient relative humidity in dissolution and recrystalliza- [116] D. Erdemir, A.Y. Lee, A.S. Myerson, Nucleation of crystals from solution: classical
tion process, Chem. Eng. Sci. 86 (5) (2013) 78–86. and two-step models, Acc. Chem. Res. 42 (5) (2009) 621–629.
[96] S. Xu, J. Wang, K. Zhang, S. Wu, S. Liu, K. Li, Nucleation behavior of eszopiclone- [117] A.A.C. Bode, P.G.M. Pulles, M. Lutz, W.J.M. Poulisse, S. Jiang, J.A.M. Meijer, Sodium
butyl acetate solutions from metastable zone widths, Chem. Eng. Sci. 155 (2016) chloride dihydrate crystals: morphology, nucleation, growth, and inhibition,
248–257. Cryst. Growth Des. 15 (1) (2015) 42.
[97] G. Han, H. Chen, D. Chen, D. Yan, Determination of critical supersaturation from [118] D.W. Rutland, Fertilizer caking: mechanisms, influential factors, and methods of
microdroplet evaporation experiments, Cryst. Growth Des. 6 (5) (2006) prevention, Fertil. Res. 30 (1991) 99–114.
1175–1180. [119] G. Lu, J.R. Third, C.R. Müller, Discrete element models for non-spherical particle sys-
[98] A. Ali, N.A. Shah, A.K.S. Aqili, A. Maqsood, Impact of evaporation rates of Cd and Te tems: from theoretical developments to applications, Chem. Eng. Sci. 127 (2015)
on structural, morphological, optical, and electrical properties of CdTe thin films 425–465.
deposited by a two-sourced evaporation technique, Cryst. Growth Des. 6 (9) [120] G.M. Walker, T.R.A. Magee, C.R.H. And, M.N. Ahmad, J.N. Fox, N.A.M. And, Caking
(2006) 2149–2154. processes in granular NPK fertilizer, Ind. Eng. Chem. Res. 37 (2) (1998) 435–438.
[99] D.I. Trauffer, A.K. Maassel, R.C. Snyder, Non-needle-like morphology of β-glycine [121] C.Y. Wu, A.C.F. Cocks, Numerical and experimental investigations of the flow of
particles formed from water solutions via monodisperse droplet evaporation, powder into a confined space, Mech. Mater. 38 (4) (2006) 304–324.
Cryst. Growth Des. 16 (4) (2016). [122] C.Y. Wu, DEM simulations of die filling during pharmaceutical tabletting,
[100] J. Mahieux, M. Sanselme, S. Harthong, C. Melan, C. Aronica, L. Guy, Preparative res- Particuology 6 (6) (2008) 412–418.
olution of (±)-bis-tetralone by means of autoseeded preferential crystallization in- [123] C. Pei, C.Y. Wu, M. Adams, D. England, S. Byard, H. Berchtold, Contact electrification
duced by solvent evaporation (ASPreCISE), Cryst. Growth Des. 13 (8) (2013) and charge distribution on elongated particles in a vibrating container, Chem. Eng.
3621–3631. Sci. 125 (2014) 238–247.
[101] D. Ehre, K. Fang, J.E. Aber, S. Arnold, M.D. Ward, B.A. Garetz, Polymorphism in con- [124] C. Pei, C.Y. Wu, E. David, B. Stephen, H. Berchtold, M. Adams, DEM-CFD modeling of
tainerless crystallization, Cryst. Growth Des. 11 (10) (2011) 4572–4580. particle systems with long-range electrostatic interactions, AICHE J. 61 (6) (2015)
[102] M. Klimakow, K. Rademann, F. Emmerling, Toward novel pseudo-polymorphs of 1792–1803.
nifedipine: elucidation of a slow crystallization process, Cryst. Growth Des. 10 [125] C. Sauter, M.A. Emin, H.P. Schuchmann, S. Tavman, Influence of hydrostatic pres-
(6) (2010) 2693–2698. sure and sound amplitude on the ultrasound induced dispersion and de-
[103] S. Wu, M. Chen, K. Li, S. Xu, B. Yu, S. Liu, B. Hou, J. Gong, Solvent penetration medi- agglomeration of nanoparticles, Ultrason. Sonochem. 15 (4) (2008) 517–523.
ated phase transformation for the preparation of aggregated particles with well- [126] R.A. Lipasek, L.S. Taylor, L.J. Mauer, Effects of anticaking agents and relative humid-
defined shape, CrystEngComm 18 (48) (2016) 9223–9226. ity on the physical and chemical stability of powdered vitamin C, Food Chem. 76
[104] M. Volmer, Crystal growth, Z. Phys. 9 (1922) 193. (2011) 1062–1074.
[105] W. Kossel, Zur Theorie des Kristallwachstums, Nachr. Ges. Go¨ tt., 2, 1927, [127] R.A. Lipasek, J.C. Ortiz, L.S. Taylor, L.J. Mauer, Effects of anticaking agents and stor-
pp. 135–145. age conditions on the moisture sorption, caking, and flowability of deliquescent in-
[106] I.N. Stranski, Zur Theorie des Kristallwachstums, Z. Phys. Chem. 136 (1928) gredients, Food Res. Int. 45 (1) (2012) 369–380.
259–278. [128] A.A.C. Bode, M. Verschuren, M. Jansen, S. Jiang, J.A.M. Meijer, W.J.P.V. Enckevort, In-
[107] M. Ohara, P.C. Reid, Modelling Crystal Growth Rates From Solution, Prentice Hall, fluence of anticaking agents on the caking of sodium chloride at the powder and
Englewood Cliffs, 1973. two-crystal scale, Powder Technol. 277 (2015) 262–267.

Please cite this article as: M. Chen, et al., Caking of crystals: Characterization, mechanisms and prevention, Powder Technol. (2017), http://
dx.doi.org/10.1016/j.powtec.2017.04.052

You might also like