Composite inorganic membranes containing nanoparticles of hydrated zirconium dioxide for electrodialytic separation
© Dzyazko et al.; licensee Springer. 2014
Received: 6 December 2013
Accepted: 6 May 2014
Published: 29 May 2014
The aim of the work was to elucidate the nature of charge-selective properties of macroporous composite inorganic membranes modified with nanoparticles of hydrated zirconium dioxide. The membranes have been investigated using methods of standard contact porosimetry, potentiometry, electron microscopy and small-angle X-ray scattering. The ion exchanger has been found to deposit inside pores of ceramics. Differential curves of pore volume distribution have been resolved using Lorentz functions; each maximum has been related to structure elements of the matrix and ion exchanger by means of calculations according to homogeneous and heterogeneous geometrical models. It was found that the voids, the radius of which is 4 to 8 nm, are responsible for charge selectivity of the composite membranes. These pores are formed due to blocking of macropores of ceramics with aggregates of nanoparticles of the ion exchanger; the radius of these aggregates is 20 to 24 nm. The membranes were applied to desalination of the solution containing NaCl. The removal degree of the salt from the solution reached 95% and 9% for the composite and unmodified membranes, respectively.
KeywordsComposite ion exchange membranes Hydrated zirconium dioxide Nanoparticles Standard contact porosimetry Electrodialysis
Inorganic membranes can operate at high temperatures and in aggressive media; moreover, they are stable against fouling with organic matters [1, 2]. Since these materials possess remarkable properties, they are attractive for separation processes particularly for electromembrane techniques . However, application of ceramic separators to electromembrane processes is limited by an absence of charge selectivity in spite of a nanoporous active layer. This is due to extremely low ion exchange capacity (low surface charge density) of ceramics, since these materials are produced at high temperature , which does not provide retention of functional groups.
The conditions of thermal treatment of the membranes provided ion exchange ability of HZD.
Pores of 190 nm dominated in pristine ceramics. After modification, their size decreased to 80 nm [6, 7] indicating formation of an active layer inside the pores of ceramics, opposite to known inorganic materials for baromembrane separation [1, 2]. This location of the active layer provides its mechanical durability.
Predominant pores of the composite membranes [6, 7] cannot provide overlapping of intraporous diffusion double electric layers. In spite of this, the membranes were shown to possess charge selectivity. They demonstrate membrane potential in rather concentrated acid solutions . When the composite separators are applied to electrodialysis, the ion transport through these separators is due to migration of counter ions and electrolyte diffusion during electrodialysis . At the same time, no migration of co-ions through these separators was found.
Many types of ceramics contain larger pores (up to several microns) in comparison with the material investigated in [6, 7]. The aims of the work involve formation of the inner active layer in coarse-pored membranes, ascertainment of the cause of their charge selectivity and application of these materials to electromembrane separation.
A method of standard contact porosimetry (SCP) was applied to membrane investigation. The method allows us to obtain pore size distribution in a wide interval of 1 nm to 300 μm as well as total volume of micropores of 0.3 to 1 nm [8–11]. The SCP method is non-destructive, since it does not require high pressure compared to mercury porosimetry. Thus, small pores can be determined more exactly. Moreover, analysis of integral pore size distribution gives a possibility to determine particle size using geometrical models [12–14]. However, in the case of composites, the particle size of their components can be close to each other; as a result, the constituents cannot be recognized. Thus, the next important task of the work is to develop an approach for analysis of pore size distributions for composite materials.
Synthesis of the composite membranes
Planar ceramic membranes (matrix) based on TiO2 (TAMI GmbH, Hermsdorf, Germany), which contain no active layer, were used. Sol of insoluble zirconium hydroxocomplexes was prepared by adding a NH4OH solution (1,000 mol m−3) to a 1 M ZrOCl2 solution (1,000 mol m−3) at 353 K followed by boiling for 48 h and storage for 48 days at 298 K [6, 7]. Sol was analysed with a dynamic light-scattering method using a Zetasizer Nano ZS device (Malvern Instruments, Worcestershire, UK). Stability of particle distribution has been found after long-term storage.
The membrane was impregnated with sol, treated with a NH4OH solution (1,000 mol m−3), dried at ≈ 298 K and heated at 423 K [6, 7]. A layer of the ion exchanger was removed from the outer surface of the membrane with ultrasonic activation at 30 kHz. The procedure, which involves impregnation, HZD deposition, drying, heating and ultrasonic treatment, was repeated two and seven times. The samples were marked as TiO2 (matrix), TiO2-HZD-2 and TiO2-HZD-7 (modified membranes). Similar growth of HZD content (2.2 to 2.4 mass%) was reached both for TiO2-HZD-2 (in comparison with the matrix) and TiO2-HZD-7 (in comparison with TiO2-HZD-2).
After dehydration of sol at room temperature, its solid constituent was investigated using a JEOL JEM 1230 transmission electron microscope (JEOL Ltd., Tokyo, Japan). Finely dispersed powders obtained both from initial and modified membranes were also researched. Before the investigations, the powders of ceramics were treated with a CH3COOH solution (100 mol m−3) to shade the modifier particles.
Transverse section of the membranes was investigated using a Zeiss EVO 50XVP scanning electron microscope (Carl Zeiss AG, Oberkochen, Germany).
Small-angle X-ray scattering
Finely dispersed powders of the membranes were inserted into cuvettes, the thickness of which was 0.1 to 0.2 mm, with 17-μm-thick Mylar windows. Small-angle X-ray scattering (SAXS) curves were obtained in a vacuum Kratky camera using a Cu-anode tube. Recording of SAXS data has been carried out under the conditions of multiple scanning of a scintillation detector at scattering angles of 0.03° to 4.0°. The first treatment of the SAXS data was carried out by means of the FFSAXS11 program. The exclusion of parasitic scattering by the camera and cuvette windows, normalization of the scattered intensity to absolute units, and the introduction of the collimation correction were performed.
Standard contact porosimetry
The membranes were heated at 423 K before the measurements. Octane was used as a working liquid [8–11]. The curves of differential pore volume (V) distribution (, where r is the pore radius) were resolved by Lorentz components using the PeakFit v. 4.12 program. Treatment of the curves involved resolution within the intervals of pore radius of 1 to 100 nm and 1 to 105 nm and comparison of the data for peaks with a maximum at ≈ 100 nm. Data adequacy is confirmed by coincidence of these maxima in two diapasons and high correlation coefficient (0.99). This procedure was necessary because the values are rather low at 1 to 100 nm.
The particle density of the membranes (ρp) was determined using a pycnometer method , and the bulk density (ρb) was estimated from geometrical parameters.
Sorption capacity and potentiometric measurements
Ion exchange capacity of the membranes has been determined by their treatment with a HCl solution (100 mol m−3), washing with deionized water followed by treatment with a NaOH solution (100 mol m−3) and analysis of the eluate using an I-160 M potentiometer and Cl−-selective electrode. The solution was neutralised with HNO3 before the measurements.
The difference of was found, and then its dependence on a± (i.e. on activity of more concentrated solution, a2) was plotted. At last, the transport number was calculated from a slope of the curve.
A solution containing NaCl (10 mol m−3), the volume of which was 50 cm3, circulated from the desalination compartment with a flow velocity of 1 cm3 s−1 (first liquid line). The second line provided circulation of the solution, which contained initially K2SO4 (1,000 mol m−3), through the cathode and anode compartments (second line). At last, a H2SO4 solution (100 mol m−3) circulated through the concentration compartment. The content of Cl− and Na+ species in the solution being purified was controlled by means of ion-selective electrodes. The removal degree of NaCl from the solution was calculated as , where C is the concentration at time τ and Ci is the initial concentration. The current efficiency was calculated as where z is the charge number, n is the amount of electrolyte removed from the solution, i is the current density and A is the membrane area.
Sol of zirconium hydroxocomplexes
During sol formation, fragmentation and defragmentation of nanoparticles occur simultaneously . As a result, sol can contain several types of particles . The first one is non-aggregated particles; their merging leads to formation of larger ones.
Structure of membranes
Porous structure of the composite membranes
Increase of mass,%
Density, kg m−3
Volume of micropores, cm3g−1
3.0 × 10−4
1.5 × 10−3
5.0 × 10−3
where Δρ is the difference of electron densities between the particle and its environment, and Rg is the gyration radius, which has been determined from the slope of the linear part of lnI − q2 curve at q = 1.1 to 1.6 nm−1 (inset of Figure 5). The particle radius (rp) was calculated as 1.29Rg[21, 22]. It was found, that rp = 3 nm.
The logI − logq curve (where I is the intensity, q is the scattering vector), which has been obtained for pristine ceramics, is characterized by a long straight part within the interval of scattering vector of 2.82 × 10−2 to 1.1 nm−1. This interval corresponds to particles II of the ceramic matrix. The slope of the curve is −4; this indicates smooth surface of these particles, which include no constituents [21, 22]. The curves demonstrate deviation from linearity under low q values; thus, the order of particle size is about 100 nm. Larger particles cannot be determined with a SAXS method.
Regarding the modified membranes, a small change of the slope of the linear part (q = 2.82 × 10−2 to 1.1 nm−1) has been found. Thus, deposition of the modifier on particles II is inconsiderable. However, a change of slope of the lnI − q 2 curve at wider angles indicates the presence of HZD particles, which are smaller, than particles I of the matrix.
The results obtained with a pycnometer method allow us to determine porosity of the samples. Modification of the matrix causes an increase of bulk density of the membranes; however, no change of particle density has been found. Thus, the particle densities of the ion exchanger and matrix are equal. Porosity (ϵm for the initial matrix and for the modified membranes) has been calculated as . The porosity decreases in the order: TiO2 > TiO2-HZD-7 > TiO2-HZD-2.
The ratio of values is 1:3.9 for TiO2-HZD-2 and TiO2-HZD-7 membranes, respectively (here, Vmicr and are the volume of micropores for pristine and modified membranes, respectively). The ratio of (here, m and m l are the mass of matrix and modified membrane, respectively) is 1:1.9. This is evidently due to different porous structures of HZD: more compact structure is attributed to the TiO2-HZD-2 sample.
Parameters of globular model for the matrix and ion exchanger layer
S, m2 kg−1
1.05 × 105
2.09 × 105
CFC or HXG
CFC or HXG
, , m2 kg−1
7.77 × 105
2.27 × 105
3.06 × 104
3.88 × 104
r g , nm
r n a, nm
r c a, nm
Calculation of porous structure according to globular models
Both homogeneous and heterogeneous globular models were applied to relate the maxima either to the matrix or to ion exchanger. The models have been developed by A.P. Karnaukhov; their main principles are described in [12–14]. Parameters of the models are radii of globules (rp), pore necks (rn) and pore cavities (rc); the values of surface and porosity are also used. The magnitudes of rn and rc are calculated using special factors for each type of globule packing: rn = 0.41rp and rc = 0.73rp for simple cubic (SC), rn = 0.22rp and rc = 0.29rp for body-centred cubic (BCC), and rn = 0.15rp and rc = 0.41rp for hexagonal (HXG) or face-centred cubic packing (FCC). A packing type is determined from the porosity (see Table 2).
According to the homogeneous model, the effective particle size was calculated as . The heterogeneous model provides analysis of integral pore size distributions [12–14]. Porosity caused by different types of particles is determined according to each semi-wave. In the case of composite materials, it is difficult to recognize their components, when sizes of the particles are close to each other. We have proposed resolution of differential pore size distributions by Lorentz components; these functions provide the best agreement of experimental and calculated curves. The globular model was assumed to give pairs of peaks: the first maximum corresponds to narrowing of pores between globules (pore necks), and the second one is related to their widening (pore cavities). Then, the porosity, which is attributed to the peak, was found by means of peak integration. The surface of each type of pores was found as (matrix) and (ion exchanger), where ϵ or are the total porosity, and ϵ p is the porosity due to each type of particles.
Two additional peaks (1 to 3 nm) due to HZD are visible for modified membranes (see Figure 7b,c). Calculations give nanosized particles I, which evidently form a structure of the ion exchanger (particles I). Similar results were obtained using the homogeneous model. These particles are evidently associated into aggregates (particles II); pores between them give maxima at 8 nm for TiO2-HZD-2 and 4 and 6 nm for TiO2-HZD-7. Evidently, there are only HZD aggregates inside the matrix, since the SAXS data indicate no considerable change of surface of particles II of the matrix. Indeed, the size of particles II of the modifier is larger than the pores, which are formed by particles II of the matrix. In the case of TiO2-HZD-2, the maxima for necks and cavities are overlapped with a peak attributed to the matrix and cannot be separated. A shift of the peak at 39 nm (TiO2) to 52 nm (TiO2-HZD-7) has been found. This indicates formation of larger particles III; their size can be estimated approximately from the peak at 52 nm, which is related to pore necks. These particles are evidently located in the cavities of pores, which are caused by the largest particles III of the matrix. The peaks at r > 100 nm for modified membranes are shifted towards lower r values in comparison with the matrix. This indicates HZD deposition inside macropores of the ceramics.
Potentiometric transport numbers of counter ions
Potentiometric measurements give additional information about the membrane structure. No membrane potential (Em) has been registered for the matrix. Em > 0 V in the case of modified samples. Since the membranes show anion exchange ability in acidic media [6, 7], Cl− and H+ species are considered as counter- and co-ions, respectively.
where t is the transport number of Cl− in a solution, k is the shape coefficient (k = 2.8 for pores between globules), η is the surface charge density and C is the average value of concentrations of the solutions from two sides of the membranes. The surface charge density was estimated from sorption measurements as 0.07 C m−2 (TiO2-HZD-2) and 0.18 C m−2 (TiO2-HZD-7).
Formula (7) gives the transport number at which concentrations of the solutions from two sides of the membrane (C1 and C2) are close to each other. The r value was plotted as a function of C2-C1. Extrapolation of the curve to C2-C1 = 0 evidently gives the ‘real’ r magnitude, which has been estimated as 8 (TiO2-HZD-2) and 2 (TiO2-HZD-7) nm (Figure 8).
where ϵ0 is the dielectric permittivity of vacuum, and ϵ is the dielectric constant (80 for water). The concentration, which corresponds to tm = 1, was found by extrapolation of tm − C curves (inset of Figure 8). The r values were estimated as 7 nm (TiO2-HZD-2) and 4 nm (TiO2-HZD-7).
These ‘corks’ isolate macropores, which are recognized with the porosimetry method as predominant. Large particles of sol can penetrate the matrix during the first modification procedure. After blocking of the matrix pores, only the smallest particles are able to enter the membrane; moreover, they form the loosening structure of the ion exchanger.
Electrodialysis of the solution containing NaCl
After 5 min
After 30 min
After 60 min
As seen from the table, the current efficiency (CE) decreased in time due to solution depletion. The highest removal degree (RD) and current efficiency were found for the TiO2-HZD-7 membrane. This membrane is characterized by the smallest size of pores, which determine charge selectivity. Moreover, the highest surface charge density is reached for this separator.
The composite inorganic membranes, which contain the active layer of the HZD layer inside coarse-pored ceramics, have been obtained. This has been proved by means of SEM, TEM and SAXS technique. The SCP method followed by resolution of differential pore size distribution, calculations according to homogeneous and heterogeneous geometrical models and potentiometric measurements allow us to determine structure of composite membranes. The approach, which is based on analysis of differential pore size distribution, gives a possibility to recognize each component of a composite. Application of integral pore distribution [12–14] is difficult, when the particle sizes of the constituents are close to each other.
The ceramic matrix is formed mainly with particles of micron size, which are distorted due to annealing and pressure. The ion exchanger consists of nanosized particles, the radius of which is 3 to 5 nm. The nanoparticles form aggregates (rp = 20 to 23 nm). The larger particles form pores, which are responsible for charge selectivity. Radii of narrowing of these pores have been estimated as 4 to 8 nm; this is in agreement with porosimetry data. Charge selectivity is also due to ion exchange ability of HZD, which is retained under thermal treatment of the membranes. The materials can be used for electromembrane separation; the modified membranes demonstrate higher desalination degree and current efficiency in comparison with the pristine separator. Mechanical stability of the active layer is provided by its location inside pores of ceramics. As expected, the membranes can be used in aggressive media as well as for treatment of solutions containing organic substances.
List of symbols
A area (m2)
a activity (mol m−3)
C concentration (mol m−3)
D diffusion coefficient (m2 s−1)
Em membrane potential (V)
F Faraday constant (96,485 A s mol−1)
I intensity (cm−1)
i current density (A m−2)
ilim limiting current density (A m−2)
k shape coefficient (dimensionless)
km mass transport coefficient (m s−1)
m mass of matrix (kg)
m l mass of modified membrane (kg)
n amount of species (mol)
q scattering vector (nm−1)
R gas constant (8.31 J mol−1 K−1)
Rg gyration radius (nm)
r radius of pores (m, nm)
r c radius of pore cavities (m, nm)
r n radius of pore necks (m, nm)
r p radius of globules (m, nm)
S surface (m2 kg−1)
Sm surface of a composite membrane (m2 kg−1)
T temperature (K)
t transport number through the solution (dimensionless)
tm transport number through the membrane (dimensionless)
V pore volume (cm3 g−1)
Vmicr volume of micropores in a matrix (cm3 g−1)
V/micr volume of micropores in a matrix (cm3 g−1)
z charge number (dimensionless)
ϵ porosity of a matrix (dimensionless)
ϵ/ porosity of a modified membrane (dimensionless)
ϵd dielectric constant (dimensionless);
ϵp porosity due to particles of chosen size (dimensionless)
ϵ0 dielectric permittivity of free space (8.85 × 10−12 F m−1)
η surface charge density (C m−2)
ν viscosity (m2 s−1)
ρ electron density (dimensionless)
ρp particle density (kg m−3)
ρb bulk density (kg m−3)
τ time (s)
ω linear flow velocity (m s−1)
Re Reynolds number (dimensionless)
Sc Schmidt number (dimensionless)
Sh Sherwood number (dimensionless)
current efficiency (%)
hydrated zirconium dioxide
removal degree (%)
standard contact porosimetry.
The work was supported by projects within the framework of programs supported by the government of Ukraine ‘Nanotechnologies and nanomaterials’ (grant no. 184.108.40.206) and by the National Academy of Science of Ukraine ‘Problems of stabile development, rational nature management and environmental protection’ (grant no. 30-12) and ‘Fundamental problems of creation of new materials for chemical industry’ (grant no. 49/12).
- Buekenhoudt A: Stability of porous ceramic membranes. Membr Sci Technol 2008, 13: 1.View ArticleGoogle Scholar
- Bose S, Das C: Preparation and characterization of low cost tubular ceramic support membranes using sawdust as a pore-former. Mater Let 2013, 110: 152.View ArticleGoogle Scholar
- Martí-Calatayud MC, García-Gabaldón M, Pérez-Herranz V, Sales S, Mestre S: Synthesis and electrochemical behavior of ceramic cation-exchange membranes based on zirconium phosphate. Ceram Intern 2013, 39: 4045. 10.1016/j.ceramint.2012.10.255View ArticleGoogle Scholar
- Ghosh D, Sinha MK, Purkait MK: A comparative analysis of low-cost ceramic membrane preparation for effective fluoride removal using hybrid technique. Desalination 2013, 327: 2.View ArticleGoogle Scholar
- Amphlett CB: Inorganic Ion-Exchangers. New York: Elsevier; 1964.Google Scholar
- Dzyaz’ko YS, Belyakov VN, Stefanyak NV, Vasilyuk SL: Anion-exchange properties of composite ceramic membranes containing hydrated zirconium dioxide. Russ J Appl Chem 2006, 80: 769. 10.1134/S0036024406050177View ArticleGoogle Scholar
- Dzyazko YS, Mahmoud A, Lapicque F, Belyakov VN: Cr(VI) transport through ceramic ion-exchange membranes for treatment of industrial wastewaters. J Appl Electrochem 2007, 37: 209. 10.1007/s10800-006-9243-7View ArticleGoogle Scholar
- Volfkovich YM, Sosenkin VE, Bagotzky VS: Structural and wetting properties of fuel cell components. J Power Sources 2010, 195: 5429. 10.1016/j.jpowsour.2010.03.002View ArticleGoogle Scholar
- Volfkovich YM, Sakars AV, Volinsky AA: Application of the standard porosimetry method for nanomaterials. Int J Nanotechnol 2005, 2: 292. 10.1504/IJNT.2005.008066View ArticleGoogle Scholar
- Volfkovich YM, Bagotzky VS: The method of standard porosimetry: 1. Principles and possibilities. J Power Sources 1994, 48: 327. 10.1016/0378-7753(94)80029-4View ArticleGoogle Scholar
- Volfkovich YM, Bagotzky VS, Sosenkin VE, Blinov IA: The standard contact porosimetry. Colloids Surf A: Physicochem Eng Aspect 2001, 187–188: 349.View ArticleGoogle Scholar
- Szczygieł J: Optimising the porous structure of heterogeneous reforming catalysts with a globular model of the grain. Comp Chem Eng 2011, 35: 2334. 10.1016/j.compchemeng.2011.06.004View ArticleGoogle Scholar
- Gierak A, Leboda R, Tracz E: Topography and morphology of the carbon deposit obtained by pyrolysis of methylene chloride on a silica gel surface. J Analyt Appl Pyrolysis 1988, 13: 89. 10.1016/0165-2370(88)80050-0View ArticleGoogle Scholar
- Leboda R, Mendyk E, Gierak A, Tertykh VA: Hydrothermal modification of silica gels (xerogels) 2. Effect of the duration of treatment on their porous structure. Colloids and Surfaces A: Physicochem. Eng Aspect 1995, 105: 191. 10.1016/0927-7757(95)03272-XView ArticleGoogle Scholar
- Jones FE, Schoonover RM: Handbook of Mass Measurements. London: CRC Press; 2002.View ArticleGoogle Scholar
- Helfferich F: Ion Exchange. New York: Dover; 1995.Google Scholar
- Berezina NP, Kononenko NA, Dyomina OA, Gnusin NP: Characterization of ion-exchange membrane materials: properties vs structure. Adv Colloid Interface Sci 2008, 139: 3. 10.1016/j.cis.2008.01.002View ArticleGoogle Scholar
- Brinker CJ, Scherer GW: Sol–Gel Science: The Physics and Chemistry of Sol–Gel Process. Amsterdam: Elsevier; 1990.Google Scholar
- Alves-Rosa MA, Martins L, Pulcinelli SH, Santilli CV: Design of microstructure of zirconia foams from the emulsion template properties. Soft Matter 2013, 9: 550. 10.1039/c2sm26842fView ArticleGoogle Scholar
- Guinier A, Fournet G: Small-Angle Scattering of X-Rays. New York: Wiley; 1955.Google Scholar
- Fagherazzi G, Ploizzi S, Bettinelli M, Speghini A: Yttria-based nano-sized powders: a new class of fractal materials obtained by combustion synthesis. J Mater Res 2000, 15: 586. 10.1557/JMR.2000.0087View ArticleGoogle Scholar
- Sastry PU, Sen D, Mazumder S, Chandrasekaran S: Fractal behavior of nanocrystalline ceria–yttria solid solution. J Solid State Chem 2003, 176: 57. 10.1016/S0022-4596(03)00344-XView ArticleGoogle Scholar
- Volfkovich YM: Influence of the electric double layer on the internal interface in an ion exchanger on its electrochemical and sorption properties. Soviet Electrochemistry 1984, 20: 621.Google Scholar
- Robinson RA, Stokes RH: Electrolyte Solutions. Mineola NY: Dover; 2002.Google Scholar
- Walsh F: A First Course in Electrochemical Engineering. London: Alresford Press; 1993.Google Scholar
- Parsons R: Handbook of Electrochemical Constants. London: Butterworth Scientific Publications; 1959.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.